• branched-chain aminotransferase;
  • glutamate;
  • nitrogen shuttle;
  • retina;
  • brain;
  • gabapentin


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The relationship between neuronal glutamate turnover, the glutamate/glutamine cycle and de novo glutamate synthesis was examined using two different model systems, freshly dissected rat retinas ex vivo and in vivo perfused rat brains. In the ex vivo rat retina, dual kinetic control of de novo glutamate synthesis by pyruvate carboxylation and transamination of α-ketoglutarate to glutamate was demonstrated. Rate limitation at the transaminase step is likely imposed by the limited supply of amino acids which provide the α-amino group to glutamate. Measurements of synthesis of 14C-glutamate and of 14C-glutamine from H14CO3 have shown that 14C-amino acid synthesis increased 70% by raising medium pyruvate from 0.2 to 5 mm. The specific radioactivity of 14C-glutamine indicated that ∼30% of glutamine was derived from 14CO2 fixation. Using gabapentin, an inhibitor of the cytosolic branched-chain aminotransferase, synthesis of 14C-glutamate and 14C-glutamine from H14CO3 was inhibited by 31%. These results suggest that transamination of α-ketoglutarate to glutamate in Müller cells is slow, the supply of branched-chain amino acids may limit flux, and that branched-chain amino acids are an obligatory source of the nitrogen required for optimal rates of de novo glutamate synthesis. Kinetic analysis suggests that the glutamate/glutamine cycle accounts for 15% of total neuronal glutamate turnover in the ex vivo retina. To examine the contribution of the glutamate/glutamine cycle to glutamate turnover in the whole brain in vivo, rats were infused intravenously with H14CO3. 14C-metabolites in brain extracts were measured to determine net incorporation of 14CO2 and specific radioactivity of glutamate and glutamine. The results indicate that 23% of glutamine in the brain in vivo is derived from 14CO2 fixation. Using published values for whole brain neuronal glutamate turnover, we calculated that the glutamate/glutamine cycle accounts for ∼60% of total neuronal turnover. Finally, differences between glutamine/glutamate cycle rates in these two model systems suggest that the cycle is closely linked to neuronal activity.

Abbreviations used

branched-chain amino acids


branched-chain aminotransferase


cytosolic branched-chain aminotransferase


mitochondrial branched-chain aminotransferase


branched-chain α-keto acids


phosphate-buffered saline

The glutamate/glutamine cycle in neural tissue efficiently prevents excessive accumulation of glutamate in the interstitium which, if unchecked, would induce excitotoxic neurodegeneration. This pathway involves the synaptic release of glutamate from neurons, rapid and efficient glutamate uptake from the synaptic space by glia, conversion of glutamate to glutamine by glial glutamine synthetase followed by release of glutamine to the interstitium and uptake by the neurons for conversion to glutamate (Shank and Aprison 1981; Cooper and Plum 1987). It was thought that cycle operation involved minimal loss of glutamine or glutamate carbon skeleton (Lapidot and Gopher 1994; Sibson et al. 1997; Shen et al. 1999).

Nevertheless, studies by Yu et al. (1983), Sonnewald et al. (1993, 1996) and Hutson et al. (1998) of glutamate metabolism in cultured neonatal astrocytes indicate that up to 30% of the glutamate taken up by the astrocytes is not converted directly to glutamine but is transaminated to α-ketoglutarate and then, via citric acid cycle enzymes, to malate and oxaloacetate, which in turn may be decarboxylated to pyruvate (cf. Fig. 1). Studies in intact brains have so far been unable to demonstrate unequivocally that citric acid cycle intermediates are decarboxylated in the glia. In one study, the results were inconclusive (Haberg et al. 1998) and in another the authors localized the citric acid cycle decarboxylation to neuronal cells (Cruz et al. 1998). Irrespective of the site of decarboxylation (or pyruvate cycling), equivalent rates of pyruvate carboxylation (anaplerosis) must take place to recover the lost carbon of glutamate. Although pyruvate carboxylase is expressed abundantly in the brain and localized to glia (Shank et al. 1993), the contribution of pyruvate carboxylase products to the maintenance of glutamate levels has been controversial (Lapidot and Gopher 1994; Sibson et al. 1997; Gruetter et al. 1998; Griffin et al. 1999; Shen et al. 1999; Hassel and Brathe, 2000a,b).


Figure 1. Hypothetical branched-chain amino acid (BCAA) shuttle. The shuttle proper is highlighted. BCAA in the glia can transaminate with α-ketoglutarate to form glutamate. The product branched-chain α-keto acids (BCKA) diffuse across the interstitium (shown as gray) and enter the neurons where they are converted back to BCAA by transamination with glutamate via the cytosolic branched-chain aminotransferase (BCATc). Other abbreviations are: BCATm, mitochondrial branched-chain aminotransferase; GDH, glutamate dehydrogenase; Gln, glutamine; Gln'ase, glutaminase; Glu, glutamate; αKG, α-ketoglutarate; PC, pyruvate carboxylase; pyr, pyruvate.

Download figure to PowerPoint

Our previous studies of the rate of anaplerosis in neonatal cultured astrocytes and the resultant accumulation of citric acid cycle intermediates showed that pyruvate carboxylase was active and dependent on levels of added pyruvate (Gamberino et al. 1997; Hutson et al. 1998). The study also indicated that the conversion of the excess citric acid cycle intermediates to glutamate could be severely limited by lack of a source of nitrogen. Because NH3 could not supply the needed nitrogen (Gamberino et al. 1997), we proposed that branched-chain amino acids (BCAA) might be the physiological source for the α-amino nitrogen of glutamate (Hutson et al. 1998). This proposal is supported by the work of Yudkoff and coworkers (Yudkoff et al. 1996) in cultured brain cells as well as by 15N studies (Kanamori et al. 1998) in the intact brain. Subsequent studies of brain branched-chain amino acid transaminase (BCAT) isoenzymes (Bixel et al. 1997; Hutson et al. 1998) have led to the suggestion that a BCAT-dependent nitrogen shuttle catalyzes the conversion of astrocyte citric acid cycle intermediates to glutamate. In the proposed shuttle, net amination of branched-chain α-keto acids (BCKA) to BCAA in neurons counterbalances the net synthesis of glutamate from α-ketoglutarate in glia (Fig. 1).

In an effort to resolve discrepancies between studies of intact brain (Sibson et al. 1997; Cruz et al. 1998; Haberg et al. 1998; Hassel and Brathe, 2000a,b) and studies of cultured astrocytes (Shank et al. 1985, 1993; Sonnewald et al. 1993; Gamberino et al. 1997; Hutson et al. 1998), we employed intact freshly isolated rat retinas to study relationships between glutamate metabolism and CO2 fixation. Compared with brain, retina provides a model of intact neuronal–glial interactions that is more accessible to pharmacologic manipulations and acute metabolic determinations. To determine whether ex vivo retinal glutamate metabolism was similar to that in the whole brain, the retinal studies were supplemented by in vivo studies of rat brain. The results demonstrate for the first time that both pyruvate carboxylase and the BCAT isoenzymes are obligatorily involved in maintenance of glutamate levels in glutamatergic tissue.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Experimental animals

Sprague Dawley rats (200–400 g) were used for all experiments. Rats were housed under a 12-h light/dark cycle and allowed free access to food and water. The experimental protocols were approved by the institutional review committees of the Pennsylvania State University College of Medicine and of the Wake Forest University School of Medicine. Animals were routinely treated in accordance with the guidelines published in the NIH Guide for the Care and Use of Laboratory Animals.

CO2 fixation in ex vivo retinas

After anesthetizing the rats with Nembutal, their eyes were enucleated and retinas dissected in ice-cold buffer. Freshly isolated half-retinas were pre-incubated at 37°C for 3 min in 1 mL buffer A containing 118 mm NaCl, 4.7 mm KCl, 2.5 mm CaCl2, 1.2 mm KH2PO4, 1.17 mm MgSO4, 20 mm HEPES, 5 mm glucose, 0.05 mm NH4Cl and 25 mm NaH14CO3 (pH 7.4). Pyruvate (0.2 or 5 mm) was also included in the buffer, which was equilibrated with 95% O2, 5% CO2. Incubations were initiated by the addition of 150 µCi/mL H14CO3, after which the vessels were closed immediately to the atmosphere. Most incubations continued for 20 min, although in some experiments times were varied from 10 to 60 min. Reactions were stopped by removing the half retinas from the buffer and placing them in 2% perchloric acid. The buffer was acidified separately with perchloric acid (final concentration 2%) and unreacted 14CO2 was allowed to diffuse out of the acidified samples.

The total amount of product (non-volatile 14C metabolites) was determined by measuring the radioactivity in aliquots of the medium and tissue samples. Except as noted in the text, individual metabolite 14C data are reported as the sum of the radioactivity in the medium plus tissue extracts. Metabolites in the medium and retina extracts were separated by Dowex-1 acetate chromatography as previously described (Gamberino et al. 1997), and quantified by scintillation counting. The chromatographic procedure separates glutamine, glutamate, aspartate and lactic acid (Gamberino et al. 1997). The citric acid cycle intermediates remain on the Dowex columns and are estimated as the difference between the total counts in the sample and the counts that are eluted. This procedure has been validated in a previous publication (Gamberino et al. 1997). Before chromatography, 14C-pyruvate in the samples was converted stoichiometrically to lactate by treating the samples with excess NADH and lactate dehydrogenase. Therefore, the single lactate peak contains the 14C-labeling of both lactate and pyruvate. To determine specific radioactivity of glutamate and glutamine, the mass amount of the glutamate was measured fluorometrically using a standard enzymatic procedure (Williamson and Corkey 1969). Glutamine was measured after chromatographic separation from glutamate followed by a luminometric determination (Gamberino et al. 1997). ATP and creatine phosphate were measured fluorometrically by enzymatic assay (Williamson and Corkey 1969).

CO2 fixation in the in vivo brain

Rats (250–300 g) were anesthetized with an intramuscular injection of ketamine and xylazine (90 and 9 mg protein per kg body weight, respectively), and sterile surgery was performed to implant catheters in the carotid artery and jugular vein (Crist et al. 1998). The animals were permitted to recover from surgery for 24 h and allowed free access to food and water. Awake animals were infused through the jugular cannula using a priming dose of H14CO3 (0.75 mCi) followed by steady-state infusion of 1.4 mCi/h in buffered saline. During the infusion, blood samples were taken every 15 min and acid volatile (14CO2) radioactivity determined. Blood H14CO3 specific activity was maintained at a constant value in each animal by this protocol. The mean blood H14CO3 specific radioactivity for each rat, which varied from 146 to 189 d.p.m./nmol, was used to calculate the nmoles of H14CO3 incorporated into each brain metabolite.

At the end of the 60 min H14CO3 infusion, the rats were decapitated and their heads dropped rapidly into liquid nitrogen. The frozen brain tissue was removed from the skull, extracted with 6% perchloric acid, and 14C-labeled metabolites in the neutralized perchloric acid extracts were separated by chromatography on Dowex-1-acetate (Gamberino et al. 1997). Glutamine was eluted together with glucose and other neutral amino acids. To quantify 14C-glutamine specifically, an aliquot of the neutral fraction was incubated with glutaminase (Gamberino et al. 1997) and then chromatographed to measure the product, 14C-glutamate. To quantify 14C-glucose, a separate aliquot of the neutral fraction was incubated with hexokinase (5 U/mL). The product, 14C-glucose-6-phosphate, remained on the column when the incubation mixture was re-chromatographed and could be assessed indirectly by a decrease in radioactivity of the second compared with the first H2O elution or directly by the appearance of a peak of 14C-glucose-6-phosphate during elution. An aliquot of each neutralized extract was also chromatographed on a Dowex-1-chloride column to separate citric acid cycle intermediates (Gamberino et al. 1997). In this way all metabolites were completely separated and quantified. Recovery was greater than 95%.

Pyruvate carboxylase assay

Pyruvate carboxylase activity was assayed under optimal conditions in fresh samples of sonicated whole brain and excised retinas using the method described by Wilbur and Patel (1974).

Preparation of antibodies and western blotting

BCAT isoenzyme-specific antisera were raised in rabbits as described by Wallin et al. (1990). Purified recombinant human BCATm (Davoodi et al. 1998) and a purified rat BCATc N-terminal peptide were used as antigens. The rat BCATc peptide expression plasmid was prepared by using a polymerase chain reaction to obtain a rat BCATc partial cDNA encoding the first 62 amino acid residues of the rat BCATc protein (Hutson et al. 1995). The sense primer 5′-TTAGCCATATGGCCTACTTGTC-3′ contained an NdeI restriction site and the antisense primer 5′-ATATTCTCGAGTCATTAGACATCGGC-3′ contained an XhoI site. For increased accuracy, a polymerase mixture of Taq and Pwo polymerases (Roche Molecular Biochemicals, Indianapolis, IN, USA) was used. The amplified product was cloned into the pT7 vector (Novagen, Madison, WI, USA). The insert was sequenced on both strands, and the fidelity was verified by comparison with the original rat BCATc cDNA clone. The rat BCATc peptide plasmid cDNA was engineered by ligating the rat BCATc partial cDNA sequence into the pET-28a vector (Novagen) cut with NdeI and XhoI. Expression of the rat BCATc peptide, extraction and purification of the recombinant BCATc peptide through the Ni-NTA resin (Qiagen, Chatsworth, CA, USA) step were as described by Davoodi et al. (1998). After removal of the His-tag (Davoodi et al. 1998), the BCATc peptide was purified to homogeneity by gel-filtration on a Sephadex g-75 column (Sigma, St Louis, MO, USA).

For preparation of the affinity-purified antibodies, the antisera were each made 50% saturated with ammonium sulfate, and centrifuged for 10 min at 10 000 g. The supernatants were discarded, and the pellets were washed in phosphate-buffered saline (PBS). Subsequently each protein pellet was dissolved in PBS and loaded onto an affinity resin having the respective antigen as a ligand. The column was washed with PBS and the fraction of antigen-specific antibodies was eluted in 0.1 m acetate buffer (pH 4.0) containing 4 m urea and 0.5 m NaCl (Stanton et al. 1991). The affinity-purified antibodies were dialyzed immediately at 4°C against glycerol : water (1 : 1). The dialyzed antibodies were aliquoted and stored at −80°C. Human BCATm-Sepharose and rat BCATc peptide-Sepharose were prepared by coupling the purified human recombinant BCAT isoenzyme or purified rat BCATc peptide to Affigel 10 support (BioRad, Richmond, CA, USA) according to the manufacturer's directions.

SDS-PAGE was performed as described by Laemmli (1970) using 10% gels. For immunoblotting, proteins were transferred to Immobilon P membranes. Membranes were blocked with 5% BSA and incubated with immunoaffinity-purified rabbit anti-rat BCATc peptide antibody (1 : 1000 dilution) or immunoaffinity purified rabbit anti-human BCATm antibody (1 : 1000 dilution). The immunoreactive protein bands were visualized using the ECL system according to the manufacturer's instructions (Amersham, Arlington Heights, IL, USA).

Protein determinations

Perchloric acid-treated retinas from each incubation were sonicated and centrifuged. The aqueous perchloric extract (1 mL) was carefully decanted and 1 mL of 1 m NaOH was added to the precipitated protein and the solubilized protein assayed immediately using the BioRad reagent. Each half-retina contained ∼0.4 mg protein. Pulverized frozen brain tissue was extracted with perchloric acid by standard techniques (Williamson and Corkey 1969). The protein was separated by centrifugation from the aqueous perchloric acid extract, solubilized in 1 N NaOH, and assayed immediately using the BioRad reagent.


The H14CO3 was obtained from New England Nuclear Life Sciences Products (Boston, MA, USA). Prior to use in experimental protocols, it was separated from non-volatile contaminants by acidifying the NaH14CO3 in a closed vessel containing a well with 50% molar excess NaOH. The yield of acid-volatile H14CO3 collected from the well was > 90%.

Calculations and statistics

Data are reported in text and figures by determining the d.p.m. per mg protein in a particular metabolite and dividing by the specific activity of medium H14CO3. Therefore data for each metabolite are reported as nmol of H14CO3 per mg protein. Values shown in tables and figures are means ± standard errors of the means. Statistical significance was judged by paired t-test's with p < 0.05 considered to be significant. Numbers of independent determinations (n) are provided in the figure and table legends. Data were also evaluated by a non-parametric technique, which is known as both the Mann–Whitney U-test and the Wilcoxon Rank sum test. No changes in evaluations of statistical significance were revealed by this additional test.


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Pyruvate carboxylase levels are similar in retina and whole brain

Patel demonstrated that pyruvate carboxylase is the major CO2 fixing enzyme in the adult rodent brain (Patel 1989). To judge whether this might also be true in the retina, the activity of pyruvate carboxylase was assayed under optimal conditions in sonicated retinal extracts and found to be similar to that in sonicated whole brain extracts. These activities in the neocortex and rat retina sonicates were 2.9 ± 0.1 nmol H14CO3 fixed/min/mg protein and 3.4 ± 0.4 nmol H14CO3 fixed/min/mg protein, respectively, suggesting that the total activity of pyruvate carboxylase is just as high in the retina as in the whole brain.

Rates of both 14CO2 fixation and 14C-glutamate appearance are a function of pyruvate levels in the retina

In preliminary experiments, 14CO2 fixation, levels of ATP and creatine phosphate were examined as a function of time (10, 20, 40 and 60 min) in retinas incubated with 0.2 mm pyruvate. Creatine phosphate and ATP levels were 16 ± 1.1 nmol/mg protein and 20 ± 0.2 nmol/mg protein, respectively, and remained constant for 60 min. These results indicate that oxygenation and pH control in the retinal preparation are adequate for incubation periods of 1 h. Total CO2 fixation (5.50 ± 0.14 nmol/mg protein) peaked between 20 and 40 min but was already 76% of maximal (4.2 ± 0.1 nmol/mg protein) at 10 min. Because total CO2 fixation did not change or decreased slightly (10–15%) after 20 min, the kinetic pattern indicates that the rate of CO2 fixation was constant, but that by 20 min continuous breakdown of non-volatile 14C products to 14CO2 + H2O produced a near steady-state level of glutamate and glutamine specific radioactivity. Thus, the total amount of 14C-glutamate and glutamine achieved at steady state is influenced not only by the rate of synthesis of 14C-glutamate but also by its rate of breakdown.

To assess the control of pyruvate carboxylation over glutamate synthesis, retinas were incubated for 20 min with either 0.2 or 5 mm added pyruvate and intracellular and extracellular metabolites were analyzed. The results are summarized in Table 1 and Fig. 3. Raising the pyruvate concentration resulted in an 80% increase in the total 14CO2 fixed from 5.27 ± 0.45 to 9.50 ± 0.74 nmol H14CO3 per mg protein (p < 0.002). Almost all of the 14C-glutamate was intracellular, whereas almost all of the 14C-lactate was in the extracellular compartment. At least 90% of the 14C-aspartate and all 14C-labeled TCA cycle intermediates were intracellular. Figure 3 shows the sum of intracellular plus extracellular 14C-glutamate plus 14C-glutamine (Fig. 3a) and 14C-lactate (14C-lactate + 14C-pyruvate, see Materials and methods) (Fig. 3b). Increasing unlabeled pyruvate increased the sum of the intracellular plus extracellular 14C-glutamate plus 14C-glutamine by 70% (Fig. 3a), whereas H14CO3 incorporation into pyruvate plus lactate increased by 107% with citric acid cycle intermediates increasing by 126%. The 14C cycled into pyruvate and lactate represents glial recycling through the pyruvate/malate cycle enzymes (pyruvate carboxylase, malic enzyme and phosphoenolpyruvate carboxykinase). We conclude that the incorporation of 14CO2 into pyruvate (pyruvate cycling) takes place in the glia because decarboxylation of malate or oxaloacetate in the neurons could not have produced 14C-lactate or 14C pyruvate if the source of the 14C was CO2. As shown in Fig. 2, the neuronal glutamate is labeled by H14CO3 only in the carboxyl group adjacent to the carbonyl (1-C). When the glutamate is transaminated to α-ketoglutarate and subsequently oxidized, the labeled carbon is converted to 14CO2 and all subsequent citric acid cycle intermediates: succinate, fumarate, malate, and oxaloacetate are unlabeled and therefore could not produce 14C-lactate.

Table 1.  The influence of medium pyruvate concentration on incorporation of H14CO3 into retinal metabolites
Pyruvate concentrationCompartmentGln*Glu*Asp*Lact*TCA cycle*Total
  1. The experiment is the same as that shown in Fig. 3. The values shown (nmol H14CO3/mg protein) were obtained by dividing the d.p.m. in each metabolite identified chromatographically by the medium H14CO3 specific radioactivity (d.p.m./nmol). Values shown are mean ± SEM, n = 4. ap < 0.03 bp < 0.002. *nmol H14CO3 per mg protein.

0.2 mmIntracellular0.59 ± 0.051.47 ± 0.290.85 ± 0.060.16 ± 0.061.02 ± 0.264.08 ± 0.41
Extracellular0.61 ± 0.080.04 ± 0.010.08 ± 0.010.55 ± 0.0301.19 ± 0.13
Sum1.20 ± 0.091.51 ± 0.290.93 ± 0.080.71 ± 0.061.02 ± 0.265.27 ± 0.45
5.0 mmIntracellular0.91 ± 0.132.67 ± 0.29a1.08 ± 0.120.29 ± 0.042.31 ± 0.13b7.26 ± 0.61b
Extracellular0.96 ± 0.140.06 ± 0.010.08 ± 0.011.18 ± 0.11b02.24 ± 0.21b
Sum1.87 ± 0.23a2.73 ± 0.28a1.16 ± 0.121.47 ± 0.13b2.31 ± 0.139.50 ± 0.74b

Figure 3. Effect of pyruvate concentration on incorporation of H14CO3 into (a) glutamate plus glutamine and (b) pyruvate plus lactate by ex vivo rat retinas. Hemi-retinas were incubated at 37°C with 25 mm H14CO3 for 20 min as described in Materials and methods. Values shown are mean ± SEM (n = 4, *p < 0.02 and **p < 0.001).

Download figure to PowerPoint


Figure 2. Diagram of the labeling of metabolites by the pyruvate carboxylase (PC) reaction in the de novo synthesis of glutamate. The position of the 14C-label from H14CO3 is shown by the solid circle. Additional abbreviations: Ala, alanine; fum, fumarate; Gln, glutamine; Glu, glutamate; KG, α-ketoglutarate; mal, malate; OAA, oxaloacetate; pyr, pyruvate; PDH, pyruvate dehydrogenase; suc, succinate.

Download figure to PowerPoint

These results demonstrate for the first time that the citric acid intermediates malate and oxaloacetate are decarboxylated by glial enzymes in intact neural tissue. Active pyruvate cycling has been shown previously in isolated cultured astrocytes by many workers (Yu et al. 1983; Sonnewald et al. 1993, 1996; Hutson et al. 1998). The results of Fig. 3 also show that the increase in flux through pyruvate carboxylase induces a larger change in pyruvate recycling than in de novo synthesis of retinal glutamate, suggesting that incorporation of 14CO2 into glutamate can be slower than into citric acid cycle intermediates, and into pyruvate and lactate.

A large fraction of total glutamine synthesis is derived from pyruvate carboxylation

The importance of pyruvate carboxylation as a source for replenishing the glutamate released from retinal nerve endings during neuronal transmission was assessed by determining the specific radioactivity of glutamine in retinal tissue incubated with medium containing H14CO3. A fixed proportion of glutamine synthesis in the retina derives from unlabeled glutamate released by neurons and taken up by Müller glia. The remainder comes from glial α-ketoglutarate synthesized anaplerotically from pyruvate and 14CO2. Therefore, the fraction of total glutamine derived from H14CO3 provides an estimate of the proportion of glutamine from pyruvate carboxylation, i.e. de novo synthesis.

As shown in Fig. 2, the product of pyruvate carboxylase in the glia is [1,4-14C]-oxaloacetate. Both the 1 and 4 positions of oxaloacetate are labeled because the 14C initially incorporated into the 4-C position of oxaloacetate is randomized between the 1 and 4 positions by equilibration of oxalocetate with symmetrical fumarate (Sterniczuk et al. 1991; Hassel and Brathe, 2000a,b). The labeled oxaloacetate condenses with pyruvate to form citrate, which is then isomerized to isocitrate. Next [1,4-14C]isocitrate is oxidized and decarboxylated to [1-14C]α-ketoglutarate. Therefore the α-ketoglutarate synthesized from 14CO2 and pyruvate has approximately half of the specific radioactivity of the original substrate H14CO3, depending on the degree of randomization of the label in oxaloacetate. This is also true of the glutamate and glutamine synthesized from the [1-14C]α-ketoglutarate.

The specific radioactivity of glutamine was measured in the retinal tissue and medium extracts incubated with 0.2 mm pyruvate and H14CO3 for 10, 20, 40 and 60 min. Although approximately half of the 14C-glutamine is found in the medium, a much larger fraction of the total glutamine mass was found in the medium. We found that much of the glutamine present in the in vivo retina leaks out of the retina during the 3-min incubation prior to addition of H14CO3 (data not shown). This conclusion is also supported by the recent study by Winkler et al. (1999). Specific radioactivity of the intraretinal pool of glutamine was constant between 10 and 60 min. After 20 min of incubation the mass amount of glutamine in the tissue was 4.50 ± 0.44 nmol/mg protein (n = 4), and its specific radioactivity was 1970 ± 115 d.p.m./nmol or 15.8 ± 0. 9% of the bicarbonate specific radioactivity. By taking into account the randomization of the label in the 1-C and 4-C positions of oxaloacetate, which results in a loss of approximately 50% of the 14C-label in the TCA cycle, we calculated that about 32% of the glutamine was synthesized from H14CO3 via pyruvate carboxylase.

The specific radioactivity of retinal glutamate (269 ± 8 d.p.m./nmol) was lower than that of glutamine (1970 d.p.m./nmol), but the mass amount of glutamate was higher (66.0 ± 1.0 nmol/mg protein) and constant as a function of time. Thus, glutamate specific radioactivity was only 14% of the specific radioactivity of glutamine. This implies that formation of 14C-glutamate from 14C-labeled glutamine may be about 14% of the rate of formation of unlabeled glutamate from the neuronal citric acid cycle.

To calculate absolute fluxes rather than percentages, the rate of turnover of the neuronal glutamate is needed. The H14CO3 labels only the 1-C of glutamate and glutamine (see Fig. 2). Therefore, when glutamate equilibrates with α-ketoglutarate and is oxidized in the citric acid cycle, the 14C in the 1-C position is irreversibly lost in the first turn of the cycle. Therefore, a knowledge of the first-order rate constant of glutamate turnover (kA) and the steady-state fraction of glutamine derived from pyruvate carboxylation allows one to calculate flux through the glutamine cycle. The details of the calculation are as follows:

First-order rate constant of glutamate turnover (per min) = Ka

Measured mass of glutamate (nmol per mg protein) = A

Rate of turnover of the neuronal glutamate pool (nmol per min) = Ka (A)

First order rate constant of glutaminase (per min) = kB

Measured mass of glutamine (nmol per mg protein) = B

nmol of glutamate derived from H14CO3 (nmol H14CO3in glutamate/mg protein)=A*

nmol of glutamine derived from H14CO3 (nmol H14CO3 −in glutamine/mg protein) = B*

The fraction of glutamine derived from CO2 fixation = B*/B. [See above and Fig. 1 for why glutamate and glutamine d.p.m./mg protein is multiplied by 2 and divided by specific radioactivity of H14CO3 to obtain A* and B*.]


kA(A) = nmol of glutamate (mass) exiting the glutamate pool per min

kA(A*) = nmol of glutamate derived from H14CO3 exitingglutamate pool per min

kB(B*) = rate of entry of H14CO3 into glutamate pool (via14C − glutamine and neuronal glutaminase)

At isotopic steady state:

kA(A*) = kB(B*)

kB(B) = kB(B*) · B/B* = the rate of the glutamate/ glutamine cycle.

Subsequent experiments were carried out to estimate the first-order rate constant for turnover of the retinal neuronal glutamate pool (kA). The pool was labeled for 20 min by incubating retinas with H14CO3 and 0.2 mm pyruvate, then retinas were transferred to fresh buffer containing unlabeled HCO3. The decline in retina glutamate specific radioactivity versus time shown in Figs 4(a) and (b) shows the first-order decline of 14C-label in the retinal glutamate pool. The mass amount of glutamate was constant at 66 ± 3 nmol/mg protein. The first-order rate constant of glutamate turnover (kA) calculated from the data shown in Fig. 4 was 0.068 per min. Multiplying 2 × 14C-labeled glutamate (A*) at t = 0 (1.8 × 2 nmol/mg protein), the rate of loss of 14C-glutamate from the glutamate pool was 0.24 nmol/min. Because at t = 0 the radioactivity of glutamate and glutamine were at steady state, the rate of 14C-glutamine conversion to glutamate must also be 0.24 nmol/min. B* was 0.81 nmol/mg protein at t = 0. Because kA (A*) = kB (B*) the value of kB was 0.15. Since 32% of the glutamine is derived from H14CO3 while 68% enters the glutamine pool from unlabeled neuronal glutamate, the total rate of conversion of intraretinal glutamine to glutamate kB(B*)(B/B*) is 0.68 nmol/min/mg protein. Therefore, at steady state this is the approximate rate of the glutamate/glutamine cycle. Compared with total glutamate turnover (0.068 × 66 nmol/mg protein = 4.5 nmol/min/mg protein), the glutamate/glutamine cycle accounts for 15% of the glutamate turnover. The results are summarized in Table 2.


Figure 4. Determination of the first-order rate constant for glutamate turnover in the ex vivo retina. Half-retinas were incubated for 20 min with 25 mm H14CO3 as described in Fig. 3. Then the retinas were transferred to buffer containing unlabeled HCO3. Incubations were stopped at the times shown and the H14CO3 incorporated into retinal glutamate (Glu) determined as described under Materials and methods. (a) nmol of H14CO3 in glutamate plotted as a function of time; (b) natural log (ln) of the nmol of H14CO3 in glutamate plotted as function of time.

Download figure to PowerPoint

Table 2.  Calculated metabolic fluxes for glutamate metabolism in brain and retina
 EquationBrain nmol/min/ mg proteinRetina nmol/min/ mg protein
  1. Definition of kA, kB, A, A*, B, and B* are in the text. Metabolic fluxes are calculated on the basis that kA (retina) = 0.068 min−1 (Fig. 4) and kA brain = 0.09 min−1 (Sibson et al. 1997). Values for A and A* (brain) are shown in Table 3 and measured values of retinal A (66 nmol/mg protein) and A* (2 × 1.8 nmol/mg protein) are in the text. It is assumed that measurements were made at isotopic steady state where kAA* = kBB*.

Turnover neuronal glutamate poolkA(A)10.54.5
Synthesis of neuronal glutamate from H14CO3kA(A)*1.420.24
Rate of entry of H14CO3 into glutamate pool (via 14C glutamine)kB (B*)1.420.24
Glutamate/glutamine cycle flux(kBB*)/(B/B*)6.170.68
Fraction of total glutamine derived from H14CO3B*/B0.230.32

Synthesis of retinal glutamate and glutamine is partially dependent on BCAA transamination

Our recent studies of glutamate synthesis in cultured astrocytes suggested that the conversion of α-ketoglutarate to glutamate could act as a rate-controlling step in anaplerotic synthesis of glutamate (Gamberino et al. 1997). Several groups (Yudkoff 1997; Kanamori et al. 1998) including our own (Gamberino et al. 1997) have indicated that the α-amino group of glutamate cannot be supplied as ammonia and must therefore be supplied by transamination of α-ketoglutarate with another amino acid. BCAA are likely nitrogen contributors because they can cross the blood–brain and blood–retinal barriers efficiently. Hutson et al. (1998) have proposed that a nitrogen shuttle between neurons and glia that involves BCAA and two BCAT isoenzymes, one in neuronal cytosol (BCATc) and one in glial mitochondria (BCATm), can provide the needed nitrogen. According to this hypothetical shuttle, which is a modification of the BCAA shuttle proposed by Yudkoff and coworkers (Yudkoff et al. 1996; Yudkoff 1997) and Bixel et al. (1997), neuronal BCATc facilitates glial BCATm in conversion of α-ketoglutarate to glutamate by regenerating BCAA in the neurons (Fig. 1).

As a first step towards investigating the role of the BCAT isoenzymes in de novo glutamate synthesis in the retina, immunoblotting was used to determine whether the BCAT isoenzymes are expressed in rat retina. As shown in Fig. 5, both BCATc and BCATm are expressed in the retina at levels that are comparable to those found in whole brain homogenates. Therefore, freshly dissected rat retina is an appropriate intact neural system in which to test whether or not BCAA and BCAT are involved in nitrogen shuttling between neurons and glia.


Figure 5. Expression of branched-chain aminotransferase (BCAT) isoenzymes in rat retina. Tissue extraction, SDS-PAGE and immunoblotting were performed as described in Materials and methods. Immunoaffinity purified rabbit rat cytosolic branched-chain aminotransferase (BCATc) peptide antibody (1 : 1000) and rabbit immunoaffinity purified human mitochondrial branched-chain aminotransferase (BCATm) antibody (1 : 1000) were used.

Download figure to PowerPoint

The leucine analog gabapentin is a neuroactive drug that inhibits BCATc but not BCATm (Hutson et al. 1998). We confirmed that gabapentin inhibits the transamination of leucine in freshly dissected rat retinas (data not shown). We measured the effect of gabapentin and BCAA on retina synthesis of 14C-glutamate plus 14C-glutamine from H14CO3. As shown in Fig. 6, addition of all three BCAA (200 µm each) alone had no effect on 14C-glutamate plus 14C-retina glutamine synthesis. However, addition of gabapentin (1.0 mm) reduced the formation of 14C-glutamate plus 14C-glutamine to 71% of control values (2.26 ± 0.05 versus 3.18 ± 0.12 nmol/mg protein, p < 0.001). The addition of the three BCAA to retinas incubated with gabapentin (BCAA/gabapentin ratio of 0.6) clearly antagonized the inhibitory effect of gabapentin on 14C-glutamate plus 14C-glutamine synthesis (89% of control without gabapentin). There was no effect of gabapentin on either ATP or creatine phosphate concentrations, indicating that gabapentin did not impair retinal bioenergetic processes.


Figure 6. The effect of gabapentin (GP) and branched-chain amino acids (BCAA) on incorporation of H14CO3 into glutamate plus glutamine (14C-Glu + 14C-Gln) by ex vivo rat retinas. Half-retinas were incubated at 37°C under standard conditions with 25 mm H14CO3 and 0.2 mm pyruvate. When added, GP was 1 mm and BCAA – leucine, isoleucine and valine – were 0.2 mm each. Values shown are means ± SEM (n = 3–8, *p < 0.001).

Download figure to PowerPoint

Further experiments were carried out to investigate dose–response properties of gabapentin. As shown in Fig. 7, the inhibition of de novo glutamate synthesis by gabapentin was nearly maximal at 0.2 mm drug, concentrations that are reached in vivo in the rat brain (Welty et al. 1995). The tendency of gabapentin to increase citric acid cycle intermediates and lactate is also consistent with the hypothesis. Gabapentin slows the conversion of α-ketoglutarate to glutamate, probably by decreasing the supply of leucine to the Müller cells (see Fig. 1).


Figure 7. Influence of gabapentin concentration on incorporation of H14CO3 into glutamate and glutamine (Glu + C-Gln) the sum of lactate (Lac) and citric acid cycle metabolites (TCA) in ex vivo rat retinas. Conditions were the same as in Fig. 6 except for the concentration of gabapentin. Values shown are means ± SEM (n = 6–7, *p < 0.001).

Download figure to PowerPoint

De novo synthesis of glutamate and glutamine in whole brain is similar to that observed in the retina

In order to estimate the likelihood that studies of retinal glutamate metabolism might be relevant to glutamate metabolism in the whole brain, incorporation of H14CO3 into metabolites of the brains of intact, awake rats infused for 1 h with H14CO3 was measured. Specific radioactivities of H14CO3 in the blood were measured at 15-min intervals, and the non-volatile 14C remaining in the plasma were determined. The non-volatile radiolabel increased linearly in serum so that it accounted for approximately 19% of the total counts at the end of 1 h. Ninety-five per cent of the non-volatile counts were in glucose, and only negligible amounts in glutamine and lactate. Their specific activities were so low they would not have influenced the results. Hepatic gluconeogenesis accounts for the conversion of H14CO3 to glucose in the blood. The serum glucose specific radioactivity per three carbon unit remained below 20% of the H14CO3 specific activity. Analysis of brain metabolites indicated that the specific radioactivities of brain and serum glucose were the same. The average specific radioactivity of the brain lactate/pyruvate was also the same per three carbon unit and was 18% of the mean serum H14CO3 specific radioactivity (162.6 ± 8.0 d.p.m./nmol, n = 5 rats). Mitochondrial pyruvate specific radioactivity might have been slightly higher than the average because of pyruvate recycling, but we assumed an average value in our calculations

Because the (1-14C)-labeled lactate and the H14CO3 contribute 14C equally to the anaplerotic synthesis of oxalacetate from 14CO2, we used the sum of the constant specific radioactivity of H14CO3 plus the average lactate specific activity to calculate the nmol of pyruvate carboxylase product in each metabolite. To determine the fraction of each metabolite derived from 14CO2 fixation (pyruvate carboxylase reaction), the total nmol H14CO3 in each metabolite fraction was divided by the mass amount of each metabolite (see Table 3).

Table 3.  Incorporation of the products of pyruvate carboxylase into brain metabolites
 14C-labeled metabolite
ProductH14CO3 incorporated(14C-labeled metabolite) (nmol/mg protein)Metabolite mass (nmol/mg protein)Metabolite mass
  1. The values shown under 14C-labeled metabolite were obtained by dividing the d.p.m. in each metabolite identified chromatographically by the mean plasma H14CO3 specific activity (d.p.m./nmol). The mass of each metabolite was determined by enzymatic assay as described in Materials and methods. Note that the fraction of the 14C-label in the glutamine pool is higher than that of glutamate. Therefore, the glutamate is in a large compartment which is partially synthesized by the breakdown of glutamine, i.e. the 14C in glutamate is the product of 14C glutamine. Values shown are means ± SEM (n = 5).

Glutamate7.91 ± 0.22116.7 ± 1.10.068 ± 0.0014
Glutamine7.53 ± 0.2366.32 ± 3.80.114 ± 0.0054
Aspartate2.57 ± 0.1434.74 ± 1.10.074 ± 0.002
α-ketoglutarate0.26 ± 0.10
Citrate1.21 ± 0.12
Unknown neutral metabolite5.51 ± 0.85
Alaninen.s.3.52 ± 0.13 

Assuming steady state and using a published value for neuronal glutamate turnover in whole brain of 0.09 nmol/mg protein per min (Sibson et al. 1998), fluxes for the glutamate/glutamine cycle and entry of the 14C-labeled products of pyruvate carboxylase into brain glutamate were calculated. The results for retina and whole brain are compared in Table 2. As in the retina, in brain, anaplerotic synthesis of glutamate from H14CO3 constitutes a significant fraction of the input into the glutamate/glutamine cycle. Also, glutamate/glutamine cycle flux is faster relative to total glutamate turnover in the intact in vivo brain than in the excised retina (Table 2).


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In this study we have demonstrated that in whole brain ≈20% and in retina ≈30% of the glutamine input into the glutamate/glutamine cycle is provided by de novo synthesis of glutamate from H14CO3 and pyruvate. The rest of the glutamate for glutamine synthesis comes from the nerve cells (glutamate released during neurotransmission). We have also demonstrated that in the retina there are two sites of control of flux through the de novo synthesis pathway: (1) pyruvate carboxylase; and (2) transamination of α-ketoglutarate.

Because of oxidative losses of glutamate from the glia (Müller cells), continued operation of the glutamate/glutamine cycle requires input of newly synthesized glutamate. Glutamine synthetase is an integral part of the glutamate/glutamine cycle (Pow and Robinson 1994) and is required to maintain levels of neuronal glutamate essential for neurotransmission (Pow and Robinson 1994). Inhibition of glutamine synthetase for 90 min depletes, to negligible levels, the neuronal glutamate pool which comprises more than 90% of total retinal glutamate (Ottersen et al. 1992; Pow and Robinson 1994). This finding demonstrates the essential role of glutamine synthetase and the flux through the glutamate/glutamine cycle for neurotransmitter homeostasis. De novo synthesis of glutamate occurs only in glia. Interconversion of α-ketoglutarate to glutamate occurs in neurons, as in most cells, but neurons lack sufficient activity of pyruvate carboxylase to replenish the α-ketoglutarate that exits the citric cycle during formation of glutamate. Instead net synthesis of glutamate in the neurons must be catalyzed by glutaminase using glutamine synthesized in the glia.

Although our results show that increasing the concentration of pyruvate in retinal incubations increases the appearance of 14C-label in non-volatile products, much of the 14CO2 fixed does not go to glutamate and glutamine but instead is recycled to pyruvate/lactate via the pyruvate/malate cycle enzyme reactions. Thus, when retinas are incubated with H14CO3 and 5 mm rather than 0.2 mm pyruvate, the magnitude of the increase in 14C-label incorporated into lactate/pyruvate (107%) is greater than the magnitude of the increase in 14C incorporated into glutamate and glutamine (70%).

There has been controversy about whether this conversion of the citric acid cycle intermediates to pyruvate and lactate occurs in intact neural tissue and, if so, whether the decarboxylating enzymes are active in glia. The pathway is not shown in current text books, nor does it appear in published models of the brain glutamate/glutamine cycle (Sibson et al. 1997). When the question of pyruvate recycling has been addressed, one recent study (Cruz et al. 1998) concludes from carbon 13 isotopomer analysis that the pyruvate recycling occurs principally in neurons, not glia. In another study, the data are inconclusive (Haberg et al. 1998). Determination of the site of this decarboxylation is important, because decarboxylation of malate to pyruvate in glia would permit this process to act as a ‘safety valve’ for the disposal of excess glutamate transported into the glia following synaptic transmission. The data provided in this manuscript show that pyruvate cycling does occur in glia and that it catalyzes a significant pathway for removal of citric acid cycle intermediates from glia.

The flexibility provided by the effective reversal of CO2 fixation by the pyruvate malate cycle, and pyruvate cycling, provides the opportunity for a second site of control in the pathway. We have evidence that this occurs at the step that converts α-ketoglutarate to glutamate. BCATm is the predominant isoenzyme in primary cultures of astroglia (Bixel et al. 1997, in press), and we proposed that the important enzyme catalysing that conversion was glial BCATm and that net flux through that step was dependent on the supply of BCAA (Hutson et al. 1998). The hypothesis that BCAA are important for de novo glutamate synthesis and that in intact brain and retinal tissue BCAA are continuously supplied by a nitrogen shuttle between glia and neurons is illustrated in Fig. 2. In this hypothetical scheme BCATc is expressed in neurons (Bixel et al., in press) and is responsible for regenerating BCAA from BCKA generated in the glia during de novo glutamate synthesis.

The inhibition of retinal de novo glutamate synthesis by the competitive BCATc-specific inhibitor, gabapentin (Hutson et al. 1998), is the first demonstration that inhibition of BCATc can affect this pathway in an intact neural preparation. The ability of gabapentin to slow the conversion of citric acid cycle intermediates to glutamine and glutamate strongly suggests that neuronal synthesis of BCAA by BCATc, i.e. the BCAA shuttle, is necessary for optimal rates of amination of α-ketoglutarate to glutamate in Müller cells (Fig. 2). Competition between BCAA and gabapentin is also consistent with the shuttle and the known kinetic properties of the drug. Thus, the BCAA shuttle may provide a significant fraction of the nitrogen for glutamate synthesis in the retina and perhaps in the whole brain (Kanamori et al. 1998)

Unlike previous reports (Shank et al. 1993; Lapidot and Gopher 1994; Sibson et al. 1997; Gruetter et al. 1998; Griffin et al. 1999), this study provides a direct estimate of the fraction of the glutamate/glutamine cycle that is dependent on de novo synthesis from H14CO3 in the ex vivo retina (0.32) and in whole brain in vivo (0.23) (Table 2). Earlier estimates of this value in rat brain come from isotopomer analysis of NMR spectra following in vivo infusion of 13C-glucose (Shank et al. 1985, 1993; Lapidot and Gopher 1994; Sibson et al. 1997; Gruetter et al. 1998; Griffin et al. 1999). Estimates using these NMR methods have generally been lower than the present one (23%), and range from less than 1% (Sibson et al. 1997) to 5–10% (Lapidot and Gopher 1994; Shen et al. 1999), to 20–25% (Shank et al. 1993; Griffin et al. 1999), depending largely on the initial assumptions of the mathematical models employed. Both our estimate and the values reported in the latter NMR studies (Shank et al. 1993; Griffin et al. 1999) indicate that glial pyruvate carboxylation and the conversion of citric acid cycle intermediates to glutamate and glutamine are quantitatively important for retinal and for whole brain function.

Nevertheless, a recent study by Hassel and Brathe, (2000a,b) has questioned the concept that brain anaplerosis is catalyzed largely by glial pyruvate carboxylase. Conversion of 14CO2 to glutamate was measured in separate glial and neuronal cell cultures after 1 h of incubation with H14CO3. Total CO2 fixation was not reported, but more 14C was found in neuronal cell glutamate than in glial cell glutamate, and the conclusion was drawn that more anaplerosis takes place in neurons than in astrocytes. However, the reported amount of 14CO2 fixed in glutamate after 1 h (0.3 nmol/mg protein in neuronal glutamate and 0.1 nmol/mg protein in glia glutamate) is orders of magnitude smaller than that found by other workers (Kaufman and Driscoll 1992; Gamberino et al. 1997) (> 30–40 nmol/mg protein in 20 min). After a single injection of 1-14C pyruvate or H14CO3 into the striatum of mice, the reported 14C in glutamine was less than that in glutamate (glutamine ∼60% of the glutamate specific radioactivity) (Hassel and Brathe, 2000a,b). Theoretically, if anaplerosis takes place in the glia (where glutamine is obligatorily synthesized), glutamine specific radioactivity should be higher than that of glutamate. It is hard to reconcile the above results with the results in our study or with the earlier findings of Waelsch et al. (1964), who also administered H14CO3 intravenously to sedated cats maintaining constant specific activity and found that glutamine specific radioactivity was greater than that of glutamate. Because the specific radioactivities of glutamate and glutamine reported by Hassel and Brathe (2000a,b) were far lower than those reported here and those reported by Waelsch et al. (1964), the discrepancy may be technical.

Finally, although our results suggest that anaplerotic 14CO2 fixation is of similar importance in whole brain and excised retinas, there is a striking difference between the two preparations in actual flux through the glutamate/glutamine cycle relative to neuronal glutamate turnover. In the brain, glutamate/glutamine cycle flux accounts for more than half of the neuronal glutamate turnover whereas in the ex vivo retina it accounts for only approximately 15% of the neuronal glutamate turnover. Flux through the glutamate/glutamine cycle probably reflects the rate of release of glutamate from nerve endings. Because normal room light suppresses neuronal activity in the retina and neuronal activity is likely to be high in the brain of awake rats, the observed retina versus brain difference in glutamate/glutamine cycle flux supports the hypothesis that the rate of the glutamate/glutamine cycle is dependent on neuronal activity, as suggested recently by Sibson et al. (1998).


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The studies reported here were funded in part by grants #197038 and 1–199–678 from the Juvenile Diabetes Foundation International (KFL and EL), by NIH DK34738 (SMH), and a pilot grant from the Center for Investigative Neuroscience, Wake Forest University School of Medicine (SMH).


  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • Bixel M. G., Hutson S. M. & Hamprecht B. (1997) Cellular distribution of branched-chain amino acid aminotransferase isoenzymes among rat brain glial cells in culture. J. Histochem. Cytochem. 45, 685694.
  • Bixel M. G., Simomura Y., Hutson S. M. & Hamprecht B. (2001) Distribution of key enzymes of branched chain amino acids in glial and neuronal cells in culture. J. Histochem. Cytochem. in press .
  • Cooper A. J. & Plum F. (1987) Biochemistry and physiology of brain ammonia. Physiol. Rev. 67, 440519.
  • Crist G. H., Xu B., LaNoue K. F. & Lang C. H. (1998) Tissue specific effects of in vivo adenosine receptor blockade on glucose uptake in Zucker rats. FASEB J. 12, 13011308.
  • Cruz F., Scott S. R., Barroso I., Satisteban P. & Cerdan S. (1998) Ontogeny and cellular localization of the pyruvate cycling system in rat brain. J. Neurochem. 70, 26132619.
  • Davoodi J., Drown P. M., Bledsoe R. K., Wallin R., Reinhart G. D. & Hutson S. M. (1998) Overexpression and characterization of the human mitochondrial and cytosolic branched-chain aminotransferases. J. Biol. Chem. 273, 49824989.
  • Gamberino W. C., Berkich D. A., Lynch C. J., Xu B. & LaNoue K. F. (1997) Role of pyruvate carboxylase in facilitation of synthesis of glutamate and glutamine in cultured astrocytes. J. Neurochem. 69, 23122325.
  • Griffin J. L., Rae C., Radda G. K. & Matthews P. M. (1999) Delayed labelling of brain glutamate after an intra-arterial [13C]glucose bolus: evidence for aerobic metabolism of guinea pig brain glycogen store. Biochim. Biophys. Acta 1450, 297307.
  • Gruetter R., Seaquist E. R., Kim S. & Ugurbil K. (1998) Localized in vivo 13C-NMR of glutamate metabolism in the human brain: initial results at 4 tesla. Dev. Neurosci. 20, 380388.
  • Haberg A., Qu H., Bakken I. J., Sande L. M., White L. R., Haraldseth O., Unsgaard G., Assley J. & Sonnewald U. (1998) In vivo and ex vivo13C-NMR spectroscopy studies of pyruvate recycling in brain. Dev. Neurosci. 20, 389398.
  • Hassel B. & Brathe A. (2000a) Cerebral metabolism of lactate in vivo: evidence for neuronal pyruvate carboxylation. J. Cereb. Blood Flow Metab. 20, 327336.
  • Hassel B. & Brathe A. (2000b) Neuronal pyruvate carboxylation supports formation of transmitter glutamate. J. Neurosci. 20, 13421347.
  • Hutson S. M., Bledsoe R. K., Hall T. R. & Dawson P. A. (1995) Cloning and expression of the mammalian cytosolic branched chain aminotransferase isoenzyme. J. Biol. Chem. 270, 3034430352.
  • Hutson S. M., Berkich D. A., Drown P., Xu B. & LaNoue K. F. (1998) Role of branched-chain aminotransferase isoenzymes and gabapentin in neurotransmitter metabolism. J. Neurochem. 71, 863874.
  • Kanamori K., Ross B. D. & Kondrat R. W. (1998) Rate of glutamate synthesis from leucine in rat brain measured in vivo by 15N-NMR. J. Neurochem. 70, 13041315.
  • Kaufman E. E. & Driscoll B. F. (1992) Carbon dioxide fixation in neuronal and astroglial cells in culture. J. Neurochem. 58, 258262.
  • Laemmli U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680685.
  • Lapidot A. & Gopher A. (1994) Cerebral metabolic compartmentation. Estimation of glucose flux via pyruvate carboxylase/pyruvate dehydrogenase by 13C NMR isotopomer analysis of D-[U-13C]glucose metabolites. J. Biol. Chem. 269, 2719827208.
  • Ottersen O. P., Zhang N. & Walberg F. (1992) Metabolic compartmentation of glutamate and glutamine: morphological evidence obtained by quantitative immunocytochemistry in rat cerebellum. Neuroscience 46, 519534.
  • Patel M. S. (1989) CO2-fixing enzymes, in Neuromethods: Carbohydrate and Energy Metabolism ( BoultonA. A. and BakerG. D., eds), pp. 309340. The Humana Press, Inc., Clifton, NJ.
  • Pow D. V. & Robinson S. R. (1994) Glutamate in some retinal neurons is derived solely from glia. Neuroscience 60, 355366.
  • Shank R. P. & Aprison M. H. (1981) Present status and significance of the glutamine cycle in neural tissues. Life Sci. 28, 837842.
  • Shank R. P., Bennett G. S., Freytag S. O. & Campbell G. L. (1985) Pyruvate carboxylase: an astrocyte-specific enzyme implicated in the replenishment of amino acid neurotransmitter pools. Brain Res. 329, 364367.
  • Shank R. P., Leo G. C. & Zielke H. R. (1993) Cerebral metabolic compartmentation as revealed by nuclear magnetic resonance analysis of D-[1–13C]glucose metabolism. J. Neurochem. 61, 315323.
  • Shen J., Petersen K. F., Behar K. L., Brown P., Nixon T. W., Mason G. F., Petroff O. A., Shulman G. I., Shulman R. G. & Rothman D. L. (1999) Determination of the rate of the glutamate/glutamine cycle in the human brain by in vivo 13C NMR. Proc. Natl Acad. Sci. USA 96, 82358240.DOI: 10.1073/pnas.96.14.8235
  • Sibson N. R., Dhankhar A., Mason G. F., Behar K. L., Rothman D. L. & Shulman R. G. (1997) In vivo 13C NMR measurements of cerebral glutamine synthesis as evidence for glutamate–glutamine cycling. Proc. Natl Acad. Sci. USA 94, 26992704.DOI: 10.1073/pnas.94.6.2699
  • Sibson N. R., Dhankhar A., Mason G. F., Rothman D. L., Behar K. L. & Shulman R. G. (1998) Stoichiometric coupling of brain glucose metabolism and glutamatergic neuronal activity. Proc. Natl Acad. Sci. USA 95, 316321.DOI: 10.1073/pnas.95.1.316
  • Sonnewald U., Westergaard N., Petersen S. B., Unsgard G. & Shousboe A. (1993) Metabolism of [U-13C]glutamate in astrocytes studied by 13C NMR spectroscopy: incorporation of more label into lactate than into glutamine demonstrates the importance of the tricarboxylic acid cycle. J. Neurochem. 61, 11791182.
  • Sonnewald U., Westergaard N., Jones P., Taylor A., Bachelard H. S. & Schousboe A. (1996) Metabolism of [U-13C5] glutamine in cultured astrocytes studied by NMR spectroscopy: first evidence of astrocytic pyruvate recycling. J. Neurochem. 67, 25662572.
  • Stanton C., Taylor R. & Wallin R. (1991) Processing of prothrombin in the secretory pathway. Biochem. J. 277, 5965.
  • Sterniczuk A., Hreniuk S., Scaduto R. C. Jr & LaNoue K. F. (1991) Effect of phenylephrine on pyruvate dehydrogenase in fasting rat livers. Eur. J. Biochem. 196, 151157.
  • Waelsch H., Berl S., Rossi C. A., Clarke D. D. & Purpura D. P. (1964) Quantitative aspects of CO2 fixation in mammalian brain in vivo. J. Neurochem. 11, 717728.
  • Wallin R., Hall T. R. & Hutson S. M. (1990) Purification of branched chain aminotransferase from rat heart mitochondria. J. Biol. Chem. 265, 60196024.
  • Welty D. F., Schielke G. P. & Rothstein J. D. (1995) Potential treatment of amyotrophic lateral sclerosis with gabapentin: a hypothesis. Ann. Pharmacother. 29, 11641167.
  • Wilbur D. O. & Patel M. S. (1974) Development of mitochondrial pyruvate metabolism in rat brain. J. Neurochem. 22, 709715.
  • Williamson J. R. & Corkey B. E. (1969) Assays of intermediates of the citric acid cycle and related compounds by fluorometric enzyme methods, in Methods of Enzymology ( LowensteinJ. M., ed.), pp. 435513. Academic Press, New York.
  • Winkler B. S., Kapousta-Bruneau N., Arnold M. J. & Green D. G. (1999) Effects of inhibiting glutamine synthetase and blocking glutamate uptake on B-wave generation in the isolated rat retina. Visual Neurosci. 16, 345353.
  • Yu A. C., Drejer J., Hertz L. & Schousboe A. (1983) Pyruvate carboxylase activity in primary cultures of astrocytes and neurons. J. Neurochem. 41, 14841487.
  • Yudkoff M. (1997) Brain metabolism of branched-chain amino acids. Glia 21, 9298.DOI: 10.1002/(sici)1098-1136(199709)21:1<92::aid-glia10>;2-w
  • Yudkoff M., Daikhin Y., Grunstein L., Nissim I., Stern J., Pleasure D. & Nissim I. (1996) Astrocyte leucine metabolism: significance of branched-chain amino acid transamination. J. Neurochem. 66, 378385.