Group I and II metabotropic glutamate receptors alter brain cortical metabolic and glutamate/glutamine cycle activity: a 13C NMR spectroscopy and metabolomic study


Address correspondence and reprint requests to Dr Caroline Rae, School of Molecular & Microbial Biosciences, The University of Sydney, NSW 2006, Australia. E-mail:


Metabotropic glutamate receptors (mGluR) modulate neuronal function. Here, we tested the effect on metabolism of a range of Group I and II mGluR ligands in Guinea pig brain cortical tissue slices, applying 13C NMR spectroscopy and metabolomic analysis using multivariate statistics. The effects of Group I agonists (S)-3,5-dihydroxyphenylglycine (DHPG) and (RS)-2-chloro-5-hydroxyphenylglycine (CHPG) depended upon concentration and were mostly stimulatory, increasing both net metabolic flux through the Krebs cycle and glutamate/glutamine cycle activity. Only the higher (50 µm) concentrations of CHPG had the opposite effect. The Group I antagonist (RS)-1-aminoindan-1,5-dicarboxylic acid (AIDA), consistent with its neuroprotective role, caused significant decreases in metabolism. With principal components analysis of the metabolic profiles generated by these ligands, the effects could be separated by two principal components. Agonists at Group II mGluR [(2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine (DCG IV) and 2R,4R-4-aminopyrrolidine-2,4-dicarboxylate (APDC)] generally stimulated metabolism, including glutamate/glutamine cycling, although this varied with concentration. The antagonist (2S)-α-ethylglutamic acid (EGLU) stimulated astrocyte metabolism with minimal impact on glutamate/glutamine cycling. (RS)-1-Aminophosphoindan-1-carboxylic acid (APICA) decreased metabolism at 5 µm but had a stimulatory effect at 50 µm. All ligand effects were separated from control and from each other using two principal components. The ramifications of these findings are discussed.

Abbreviations used

(RS)-1-aminoindan-1,5-dicarboxylic acid




(RS)-1-aminophosphoindan-1-carboxylic acid




IV, (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine


(S)-3,5-dihydroxyphenylglycine 0.75 H2O


(2S)-α-ethylglutamic acid


metabotropic glutamate receptor




principal component


principal components analysis


sodium 2,2,3,3-tetradeutero-3-trimethylsilyl propionate

Metabotropic glutamate receptors (mGluR) are a family of G-protein-coupled glutamate receptors, whose activation produces changes with a relatively slow time course, and which are not directly linked to rapid and large changes in neuronal membrane conductance. There are currently eight known mGluR genes classified into three groups based on sequence homology, coupling mechanisms and pharmacological properties (Nakanishi 1992). Group I mGluR comprises mGluR1 and mGluR5, and transduce signals through phospholipase C activity. Group II mGluR comprises mGluR2 and mGluR3, and Group III mGluR comprises mGluR4, 6, 7 and 8. Both Groups II and III are coupled to adenylyl cyclase.

Substances that activate Group I mGluR induce neuronal depolarization and excitation and enhance glutamate release (Bordi and Ugolini 1999; Cartmell and Schoepp 2000) but can also modulate synaptic transmission by regulating GABA release (Battaglia et al. 2001; Smolders et al. 2004) or by altering the action of ionotropic glutamate receptors (Pisani et al. 1997). Agonists at Group I mGluR have been reported to be both neuroprotective and neurotoxic (Ong and Balcar 1997; Strasser et al. 1998; Nicoletti et al. 1999; Wisniewski and Car 2002), while Group I antagonists attenuate excitotoxic neuronal death in cortical cultures (Strasser et al. 1998) and have anti-convulsive effects (Guérineau et al. 1995). Group I mGluR are predominantly located perisynaptically on postsynaptic neurons (Shigemoto et al. 1993; Lujan et al. 1996).

Group II mGluR are expressed mainly in the cerebral neocortex, striatum, hippocampus, septal nuclei and cerebellar cortex (Shave et al. 2001). At the cellular level, mGluR II are found both pre- and postsynaptically (Ohishi et al. 1994; Neki et al. 1996; Cartmell and Schoepp 2000), with subtype mGluR3 found also on astrocytes (Ohishi et al. 1994; Cartmell and Schoepp 2000). Activation of Group II mGluR has consistently been shown to suppress excitatory amino acid release (Cartmell and Schoepp 2000). Activation of presynaptically-located mGluR2 and mGluR3 has been shown to be neuroprotective (Allen et al. 1999), but this has been suggested to be primarily due to the activity of astrocyte-located rather than neuronal mGluR (Moldrich et al. 2001). Subtype mGluR2 is also located perisynaptically where it has been suggested to modulate feedback under conditions of greatly enhanced glutamate release, where glutamate has ‘leaked’ from the synapse and is able to diffuse to perisynaptic sites; feedback inhibition would result in decreased glutamate release (Cartmell and Schoepp 2000). Increasing levels of the endogenous mGluR II agonist N-acetyl-l-aspartyl-l-glutamate (NAAG) could have a neuroprotective effect [(Slusher et al. 1999; Thomas et al. 2001) but see also (Pliss et al. 2000, 2003)]. Both mGluR I and II could therefore be strongly involved in the mechanisms of glutamate-related neuronal death. In view of the fact that changes in its metabolism may represent an important element of glutamate neurotoxicity (Rae et al. 2000), it seems surprising that the effects of mGluR I and mGluR II on brain metabolism, glutamate/glutamine cycling and on metabolic compartmentation remain largely unexplored.

Metabolic turnover in the brain is known to be related to neuronal activity (Badar-Goffer et al. 1992; Shank et al. 1993; Griffin et al. 1999; Sibson et al. 2001), and has been shown to be highly sensitive to perturbations in the activity of transporters (Moussa et al. 2002; Rae et al. 2003) and receptors (Rae et al. 2002; Stanton et al. 2003). Further, by analysing the pattern of isotope distribution following administration of 13C-enriched substrate, information about the metabolic compartments that are activated or inhibited by receptor or transporter modulation can be uncovered (Waagepetersen et al. 2001; Zwingmann et al. 2001; Rae et al. 2003). Here, we have used 13C NMR spectroscopic analysis of metabolic profiles subsequent to incubation of brain cortical tissue slices with mGluR-specific ligands and [3-13C]pyruvate, coupled with principal components analysis (PCA) of the data, to examine the role of Group I and II mGluR in modulating metabolism.

Materials and methods


Guinea pigs (Dunkin–Hartley), weighing 400–800 g were fed ad libitum on standard Guinea pig/rabbit pellets, with fresh cabbage leaves and lucerne hay roughage. Water bottles were supplemented with l-ascorbic acid (1 g/L). Animals were maintained on a 12 h light/dark cycle. All experiments were conducted in accordance with the guidelines of the National Health and Medical Research Council (NHMRC) of Australia and were approved by the institutional Animal Care Ethics Committee.

d-[1-13C]Glucose (99.9%), sodium [3-13C]pyruvate and sodium [13C]formate were purchased from Cambridge Isotope Laboratories Inc. (Andover, MA, USA). (S)-3,5-Dihydroxyphenylglycine (DHPG), (RS)-1-aminoindan-1,5-dicarboxylic acid (AIDA), 2S,2′R,3′R-2-(2′,3′-dicarboxycyclopropyl)glycine (DCG IV), 2S-α-ethylglutamic acid (EGLU), (RS)-1-aminophosphoindan-1-carboxylic acid (APICA), 2R,4R-4-aminopyrrolidine-2,4-dicarboxylate (APDC) and (RS)-2-chloro-5-hydroxyphenylglycine (CHPG) were purchased from Tocris Cookson (Avonmouth, UK). All other reagents were of AR grade.

Preparation of brain cortical tissue slices

Guinea pigs were killed by cervical dislocation. The brain was rapidly removed from the cranial vault. The cortex was dissected and chopped into 350 µm slices in the paraxial plane using a McIlwain tissue chopper (The Mickle Laboratory Engineering Company, Gomshall, UK). The resulting slices were immediately washed three times in a modified Krebs–Henseleit buffer [124 mm NaCl, 5 mm KCl, 1.2 mm KH2PO4, 1.2 mm CaCl2, 1.2 mm MgSO4 and 26 mm NaHCO3 (Badar-Goffer et al. 1990)], resuspended for 1 h in fresh buffer containing 10 mm unlabelled glucose, and gassed with 95% O2/5% CO2 in a shaking water bath, maintained at 37°C, to allow metabolic recovery (McIlwain and Bachelard 1985). Slices were then washed three times in glucose-free buffer and resuspended in fresh buffer with the substrate of choice.

Modulation of mGluR activity

To determine the effect of modulation of mGluR activity on metabolic activity, slices were incubated with 2 mm sodium [3-13C]pyruvate (control) and also, in the case of ligand treatment groups, with one of two concentrations of the ligand. These included 5 and 50 µm DHPG [Group I mGluR agonist (Schoepp et al. 1994)], 5 and 50 µm CHPG [mGlu5 specific agonist (Doherty et al. 1997)], 5 and 30 µm AIDA [Group I mGluR antagonist (Pellicciari et al. 1995)], 2 and 20 µm DCG IV [Group II mGluR agonist (Cartmell et al. 1998)], 5 and 50 µm APDC [Group II specific agonist (Schoepp et al. 1999)], 5 and 50 µm APICA [Group II mGluR antagonist (Ma et al. 1997)], and 5 and 50 µm EGLU [Group II mGluR antagonist (Jane et al. 1996)]. The concentrations of ligands were chosen to reflect the affinity of mGluR agonists and antagonists and to separate other possible non-specific activities. As we have observed that 13C from [3-13C]pyruvate labels metabolic pools differently to d-[1-13C]glucose (Rae et al. 2003), an experiment was also performed using d-[1-13C]glucose (control) incubated with 5 or 50 µm EGLU.

Slices were incubated for 1 h with [3-13C]pyruvate or for 90 min with d-[1-13C]glucose and the experiment was stopped as indicated below.

To monitor cell viability in the presence of mGluR ligands, the intracellular [K+] was measured by atomic absorption spectrometry (SpecterAA 20 Plus Varian Spectrometer; Varian, Victoria, Australia) using the supernatant fluid obtained after the slices were lysed in MilliQ water, homogenized and centrifuged. The protein concentration of the pellet was measured by the Lowry technique and used to standardize the measured [K+]. Experimental samples (n = 10) were measured in triplicate and mean values recorded for each sample. Data were tested for statistical significance using the Newman–Keuls test (GraphPad Prism software Version 3, San Diego, USA). An intracellular [K+] > 80 mm was considered to represent viable tissue (McIlwain and Bachelard 1985).

Preparation of samples and NMR analysis

On completion of the incubation period, slices were removed from the incubation buffer by rapid filtration and extracted in ice-cold 6% perchloric acid. Extracts were neutralised to pH 7.2 with KOH, lyophilized, and the pellet retained for protein estimation by the Lowry technique. Lyophilized supernatant fluids were stored at − 20°C until required for NMR analysis. Samples were resuspended in 0.65 mL D2O containing 2 mm sodium [13C]formate as an internal intensity and chemical shift reference (13C δ 171.8). 13C [1H-decoupled] spectra (typically 14 000–18 000 transients, duty cycle 4 s, 83 300 data points) were acquired at 9.4 T on a Bruker DRX-400 WB spectrometer (Bruker Biospin, Karlsruhe, Germany) using a 5 mm dual 1H/13C probe. Fully relaxed 1H and 1H[13C-decoupled] spectra [duty cycle 30 s, WURST-40 (Kupce and Freeman 1995) with a 112-step phase cycle (Skinner and Bendall 1997), decoupling during acquisition] were obtained at 600.13 MHz on a Bruker DRX-600 spectrometer with a broadband inverse xyz-gradient probe. Assignments were aided by reference to standard spectra, coupling constant analysis, and by the acquisition of heteronuclear single quantum coherence and heteronuclear multiple-bond correlation spectra with gradient selection, employing standard Bruker pulse sequences (Willker et al. 1993) and a broadband inverse xyz-gradient probe at 600.13 MHz.

Following zero filling to 128 k, [1H-decoupled]13C spectra were transformed using 3 Hz exponential line-broadening and peak areas were determined by integration using standard Bruker software (xwinnmr, Version 3.5). Peak areas were adjusted for nuclear Overhauser effect, saturation and natural abundance effects, and quantified by reference to the area of the internal standard resonance of [13C]formate. Glu C3 was not quantified when [3-13C]pyruvate was used as substrate due to resonance overlap with pyruvate dimer, as described previously (Rae et al. 2000). Metabolite pool sizes (lactate, alanine, GABA, glutamate, glutamine and aspartate) were determined by integration of resonances in fully relaxed 600 MHz 1H[13C-decoupled] spectra using [13C]formate as the internal intensity reference.

Experimental data (n = 4) are given as means (standard deviation). Statistical analysis was carried out using anova for comparing ligand-treated metabolism with control, followed by non-parametric (Mann–Whitney U-test) tests where statistical significance was indicated by Scheffe F-test (statview Student). Significance was assumed at α = 0.05.

Slice oxygen consumption and lactate production

Oxygen consumption by tissue slices incubated with mGluR ligands (apart from APDC and CHPG which were not tested) was measured using a Rank oxygen electrode (Rank Brothers, Cambridge, UK). Tissue slices were prepared and allowed to recover as described above. Aliquots of slices were then transferred into the electrode with the mGluR ligand of interest and the rate of slice oxygen consumption measured. Slices were recovered following the experiment for measurement of protein content, which was used to standardize oxygen consumption rates. For this experiment, n = 4.

Lactate production by the slices was measured by sampling the buffer in which slices were incubated with 5 mm glucose and the same concentrations of mGluR ligands (apart from APDC and CHPG which were not tested) used in the 13C-labelling experiments. Samples were taken in duplicate every 5 min for 1 h and the lactate content assayed using an automated lactate assay (Roche Diagnostics Systems, NSW, Australia) on a Cobas Fara spectrophotometer (Roche Diagnostic Systems). Rates of lactate production were standardized for slice protein content.

Pattern recognition of the data

Principal components analysis (PCA) was applied to the concentrations of 13C-labelled glutamate (C2 and C4 positions), GABA (C2), lactate (C3), glutamine (C4), aspartate (C2, C3, and the difference C2–C3) and alanine (C3), as well as the total concentrations of lactate, alanine, glutamate, glutamine, aspartate and GABA for both the mGluR I and mGluR II ligands. These variables were imported in Simca-P (Umetrics, Umeå, Sweden). Prior to PCA, the data were scaled using autoscaling. This scaling subtracts the mean value of the variable from each measured variable for the total data set, and divides this number by the variance across that variable. This autoscaling routine ensures that each variable contributes equally to the resultant PCA model. Without this scaling, the resultant PCA models would be most influenced by metabolites with high concentration or labelling, rather than all the variables in the total data set. The scores of the first three components of the PCA models were investigated to identify clustering of the treatment groups in this reduced dimensionality. Where clustering was apparent, the contributions of the variables to these principal components (PC) were investigated to identify which metabolites contributed the most to the separations identified.


Group I mGluR agonist, DHPG

The mGluR agonist DHPG significantly (p < 0.05) increased net flux of 13C from [3-13C]pyruvate into Glu C2 and C4, Gln C4, GABA C2, and Asp C2 and C3 (Fig. 1a). The total metabolite pool sizes of aspartate [2.37 (0.37) µm/100 mg protein, control; 3.58 (0.79) 5 µm DHPG; 3.13 (0.44) 50 µm DHPG] and glutamine [1.15 (0.10) µm/100 mg protein, control; 2.44 (0.43) 5 µm DHPG; 2.23 (0.27) 50 µm DHPG] were also significantly (p < 0.05) increased, while those of glutamate, lactate, GABA and Ala were unchanged. In keeping with these effects, the fractional enrichments of Glu C2 [6.1 (1.3)%, control; 11.2 (1.0) 5 µm DHPG; 7.5 (2.3) 50 µm DHPG] and Glu C4 [19.1 (1.6)% control; 30.9 (1.0) 5 µm DHPG; 27.5 (2.5) 50 µm DHPG], GABA C2 [8.0 (1.5)% control; 16.8 (2.1) 5 µm DHPG; 12.5 (1.8) 50 µm DHPG], and Asp C2 [9.3 (1.4)% control; 16.7 (3.4) 5 µm DHPG; 13.5 (3.3) 50 µm DHPG] and Asp C3 [11.3 (1.4)% control; 20.2 (3.6) 5 µm DHPG; 16.8 (3.1) 50 µm DHPG] were also significantly increased (p < 0.05). The metabolic profile in the presence of 5 µm DHPG was consistent with stimulation of both Krebs cycle flux (as seen in the increased flux into Glu C2 and C4, and Asp C2 and C3, and increased metabolic pool size of aspartate) and glutamate/glutamine cycle activity (as seen in the increased flux into Glu C4, Gln C4 and Ala C3). In the presence of 50 µm DHPG, while glutamate/glutamine cycle rates remained increased compared with control (as seen in the increased flux into Glu C4, Gln C4 and Ala C3), the flux into the Krebs cycle was decreased compared with that in the presence of 5 µm DHPG (seen in the decrease in flux into Glu C2, Asp C2 and C3), although this decrease was not sufficient to impact upon metabolite pool sizes.

Figure 1.

Effects of Group I mGluR agonist and antagonist on net flux of 13C in brain cortical tissue slices incubated for 1 h with sodium [3-13C]pyruvate. (a) Slices were incubated with 5 (single hatching) or 50 (double hatching) µm DHPG, shown compared with control (no hatching). *Significantly different to control; # significantly different to 50 µm DHPG. (b) Same as for (a) but with 5 (single hatching) or 50 (double hatching) µm CHPG. (c) Slices were incubated with 5 (single hatching) or 30 (double hatching) µm AIDA, shown compared with control (no hatching). *Significantly different to control; #significantly different to 30 µm AIDA. n = 4, data are shown as means and error bars represent standard deviations.

Resonances attributable to DHPG were observed in 600.13 MHz 1H spectra of extracts of slices incubated at 50 and 5 µm DHPG (Fig. 2). Quantification showed that DHPG was accumulated by slices to a concentration of 0.34 (0.05) µm/100 mg protein (incubation with 50 µm DHPG) and 0.07 (0.03) µm/100 mg protein (5 µm DHPG).

Figure 2.

Section of 600.13 MHz 1H spectra showing resonances assigned to DHPG. (a) Spectrum of DHPG in D2O, referenced to sodium 2,2,3,3-tetradeutero-3-trimethylsilyl propionate (TSP) at 0.00 p.p.m. (b) Extract of brain slices incubated with 50 µm DHPG. (c) Control spectrum. Spectra in (b) and (c) represented the sum of 16 transients acquired across 64 K data points, using a duty cycle of 30 s. Free induction decays were transformed using 0.5 Hz exponential multiplication.

Group I mGluR agonist, CHPG

The mGluR5 selective agonist CHPG (Doherty et al. 1997) was tested to determine whether the effects seen with DHPG were due to activity at mGluR1 or mGluR5. Incubation of cortical brain tissue slices with 5 µm CHPG generally stimulated net flux of 13C into all metabolites measured, although these increases were not individually statistically significant compared with controls (Fig. 1b). By contrast, incubation with 50 µm CHPG (Fig. 1b) resulted in potent inhibition of flux through the Krebs cycle (as indicated by decreased flux into Glu C2, Asp C2 and C3) and inhibition of glutamate/glutamine cycling (as indicated by decreased flux into Glu C4 and Ala C3. Gln C4 was not detected in any 13C spectrum of extracts incubated with 50 µm CHPG). Flux into GABA C2 was also decreased.

The total metabolic pool sizes decreased in the presence of both concentrations of CHPG, with significant falls in Glu [6.26 (0.54) µm/100 mg protein, control; 4.51 (1.70) 5 µm CHPG; 3.86 (0.91) 50 µm CHPG], GABA [2.30 (0.22) µm/100 mg protein, control; 1.55 (0.48) 5 µm CHPG; 1.09 (0.25) 50 µm CHPG], Asp [2.37 (0.37) µm/100 mg protein, control; 1.36 (0.49) 5 µm CHPG; 0.89 (022) 50 µm CHPG], Gln [1.15 (0.10) µm/100 mg protein, control; 0.63 (0.26) 5 µm CHPG; 0.20 (0.03) 50 µm CHPG] and Ala [1.25 (0.08) µm/100 mg protein, control; 0.84 (0.28) 5 µm CHPG; 0.57 (0.15) 50 µm CHPG].

No resonances attributable to CHPG were seen in 1H NMR spectra of extracts of brain cortical tissue slices incubated with CHPG [αCHδ = 5.106, H-2, H-4 multiplet δ = 6.93–6.96, H-5 multiplet δ = 7.407, acquired in 200 mm phosphate buffer, pH 7.2, chemical shifts relative to sodium 2,2,3,3-tetradeutero-3-trimethylsilyl propionate (TSP) at δ =0.00].

Group I mGluR antagonist, AIDA

The mGluR I antagonist AIDA, at 30 µm, significantly decreased net flux of 13C from [3-13C]pyruvate into Glu C2, Glu C4, GABA C2, Gln C4, Asp C2 and Asp C3, Ala C3 and lactate C3 (Fig. 1c), while AIDA at 5 µm had little significant effect on net 13C fluxes (Fig. 1c). AIDA at 30 µm also resulted in a significant (p < 0.05) decrease in the metabolite pool sizes of lactate [2.19 (0.34), control; 0.80 (0.20) µm/100 mg protein], glutamate [6.49 (0.63), control; 3.41 (0.81) µm/100 mg protein], GABA [2.25 (0.48), control; 1.06 (0.26) µm/100 mg protein], aspartate [2.64 (0.89), control; 1.06 (0.26) µm/100 mg protein] and alanine [1.08 (0.21), control; 0.43 (0.08) µm/100 mg protein].

Principal components analysis of Group I mGluR ligand effects on metabolic activity

PCA was applied to the data set consisting of 13C-labelling patterns and metabolite pool sizes (Fig. 3a). The data from ligand-free slices (control) tightly clustered, demonstrating the high reproducibility of our experimental approach. All ligands and doses, apart from 5 µm AIDA, were successfully separated from the control group across two PCs (Table 1). PC1 represented 67% of the total variance of the data and clearly separated the AIDA and CHPG groups from the DHPG group, with the high dose groups having eigenvalues of − 4.6 ± 0.93, − 4.57 ± 0.8 and 3.90 ± 1.3, respectively (p < 0.001 for difference between the AIDA and CHPG/DHPG groups). Examining the loadings for the PCA plot, the correlations represented by PC1 were relative increases in lactate C3, the difference between Asp C3 and C2, decreases in Glu C4 and C2, Gln C4, and Asp C2 and C3. The high dose DHPG group had a high score in PC1, demonstrating an increase in labelling of lactate, glutamate, aspartate and glutamine compared with the control group. CHPG and AIDA, both at 50 µm, had negative scores in PC1, showing that the labelling of lactate C3, Glu C4 and Glu C2, Gln C4, and Asp C2 and C3 were all decreased compared with the control group (p < 0.001). The AIDA and CHPG groups were separated along PC2 (p < 0.001; Table 1) while AIDA and control groups were separated by PC3 (p < 0.001; Table 1, PCA plot not shown). This demonstrated that each of the Group I mGluR-specific ligands had a distinct effect on metabolism.

Figure 3.

PCA of the 13C labelling and total metabolite concentrations from the study for DHPG and AIDA using autoscaled data. (a) Group I mGluR. The figure shows the first two principal components of the analysis, and demonstrates clear clustering of the control and high dose AIDA group. While the low dose AIDA group appeared to cluster with the control data in the middle of the PCA plot, the low dose DHPG group was more different from the control group compared with the DHPG high dose. Open squares, control; open triangles, 5 µm DHPG, black triangles, 50 µm DHPG; open diamonds, 5 µm CHPG, black diamonds, 50 µm CHPG; grey circles, 5 µm AIDA, black circles, 30 µm AIDA. (b) Group II mGluR. The figure shows the first two principal components of the analysis, and demonstrates clear clustering of the control group. PC1 separated both doses of EGLU from the other ligand exposure groups. The other ligands demonstrated an increase in PC2 for both doses, apart from the low dose of APDC. Thus, each ligand at both concentrations had a marked effect on metabolism when compared with the control group. Open squares, control data; large open diamonds, 2 µm DCG IV, large black diamonds, 20 µm DCG IV; small open diamonds, 5 µm APDC, small black diamonds, 50 µm APDC; open triangles, 5 µm APICA, black triangles, 50 µm APICA; open circles, 5 µm EGLU, black circles, 50 µm EGLU. The large ellipse represents 95% confidence interval (Hotellings score). The dashed lines define subset groupings to make the figure more readable and are not representative of any statistical certainty.

Table 1.  Variables contained in each principal component. Also listed, the amount of variance in the data which is described by each principal component
 Group 1 mGluRVarianceGroup II mGluRVariance
PC1↑ Glu C2, Glu C4, Asp C2, Asp C3, total Glu, total Gln67%↑ Glu C4, Gln C4, Ala C3, Asp C3, Asp C2,
PC2↑ Glu C2, Glu C4, total Glu, Asp, Gln ↓ lactate C3, Gln C4, Ala C3, total lactate, total Ala12%↑Asp C3 ↓ lactate C3, total lactate, total Glu, total Ala37%
PC3↑ Glu C4, Gln C4
↓ total Glu, total lactate, total Ala

Group II mGluR agonist, DCG IV

Incubation of Guinea pig cortical tissue slices with DCG IV (2 µm) resulted in significantly (p < 0.05) increased net flux of 13C from [3-13C]pyruvate into Glu C4, Gln C4, Ala C3, Asp C2 and C3, and lactate C3. By contrast, net flux into GABA C2 was significantly decreased (Fig. 4a). Similar effects were seen when 20 µm DCG IV was used (Fig. 4a). DCG IV had little or no effect on metabolite pool sizes, although 2 µm DCG IV increased the total glutamine pool size in comparison with control [control 1.21 (0.17); 2 µm DCG IV 1.61 (0.27)].

Figure 4.

Effects of Group 2 mGluR agonist and antagonists on net flux of 13C in brain cortical tissue slices incubated for 1 h with sodium [3-13C]pyruvate. (a) Slices were incubated with 2 (single hatching) or 20 (double hatching) µm DCG IV, shown compared with control (no hatching). *Significantly different to control; #significantly different to 20 µm DCG IV. (b) Slices were incubated with 5 (single hatching) or 50 (double hatching) µm APDC, shown compared with control (no hatching). *Significantly different to control; #significantly different to 50 µm APDC. (c) Slices were incubated with 5 (single hatching) or 50 (double hatching) µm APICA, shown compared with control (no hatching). *Significantly different to control; #significantly different to 50 µm APICA. (d) Slices were incubated with 5 (single hatching) or 50 (double hatching) µm EGLU, shown compared with control (no hatching). *Significantly different to control; #significantly different to 50 µm EGLU. n = 4, data are shown as means and error bars represent standard deviations.

Alone among all the mGluR ligands used in this study, 20 µm DCG IV significantly decreased tissue slice oxygen consumption [6.16 (0.61) control; 3.7 (0.36) 20 µm DCG IV]. Both concentrations of DCG IV significantly increased slice lactate production rates [11.7 (1.1) µm/min control; 29.3 (0.5) µm/min 2 µm DCG IV; 48.4 (1.5) µm/min 20 µm DCG IV].

Group II mGluR agonist, APDC

The Group II specific agonist APDC, at 5 µm (Fig. 4b), stimulated flux into the glutamate/glutamine cycle (shown by increased net flux into Glu C4, Gln C4 and Ala C3) with little effect on flux into the Krebs cycle (Glu C2, Asp C2 and C3). There was a mild stimulatory effect on glutamate/glutamine cycling at 50 µm APDC, seen through the significant increase in flux into Gln C4. Metabolite pool sizes were affected by 5 µm APDC, with significant increases in all metabolites compared with control and also, in the case of Glu, GABA, Asp and Gln, when compared with 50 µm APDC [lactate 2.29 (0.44), 4.12 (1.08); Glu 6.26 (0.54), 10.14 (2.29); GABA 2.30 (0.22), 3.93 (0.61); Asp 2.37 (0.37), 3.68 (0.55); Gln 1.15 (0.10), 2.46 (0.34); Ala 1.25 (0.08), 1.67 (0.45); values given as control and 5 µm APDC, respectively].

Group II mGluR antagonist, APICA

APICA at 5 µm produced decreased net flux into Glu C2 and C4, GABA C2, Gln C4, lactate C3 and Ala C3 (Fig. 4c) with no effect on flux into Asp C2 and C3. The total metabolite pools of lactate [2.17 (0.20) control; 1.21 (0.27) 5 µm APICA], glutamate (7.31 (0.65) control; 3.95 (0.67) 5 µm APICA], GABA [2.23 (0.26) control; 1.35 (0.18) 5 µm APICA] and alanine [1.12 (0.10) control; 0.55 (0.11) 5 µm APICA] were also significantly lower.

By contrast, incubation of tissue slices with 50 µm APICA resulted in significantly increased net flux into Glu C2 and C4, GABA C2, Lac C3, Asp C3 and Ala C3 (Fig. 4c). Only the pool size of glutamine increased significantly following exposure to 50 µm APICA [1.12 (0.26) control; 1.95 (0.38) 50 µm APICA].

Group II mGluR antagonist, EGLU

Incubation of Guinea pig cortical tissue slices with 5 or 50 µm EGLU resulted in significant increases in net flux of 13C from [3-13C]pyruvate into Glu C2, Glu C4, GABA C2, and Asp C2 and C3. Net flux into Ala C3 and lactate C3 was decreased (Fig. 4d) and there was no significant effect on net flux into Gln C4. The ratio of Asp C3 to C2, which, in the absence of carbon fixing by pyruvate carboxylase should be equal, was increased by EGLU, indicating increased anaplerotic input into the Krebs cycle. Incubation with 5 µm and 50 µm EGLU resulted in significant increases in the metabolite pool sizes of Asp [2.30 (0.47) control; 6.11 (1.05) 5 µm EGLU; 4.97 (1.20) 50 µm EGLU] and Gln [1.11 (0.20) control; 3.22 (0.65) 5 µm EGLU; 3.51 (0.78) 50 µm EGLU] and significant decreases in the pool size of Ala [1.06 (0.13) control; 0.49 (0.17) 5 µm EGLU; 0.73 (0.17) 50 µm EGLU] and lactate [2.27 (0.43) control; 1.39 (0.19) 5 µm EGLU; 1.79 (0.37) 50 µm EGLU].

In contrast to the effect shown with [3-13C]pyruvate as substrate, incubation of brain cortical tissue slices with 5 mm[1-13C]glucose, which preferentially labels different metabolic compartments to pyruvate (Zwingmann et al. 2001; Rae et al. 2003), showed a different metabolic profile (Fig. 5). Although incorporation of 13C label into Glu C4, Gln C4 and Asp C3 was significantly increased by 50 µm EGLU, the relative increase in net flux was much less than that demonstrated with pyruvate (Fig. 4d), indicating that the stimulatory effects of EGLU on the Krebs cycle were acting primarily on the metabolic compartment into which pyruvate is accumulated. In the presence of [1-13C]glucose, the metabolite pool size of glutamate was significantly increased by 50 µm EGLU, while that of GABA was significantly increased by 5 µm EGLU.

Figure 5.

Effects of Group 2 mGluR antagonist EGLU on net flux of 13C in brain cortical tissue slices incubated for 90 min with d-[1-13C]glucose. Slices were incubated with 5 (single hatching) or 50 (double hatching) µm EGLU, shown compared with control (no hatching). *Significantly different to control; #significantly different to 50 µm EGLU. n = 4, data are shown as means and error bars represent standard deviations.

PCA of Group II mGluR ligand effects on metabolic activity

Three PCs described 94% of the variation in the data set (Fig. 3b). The first two PCs readily separated the four ligand groups, with PC1 classifying both low and high dose EGLU groups (p < 0.001, EGLU vs. control) and PC2 separating the other groups (Table 1). EGLU caused relative increases in Glu C2 and C4, GABA C2, Asp C2 and C3 (and, to a lesser extent, the difference in labelling between Asp C2 and C3), and the total glutamate, glutamine and aspartate concentrations, and decreases in lactate C3, Ala C3, and total lactate and alanine. The other high dose groups were separated by an increase in the score of PC2, caused by relative increases in lactate C3, Ala C3 and Gln C4 labelling, and increases in total lactate, alanine and GABA pool sizes. The low dose APDC group was the only group to have a decreased score in PC2, caused by a relative increase in the difference between Asp C2 and C3 labelling. At the higher dose this effect disappeared, with the 50 µm APDC dose producing a metabolic profile closer to that demonstrated by 2 µm DCG IV than by 5 µm APDC.

Intracellular [K+]

There was no significant effect of any of the ligands tested on intracellular [K+]. All values were within the normal range (data not shown).


Group I mGluR

While antagonists at Group I mGluR are consistently neuroprotective (Nicoletti et al. 1999), agonists have shown varied results. This can be explained by variations in the exposure times, concentrations of ligands and ambient glutamate concentration when using a variety of experimental models. In addition, adjacent astrocytes may vary in their ability to transport (i.e. remove) glutamate and/or to release neuroprotective or neurotoxic factors (Nicoletti et al. 1999). Such factors are difficult to control in vivo while in cultured cells, the conditions may not be representative of those found in the living brain. Brain slices preserve many features of the brain tissue in vivo, but the experimental design offers much better control of ligand concentrations and more detailed analysis of their effects at the level of brain metabolism. How do the findings of this work fit with observations in vivo and in cultured cells?

The Group I mGluR antagonist AIDA produced a general inhibitory effect on metabolism, with flux into Glu C2, Asp C2 and C3 (labelled on first turn of the Krebs cycle), and metabolite pool sizes both significantly decreased, while internal [K+] was maintained, indicating that the brain slices were still viable despite the reduction in metabolism. Such a metabolic response is consistent with decreased activity, which might be expected to be neuroprotective (Ichord et al. 2001). AIDA has been shown to reduce neuronal death both in vivo and in vitro following ischaemia (Pellegrini-Giampietro et al. 1999). This action may be mediated by enhancement of GABA release and hence, increased inhibition of neurons (Battaglia et al. 2001; Cozzi et al. 2002; Saransaari and Oja 2003; Smolders et al. 2004).

DHPG was the first agonist shown to be Group I mGluR selective (Schoepp et al. 1994). It is active at both mGluR1 and mGluR5 (Wisniewski and Car 2002). DHPG has been shown to result in release of glutamate in spinal motor neurons, and to increase hippocampal CA1 neuronal excitability. There are reports of DHPG both causing neuronal damage (Ong and Balcar 1997) and protecting against toxicity induced by low concentrations of glutamate (Sagara and Schubert 1998). DHPG has been suggested to induce long-term depression by decreasing the numbers of NMDA and AMPA receptors present in the membrane (Snyder et al. 2001). CHPG has been shown to be neuroprotective against focal cerebral ischaemia (Bao et al. 2001), perhaps through the anti-apoptotic activity of mGluR5.

In the present study, DHPG produced a stimulatory effect that was greater at the 5 µm than at the 50 µm concentration. DHPG does not interact with GluR other than mGluR I and is not handled by glutamate transport (Lieb et al. 2000; Fazal et al. 2003). That does not mean, however, that DHPG does not stimulate/inhibit non-glutamate receptors, or that it is not accumulated by non-glutamate transport systems. Additional, non-glutamatergic effects cannot therefore be ruled out. DHPG has previously been shown to produce changes in GABAergic activity that are dose-dependent (Drew and Vaughan 2004). Also, DHPG has recently been shown to modulate the expression of the glutamate transporters (Aronica et al. 2003), although it is not known whether this leads to changes in glutamate transport rates.

In contrast to DHPG, CHPG, particularly at 50 µm, had a largely inhibitory effect on both metabolism and on net flux into glutamate/glutamine cycle metabolites (Fig. 3). DHPG is a more potent agonist than CHPG (EC50 in the range of 2–20 for DHPG vs. 750 µm for CHPG), but CHPG is known to have a marked preference for mGluR5, being practically inactive at mGluR1 (Schoepp et al. 1999).

The actions of AIDA, a selective mGluR1a antagonist (Moroni et al. 1997), are consistent with reported neuroprotective roles for this ligand (Strasser et al. 1998; Bruno et al. 2001), acting not only to decrease metabolic activity (net flux through the Krebs cycle) but also to decrease glutamate/glutamine cycle activity. Neither DHPG nor CHPG act unequivocally in the direction opposing the actions of AIDA; a simple mGluR1 agonist that could be used to test whether a specific activation of mGluR1 would have an effect on metabolism exactly opposite to that produced by AIDA is not yet available.

Group II mGluR

DCG IV produces agonist effects at Group II mGluR, but also at NMDA receptors (Wilsch et al. 1994; Uyama et al. 1997), as well as displaying antagonist activity at Groups III (20–40 µm) and I (> 100 µm) (Brabet et al. 1998). The effects that DCG IV produces are thus highly dependent on the dose and receptor constituents of the tissue against which it is employed (Venero et al. 2002). The net flux profile generated by 2 µm DCG IV (Fig. 4a) is similar to that produced by 5 µm APDC (Fig. 4b), but the two ligands at this concentration can be readily differentiated by PC2 (Fig. 5), in this case influenced mostly by significant differences in the effects of each ligand on metabolite pool sizes.

At 20 µm, DCG IV is clearly having effects at receptors other than Group II mGluR, as the dose significantly decreased tissue slice oxygen consumption and increased lactate production rates around fourfold. This could be an early indication of NMDA receptor-mediated neurotoxicity that could be triggered off at higher concentrations of DCG IV (Kwak et al. 1996; Mutel et al. 1998). The lack of significant effect on tissue [K+], indicating that the slices were still viable, is not incompatible with such an interpretation because the actual cell death mediated by NMDA receptors may be delayed by many hours (Strijbos et al. 1996).

Broadly, in the literature, agonists at Group II mGluR have been reported to decrease neuronal activity (Cartmell and Schoepp 2000). In keeping with this, neither 2 µm DCG IV nor APDC produced increased flux into the Krebs cycle. This is indicated by a lack of increased net flux into Glu C2, which is labelled on the second turn on the Krebs cycle. Net flux into Glu C4, Gln C4 and Ala C3 is increased, suggesting that net flux into the glutamate/glutamine cycle is increased by both concentrations of DCG IV and 5 µm APDC. Finally, net flux into Asp C2 and C3 is increased, as is flux into Lac C3, again suggestive of decreased Krebs cycle rates, with pyruvate being converted instead to lactate, and oxaloacetate being converted to aspartate.

EGLU showed strong stimulation of net flux into the Krebs cycle, increased metabolite pool sizes and also increased anaplerotic activity in glial cells. EGLU acts similarly potently on both mGluR2 and mGluR3; the increased glial pyruvate carboxylase activity may be due to modulation of mGluR3 located on glial cells (Winder and Conn 1996). EGLU caused the opposite effect on flux into the glutamate/glutamine cycle to DCG IV and APDC, with no change or significantly decreased flux into Gln C4 and Ala C3. However, when [1-13C]glucose was used as a substrate it became apparent that the increased Krebs cycle activity seen in Fig. 4(d) was primarily in a compartment accessible to high concentrations of pyruvate. At the high concentration (2 mm) of [3-13C] pyruvate used, 13C enters compartments it would not normally enter in significant quantities (Zwingmann et al. 2001; Rae et al. 2003), such as when it is supplied as [1-13C]glucose. Given also the increased anaplerotic activity, as seen through increased labelling of Asp C3 compared with Asp C2, the most plausible explanation is that EGLU is acting primarily on an astrocytic cell body compartment (Rae et al. 2003). EGLU has previously been reported to have activity in vivo which is not consistent with mGluR activity (Pliss et al. 2000) and this is reflected in the separation of EGLU effects by PC1 (Fig. 3b).

In contrast to EGLU, the metabolic effects produced by APICA were highly dose-dependent. The IC50 for APICA at mGluR2 has been measured at 30 µm, and APICA has been reported to have inverse agonist activity at Group II mGluR, as does EGLU (Ma et al. 1997). In rat spinal cord, APICA has been reported to show antagonistic activity at Group III mGluR, unlike EGLU which is not active, although neither EGLU nor APICA have been tested thoroughly against all mGluR clones (Schoepp et al. 1999). The effects of the two Group II antagonists were clearly differentiated by PC1 (Fig. 3b), indicating that despite similar potencies at mGluR2 and mGluR3, their activities at other sites are diverse.

Experimental approach and data analysis

PCA provides a novel feature in the evaluation of complex data such as that obtained by analysing metabolism using 13C and 1H NMR spectroscopy. The results from this study suggest that the present experimental approach is highly reproducible, as demonstrated by the tight clusterings of the treatment groups, and the multiparametric nature of the observations makes it possible to uncover subtle differences among the data sets (the metabolic ‘profiles’) rather than relying on a univariate description of the data. This can yield additional insights into the mechanisms associated with individual receptors and/or their ligands, and add to the power of the approach.

PCA of data showed that Group I mGluR ligands produced effects mostly characterized by changes in labelling to glutamate and aspartate, and changes in the total pool size of glutamate and glutamine (PC1; Table 1). This can be explained by these ligands affecting the flux through the Krebs cycle rather than flux through the glutamate/glutamine cycle. Group II mGluR ligands, on the other hand, had a more potent effect on flux into the glutamate/glutamine cycle (Fig. 4; Table 1) as well as general metabolic flux. Further, the effects of individual ligands on metabolism could easily be characterized and separated using PCA. This work illustrates that metabolic profile analysis is a sensitive method for differentiating mGluR ligand effects and for detecting metabolic consequences of mGluR perturbation. The distinct differences between individual ligands active on the same receptor indicates that caution must be employed when using these ligands as they are plainly not interchangeable.


This work was supported by The University of Sydney, the Australian Health Management Fund, and by the Australian National Health and Medical Research Council (Fellowship to CR). The authors thank Mr James McQuillan for expert research assistance and Mr William Lowe for technical support.