Steady-state in vivo glutamate dehydrogenase activity in rat brain measured by 15N NMR.

The in vivo activity of glutamate dehydrogenase (GDH) in the direction of reductive amination was measured in rat brain at steady-state concentrations of brain ammonia and glutamate after intravenous infusion of the substrate 15NH4+. The in vivo rate was determined from the steady-state fractional 15N enrichment of brain ammonia, measured by selective observation of 15NH4+ protons in brain extract by 1H-15N heteronuclear multiple-quantum coherence transfer NMR, and the rate of increase of brain [15N]glutamate and [2-15N]glutamine measured by 15N NMR. The in vivo GDH activity was 0.76-1.17 mumol/h/g, and 1.1-1.2 mumol/h/g at 1.0 +/- 0.17 mumol/g. Comparison of the observed in vivo GDH activity with the in vivo rates of glutamine synthesis and of phosphate-activated glutaminase suggests that, under mild hyperammonemia, GDH-catalyzed de novo synthesis can provide a minimum of 19% of the glutamate pool that is recycled from neurons to astrocytes through the glutamate-glutamine cycle.

The in vivo activity of glutamate dehydrogenase (GDH) in the direction of reductive amination was measured in rat brain at steady-state concentrations of brain ammonia and glutamate after intravenous infusion of the substrate 15 NH 4 ؉ . The in vivo rate was determined from the steady-state fractional 15 N enrichment of brain ammonia, measured by selective observation of 15  Glutamate dehydrogenase (GDH) 1 catalyzes the reversible reaction below. NH 3 ϩ 2-oxoglutarate ϩ NAD͑P͒H º glutamate ϩ NAD͑P͒ ϩ REACTION 1 This mitochondrial enzyme is present at a high level in rat brain, with an in vitro activity of 900 mol/h/g (1). GDH is believed to contribute to the synthesis of the metabolic and neurotransmitter pools of glutamate. While glutamine is also an important precursor of the neurotransmitter glutamate (2)(3)(4), there is evidence to suggest that the glutamate-glutamine cycle is not operating in a stoichiometric manner (5), and some de novo synthesis of glutamate from glucose is required to maintain the neurotransmitter pool (5,6). GDH is a likely candidate for this role, since the equilibrium of the reaction favors the formation of glutamate.
While GDH activities measured in cultured astrocytes and synaptosomal preparations have provided useful information (7)(8)(9), it is also important to measure the activity in intact brain, because the distribution of this enzyme according to cell type is controversial. Biochemical and histochemical studies show higher GDH levels in neurons (10 -12), while immunocytochemical studies show highest GDH immunoreactivity in astrocytes (13) with only 15% reactivity in neurons (14). Moreover, a recent review of glutamate metabolism in mammalian brain (6) suggests that the rapid labeling of brain glutamate after intravenous injection of 14 C-labeled glucose (15,16) and the slow labeling of brain glutamate after short term 15 NH 4 ϩ or 13 NH 4 ϩ infusion (17,18) give conflicting pictures of the role of GDH in glutamate synthesis. The classical study by Berl et al. (17) on the 15 N labeling of brain glutamate and glutamine in 15 NH 4 ϩ -infused cat led to the important concept of compartmentation of brain glutamate metabolism. A subsequent 13 N study in normal rat brain (18) confirmed the labeling pattern. However, these short term (10 -25 min) labeling experiments did not yield a rate for GDH-catalyzed glutamate synthesis. To our knowledge, measurement of steady-state in vivo GDH activity in mammalian brain has not been reported. We report here measurement of in vivo GDH activity in the direction of reductive amination at steady-state concentrations of brain ammonia and glutamate in rats given 15 NH 4 ϩ infusion for 1-6 h, using 15 N NMR in combination with biochemical techniques. 15 N NMR was previously used for noninvasive monitoring of 15 N-metabolites to determine in vivo glutamine synthetase (GS) (19) and phosphate-activated glutaminase (20) activities in rat brain. In the present study, after 15 N enrichment in vivo, brain [ 15 N]glutamate and [2-15 N]glutamine were quantified in vitro for better spectral resolution. The results are discussed in relation to the role of GDH in glutamate replenishment. An explanation is offered for the apparent discrepancy between results obtained with labeled glucose and ammonia on the role of GDH in glutamate synthesis.

Animal Preparation and Ammonia Infusion-Male
Wistar rats (250 -300 g) were anesthetized by the intraperitoneal injection of sodium pentobarbital (Nembutal; 40 mg/kg body weight), and prepared for ammonia infusion through the femoral vein (21). Two infusion protocols were used to achieve different steady-state brain ammonia concentrations for measurement of in vivo GDH activities. One group of rats (Group I) was given infusion of 15 (19,20). 15 1 The abbreviations used are: GDH, glutamate dehydrogenase; GS, glutamine synthetase; GABA, ␥-aminobutyric acid; Glx, glutamate ϩ glutamine.
For in vitro study, the rat was sacrificed at the time indicated after 15 NH 4 ϩ infusion or 14 NH 4 ϩ chase. With the anesthetized rat still breathing, the cranium was opened by scissor dissection and the brain was removed in toto and rapidly frozen in liquid nitrogen for preparation of a perchloric acid extract, as described previously (21). This procedure takes Ͻ30 s and is as fast as similar dissection-freezing methods (22,23) that have been shown to yield brain ammonia, glutamate, and glutamine concentrations that are in good agreement with those obtained by freeze-blowing (24). [ 15 N]glutamate, [2-15 N]glutamine, and [5-15 N]glutamine in the brain extract were identified by 15 N NMR and quantified from the observed peak intensity (measured as integrated peak area) by comparison with those of standards, as described previously (19,20). The 15 NH 4 ϩ concentration in brain extract was measured by selective observation of the 15 NH 4 ϩ protons by 1 H-15 N heteronuclear multiple quantum coherence NMR at 200 MHz for 1 H as described previously (20,25), with the following modification. A sample volume of 3.7 ml was used to accommodate the entire brain extract (ϳ4 ml/brain). To quantify brain 15 NH 4 ϩ , a new standard curve was constructed by measuring the 1 H peak intensities for 300, 500, and 700 nmol of 15 NH 4 Cl dissolved in 3.7 ml of unlabeled brain extract at pH 3.3. For quantification of brain 15 NH 4 ϩ and [2-15 N]glutamate ϩ glutamine, the brain extract prepared from a single rat was used in each experiment, for the number of rats indicated in Table I. For resolution of [ 15 N]glutamate and [2][3][4][5][6][7][8][9][10][11][12][13][14][15] N]glutamine at pH 9.1, the brain extract for each NMR experiment was prepared from a single rat (Group II), or pooled from 2 or 3 rats that were infused at the same rate for the same duration, to increase sensitivity (Group I). In the latter case, the number of separate experiments is shown in parentheses in Table I. The concentrations of total [ 14 Nϩ 15 N]ammonia, -glutamate, and -glutamine in the brain extracts were measured enzymatically according to published procedures (26 -28). Fig. 1A shows an in vivo 15 N NMR spectrum obtained from the head of an anesthetized rat after 3 h of 15 NH 4 ϩ infusion at the rate of 2.3 mmol/h/kg weight. We have previously shown that the peaks for [5-15 N]glutamine (Ϫ271 ppm) and [2-15 N]glutamate/glutamine (Ϫ342.1 ppm) arise exclusively from the brain (29). Fig.  1B shows an 15 N NMR spectrum of the perchloric acid extract of the brain of a rat after 4.2 h of 15 NH 4 ϩ infusion at the rate of 3.3 mmol/h/kg weight, and Fig. 1C the spectrum of an extract after 2.9 h of infusion at the rate of 2.3 mmol/h/kg. In vivo and in extracts, at pH 7, the [ 15 N]glutamate peak overlaps the [2-15 N]glutamine peak at Ϫ342.1 ppm. The two peaks can be well resolved at pH 9.1, as described previously (20), and will be shown later in this work. However, for measurement of in vivo GDH activity, we need to monitor the increase, not only of brain [ 15 [2-15 N]Glx increased linearly with time in both groups. From the slope of the least-squares line through the plots, the rate of increase was determined to be 0.295 mol/h per g of brain for Group I and 0.50 mol/h/g for Group II. 15 N Enrichment of Brain Ammonia- Table I shows [5-15 N]glutamine (Ϫ271 ppm) and [2-15 N]glutamate/glutamine (Ϫ342.1 ppm) arise from the brain (29). The peak at Ϫ306 ppm is [ 15 N]urea, which was synthesized in the liver and circulating in the blood to enter various tissues in the head including the brain. B-D, a spectrum of the perchloric acid extract of the brain of a rat after 15  arate, the second substrate of the GDH reaction, is reported to be unaffected by acute or chronic hyperammonemia (30,31), and hence can reasonably be assumed to be at steady state under our experimental condition.
Another method of estimating the in vivo GDH activity is to determine the rate of 15 N flux through Glx 2-N relative to that through glutamine 5-N (catalyzed by GS) when 15 N is chased by 14 N at steady state. This flux ratio, combined with the known in vivo GS activity in rat brain (32), permits estimation of the GDH activity. Fig. 1D shows an 15 N spectrum of the brain extract of a rat given 15  N]peak intensity ratio changed substantially between prechase (Fig. 1C) and postchase (Fig. 1D) period. This suggests that the flux of 15 N through 2-N is much slower than that through 5-N. Table I shows the progressive increase in [2][3][4][5][6][7][8][9][10][11][12][13][14][15] N]Glx/[5-15 N]Gln concentration ratio (determined from their observed concentrations in brain extracts), during the chase period for Group I. The concentration ratio changed from 0.38 Ϯ 0.06 (n ϭ 4) before the chase to 1.45 Ϯ 0.33 (n ϭ 3) after the 3.2 h chase. This 3.8-fold increase shows that the GDHcatalyzed flux of 15 N through 2-N was 3.8-fold slower than the GS-catalyzed flux through 5-15 N during the steady-state chase period. The in vivo GS activity, determined from the rate of decrease of [5-15 N]glutamine during the same chase period, was 3.3-4.4 mol/h/g (32). Hence, in vivo GDH activity estimated by this method is 0.86 -1.17 mol/h/g. This rate is in reasonable agreement with the in vivo rate of GDH obtained by the first method for Group I. For Group II, the second method was not attempted because of insufficient numbers of postchase data. The in vivo rates of GDH reaction for each group, determined as described above, are listed in Table I. 15 N Enrichments of Glutamate and Glutamine 2-N- Fig. 1E shows an 15 N spectrum, obtained at pH 9.1, of the brain extract of rats given 15 15 N enrichments of brain glutamate and glutamine 2-N in Groups I and II were calculated and are shown in Table  I. It was previously shown that total brain glutamine reaches steady-state levels of 8.5 Ϯ 0.96 (Group I) and 9.8 Ϯ 0.86 mol/g (Group II) after 3-4 h of ammonia infusion (32). It is interesting that, after 3-4 h of 15 NH 4 ϩ infusion, the 15 N enrichment of brain glutamate is higher than that of glutamine 2-N. DISCUSSION Our results show that in vivo GDH activity in the direction of reductive amination in rat brain is 0.76 -1.17 mol/h/g at steady-state brain ammonia level of 0.87 Ϯ 0.18 mol/g, and 1.1-1.2 mol/h/g at 1.0 Ϯ 0.17 mol/g. The low in vivo activity compared to the reported in vitro activity measured at enzymesaturating concentrations of the substrates, 900 mol/h/g (1), is most probably the result of the low in situ concentrations of ammonia and 2-oxoglutarate (0.23 Ϯ 0.05 mM) (30,31) relative to the K m values of the enzyme, 10 -18 mM for NH 4 ϩ and 0.2-1.5 mM for 2-oxoglutarate (33,34).
Carbon Versus Nitrogen Labeling-It has been suggested that the rapid labeling of brain glutamate after intravenous injection of [ 14 C]glucose (15,16,35) and the slow labeling of brain glutamate after 15 NH 4 ϩ or 13 NH 4 ϩ administration (17, 18) lead to conflicting conclusions on the role of GDH in glutamate synthesis (6). In reality, both carbon and nitrogen labeling experiments are correct, but use of labeled ammonia leads to measurement of GDH activity while use of labeled glucose yields the rate of 2-oxoglutarate-glutamate exchange, as shown by Mason et al. (36,37) using in vivo 13 C NMR. This exchange is catalyzed by transaminases, including aspartate aminotransferase, as well as by GDH. Aspartate aminotransferase is present in the brain at a much higher level than GDH and is near equilibrium (38). Balá zs and Haslam (39) showed that the rapid 14 C labeling of brain glutamate from labeled glucose mainly reflects aspartate aminotransferase-catalyzed isotopic exchange between 2-oxoglutarate and glutamate. After intravenous injection of [ 14  equilibrium is reached, without net glutamate synthesis (39). Labeling with ammonia, on the other hand, permits measurement of the rate of GDH-catalyzed glutamate synthesis from 2-oxoglutarate and ammonia.
Garfinkel (40) combined data from 14 C-glucose labeling (15,35,41) and 15 NH 4 ϩ labeling experiments (17) to calculate the rates of 104 reactions involved in the tricarboxylic acid cycle and amino acid metabolism. The rates were adjusted to provide the best overall fit to the reported specific isotopic enrichments of the brain metabolites, using a two-compartment brain model. The calculated aspartate aminotransferase activity was very high, 27-240 mol/h/g, while GDH activity (reductive amination) was 1.08 mol/h/g for the large compartment and 7.8 mol/h/g for the small compartment. The large compartment was thought to contain 4 -10 times as much of the metabolites as the small compartment. On that assumption, GDH activity for the whole brain is expected to be about 1.7-2.8 mol/h/g, which is only slightly higher than the rate reported here. These considerations, together with the experimental result reported here, strongly suggest that in vivo GDH activity in the direction of reductive amination in rat brain is of the order of 1 mol/h/g of brain.
Net Glutamate Synthesis by GDH-The GDH reaction is reversible and the whole-brain glutamate level is observed to be at steady state. It is therefore important to consider whether (a) the observed GDH-catalyzed [ 15 N]glutamate synthesis is offset by oxidative deamination of [ 14 N]glutamate at an equal rate, resulting in no net GDH-catalyzed glutamate synthesis, or (b) there is net synthesis by GDH but glutamate is utilized by other pathways, such as GS (conversion to glutamine) and glutamate decarboxylase (conversion to GABA). The following considerations suggest that GDH-catalyzed glutamate catabolism is negligible in the brain. The equilibrium of GDH reaction favors glutamate formation, particularly in brain compartments with a low glutamate level such as astrocytes and GABAergic neurons (42). Glutamatergic neurons have a high glutamate level, but the highest concentration occurs in the presynaptic terminal where the transmitter glutamate pool is sequestered in vesicles (43). In neuronal perikaryon, mitochondria do not show a gradient of glutamate-like reactivity over the cytoplasm (43). Neuronal GDH level, on the other hand, is highest in somatic zones and dendritic processes and lower in axon terminals (12). Hence, glutamate concentration at the site of GDH in mitochondria is probably not high enough to overcome the unfavorable equilibrium. Rat brain mitochondria in-cubated with 10 mM glutamate produced no detectable ammonia (44), while synaptosomal preparation incubated with [ 15 N]glutamate produced 15 NH 4 ϩ at the rate of only 0.2 nmol/mg protein in 30 min (9). In contrast, in intact brain, 15 NH 4 ϩ and 13 NH 4 ϩ label GABA and glutathione nitrogens (17,18). These considerations strongly suggest that the GDH-catalyzed [ 15 N]glutamate synthesis reported here reflects net synthesis, which can replenish the metabolic and neurotransmitter pools of glutamate.
Role of GDH in Glutamate Replenishment-The glutamateglutamine cycle is not operating in a stoichiometric manner (5), and some de novo synthesis of glutamate from glucose is needed to maintain the neurotransmitter pool (5,6). The relative contributions of glutamate synthesized de novo by GDH and of glutamine-derived glutamate to the pool can be estimated by comparing the observed in vivo GDH activity, 0.76 -1.17 mol/h/g, with the in vivo rate of glutamine synthesis and utilization, 3.3 Ϯ 0.3 mol/h/g, also measured under identical mildly hyperammonemic condition (32), and of phosphate-activated glutaminase, 1.1 Ϯ 0.2 mol/h/g (20). These in vivo rates show that glutamate is converted to glutamine in astrocytes at the rate of 3.3 Ϯ 0.3 mol/h/g. After migration to neurons, glutamine is reconverted to glutamate by phosphate-activated glutaminase at the rate of 1.1 Ϯ 0.2 mol/h/g. Amidotransferases involved in purine and pyrimidine synthesis also release glutamate from glutamine. The maximum possible rate of glutamine conversion to glutamate is the experimentally measured rate of total glutamine utilization, which, at steady state, is equal to the rate of glutamine synthesis (32). Hence the in vivo rate of glutamine conversion to glutamate is between 1.1 Ϯ 0.2 mol/h/g (phosphate-activated glutaminase) and 3.3 Ϯ 0.3 mol/h/g (total glutamine utilization rate). Comparison with the observed in vivo GDH activity shows that GDH-catalyzed de novo synthesis can provide at least 0.76/(0.76 ϩ 3.3) ϫ 100% ϭ 19% of the glutamate pool that is recycled from neurons to astrocytes through the glutamate-glutamine cycle. Thus, the role of GDH in glutamate replenishment can be significant. The source of carbon for the GDH reaction is likely to be glucose, but the rate of net conversion of 2-oxoglutarate to glutamate cannot be significantly greater than that identified here for GDH. Acknowledgment-We are grateful to Emily L. Kuo for assistance with perchloric acid extraction and enzymatic assay for brain ammonia.  15 N enrichments of brain metabolites and in vivo GDH activities in 15 NH 4 ϩ -infused rats Tissue levels (mol/g) and 15 N enrichments (%) of brain ammonia, glutamate, and glutamine in 15 NH 4 ϩ -infused rats are shown as the mean Ϯ standard deviation for the number of rats shown in parentheses for n Ͼ 3. For n ϭ 2, the range is shown. Also shown are [2-15 N]Glx/ [5-15 N]Gln ratios for rats given 15