Glutamatergic and GABAergic TCA cycle and neurotransmitter cycling fluxes in different regions of mouse brain

Abstract

The ¹³C nuclear magnetic resonance (NMR) studies together with the infusion of ¹³C-labeled substrates in rats and humans have provided important insight into brain energy metabolism. In the present study, we have extended a three-compartment metabolic model in mouse to investigate glutamatergic and GABAergic tricarboxylic acid (TCA) cycle and neurotransmitter cycle fluxes across different regions of the brain. The ¹³C turnover of amino acids from [1,6-¹³C₂]glucose was monitored ex vivo using ¹H-[¹³C]-NMR spectroscopy. The astroglial glutamate pool size, one of the important parameters of the model, was estimated by a short infusion of [2-¹³C]acetate. The ratio V_cyc/V_TCA was calculated from the steady-state acetate experiment. The ¹³C turnover curves of [4-¹³C]/[3-¹³C]glutamate, [4-¹³C]glutamine, [2-¹³C]/[3-¹³C]GABA, and [3-¹³C]aspartate from [1,6-¹³C₂]glucose were analyzed using a three-compartment metabolic model to estimate the rates of the TCA cycle and neurotransmitter cycle associated with glutamatergic and GABAergic neurons. The glutamatergic TCA cycle rate was found to be highest in the cerebral cortex (0.91±0.05 μmol/g per minute) and least in the hippocampal region (0.64±0.07 μmol/g per minute) of the mouse brain. In contrast, the GABAergic TCA cycle flux was found to be highest in the thalamus–hypothalamus (0.28±0.01 μmol/g per minute) and least in the cerebral cortex (0.24±0.02 μmol/g per minute). These findings indicate that the energetics of excitatory and inhibitory function is distinct across the mouse brain.

Introduction

Glutamate (Glu) and gamma-aminobutyric acid (GABA) are the most abundant neurotransmitters in the cerebral cortex and are responsible for the excitatory and inhibitory neurotransmissions in the matured central nervous system. These neurotransmitters are involved in many functions such as motor behavior, cognition, and emotion.^{1, 2} Perturbation in glutamatergic and GABAergic neurotransmission is associated with several neurological and psychiatric disorders.³ In vivo ¹³C nuclear magnetic resonance (NMR) studies in rat and human brain have established that the Glu-glutamine (Gln) cycle accounts for a major fraction (>80%) of Gln synthesis. Experiments conducted in the rat brain under conditions of graded anesthesia have shown that rates of neurotransmitter cycle and neuronal glucose oxidation are stoichiometrically coupled through the entire level of brain activity, thus indicating that neurotransmitter energetics is supported by oxidative glucose metabolism.^{4, 5 13}C NMR studies in the rat brain have indicated that GABA also contributes significantly to neuronal glucose oxidation and neurotransmitter cycling flux.^{4, 6}

Extensive ¹³C NMR studies of cerebral metabolism in rat^{4, 5, 7, 8, 9, 10} and human^{11, 12, 13} brain have given important insights into neurotransmitter energetics. In vivo ¹³C NMR and autoradiography studies conducted in rat brain indicated that metabolism is not uniform across the brain.^{7, 14, 15} Successful completion of the human genome project led to the development of transgenic models of various cerebral disorders in mice. However, cerebral metabolism in mice has not been understood quantitatively. The understanding of regional energetics of glutamatergic and GABAergic neurotransmission in normal mice is indispensible to gain insight into the pathophysiology of various human neurological disorders using transgenic mouse models.

In the present study, we have extended the three-compartment metabolic model⁴ in mice to investigate glutamatergic and GABAergic fluxes in cerebral cortex (Cx), hippocampus (Hip), striatum (Str), thalamus–hypothalamus (THt), and cerebellum (Cb) regions. The ¹³C turnover of amino acids from [1,6-¹³C₂]glucose was measured ex vivo using ¹H-[¹³C]-NMR spectroscopy. The astroglial Glu pool size was estimated using a short-time infusion of [2-¹³C]acetate. The ratio V_cyc/V_TCA for glutamatergic and GABAergic neurons was measured from the steady-state [2-¹³C]acetate measurement. The ¹³C turnover of cerebral amino acids from [1,6-¹³C₂]glucose was analyzed using a three-compartment metabolic model to determine metabolic rates associated with the glutamatergic and GABAergic pathway in different regions of the mouse brain. Our findings indicate that the glucose oxidation and neurotransmitter cycling fluxes of glutamatergic and GABAergic neurons vary across the different regions in the mouse brain.

Materials and Methods

Animal Preparation

All experiments were conducted in accordance with the protocols approved by Institutional Animal Ethics Committee of CCMB. ARRIVE guidelines were followed in the preparation of the manuscript. Thirty-five male C57BL6 (2 months old) mice were fasted for 12 hours before infusion of ¹³C-labeled substrates. Mice were anesthetized with urethane (1.5 g/kg, intraperitoneally). Core body temperature of the animal was measured using a rectal probe and maintained at ∼37°C by placing it supine on a heating pad connected to a temperature-regulated circulating water bath. The tail vein was cannulated for infusion of ¹³C-labeled glucose or acetate. Animal respiration was monitored throughout the experiment by using the BIOPAC device (Santa Barbara, CA, USA) interfaced to a computer.

Infusion of [1,6-¹³C₂]Glucose and [2-¹³C]Acetate

A solution of [1,6-¹³C₂]glucose (99 atom%, Cambridge isotope, Andover, MA, USA) was infused in mice through the tail vein for 7, 15, 30, 60, and 90 minutes (n=5 for each time point) after 45 minutes of urethane anesthesia. The [1,6-¹³C₂]glucose (0.225 mol/L) dissolved in deionized water was delivered as a bolus for the first 15 seconds, followed by an exponentially decreasing infusion rate every 30 seconds for the next 8 minutes, whereupon the infusion rate was constant until in situ brain freezing.¹⁶ The infusion rate of [1,6-¹³C₂]glucose at steady state (⩾8.25 minutes) was 15 μmol/kg per minute. [2-¹³C]acetate (1 mol/L) and unlabeled glucose (0.225 mol/L), dissolved in deionized water and pH adjusted to 7.0, were administered to mice (n=5) by bolus-variable infusion rate as described previously.⁹ The [2-¹³C]acetate infusion rate was 0.2 mmol/kg per minute at the steady state. The infusion was terminated at 90 minutes, which was sufficient to attain the steady-state ¹³C enrichments of Glu_C4, GABA_C2, and Gln_C4. In addition, mice (n=5) were also infused with [2-¹³C]acetate for 3 minutes to estimate the size of astroglial Glu pool. Blood was collected from sinus orbitus using the fine capillary just before the termination of infusion. Plasma was collected after centrifugation of the blood and stored at −80°C for subsequent analysis. At the end of the infusion, mouse brain was frozen in situ using liquid N₂.

Preparation of Brain Extracts for Nuclear Magnetic Resonance Analysis

The mouse brain was dissected in a cryostat (maintained at −20°C) to isolate cerebellum, cerebral cortex, hippocampus, striatum, and thalamus–hypothalamus regions. Metabolites were extracted from frozen tissue as described previously.¹⁷ In brief, the frozen tissue was weighed (Supplementary Table 2S) and powdered with 0.1 N HCl in methanol (1:2 w/v) in a dry ice/ethanol bath. The [2-¹³C]glycine (100 μl; 2 mmol/L) was added as an internal concentration reference. The tissue powder was thoroughly homogenized with ice-cold ethanol (1:6 w/v; 60% ethanol), and the homogenate was clarified by centrifugation at 20,000 g. The tissue extract was passed through the chelex column, and the pH of the extract was adjusted to 7.0. The extract was lyophilized and dissolved in phosphate-buffered (50 mmol/L, pH=7.0) deuterium oxide containing sodium 3-trimethylsilyl[2,2,3,3-D4]-propionate (TSP) (0.5 mmol/L) as a chemical shift reference.

The brain extracts obtained from short-time infusion of [2-¹³C]acetate were passed through an AG 1-X8 anion exchange column to separate Gln and Glu. Gln and neutral molecules were eluted using deionized water, whereas Glu and anionic molecules were eluted using 2 mol/L acetic acid.¹⁷ Both fractions were lyophilized and resuspended in a phosphate-buffered (50 mmol/L, pH=7) deuterium oxide solution containing 0.5 mmol/L TSP for further ¹H-[¹³C]-NMR analysis.

Nuclear Magnetic Resonance Analysis of Brain Extract and Plasma

All NMR measurements were carried out on a 600-MHz (Bruker Biospin, Rheinstetten, Germany) spectrometer. The ¹H-[¹³C]-NMR spectra of tissue extracts were acquired using a pulse sequence that incorporates adiabatic pulses for ¹H and ¹³C frequencies.^{16, 18} Concentrations of metabolites were determined relative to [2-¹³C]glycine, added during tissue extraction as an internal concentration reference. The concentration thus obtained represents weighted average of gray matter, white matter, and cerebrospinal fluid in the given brain region. The ¹³C atom percent enrichment of metabolites at different carbon positions was determined as the ratio of the peak areas in the ¹H-[¹³C]-NMR difference spectrum (2 × ¹³C only) to the nonedited spectrum (¹²C+¹³C). The percent enrichment of metabolites was corrected for ¹³C natural abundance by subtraction of 1.1%.

Blood plasma was mixed with deuterium oxide containing 0.5 mmol/L sodium formate and passed through a centrifugal filter (10-kDa cut off) to remove macromolecules. Plasma samples were analyzed using ¹H NMR spectroscopy to determine the total concentration and ¹³C enrichment of glucose and acetate. Water resonance was suppressed by continuous irradiation during the relaxation delay. The concentration of glucose and acetate was determined using formate as reference. The isotopic ¹³C enrichment of glucose-C1α (5.2 p.p.m.) and acetate-C2 (1.9  p.p.m.) was calculated by dividing the areas of the ¹³C satellites with the total area (¹²C+¹³C).

Determination of Glutamate Pool in Neurons and Astroglia

Astroglial Glu pool size, which is required for the three-compartment metabolic modeling, is not established across different regions of the mouse brain. The pool size of Glu in astroglia was estimated by using a very short-time infusion of [2-¹³C]acetate. The measured Glu labeling from [2-¹³C]acetate is the weighted average of neuronal and astroglial fraction:

where Glu_C4(g) and Glu_C4(n) are the percent ¹³C enrichment of astroglial and neuronal Glu pool, and ‘f ' is the fraction of astroglial Glu. As astroglial Glu is the precursor for Gln and Glu_C4(g)∼Gln_C4 for a short-time infusion of [2-¹³C]acetate, Glu_C4(n)∼0. Hence, Equation (1) is reduced to the following:

Determination of V_cyc/V_TCA from [2-¹³C]Acetate Studies

The ratio of V_cyc/V_TCA for glutamatergic and GABAergic neurons was calculated in different regions of the brain by steady-state labeling of Glu_C4, GABA_C2, and Gln_C4 from [2-¹³C]acetate.⁴ For glutamatergic neurons, the ratio of Glu–Gln cycle to glutamatergic TCA cycle fluxes (V_cyc(Glu-Gln)/V_TCA(Glu)) was calculated using the following equation:

where Glu_C4 and Gln_C4 are the labeling of neuronal [4-¹³C]glutamate and astroglial [4-¹³C]glutamine, respectively, from [2-¹³C]acetate at the steady state. Glutamine and GABA were assumed to be entirely localized in astroglia and GABAergic neurons, respectively, whereas Glu was distributed in glutamatergic neurons, GABAergic neurons, and astroglia in the ratio 82:2:16 in cerebral cortex, as determined using Equation (2) (Table 1). Enrichment of neuronal [4-¹³C]glutamate was calculated under the assumption that at the steady state ¹³C enrichment of astroglial Glu is equal to Gln. The ¹³C Labeling of amino acids from [1-¹³C]/[6-¹³C]glucose generated from [2-¹³C]acetate via gluconeogenesis was corrected by subtracting the [3-¹³C]lactate enrichment.

Table 1. Astroglial Glu pool and V _cyc/V _TCA in different regions of mouse brain.

	3 minutes		90 minutes				Vcyc/VTCA
Brain region	GluC4	GlnC4	GluC4	GABAC2	GlnC4	fGlua	Glutamatergic	GABAergic
Cx	0.54±0.07	3.23±0.35	7.2±0.1	5.7±0.35	15.3±1.1	0.16±0.01	0.39±0.04	0.43±0.02
Hip	0.43±0.16	3.6±0.63	6.0±1.1	4.1±0.7	13.3±1.6	0.12±0.04	0.49±0.03	0.35±0.02
Str	0.46±0.13	3.06±0.5	5.0±0.7	4.5±0.7	13.5±2.0	0.15±0.04	0.24±0.03	0.43±0.02
THt	0.59±0.14	4.52±0.17	5.8±0.8	5.3±0.3	14.5±1.2	0.13±0.03	0.30±0.03	0.44±0.03
Cb	0.26±0.06	2.90±0.53	6.6±0.6	4.9±0.8	15.1±1.7	0.09±0.01	0.40±0.03	0.33±0.02

Abbreviations: Cb, cerebellum; Cx, cortex; Hip, hippocampus; Str, striatum; THt, thalamus–hypothalamus.

Astroglial Glu pool size (fGlu_a) was estimated by using Equation (2) and ¹³C labeling of amino acids obtained during a short-time infusion (3 minutes) of [2-¹³C]acetate. The ratio, V_cyc/V_TCA for glutamatergic and GABAergic neurons was estimated using Equations (3) and (4), and labeling of amino acids from [2-¹³C]acetate at steady state (90 minutes). All values are presented as mean±s.d.

Similarly, the ratio of GABA-Gln cycle to GABAergic TCA cycle (V_{cyc(GABA-Gln)}/V_TCA(GABA)) was calculated as follows:

where GABA_C2 is the steady-state ¹³C labeling of [2-¹³C]GABA from [2-¹³C]acetate.

Determination of Metabolic Rates

The ¹³C turnover of amino acids from [1,6-¹³C₂]glucose was constructed using the measured ¹³C labeling in tissue extract for different infusion times. Metabolic rates were determined by fitting a three-compartment metabolic model (Figure 1; glutamatergic neurons, GABAergic neurons, and astroglia) to the ¹³C turnover of Glu_C4, Glu_C3, GABA_C2, GABA_C3, Gln_C4, and Asp_C3 from [1,6-¹³C₂]glucose.⁴ The metabolic model consists of a series of coupled differential equations reflecting mass balance and ¹³C isotope flowing from [1,6-¹³C₂]glucose (Supplementary Table 1S) into neuronal and astroglial amino acids using the CWave software package executed in Matlab (Mathworks, Natick, MA, USA).¹⁹ The mass and isotope balance equations for the three-compartment metabolic model were similar to those described previously in detail.⁴ The differential equations were solved using the Runge–Kutta algorithm, and the fitting was done using a Levenberg–Marquardt algorithm.²⁰ The cerebral metabolic fluxes were determined from the best fits of the model to the ¹³C turnover data. In the case of the cerebral cortex, Glu pool was distributed among glutamatergic neurons (82%), GABAergic neurons (2%), and astroglia (16%) (Table 1). The ratios V_{cyc(Glu–Gln)}/V_TCA(Glu) and V_{cyc(GABA-Gln)}/V_TCA(GABA) obtained from Equations 3 and 4 were used as constraints to relate the V_cyc to V_TCA when fitting the model to ¹³C turnover of cerebral amino acids from [1,6-¹³C₂]glucose. The uncertainties in metabolic rates were obtained by Monte Carlo analysis, where 25 values of each rates were obtained by fitting the model to the randomly generated ¹³C turnover curve of amino acids.

Figure 1.

A three-compartment metabolic model showing the ¹³C labeling of cerebral metabolites from [1,6-¹³C₂]glucose and [2-¹³C]acetate. Although [1,6-¹³C₂]glucose is metabolized in the neurons and astroglia, [2-¹³C]acetate is selectively transported and metabolized in astroglia. Metabolism of [1,6-¹³C₂]glucose via glutamatergic and GABAergic tricarboxylic acid (TCA) cycle labels Glu_C4. In GABAergic neurons, Glu_C4 is decarboxylated to GABA_C2. Labeling of Gln_C4 occurs from Glu_C4 and GABA_C2 via Glu–Gln and GABA-Gln cycle, respectively. Further metabolism in the TCA cycle labels Glu_C2/3, Gln_C2/3, GABA_C3/4, and Asp_C2/3. The [2-¹³C]acetate metabolism in astroglia labels Gln_C4 by combined action of astroglial TCA cycle and glutamine synthetase activity. Glu_C4 and GABA_C2 are labeled from Gln_C4 via Glu–Gln and GABA-Gln substrates cycling, respectively, between astroglia and neurons. Subsequent metabolism through TCA cycle incorporates labels into Glu_C2/3, Gln_C2/3, GABA_C3/4, and Asp_C2/3V_{cyc(GABA-Gln)}, GABA-Gln cycling flux; V_cyc(Glu-Gln), Glu–Gln cycling flux; V_gad, glutamate decarboxylase flux; V_gln, rate of glutamine synthesis; V_pc, pyruvate carboxylase flux; V_shunt, flux of GABA shunt; V_TCA(A), astroglial TCA cycle flux; V_TCA(GABA), GABAergic TCA cycle flux; V_TCA(Glu), glutamatergic TCA cycle flux.

Statistical Analysis

The statistical significance of differences for metabolite levels and metabolic rates among different brain regions was determined using analysis of variance. All results are presented as mean±s.d.

Results

Plasma Glucose and Acetate Level and Enrichment

Infusion of [1,6-¹³C₂]glucose led to a rapid increase in total glucose level from 8.8±0.7 mmol/L to 15.9±1.7 mmol/L within 7 minutes and remained elevated during the entire course of experiment. The ¹³C enrichment of glucose-C1 also increased rapidly to 37.9% in 7 minutes, and thereafter remained at ∼40% throughout the infusion of [1,6-¹³C₂]glucose. Plasma acetate level and ¹³C enrichment were found to be 3.7±0.8 mmol/L and 81.2±5.2%, respectively, after 90 minutes of [2-¹³C]acetate infusion.

Level of Metabolites in Different Regions of the Mouse Brain

Metabolite levels in different brain regions were measured from nonedited ¹H-[¹³C]-NMR spectrum (Figure 2 upper panel). The levels of metabolites were found to vary across the brain and were distinct in the different regions of the brain (Figure 3). The levels of Glu (13.8±0.1 μmol/g) and N-acetylaspartate (NAA) (8.1±0.2 μmol/g) were highest in the cerebral cortex. However, the level of GABA was highest in the thalamus–hypothalamus (4.3±0.1 μmol/g) and least in the cerebellum (2.3±0.1 μmol/g). The level of inositol was higher in the thalamus–hypothalamus (8.4±0.2 μmol/g) and striatum (8.3±0.3 μmol/g) and least in the cerebral cortex (6.5±0.2 μmol/g). Glutamine level was highest in the cerebral cortex (5.5±0.1 μmol/g) and lowest in the striatum (4.8±0.1 μmol/g).

Figure 2.

Typical ¹H-[¹³C]-nuclear magnetic resonance (NMR) spectra from cortical tissue extract prepared after 90 min of (A) [1,6-¹³C₂]glucose and (B) [2-¹³C]acetate infusion. Upper panel, nonedited ¹H-[¹²C+¹³C] spectrum represents total concentration of neurometabolites. The lower panel represents ¹³C-edited spectrum showing labeling from [1,6-¹³C₂]glucose (left panel) and [2-¹³C]acetate (right panel). Peak labeling is as follows: Ala_C3, alanine-C3; Asp_C3, aspartate-C3; GABA_C2, GABA-C2; Gln_C4, glutamine-C4; Glu_C4, glutamate-C4.

Figure 3.

Metabolite concentration (μmol/g) in different brain regions. The concentration of metabolites was determined from ¹H-[¹³C+¹²C]-nuclear magnetic resonance spectrum using [2-¹³C]glycine as the internal reference. The values are mean±s.e.m. (n=25). *P<0.05, **P<0.001 when compared with respective cortical value.

Labeling of Cerebral Amino Acids from [1,6-¹³C₂]Glucose and [2-¹³C]Acetate

Typical ¹H-[¹³C]-NMR spectra obtained from the cerebral cortex of the mouse infused with either [1,6-¹³C₂]glucose or [2-¹³C]acetate for 90 minutes are shown in Figure 2. The ¹³C intensity of amino acids from [1,6-¹³C₂]glucose (Figure 2A lower panel) is much higher than that of [2-¹³C]acetate (Figure 2B lower panel), thus suggesting glucose as the preferred energy substrate over acetate in the brain. Further, the magnitude of labeling of Glu_C4 and Gln_C4 from [2-¹³C]acetate is opposite to that of [1,6-¹³C₂]glucose. In the cerebral cortex, ¹³C enrichment of Gln_C4 (15.3±1.1%) was found to be significantly higher than Glu_C4 (7.2±0.1%) and GABA_C2 (5.7±0.3%) in [2-¹³C]acetate experiment, whereas labeling of Gln_C4 (23.6±0.5%) was lower than Glu_C4 (36.1±0.5%) and GABA_C2 (29.6±1.1%) in [1,6-¹³C₂]glucose experiment (Supplementary Table 2S). Moreover, labeling of amino acids from glucose was observed to be higher than acetate. The higher labeling of Gln_C4 than Glu_C4 from [2-¹³C]acetate is in accordance with the preferential utilization of acetate in astrocytes, whereas higher labeling of Glu_C4 and GABA_C2 over Gln_C4 from glucose suggests preferential utilization of glucose by neurons.

Spectral time course of ¹³C labeling of cortical amino acids from [1,6-¹³C₂]glucose is shown in Figure 4. Early-time-point infusion spectra (7 and 15 minutes) show well-resolved signals from Glu_C4 and GABA_C2, which are labeled via the first turn of the TCA cycle; furthermore, later-time-point spectra (30, 60, and 90 minutes) exhibited signals from Glu_C3, Asp_C3, GABA_C3, and GABA_C4, which are labeled in the subsequent turn of the TCA cycle. Spectral time course displayed well-resolved signals from Glu_C4,C3, Gln_C4, GABA_C2,C3,C4, and Asp_C3, which increased with time. The ¹³C turnover of amino acids from [1,6-¹³C₂]glucose at different carbon positions, constructed by plotting the normalized ¹³C enrichment of amino acids with plasma glucose enrichment with time, was used for metabolic flux analysis.

Figure 4.

Representative ¹H-[¹³C]-nuclear magnetic resonance (NMR) spectra of cortical extract with time showing the ¹³C labeling of cortical metabolites from [1,6-¹³C₂]glucose. Mice were infused with [1,6-¹³C₂]glucose for a predefined duration, and ¹H-[¹³C]-NMR spectra were recorded from cortical tissue extracts. Peak labeling is the same as in Figure 3.

Glutamate Pool Distribution in Neurons and Astroglia

Glutamatergic and GABAergic TCA cycle and neurotransmitter cycle fluxes in different regions of the mouse brain were evaluated by fitting a three-compartment metabolic model (Figure 1; Supplementary Table 1S) to the ¹³C turnover of amino acids from [1,6-¹³C₂]glucose. One of the important parameters in the model is the astroglial Glu pool size, which is not established in different regions of the mouse brain. Astroglial Glu pool was estimated from a short infusion of [2-¹³C]acetate and was found to vary from 9% to 16% in different regions of the mouse brain (Table 1). Astroglial Glu pool was found to be highest in the cerebral cortex (16%) and least in the cerebellum (9%), whereas other regions (hippocampus 12% thalamus–hypothalamus 13% striatum 15%) showed an intermediate value.

Glutamatergic and GABAergic Fluxes

The ratio of V_cyc/V_TCA for glutamatergic and GABAergic neurons for different regions of the mouse brain was calculated using Equations 3 and 4, respectively, from the steady-state ¹³C labeling of amino acids from [2-¹³C]acetate (Table 1). The ratio V_cyc/V_TCA for glutamatergic neurons was found to be highest for the hippocampus (0.49±0.03) and least for the striatum (0.24±0.03) (Table 1). For GABAergic neuron, V_cyc/V_TCA was higher in the thalamus–hypothalamus (0.44±0.03), cortex (0.43±0.02), and striatum (0.43±0.03), and lowest in the cerebellum (0.33±0.02).

The fit of the metabolic model to ¹³C turnover of amino acids in the cerebral cortex is depicted in Figure 5. The random distribution of residual together with low value of the least-square standard deviation between measured and predicted turnover suggests a good fit of the model to the measured data. The quality of the fit of the model to ¹³C turnover of amino acids was similar for other brain regions, as revealed by similar least-square standard deviation. The glutamatergic TCA cycle rate in the cerebral cortex (0.91±0.05 μmol/g per minute) was observed to be significantly higher (F[1,48]=231, P=5.3e-²⁰) than other brain regions (Figure 6, Supplementary Table 3S). The TCA cycle flux decreased in the order of cerebral cortex>cerebellum (0.81±0.09 μmol/g per minute)> striatum (0.75±0.05 μmol/g per minute)∼thalamus–hypothalamus (0.73±0.05 μmol/g per minute)>hippocampus (0.64±0.07 μmol/g per minute). The Glu–Gln cycle rate was higher (F[1,48]=14, P=0.0004) in the cerebral cortex (0.36±0.02 μmol/g per minute) as compared with the other regions. The glutamatergic neurotransmitter cycle rate was in the following order: cerebral cortex>cerebellum (0.32±0.04 μmol/g per minute)∼hippocampus (0.31±0.04 μmol/g per minute)>thalamus–hypothalamus (0.22±0.02 μmol/g per minute)>striatum (0.18±0.01 μmol/g per minute).

Figure 5.

The fit of the metabolic model to the ¹³C turnover of cortical (A) GluC4 and GluC3, (B) GABAC2 and GABAC3, (C) GlnC4 and AspC3 from [1,6-¹³C₂]glucose. Mice were infused with [1,6-¹³C₂]glucose for a predefined duration. The ¹³C enrichment of cortical amino acids was measured in extract using ¹H-[¹³C]-nuclear magnetic resonance spectroscopy. The ¹³C concentration was obtained by multiplying the total concentration with corresponding ¹³C enrichment. Residual (R) between measured and fitted time course was obtained as follows: (measured−calculated) × (calculated)^−1/2. Symbols represent the measured ¹³C labeling, whereas lines show the best fit of the three-compartment metabolic model to the measured data.

Figure 6.

Cerebral metabolic rates in different regions of the mouse brain. (A) tricarboxylic acid (TCA) cycle flux, (B) Neurotransmitter cycling flux. Metabolic rates were calculated by best fit of the three-compartment metabolic model to the measured ¹³C turnover of cerebral amino acids from [1,6-¹³C₂]glucose. The values are mean±s.d. *P<0.01 glutamatergic flux; ^#P<0.05, ^##P<0.01 GABAergic rate; ^∞P<0.01 astroglial TCA cycle rate when compared with the corresponding cortical rate.

The GABAergic TCA cycle flux was significantly (F[1,48]=5.2, P=0.026) higher in the thalamus–hypothalamus (0.28±0.01 μmol/g per minute) than in other brain regions. The V_TCA(GABA) decreased in the following order: thalamus–hypothalamus>striatum (0.27±0.02 μmol/g per minute)> hippocampus (0.26±0.04μmol/g per minute)>cerebellum (0.25±0.03 μmol/g per minute)> cerebral cortex (0.24±0.02 μmol/g per minute) (Figure 6, Supplementary Table 3S). The rate of GABA-Gln cycle in the thalamus–hypothalamus (0.12±0.01 μmol/g per minute) was found to be significantly (F[1,48]=11.89, P=0.0011) higher than that in other brain regions. The GABAergic neurotransmission decreased in the following order: thalamus–hypothalamus>striatum (0.11±0.01 μmol/g per minute)>cerebral cortex (0.10±0.01 μmol/g per minute)>hippocampus (0.09±0.01 μmol/g per minute)>cerebellum (0.08±0.01 μmol/g per minute). The astroglial TCA cycle flux (V_TCA(A)) was also found to vary across the different regions of the brain in the following order: cerebellum (0.28±0.03 μmol/g per minute)>hippocampus (0.26±0.05 μmol/g per minute)∼thalamus–hypothalamus (0.26±0.01 μmol/g per minute)∼cerebral cortex (0.25±0.03 μmol/g per minute)>striatum (0.19±0.02 μmol/g per minute).

Discussion

The current study provides, for the first time, the quantitative fluxes associated with different metabolic pathways across the mouse brain. A particular region of the brain is attributed to specific function, which depends on the strength of the neuronal firing of the task associated with the brain region. Hence, the magnitude of energy consumed and neurotransmitter cycling rate is expected to vary across the brain. Usually, a particular region of the brain is affected in a given neurological condition. Hence, the understanding of neurotransmitter energetics across the brain may be useful for the diagnosis of the disease and to understand the disease manifestation during the treatment. In the present study, the three-compartment metabolic model was extended in mice to quantify the glutamatergic and GABAergic TCA cycle and neurotransmitter cycle rates in different regions of the mouse brain. To the best of our knowledge, this is the first comprehensive study that has evaluated the glutamatergic and GABAergic TCA cycle and neurotransmitter cycle flux across the brain of young adult mice. Our findings indicate that both the glutamatergic and GABAergic TCA cycle and neurotransmitter cycle rates are very distinct across different brain regions.

Neurochemical Profile in the Mouse Brain

Neurochemical profile has been extensively evaluated in the rat brain using ex vivo and in vivo ¹H NMR spectroscopy.^{15, 21, 22} These studies have indicated inter-regional variability in the level of neurometabolites. Measurement of neurochemical profile across mouse brain has been challenging owing to the small size of the brain. However, recently, mouse brain ¹H MR spectroscopy has gained momentum because of the improvement in NMR hardware and increased availability of genetically engineered mice for the investigation of different human diseases. A range of values has been reported for cortical Glu (10.3 to 12.4 μmol/g), GABA (1.8 to 2.0 μmol/g), NAA (8.2 to 9.2 μmol/g), Ins (4.2 to 4.6 μmol/g), and other metabolites.^{23, 24, 25, 26} In the present study, the level of cortical Glu (13.8 μmol/g), GABA (2.4 μmol/g), NAA (8.1 μmol/g), and Ins (6.5 μmol/g) is slightly higher than in vivo values, but significantly lower than ex vivo values.²⁷ Although absolute levels of metabolites are reported to vary across different studies, there seems to be a specific pattern for variation in the level of neurometabolites across the brain. Our data indicate that Glu level decreases in the following order: the cerebral cortex>hippocampus>striatum∼thalamus–hypothalamus>cerebellum. The level of GABA was found to increase in the following order: cerebral cortex∼cerebellum>hippocampus>striatum>thalamus–hypothalamus. The observed patterns of variation in the level of neurometabolite across the brain were found to be in agreement with the previous in vivo studies in mice.^{24, 25, 26}

Glu and NAA levels were found to be highest in the cortical region, indicating more excitatory synapses and density of neurons in the cerebral cortex. The finding of higher level of Glu in the cortical region is consistent with earlier reports, which stated that the excitatory synapses outnumber the inhibitory synapses in the cerebral cortex.^{28, 29} Higher level of GABA in the thalamus–hypothalamus and striatum may be due to higher density of GABAergic neurons/synapses in these regions. Indeed, the hypothalamus and striatum in rats have been shown to have higher densities of GABAergic neurons when compared with other brain regions.³⁰ Nuclear magnetic resonance studies have also suggested a higher value of GABA in the hypothalamus when compared with other brain regions.^{15, 25} The level of inositol is comparatively higher in the striatal and thalamic regions, suggesting that these regions may have a higher astroglial population.

The Metabolic Model

The metabolic model used for the analysis of the ¹³C turnover data in the present study is the same as that used in the rat brain.⁴ The model is well parameterized for rat cerebral cortex and has not yet been tested for other brain regions, which may have different proportions of cellular population. Glutamate, GABA, Gln, and aspartate (Asp) are readily labeled from [1,6-¹³C₂]glucose or [2-¹³C]acetate in different regions of the mouse brain, indicating that trafficking of neurotransmitters, Glu and GABA, between neurons and astrocytes is operational across the mouse brain. The total pool of Glu, GABA, Gln, and Asp was measured across different brain regions in the present study. Astroglial Glu pool size, an important parameter of the model, was determined using a short-time infusion of [2-¹³C]acetate. Similar to rat brain, GABA and Gln were assumed to be localized exclusively in GABAergic neurons and astrocytes, respectively. The pool size of α-ketoglutarate and oxaloacetate was the same as that in Patel et al.⁴

Glucose Metabolism and Neurotransmitter Cycling in the Cerebral Cortex

Over the past two decades, glucose metabolism and neurotransmitter cycle have been studied in rat^{4, 7, 8, 9, 10} and human brain.^{11, 12, 13} It has been well established that cerebral metabolic rates depend on the type and dose of anesthetics used, which affect the brain activity.^{5, 17} Urethane (1.5 mg/kg) may be treated as a light anesthetic; hence, studies conducted under light anesthetics or awake conditions are considered for further discussion. A very recent study conducted in an awake rat has indicated cortical glucose oxidation in the range of 0.45 to 0.55 μmol/g per minute.¹⁵ However, another study conducted in the whole brain has reported a very high value (0.91 μmol/g per minute) for neuronal glucose oxidation.³¹ Moreover, a three-compartment modeling of ¹³C turnover of amino acids from glucose in halothane-anesthetized rat has revealed the rate of total (glutamatergic+GABAergic) neuronal glucose oxidation to be 0.60 μmol/g per minute.⁴ The total neuronal glucose oxidation, 0.55±0.04 μmol/g per minute (1.11±0.07÷2), obtained in the present study in mouse cerebral cortex was found to be lower than the reported value under awake conditions;³¹ however, it is very close to the rates in halothane-anesthetized rats.⁴

In an awake rat cortex, total neurotransmitter cycling flux has been reported to be in the range from 0.47 to 0.57 μmol/g per minute.^{15, 31} The total cortical neurotransmitter cycling has been reported to be 0.57 μmol/g per minute in halothane-anesthetized rat.⁴ Our finding of total neurotransmitter cycle flux (0.45±0.03 μmol/g per minute) in the mice cerebral cortex was found to be in good agreement with these values. In the rat cerebral cortex, neurotransmitter cycle flux has been reported to be 80% of the neuronal glucose oxidation.^{5, 17} In the present study, neurotransmitter cycle rate was found to be ∼83% of the neuronal glucose oxidation, suggesting that in the mouse cortex also most of the energy is used for supporting the processes associated with neurotransmission.

Glucose Metabolism and Neurotransmitter Cycling in Different Brain Regions

Until now, most of the regional cerebral metabolic studies have been carried out in rats,^{15, 32} and very little information is available for regional glucose metabolism and neurotransmitter cycling flux in brain of mice. The ¹⁴C deoxyglucose measurements of cerebral glucose utilization has indicated a decrease in the local glucose consumption in the following order: cerebral cortex>thalamus–striatum>hippocampus.¹⁴ The ¹³C NMR study in rats has indicated that glucose oxidation and neurotransmitter cycle flux decreased in the order of cerebral cortex>corpus callosum>sub-cortex in rat brain,⁷ suggesting that total glucose oxidation and neurotransmitter cycle might decrease in the order of cortex>striatum∼thalamus>hippocampus. Moreover, glucose oxidation was reported to decrease in the order of cerebral cortex>striatum∼cerebellum∼hippocampus>thalamus–hypothalamus in rats in awake condition.¹⁵ Our data indicate that neuronal glucose oxidation in the mouse brain decreases in the order of cerebral cortex>cerebellum>striatum∼thalamus–hypothalamus>hippocampus (Supplementary Table 3S). The slight variation in the pattern of metabolic flux in the current study may be due to differences in the animal species studied.

Contribution of Excitatory and Inhibitory Fluxes in Different Regions of the Brain

The contribution of GABA to total neurotransmission and neuronal glucose oxidation has been evaluated in anesthetized rat cortex. The glucose oxidation by glutamatergic and GABAergic neurons in rat cerebral cortex under halothane anesthesia was 0.50 and 0.11 μmol/g per minute, respectively, and the corresponding neurotransmitter cycle fluxes were 0.45 and 0.14 μmol/g per minute, respectively.⁴ The contribution of GABA to total neuronal glucose oxidation and neurotransmitter cycling fluxes was estimated to be 18% and 23%, respectively.^{4, 6} The present study indicates that the contribution of GABA to total neuronal TCA cycle is 21% (glutamatergic 0.91; GABAergic 0.24 μmol/g per minute), whereas it is 22% (glutamatergic 0.36 μmol/g per minute; GABAergic 0.10 μmol/g per minute) for the neurotransmitter cycle in mouse cerebral cortex, which is similar to the values reported in rat cortex. Further, our findings also indicate that the contribution of GABA to neurotransmitter cycle flux increases in the order of cerebellum (20%)<cerebral cortex (22%)∼hippocampus (22%)<thalamus–hypothalamus (36%)<striatum (39%) (Figure 6B, Supplementary Table 3S). The glutamatergic neurotransmission increases in the following order: striatum<thalamus–hypothalamus<hippocampus<cerebellum<cerebral cortex (Figure 6A). The finding of maximum contribution of GABA to neurotransmitter cycling in striatum and thalamus is in good agreement with the higher density of GABAergic neurons in these regions.^{30, 33}

Neurons and astrocytes function in a coordinated manner to achieve optimum brain functions. The total energy demand of the brain to carry out neurotransmitter activity is the sum total of TCA cycle flux of neurons and astrocytes. The data from the present study revealed that astroglial TCA cycle flux was highest in the hippocampus (22% V_TCA(A) 0.26 μmol/g per minute, V_TCA(Glu) 0.64 μmol/g per minute, V_TCA(GABA) 0.26 μmol/g per minute), followed by the cerebellum (21%)∼thalamus–hypothalamus (21%)>cerebral cortex (18%)>striatum (16%).

Implications for Functional Neuroimaging

Traditionally, functional studies using fMRI and PET were concerned with the energy requirement of excitatory neurons.^{34, 35} Inhibition of the human motor cortex evoked no measurable changes in the blood-oxygenation-level-dependent signal in the no-go condition, indicating that inhibition is less metabolically demanding.³⁶ However, energy budget analysis has suggested that the energetic requirement for inhibitory neurons may be similar to its excitatory counterpart.³⁷ Recently, the relationship between baseline GABA levels and functional signal in human visual cortex has been investigated.³⁸ The finding indicates that blood-oxygenation-level-dependent fMRI signal is inversely correlated with baseline GABA levels, thus predicting that the functional response in the thalamus, where GABA level is significantly higher than other brain regions, will be lesser. In vivo studies showing increased ¹⁴C₂-deoxyglucose uptake in areas densely populated by inhibitory synapses during selective stimulation of inhibitory pathways further strengthens the hypothesis of similar energetics requirement for excitatory and inhibitory neurons.³⁹ The high energy demand for inhibitory synapses has been further established by the observation of rise in local field potentials and cerebral blood flow during simultaneous stimulations of the inhibitory parallel fibers and excitatory climbing fibers in cerebellar neurons.⁴⁰ In the rat cerebral cortex, inhibitory and excitatory neurons are estimated to be 15 to 30% and 70 to 85% of total neurons, respectively. Further, studies in rat^{4, 6} and mouse cerebral cortex (present study) indicated that GABAergic neurons account for ∼20% of neuronal TCA cycle flux. Moreover, the GABAergic rate has been shown to increase from isoelectricity to higher activity.⁴ The current data indicate that the contribution of GABA to neurotransmitter activity increases from cerebellum (20%)<cerebral cortex (22%)∼hippocampus (22%)<thalamus–hypothalamus (36%)<striatum (39%). Hence, a selective activation of inhibitory neurons is expected to result in a higher functional signal in the striatal and thalamic regions and a least functional signal in the cerebellum and cerebral cortex. The functional output due to the stimulation of inhibitory synapses in other brain regions will be in between the cerebellum and the striatum. However, the magnitude of the functional signal in a given brain region and paradigm will depend on the collective response of the excitatory and inhibitory synapses to the stimulus.

Limitations

The metabolic analysis carried out in the current study involves macrostructures such as the cerebral cortex, cerebellum, hippocampus, hypothalamus–thalamus, and striatum. It will be interesting to study the metabolic fluxes in microstructures such as the amygdala and substantia nigra, which are involved in different disorders such as addictions and Parkinson's disease. The astroglial Glu pool was estimated using a strategy of short-time infusion of [2-¹³C]acetate. The estimation of astroglial Glu pool involves the assumption that ¹³C label is not transferred from astroglial Gln to neuronal Glu. However, small leakage of ¹³C label into neurons cannot be ruled out. Furthermore, assumption of the labeling of glial Glu pool equal to Gln may also not be fully valid in a short-time infusion. Hence, the fraction of astroglial Glu pool obtained by Equation (2) will be overestimated to the extent that the ¹³C label is transferred into neuronal Glu.

Conclusion

Our findings reveal that the metabolite levels are very different across the various regions of the mouse brain. Cortical regions exhibited the highest level of Glu and NAA, which shows a higher number of excitatory neurons and density, whereas the thalamic and striatal regions exhibited higher GABAergic density. The finding of higher ¹³C labeling of cerebral amino acids from glucose than acetate is consistent with the view of glucose as the preferred energy substrate for the brain than acetate. The neurotransmitter cycle in the mouse brain was found to decrease in the order of cerebral cortex>cerebellum>hippocampus>thalamus–hypothalamus>striatum. Glutamatergic activity was found to be highest in the cerebral cortex, whereas the GABAergic activity was observed to be maximum in the subcortical regions. To the best of our knowledge, the present investigation is the first comprehensive study that has evaluated the glutamatergic and GABAergic TCA cycle and neurotransmitter cycle fluxes across different regions of the mouse brain. These findings indicate that both the glutamatergic and GABAergic TCA cycle and neurotransmitter cycle rates are very distinct across different brain regions.

The authors declare no conflict of interest.