Heptanoate as a neural fuel: energetic and neurotransmitter precursors in normal and glucose transporter I-deficient (G1D) brain

Journal of Cerebral Blood Flow & Metabolism (2013) 33, 175–182 & 2013 ISCBFM All rights reserved 0271-678X/13 $32.00 www.jcbfm.com

ORIGINAL ARTICLE

Heptanoate as a neural fuel: energetic and neurotransmitter precursors in normal and glucose transporter I-deficient (G1D) brain Isaac Marin-Valencia1, Levi B Good1, Qian Ma1, Craig R Malloy2,3,4 and Juan M Pascual1,5,6 It has been postulated that triheptanoin can ameliorate seizures by supplying the tricarboxylic acid cycle with both acetyl-CoA for energy production and propionyl-CoA to replenish cycle intermediates. These potential effects may also be important in other disorders associated with impaired glucose metabolism because glucose supplies, in addition to acetyl-CoA, pyruvate, which fulfills biosynthetic demands via carboxylation. In patients with glucose transporter type I deficiency (G1D), ketogenic diet fat (a source only of acetyl-CoA) reduces seizures, but other symptoms persist, providing the motivation for studying heptanoate metabolism. In this work, metabolism of infused [5,6,7-13C3]heptanoate was examined in the normal mouse brain and in G1D by 13C-nuclear magnetic resonance spectroscopy, gas chromatography-mass spectrometry (GC-MS), and liquid chromatography-mass spectrometry (LC-MS). In both groups, plasma glucose was enriched in 13C, confirming gluconeogenesis from heptanoate. Acetyl-CoA and glutamine levels became significantly higher in the brain of G1D mice relative to normal mice. In addition, brain glutamine concentration and 13C enrichment were also greater when compared with glutamate in both animal groups, suggesting that heptanoate and/or C5 ketones are primarily metabolized by glia. These results enlighten the mechanism of heptanoate metabolism in the normal and glucosedeficient brain and encourage further studies to elucidate its potential antiepileptic effects in disorders of energy metabolism. Journal of Cerebral Blood Flow & Metabolism (2013) 33, 175–182; doi:10.1038/jcbfm.2012.151; published online 17 October 2012 Keywords: glucose; glutamine; nuclear magnetic resonance; transporter

INTRODUCTION Metabolism of blood-borne glucose provides the majority of energy consumed by the brain and also the substrates for neurotransmitter synthesis via the tricarboxylic acid (TCA) cycle.1 Defects of brain glucose metabolism, such as glucose transporter type I deficiency (G1D), commonly lead to neuronal hyperexcitability manifested as epilepsy as well as delayed myelin deposition and brain maturation.2,3 Alternative brain fuels, such as ketones produced from a high-fat content ketogenic diet, can ameliorate epilepsy in G1D patients4 but, at best, this therapy has a modest impact on intellectual disability, ataxia (movement incoordination), and spasticity (motor neuron dysfunction) in most cases.5,6 Food-derived ketones (bhydroxybutyrate and acetoacetate) contain an even number of carbons that provide acetyl-CoA, which can be oxidized in the TCA cycle.7,8 However, even-carbon ketones cannot provide substrate for net biosynthesis because the input of two carbons provided by each acetyl-CoA molecule is lost at the completion of one TCA cycle. Consequently, supplementation with even-carbon fatty acids or their derivative substrates cannot support neurotransmitter synthesis or other synthetic needs.

In contrast, anaplerosis, the combined flux through all reactions converging into the TCA cycle that have the effect of increasing the mass of TCA cycle intermediates, can support net biosynthesis. For example, pyruvate carboxylase (EC 6.4.1.1) is a key anaplerotic enzyme in the brain tissue. However, the availability of pyruvate is likely to be compromised in disorders associated with diminished brain glucose transport and reduced glycolysis. An alternative route of entry into the TCA cycle is via the reaction catalyzed by propionyl-CoA carboxylase (EC 6.4.1.3). This reaction is relevant because exogenously administered odd-carbon triglycerides such as triheptanoin could provide substrate for anaplerosis via the b-oxidation of carbons 1 to 4, which generates two molecules of acetyl-CoA and one molecule of propionyl-CoA. The latter can enter the TCA cycle through propionyl-CoA carboxylase.9–11 Triheptanoin has been previously used in disorders of mitochondrial fat oxidation, glycogen storage, and pyruvate metabolism.10,12 Additionally, it has been reported that heptanoate derived from triheptanoic diet may exert anticonvulsant effects in epileptic animal models while refilling depleted brain TCA cycle intermediates.11 As illustrated in Figure 1, the standard, whole-organ notion that explains these

1 Rare Brain Disorders Clinic and Laboratory, Department of Neurology and Neurotherapeutics, Dallas, Texas, USA; 2Department of Radiology, The University of Texas Southwestern Medical Center, Dallas, Texas, USA; 3Advanced Imaging Research Center, The University of Texas Southwestern Medical Center, Dallas, Texas, USA; 4Department of Internal Medicine, The University of Texas Southwestern Medical Center, Dallas, Texas, USA; 5Department of Physiology, The University of Texas Southwestern Medical Center, Dallas, Texas, USA and 6Department of Pediatrics, The University of Texas Southwestern Medical Center, Dallas, Texas, USA. Correspondence: Professor JM Pascual, Rare Brain Disorders Clinic and Laboratory, Department of Neurology and Neurotherapeutics, The University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390-8813, USA. E-mail: [email protected] This study was supported by Fundacio´n Caja Madrid (IMV) and NIH Grants NS077015 (JMP), RR002584 and EB000461 (CRM, JMP), F32NS065640 (LBG), and Dallas Women’s Foundation (Billingsley Fund) (JMP). Received 25 April 2012; revised 27 August 2012; accepted 3 September 2012; published online 17 October 2012

Brain heptanoate metabolism I Marin-Valencia et al

176

Figure 1. Schematic diagram of [5,6,7-13C3]heptanoate and [3,4,5-13C]C5-ketone metabolism in the liver and brain. Standard metabolic model depicting that dietary [5,6,7-13C3]heptanoate can be metabolized in the liver to produce [3,4,5-13C3]C5-ketones, which are subsequently released to the bloodstream. C5-ketones cross the blood–brain barrier through monocarboxylate transporters and are metabolized in the brain. On the other hand, heptanoate can directly cross the blood–brain barrier by passive diffusion and be oxidized in the brain as well. In the brain, [5,6,7-13C3]heptanoate and [3,4,5-13C3]C5-ketones produce unlabeled acetyl-CoA (2 molecules from heptanoate, one from C5-ketones) and one molecule of [1,2,3-13C3]propionyl-CoA. Acetyl-CoA is further oxidized in the tricarboxylic acid (TCA) cycle and [1,2,3-13C3]propionyl-CoA can be converted to [1,2,3-13C3]methyl-malonyl-CoA and this can be converted into [1,2,3-13C3]succinyl-CoA or [2,3,4-13C3]succinyl-CoA, since this molecule is structurally symmetrical. As a consequence of the first cycle through the TCA cycle, glutamate and glutamine will be labeled in carbons 2 and 3 or in carbons 1,2 and 3. Note that carbons 4 and 5 of glutamate are derived directly from acetyl-CoA and carbons 1, 2 and 3 originate from oxaloacetate. Potential hepatic or glial 4-carbon ketone body formation and subsequent metabolism by the brain are not represented. Hpt, heptanoate; C5-KB, C5 ketone bodies; Ac-CoA, acetyl-CoA; CIT, citrate; a-KG, a-ketoglutarate; GLU, glutamate; GLN, glutamine; Prop-CoA, propionyl-CoA; SucCoA, succinyl-CoA. C#: carbon labeled in position #. Open circle: carbon 12, closed circle: carbon 13.

findings is that dietary triheptanoin is metabolized to plasma heptanoate and C5 ketones (b-ketopentanoate and b-hydroxypentanoate),10 both of which can enter the TCA cycle at the level of acetyl-CoA and propionyl-CoA. According to this principle, heptanoate metabolized directly in the brain or partially in the periphery to C5 ketones, followed by transfer to the brain11,13 could drive anaplerosis both in the normal brain and in the setting of glucose transport defects or insufficient glucose accumulation. The liver, however, can avidly scavenge odd-carbon fatty acids for gluconeogenesis, which may also result in therapeutic benefit by increasing glucose availability in the bloodstream. Thus, in spite of the intense interest in the possible benefits of heptanoate that these contentions have kindled, there is no direct evidence that either heptanoate itself or C5 ketones contribute to metabolism through anaplerosis in the brain in vivo. Analysis of heptanoate metabolism in vivo using 14C or 13C tracer methods is challenging because glucose generated in the liver can be enriched with the tracer and metabolism of these liver-13C-enriched glucose molecules by the brain can complicate the analysis of the metabolism of heptanoate or pentanoate equivalents. An alternative is to select a predicted heptanoate labeling pattern that allows for the distinction between the oxidation of glucose (derived from heptanoate) in the brain from the oxidation of heptanoate itself. In this work, we sought to characterize the metabolism of [5,6,7-13C3]heptanoate in the brain of normal conscious mice and of a transgenic antisense mouse model of G1D that exhibits the common epileptic phenotype typical of the human glucose transporter defect.14 Administration of heptanoate labeled in this fashion is followed by oxidation of [5,6,7-13C3]heptanoate or [3,4,5-13C3]C5-ketone derivatives in the Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

brain without generation of [1,2-13C2]acetyl-CoA, and, consequently, without enrichment of carbons 4 or 5 of glutamine or glutamate. Using this approach, we demonstrate that plasma glucose becomes enriched with 13C in the presence of intravenously infused [5,6,7-13C3]heptanoate as expected, indicating that the understanding of any effects of heptanoate therapy must consider its effects on glucose production. Nevertheless, the fractional 13C enrichment in plasma glucose was small. The labeling patterns in glutamate and glutamine in the brain tissue demonstrate that heptanoate and/or C5 ketones are metabolized in the brain in both animal groups, and glucose generated by gluconeogenesis from 13C-enriched heptanoate is also oxidized by the brain of both animal groups. In both mice groups, glutamine concentration, its overall 13C resonance, and its 13 C enrichment (primarily M þ 3) were higher than that of glutamate, indicating that heptanoate or its derivatives are directly metabolized in the brain, in a reaction thought to occur primarily in glia.

MATERIALS AND METHODS Infusion of [5,6,7-13C]Heptanoate This study was approved by the Institutional Animal Care and Use Committee of UT Southwestern Medical Center. Normal C57BL/6J mice (n ¼ 4, weight ¼ 21.5±0.6 g, age ¼ 4.2±0.2 months; all ranges and errors hereafter are expressed as standard error of the mean (s.e.m.) unless otherwise indicated) and transgenic GLUT-1 mice (n ¼ 6, weight: 17.9±3.5 g, age: 3.1±0.3 months)14 were used for the study. A cannulation and infusion protocol was used as previously described.15,16 Briefly, the right jugular vein was aseptically cannulated under intra& 2013 ISCBFM

Brain heptanoate metabolism I Marin-Valencia et al

177 peritoneal anesthesia provided by ketamine (100 mg/kg) and xylazine (10 mg/kg). The catheters were filled with glycerol and heparin (500 U/mL) to prevent clotting. After recovery, mice were individually housed under standard animal care conditions with ad libitum access to water and food. Three days post-cannulation, the mice were habituated and confined to a cylindrical Lucite cage to prevent ambulation and the catheter purged with 2 mL saline solution. A 150-mmol/L solution of [5,6,7-13C3]heptanoate (Na salt, 99% 13C enrichment, Sigma-Aldrich St Louis, MO, USA) was prepared (pH: 7.5 to 7.8). Heptanoate was then intravenously infused for 35 minutes in each conscious mouse at a rate that provided 50% of the caloric requirement. The infusion rates, calculated according to the weight of each mouse and the caloric content of heptanoate (7.7 kcal/g) as previously described,17 were 95.9±3.9 mmol of heptanoate per minute per kg. Animals were decapitated at the end of infusion and forebrains were rapidly removed and hemispheres manually dissected (o15 seconds), frozen in liquid nitrogen and stored at  85 1C. One hemisphere was used for nuclear magnetic resonance (NMR) studies and the other hemisphere for gas chromatography-mass spectrometry (GC-MS)/liquid chromatographymass spectrometry (LC-MS) analysis. Blood was simultaneously collected from the cervical stump, placed in a precooled microfuge tube with 2 mL of heparin (500 U/L), centrifuged at 3,300 g for 3 minutes to extract the plasma. The plasma was frozen in liquid nitrogen and stored at  85 1C. Blood glucose concentration was measured before and after the infusion with a Contour glucometer (Bayer Healthcare, Tarrytown, NY, USA)18 from each mouse tail.

Preparation of Tissue Extracts The perchloric acid extraction protocol was previously reported by us.15 Briefly, the brain hemispheres (B150 mg) were finely grounded in a mortar under liquid nitrogen. Perchloric acid (4%; 1:4 w/v) was added to each sample, followed by centrifugation at 47,800 g for 15 minutes. The supernatant was transferred to a new tube where chloroform/ tri-noctylamine (78%/22%; v/v) was added in 1:2 volumetric ratio to a pH of 6. The samples were centrifuged at 3,300 g for 15 minutes, the aqueous phase removed and transferred to a microfuge tube and then lyophilized. In all, 200 mL of deuterium oxide (99.96%; Cambridge Isotope Laboratories, Andover, MA, USA) was added to each sample and the pH adjusted to 7.0 with 2 to 3 mL of 1 M sodium deuteroxide (99.5%; Cambridge Isotope Laboratories). The pH-neutral samples were then centrifuged at 18,400 g for 1 minute and the supernatant removed and placed into a 3-mm NMR tube for subsequent NMR analysis.

Nuclear Magnetic Resonance Spectroscopy 13 C-nuclear magnetic resonance spectra were obtained as previously reported.15 In brief, proton-decoupled 13C spectra were acquired on a 600MHz Oxford magnet and Varian VNMRS Direct Drive console using a 3-mm broadband probe (Varian, Palo Alto, CA, USA). Proton decoupling was performed using a Waltz-16 sequence. 13C-nuclear magnetic resonance spectroscopy parameters included a 451 flip angle per transient, a relaxation delay of 1.5 seconds, an acquisition time of 1.5 seconds, and a spectral width of 36.7 kHz. Samples were spun at 20 Hz and 25 1C. A heteronuclear 2H lock was used to compensate for magnet drift during data acquisition. To achieve adequate signal-to-noise in the brain spectra, the number of scans acquired were typically 10,000 to 15,000.

acyl-CoA as previously described.21,22 Plasma samples were assayed for the mass distribution of glucose as previously reported.23–25 Correction of measured mass isotopomer distributions for natural enrichment was performed using the CORMAT software.26

Statistical Analysis Statistical analyses were performed using a two-sample t-test not assuming equal variances using SPSS Graduate Pack version 18.0 (SPSS, Chicago, IL, USA). All data are reported as mean±s.e.m. except where stated.

RESULTS 13 C Enrichment of Plasma Glucose Before infusion, blood glucose levels were 7.2±0.5 and 5.8±0.4 mmol/L in normal and G1D mice, respectively, and were not statistically different between the two genotypes. Following infusion, blood glucose levels were significantly elevated both in normal mice, at 15.3±2.4 mmol/L, and in G1D animals, at 9.6±1.2 mmol/L, again without statistical difference between the two genotypes. The presence of significant excess enrichment over natural abundance enrichment, illustrated in Figure 2, demonstrates gluconeogenesis from [5,6,7-13C3]heptanoate in the liver. Of note, the most highly enriched 13C-glucose fraction was M þ 2 and M þ 3, as expected, since gluconeogenesis from [1,2,3-13C3]succinyl-CoA leads to the generation of oxaloacetate labeled in carbons 1,2,3 or 2,3,4, and oxaloacetate in turn produces phosphoenolpyruvate labeled in carbons 1,2,3 or 2 and 3. No differences were detected in glucose enrichment between normal control and G1D animal groups (Figure 2). 13

C-Nuclear Magnetic Resonance Spectra Typical proton-decoupled 13C spectra from extracts of the brain hemispheres are shown in Figure 3 for control and G1D mice. In both groups of animals, the glutamate resonance at carbon 4 (34.2 p.p.m.) displayed a prominent doublet due to spin–spin coupling between carbons 4 and 5 (J45), which was comparable between both animal groups (Figure 4D). The measured coupling constant, B55 Hz, excludes a contribution from coupling between carbons 3 and 4. Consequently, the spectra of glutamate demonstrate that [1,2-13C2]acetyl-CoA was generated and oxidized in the brain. Since glucose was multiply labeled as a consequence of gluconeogenesis and this labeling pattern of heptanoate (or its 5-carbon products) cannot generate enriched acetyl-CoA, these spectra demonstrate that 13C-enriched

Analysis of Nuclear Magnetic Resonance Spectra Nuclear magnetic resonance spectra were analyzed as previously described.15 Briefly, analyses were performed with ACD/Spec Manager 11.0 software (Advanced Chemistry Development, Toronto, ON, Canada). Time-series free induction decays were zero-filled and multiplied by an exponential weighting function before Fourier transformation. Metabolite peaks were then identified based on chemical shift position referenced to the glutamate C4 singlet at 34.2 p.p.m. Each peak was then fitted with a Gauss-Lorentz function and the area measurements for each fitted resonance peak and their multiplets estimated. For each isotopomer, multiplet areas were defined as a fraction of the total resonance area for that carbon signal.19,20

GC-MS/LC-MS Analysis The brain hemispheres (B150 mg) were assayed at the Mouse Metabolic Phenotypic Center at Case Western Reserve University for the concentration and/or mass analysis of TCA cycle intermediates, amino acids, and & 2013 ISCBFM

Figure 2. Mass isotopomers of 13C-glucose in plasma. Plasma 13Cglucose was enriched from [5,6,7-13C3]heptanoate in both normal and mutant (glucose transporter type I deficiency, G1D) animal groups. Data were corrected for the natural abundance of 13C. Glucose was mostly enriched in M þ 2 (Control: 4.63%±1.5%; G1D: 5.28%±1.22%) and M þ 3 (Control: 3.65%±0.86%; G1D: 3.45%±0.69%). No statistical differences were detected between both animal groups. Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

Brain heptanoate metabolism I Marin-Valencia et al

178

Figure 3. 13C-NMR (nuclear magnetic resonance) spectra of brain hemispheres from normal and glucose transporter type I deficiency (G1D) mice. Proton-decoupled 13C-NMR spectra were acquired from extracts of brain tissue taken from animals infused with [5,6,7-13C3]heptanoate. Spectra from both animals demonstrate 13C labeling in brain glutamate and glutamine in carbons 2, 3 and 4 and lactate in carbons 2 and 3. In both animals, the glutamate and glutamine 13C labeling pattern differed from one another in all carbon positions, suggesting that heptanoate metabolism is compartmentalized in the brain. In carbons 2 and 3, the total signal from 13C-multiplets was more prominent in glutamine compared with glutamate. In carbon 4, the glutamate resonance was greater than that of glutamine. A doublet due to J45 (arising from coupling between carbons 4 and 5) was present in glutamate but not glutamine. Of note, the resonance of glutamine C2 doublet D2,3 (purple arrows) was statistically higher in G1D animals compared with controls. The lactate resonance was also prominent in the 13C spectrum of both animals. In lactate C2 and C3, the doublet D2,3 was the most abundant signal, indicating other 13C substrates (i.e., glucose) may be oxidized in the brain as a result of gluconeogenesis from [5,6,7-13C3]heptanoate. 1. N-acetyl-aspartate C2, 2. Aspartate C2, 3. Taurine C1, 4. N-acetyl-aspartate C3, 5. GABA C4, 6. Creatine C2, 7. Aspartate C3, 8. Taurine C2, 9. GABA C2, 10. GABA C3, 11. N-acetyl-aspartate C6. GLU, glutamate; GLN, glutamine; LAC, lactate. C#: carbon labeled in position #. Sx, singlet; Dxx, doublet; T, triplet; Q, quartet. The color reproduction of this figure is available on the Journal of Cerebral Blood Flow and Metabolism journal online.

heptanoate supplies gluconeogenesis and that the resulting 13Cenriched glucose is oxidized in the brain of both control and G1D animals. The fact that the lactate carbon 2 and 3 also contains a doublet due to J23 (and a quartet in case lactate C2) further demonstrates that the upstream pyruvate pool was multiply Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

labeled. Interestingly, a doublet due to J45 was not detected in the glutamine C4 signal in any sample. In general, the labeling pattern of glutamate and glutamine was different in G1D compared with wild-type animals. In carbon 2, the overall 13C resonance of glutamine was significantly higher than that of glutamate (Figure 4A) in both animal groups but the 13C resonance in glutamate C4 was greater than glutamine C4 (Figure 3). As illustrated in Figures 3 and 4B, the fractional amount of glutamine doublet D2,3 was significantly larger (Po0.01) in G1D mice relative to normal, and in both animals it was greater than that of the glutamate doublet D2,3. The quartet resonance was also larger in glutamine C2 relative to glutamate (Figures 3 and 4B), although the difference was statistically significant (Po0.01) only for G1D mice. In carbon 3, the glutamine 13C resonance was also greater than the glutamate resonance (Figures 3 and 4A), with no difference between animal groups. As it occurred with carbon 2, the labeling pattern was notably different between both molecules: The doublet (D) in carbon 3 can arise due to J23 or J34 since the coupling constants are virtually identical. The fractional amount of the glutamine C3 doublet was greater relative to that of glutamate and equivalent in both animals (Po0.001) (Figure 4C). Since the doublet D3,4 was not detected in glutamate C4 or glutamine C4 (Figure 3), most, if not all, of the doublet C3 in both molecules derives from the doublet due to J23. [4,5-13C2]glutamine was not detected, even though this metabolite can be observed after infusion of [1,2-13C2]glucose. In addition to a limited aspartate pool (described in further detail below), this may be due to dilution of [1,2-13C2]acetyl-CoA (derived from plasma glucose) with unlabeled acetyl-CoA inside the astrocyte generated after astrocytic degradation of heptanoate. Thus, (astrocytic) acetyl-CoA may derive both from labeled plasma glucose and from unlabeled heptanoate. This can be more manifest in G1D, where heptanoate may become the main contributor to acetyl-CoA generation because of reduced brain glucose accumulation.14,27,28 Within each genetic group, the 13C-multiplet spectra of glutamate and g-aminobutyric acid (GABA) were comparable, suggesting that the GABA pool is derived directly from the glutamate pool (Supplementary Figure 1) in the neuronal compartment since GABA is primarily produced in GABAergic neurons.29 Particularly, GABA C3 and C4 contained a doublet (D3,4) that derives from glutamate labeled in carbons 2,3 and in carbons 1,2 and 3. Although this labeling pattern is consistent with metabolism of [5,6,7-13C3]heptanoate to [1,2,3-13C3]propionyl-CoA followed by entry into the TCA cycle, the identical pattern could occur as a consequence of carboxylation of [1,2,3-13C3]pyruvate derived from glucose. Mass Spectrometry of Brain Glutamate, Glutamine, Succinyl-CoA, and Propionyl-CoA The enrichment of glutamate and glutamine from M þ 0 to M þ 3 is summarized in Table 1. The enrichment of glutamine M þ 3 was approximately three times higher than the enrichment of glutamate M þ 3 (Po0.01) in both animal groups. The glutamine M þ 2 enrichment was also greater than that of glutamate M þ 2 but was not statistically significant (Table 1). No differences were detected on M þ 1. On the other hand, the 13C enrichment of M þ 3 of propionyl-CoA and succinyl-CoA in the brain of control and G1D mice were 70.75%±5.51% and 70.16%±6.51%, and 2.14%±0.31% and 2.28%±1.08%, respectively, with no differences between both animal groups. Metabolism of 13C-Glucose Relative to 13C-Heptanoate The presence of multiplets in carbons 2 and 3 of both glutamate and glutamine in the NMR spectra is consistent with the generation of [1,2,3-13C3]propionyl-CoA from either [5,6,7-13C3]heptanoate and/or [3,4,5-13C3]C5 ketones. This interpretation is & 2013 ISCBFM

Brain heptanoate metabolism I Marin-Valencia et al

179 Table 1.

Mass isotopomers of glutamate and glutamine in the brain

MPE (%)

Mþ0 Mþ1 Mþ2 Mþ3

Normal

G1D

GLU

GLN

GLU

GLN

92.2±0.77 2.87±0.34 3.83±0.28 1.3±0.19

87.95±1.49* 2.76±0.48 5.5±0.65 4.31±0.48**

93.66±0.98 2.49±0.32 3.12± 0.44 0.88±0.25

90.56±0.95* 2±0.28 4.35±0.42 3.4±0.33**

G1D, glucose transporter type I deficiency; GLN, glutamine; GLU, glutamate; MPE, mole percent excess. Differences in the mass isotopomer distribution between glutamate and glutamine in normal and mutant G1D animal groups. Strikingly, the abundance of M þ 3 was approximately three times higher in glutamine relative to glutamate, which suggests that [5,6,7-13C3]heptanoate enriched mostly the glutamine pool. The abundance of M þ 2 was higher in glutamine but not statistically different from the glutamate M þ 2 abundance. No differences in glutamate and glutamine enrichment were detected between animal groups. *Po0.05, **Po0.01.

consistent with the presence of M þ 3 succinyl-CoA detected by mass spectrometry. However, the spectra of glutamate (specifically, the doublet due to J45) demonstrates that pyruvate in the brain was multiply labeled. Therefore, an alternative interpretation of both the 13C-NMR spectra in Figure 3 and the mass spectrometry observations is simply that 13C-enriched glucose derived via gluconeogenesis from [5,6,7-13C3]heptanoate provides the substrate for both pyruvate carboxylation as well as energy production in the brain. For example M þ 3 succinyl-CoA could arise exclusively from oxidation of [1,2-13C2]acetyl-CoA, from carboxylation of [1,2,3-13C3]pyruvate, or from innumerable combinations of enrichments and fluxes. The measured 13C enrichment in plasma glucose shown in Figure 2 fixes the upper limit on the amount of either [2,3-13C]pyruvate or [1,2,3-13C]pyruvate that can be present in the brain (as these are the two labeling patterns that can be metabolized to [1,2-13C2]acetyl-CoA). The measured M þ 3 in propionyl-CoA, B70% in both animal groups, proves that carbon skeletons from [5,6,7-13C3]heptanoate and/or [3,4,5-13C3]C5-ketones are metabolized in the brain to the level of propionyl-CoA. The detection of M þ 3 propionyl-CoA does not, in itself, demonstrate significant flux from heptanoate carbons through propionyl-CoA carboxylase. However, metabolism of 13Cenriched glucose under these conditions is not consistent with the observed mass and NMR spectra and, for this reason, we conclude that [5,6,7-13C3]heptanoate provides carbon directly for anaplerosis in the brain in both groups of animals.

Figure 4. Analysis of 13C spectra in normal and glucose transporter type I deficiency (G1D) mice. (A) Overall, the 13C abundance of glutamine in carbons 2 and 3 was greater than glutamate; this difference was more prominent in G1D mice although did not reach statistical significance. Carbon 4 was not illustrated since glutamine C4 did not contain multiplets. (B) The fractional amount of the glutamine C3 doublet was significantly higher than the glutamate C3 doublet in both mice with no difference between animal groups. (C) In carbon 2, the glutamine doublet D2,3 was statistically more abundant in G1D mice relative to normal and greater than the glutamate doublet D2,3 in both animals. No differences were observed in the fractional amount of the doublet D1,2 within and between animal groups. The quartet (Q) was more prominent in glutamine than glutamate, although it was statistically significant only in the G1D group. (D) No difference was detected between animals relative to the fractional amount of glutamate doublet D4,5. Values are expressed as mean±s.e.m. **Po0.01,***Po0.001. & 2013 ISCBFM

Concentration of Tricarboxylic Acid Cycle Intermediates, Amino Acids, and Acetyl-CoA To determine the impact of heptanoate infusion on intermediates and products of the brain TCA cycle, their concentration was measured in the contralateral brain hemisphere in both animal groups as depicted in Table 2. Strikingly, brain acetyl-CoA levels were statistically higher in the G1D group relative to the normal control animals, which indicates that heptanoate or C5 ketones derivatives increase acetyl-CoA concentrations in the transgenic G1D mouse above those found in the normal mouse. Of note, basal (uninfused) brain amino-acid contents are preserved in G1D relative to normal mice14 (unpublished data). In agreement with the NMR data, the levels of glutamine were higher than that of glutamate in both animal groups and more markedly so in G1D mice, which is consistent with the notion that heptanoate and C5 ketones are primarily metabolized by glia. In addition, glutamine levels were significantly higher in the G1D mouse compared with Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

Brain heptanoate metabolism I Marin-Valencia et al

180 Table 2.

Concentration of amino acids, TCA cycle intermediates, and acetyl-CoA in the brain following heptanoate infusion Substrates

Normal

G1D

Ac-CoA Glutamate Glutamine GABA Succinate Fumarate Malate

3.8±0.12 3.96±0.41 4.50±0.16 3.16±0.45 0.15±0.02 0.22±0.03 0.20±0.01

4.9±0.29*** 3.88±0.21 6.00±0.17*** 3.43±0.10 0.14±0.01 0.18±0.02 0.19±0.01

G1D, glucose transporter type I deficiency; GABA, g-aminobutyric acid; TCA, tricarboxylic acid. The concentration of acetyl-CoA (nmol/g of tissue) was higher in the G1D group relative to the normal group. The concentration of glutamine (mmol/g of tissue) was greater in G1D mice relative to normal mice and in both animal groups glutamine levels were higher than glutamate levels (determined in mmol/g of tissue), which suggests that heptanoate and/or C5-KB are primarily metabolized by glia and that this process is more robust in mutant animals. No differences were detected in the levels of glutamate or GABA between animals. The concentration of TCA cycle intermediates (mmol/g of tissue) such as succinate, fumarate and malate was comparable between both groups. ***Po0.001.

normal, which correlates with the presence of reactive astrocytes with preserved expression of glutamine synthetase in the G1D brain, probably as a potential compensatory mechanism of glucose transporter deficiency in these cells,14 No differences were identified in the levels of glutamate or GABA across animal groups. The concentration of succinate, fumarate, and malate was also equivalent between both animal groups. DISCUSSION Since its first use to treat fatty acid oxidation defects,9 triheptanoin has been applied in multiple energy metabolism disorders in which anaplerosis is thought to be impaired.10,20,30,31 The rationale for the use of triheptanoin in patients with long-chain fatty acid oxidation defects was that the poor response to even-carbon medium-chain triglycerides (MCT) may be due to the loss of TCA cycle intermediates from skeletal muscle and heart mitochondria. The resulting oxaloacetate depletion would then limit the oxidation of MCT-derived acetyl-CoA. Triheptanoin may be a more effective therapy in this condition because it provides propionyl-CoA (which is anaplerotic by replenishing the oxaloacetate pool via succinyl-CoA) and acetyl-CoA, which can be further oxidized in the TCA cycle. More recently, it was hypothesized that certain forms of epilepsy are associated with depletion of TCA cycle intermediates.11 Triheptanoin exhibited anticonvulsant effects in animal models of chronic epilepsy by increasing the seizure threshold.11 However, the mechanisms of triheptanoin metabolism in the brain remained to be elucidated. In this study, the metabolism of heptanoate was examined in the conscious brain of normal mice and in a transgenic mouse model of the common epileptic phenotype of G1D. A central problem was to distinguish the direct metabolism in the brain of heptanoate (or equivalent molecules such as pentanoate) from the metabolism of heptanaote-derived glucose in the brain. Since both pathways provide for delivery of 13C into the TCA cycle, a simple detection of spin-coupled multiplets is not discriminative. Furthermore, 13C-enriched glucose could generate 13C-enriched pyruvate that enters TCA cycle via an anaplerotic pathway. In this study multiply enriched glutamate and glutamine was detected by both mass spectrometry and 13C-NMR spectroscopy. This study demonstrates both indirect metabolism of [5,6,7-13C3]heptanaote via initial conversion to multiply enriched glucose, as well as direct metabolism of [5,6,7-13C3]heptanoate Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

and/or [3,4,5-13C3]C5-ketones via anaplerosis in the brain. The results further suggest that heptanoate and/or C5 ketones are metabolized primarily through the glial TCA cycle and that relatively little utilization occurred in neurons in both animals (Figure 5). This conclusion is based on the finding that 13C-labeled heptanoate and/or C5 ketones were preferentially metabolized to glutamine, as the concentration and the overall 13C enrichment of glutamine (primarily M þ 3) was larger than that of glutamate (Table 1). The preferential labeling of glutamine indicates that odd-carbon fatty acids were metabolized in cells that express glutamine synthetase (EC 6.3.1.2), such as astroglia and oligodendroglia.32,33 These findings are in agreement with the general contention that fatty acids are primarily metabolized in glia.34,35 For example, the injection of 14C-labeled propionate to the striatum or cortex of mice resulted in five to six times higher labeling of glutamine relative glutamate.34 Ebert et al.35 reported that the metabolism of 13C-octanoate in the brain occurs predominantly in the glial compartment based on the greater enrichment of glutamine over glutamate. Together, these findings suggest that the potential anticonvulsant effect of triheptanoin may begin or take place primarily in glia. Conversely, heptanoate-derived 13C-glucose was primarily metabolized in neurons, as demonstrated by the characteristic coupling in the carbon 4 in glutamate, but not in glutamine (Figures 3 and 5). Glia may also have the capacity to generate ketones,36 a phenomenon observed in cultured immature astrocytes,37 while hepatic ketogenesis could in principle proceed from labeled glucose originated by gluconeogenesis and translate into ketonemia for labeled hydroxybutyrate and acetoacetate. Although glucose is not a ketogenic substrate physiologically or under conditions that favor an increase in plasma glucose (such as heptanoate infusion; see below), circulating hydroxybutyrate and acetoacetate can be consumed by astrocytes, such that part of the observed labeling in glutamate and glutamine may also arise from these substrates. The gluconeogenic properties of heptanoate have been previously observed in rats infused with 13C-heptanoate, in which the rate of glucose synthesis in the liver manifested an increase of about 1.5 times compared with rats infused with saline.17 Under these experimental conditions, only glutamate C4 (and not glutamine C4) exhibited a prominent doublet D4,5, which, together with the fact that glucose became predominantly labeled in M þ 2, indicates that heptanoate-derived 13C-glucose was largely metabolized in neurons. On the other hand, the finding that the 13 C-multiplet pattern was equivalent for glutamate and GABA in all carbons indicates that GABA is produced from glucose-derived and also from heptanoate-derived glutamate in neurons in both animal groups (Supplementary Figure 1). The lack of labeling in glutamine may also reflect diminished transamination of labeled aspartate,38 despite the normal aspartate brain contents found in uninfused control and G1D mice.14 Metabolism of heptanoate in the brain differed between normal and G1D mice. GLUT-1 serves as the facilitative transporter for glucose across the blood–brain barrier and astrocyte membrane. It has been reported that GLUT-1 deficiency induces reactive astrocytosis,39 defined as an increase in the number and the size of cells expressing glial fibrillary acidic protein.40 Reactive astrocytes that retain expression of glutamine synthetase are present in the brain of G1D mice,14 consistent with a greater rate of glutamine synthesis from heptanoate in G1D relative to normal brain. Anaplerosis from propionyl-CoA precursors, such as heptanoate, is of potential interest based on its capacity to replenish TCA cycle intermediates and the transmitter glutamate and its precursor glutamine.12,13,34 It has been hypothesized that, in certain forms of epilepsy, TCA cycle intermediates may be deficient in the context of neuronal hyperexcitability.11 Willis et al.11 reported that dietary triheptanoin restored malate levels in the brain of pilocarpineinduced status epilepticus mice, who also experienced an increase & 2013 ISCBFM

Brain heptanoate metabolism I Marin-Valencia et al

181

Figure 5. Heptanoate stimulation of brain metabolism. The higher concentration, 13C resonance and 13C enrichment of glutamine relative to glutamate indicates that [5,6,7-13C3]heptanoate supplied [1,2,3-13C3]propionyl-CoA in the glial compartment, either directly or after conversion to C5 ketones in the liver. The presence of J45 in glutamate and not in glutamine, and the measured 13C enrichment in plasma glucose indicates that [5,6,7-13C3]heptanoate was metabolized to multiply enriched glucose followed by oxidation via pyruvate dehydrogenase primarily by neurons. This labeling pattern is also compatible with metabolism of labeled glucose to [1,2-13C2]acetyl-CoA by glia followed by dilution with unlabeled acetyl-CoA derived from heptanoate oxidation. Thus, (astrocytic) acetyl-CoA may be derived both from labeled plasma glucose and from unlabeled carbons of heptanoate. The different labeling pattern of glutamine relative to glutamate supports the notion that heptanoate and/or C5 ketone metabolism is compartmentalized in the brain. Additionally, the GABA doublet D3,4 was observed in the 13C spectra (Supplementary Figure 1), which indicates that part of the glial glutamine produced from heptanoate is transferred to neurons. Any potential contribution of hepatic or glial 4-carbon ketone body formation to the labeling patterns observed in the brain tissue is not represented. Hpt, heptanoate; C5-KB, C5 ketone bodies; Glc, glucose; PYR, pyruvate; Ac-CoA, acetyl-CoA; CIT, citrate; a-KG, a-ketoglutarate; GLU, glutamate; GLN, glutamine; Prop-CoA, propionyl-CoA; SucCoA, succinyl-CoA. C#: carbon labeled in position #. Open circle: carbon 12, closed circle: carbon 13.

in seizure threshold. However, the rate by which propionyl-CoA is converted to succinyl-CoA is assumed to be low based on the small detectable 13C labeling of glutamate and glutamine upon injection of 13C-propionate or [U-13C]isoleucine.34,41 In agreement with these studies, the enrichment of succinyl-CoA M þ 3 relative to its precursor propionyl-CoA M þ 3 measured in this study was B3% in the brain of both animals, indicating that the rate of conversion of propionyl-CoA to succinyl-CoA is probably small in both normal and G1D brains. In conclusion, the present work aimed to characterize the metabolism of heptanoate in the conscious brain of normal and G1D mice. The simplest interpretation of the results (Figure 5) support several conclusions: First, heptanoate is a substrate for gluconeogenesis and plasma glucose derived from heptanoate is oxidized in the brain. Any metabolic (or therapeutic, especially when referring to disorders associated with impaired brain glucose accumulation) effects attributed to heptanoate must consider this pathway in vivo. Second, heptanoate itself (or C5 derivatives) may supply propionyl-CoA in the brain and thereby directly support & 2013 ISCBFM

biosynthetic reactions that cannot be supplied by even-carbon ketones, even in the setting of even-carbon ketogenesis by liver or glia. Finally, heptanoate and/or its by-product C5 ketones are primarily metabolized by glia. These results support the notion that the anticonvulsant effects of triheptanoin may stem primarily from glial metabolism. Further studies are necessary to elucidate the clinical effects of triheptanoin in G1D and its potential applications in other brain energy metabolism disorders.

DISCLOSURE/CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS The authors are grateful to the National Mouse Metabolic Phenotyping Center at Case Western Reserve University (supported by NIDDK 1U24DK076174) and to Drs Michelle Puchowicz and Henri Brunengraber for advice and for the assay of brain

Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

Brain heptanoate metabolism I Marin-Valencia et al

182 metabolites. The authors also acknowledge helpful advice from and discussions with Dr Charles R Roe.

REFERENCES 1 Schurr A. Neuronal energy requirements. In: Walz W (ed) The Neuronal Environment: Brain Homeostasis in Health and Disease. Humana Press: New Jersey, 2002, pp 25. 2 Leary LD, Wang D, Nordli Jr. DR, Engelstad K, De Vivo DC. Seizure characterization and electroencephalographic features in Glut-1 deficiency syndrome. Epilepsia 2003; 44: 701–707. 3 Pascual JM, Campistol J, Gil-Nagel A. Epilepsy in inherited metabolic disorders. Neurologist 2008; 14(6 Suppl 1): S2–S14. 4 Wang D, Pascual JM, Yang H, Engelstad K, Jhung S, Sun RP et al. Glut-1 deficiency syndrome: clinical, genetic, and therapeutic aspects. Ann Neurol 2005; 57: 111–118. 5 Wang D, Pascual JM, Yang H, Engelstad K, Mao X, Cheng J et al. A mouse model for Glut-1 haploinsufficiency. Hum Mol Genet 2006; 15: 1169–1179. 6 Gruetter R, Seaquist ER, Ugurbil K. A mathematical model of compartmentalized neurotransmitter metabolism in the human brain. Am J Physiol Endocrinol Metab 2001; 281: E100–112. 7 Yudkoff M, Daikhin Y, Melo TM, Nissim I, Sonnewald U. The ketogenic diet and brain metabolism of amino acids: relationship to the anticonvulsant effect. Annu Rev Nutr 2007; 27: 415–430. 8 Bough K. Energy metabolism as part of the anticonvulsant mechanism of the ketogenic diet. Epilepsia 2008; 49(Suppl 8): 91–93. 9 Roe CR, Sweetman L, Roe DS, David F, Brunengraber H. Treatment of cardiomyopathy and rhabdomyolysis in long-chain fat oxidation disorders using an anaplerotic odd-chain triglyceride. J Clin Invest 2002; 110: 259–269. 10 Roe CR, Mochel F. Anaplerotic diet therapy in inherited metabolic disease: therapeutic potential. J Inherit Metab Dis 2006; 29: 332–340. 11 Willis S, Stoll J, Sweetman L, Borges K. Anticonvulsant effects of a triheptanoin diet in two mouse chronic seizure models. Neurobiol Dis 2010; 40: 565–572. 12 Marin-Valencia I, Roe CR, Pascual JM. Pyruvate carboxylase deficiency: mechanisms, mimics and anaplerosis. Mol Genet Metab 2010; 101: 9–17. 13 Borges K, Sonnewald U. Triheptanoin-A medium chain triglyceride with odd chain fatty acids: a new anaplerotic anticonvulsant treatment? Epilepsy Res 2011. 14 Marin-Valencia I, Good LB, Ma Q, Duarte J, Bottiglieri T, Sinton CM et al. Glut1 deficiency (G1D): epilepsy and metabolic dysfunction in a mouse model of the most common human phenotype. Neurobiol Dis 2012; 48: 92–101. 15 Marin-Valencia I, Good LB, Ma Q, Jeffrey FM, Malloy CR, Pascual JM. High-resolution detection of (1)(3)C multiplets from the conscious mouse brain by ex vivo NMR spectroscopy. J Neurosci Methods 2012; 203: 50–55. 16 Marin-Valencia I, Good LB, Ma Q, Malloy CR, Patel MS, Pascual JM. Cortical metabolism in pyruvate dehydrogenase deficiency revealed by ex vivo multiplet (13)C NMR of the adult mouse brain. Neurochem Int 2012; 100: 239–244. 17 Gu L, Zhang GF, Kombu RS, Allen F, Kutz G, Brewer WU et al. Parenteral and enteral metabolism of anaplerotic triheptanoin in normal rats. II. Effects on lipolysis, glucose production, and liver acyl-CoA profile. Am J Physiol Endocrinol Metab 2010; 298: E362–371. 18 Suh JM, Zeve D, McKay R, Seo J, Salo Z, Li R et al. Adipose is a conserved dosagesensitive antiobesity gene. Cell Metab 2007; 6: 195–207. 19 Malloy ADSaCM. 13C Isotopomer analysis of glutamate. A NMR method to probe metabolic pathways intersecting in the citric acid cycle. In Berliner P-MLR L. J (ed) Biological Magnetic Resonance. In Vivo Carbon-13 NMR vol. 15. Kluwer Academic/ Plenum Publishers: New York, 1998, pp 59–97. 20 Roe CR, Yang BZ, Brunengraber H, Roe DS, Wallace M, Garritson BK. Carnitine palmitoyltransferase II deficiency: successful anaplerotic diet therapy. Neurology 2008; 71: 260–264.

21 Zhang GF, Kombu RS, Kasumov T, Han Y, Sadhukhan S, Zhang J et al. Catabolism of 4-hydroxyacids and 4-hydroxynonenal via 4-hydroxy-4-phosphoacyl-CoAs. J Biol Chem 2009; 284: 33521–33534. 22 Kombu RS, Brunengraber H, Puchowicz MA. Analysis of the citric acid cycle intermediates using gas chromatography-mass spectrometry. Methods Mol Biol 2011; 708: 147–157. 23 Kinman RP, Kasumov T, Jobbins KA, Thomas KR, Adams JE, Brunengraber LN et al. Parenteral and enteral metabolism of anaplerotic triheptanoin in normal rats. Am J Physiol Endocrinol Metab 2006; 291: E860–866. 24 Leclerc J, Des Rosiers C, Montgomery JA, Brunet J, Ste-Marie L, Reider MW et al. Metabolism of R-beta-hydroxypentanoate and of beta-ketopentanoate in conscious dogs. Am J Physiol 1995; 268(3 Pt 1): E446–452. 25 Previs SF, Hallowell PT, Neimanis KD, David F, Brunengraber H. Limitations of the mass isotopomer distribution analysis of glucose to study gluconeogenesis. Heterogeneity of glucose labeling in incubated hepatocytes. J Biol Chem 1998; 273: 16853–16859. 26 Fernandez CA, Des Rosiers C, Previs SF, David F, Brunengraber H. Correction of 13C mass isotopomer distributions for natural stable isotope abundance. J Mass Spectrom 1996; 31: 255–262. 27 Pascual JM, Van Heertum RL, Wang D, Engelstad K, De Vivo DC. Imaging the metabolic footprint of Glut1 deficiency on the brain. Ann Neurol 2002; 52: 458–464. 28 Pascual JM, Wang D, Hinton V, Engelstad K, Saxena CM, Van Heertum RL et al. Brain glucose supply and the syndrome of infantile neuroglycopenia. Arch Neurol 2007; 64: 507–513. 29 Fogarty M, Grist M, Gelman D, Marin O, Pachnis V, Kessaris N. Spatial genetic patterning of the embryonic neuroepithelium generates GABAergic interneuron diversity in the adult cortex. J Neurosci 2007; 27: 10935–10946. 30 Mochel F, DeLonlay P, Touati G, Brunengraber H, Kinman RP, Rabier D et al. Pyruvate carboxylase deficiency: clinical and biochemical response to anaplerotic diet therapy. Mol Genet Metab 2005; 84: 305–312. 31 Roe CR, Bottiglieri T, Wallace M, Arning E, Martin A. Adult Polyglucosan Body Disease (APBD): anaplerotic diet therapy (Triheptanoin) and demonstration of defective methylation pathways. Mol Genet Metab 2010; 101: 246–252. 32 Martinez-Hernandez A, Bell KP, Norenberg MD. Glutamine synthetase: glial localization in brain. Science 1977; 195: 1356–1358. 33 Warringa RA, van Berlo MF, Klein W, Lopes-Cardozo M. Cellular location of glutamine synthetase and lactate dehydrogenase in oligodendrocyte-enriched cultures from rat brain. J Neurochem 1988; 50: 1461–1468. 34 Nguyen NH, Morland C, Gonzalez SV, Rise F, Storm-Mathisen J, Gundersen V et al. Propionate increases neuronal histone acetylation, but is metabolized oxidatively by glia. Relevance for propionic acidemia. J Neurochem 2007; 101: 806–814. 35 Ebert D, Haller RG, Walton ME. Energy contribution of octanoate to intact rat brain metabolism measured by 13C nuclear magnetic resonance spectroscopy. J Neurosci 2003; 23: 5928–5935. 36 Guzman M, Blazquez C. Is there an astrocyte-neuron ketone body shuttle? Trends Endocrinol Metab 2001; 12: 169–173. 37 Auestad N, Korsak RA, Morrow JW, Edmond J. Fatty acid oxidation and ketogenesis by astrocytes in primary culture. J Neurochem 1991; 56: 1376–1386. 38 Pardo B, Rodrigues TB, Contreras L, Garzo´n M, Llorente-Folch I, Kobayashi K et al. Brain glutamine synthesis requires neuronal-born aspartate as amino donor for glial glutamate formation. J Cereb Blood Flow Metab 2011; 31: 90–101. 39 Ullner PM, Di Nardo A, Goldman JE, Schobel S, Yang H, Engelstad K et al. Murine Glut-1 transporter haploinsufficiency: postnatal deceleration of brain weight and reactive astrocytosis. Neurobiol Dis 2009; 36: 60–69. 40 Eddleston M, Mucke L. Molecular profile of reactive astrocytes--implications for their role in neurologic disease. Neuroscience 1993; 54: 15–36. 41 Bak LK, Iversen P, Sorensen M, Keiding S, Vilstrup H, Ott P et al. Metabolic fate of isoleucine in a rat model of hepatic encephalopathy and in cultured neural cells exposed to ammonia. Metab Brain Dis 2009; 24: 135–145.

Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website (http:// www.nature.com/jcbfm)

Journal of Cerebral Blood Flow & Metabolism (2013), 175 – 182

& 2013 ISCBFM