Brain metabolism of exogenous pyruvate


Address correspondence and reprint requests to Bjørnar Hassel, Norwegian Defence Research Establishment, PO Box 25, N-2027 Kjeller, Norway. E-mail:


Pyruvate given in large doses may be neuroprotective in stroke, but it is not known to what degree the brain metabolizes pyruvate. Intravenous injection of [3-13C]pyruvate led to dose-dependent labelling of cerebral metabolites so that at 5 min after injection of 18 mmoles [3-13C]pyruvate/kg (2 g sodium pyruvate/kg), approximately 20% of brain glutamate and GABA were labelled, as could be detected by 13C nuclear magnetic resonance spectrometry ex vivo. Pyruvate, 9 mmoles/kg, was equivalent to glucose, 9 mmoles/kg, as a substrate for cerebral tricarboxylic acid (TCA) cycle activity. Inhibition of the glial TCA cycle with fluoroacetate did not affect formation of [4-13C]glutamate or [2-13C]GABA from [3-13C]pyruvate, but reduced formation of [4-13C]glutamine by 50%, indicating predominantly neuronal metabolism of exogenous pyruvate. Extensive formation of [3-13C]lactate from [2-13C]pyruvate demonstrated reversible carboxylation of pyruvate to malate and equilibration with fumarate, presumably in neurones, but anaplerotic formation of TCA cycle intermediates from exogenous pyruvate could not be detected. Too rapid injection of large amounts of pyruvate led to seizure activity, respiratory arrest and death. We conclude that exogenous pyruvate is an excellent energy substrate for neurones in vivo, but that care must be taken to avoid the seizure-inducing effect of pyruvate given in large doses.

Abbreviations used







nuclear magnetic resonance



TCA cycle

tricarboxylic acid cycle

Pyruvate has gained interest as a neuroprotective agent. Administration of pyruvate has been shown to ameliorate nerve cell damage in rat models of reversible cerebral and retinal ischemia (Lee et al. 2001; Yoo et al. 2004), and in in vitro models of excitotoxicity (Maus et al. 1999; Gramsbergen et al. 2000). However, in a model of irreversible cerebral ischemia, pyruvate had an adverse effect on neuronal survival (Gonzalez-Falcon et al. 2003). A neuroprotective effect of pyruvate would be expected to be linked to its role as an energy substrate but in one study on the neurotoxicity of hydrogen peroxide in vitro, pyruvate acted as a scavenger of H2O2 rather than as an energy substrate (Desagher et al. 1997).

There has been a notion that pyruvate would not enter the brain to a degree that would allow it to be of quantitative importance in cerebral energy metabolism (Pardridge and Oldendorf 1977; Cremer et al. 1979). This view was based on studies in anaesthetized animals, and subsequent studies in awake animals showed that the passage of pyruvate across the blood–brain barrier may be substantial (Miller and Oldendorf 1986). In a study on reversible ischemia in rat brain, as much as 1 g of sodium pyruvate was given per kg of bodyweight, and this dosage was superior to a dose of 0.5 g pyruvate/kg bodyweight with respect to neuroprotective effect (Lee et al. 2001). It is open to question whether such doses differ with respect to the amount of pyruvate that enters the brain, an aspect of pyruvate therapy that remains to be elucidated. Furthermore, in experiments designed to evaluate the neuroprotective effect of pyruvate in vivo, the pyruvate was given intraperitoneally (Lee et al. 2001; Gonzalez-Falcon et al. 2003; Yoo et al. 2004). This route of administration leads to more hepatic metabolism of the injected pyruvate than the intravenous (i.v.) route, and probably entails substantial conversion of pyruvate into glucose. In a clinical setting the route of administration would be i.v.

Against this background, information on the cerebral handling of large amounts of pyruvate after i.v. injection is warranted. We have studied the cerebral metabolism of 13C-labelled pyruvate given i.v. in large amounts to awake mice with the use of ex vivo nuclear magnetic resonance (NMR) spectroscopy. The solutions of sodium pyruvate that were injected were hypertonic (up to 1.8 m) to avoid volume overload. The cerebral metabolism of pyruvate was compared with that of acetate, an energy substrate that is metabolized by glial cells (Van den Berg et al. 1969; Cerdan et al. 1990; Hassel et al. 1997), and with glucose, the physiological energy substrate of all brain cells. To study the compartmentation of pyruvate metabolism in the brain we used fluoroacetate (F-acetate), which, after conversion into fluorocitrate, causes irreversible inhibition of aconitase (Peters 1957) of the glial TCA cycle (Muir et al. 1986; Hassel et al. 1997). We have previously shown that F-acetate does not interfere with transport of acetate into the brain, or with the initial metabolism of acetate into citrate (Hassel et al. 1997).

Experimental procedures

Male and female NMRI mice, 20–25 g, were from M & B, Ry (DK). Animal handling adhered strictly to national and institutional guidelines for animal research. [3-13C]Pyruvate, [2-13C]pyruvate, [1-13C]glucose and [1,2-13C]acetate, all compounds with 99%13C enrichment, were from Isotec (Miamisburg, OH, USA). The compounds were analysed by 13C NMR spectroscopy, and their purity was confirmed.

Dose-labelling and time-labelling studies

Mice that were fasted overnight (with free access to water) received sodium [3-13C]pyruvate in doses of 2.25 mmoles/kg, 4.5 mmoles/kg, 9 mmoles/kg or 18 mmoles/kg, corresponding to 0.25 g/kg, 0.5 g/kg, 1 g/kg or 2 g/kg, respectively, a dose range that covers the doses used in studies on the effect of sodium pyruvate in cerebral ischemia (Lee et al. 2001; Gonzalez-Falcon et al. 2003). The injection volume was 10 µL/g bodyweight, and was injected into a tail vein over 30 s. When the osmolarity of the injection solution was below 300 mOsm, it was corrected with sodium chloride. The pH was maintained at 7.

To study the dose-labelling relationship for [1-13C]glucose, mice received 1.125 mmoles/kg, 2.25 mmoles/kg, 4.5 mmoles/kg or 9 mmoles/kg. In terms of pyruvate equivalents, these doses corresponded to the above doses of [3-13C]pyruvate, since one molecule of glucose yields two molecules of pyruvate. It should be noted that [1-13C]glucose yields one molecule of [3-13C]pyruvate and one molecule of unlabelled pyruvate.

To study the 13C-labelling of cerebral amino acids from [3-13C]pyruvate with time, animals received [3-13C]pyruvate, 9 mmoles/kg, and were killed at 2, 5, 10 or 30 min.

To determine whether [3-13C]pyruvate as such would reach the brain, mice were anaesthetized with (per kg bodyweight) fentanyl citrate (0.2 mg), fluanisone (10 mg) and midazolam (5 mg). They then received heparin, 10 U subcutaneously (s.c.), to prevent blood clotting and allow bleeding, and at 10 min they received [3-13C]pyruvate, 9 mmoles/kg, in a tail vein. Some of the animals were bled from an incision in the tail at various time points after injection of [3-13C]pyruvate. Blood was collected on ice, the volume was measured and 1 mL ice-cold perchloric acid, 3.5% (v/v), was added. Some of the anaesthetized animals had the skin over the scalp removed, and liquid N2 was poured onto the skull through a plastic funnel 1 min after injection of [3-13C]pyruvate.

Metabolism of 13C-labelled pyruvate and acetate during F-acetate treatment: liver and brain metabolism of [2-13C]pyruvate

Mice that were fasted overnight were pre-treated with sodium F-acetate, 100 mg/kg s.c. (1 mmol/kg), dissolved at 100 mm in 50 mm NaCl, which gave an injection volume of 10 µL/g bodyweight. This dose had previously been shown not to affect the brain content of ATP (Goldberg et al. 1966). Control mice received an injection of NaCl, 150 mM. At 15 min the animals received [3-13C]pyruvate or [1,2-13C]acetate, 9 mmoles/kg, and were killed after another 10 min.

Mice that were either fed or fasted overnight received sodium [2-13C]pyruvate, 9 mmoles/kg, and were killed at 5 min. Brains and livers were sampled.

Preparation of tissue extracts and analysis by 13C NMR spectroscopy

Awake animals were killed by cervical dislocation and decapitation, and the heads were immediately dropped in liquid N2 (approximately 1 s delay from death). Blood was collected from the severed vessels, and livers were rapidly removed and dropped in liquid N2. Brains were removed in the frozen state and homogenized in 4 mL perchloric acid, 3.5%. Livers were treated similarly. Protein was removed by centrifugation, supernatant fluids were neutralized with KOH, and the precipitating KClO4 was removed by centrifugation. Extracts were lyophilized to dryness and dissolved in 500 µL D2O with dioxane, 0.1%, as internal concentration standard. Inverse-gated 13C NMR spectroscopy was performed on a Bruker Avance DRX 500 (Zürich, Switzerland) as described previously (Hassel and Bråthe 2000a; Nguyen et al. 2003), with 10 240 scans per sample and a pulse angle of 30°. The spectral width was 39.6 kHz, and 64 000 data points were used.

The percent 13C enrichment was calculated after subtraction of naturally abundant 13C as previously described (Badar-Goffer et al. 1990). The pool sizes of amino acids were analysed by HPLC (HP 1100) after pre-column derivatization with o-phthaldialdehyde, and serum and tissue levels of lactate and glucose were determined by reflectance spectrophotometry (Kodak DT60, Rochester, NY, USA), as previously described (Hassel et al. 1997). The level of [3-13C]pyruvate in serum was quantified from 13C NMR spectra.

Electroencephalography (EEG)

Mice were anaesthetized with (per kg bodyweight) fentanyl citrate (0.2 mg), fluanisone (10 mg) and midazolam (5 mg). The skull was exposed and a hole was drilled 2 mm posterior to bregma, 1 mm to the right of the midline. As electrode, a plastic-coated silver thread (Johnson Matthey Metals, London, UK), 0.3 mm outer diameter and 10 cm long, was inserted into the right hippocampus 2 mm deep of dura. Another hole was drilled 0.6 mm to the left of bregma, and a silver thread was inserted with its tip immediately above dura. The electrodes were secured with dental cement, the skin incision was sutured and the animals were caged individually. EEG registration was performed the day after surgery. The plastic covering was scraped off the tips of the electrodes, which were connected to a Grass polygraph model 7 (Quincy, MA, USA), and signals were detected with a differential amplifier. The animals moved freely on a table but were held by the tail (with their feet on the table) during injection of unlabelled sodium pyruvate, 9 mmoles/kg (0.9 m), or physiological saline into a tail vein.

Data presentation and statistics

Data are presented as means ± SEM values. Differences between groups were analysed by Student's t-test, paired or unpaired as appropriate.


13C Enrichment of metabolites in serum after injection of [3-13C]pyruvate

Serum concentrations of [3-13C]pyruvate and [3-13C]lactate were investigated in anaesthetized mice, which could be bled without inducing hypoxia. [3-13C]Pyruvate was the main 13C-labelled compound in serum 1 min after injection of [3-13C]pyruvate (Fig. 1). A little [3-13C]lactate and [3-13C]alanine were seen, but no [13C]glucose. In brains that were funnel frozen in situ 1 min after injection of [3-13C]pyruvate, the amount of [3-13C]lactate exceeded that of [3-13C]pyruvate (Fig. 1), suggesting that pyruvate was largely reduced after entering the brain. At 5 min the level of [3-13C]lactate in serum exceeded that of [3-13C]pyruvate (Fig. 1).

Figure 1.

13C NMR spectra of serum and brain after i.v. injection of [3-13C]pyruvate. The animals were given 9 mmoles [3-13C]pyruvate/kg under anaesthesia. At 1 or 5 (insert) min, an incision was made in the tail, and blood was collected for 20 s. Insert: Brain spectrum from a mouse that received [3-13C]pyruvate under anaesthesia. The brain was frozen in situ with liquid N2 while the animal was breathing. Note that at 1 min the pyruvate peak is larger than the lactate peak in serum, while in the brain the reverse is true. Ala, alanine; D, dioxane (internal standard); Lac, lactate; Pyr, pyruvate (keto and enol forms).

In spite of the high concentration of the injected [3-13C]pyruvate, serum collected from anaesthetized mice between 10 and 30 s after injection had a concentration of [3-13C]pyruvate of only 15 ± 1 mm, indicating rapid efflux of pyruvate from the circulation. At 1, 2 and 5 min after injection, the serum concentration of [3-13C]pyruvate was 6.7 ± 0.8 mm, 3.2 ± 0.3 mm and 1.3 ± 0.2 mm, respectively.

At 2 min after injection of [3-13C]pyruvate, serum lactate was 2.1 ± 0.4 mm with a 13C enrichment of the C3 of 80 ± 8%. At 5 min, serum lactate was 8 ± 1 mm, with a 13C enrichment of the C3 of 27 ± 2%. At 30 min, serum lactate was 3.1 ± 0.3 mm, with a 13C enrichment of the C3 of 11 ± 1%. Therefore, [3-13C]pyruvate is probably the main substrate to enter the brain during the first few minutes after injection; with time, [3-13C]lactate becomes a more important contribution.

At 2 min after injection of [3-13C]pyruvate, serum glucose was 5.6 ± 0.4 mm, with no observable 13C enrichment. At 5 min, serum glucose was 8.4 ± 0.6 mm, with an enrichment of the C1 position of 5.1 ± 0.8%. At 30 min, the serum concentration of glucose was 9.0 ± 0.7 mm, with an enrichment of the C1 position of 12 ± 1%; this position (together with glucose C6, which will be labelled to the same degree) corresponds to the C3 of pyruvate.

Cerebral metabolism of [3-13C]pyruvate in awake animals

Injection of [3-13C]pyruvate led to robust labelling of the C4 of glutamate and glutamine, the C2 of GABA, and the C3 of alanine and lactate in the brain (Fig. 2). Similarly, injection of [2-13C]pyruvate led to labelling of glutamate C5, GABA C1, and the C2 of alanine and lactate (Fig. 2). Glutamine C5 could not be reliably distinguished from aspartate C1.

Figure 2.

13C NMR spectra of brain after injection of [3-13C]pyruvate or [2-13C]pyruvate. Awake mice received 9 mmoles/kg [3-13C]pyruvate (upper panel) or [2-13C]pyruvate (lower panel) i.v. Survival time was 5 min. [3-13C]Pyruvate labels the C3 of lactate (Lac) and alanine (Ala), the C4 of glutamate (Glu) and glutamine (Gln), and GABA C2, but not glutamate C5 or the C1 of GABA or lactate (insert). [2-13C]Pyruvate labels lactate and alanine C2, and glutamate C5, GABA C1 and lactate C1 (insert). D is dioxane (internal standard). The peak corresponding to Glu C4 in lower spectrum represents naturally abundant 13C-glutamate.

The 13C labelling of cerebral metabolites increased with increasing doses of [3-13C]pyruvate, so that at 5 min after injection of the maximum dose of sodium [3-13C]pyruvate, 18 mmoles/kg or 2 g/kg, approximately 20% of brain glutamate and GABA was labelled in the C4 and C2 positions, respectively (Fig. 3a). Male and female mice were compared with respect to 13C labelling of cerebral metabolites at 5 min after injection of [3-13C]pyruvate, 9 mmoles/kg, and no differences were detected (data for male mice not shown).

Figure 3.

Dose-labelling relationships for [3-13C]pyruvate and [1-13C]glucose. (a) Awake mice received i.v. injections of [3-13C]pyruvate, 2.25, 4.5, 9 and 18 mmoles/kg bodyweight; these doses correspond to 0.25, 0.5, 1 and 2 g sodium [3-13C]pyruvate/kg bodyweight, respectively. (b) Awake mice received i.v. injections of [1-13C]glucose at various doses. Survival time was 5 min for all groups. Data are percent 13C enrichment, means ± SEM values, n = 5 per group.

The 13C labelling of cerebral amino acids increased from 2 to 5 min after injection of [3-13C]pyruvate; it remained fairly stable for another 5 min (Fig. 4) except for glutamine C4, the 13C enrichment of which continued to increase. From 10 to 30 min after injection, the 13C enrichments tended to decrease (Fig. 4).

Figure 4.

Time curve for the 13C-labelling of glutamate C4, GABA C2, glutamine C4 and alanine C3 from [3-13C]pyruvate. Awake mice received [3-13C]pyruvate, 9 mmoles/kg, i.v., and were killed at 2, 5, 10 or 30 min. Data are percent 13C enrichment, means ± SEM values, n = 5 per time-point.

Comparison of exogenous glucose and pyruvate as cerebral energy substrates

Injection of [1-13C]glucose led to dose-dependent labelling of cerebral amino acids (Fig. 3b). At the lower doses, e.g. 2.25 mmoles/kg, [1-13C]glucose gave a higher 13C enrichment of brain glutamate C4 than did [3-13C]pyruvate (Fig. 3a). However, at the higher doses, e.g. 9 mmoles/kg, the two substrates gave the same 13C enrichment. Because [1-13C]glucose gives rise to two molecules of pyruvate only one of which is 13C-labelled, it may be more correct to compare the 13C enrichment achieved with 9 mmoles [1-13C]glucose/kg to that achieved with 18 mmoles [3-13C]pyruvate/kg. [3-13C]Pyruvate, 18 mmoles/kg, gave a 13C enrichment of glutamate C4 of 23 ± 2%, whereas [1-13C]glucose, 9 mmoles/kg, gave a 13C enrichment of glutamate C4 of 13 ± 2%. The latter value should be multiplied by two to account for the unlabelled pyruvate originating from [1-13C]glucose, which gives 26 ± 4%, a value similar to that achieved with [3-13C]pyruvate, 18 mmoles/kg.

Behavioural and electroencephalographic responses to injection of pyruvate

Injection of [3-13C]pyruvate had to be performed slowly, over 30 s. Too rapid injection of the 1 and 2 m solutions of [3-13C]pyruvate would lead to symptoms ranging from hind limb paralysis, which disappeared within 10 s, to generalized clonic convulsions, respiratory arrest and death within a few seconds after the injection. Irrespective of these reactions, animals that received the 1 and 2 m solutions of [3-13C]pyruvate became passive and did not interact with their littermates for about 1–2 min, after which behaviour normalized. These reactions were not seen with the 0.25 and 0.5 m solutions of [3-13C]pyruvate, or when [1-13C]glucose or [1,2-13C]acetate were given.

Hippocampal electroencephalography (EEG) recordings during injection of unlabelled sodium pyruvate, 9 mmoles/kg, over approximately 5 s showed transient synchronization, which, in six out of eight animals, was accompanied by clonic convulsions and respiratory arrest (Fig. 5). In most cases the animals began breathing again spontaneously. In two animals, pyruvate administration caused seizure activity in the EEG recordings without accompanying convulsions (Fig. 5). Injection of physiological saline did not cause EEG alterations (not shown).

Figure 5.

Effect of rapid pyruvate injection on hippocampal EEG. Mice had electrodes in the right hippocampus. Sodium pyruvate, 9 mmoles/kg (0.9 m solution), was given i.v. into a tail vein over approximately 5 s. The upper panel shows EEG synchronization accompanying convulsions, followed by respiratory arrest, which led to EEG activity being diminished. The lower panel shows movement artefacts after pyruvate-induced clonic seizures, followed by transient seizure activity in a non-convulsive phase before normalization of the EEG.

Effects of F-acetate on metabolism of 13C-labelled acetate and pyruvate

In order to study which type of brain cell metabolizes exogenous pyruvate, we pre-treated mice with F-acetate, which blocks the TCA cycle of glial cells (Muir et al. 1986; Hassel et al. 1997), and compared the effect of F-acetate on pyruvate metabolism with its effect on acetate metabolism.

Injection of [1,2-13C]acetate, 9 mmoles/kg, led to a higher percent 13C enrichment of glutamine than of glutamate (Table 1), as would be expected from a substrate that selectively enters glial cells (Van den Berg et al. 1969; Cerdan et al. 1990; Hassel et al. 1997) since glutamine synthetase in the brain is localized in glial cells, not in neurones (Martinez-Hernandez et al. 1977; Tansey et al. 1991). Pre-treatment with sodium F-acetate, 100 mg/kg (1 mmole/kg), virtually abolished labelling from [1,2-13C]acetate, and only a faint labelling of glutamate could be seen (Table 1). The behavioural response to F-acetate was motor inactivity; 5 min after injection of F-acetate the animals lay down and could barely be stimulated to walk about the cage.

Table 1.  Effect of the gliotoxin F-acetate on the 13C labelling of brain amino acids from [1,2-13C]acetate and [3-13C]pyruvate. Awake mice received sodium F-acetate, 1 mmole/kg or saline s.c. At 15 min they received [1,2-13C]acetate or [3-13C]pyruvate, 9 mmoles/kg, and were killed after another 10 min. Data are percent 13C enrichment, means ± SEM values, n = 4–5 per group
 Glu C4GABA C2Gln C4
  • *

    Different from control value, p = 0.007; n.d. not detectable.

[1,2-13C] Acetate
 Control3.8 ± 0.74.2 ± 0.811.1 ± 0.8
 F-ac0.5 ± 0.1n.d.n.d.
 Control12 ± 112 ± 24.4 ± 0.5
 F-ac13 ± 112 ± 12.1 ± 0.5*

In contrast to the findings with [1,2-13C]acetate, F-acetate treatment did not affect labelling of glutamate or GABA from [3-13C]pyruvate (Table 1). This strongly suggests the predominantly neuronal metabolism of exogenous pyruvate in the brain. F-acetate treatment did, however, reduce labelling of glutamine by about 50% (Table 1). F-acetate treatment reduced the level of aspartate by 16%, from 30 ± 1 mmoles/mg protein to 25 ± 1 mmoles/mg protein (p = 0.02). The levels of glutamate (95 ± 6 mmoles/mg protein), GABA (16 ± 1 nmoles/mg protein), glutamine (40 ± 4 mmoles/mg protein) and alanine (5.8 ± 0.4 mmoles/mg protein) were not affected by F-acetate.

Reversible carboxylation of pyruvate to malate in the brain

At 5 min after injection of [3-13C]pyruvate, considerable amounts of C2-labelled lactate were present in the brain extracts (Fig. 6). This was the case even though very little C2-labelled lactate was seen in the serum at this time-point (Fig. 1), indicating formation of C2-labelled lactate in the brain itself. This finding may reflect two different metabolic pathways. First, [3-13C]pyruvate could be carboxylated to [3-13C]malate, which could equilibrate with the symmetrical fumarate leading to formation of [2-13C]malate; this [2-13C]malate could be decarboxylated to [2-13C]pyruvate and [2-13C]lactate. Alternatively, [3-13C]pyruvate could be metabolized through pyruvate dehydrogenase and the TCA cycle with formation of [2-13C]malate, which could be decarboxylated to [2-13C]pyruvate and [2-13C]lactate. In order to address this problem we injected [2-13C]pyruvate i.v. into awake mice to determine whether [3-13C]lactate would be formed in the brain. This isotopomer can only be formed through reversible carboxylation (Fig. 7a); it cannot be formed through pyruvate dehydrogenase since this pathway would yield only [1-13C] and [4-13C]malate (Fig. 7b), which, upon decarboxylation, would yield [1-13C]pyruvate (and [1-13C]lactate) and 13CO2, respectively. At 5 min after injection of [2-13C]pyruvate, 9 mmoles/kg, a conspicuous peak of [3-13C]lactate could be seen in the brain extracts (Fig. 6). The 13C enrichment of lactate C2, C3 and C1 was approximately 50%, 10% and 2.5%, respectively (Table 2), indicating extensive reversible carboxylation of pyruvate to malate and equilibration with fumarate. Virtually no lactate labelled in the C2 position was seen when [1-13C]glucose was administered, although the C3 position was strongly labelled (Fig. 6).

Figure 6.

Formation of [2-13C]lactate from [3-13C]pyruvate (lower spectrum), [3-13C]lactate from [2-13C]pyruvate (middle spectrum) and lack of formation of [2-13C]lactate from [1-13C]glucose (upper spectrum; glucose peaks are truncated). Mice received 9 mmoles/kg of either [3-13C]pyruvate, [2-13C]pyruvate or [1-13C]glucose and were killed after 5 min. Glc C1, the β- and α-enantiomers of glucose; Lac C2 and Lac C3, the 2nd and 3rd carbons of lactate, respectively.

Figure 7.

Metabolism of [2-13C]pyruvate through carboxylation and pyruvate dehydrogenase. (a) Formation of [3-13C]pyruvate from [2-13C]pyruvate through reversible carboxylation to [2-13C]malate, equilibration with fumarate and formation of [3-13C]malate, which is decarboxylated to [3-13C]pyruvate, from which [3-13C]lactate is formed (seen in both brain and liver). (b) Metabolism of [2-13C]pyruvate through pyruvate dehydrogenase, leading to formation of [5-13C]glutamate (seen in both brain and liver), and to [1-13C]- and [4-13C]malate, which, after decarboxylation, yield [1-13C]pyruvate and 13CO2, respectively. (c) Carboxylation of [2-13C]pyruvate leads to anaplerotic formation of [3-13C]α-ketoglutarate and -glutamate (seen only in liver). Ac∼CoA, acetyl∼CoA; Fum, fumarate; Glu, glutamate; α-kg, α-ketoglutarate; Mal, malate; Pyr, pyruvate.

Table 2.  Effect of fasting on 13C labelling of individual lactate and glutamate carbons in brain and liver. Fasted or fed animals received an i.v. injection of [2-13C]pyruvate, 9 mmoles/kg, and were killed at 5 min. Data are percentage of 13C enrichment (seven left data columns), the glutamate C3/C513C enrichment ratio, and nmoles glutamate/mg protein, mean ± SEM values, n = 4–5.
 [13C]Lactate[13C]Glutamate Glutamate
nmoles/mg protein
  1. *Different from corresponding value in fasted animals, p ≤ 0.02 (unpaired t-test); a different from value for the C3 position; p < 0.02 (paired t-test); n.d. not detectable.

 Fasted2.5 ± 0.551 ± 410 ± 1n.d.n.d.1.0 ± 0.112 ± 1080 ± 2
 Fed0.7 ± 0.2*21 ± 1*2.1 ± 0.3*n.d.n.d.n.d.4.5 ± 0.5*091 ± 3*
 Fasted9 ± 294 ± 316 ± 14.8 ± 0.8a6 ± 1n.d.3 ± 12.2 ± 0.510 ± 0.7
 Fed9 ± 276 ± 98.1 ± 0.7*1.8 ± 0.2*a2.2 ± 0.2*n.d.4.3 ± 0.30.50 ± 0.01*12 ± 0.7

Pyruvate recycling in the brain

The C2-labelled pyruvate formed from the injected [3-13C]pyruvate was recycled through pyruvate dehydrogenase and the TCA cycle, as could be seen from the labelling of the 5th carbon position of glutamate (Table 3; Fig. 7b; see Fig. 2 for 13C NMR spectrum of glutamate C5). At the early time points, 5 and 10 min after injection of [3-13C]pyruvate, the 13C enrichment of glutamate C5 was approximately one tenth that of glutamate C4. The C3-labelled pyruvate that was formed when [2-13C]pyruvate was injected was also metabolized through pyruvate dehydrogenase, as could be seen from the formation of [4-13C]glutamate, which was one tenth that of glutamate C5 (Table 3). No labelling of glutamine C4 was seen in these animals. In mice that received [1-13C]glucose, no C5-labelled glutamate was seen at either 5 or 15 min after injection (see Table 3 for data at 15 min), in agreement with the lack of C2-labelled lactate after injection of [1-13C]glucose.

Table 3.  Pyruvate recycling seen as formation of [5-13C]glutamate after injection of [3-13C]pyruvate, and as formation of [4-13C]glutamate from [2-13C]pyruvate. Awake mice received 9 mmoles/kg of [3-13C]pyruvate, [2-13C]pyruvate or [1-13C]glucose, and were killed at various time-points. Data are means ± SEM values, n = 4–5 in each group
PrecursorSurvivalGlu C4Glu C5
[3-13C]Pyruvate5 min12.6 ± 0.11.4 ± 0.1
[3-13C]Pyruvate10 min10 ± 0.81.1 ± 0.1
[2-13C]Pyruvate5 min1.4 ± 0.213 ± 1
[1-13C]Glucose15 min13 ± 20

Liver metabolism of [2-13C]pyruvate in the fed and fasted state

We wished to determine whether the nutritional state of the animals had any influence on the liver metabolism of 13C-labelled pyruvate, since fasting induces gluconeogenesis (e.g. Exton et al. 1972), which is assumed to affect labelling of cerebral metabolites (Fitzpatrick et al. 1990). However, liver perfusion is reduced in the fasting state (Hausken et al. 1998), which could reduce hepatic metabolism of the injected substrate.

At 5 min after injection of [2-13C]pyruvate, the 13C enrichment of the C5 position of liver glutamate was not significantly different in fed and fasted animals (Table 2). Formation of [5-13C]glutamate from [2-13C]pyruvate reflects the activity of pyruvate dehydrogenase (Fig. 7b). In contrast, the 13C enrichment of the C2 and C3 positions in glutamate was higher in the fasted than in the fed state (Table 2). Formation of [3-13C]glutamate from [2-13C]pyruvate reflects pyruvate carboxylation (Fig. 7c), an early step in gluconeogenesis. Formation of [2-13C]glutamate from [2-13C]pyruvate reflects carboxylation of the [3-13C]pyruvate that is formed after reversible carboxylation of [2-13C]pyruvate to malate and equilibration with fumarate (Fig. 7a). The C4 position of liver glutamate was not labelled in either group. The effect of fasting, to increase pyruvate carboxylation, was reflected in the C3/C5 enrichment ratio, which was 4.4 times higher in fasted than in fed animals (Table 2). The level of glutamate in the liver was not significantly different in the two groups (Table 2).

At 5 min after injection of [2-13C]pyruvate, serum glucose was 8.4 ± 0.7 mm and 10.4 ± 0.6 mm in fasted and fed animals, respectively (p = 0.03), and it was weakly labelled. In fasted animals, serum glucose C1 and C2 had enrichments of 2.3 ± 0.5% and 5.1 ± 0.8%, respectively, and in fed animals, serum glucose C1 and C2 had enrichments of 1.1 ± 0.7% and 2.5 ± 0.7%, respectively. The enrichment of glucose C2 was significantly higher in fasted animals (p = 0.048). Thus, although fasting increased hepatic pyruvate carboxylation and gluconeogenesis, it could not have influenced 13C labelling in the brain significantly during the first 5 min after injection of [13C]pyruvate.

Serum lactate was lower in fasted animals, 13.4 ± 0.4 mm versus 16.7 ± 0.8 mm in fed animals (p = 0.005), but the 13C enrichment was higher in fasted animals, 71 ± 3% versus 46 ± 9% in fed animals (p = 0.02).

Another observation in liver was that the C3 of glutamate was more highly enriched from [2-13C]pyruvate than the C2 in both the fed and fasted state (Table 2). This finding suggests incomplete scrambling (Fig. 7a) of the product of pyruvate carboxylation in the fumarate step of the hepatic TCA cycle (Heath and Rose 1985).

Comparison of brain metabolism of [2–13C]pyruvate in fasted and fed animals

Five minutes after injection of [2-13C]pyruvate, the 13C enrichment of glutamate C5 in the brain was 2.7 times higher in fasted than in fed animals (Table 2). This effect of fasting contrasted with the findings in liver. Fasting reduced the pool size of brain glutamate by 12% (Table 2). Reversible carboxylation and equilibration with fumarate was also higher in fasted animals (Table 2): The lactate C3/C2 labelling ratio, which can be used as a measure of the degree of reversible pyruvate carboxylation, was 20 ± 1 in fasted animals and in fed animals, 10 ± 1 (p = 9 × 10−4).

No labelling of the C2 or C3 of glutamate (Table 2) or glutamine, or of the corresponding C3 or C4 in GABA, was detected after administration of [2-13C]pyruvate. Such labelling could have been a sign that the injected [2-13C]pyruvate was carboxylated to be used for net (anaplerotic) synthesis of TCA cycle intermediates (Fig. 7c).


Exogenous pyruvate is an excellent substrate for cerebral energy metabolism

Exogenous pyruvate was shown to be an excellent substrate for cerebral TCA cycle activity in this study. At the higher doses, 9 and 18 mmoles/kg (or 1 and 2 g sodium pyruvate/kg), the formation of cerebral metabolites from 13C-labelled pyruvate was as effective as that from glucose. This finding agrees with those of Miller and Oldendorf (1986) who found that the Vmax for transport of pyruvate across the blood–brain barrier in conscious rats was close to that for glucose. The injected pyruvate was to some degree converted into lactate in serum. With respect to transport across the blood–brain barrier, pyruvate and lactate are fairly similar (Pardridge and Oldendorf 1977), and conversion into serum lactate would probably not affect the role of pyruvate as a potential cerebral energy source.

The present finding could support a role for exogenous pyruvate as a cerebral energy substrate during energy deficiency, as suggested by the neuroprotective effect of pyruvate in reversible ischemia (Lee et al. 2001; Yoo et al. 2004). It has been shown that hyperglycemia may be detrimental in stroke patients (Kiers et al. 1992; Capes et al. 2001); therefore, administration of glucose may have adverse effects in the acute phase of cerebral ischemia. Administration of sodium pyruvate rather than glucose could be advantageous, since pyruvate is an excellent energy substrate for the brain and since pyruvate as a base would not aggravate tissue acidosis, as could glucose by being metabolized to lactic acid (Parsons et al. 2002). However, pyruvate administration may also have adverse effects. The pro-convulsive effect of pyruvate seen in this study suggests that pyruvate administration could lower seizure threshold in stroke patients. Seizures, convulsive or non-convulsive, may occur in the acute phase of stroke and are associated with a worsening of neurological outcome (Vespa et al. 2003). Further, peri-infarct depolarization causes brain infarcts to increase in size (e.g. Hartings et al. 2003). If pyruvate administration increases neuronal excitability, as suggested by its pro-convulsive effect, and stimulates peri-infarct depolarization, such an effect may explain the increase in infarct size that was seen when pyruvate was given to rats with irreversible focal ischemia (Gonzalez-Falcon et al. 2003).

The mechanism behind the pro-convulsive effect of pyruvate may be related to the ability of pyruvate to chelate free calcium ions. Lowering extracellular free Ca2+ increases neuronal excitability and lowers seizure threshold (Pumain et al. 1983; Stringer and Lothman 1988). The dissociation constant for calcium pyruvate is 0.083 M; that for calcium lactate is 0.035 M (Davies 1938). A pro-convulsive effect of lactate has been reported previously (Hassel and Bråthe 2000a). With their rapid entry into the brain, pyruvate and lactate may transiently reach high concentrations in the extracellular fluid and increase neuronal excitability through chelation of calcium ions.

Exogenous pyruvate is predominantly metabolized by neurones

From the results obtained with F-acetate, an inhibitor of the glial TCA cycle (Muir et al. 1986; Hassel et al. 1997), it seems clear that exogenous pyruvate is predominantly metabolized by neurones, since the formation of [4-13C]glutamate and [2-13C]GABA from [3-13C]pyruvate was unaffected by F-acetate treatment. This conclusion agrees with previous studies which showed that radiolabelled pyruvate preferentially labelled glutamate over glutamine (O'Neal and Koeppe 1966; Hassel and Bråthe 2000b), a suggestion that metabolism of exogenous pyruvate mostly occurs in cells that do not express glutamine synthetase, i.e. neurones (Martinez-Hernandez et al. 1977; Tansey et al. 1991). The formation of [4-13C]glutamine from [3-13C]pyruvate in F-acetate-treated animals probably reflected glial uptake of [4-13C]glutamate formed in neurones. This glutamate would be expected to be transmitter glutamate, but its vesicular origin remains to be proven. F-acetate treatment caused a 50% reduction in the formation of glutamine from [3-13C]pyruvate. This reduction could reflect inhibition of glial metabolism of [3-13C]pyruvate; alternatively, it may reflect reduced release of [4-13C]glutamate from neurones.

Reversible pyruvate carboxylation and pyruvate recycling in the brain

We show in this study that exogenous pyruvate undergoes reversible carboxylation in the brain, and that the malate that is formed through this reaction undergoes equilibration with fumarate. This chain of reactions was suggested in a previous study by the cerebral formation of [2-13C]lactate from [3-13C]lactate (Hassel and Bråthe 2000a); in the present study, it was confirmed by the formation of [3-13C]lactate from [2-13C]pyruvate. From the predominantly neuronal metabolism of pyruvate it may be assumed that the reversible carboxylation and equilibration with fumarate occurs mainly in neurones, which agrees with the neuronal expression of mitochondrial malic enzyme (Vogel et al. 1998; McKenna et al. 2000). In spite of this, a glial contribution cannot be excluded, since glia express cytosolic malic enzyme (Kurz et al. 1993). The [3-13C]pyruvate, which was formed from the injected [213C]pyruvate, was metabolized through pyruvate dehydrogenase and the TCA cycle, as could be seen from the formation of [4-13C]glutamate. This is an example of pyruvate recycling, which, as originally proposed by Cerdan et al. (Cerdan et al. 1990; Künnecke et al. 1993), probably takes place in neurones and to a lesser extent in astrocytes, since formation of [4-13C]glutamine from [2-13C]pyruvate could not be seen. We cannot, however, exclude the possibility that pyruvate recycling also takes place in glia, as seen in cultured astrocytes (Sonnewald et al. 1996; Waagepetersen et al. 2002), since [13C]pyruvate did not appear to enter glia to any great extent.

Lack of signs of cerebral anaplerosis from exogenous pyruvate

Even though [2-13C]pyruvate was carboxylated in neurones, as discussed above, we did not see signs that the products of this carboxylation (malate and fumarate) were metabolized further in the TCA cycle. Such metabolism would have led to the labelling of the C2 and C3 positions of glutamate (Fig. 7c), and it would have indicated anaplerotic metabolism of the injected [2-13C]pyruvate. As discussed extensively by Merle et al. (2002), 13C labelling through pyruvate carboxylation will change with duration of exposure to 13C labelled substrates. In our study we did not use data obtained at late time-points after injection of [2-13C]pyruvate to assess anaplerotic activity, since hepatic gluconeogenesis became prominent with time and could have influenced labelling of cerebral metabolites.

Nutritional state affects brain metabolism of exogenous pyruvate

Fasting had two obvious effects on brain metabolism of [2-13C]pyruvate. First, the 13C enrichment of brain glutamate C5 was 2.7 times higher in fasted than in fed animals. This finding suggests an increase in cerebral uptake of monocarboxylates during fasting, as has been proposed earlier (Pan et al. 2001). The difference in 13C enrichment of brain glutamate between fasted and fed animals could not fully be explained by differences in the concentration and 13C enrichment of serum lactate, which was the predominant 13C-labelled metabolite in serum of these animals; the concentration of serum [2-13C]lactate at 5 min after injection of [2-13C]pyruvate was 9.2 mm and 7.8 mm in fasted and fed animals, respectively, as can be calculated from the mean values for the serum concentrations of lactate (13 mm and 17 mm) and the 13C enrichment (71% and 46%). Nor were the differences in the concentration and 13C enrichment of serum glucose sufficient to explain the differences between the two groups. However, because these data apply only to the situation 5 min after injection of [2-13C]pyruvate, it is not possible to quantify accurately the increase in pyruvate uptake caused by fasting. A second effect of fasting was to increase the fraction of [2-13C]pyruvate that underwent reversible carboxylation and equilibration with fumarate. The lactate C3/C2 labelling ratio in fasted animals was twice that in fed animals, suggesting either that this process increases with an increasing concentration of pyruvate in the brain, or that the activity of malic enzyme increases during fasting. Pyruvate recycling has previously been shown to increase in the fasted state (Hassel and Sonnewald 1995).

The effect of the nutritional state on pyruvate metabolism in liver

In the liver it was clear that fasting caused up-regulation of hepatic pyruvate carboxylation, an initial step in gluconeogenesis, as would be expected from previous studies (Exton et al. 1972). We base this conclusion on the observation that the percent 13C enrichment of the C2 and C3 positions in glutamate from [2-13C]pyruvate was greater in the fasted than in the fed state. The greater hepatic gluconeogenetic activity in fasted animals resulted in higher levels of 13C-labelled serum glucose in fasted than in fed animals, but these levels were low compared with those of 13C-labelled pyruvate and lactate at early time-points after injection. Therefore, we do not believe that hepatic gluconeogenesis significantly affected our data on the early phase of cerebral pyruvate metabolism.

The higher 13C enrichment of glutamate C3 than of the C2 from [2-13C]pyruvate agrees with previous findings that the product of pyruvate carboxylation in liver does not undergo complete equilibration with fumarate (Heath and Rose 1985). This finding may be explained if the malate or oxaloacetate formed through pyruvate carboxylation is metabolized directly to citrate without prior equilibration with the symmetric fumarate over fumarase.