Department of Pharmacology, School of Biomedical Sciences, The University of Queensland, St. Lucia, QLD, Australia
Address correspondence and reprint requests to Karin Borges, Department of Pharmacology, School of Biomedical Sciences, The University of Queensland, Skerman Building 65, St Lucia QLD 4072, Australia. E-mail: firstname.lastname@example.org
Triheptanoin, the triglyceride of heptanoate, is anticonvulsant in various epilepsy models. It is thought to improve energy metabolism in the epileptic brain by re-filling the tricarboxylic acid (TCA) cycle with C4-intermediates (anaplerosis). Here, we injected mice with [1,2-13C]glucose 3.5–4 weeks after pilocarpine-induced status epilepticus (SE) fed either a control or triheptanoin diet. Amounts of metabolites and incorporations of 13C were determined in extracts of cerebral cortices and hippocampal formation and enzyme activity and mRNA expression were quantified. The percentage enrichment with two 13C atoms in malate, citrate, succinate, and GABA was reduced in hippocampal formation of control-fed SE compared with control mice. Except for succinate, these reductions were not found in triheptanoin-fed SE mice, indicating that triheptanoin prevented a decrease of TCA cycle capacity. Compared to those on control diet, triheptanoin-fed SE mice showed few changes in most other metabolite levels and their 13C labeling. Reduced pyruvate carboxylase mRNA and enzyme activity in forebrains and decreased [2,3-13C]aspartate amounts in cortex suggest a pyruvate carboxylation independent source of C-4 TCA cycle intermediates. Most likely anaplerosis was kept unchanged by carboxylation of propionyl-CoA derived from heptanoate. Further studies are proposed to fully understand triheptanoin's effects on neuroglial metabolism and interaction.
In the hippocampal formation (HF) of a mouse epilepsy model, formation of citrate, GABA, succinate, fumarate, and malate from 13C-labeled glucose is reduced. Triheptanoin, the triglyceride of heptanoate, inhibits pyruvate carboxylase activity (PC, green arrow), but restores some of these metabolite levels (red arrows). The refilling of the TCA cycle via carboxylation of propionyl-CoA is likely to contribute to triheptanoin's anticonvulsant effects.
Epilepsy is a neurological disease affecting 4–10 per 1000 people per year globally. Despite progress in drug development, pharmacotherapy fails to satisfactorily control seizure symptoms in 30% of patients (Duncan et al. 2006). The high-fat, low carbohydrate ketogenic diet has been used successfully in children with intractable epilepsy (Neal et al. 2008), although it has been associated with issues of non-compliance and adverse health effects (Kang et al. 2004). A novel approach in the treatment of epilepsy is diet supplementation with the tasteless and colorless oil triheptanoin, a triglyceride of the medium chain fatty acid heptanoic acid. It is successfully being used in treating long-chain fatty acid oxidation disorders (Roe et al. 2002) and normalizing metabolic and biochemical symptoms of pyruvate carboxylase deficiency (Mochel et al. 2005). Odd numbered medium chain fatty acids produce anaplerotic propionyl-CoA in addition to acetyl-CoA (Brunengraber and Roe 2006; Roe and Mochel 2006). It was hypothesized that the anaplerotic property might contribute to the treatment of disorders with metabolic impairments such as epilepsy (Willis et al. 2010; Borges and Sonnewald 2012), in which excessive excitation of neurons and release of glutamate might cause a decrease in the tricarboxylic acid (TCA) cycle intermediate pool size and turnover and thus compromise mitochondrial oxidation of acetyl-CoA. By replenishing the TCA cycle with intermediates via entry of succinyl-CoA and providing additional acetyl-CoA, triheptanoin was hypothesized to improve mitochondrial energy and neurotransmitter metabolism (Roe et al. 2002; Borges and Sonnewald 2012).
Triheptanoin has displayed anticonvulsant properties in various mouse models of epilepsy. It delayed development of corneal kindled seizures and increased the second hit pentylenetetrazole (PTZ) tonic seizure threshold at the chronic epileptic stage of a mouse pilocarpine model (Willis et al. 2010). In mouse models of acute seizures induced by flurothyl, PTZ and 6 Hz maximal electroshock, the effects of triheptanoin were rather moderate and inconsistent, though it increased the threshold in the maximal electroshock model (Thomas et al. 2012) and in some experiments in the 6 Hz model (McDonald et al.,2013). Its anticonvulsant property in the syndrome-specific genetic mouse model of generalized absence epilepsy was also reported recently (Kim et al. 2013). Furthermore, triheptanoin added to a ketogenic diet reduced the velocity of propagation of cortical spreading depression similar to ketogenic diet in young rats (de Almeida Rabello Oliveira et al. 2008). Clinically, epilepsies that are caused by deficiencies of glucose transporter 1, pyruvate carboxylase, pyruvate dehydrogenase or complex I will most likely benefit from triheptanoin therapy.
In this study, we used the pilocarpine mouse model of temporal lobe epilepsy (TLE), in which systemic administration of pilocarpine leads to acute status epilepticus (SE), followed by recurrent spontaneous seizures. This model reproduces electroencephalographical, behavioral, and anatomical alterations and characteristics of human TLE in the mouse (Turski et al. 1984; Shibley and Smith 2002; Borges et al. 2003; Mazzuferi et al. 2012; Kharatishvili et al. 2013). The hippocampal formation appears to be one of the important brain areas involved in this epilepsy model and in many human TLEs, because spontaneous recurrent seizures, cell death, gliosis, mossy fiber sprouting, inflammation, and metabolic alterations are found in this brain area. Recently, we have characterized the metabolic profile in the cerebral cortex and hippocampal formation (HF) of CD1 mice 3.5–4 weeks after pilocarpine-induced SE. Reduction in glutamate content and mitochondrial metabolic dysfunction in the HF as well as reduced turnover of important metabolites and intermediates of the TCA cycle in the cortex and HF were the salient metabolic characteristics in these mice (Smeland et al. 2013). Importantly, the decreased production of glutamate and glutamine via the pyruvate carboxylase pathway in the cortex of this epilepsy model demonstrates dysfunctional astrocytic metabolism necessary to refill the TCA cycle intermediate pool size.
During absorption from the gastro-intestinal tract, triheptanoin is hydrolyzed into glycerol and heptanoate, which then diffuse directly into the blood. Heptanoate taken up by hepatocytes is handled in various ways. Its carbon atoms can be used in gluconeogenesis and exported as glucose into the blood for use by peripheral tissues, including the brain, and the formation and release into the blood of C-5 ketone bodies, β-hydroxypentanoate and β-ketopentanoate (Kinman et al. 2006; Gu et al. 2010; Marin-Valencia et al. 2013). Heptanoate (and/or its C-5 ketone products) and glucose from gluconeogenesis in hepatocytes have recently been reported to be the components that the brain utilizes when heptanoate is infused (Marin-Valencia et al. 2013).
To gain insight into the mechanisms by which triheptanoin exerts its anticonvulsant effects, we have here investigated whether its supplementation influences the metabolic alterations caused by pilocarpine-induced SE in mice. [1,2-13C]glucose is an excellent substrate to distinguish between metabolism by pyruvate carboxylation (PCX) and pyruvate dehydrogenation (PDH; Fig. 1) and was used to investigate these metabolic pathways. Moreover, to shed light on the changes in metabolism, we investigated the gene expression and activity of key enzymes involved.
Materials and methods
Male CD1 mice, 7- to 8-week old, (Animal Resources Center, Canningvale, Western Australia, Australia) were used. All animals were housed in individual cages under a 12-h light:dark cycle and were adapted to these conditions for at least 1 week before being used in the experiments. After pilocarpine and saline treatment animals were randomly assigned to either a control diet (SF11-027 diet, Specialty Feeds, Glen Forrest, Western Australia, Australia) or 35% of calories triheptanoin diet (SF11-028). The two diets were matched in protein, mineral, vitamin, and antioxidant content. Triheptanoin replaced 15% sucrose (w/w), some of the complex carbohydrates and most of the long chain fats from the control diet. Food and water were available ad libitum. The experiments followed the guidelines of the Queensland Animal Care and Protection Act 2001 and were approved by the University of Queensland's Animal Ethics Committee. All efforts were made to minimize the suffering and the number of animals. This work is written according to the ARRIVE guidelines (http://www.nc3rs.org/ARRIVE).
To induce SE, 35 CD1 mice, weighing 25–40 g, were injected with pilocarpine (330–345 mg/kg s.c. in 0.9% saline; Sigma Aldrich, St Louis, MO, USA) (unless otherwise stated reagents are from Sigma Aldrich). Behavioral SE, as defined by continuous seizure activity consisting mainly of whole-body continuous clonic seizures, was experienced by 23 pilocarpine-injected mice (66%), while nine mice died and three showed no SE. Fifteen to thirty minutes before pilocarpine injections, animals were administered with methylscopolamine (2 mg/kg i.p. in 0.9% NaCl) to minimize peripheral side effects. Ninety minutes after pilocarpine administration, animals were injected with pentobarbital (22.5 mg/kg i.p. in 0.9% NaCl; Provet, Northgate, Queensland, Australia) followed by 1 mL 4% dextrose in 0.18% saline (s.c.). After SE, eleven of the SE mice were placed on triheptanoin diet and twelve mice on control diet. In addition, mice were hand-fed moistened cookies and injected with 4% dextrose in 0.18% saline twice a day for about 3 days and thereafter when needed. Twice daily, mice were monitored and body weight and appearance were recorded. As a result of the high variability of seizure frequencies in these mice (Borges et al. 2003), no systematic observations were done to compare spontaneous seizure frequencies in SE mice. Twenty-one mice with SE were observed to have handling induced seizures and altered behavior, ten of the triheptanoin fed mice and eleven on the control diet. In addition, they showed lack of nest building and all mice had lost 10–20% of weight after SE. One SE mouse on triheptanoin and one on control diet died suddenly overnight in their home cages at 10 and 21 days, respectively. Autopsies did not reveal causes of death and therefore sudden unexplained death in epilepsy or severe seizures are likely. For this experiment, 22 mice received methylscopolamine and pentobarbital only and 0.9% saline instead of pilocarpine. Afterward, eleven of these sham mice each were placed on either triheptanoin or control diets.
In a separate experiment, SE mice were produced for gene expression and enzyme activity analysis according to same protocol as above. Of 45 CD1 mice injected with pilocarpine, 20 developed behavioral SE (44%) over the 90 min observation period, 18 did not develop SE (no SE mice), and seven died. In the following 24 hours after the termination of SE, three of the SE mice died. As reported above, all mice that developed SE lost 10–20% of body weight following SE and lacked nest building. Of the mice that developed SE, nine were randomly placed on the 35% triheptanoin diet, and eight on the control diet. Ten of the no SE mice were fed control diet, whereas eight were fed triheptanoin-diet. In this experiment, mice that did not experience SE (no SE) were used as control, because they do not experience spontaneous seizures or interictal spikes (Kharatishvili et al. 2013) and their properties have always been similar to saline injected control mice in the past (Borges et al. 2003; Willis et al. 2010).
When monitored by video-electro-encephalography, all male CD1 mice subjected to our pilocarpine-SE model developed spontaneous recurrent seizures and/or interictal spikes in another laboratory at our university (Kharatishvili et al. 2013) similar to a previous report in the same mouse strain (Shibley and Smith 2002).
[1,2-13C]Glucose injections and tissue extraction
At 3.5–4 weeks after SE, animals were injected in random order with [1,2-13C]glucose (543 mg/kg i.p.; 99% 13C; Cambridge Isotope Laboratories, Woburn, MA, USA). To instantaneously inactivate brain metabolic reactions, mice were subjected to microwave fixation of their heads at 5 kW for 0.80–0.86 sec (Model MMW-05, Muromachi, Tokyo, Japan) 15 min after glucose injection. The time interval of 15 min between 13C glucose injection and microwave fixation ensures substantial 13C label incorporation in brain metabolites without washout (pilot experiments, results not shown). After microwave fixation, mice were decapitated, trunk blood was collected, and cerebral cortices and HF were dissected and stored at −80°C until extraction. The HF included the dentate gyrus, hippocampus proper, subiculum, but not entorhinal cortex. The blood samples were centrifuged at 1000 g for 5 min and the serum was later analyzed for total amounts of glucose using the glucose oxidase method (Sigma) and the amount of [1,2]glucose using 1H NMR. The brain samples were masked and subjected to a water/methanol-chloroform extraction method as previously described (Le Belle et al. 2002). L-2-aminobutyric acid (α-ABA) was added as an internal standard for HPLC analysis and samples were then homogenized in 200 mL of methanol using a Vibra Cell sonicator (Model VCX 750; Sonics and Materials, Newtown, CT, USA). Samples were lyophilized and all subsequent analyses were performed by investigators masked to the treatment.
To quantify the amounts of metabolites in cerebral cortex and HF, we used HPLC and 1H NMR spectroscopy. We determined the amounts of 13C in cortical metabolites using 13C NMR spectroscopy and the percentage of 13C enrichment in HF metabolites with GC-MS, owing to the small size of the mouse HF. The percent enrichment with 13C isotopomers in glucose, alanine, and lactate in both brain regions was quantified using 1H NMR spectroscopy.
1H and 13C NMR spectroscopy
Lyophilized samples were dissolved in 120 mL of D2O (99.9%; Cambridge Isotope Laboratories) containing 0.29 g/L 2,2,3.3-d(4)-3-(trimethylsilyl)propionic acid sodium salt (98%; Alfa Aesar, Karlsruhe, Germany) and 0.10% ethylene glycol (Merck, Darmstadt, Germany) as internal standards for quantification. Samples were transferred to SampleJet tubes (3.0 × 103.5 mm2) for insertion into the SampleJet autosampler (Bruker BioSpin GmbH, Rheinstetten, Germany). Next, samples were analyzed using a QCI CryoProbe 600 MHz ultrashielded Plus magnet (Bruker BioSpin GmbH). We recorded 1H and 13C NMR spectra at 20°C. 1H NMR spectra were acquired with the following parameters: pulse angle of 90°, acquisition time of 2.66 s, and relaxation delay of 10 s. The number of scans was 256. Proton-decoupled 13C NMR spectra were acquired with the following parameters: pulse angle of 30°, acquisition time of 1.65 s, and a relaxation delay of 0.5 s, 30 kHz spectral width with 98 K data points. The number of scans needed to obtain appropriate signal to noise ratios was approximately 10 000. Relevant peaks in the spectra were identified and integrated using TopSpin 3.0 software (Bruker BioSpin GmbH). The total amounts and 13C labeling of metabolites were calculated from the integrals of the peak areas using internal standards 2,2,3.3- d(4)-3-(trimethylsilyl)propionic acid sodium salt for 1H NMR spectra and ethylene glycol for 13C NMR spectra. Integrals from 1H spectra were corrected for number of protons constituting the peak. Integrals from 13C spectra were corrected for nuclear Overhauser enhancement and relaxation effects relative to the internal standard, and for the 1.1% natural abundance of 13C using the data obtained from 1H NMR spectroscopy or HPLC. Percent 13C enrichment of glucose, lactate, and alanine with [1,2-13C]glucose, [2,3-13C]lactate and [2,3-13C]alanine, respectively, was calculated using formulas and considerations as reported in Smeland et al. (2013).
We determined the total amounts of amino acids using a Hewlett Packard 1100 System (Agilent Technologies, Santa Clara, CA, USA) with fluorescence detection, after derivatization with o-phthaldialdehyde. The components were separated with a ZORBAX SB-C18 (4.6 × 150 mm, 3.5 micron) column from Agilent using 50 mM sodium phosphate buffer (pH 5.9) with 2.5% tetrahydrofurane and methanol (98.75%) with tetrahydrofurane (1.25%) as eluents. Compounds were quantified by comparison with a standard curve derived from a standard solution of metabolites run repeatedly at 15 sample intervals and corrected using α-ABA as an internal standard.
Aliquots of the samples were dissolved in 0.05 mM HCl, followed by lyophilization. Organic acids and amino acids were then extracted into an organic phase of ethanol and benzene, dried under air, and reconstituted in N,N-dimethylformamide before derivatization with N-Methyl-N-(t-butyldimethylsilyl)trifluoroacetamide in 1% t-butyldimethyl-chlorosilane. We analyzed compounds using an Agilent 6890N gas chromatograph (Agilent Technologies) linked to an Agilent 5975B mass spectrometer (Agilent Technologies) with an electron ionization source. Results were corrected for natural abundance of 13C using standard solutions that were acquired concurrently with the samples.
Interpretation of 13C-labeling Patterns from Metabolism of [1,2-13C]glucose
To interpret the 13C incorporation results, it is necessary to analyze the metabolism of [1,2-13C]glucose (Fig. 1). Via glycolysis, [1,2-13C]glucose is metabolized to [2,3-13C]pyruvate, which can be converted to [2,3-13C]lactate, [2,3-13C]alanine, or enter the TCA cycle via PDH as [1,2-13C]acetyl-CoA. Metabolism of [1,2-13C]acetyl-CoA in the TCA cycle gives rise to [4,5-13C]-2-oxoglutarate, which is a precursor for [4,5-13C]glutamate. [4,5-13C]glutamate may subsequently be converted to [4,5-13C]glutamine by the exclusively glial enzyme glutamine synthetase (Norenberg and Martinez-Hernandez 1979) or to [1,2-13C]GABA in γ-aminobutyric acid (GABA)ergic neurons. In astrocytes, [2,3-13C]pyruvate may also be converted to [2,3-13C]oxaloacetate via PCX (Patel 1974). This can lead to the formation of [2,3-13C]-2-oxoglutarate and eventually [2,3-13C]glutamate, [2,3-13C]glutamine and [3,4-13C]GABA. Further metabolism of [2,3-13C]-2-oxoglutarate and [4,5-13C]-2-oxoglutarate in the TCA cycle yields labeled oxaloacetate, which can be transaminated to [1,2-13C]aspartate or [3,4-13C]aspartate. If labeled metabolites remain in the TCA, different labeling patterns in amino acids emerge, which are not discussed in this article.
Quantitative real-time polymerase chain reactions
After isoflurane anesthesia and decapitation, forebrains were removed from mice three weeks after SE or no SE and homogenized in Trizol (Life Technologies, Australia). Total RNA was isolated and treated with DNase I. Using random hexamers and the Tetro cDNA Synthesis Kit (Bioline, Australia) 2 μg of this RNA were reverse transcribed. Gene expression was quantified using qRT-PCR with gene specific primers (Table S1), FastStart Universal SYBR Green Master mix (Roche, Castle Hill, NSW, Australia), the Eppendorf epMotion 5075 Robotics System and ViiA™7 Real-Time PCR System (Applied Biosystems, Mulgrave, VIC., Australia). The efficiency of each primer pair was taken into consideration in the calculation of fold expression (∆Ct) of gene of interest (goi) relative to the geometric mean of two housekeeping genes Tbp (TATA binding protein) and Hmbs (Hydroxymethylbilane synthase) as in the formula adapted from Vandesompele et al. (2002):
Enzyme activity assays
The other halves of the forebrains from the mice used for qRT-PCR were homogenized in a 0.32 M sucrose, 50 mM Tris-HCl solution pH 7.5 with 0.1% Triton X-100. To collect crude mitochondrial pellet samples were centrifuged at 4000 g for 20 min at 4°C. The supernatant was collected and centrifuged at 20 000 g for 20 min at 4°C. The supernatant was discarded and the pellet resuspended in extraction buffer.
The activities of oxoglutarate dehydrogenase (OGDH) complex and PDH complex were measured by continuous spectrophotometric assays at 25°C (Lai and Cooper 1986). The assay mixture contained 50 mM Tris-HCl (pH 8 for PDH, and pH 7.4 for OGDH assay), 0.2 mM sodium CoA, 2 mM Nicotinamide adenine dinucleotide, 0.5 mM thiamine pyrophosphate, 0.5 mM magnesium chloride, 10 mM dithiothreitol, and 20 μL of tissue sample. Reactions were initiated by the addition of 10 mM oxoglutarate to measure OGDH activity, and 10 mM concentration of sodium pyruvate for PDH activity.
PCX enzyme activity assays were measured by a continuous colorimetric coupled enzyme assay at 30°C. The final concentration of all reagents in the reaction mixture were 50 mM Tris-HCl buffer (pH 8.0), 50 mM sodium bicarbonate, 5 mM magnesium chloride, 0.1 mM acetyl-CoA, 5 mM sodium pyruvate, 5 mM adenosine triphosphate, 0.5 mM 5,5'-dithiobis-(2-nitrobenzoic acid), and 1 unit of citrate synthase with 20 μL of the crude mitochondrial extract. The reaction was initiated with the addition of acetyl-CoA. The rate of change in absorbance was measured at a wavelength of 412 nm for 15 min using the Sunrise Tecan microplate reader (Tecan Group Ltd., Mannedorf, Switzerland), which represents the amount of 5,5'-dithiobis-(2-nitrobenzoic acid) reduced by the free CoA produced by citrate synthase.
All activity rates were corrected to milligrams of protein in samples, using the bicinchoninic acid protein Assay kit (Thermo Scientific, Rockford, IL, USA).
For statistics, we used one-way anova followed by a least significant difference post hoc test with p < 0.05 regarded as significant. This statistical test does not correct for multiple comparisons in order to avoid false negative results. Instead, it may increase the number of false positive results, especially given the high number of comparisons performed. The 2 × 2 design of the two-way anova test was also applied to the results when a need to compare by disease and triheptanoin supplementation was beneficial. Data are represented as mean ± SEM. We chose to use 11–12 animals in each group based on previous experience on variations of metabolite levels in epilepsy models. Because of faulty 13C injections, post-mortem effects and occasional over-fixation of the brain tissue, the number of samples for each group varied between analytical methods.
The total blood glucose level averages in each group ranged from 14.56 to 16.9 mmol/L, while the percent enrichment with [1,2-13C]glucose varied from 34 to 43%. The amount of total glucose and percent enrichment with [1,2-13C]glucose in blood were not affected by SE or triheptanoin supplementation (both anovas, p < 0.05). Table 1 shows the amount of glucose, lactate and alanine and 13C labeling and percent 13C enrichment with [1,2-13C]glucose, [2,3-13C]lactate and [2,3-13C]alanine in the cortex and HF. Triheptanoin supplementation did not significantly alter glucose content in the sham or SE mice in both brain areas. In the cortex, a 50% (p = 0.049, One-way anova; p = 0.018, post hoc test) increase in total glucose in SE mice fed triheptanoin compared to the sham mice fed triheptanoin was observed. Two-way anova 2 × 2 factorial interaction analysis revealed a significant effect of triheptanoin on the total glucose content in the cortex of SE and sham mice (p = 0.018). A similar interaction was also observed with [1,2-13C]glucose (p = 0.031). In the HF, total glucose content was increased (p = 0.017) by triheptanoin compared to control diet in SE mice, with a similar trend (p = 0.053) in the amount of [1,2-13C]glucose. However, the percent enrichment with [1,2-13C]glucose was not affected by triheptanoin in both the brain areas. There were no significant alterations in the total, labeled lactate amounts or its percent 13C enrichment by SE or the triheptanoin diet in both brain areas. The total and labeled alanine levels, but not the percent enrichments, were decreased by SE in HF while triheptanoin did not have any effect on any of these variables in both the SE and sham mice. Pentose Phosphate Pathway activity could not be detected in any group, indicating that this pathway was not very active.
Table 1. [1,2-13C]glucose metabolism in the cortex and hippocampal formation of sham and pilocarpine-induced status epilepticus (SE) mice fed with either control (CON) or triheptanoin (TRIH) diet
13C percent excess enrichments and amounts of 13C labeled and total glucose, lactate, and alanine. Mice in all the four groups were injected with [1,2-13C]glucose 15 min before microwave fixation of the head. Data represent mean ± SEM and statistical significance is indicated in one-way anova by X (for the cortex) and Y (for the HF) (p < 0.05); followed by the LSD post hoc test indicated when p < 0.05 compared to (a) Sham-CON, (b) Sham-TRIH and (c) SE-CON groups.
n = 8, 9, 9 and 10 in Sham-CON, Sham-TRIH, SE-CON, SE-TRIH groups in the cortex and n = 11 Sham-CON and n = 9 for other three groups in HF.
n = 7 Sham-TRIH and n = 8 SE-CON groups in the cortex.
13C labeling of TCA cycle metabolites and intermediates in the cortex
13C labeling of TCA metabolites via PDH in the cortex
The amount of [4,5-13C]glutamate and [4,5-13C]glutamine was reduced in the SE compared to the sham mice, indicating reduced PDH activity. The triheptanoin diet did not significantly affect the amount of 13C labeling in neither the SE nor the sham groups (Fig. 2A). The amount of [1,2-13C]GABA was not affected by SE or feeding of triheptanoin. [1,2-13C]aspartate and [3,4-13C]aspartate are formed from both [4,5-13C]oxoglutarate (PDH) and [2,3-13C]oxoglutarate (PCX) via TCA cycling and could not, thus, be used to distinguish between the pathways.
13C labeling of TCA metabolites via PCX in the cortex
[1,2-13C]glucose is an excellent substrate to study pyruvate carboxylation. Carboxylation of [2,3-13C]pyruvate results in [2,3-13C]oxaloacetate formation which rapidly equilibrates with [2,3-13C]aspartate via transamination. We found a reduction of the [2,3-13C]aspartate content in the SE mice compared with the sham group (p = 0.002, One-way anova; p = 0.019, post hoc test, Fig. 2B). Triheptanoin feeding reduced its content in both the sham (p = 0.002, One-way anova; p = 0.002, post hoc test) and the SE groups (p = 0.002, One-way anova; p = 0.040, post hoc test). Furthermore, amounts of [2,3-13C]glutamate and [2,3-13C]glutamine were reduced in the SE group. Triheptanoin did not significantly affect the labeling in glutamate and glutamine in either the sham or SE groups.
% 13C enrichment with M + 2 isotopomers of TCA metabolites and intermediates in the HF
The HF tissue was too small to obtain meaningful spectra for quantification of metabolites using 13C NMR spectroscopy, therefore GC-MS was used to assess the percent 13C enrichment with M + 2 isotopomers (Fig. 3). No alterations in the % enrichment with M + 2 in glutamate, glutamine, and aspartate were found because of SE or triheptanoin feeding (not shown). The % enrichment with M + 2 was reduced in succinate (p = 0.028, One-way anova; p = 0.004, post hoc test) and GABA (p = 0.005, One-way anova; p = 0.001, post hoc test) in the SE group compared to the sham group, but was not significantly affected by triheptanoin in either group. Triheptanoin, however, increased (p = 0.017, One-way anova; p = 0.016, post hoc test) the percent enrichment with M + 2 of fumarate in the SE mice, an interaction confirmed by a 2 × 2 factorial design treatment of the data using two-way anova (p = 0.016). The % enrichment with M + 2 in malate (p = 0.050, One-way anova; p = 0.011, post hoc test) and citrate (p = 0.037, One-way anova; p = 0.008, post hoc test) in the SE mice compared to the sham mice was reduced. However, the statistical significance of these reductions was lost in the triheptanoin-fed SE group, indicating that triheptanoin may prevent the decrease in the TCA cycle intermediate pool in SE mice.
Amount of amino acids and other important metabolites in the cortex and HF
The total content of some amino acids in extracts from both the cortex and HF was determined using HPLC (Fig. 4). The total glutamate content was reduced in the HF in the SE group compared to sham by 26% (p = 0.001, One-way anova; p = 0.001, post hoc test), which was not affected by triheptanoin. Neither SE mice nor mice fed triheptanoin showed any significant alterations in amounts of other neuroactive amino acids (GABA, glutamine, aspartate and glycine).
Threonine levels were increased by 20% (p = 0.001, One-way anova; p = 0.015, post hoc test) in SE mice compared to sham mice in the HF, but this did not reach statistical significance in the cortex. Triheptanoin diet reduced threonine levels in the cortex irrespective of SE. In sham mice triheptanoin reduced the level by 17% (p = 0.001, One-way anova; p = 0.016, post hoc test), while it caused a 34% reduction (p = 0.001, One-way anova; p < 0.001, post hoc test) in SE mice. In the HF, however, triheptanoin reduced threonine to statistically significant levels only in the SE group (p = 0.001, One-way anova; p = 0.016, post hoc test). A 2 × 2 factorial design treatment of the data using a two-way anova indicated that triheptanoin affected the level of threonine in SE (p < 0.001) and not in the sham mice (p = 0.119). Lysine levels were increased in the HF of SE mice by 60% (p = 0.006, One-way anova; p = 0.007, post hoc test) compared to the sham, but not in cortex. Triheptanoin diet did not affect the levels of lysine in any group. The amount of the tripeptide glutathione was not altered in the SE mice in the cortex and the triheptanoin diet did not affect glutathione levels in either the sham or SE groups. However, glutathione was reduced by 18% (p = 0.008, One-way anova; p = 0.003, post hoc test) in the HF of SE mice compared to the sham group. No statistically significant reduction was observed between the triheptanoin-fed mice in the sham and SE groups, indicating a tendency toward prevention of glutathione reduction in SE mice, as was found before (Willis et al. 2010).
As presented in Table 2, some important metabolite levels were determined using 1H NMR spectroscopy. No significant alterations because of either SE or triheptanoin were observed in the levels of NAD(P)+, AMP, ATP, ADP, fumarate, creatine, and phosphocreatine in either brain areas. The amounts of NAA and NAD(P)H were not altered in the cortex and the same was true for the amount of succinate in the HF. The amount of NAD(P)H was increased by the triheptanoin diet in the sham group (p = 0.022, One-way anova; p = 0.023, post hoc test), though it was not altered in the SE group compared to the sham and triheptanoin diet did not cause significant alteration of the amounts in the SE group. SE increased the level of myo-inositol (compared to the sham, p < 0.001, One-way anova; p = 0.008, post hoc test) in the cortex while the increase was not statistically significant in the HF. Triheptanoin did not affect the amount of myo-inositol in the cortex of both the sham and SE mice but the level was increased in the HF of SE mice on the triheptanoin diet when compared to both sham groups (p = 0.046, One-way anova; p = 0.021 and p = 0.022, post hoc test). The amount of NAA was reduced in the HF of SE mice compared to the sham group (p = 0.004, One-way anova; p = 0.006, post hoc test), and triheptanoin did not affect the level in the sham or the SE mice.
Table 2. Metabolite levels in extracts of hippocampal formation and cortex
The amount of metabolites in brain extracts of sham mice and mice after status epilepticus (SE) in the absence (Sham-CON and SE-CON) and presence (Sham-TRIH and SE-TRIH) of triheptanoin feeding in the cortex and hippocampal formation. Mice in all the four groups were injected with [1,2-13C]glucose 15 min before microwave fixation of the head. Quantification was made using 1H-NMRS. Data represent mean ± SEM and statistical significance is indicated in one-way anova by X (for the cortex) and Y (for the HF) (p < 0.05); followed by the LSD post hoc test indicated when p < 0.05 compared to (a) Sham-CON, (b) Sham-TRIH and (c) SE-CON groups.
n = 8 in the Sham-CON, n = 5 in the Sham-TRIH and n = 7 in both the SE-CON and SE-TRIH groups in the hippocampal formation.
Three weeks after pilocarpine injections no changes were observed in gene expression or enzyme activity of Pdh, Ogdh, or Sdha between SE and no SE mice in both control and triheptanoin diet groups (p > 0.05, Fig. 5A, B). However, a reduction in the Pcx mRNA levels by 30% (p = 0.0268, One-way anova; p < 0.05, post hoc test) and PCX activity by 39% (p = 0.0417, One-way anova; p < 0.05, post hoc test) was observed in triheptanoin-fed SE mice compared to control-fed SE mice, consistent with reduced [2,3-13C] aspartate levels. No other significance was found in the gene expression or enzyme activity levels in SE and no SE fed each diet.
Anticonvulsant effects of triheptanoin have been reported in various epilepsy models. In the pilocarpine model in mice, triheptanoin increased the second hit PTZ tonic seizure threshold (Willis et al. 2010). Hence, we discuss here the effects of 35% caloric triheptanoin supplementation diet on brain energy metabolism and amino acid neuroactive amino acids homeostasis in pilocarpine-SE mice and relate this to possible mechanisms by which the diet affects susceptibility to seizures in this model.
Triheptanoin supplementation does not significantly affect glucose utilization
Glucose is the major substrate for energy and amino acid production in the adult mammalian brain and our results and other reports (Willis et al. 2010; Kim et al. 2013) show that animals on a diet containing triheptanoin have unaltered steady state plasma levels of glucose. Increased plasma glucose levels were, however, reported following short term infusion of triheptanoin (Kinman et al. 2006; Gu et al. 2010) and heptanoate (Marin-Valencia et al. 2013) in rats, probably secondary to gluconeogenic release of glucose by hepatocytes (Marin-Valencia et al. 2013). Even after infusing of 50% calories as 13C-labeled heptanoate (i.v.) only 13% of blood glucose was labeled (Marin-Valencia et al. 2013). Our results showed that glucose levels in blood and percent enrichment with [1,2-13C]glucose are not affected by SE or triheptanoin feeding. It is, hence, safe to attribute changes in glucose metabolism to processes downstream to glucose uptake into cells.
In this study, glucose contents in the HF were not altered, although a trend (p = 0.098, One-way anova; p = 0.017, post hoc test) of increase in total glucose was evident in triheptanoin-fed SE mice compared to control-fed SE mice. This trend is an indication of reduced glucose consumption, suggesting utilization of heptanoate and/or its C-5 ketone derivatives for energy metabolism in the epileptic brain in HF. The fact that astrocytes metabolize only 30% of acetyl-CoA from glucose (Hassel et al. 1995; Qu et al. 2000) and that metabolism of heptanoate and/or its C-5 ketones is possibly compartmentalized to astrocytes (Marin-Valencia et al. 2013), might have contributed in masking the significance of this effect. The glucose content in the cortex, on the other hand, was increased in both SE groups compared to sham mice fed triheptanoin but not to those fed control diet. Triheptanoin did not appear to affect glucose metabolism in the cortex of SE mice, since contents and 13C labeling of glucose, lactate, and alanine remained unaltered. Overall, it appears that the effect of triheptanoin on glucose metabolism is small in both healthy and epileptic mice, although a trend towards decreased glucose consumption was seen in the HF of SE mice.
Triheptanoin prevents reduction of 13C labeled TCA intermediates in chronically epileptic mice
Pilocarpine-SE causes extensive neuronal cell loss, astrogliosis, and mossy fiber sprouting in the HF (Turski et al. 1984; Shibley and Smith 2002; Borges et al. 2003; Mazzuferi et al. 2012). Several studies revealed reduced content of neuronal markers, particularly NAA, in the pilocarpine model in rats (Melo et al. 2005) and mice (Martinez-Hernandez et al. 1977; Smeland et al. 2013) and the kainate model in rats (Alvestad et al. 2008) replicating metabolic alterations of human TLE. Reduced content of TCA cycle intermediates, such as malate (Willis et al. 2010) and succinate (Willis et al. 2010; Smeland et al. 2013), have also been reported. Triheptanoin did not prevent the reduction of NAA in the SE group in this study, implicating no or minimal effect on neuronal mitochondrial function and/or cell death. However, the increase in the percent of M + 2 in fumarate in the SE mice fed triheptanoin compared with those on control diet (Fig. 3) indicates increased capacity in either or both the PDH and PCX pathways. This is consistent with normalized mitochondrial activity and TCA cycle turnover caused by triheptanoin in SE mice. Similar results were obtained for M + 2 in citrate and malate, which were decreased in the SE group but not different in SE mice fed triheptanoin compared to sham groups. Willis et al. (2010) have also reported that triheptanoin prevented the decrease in the levels of malate as well as the anaplerotic propionyl-CoA caused by pilocarpine-SE. The effects of triheptanoin on the percent enrichment with M + 2 isotopomers were not observed in the sham mice (this study, Willis et al. (2010)), suggesting that triheptanoin exerts more effects on the epileptic than healthy brain as was previously proposed (Willis et al. 2010).
Additional indications that triheptanoin is anaplerotic in the pilocarpine model in mice
Triheptanoin enhances oxaloacetate turnover in astrocytes
Pyruvate carboxylation is the major anaplerotic pathway in the brain and is exclusively glial (for review Sonnewald and Rae 2010). Triheptanoin did not affect the extent of anaplerosis from [1,2-13C]glucose in glutamate and glutamine but reduced it in aspartate in both the sham as well as SE groups (Fig. 2B). One possible reason for the reduced amounts of [2,3-13C]aspartate is that there is a general reduction in PCX activity and hence a reduction in the formation of [2,3-13C]oxaloacetate from [2,3-13C]pyruvate. Consistent with this finding, the expression and activity of PCX in the SE mice was reduced in triheptanoin compared to control diet in SE mice (Fig. 5). It is possible that the reduction in PCX activity is caused in response to increased anaplerosis via heptanoate-derived propionyl-CoA and the propionyl-CoA carboxylation pathway. The lack of a significant effect of triheptanoin on the anaplerotic production of [2,3-13C]glutamate and [2,3-13C]glutamine in both the sham and SE mice may reflect more channeling of oxaloacetate into the TCA cycle as a result of normalized acetyl-CoA availability (Willis et al. 2010) in the triheptanoin diet (possibly from glucose and beta-oxidation of heptanoate). Enhanced utilization of oxaloacetate has previously been reported following ketogenic diet (Yudkoff et al. 1997, 2004) and can explain the proportional increases in M + 2 labeled malate levels (Fig. 3).
Triheptanoin reduces threonine levels in chronically epileptic mice
Our findings show that in the chronic stage of the pilocarpine model the level of threonine is increased in the brain. Threonine is an essential amino acid acquired from diet. Interestingly, triheptanoin reduced the level of threonine in both sham and SE mice in the cortex. In the HF the reduction of threonine content by triheptanoin was statistically significant only in the SE and not in the sham mice, suggesting a more pronounced effect on the epileptic brain. High dietary intake of threonine is the most common cause of hyperthreoninemia, which is associated with certain forms of seizures (Reddi 1978; Clayton et al. 2003). Threonine is not metabolized extensively in the brain (Pardridge 1983), but can follow either of two catabolic pathways: threonine dehydratase ultimately producing succinyl-CoA via propionyl-CoA and threonine 3-dehydrogenase producing glycine and acetyl-CoA (Gaitonde 1975; Edgar 2005). Considering the reduced propionyl-CoA level in this model (Willis et al. 2010) and the reduced 13C labeling of TCA cycle metabolites (Fig. 2), it could be expected that threonine was metabolized via the threonine dehydratase pathway to increase propionyl-CoA levels. Our results, however, do not suggest a sink effect. Rather threonine metabolism appears to be compromised and its concentration increased in sham SE. It appears, hence, that in epileptic brains, there is not only an increased need for anaplerotic reactions, but at the same time possible impairment of utilization of threonine for anaplerotic propionyl-CoA production. The reduction in the level of threonine by triheptanoin in SE mice as well as the sham suggests enhancement of utilization of threonine possibly through its conversion into TCA cycle intermediates.
Triheptanoin supplementation does not affect neuroactive amino acids levels
Neuroactive amino acid levels (glutamate, GABA, aspartate, glutamine) in the brain of sham as well as SE mice were not affected by the three-week triheptanoin supplementation in this study. We have recently reported a reduction in glutamate and NAA levels in the HF of pilocarpine-SE mice proposing possible mitochondrial dysfunction and/or neuronal cell death (Smeland et al. 2013). As can be deduced from unaffected NAA levels, our current findings indicate that triheptanoin did not affect SE-induced cell death or mitochondrial dysfunction in neurons (Table 2). This lack of effects on neurons may be because of predominant heptanoate metabolism in astrocytes (Marin-Valencia et al. 2013).
This study confirmed the findings of previous studies that mitochondrial metabolism is impaired in the chronic stage of the pilocarpine mouse epilepsy model. Here, we detected several signs of anaplerotic activity by triheptanoin in SE mice, suggesting that “diseased” brains are more responsive to metabolic alteration by triheptanoin. The most prominent effects of triheptanoin in the brains of SE mice include reduced PCX expression and activity, which like propionyl-CoA carboxylation dependent anaplerosis is found in astrocytes. Because no major alterations were found in the levels of most other metabolites measured, it is likely that triheptanoin normalized anaplerosis producing succinate via propionyl-CoA carboxylation, which is consistent with our earlier finding that triheptanoin alleviated the reduction in propionyl-CoA levels and elevated levels of methyl-malonyl-CoA in SE mice (Willis et al. 2010). Consistent with normalized anaplerosis, we also observed percent wise increased 13C labeling of TCA cycle intermediates, indicating that triheptanoin increased TCA cycle activity in SE mice. Further studies focusing on glial metabolism seem to be necessary to fully reveal the effects of triheptanoin on the brain.
The authors thank Nicola K Thomas for help with the animals and Nina Berggaard and Tesfaye Tefera for their help with the GC-MS analyses. Karin Borges is grateful for funding by the Australian National Health and Research Council (Grants 63145 and 1044407).
Disclosure/Conflict of interest
KB has filed for a complete US patent regarding triheptanoin as a treatment for seizures. All other authors declare no conflict of interest.