METABOLIC ENERGY BALANCES IN KETOTIC RAT BRAIN

Transcription

1 METABOLIC ENERGY BALANCES IN KETOTIC RAT BRAIN by YIFAN ZHANG Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy Dissertation Advisor: Joseph C LaManna, PhD Department of Biomedical Engineering CASE WESTERN RESERVE UNIVERSITY August, 2013

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the dissertation of Yifan Zhang candidate for the Doctor of Philosophy degree*. Xin Yu, Sc. D Joseph C. LaManna, Ph.D Zhenghong Lee, Ph.D Michelle. A. Puchowicz, Ph.D Gerald. M. Saidel, Ph.D Kingman. P. Strohl, M.D (date) May 14 th, 2013 * We also certify that written approval has been obtained for any proprietary material contained therein. ii

3 Table of Contents METABOLIC ENERGY BALANCES IN KETOTIC RAT BRAIN... I DEDICATION... VII ACKNOWLEDGMENT... VIII ABSTRACT... IX LIST OF FIGURES... XI LIST OF TABLES... XIII ACRONYMS... XIV CHAPTER 1 OVERVIEW OF THE DISSERTATION... 1 CHAPTER 2 BACKGROUND BIOCHEMISTRY OF KETONE BODIES Ketone and ketone bodies Pathways and regulations of ketone bodies metabolism KETOSIS AND METHODS OF INDUCTION NEUROPROTECTION FROM KETOSIS Ketosis as a pre-conditioning for protection Ketosis as a therapy METABOLISM OF GLUCOSE AND KETONE BODIES OVERVIEW OF METHODS TO ESTIMATE THE CMR KETY-SCHMIDT METHOD (MEASUREMENT OF UPTAKE) Compartmental Modeling method (Measurement of reaction) Inherent difficulties to determine the CMR UNPUBLISHED PILOT STUDIES ON CMRGLC AND OXIDATIVE METABOLISM Common mistakes and precautions in determining the CMR glc by FDG-PET Animal Anesthesia System Development...20 iii

4 2.6.3 Unreported CMR glc data Unreported CMR glc meta-analysis data FIGURES AND TABLES CHAPTER 3 KETOSIS PROPORTIONATELY SPARES GLUCOSE UTILIZATION IN BRAIN ABSTRACT INTRODUCTION MATERIALS AND METHODS Animal Model and Diets Anesthesia and Surgery Physiological Parameters Image Acquisition and Blood Sampling Image Processing: Region and Volumes of Interest Parameter Estimation and Calculation of CMR glc RESULTS Physiological parameters Cerebral Glucose Metabolic Rates Meta-analysis of CMR glc in Ketotic Subjects DISCUSSION ACKNOWLEDGEMENTS FIGURES AND TABLES CHAPTER 4 CONTRIBUTIONS OF BRAIN GLUCOSE AND KETONE BODIES TO OXIDATIVE METABOLISM ABSTRACT INTRODUCTION iv

5 4.3 METHODS Animal Preparation and Diets Experimental Design, Tracer Preparation, and Infusions Estimation of the Contribution of Acetoacetate and Glucose to Oxidative Metabolism RESULTS AND DISCUSSIONS Acknowledgments Figures and tables...73 CHAPTER 5 KETONE BODIES SPARES GLUCOSE OXIDATIVE METABOLISM IN DIET-INDUCED KETOSIS IN RAT BRAIN ABSTRACT INTRODUCTION METHODS Animal model and diets Tracer Infusion and tissue collection Analytical method and theory of flux analysis RESULTS Physiological parameters Plasma and BHB tracer enrichments First turn of CAC metabolites fluxes Pyruvate recycling and 2 nd turns of CAC Metabolite concentrations DISCUSSION Changes of oxidative metabolism in ketosis...94 v

6 5.5.2 Shunts to neurotransmitters Alterations of pyruvate recycling ACKNOWLEDGMENT FIGURES AND TABLES CHAPTER 6 CONCLUSIONS & FUTURE WORKS INTRODUCTION ESTIMATION OF THE LUMPED CONSTANT IN KETOTIC RAT BRAINS Objective and specific aims Technical and scientific Challenges OPTIMIZING THE STABLE ISOTOPE STUDIES ON OXIDATIVE METABOLISM IN KETOSIS CONCLUSIONS FIGURES AND TABLES APPENDIX APPENDIX I SAMPLE FILES FOR PET PLASMA INPUT FUNCTIONS (.CRV) AND TIME ACTIVITY CURVES(.TAC) APPENDIX II MATLAB CODE FOR GJEDDE-PATLAK ANALYSIS APPENDIX III FDG-PET MODEL AND LC MEASUREMENT MODEL DEVELOPMENT DERIVATION OF THE COMPETITIVE REACTIONS OF GLUCOSE AND 18 FDG FINDING THE PHOSPHORYLATION RATE OF GLUCOSE AND 18 FDG LINKING THE PHOSPHORYLATION RATE OF GLUCOSE AND 18 FDG ESTIMATION OF 18FDG KINETIC CONSTANTS ESTIMATION OF THE LUMPED CONSTANT (LC) BIBLIOGRAPHY vi

7 Dedication This work is dedicated to my wife and parents. vii

8 Acknowledgment I would like to first thank my research advisor, Dr. Joseph LaManna, for his persistent professional education and support in my learning process. His deep insight in physiology is truly outstanding and of great value to my research. I would also like to thank Dr. Michelle Puchowicz and Dr. Zhenghong Lee, who had provided hand-to-hand guidance to my scientific investigation in analytical biochemistry, ketone body metabolism and radiology. Working with them was a great pleasure. I thank Dr. Xin Yu and Dr. Gerald Saidel, my academic advisor and committee member in Biomedical Engineering, who had been vigorously setting high academic standards for my coursework education, teaching experiences and professional development. Their challenges made me today. Lastly, I thank Dr. Kingman Strohl for his insightful suggestions and criticisms to my work and presentations. These are certainly to benefit me greatly as a young researcher. All my work is not possible without my colleagues and lab friends in the past five years. I would also thank my colleagues, Youzhi Kuang, Dr. Kui Xu, Edwin Vazquez, Sharon Zhang, and Lan Wang. Their technical supports are highly appreciated. I thank my lab student friends, Kevin Train, David Corn, and Donald Harris for working with me. The research projects are supported by the National Institute of Health, R01 HL , R21 NS and Mouse Metabolic Phenotyping Center, MMPC U24 DK viii

9 Metabolic Energy Balances in Ketotic Rat Brain Abstract by YIFAN ZHANG The brain normally uses glucose as its primary fuel, but is able to use ketone bodies as an alternative fuel during fasting, starvation, or feeding of high-fat, low-carb diets. Ketosis, as a physiological state, has been shown to be neuroprotective since the 1920s. The biochemical links between ketosis and neuroprotection has been of interest to clinicians and scientists. To investigate the metabolic mechanism, we hypothesized that 1) the total energy demand (glucose + ketone bodies) is constant during ketosis 2) in chronic ketosis, ketone bodies spare glucose from oxidative metabolism and shunts towards neurotransmitters. Using Positron Emission Tomography (PET) and 2-tissue compartment modeling, we show that the cerebral metabolic rate of glucose (CMR glc ) decreases linearly (9% per 1mM blood ketone body increase) in rats with diet-induced ix

10 ketosis. In another study, using Liquid Chromatography and Gas-Chromatography Mass Spectrometry (LC-MS, GC-MS), we applied carbon-13 ( 13 C) isotopic flux analysis in ketotic rat brains with either [U 13 C]-glucose or [U 13 C]-acetoacetate intravenous infusions. The data show that ketosis reduced glucose carbon flux into the citric acid cycle and γ- aminobutyric acid (GABA), whereas ketone body carbon flux increased in these pathways. In conclusion, ketone bodies partition and spare glucose oxidative metabolism in ketotic rat brain. This may lead to further understanding to neuroprotection from changes of metabolic energy balances. x

11 List of Figures Figure 2.1 Illustration of BHB and AcAc inter-conversion and oxidation to acetyl-coa. 25 Figure 2.2 Ketone bodies utilization in the brain and synthesis in the liver Figure 2.3 Schematics for 18 FDG tracer study models Figure 2.4 Sample Rat PET images, as displayed in CARIMAS2 software Figure 2.5 Experiment Set up for anesthesia system Figure 3.1 Decreased cerebral metabolic rate for glucose (CMR glc ) with increasing plasma ketone body concentrations in rats fed with ketogenic (KG) diet compared to standard diet (STD) Figure 3.2. Meta-Analysis of CMR glc reduction in ketotic subjects (human or rats) Figure 3.3. Images of Volumes of the Interest (VOI) Figure 4.1 Plasma molar enrichment (MPE %) at t = 50 min Figure 4.2 Acetyl-CoA MPE in cortical brain Figure 4.3 Contributions of glucose and AcAc to oxidative metabolism Figure 5.1 Simplified schematics of metabolite labeling patterns with [U 13 C]-Glucose or [U 13 C]-Acetoacetate (AcAc) infusion Figure 5.2 Brain metabolite M2 enrichment from [U 13 C]-glucose studies (Panel A) and [U 13 C]-Acetoacetate studies (Panel B) Figure 5.3 Brain metabolite M1 enrichment from [U 13 C]-glucose studies (Panel A) and [U 13 C]-acetoacetate studies (Panel B) Figure 5.4 Brain metabolite concentrations in rats infused with [U 13 C]-glucose (Panel A and B) or [U 13 C]- acetoacetate (Panel C and D) Figure 5.5 Theoretical schemes for M+1 metabolites generation xi

12 Figure 5.6 Chromatogram of the Citric Acid Cycle intermediates and neurotransmitters Figure 6.1 Proposed Neuron-Glial Compartmentation models for ketone metabolism studies. Figure reference from McKenna review 2007, JNR (112) xii

13 List of Tables Table 2.1 Unpublished data of meta-analysis of the studies on CMR glc during ketosis. 31 Table 2.2 Unpublished data on KG and STD rat CMR glc, with hyperoxia (100% O 2 anesthesia) Table 3.1 Physiological Parameters Table 3.2 CMR glc in the volumes of interest (VOI) Table 3.3 Macronutrients of the standard (STD) diet and ketogenic (KG) diet Table 3.4 Micronutrients of the STD diet and KG diet Table 5.1 Physiological parameters of the rats Table 5.2 Plasma and brain enrichments of glucose M+6 and ketone bodies M Table 6.1 Literature Lumped Constant (LC) numbers for 2-Deoxyglucose(DG) and 18 FDG xiii

14 Acronyms α-kg α-ketoglutarate, same as OHG, or oxoglutarate AAT Aspartate Aminotransferase AcAc Acetoacetate ASP Aspartate ATP Adenosine triphosphate BAM Blood Acquisition Module BHB β-hydroxybutyrate BBB Blood-Brain Barrier 13 C Carbon-13 CAC Citric Acid Cycle (Same as TCA, or tricarboxyl acid cycle) CBF Cerebral Blood Flow CIT Citrate CMR Cerebral Metabolic Rate CMR glc Cerebral Metabolic Rate of Glucose xiv

15 CMR ket Cerebral Metabolic Rate of Ketone Bodies CMR o2 Cerebral Metabolic Rate of Oxygen CNS Central Nervous System CPT Carnitine palmitoyltransferase CT Computed Tomography DG 2-14 C-Deoxy-Glucose EEG Electrocardiogram ETC Electron Transport Chain 18 F Fluorine-18 FDG 18 [F]-2-Fluoro-Deoxy-Glucose FDG-6-P 18 [F]-2-Fluoro-Deoxy-Glucose-6-P FUM Fumarate FW Formula Weight (or molecular weight per mole) G-6-P Glucose-6-phosphate GABA γ-aminobutyric acid xv

16 GAD Glutamate acid decarboxylase GC-MS Gas Chromatography Mass Spectrometry GLC Glucose GLN Glutamine GLU Glutamate 2 H Deuterium HPLC High-Performance Liquid Chromatography HIF Hypoxia-Inducible Factors HMG-CoA Hydroxymethylglutaryl-coA IS Internal Standard KB Ketone bodies KG Ketogenic diet LAC Lactate LC Lumped Constant LC-MS Liquid Chromatography Mass Spectrometry xvi

17 LDH Lactate dehydrogenase MAL Malate MCT Monocarboxylate Transporter MPE Molar Percent Enrichment MRI Magnetic Resonance Imaging NAA 15 N-Acetyl-Aspartate NAD+ Nicotinamide adenine dinucleotide. (Reduced form: NADH) NMR Nuclear Magnetic Resonance OAA Oxaloacetate PC Pyruvate carboxylase PDH Pyruvate dehydrogenase PET Positron Emission Tomography PYR Pyruvate ROI Region of Interest ROS Reactive Oxygen Species xvii

18 SD Standard Deviation SEM Standard error of the mean STD Standard diet SUC Succinate VOI Region of Interest xviii

19 CHAPTER 1 OVERVIEW OF THE DISSERTATION Ketone bodies (KB) are alternative energy fuel used in the brain during the state of ketosis (1) (2) (3) (4) (5). The state of ketosis is known to be neuro-protective against various pathological conditions in the brain, including cancer (6), Alzheimer s diseases (5, 7), epilepsy (8-12), traumatic brain injury (13), and stroke (14). Hypothetical explanations to the mechanisms underlying the protections have been of great interest for scientists and clinicians. There are four major schools in current literature that explained the mechanisms. i) Ketone body utilization spares glucose utilization and oxidation (3, 15-17), which is believed to be vicious following neurological insults. ii) Ketosis changes the citric acid cycle (CAC) intermediates and neurotransmitters fluxes from KB and glucose, leading to a different state of neurotransmitter synthesis and utilization (18-20). iii) KB utilization changes the regulations of key molecular factors and proteins, allowing the brain to adapt to the neurological challenges (14, 21, 22). iv). KB utilization in the brain and in the mitochondria reduces the Reactive Oxygen Species (ROS) in the Electron Transport Chain (ETC) (23, 24). This dissertation is summarizes my works on aspects (i) and (ii). The basic scientific and engineering grounds were present in chapter 2. First, fundamentals of KB and glucose metabolism in the brain are illustrated. Secondly, the neuroprotection from ketone bodies, as are reported in literature, are discussed. Thirdly, the scientific and engineering methods to estimate glucose and KB metabolism are 1

20 reviewed. Lastly, my previously unpublished experimental setup for investigating the glucose and ketone body metabolism are presented. Some of them are used in the publications in the later chapters, but were not presented in the respective chapters. In chapter 3, we present the work on ketosis effect to suppress the Cerebral Metabolic Rate of glucose (CMR glc ). We hypothesize that the total CMR of the brain from KB and glucose is constant in diet-induced ketosis. As a result, we expected to see that the CMR glc decrease during ketosis, in which KB utilization is known to increase (2, 4, 25). We used Positron Emission Tomography (PET) and 2-Tissue Compartmental modeling techniques (26, 27), and tested this hypothesis on rats fed with 3-weeks of ketogenic diet. This article has been submitted to the Journal of Cerebral Blood Flow and Metabolism (JCBFM) and was accepted for publication on May 7 th, In chapter 4, we present the work on ketosis effect to switch the glucose oxidation to acetyl-coa to KB. We hypothesized that the total glucose and ketone bodies fluxes towards the acetyl-coa is constant (downstream from CMR glc, where the first step of glycolysis, phosphorylation, was studied and presented in chapter 3). We divided the animal into 2 diet groups, standard (STD) and ketogenic (KG), and then further divided into two infusion groups We infused either [U 13 C]glucose or [U 13 C]acetoacetate intravenously to those rats(total 4 groups), and analyzed the plasma and brain homogenate with Gas Chromatography Mass Spectrometry(GC-MS) and Liquid Chromatography Mass Spectrometry(LC-MS), respectively. The results are1) from the [U 13 C] glucose groups: isotopic flux from glucose tracer contributes less to the brain 2

21 acetyl coa generation during ketosis 2) from the [U 13 C]-acetoacetate groups: isotopic flux from KB tracer contributes more to the brain acetyl-coa generation during ketosis. These data suggest that the oxidation of the fuels also switches downstream of glycolysis, in addition to glucose phosphorylation. This article has been published Oxygen Transport to Tissue 2013 (28). In chapter 5, we present the investigation on how ketosis alters glucose and ketone balances in contributions to CAC intermediates and neurotransmitters (aspect ii of the hypothetical mechanism on neuroprotection). The experimental protocol is very similar with that presented in chapter 3, but with a different set of analytical methods. Only GC- MS was performed. The data shows complex labeling patterns of the CAC intermediates, as well as neurotransmitters. We reported i) Brains CAC intermediates showed increased isotopic carbon fluxes from ketone bodies in ketosis, while carbon fluxes from glucose decreased in ketosis ii) GABA, is not normally synthesized from ketone bodies in normal conditions, shows significant flux from ketone bodies in ketosis. iii) Ketone bodies contributions to pyruvate recycling increased while glucose contributions to pyruvate recycling decreased in ketosis. A potential mechanism to explain the neuroprotection from ketosis contribution to reservation of carbon sources through the CAC (instead of being completely cleaved down to CO 2 ) and neurotransmitter recycling is discussed. This paper is to be submitted to the Journal of Neurochemistry, due by May 31 st, Finally, in the last chapter, chapter 6, a summary of the previous findings in chapter 3-5 is presented. To the best of my knowledge, I presented two more additional projects for 3

22 future research. 1) Proposal to estimate the lumped constant (LC) in rats in ketosis. LC is an important parameter used in FDG-PET experiments, to determine the CMR glc (26, 29). Our paper in press (chapter 3) findings and conclusions are based on the assumption that the LC was a constant in ketosis. The validation is a crucial point in future. 2).Proposal to further investigate the brain ketone-glucose contributions to CAC intermediates and neurotransmitters in dynamic cerebral compartments. Our article to be submitted (chapter 5) presented the isotopic flux, in unit of percent enrichment (MPE) in the brain Zhang et al 2013 (28) during ketosis. However, this assumes steady state of isotopic balances in the brain, and does not reflect the dynamic substrate utilization rate (in μmol/100g/min tissue). More complex mathematical models, which takes into neuronal-glial interactions (18, 30, 31) is proposed to address the problem. 4

23 Chapter 2 Background 2.1 Biochemistry of ketone bodies Ketone and ketone bodies Ketones are organic chemicals containing double carbon C= groups. Ketone bodies (KB) are different from ketones. They are water-soluble molecules with ketones, generated as a by-product of β-oxidation of fatty acids. Physiological ketone bodies include acetoacetate (AcAc), β-hydroxybutyrate (BHB, also known as 3-BHB) and acetone. The first two ketone bodies are frequently by the body during the state known as ketosis, as this dissertation presents in detail. The chemical structures for the KB are shown in figure 2.1. Both BHB (C 4 H 8 O 3, FW=104) and AcAc (C 4 H 6 O 3, FW=102) contain 4 backbone carbons. AcAc is the oxidized form; BHB is the reduced form. They are inter-convertible through BHB-dehydrogenase (E.C ). The ratios of BHB/AcAc are known as a redox index (32, 33) in describing physiology. AcAc+ NADH+ H + BHB+ NAD + 5

24 2.1.2 Pathways and regulations of ketone bodies metabolism Ketone bodies (KB) are synthesized in the liver and transported to various tissues for use (4, 34-37). First, long-chain fatty acid breaks down, transferring the acyl-coa to the mitochondria through carnitine palmitoyl transferase (CPT). This step is positively regulated by the CPT 1 and 2, and inhibited by malonyl-coa, the key enzyme for fatty acid synthesis. Then, in the mitochondria, the acyl-coa is oxidized to acetyl-coa. The key chemical involved is the hydroxymethylglutaryl-coa (HMG-CoA). Two of the newly formed acetyl-coa then converts to one AcetoAcetate-coA (AcAc-coA) by AcAccoA thiolase. AcAc-coA can be used to generate HMG-CoA from HMG-coA synthase. Finally, the breakdown of the HMG-CoA generates one AcAc plus a free acetyl-coa. In the BHB and AcAc synthesis, AcAc-coA synthesis is a necessary step. Ketone bodies utilizations are also starting with AcAc-coA with AcAc-coA thiolase. It is noteworthy that usually the physiological redox in the blood is greater than 1 (i.e, BHB concentration is higher than that of AcAc) (33). The majority of the ketone bodies BHB, cannot contribute to oxidative metabolism in the CAC and the generation of ATP without first been converted to AcAc and AcAc-coA. Ketone body utilization occurs in the brain and other tissues (32, 34)(Heart, kidney, muscles). Ketone bodies are present in very low concentrations (<0.1mM) in the human and rodent brain normally (3, 38, 39). Diffusion of ketone bodies through the blood brain barrier (BBB) is very low (40-42). The cerebral utilization of ketone bodies is through monocarboxyl transporters (MCT) (40) (41). The synthesis and utilization of ketone bodies are shown in figure

25 2.2 Ketosis and methods of induction Ketosis is defined as a state in which the total blood ketone body (BHB+AcAc) concentrations (alternatively, only BHB) exceed 0.5 mm (4, 25). There are three ways to induce one subject (human or mammals) with ketosis: fasting (including starvation), feeding of high-fat-low-carb diet, and acute infusion of ketone bodies (32, 43). i) Fasting or starvation. In the history of scientific investigation on ketones, this model is first tested. Long term (6-7 weeks) of fasting of obese patients were performed and reported by Owen et al in 1960s. The authors demonstrated, via Kety-Schmidt method (44) (discussed in section 2.4.1), that the cerebral arterial KB concentrations are progressively elevated in chronic fasting. Fasting generated ketosis is known to elevate the blood KB concentrations while decrease blood glucose concentrations, causing a state known as hypoglycemia (2, 15, 17, 45-47). It is apparent that during fasting, insulin level in the blood will be low, and this hormone negatively regulates the lipolysis. Ketone bodies are thus generated as a by-product of β-oxidation (33, 48, 49). The cerebral blood flow (CBF) had been reported not to change in healthy volunteers undergoing prolonged fasting and high levels of ketosis. The ph in the blood decreased from 7.40 to 7.37 in 2-day fasted humans (15), a state of metabolic acidosis from hyperketonemia in the brain may or may not present (15, 39). ii) Diet-induced ketosis. Unlike fasting, where KB are generated from depletion of blood glucose (thus pulling the demand of fatty acid oxidation, (42)), this method works by 7

26 providing excessive fatty acid supply while drastically reduces glucose supply (also indirectly pulling the demand for fatty acid oxidation). For humans, the ketogenic diets (KG) typically contain 10% or less carbohydrate and more than 70% of fat (12, 48) For experimental rodents, the KG diets contain less than 1% of carbohydrates and about 90% of fat, due to less responsiveness (mechanism not clearly known; (16, 18, 50-53). Dietinduced ketosis are not known to cause changes to blood glucose levels or causes adverse effect to blood ph (16, 18, 51-53). However, the diet-induced ketosis is known to increase the free fatty acid in the blood. Researchers are investigating into the improving the diet compositions by altering the compositions (54). iii) Ketosis from infusion of KB. This method is frequently used in NMR studies with ketosis, where the measurement sensitivity of the tracer (discussed in section 2.4.3) is low. Infusion of high load of exogenous ketones, typically more than 1mmol/kg/hr for 1-3 hours (15, 55-58) can cause human hyperketonemia. It is also known that infusion of KB can reset the coupling of CBF and CMR glc and CMR ket, thus invalidate the assumptions of uptake measurement (discussed in section 2.4.1) (56, 59). 8

27 2.3 Neuroprotection from ketosis Ketosis as a pre-conditioning for protection Ketone bodies as a pre-conditioning to protect against neuro-trauma had been reported mainly in basic science investigations (60). The reasons are that artificially induced neurological damages are easier to induce than finding subject with latency of neurological damages for treatment. Indeed, many rodent brain damage models, such as focal and global ischemia by arterial occlusion (61), served well in the investigative purposes. With these in vivo models, the neuroprotection by ketosis were commonly reported in rats: ketosis preconditioning were shown to reduce the infarct volumes (14) (61), increase angiogenesis (62), increase the threshold for seizure occurrence (63)Bough 1999), decrease the edema and improve the ATP Suzuki (33), decrease the CO 2 production from BHB (64), decrease of the contusion volumes (65), and alleviate glutamate cytotoxicities (24, 66) Ketosis as a therapy Ketosis as a therapy had been mainly tested in humans with epilepsy and less in other disease models. In human studies, the most relevant frequently used model of induction is through the ketogenic diet (9, 67, 68). 9

28 Before the 20 th century, fast-induce ketosis had been reported to treat refractory epilepsy. The diet-induced ketosis was first systematically proposed by Wilder et al in 1920 th in Mayo Clinic (1). Large-scale human trials of ketogenic diet as a therapy for epilepsy (9) showed that at least 2/3 of the subjects with epilepsy had reduced occurrence of seizure activities. It has been suggested by Gilbert et al, that (68) as high as 4mM blood ketone concentration was desired to reduce epilepsy. The mechanisms underlying the anti-convulsant effects had been explored from different aspects, including decreased neuronal excitability (69), increased ketone shunts to glutamate and GABA (19). More are discussed in chapter 5.2. Diet-induced ketosis had also been reported to be neuroprotective against Alzheimer s disease (70), traumatic brain injuries (64), reversible focal ischemia (61), and glutamate toxicity (24). For details, see review papers (5, 60). It is important to note that the therapeutic effect of the ketosis has been shown to be age dependent. In developing rats, ketone bodies had more presence in the blood than in adults (32, 41). The therapeutic effects of ketone bodies on refractory seizures had been reported to be highly effective in children (71). Finally, the therapeutic effects from ketosis, in some cases, are reversible. Seizures can be reported 1-2 weeks following the ketogenic diet (8), but consumptions of carbohydrates will immediately reverse the effect. This case may have been due to switches (back and forth) of carbohydrates (primarily glucose) and ketone bodies as a 10

29 source of neurotransmitter generation, as discussed in chapter 5 ketogenic animals infused with [U 13 C]-glucose. In other studies, children with De Vivo s disease who lacks GLUT-1 transporters, can be treated with ketogenic diet, and no reversal of the effect be observed after 2-3 year (72). Future investigations into the reversibility of the therapeutic effects from ketosis are needed. 11

30 2.4 Metabolism of glucose and ketone bodies. Glucose, a simple hexose, is widely thought to be the dominant energy substrate in the brain in normal physiological conditions (4, 35). To generate energy Adenosine Triphosphate (ATP) in the brain, glucose first need to be mobilized in the plasma and the liver, transported to the brain through the Blood Brain Barrier (BBB) by Glut-1 transporters (40, 41, 73).Then the glucose enters the cytosol to undergo the process of glycolysis. The first step would be glucose becomes phosphorylated to glucose-6- phosphate (G-6P), in the brain, mostly catalyzed by hexokinase 1 and 2 that follows Michaelis-Menten kinetics (26, 74). Unlike what occurs in the liver, glucokinase is rarely present to catalyze the glycolysis in the brain (49). Then the G-6P undergoes a series of reactions and transformations to end up with two pyruvates, each with three carbons. Pyruvate then lose one more carbon by the catalysis from pyruvate dehydrogenase and enters the Citric Acid Cycle (CAC) as acetyl-coa (enters the mitochondria). ATP is then generated from the citric acid cycle product, NADH (60). Ketones bodies are produced as a by-product from fatty-acid oxidation via HMG- CoA(60). During ketosis, the MCT transporters amount is elevated (40-42). In the tissues, ketone bodies readily enter the mitochondria to merge into the CAC by converting to AcAc, then AcAc-CoA, and then cleaving the four-carbon AcAc in the AcAc-CoA to two Acetyl groups and thus end up with two Acetyl-coA to enter the citric acid cycle. ATP is then generated from the citric acid cycling similar with the fate of glycolysis. 12

31 Both glucose and ketone metabolic pathways merge at the CAC through the acetyl-coa, with each 1 mole of substrate contributing to 2 moles of acetyl-coa. On the ATP product side, complete glycolysis from 1 mole of glucose generates moles ATP while each 1 mole of ketone oxidation generates moles of ATP (2, 42). It is an interesting topic for physiologists how substrate carbon numbers or ATP demand-supply relationship drives the interactions between glucose and ketone metabolism. In anesthetized rats, the global CMR glc is reported to be between μmol/100g/min (17, 26, 46, 47, 75, 76). On the other hand, the rate of ketone utilization in the brain, CMR ket, is defined as the rate of AcAc oxidation to AcAc-coA, in micromole per 100g wet tissue per minute (μmol/100g/min). For rats, CMR ket is about 1-20 times less than CMR glc (3, 15, 55, 57, 77, 78). The variations are probably from the methods of estimations. See section 2.5 and

32 2.5 Overview of methods to estimate the CMR The word CMR (cerebral metabolic rate) is a confusing term. In many cases, it can be represented by either Uptake or Reaction (or Metabolism ). Metabolic substrates, such as glucose, ketones, oxygen, underwent three physical processes: diffusion (through concentration gradient), convection (through blood flow), and reactions (through chemical binding, such as hemoglobin for oxygen, Monocarboxylic transporters for ketones and lactate (72) GLUT-1 and GLUT-4 for glucose). Take glucose for example, in rat brain, the glucose concentration is about 5-10 times less than that of plasma and the diffusion process was inadequate to meet the demand of glucose. In this case, the dominant, salient features to measure would be either the convection paraphrased as uptake ; or reaction paraphrased as metabolism (the process of glucose phosphorylation to glucose-6-phosphate). To measure either case, one would need to impose a tracer or measure an endogenous tracer from the blood or plasma, which will undergo all three physical processes in the brain; when both the tracer and the tracee (in this case, glucose) reach steady state in the plasma or blood, assuming that diffusion was significantly lower than the uptake or metabolism, the convection rate would approximately equal to the reaction rate (Mathematically the two processes have opposite signs, if reaction causes the tracee to decrease (79)). 14

33 2.5.1 Kety-Schmidt Method (Measurement of uptake) Kety and Schmidt pioneered in the clinical studies of CMR glc measurement methods by introducing convection-based measurement principles (44). At steady state after an inert tracer infusion, the cerebral arterial and venous concentrations, as well as the total amount of tracer disappearance can be measured repeatedly until no A-V difference can be observed (See equation below; V u is the total blood flow tracer taken by the brain from infusion to the end of study, S is the partition coefficient of blood flow tracer, C A and C V stand for cerebral glucose concentrations, respectively, u is the end time of the measurement). Assuming the tracer/tracee partition coefficient stays the same in the brain as ex vitro, the CBF can be estimated. Therefore the CMR glc = CBF (C A -C V ), divided by brain sample weight. This measurement scheme served as standard for years, and it worked for CMR glc, CMR ket, and CMR o2. However, this method had obvious drawbacks: i) the brain arterial and venous sampling is highly invasive ii) measurement would be global CMR glc, does not allow regional CMR glc measurement Compartmental Modeling method (Measurement of reaction) More than two decades later, Sokoloff et al published seminal works on estimating the CMR glc by studying the phosphorylation rate, using a trapping tracer 2-14 [C]- 15

34 Deoxyglucose (DG) and autoradiography (26). The authors postulated transport and phosphorylation rate constants (80), and assumed that the rate constants for DG and glucose held fixed ratios. By further assuming Michaelis-Menten kinetic parameters V m and K m held constant ratios between the tracer (DG) and tracee (glucose), a lumped constant (LC) was assumed. The CMR glc thus can be estimated if tissue and blood (or plasma) activities are known during the study period. CMR glc would be inversely proportional to LC if transport and phosphorylation rate constants are known. This method was verified by in vivo human studies with FDG-PET published by Phelps et al (27). The 2-DG and FDG-PET methods essentially measure the glucose phosphorylation rates. Given the high specific activities of the radio-tracers, very small of tracer volume was required, and no cerebral A-V samplings necessary. Furthermore, Gjedde and Patlak had respectively worked out graphic methods for DG and FDG compartmental models, making parameter estimations for FDG-PET more convenient (81-83). The advantage with PET versus 2-DG method comes with low remaining radioactivity after each study and readily 3-D tomographic assessment of glucose utilization without tissue collection. The disadvantage is that sophisticated mathematical models would be needed, and the justifications of the assumptions are more demanding (84). These methods served as foundations for future generalized models with PET imaging. The details of the model development can be found at appendix iii. The model schematics are shown at figure

35 2.5.3 Inherent difficulties to determine the CMR Each cerebral metabolite has its own characteristics in terms of diffusion, reaction and blood flow dependency. It is therefore very important to realize the respective implications to technical (engineering) methods that need to be tailored in addressing specific questions. Here, we specifically discuss CMR glc, CMR ket and CMR o2 estimations. CMR glc can be measured by either uptake or reaction method. Due to the availability of trapping tracer, the reaction method by FDG-PET, or by 2-DG-autoradiography, are much more convenient than the highly invasive Kety-Schmidt uptake method. However, there are two important drawbacks for the reaction methods. 1) The estimation from 2- tissue compartmental models usually underestimate the real metabolism, because the k5 terms, which specifies the loss of glucose-6-p downstreams through glycolysis, was assumed to be close to zero comparing with phosphorylation rate constant, k3 (84). 2) The lumped constant (LC) is an experimental variable that may well change with physiological conditions (see section 2.5 and section 6.2). These directly impact the final calculation of CMR glc. CMR ket is more difficult to measure than CMR glc because of the lack of a trapping tracer. Significant estimation error arises when the labeled tracer administered to the subject loses in CO 2 (as a result of oxidation) downstream. In this field, literature values obtained by uptake method (3, 15, 55, 57) and reaction method (77, 78, 85) had great 17

36 discrepancies, ranging from 0.5μmol/100g/min to about 20μmol/100g/min in normal and ketotic conditions. One alternative explanation would be that the measurement of blood flow (CBF) from Kety-Schmidt method actually altered the baseline CBF, thus the uncoupling artificially elevated the CMR ket ; on the other hand, it may have been the tracer loss to CO 2 from the non-trapping tracer measured by PET or other reaction-based methods, such that severe underestimation made the apparent CMR ket too low. In addition, the interconversions of the BHB and AcAc complicate the process of measurement of reactions, such that inhibition of the enzyme may be required for one to obtain a true reaction rate of ketone utilization (86). Finally, the process of pseudoketogenesis, though not confirmed to present in the brain, had been shown to confound the estimation of ketone utilization if uptake method was used (34) (87). CMR o2 is a key index in studying the brain metabolism. Essentially, all oxidative fuel metabolism study in the brain relies on the assumption about the CMR o2. Alternatively, the CMR o2 can be measured by fmri (BOLD) technique in combination with inversion techniques to measure the CBF (88). Another reported method is to use 17 O labeled tracers. The uptake method of CMR o2 shares the same principles for CMR ket and CMR glc Aside from the determination of the CBF, the real challenge still lies on the determination of cerebral venous concentration of the oxygen(89). For reaction-based methods, 15 O labeled H 2 O had been reported (90). Due to the technical demands for a dedicated onsite cyclotron and especially for the short half-life of 15 O (only 122 seconds), this kind of studies are not often performed (91). 18

37 2.6 Unpublished Pilot Studies on CMRglc and oxidative metabolism Common mistakes and precautions in determining the CMR glc by FDG-PET The following summary pertains to the troubleshooting experiences on rat FDG-PET experiments on KG animals (see chapter 3). 1) Levels of ketosis. It is critical that the rats needed to be fasted overnight before the initiation of ketosis by KG diet. 2) Anesthesia. Only isoflurane was allowed during FDG-PET experiments. 3) Blood gas parameters. The blood gas parameters, especially the ph, are very sensitive measurement to respiratory and metabolic acidosis. Respiratory acidosis can be from hypo or hyper ventilations. Metabolic acidosis may be due to changes of glycolysis, especially pyruvate-lactate balances (92). The purpose of the setup is to avoid the overshadowing effects of the respiratory acidosis. 5) Injection of the tracers (FDG). Bubbles must be avoided while injecting the radioactive tracers. 19

38 6) Landmarks for image analysis, definitions of ROIs. The highest uptakes of the FDG tracer are from the eyes (see figure 2.). It is easily confused with the frontal cortices. The highest radioactivities occurred at the eye-cup. 7) Parameter estimations. First, we need to make careful choice of the Lumped Constant (LC). The 2-DG and 18 FDG has different LC values. The LC values in these two conditions are reported to bear some linearity (93) in normal conditions, though the mechanism is unclear. The LC is reported to change with age, insulin infusion, and may well shift if the CBF is uncoupled with metabolism. The current best literature LC values for rats are 0.71 (for Sprague-Dawley rats), which may probably work in wistar and * fisher rats with the same age (3 months). Second, for the K 1, K * * 2, and K 3 estimation: Nonlinear fitting method requires good initial guess to obtain a good estimate. It is * * * crucial that K 1, K 2 be in the range of 0.1-1, while K 3 should be one magnitude lower. * * * Alternatively, graphic method to estimate the combination of K 1, K 2 and K 3 can be used. In both cases, the steady state CMR glc should be evaluated 45 minutes post the bolus injection of the tracer, if consistent estimation is needed (26) Animal Anesthesia System Development The experimental set-up for the estimations of the glucose and ketone metabolisms would need to follow strict physiological criteria, such that respiratory acidosis, hypoglycemia should be absent, as discussed in previous sections in chapter 2. In 20

39 addition, it is important that we manage the anesthesia level and stress levels of the animals that the anesthetized CMR glc data are consistent across different animals. Finally, the studies with the ketone and glucose metabolism needs careful amount of oxygen delivery, such that the rats receive physiological relevant amount of oxygen for brain glucose and ketone body utilizations. Figure 2.5 illustrate the final working version of the anesthesia systems. In figure 2.5, legend part b. The nose cone that interfaces the rat (a) has a bite bar (line perpendicular to the cone), that secures the front tooth; on top of that, two dotted vertical parallel lines indicates that the anesthesia line for isoflurane delivery. On the sides of the delivery tube, we drilled holes in it. When the rat exhales, those holes allows waste isoflurane and CO 2 to escape (pointed to the right, as indicated by the arrow), to section d, which has a flask filled with water (~1-1.5L). The solvency of CO 2 to water is 1.45g/L solvency in water. Normal rats have 1.5ml/min/100g co2 production (94). In a PET imaging experiment, which typically lasts about 80 minutes at the most, a rat that weight 300g produces approximately 0.36 liters of CO 2. Assuming ideal non compressible gas for CO 2 at room temperature, using Avogadro's constant such that 1mol of CO 2 would contain 22.4L, a water flask (1L, with water) would be able to contain ~ 0.7 liters. Therefore, 1L of water in that flask is sufficient for the animal to trap all of the CO 2, even if the charcoal filter did not catch any, or if the animal did not breathe out any co2 from the mouth. 21

40 Next, consider the oxygen delivery. The atmosphere contains ~20% oxygen. When awake, a rat breathes in ~20% to maintain its functions of the body and the brain. However, during anesthesia, it is important to have more than 20% oxygen (95). For this reason, we engineered the gas mixture of oxygen and room air, such that the mixture contains 20%-100% oxygen, and is fully adjustable. Our unpublished data (section 2.6.3) showed rats with unphysiological oxygen (hyperoxia) in the scan by improper delivery of the oxygen (100%). Hypoxia and hyperoxia are known to alter the cerebral glucose and ketone metabolism (48, 96). We reports that the appropriate oxygen levels vary, but should be 20-30% during the experiments. We also tested on several animals on the optimum maneuverable levels of anesthesia, and concluded that per minute respiration will not wake up the animal during the 2 hour PET scan while maintaining reasonable blood gas parameters. Finally, to minimize motion artifact from light anesthesia, a metal bite bar was inserted into the nose cone Unreported CMR glc data Before we published the papers (chapter 3-5), we had pilot studies on rats with ketosis (KG) and with standard diet (STD). All CMR glc data published after 2012 were 22

41 performed using the revised animal anesthesia system, as described in figure 2.5. This system was crucial in generating physiological normal parameters. We had failed to generate these parameters in the pilot studies (data showing in table 2.2) for two reasons: i) Anesthesia levels. Several animals were sacrificed due to improper handling of the anesthesia. High levels of isoflurane suppressed the respiration, causing respiratory acidosis by hypoventilation (97). ii) Oxygen levels. The research facility did not provide with nitrogen balance, and there the animals were anesthetized by pure oxygen (100%), thus undergoing hyperoxia (See table 2.2). Such approaches fail to provide reasonable normal physiology. In fact, two previous studies in literature, one from our lab, suggest this to be the case (16, 78). Comparing the data with hyperoxia will need normalization, since the physiology changed. In addition, higher levels of anesthesia (breathe rate ~30 in the pilot studies vs. Chapter 3 studies breathe rate ~60) by isoflurane are known to artificially elevate the CMR glc compared with pentobarbital (98). Finally, the hyperoxia causes measurement of plasma glucose level by enzymatic method (glucose oxidase, by YSI 2700, Yellow Spring, OH) to error prone (99). In sum, pilot studies of rats with hyperoxia anesthesia had respiratory acidosis, tripled Pa O2, low respiration rate. These data were not published, but served good references for study purposes. 23

42 2.6.4 Unreported CMR glc meta-analysis data As is shown in table 2.2, we performed a CMR glc meta-analysis (unpublished) with literature values by either uptake method or reaction methods. This table differs from the published one in chapter 3 for two criteria: 1) Literature values without inclusion of the blood ketone levels were discarded for the published version 2) Literature values of the cerebral A-V differences with no report for CBF, or significant changes of CBF, were not used. We deem that the reasonable levels of the ketosis and the unchanged CBF are keys to maintaining stable ketosis. 24

43 2.7 Figures and tables Figure 2.1 Illustration of BHB and AcAc inter-conversion and oxidation to acetyl-coa. Pathway schematics from (60). AcAc thiolase is the key enzyme for ketone utilization. 25

44 Figure 2.2 Ketone bodies utilization in the brain and synthesis in the liver. The transport of ketone bodies across the blood brain barrier (BBB) requires monocarboxylate transpoters (MCT). Pathways are shown in reference (18). 26

45 Figure 2.3 Schematics for 18 FDG tracer study models. 27

46 Figure 2.4 Sample Rat PET images, as displayed in CARIMAS2 software. The top left, bottom left, top right show the transversal, coronal and sagittal views of the rat brain. Bottom right image shows the 3-D view of the rat. The catheter that delivers the 18 FDG tracer had high residues and is highly visible. The highest uptake of 18 FDG in the brain occurs at the eyes. 28

47 Figure 2.5 Experiment Set up for anesthesia system. Rats are anesthetized by isoflurane and then passively breathing in the vaporized isoflurane in PET gantry. a) The rat b) customized nose cone. The dotted lines indicate tubes with holes to allow animals to exhale. c) Flask of water (used to add more vapor to vaporized isoflurane) d) Flask of water (used to absorb carbon dioxide) e) Isoflurane vaporizer f) 100% oxygen, delivered at liters per minute g) Room air pump, delivered at liters per minute h) Charcoal filter (traps waste gas). 29

48 Author s (et al) Year Journa l Al Mudall al 1995 speci es An est hes ia Tech Method ketosis method CMRglc changed? Change Change after the effect, % Ctrl KG Total keton e Neurol Rats Y 2DG KG diet NS Chang 1993 Can J Physio Kety BHB Phar Dogs Y Schmidt Infusion NA N/A 5 Corddr y 1982 J Neuro chem Rats N 2DG Crane 1985 JCBF M(101 ) Rats same as above Dalqui st 1976 Pediat r Res same as above same as above Same as above Hassel bach 1994 JCBF M Hassel bach 1996 AJP Hawki ns 1986 adult rat Infant rat adult rat Infant rat Hum an Hum an Y and N Y Y Y Y N N AJP Rats Y Hawki ns 1971 Bioche m J same same same same Adult rats suckli ng rats Y 3day fast or BHB Infusion N modified uptake method 2day fast NS NS Diffusion and Kety Schmidt Diffusion and Kety Schmidt Diffusion and Kety Schmidt Diffusion and Kety Schmidt FDG PET FDG PET + Kety modified uptake Diffusion and Kety Schmidt 2day fast 2day fast 3 day fast 3 day fast NS Not Reported Not Reported Not Reported Not Reported NS Not Rep orte d Not Rep orte d Not Repo rted Not Repo rted Not reported Not reported Comments KG diet has calorie restriction; KG diet for three weeks but calory restriction. Hypoxia Vs. Normoxia study, did not measure CMRglc. Only the extraction fraction measured. Cerebellum; N.S decrease. Only look at frontal cortex reduction to 63% after 5 days in pentobarbital anesthetized rats; Hemispheres reduction to 78.5% in hemispheres and frontal cortex unit: umol/mg DNA /min; In adults. This result is weird unit: umol/mg DNA /min; In adults This is also sort of weird Not reported unit: umol/mg DNA /min; In adults Not reported unit: umol/mg DNA /min; In adults 3.5 day Fast Yes No change LC no change. Kety's method: -24% Yes for global, but BHB N.S for infusion regional BHB utilization up 5 folds also Starve 2 Not or 4 Rep days. No orte ctrl NA NA d NA Only measured Cmrket contributes 3% Fast 2day NA fast 3 day 3.03 fast 4 day 3.34 Infuse acac 1.40 Infuse acac Did not measure CMRglc or CMRket. Only a-v concentrations 30

49 Hassel bach 1995 AJP Owen 1967 J Clin Invest Pan 2002 JCBF M Redie s 1989 AJP Ruder man 1974 Bioche Hum an Hum an (Obe se) N N Kety Schmidt Kety Schmidt Hum an N NMR Hum an N m J Rats Y Linde Y 2005 and JCBFM Rats N Mans 1987 Metab olic Br Diseas e Rats N Linde 1999 Acta Physiol Scand Rats N Prins 2009 J Neurot rauma Rats Y FDG PET + Kety Kety Schmidt Kety Schmidt modified Autoradio graphy 3.5 day s fast NA No change NA weeks fast NA hr BHB infusion NA No change of Glucose influx (unidirectional), did not measure CMRglc Not Rep orte d 7.83 Did not measure CMRglc or CMRket Unk now n 2.25 Did not measure CMRglc Day fast Yes No change LC decreased by 25% in ketosis 1-2 days fast NA N/A! Ketone Infusion Only reports A-V diff; only significant for 2days fast; Lactate down in fast rats. This is glucose oxidation, not uptake or utilization Lactate Raised in the awake, but not anesthetized. No CMRglc change 2 days fast Yes Lactate unchanged. Kety Schmidt No NA Autoradio graphy Jiang 2011 JCBFM Rats Y NMR Bentou rkia 2009 AJP Rats Y C11 PET Issad 1987 Bioche m J Rats N Cherel 1988 Metab olism Rats Y same as above Did not induce ketosis. New CBF measure method KG diet, 7 days NA KG diet as therapeutic (Not protective) Lactate up and recovered from the trauma Fast 1.5d, then inf BHB NA Lacks real control animal KG diet and fast 2day NA Autoradio graphy 2day fast NS Autoradio graphy up to 3 day fast NS Unsu re Pure Oxygen Anesthesia; only has CMRket whole body CMRglc decreased 40% Not significant different in the brain Too small to tell 1.30 glucose utilization index, not CMRglc up to 8.5 NS same as above up to 12 days fast NS weird Gjedde 1975 AJP Rats Y Kety Schmidt 5 day fast 0.10 Only measured BUI Table 2.1 Unpublished data of meta-analysis of the studies on CMR glc during ketosis. NS: not significant. NA: not available. 31

50 Rat1 Rat2 Rat3 Rat4 Rat5 Rat6 Rat7 Age P56 P60 P61 P65 P75 P79 P62 Weight grams FDG Dose Injected uci Died 140min post-surgery Dose per g uci/g Time btw fast & injct hrs Pre-Img [BHB] mm N/A Pre-Img [Glc] in whole blood mm N/A N/A N/A N/A N/A 4.7 Pre-Img [Glc] in plasma (Cp) mm N/A N/A N/A N/A Hematocrit (%) Hematocrit stable Through H90min=3 H0mi H10= Imaging? Yes 9% Yes n=42 47 yes Input function quality PET Very Delay 40sec, Heart Very Very Very Heart Good Input Function good good good Brain Image TAC quality Breathe Rate at steady state b/min Steady state sampling time min post injection 48 51, Plasma glucose At steady state mm [BHB] At steady state mm L-Lactate at steady state mm Haemoglobin Oxygen Saturation 97%-18% at steady State % ph at steady state 7.18(Wr N/A N/A ong?) Pco2 at steady state mmhg has N/A N/A Error 333 Po2 at steady state mmhg 314(3 (wrong? 7C ) 152(25c ) 104(25C ) N/A N/A ) CMRglc,trapping,ODE method (K1-k3) umol/100g/min CMRglc,nontrapping,ODE (K1- k4) umol/100g/min CMRglc, trapping, Patlak (K1-k3) umol/100g/min Table 2.2 Unpublished data on KG and STD rat CMR glc, with hyperoxia (100% O 2 anesthesia). 32

51 Chapter 3 Ketosis Proportionately Spares Glucose Utilization in Brain (This chapter is a copy from article to be published by Journal of Cerebral Blood Flow and Metabolism, 2013.) 33

52 3.1 Abstract The brain is dependent on glucose as a primary energy substrate, but is capable of utilizing ketones such as β-hydroxybutyrate and acetoacetate, as occur with fasting, starvation or chronic feeding of a ketogenic diet. The relationship between changes in cerebral metabolic rates of glucose (CMR glc ) and degree or duration of ketosis remains uncertain. To investigate if CMR glc decreases with chronic ketosis, 2-[ 18 F]fluoro-2- deoxy-d-glucose in combination with Positron Emission Tomography, was applied in anesthetized young adult rats fed three weeks of either standard or ketogenic diets. CMR glc (µmol/min/100g) was determined in the cerebral cortex and cerebellum using Gjedde-Patlak analysis. The average CMR glc significantly decreased in the cerebral cortex (23.0 ± 4.9 vs ± 4.7) and cerebellum (29.3 ± 8.6 vs ± 6.4) with increased plasma ketone bodies in the ketotic rats compared to standard diet group. The reduction of CMR glc in both brain regions correlates linearly by ~9% for each 1mM increase of total plasma ketone bodies ( mm). Together with our meta-analysis, these data revealed that the degree and duration of ketosis plays a major role in determining the corresponding change in CMR glc with ketosis. 34

53 3.2 Introduction Researchers and clinicians have been interested in brain metabolism during starvation, fasting or acute ketosis for many decades. Under physiological blood glucose concentrations the fractional contribution of ketone bodies to oxidative metabolism in adult brain has remained uncertain. During prolonged starvation, brain energy requirements have been traditionally accepted to be supplemented by ketone body oxidation. (2, 17). The conviction was founded on the rationale that under glucose sparing conditions, a large portion of oxidative energy must be derived from ketone bodies and thus resulting in reduced glucose consumption. (2, 17, 102). Historically there has been controversy amongst researchers whether there is a causal relationship between changes in cerebral metabolic rates of glucose with degree and duration of ketosis. Inconsistencies across studies were revealed when the effects of short-term fasting (or acute ketosis) on changes in cerebral metabolic rates of glucose (CMR glc ) were further explored (15, 35, 45, 76, 102) We deem that ketones are effective against pathology associated with altered glucose metabolism and inadequate regulation of salvation pathways. We hypothesize that ketone bodies are neuroprotective through the restoration in energy balance via suppression of glucose oxidation and stabilization of ATP supply. Ketone bodies, such as β- hydroxybutyrate (BHB) and acetoacetate (AcAc), are alternative energy substrates to 35

54 glucose especially important during development and glucose sparing conditions, such as with fasting, starvation and diet-induced ketosis. (2, 18, 32, 33)The relationship between energy supply and demand and the partitioning of substrate utilization between glucose and ketones in brain continues to be explored. The ketogenic diet (high-fat, very lowcarbohydrate) to induce chronic ketosis has been successfully used in the clinical setting as a therapy for intractable seizures for nearly a century. (9, 10, 12, 18, 48)However, the mechanistic link between the anticonvulsant effects and ketosis continues to be investigated and remains to be elucidated. (20) Ketone bodies as neuroprotective agents appear to related to the change in the regulation of the cell s stress responses, (21) as well as changes in oxidative (glucose) metabolism(13, 14). Neuroprotection by ketosis is thought to be associated with improved mitochondrial function, decreased reactive oxygen species, apoptotic and inflammatory mediators, and increased protective pathways. (18, 60) In the last few decades the 2-deoxy-D-glucose (2DG) or Positron Emission Tomography (PET) approaches have been applied to ketotic studies, both animal, (13) (16, 47, 51)and humans (15, 17, 56, 103). Reports of altered CMR glc as result of short-term fasting (15, 35, 45, 76, 102) or acute infusions of ketone bodies (56, 59)had generated discrepancies. What remained to be clarified was (i) whether oxidation of ketone bodies can replace glucose proportionately during acute/mild ketosis under normglycemia and (ii) the percent of glucose sparing with degree of ketosis. Some studies reported generalized, 36

55 decreasing CMR glc with 3-5 days of fasting in humans(15, 17, 103)while in other studies there were no significant changes in CMR glc (46, 51, 75, 102). The goal of this study was to estimate CMR glc in chronic ketotic rats and to determine if ketosis induces a metabolic adaptation through changes in glucose phosphorylation rates. The effects of ketosis on CMR glc in intact brain during stabilized blood glucose conditions in diet-induced ketotic rats using positron emission tomography (PET) and 2- [ 18 F] fluoro-2-deoxy-d-glucose ( 18 FDG) were determined. The rationale for using PET- FDG was based on the principle that the phosphorylation rate of 18 FDG (a trapping tracer) can be used to estimate the phosphorylation rate of glucose. In support of our findings, a retrospective analysis of historical data (meta-analysis) to resolve the inconsistencies across studies was also performed(2, 15-17, 45-47, 51, 75, 76, 102, 103). 37

56 3.3 Materials and methods Animal Model and Diets Young adult male Wistar rats were purchased from Charles River (Wilmington, MA, USA), 40 days old, and weighing ~150 grams. All procedures were performed in strict accordance with the National Institutes of Health Guide for Care and were approved by Institutional Animal Care and Use Committee of Case Western Reserve University. Body weights were measured upon arrival and on the experimental day (Table 1). Littermates were housed 3 per cage in the Case Western Reserve University Animal Resource Center with 12h-12h light-dark cycle. All rats were allowed to acclimate for 1 week prior to initiating dietary protocols. Standard rodent diet (STD) was fed to all rats during the acclimation period (Labdiet Cincinnati, OH, USA, Prolab RMH3000 5ANE) ad libitum. One week after their arrival, all rats were fasted overnight for 16 hours to deplete the liver glycogen stores and initiate ketosis. Rats were then randomly assigned to two diets, STD or Ketogenic diet (ketogenic, KG; Research Diet, New Brunswick, NJ, USA, D12369b) and fed for three weeks ad libitum until FDG-PET experiments. (62) The macro and micro-nutrient of the STD and KG diets is shown in Table 3. The original datasheets for the diets are included in supplementary Table 3 and 4. 38

57 3.3.2 Anesthesia and Surgery On the experimental day (post three weeks of diets) rats were morning fasted for 6 hours prior to PET imaging. Rats were then anesthetized with vaporized 2.5% isoflurane balanced with pure oxygen delivered through a nose cone during the surgical placement of arterial and venous catheters: right jugular catheter (MRE, mm ID and 0.084mm OD, Braintree Scientific Inc, Braintree, MA, USA) was advanced towards the atrium for 18 FDG injection and the tail artery (PE-50, 0.58mm I.D and 0.965mm O.D. Stoelting Co. Wood Dale, IL, USA) was cannulated for blood sampling during the PET imaging period. (104) Rats were then transported to the Inveon PET (Siemens, Knoxville, TN, USA) bed and maintained with a mixture of vaporized isoflurane, pure oxygen and room air. Anesthesia level (1-2%), oxygen flow rate ( liters per minute) and air flow rate ( liters per minute) were adjusted to achieve a consistent physiological status across animals. Absence of hind-leg pinch reflex was monitored throughout the PET scan to ensure depth of anesthesia. Heart rate, respiratory rate (breaths/min), plethysmography and oxygen saturation (%) were monitored (via hind leg senor) and recorded throughout the experiment using a pulse oximeter system (MouseOx, Starr life sciences, Oakmont, PA, USA) (Table 1). To maintain breath rates (~70 per minute) and normal blood gases throughout the 1.5 h imaging process, isoflurane was adjusted, as well as the oxygen percentage and flow rates. 39

58 3.3.3 Physiological Parameters Plasma glucose, lactate and total ketone bodies (BHB +AcAc) concentrations were measured pre- and post-imaging (t=0, 60 min) from a blood sample collected (0.1 ml) from the tail artery catheter. The whole blood samples were centrifuged and the plasma separated and immediately frozen in dry ice; the end-of-imaging hematocrits were also recorded. Plasma D-glucose and L-lactate were later measured by YSI 2700 Biochemistry Analyzer (YSI Inc., Yellow Springs, OH, USA) and the plasma total ketone bodies were measured by gas chromatography mass spectrometry, as previously described. 31 Arterial blood gas parameters (ph, P ao2 and P aco2 ) were measured at t=0, 45 min (ABL5 Radiometer, Copenhagen, Denmark); 45 minutes was considered the end point where CMR glc reached steady state (81, 82). Arterial blood glucose was measured at t=15, 30, and 45 min to verify the steady state plasma glucose concentration during the experiment (Precision Xtra Meter, Abbott Diabetes Care, Inc, Alameda, CA, USA). The breath and heart rates were also recorded throughout the imaging process and were used as indicators for physiological status Image Acquisition and Blood Sampling A dual-modal PET-CT device, Inveon (Siemens, Knoxville, TN), was used to image the 18 FDG activities in the brain. The rat s eyes were placed at the center of the field of view for the best spatial resolution. First, a 10-minute transmission scan was performed 40

59 before the 18 FDG tracer injection. The transmission scan generates tissue attenuation map for attenuation correction in the PET images. Then 10 ± 2Mbq/100g of 18 FDG was injected through the jugular line at time zero. Simultaneously, a 60-minute list-mode PET emission scan was started along with the automatic arterial sampling using a customized Blood Acquisition Module (BAM). The BAM device acquires the whole blood radioactivity in the first 2.5 minutes post injection, at a rate of 0.2 ml/min, specified by a connected syringe pump (Harvard Pump 11 plus, Hollisten, MA, USA). The pump was stopped at 2.5 minutes post injection. Manual sampling for arterial blood activity was performed at 3.5, 5, 7.5, 10, 15, 25, 40, 50, 60 minutes, using heparin-coated, microcapillary tubes (HT9H, Statspin, Westwood, MA, USA) with each tube s volume no more than 9µl. On the experimental day the total blood sample volumes were noted from each rat which was less than 1.4ml. After the PET emission scan, the manually sampled bloods were centrifuged (RH12, Statspin, Westwood, MA, USA). The volume inside the micro-hematocrit tubes were premeasured as 8.3μl/37mm, therefore by measuring the length of the whole blood portion, whole blood activity per volume (C * wb ), was obtained by converting the counts from a Gamma counter (LKB1282 Compugamma, LKB Instruments, Mt Waverley, Vic, Australia) and time-correct to time zero. Hematocrit tubes were then broken and the plasma radioactivity (C * p ) was also counted and corrected for the decay. The hematocrits at 3.5 (when manual sampling begins) and 60 minutes were recorded. 41

60 The first 2.5minutes of input function C wb *, was converted to C p * by a factor R. This follows C R = C * p * wb ( t = 3.5) ( t = 3.5) The manually sampled plasma radioactivity data were time-corrected to the 18 FDG injection time and the half-life (107 minutes). The BAM and manually sampled data were combined and saved in a text file for the Matlab (The MathWorks, Natick, MA, USA) program analysis. Factor R was not different between diet groups (1.7 ± 0.10 vs 1.6 ± 0.10; STD, KG, respectively) Image Processing: Region and Volumes of Interest The list mode emission data were binned to 34 frames: 6 10sec, 6 20sec, 4 30sec, 3 1min, 2 2min, 2 4min, and 5 8min. The reconstruction algorithm on the scanner was set to OSEM2D with a ramp filter supplied by the vendor of the scanner. The final images were saved as coronal, transversal and sagittal images with pixels, and the resolution was 0.78 mm in the sagittal, transversal sections and 0.79 mm in the coronal section. The value of each voxel in the reconstructed PET image sets is converted to radioactivity per volume. 42

61 The processed PET radioactivity image data were analyzed using Carimas 2 (Turku PET centre,turku, Finland), to generate the Region of Interest (ROI) and Volume of Interest (VOI) data. Both the left and right eyes were identified as the landmarks. A rat brain atlas (Paxinos and Watson, Academic Press) was used to guide the selection of the ROI and VOI. Starting from the rear of the eyecup, with slice thickness mm, the left and right entire cortical hemispheres were encircled and two separate VOI were generated. Similarly, the whole volumetric cerebellum was selected as one VOI. A separate PET-CT image set for a rat of the same range (P60-P80) of age was overlaid to verify the cortical and cerebellar regions (see Figure3.4). The two hemispheres VOI and the cerebellum VOI were saved to text files and made importable to the Matlab program as the Time Activity Curve (TAC) format Parameter Estimation and Calculation of CMR glc We developed a MatLab program to perform the parameter estimation and calculation of CMR glc. The plasma input function was interpolated to render a time resolution of 0.1 second. Then the 34-frame TAC was matched with the input function for each of the time points. The Gjedde-Patlak plots were generated and only the last 6 matching points, namely the time after 25 minutes data were used to generate the parameter K i, which follows: 43

62 k k * * 1 K i = * k2 + * 3 * k3 While k 1 * is the 18 FDG transport rate constant (/min) to the brain tissue, k 2 * is the 18 FDG reverse transport rate constant from tissue to the plasma (/min) and k 3 * is the 18 FDG phosphorylation rate constant (/min). The lumped constant in both the KG and the STD diet rats were assumed to be 0.71 (105). The 60-minute plasma glucose level, C p, was used to generate the final CMR glc, which follows: (26, 27) * Ki C CMRglc = LC p 44

63 3.4 Results Physiological parameters There were no significant differences in body weights, blood gases, physiological parameters, and plasma glucose concentrations between KG and STD diet groups following 3-weeks of feeding the diets (Table 1). As expected, plasma ketone (BHB, AcAc) concentrations were statistically higher and the plasma lactate concentrations were lower in the KG rats compared to the STD group(62). Ketosis ranged between mm as measured by total plasma ketone bodies (Figure 1); the STD group was mildly ketotic ( mm, plasma total ketone bodies) following a 6 hour fast prior to imaging. Lactate concentrations in the plasma were significantly lower in the KG diet group (0.77 ± 0.18 mm), Table Cerebral Glucose Metabolic Rates The averages of the CMR glc (µmol/100g/ min) measured in both cerebral hemispheres and cerebellums are shown in Table 2. There were no significant differences in CMR glc between the left and right hemispheres The PET analysis revealed that diet induced ketosis resulted in a significant decrease in the average CMR glc in both cerebral hemispheres and cerebellum compared to STD group. CMR glc was significantly lower in the left and right cerebral hemispheres compared to the cerebellum, in both dietary groups. 45

64 The CMR glc calculated with Gjedde-Patlak analysis was plotted as a function of the measured total plasma ketone body concentrations (BHB + AcAc; mm) (Figure 1). These data showed that cerebral (left and right hemispheres) and cerebellar CMR glc decreased with increasing ketosis. The calculated CMR glc in each region was represented by a linear decrease with increasing total plasma ketone concentrations. There were no significant differences between left and right cerebral hemispheres, (CMR glcright = -2.9 ([BHB]+[AcAc]) ; R² = 0.59); whereas the cerebellar region was significantly higher (CMR glc = -3.7 ([BHB]+[AcAc]) ; R² = 0.59) compared to Cerebral Cortex. These data highlight the proportional change in CMR glc with increasing ketosis; thus for every 1mM increase in total plasma ketone bodies CMR glc decreases by ~9% Meta-analysis of CMR glc in Ketotic Subjects Meta-Analysis of CMR glc reduction in ketotic subjects (human or rats) was shown in Figure 2. All data were collected from previously reported studies where CMR glc was measured and level of ketosis was reported. CMR glc data from the ketotic subjects were normalized to the non-ketotic controls (%) and then graphed against the total blood ketone body concentrations (mm). The normalized glucose utilization rate decreased ~ 9% for each 1mM increase of the total blood ketone bodies. A summary of these data collected from the various studies measuring CMR glc and blood ketone concentrations 46

65 includes (see Figure 3.2 legend for details): PET-FDG studies conducted in fasted humans showing a 27% decrease in CMR glc following 3.5 days of fasting (45), in humans that were fasted for 3 weeks the authors reported a 46% decrease in CMR glc relative to the non-fasted baseline conditions (17). Other studies using different methodologies for assessing glucose utilization in ketotic rats showed similar decreases. In one study where [6-14 C] glucose and autoradiography was applied, glucose utilization decreased 12% in conscious 2-day fasted rats with mild ketosis(76). 47

66 3.5 Discussion We report here, in diet-induced ketotic rats, decreases in CMR glc highly correlate with both the level and the duration of the ketosis. These data revealed that the degree and duration of ketosis play a major role in determining corresponding changes in CMR glc with ketosis. We also present a retrospective analysis of historical data (meta-analysis) that appears to reconcile the inconsistencies from previous studies which supports our conclusion. The brain s ability to switch from glucose oxidation towards ketone bodies requires a type of cerebral metabolic adaptation. This process is not well understood but is thought to be highly associated with the duration and level of ketosis(14, 32, 40, 41).Ketones are considered to supply up to 70% of the total energy demands once maximal metabolic adaptation occurs(2). Blood ketones become elevated during prolonged fasting or with a ketogenic diet reaching a state ketosis and glucose sparing. During this process, monocaboxylic transporters (MCT) up-regulate at the blood brain barrier with increasing demand for ketone utilization by brain (40, 41). Recently, investigators have recognized additional therapeutic properties of ketosis, such as neuroprotection following stroke or injury(14, 60).What remains unclear is whether the neuroprotective or therapeutic properties of ketosis is as a result of changes in the regulation of metabolic signaling pathways. These would include those associated with 48

67 enzyme-catalyzed steps involved with glucose regulation (5) or glucose independent pathways, such as the Nrf2 pathway (a "responder" to cellular stress) (21). In this study we questioned whether cerebral metabolism of ketone bodies (CMR ket ) replaces CMR glc following three weeks of diet-induced ketosis. Previous studies measuring CMR glc in ketotic subjects report either changes in CMR glc with ketosis or failure to detect significant changes(2, 15-17, 45-47, 51, 75, 76, 102, 103).Historically, it has been established that brain can utilize ketone bodies under ketotic conditions(2) (5, 9, 10, 12, 20, 32, 35). However, corresponding changes in CMR glc during metabolic adaptation to ketosis has not been clearly described. Using PET-FDG imaging, the focus of this study was to determine if CMR glc decreases with increasing ketosis in adult anesthetized diet-induced ketotic rats. CMR glc in cerebral hemispheres and cerebellum decreased with increasing ketosis ( mm) in rats fed either STD or KG diets for 3 weeks. These data are consistent with the conclusions described in the classic human study by Owen et al (2). Their study was the first to highlight that the brain can switch from glucose oxidation to ketone body oxidation with chronic ketosis. Most revealing to us was a previous study using a similar rat model of diet-induced ketosis to measure changes CMR glc (51). The study failed to detect significant changes in CMR glc even though the duration and method of induction of ketosis was similar. However, the level of ketosis was 4-fold lower making it difficult to detect a corresponding change in CMR glc with ketosis. 49

68 The most striking information obtained from our study was the correlative finding that CMR glc decreased 9% with every 1% increase in total plasma ketones. Although not previously reported as such, the results of this study are consistent with previous studies measuring CMR glc in ketotic subjects, as our meta-analysis (Figure 3.2) also showed the same linear association between level of ketosis and corresponding changes in CMR glc. The meta-analysis supports our current findings and has brought new insight into previous studies (as authors interpretations led to discrepancies or incomplete conclusions). One explanation for the discrepancies may be the difficulty to distinguish small changes in CMR glc with a small degree of ketosis. This was the case with our previous study in diet-induced ketotic rats where CMR glc was assessed using 2-DG(51). The level of ketosis was less than 1mM, making it difficult to detect a less than 9% decrease in glucose utilization using a non-imaging compartment modeling method. Another consideration is the induction of ketosis through acute ketone body infusions. The main difficulty to this approach is the lack of metabolic adaptation to ketosis (41, 62). We have previously shown metabolic adaptation to ketosis is directly associated with duration of ketosis and level of ketosis (13, 62).Thus, in some studies using acute infusions of ketones to mimic ketotic conditions the outcome failed to show decreases (or consistency) in glucose utilization (56). An exception might be in studies where low 50

69 doses were given following short-term fasting, but the analytical approach often requires a higher degree of sensitivity for detecting small changes in CMR glc (57). Variabilities in experimental models such as, physiological status, level of ketosis via metabolic adaptation, and analytical approach play a key role in the measured outcome. The emphasis of our current study was to use PET imaging together with our diet-induced rat model of ketosis to measure detectable changes in CMR glc. In summary, CMR glc decreased ~9% in both the cortex and the cerebellum for each 1mM increase in blood ketone bodies, which is consistent with diet-induced ketosis, as well as long and short term fasted ketosis. We attribute previous discrepancies to i) the failure to detect significant differences within and across studies, ii) in adequate metabolic adaptation to ketosis, and iii) difficulty in establishing and/or maintaining a higher degree of ketosis. Our work puts historical data into a current perspective by reconciling the inconsistencies from previous studies where little or no change in CMR glc with ketosis was reported. Nevertheless, the maximum percent ketone bodies that can replace glucose oxidation still need to be determined. A quantitative understanding of CMR glc and CMR ket under different durations and degrees of ketosis would elucidate the energy balance between glucose and ketone bodies. 51

70 3.6 Acknowledgements We would thank the CASE Mouse Metabolic Phenotyping Center (MMPC; U24 DK76174) for assisting with GC-MS assays. This research has been supported by the National Institutes of Health, R01 HL , R21 NS

71 3.7 Figures and Tables Figure 3.1 Decreased cerebral metabolic rate for glucose (CMR glc ) with increasing plasma ketone body concentrations in rats fed with ketogenic (KG) diet compared to standard diet (STD). Volumes of the interest (VOI) were defined as the left and right hemispheres (panel A; open circle, left and closed circle, right) and cerebellum (panel B). The CMR glc in each region was calculated with Gjedde-Patlak analysis and plotted as a function of the measured plasma total ketone body concentrations; the equation CMR glc = [slope 53

72 plasma total ketone concentrations + CMR glc at non-ketogenic state] corresponds to the linear correlation; goodness of fit was represented as the coefficient of determination, R 2, which reflected ~0.61 for each VOI. The STD diet group (n=9) total plasma ketone bodies were less than 0.87 mm and the KG diet group (n=10) was greater than 3.0 mm. These results demonstrate that CMR glc decreased ~9% for each 1 mm increase in total plasma ketone body concentration in ketotic rats induced by 3 weeks of KG diet. 54

73 Figure 3.2. Meta-Analysis of CMR glc reduction in ketotic subjects (human or rats). All data were collected from previously reported studies where CMR glc was measured and level of ketosis was reported. Data were normalized (%) against control state (nonfasted, non-diabetic conditions) and graphed as a function of total blood ketone bodies level (mm). The study, method and reported outcome is noted for each point: (a) Data from Al-Mudallal et al, (51) ketosis by KG diet in rat, 2-DG method; no significant cortical change in CMR glc, (b) Data from Corddry et al, (47) 3days fasted rats, 2-DG method; frontal cortical change, not significant, (c) &(d) Data from Dalquist et al, (46) 3 Days fasted rats, A-V uptake method; no significant change, (e) Data from Hasselbach et al, (15) 3.5 days fasted humans, PET-FDG imaging; significant reduction, (f).data from 55

74 Owen et al, (2) 5-6 weeks fasted obese human subjects, A-V uptake method, CMR glc, was indirectly calculated by O 2 consumption; significant change, (g) Data from Redies et al, (17) days fasted obese human subjects, PET-FDG and A-V uptake method; significant CMR glc reduction, (h) Data from Ruderman et al, (45)1-2 days fasted rats, A- V uptake method; trended significant, (i) Data from Mans et al, (76)2 days fasted rats, compartmental modeling with non-trapping tracer (autoradiography); significant reduction, (j) Data from Issad et al, (75) 2 days fasted rats, (autoradiography), no significant change, (k) Data from Cherel et al, (102) 6 days fasted rats, modified 2DG method, no significant change. The meta-analysis plot shows a linear relationship between CMR glc and level of ketosis in human or rat subjects. For each 1mM of total blood ketone concentration increase there was approximately a 9 % decrease in CMR glc. 56

75 Figure 3.3. Images of Volumes of the Interest (VOI). VOI drawings were performed using CARIMAS2 software with the aid of a CT anatomical image set and a rat brain atlas. Top panels (yellow) indicate volume of left hemisphere in the transversal and sagittal planes. The bottom panels (pink) represent volume of cerebellum in the transversal and sagittal planes. VOI of right hemisphere are not shown. VOI drawings were performed using CARIMAS2 software. 57

76 Table 3.1 Physiological Parameters Age (days) 72 ± 6 77 ± 7 Weight (g) 358 ± ± 26 Blood Gas Parameters a ph 7.35 ± ± 0.03 PaO 2 (mm Hg) 112 ± ± 20 PaCO 2 (mm Hg) 53 ± 5 50 ± 5 Physiological Parameters b Breath Rate (Br/min) 63 ± 3 63 ± 5 Hematocrit (%) 48 ± 2 47 ± 2 Heart Rate (Beats/min) 378 ± ± 37 Plasma Metabolic Parameters b BHB (mm) 0.36 ± ± 0.85 * AcAc (mm) 0.21 ± ± 0.31 * BHB+AcAc (mm) 0.51 ± ± 1.04 * BHB/AcAc ratio 1.74 ± ± 1.19 * L-Lactate (mm) 1.00 ± ± 0.18 * D-Glucose (mm) ± ± 1.87 * P<0.05 compared to STD diet group. a Measured at 45 minutes post 18 FDG injection b Measured 60 minutes post 18 FDG injection Physiological parameters and concentrations of metabolites in plasma in rats fed standard or ketogenic diets. Values are the means ± standard deviations. n, number of rats. Young adult rats were fed either standard (STD) or ketogenic diet (KG) for three weeks prior to measurements of CMR glc. 58

77 Table 3.2 CMR glc in the volumes of interest (VOI) VOI STD diet (n=9) KG diet (n=10) Left Hemisphere 33.5 ± ± 4.8 * Right Hemisphere 32.3 ± ± 5.2 * Hemispheres Average 32.9 ± ± 4.9 * Cerebellum 41.2 ± 6.4 ** *, ** 29.3 ± 8.6 *P<0.05 compared to STD diet group, **P<0.05 compared to hemispheres CMR glc (µmol/100g/min) Cerebral metabolic rate for glucose (CMR glc ) by positron emission tomography and 2- [ 18 F] fluoro-2-deoxy-d-glucose in rats fed standard or ketogenic diets. CMR glc (µmol/ 100g/ min), at t=60 min post 18 FDG injection. Values are the means± (SD); n, number of rats. *Significance p < 0.05 relative to STD. **Significance p < 0.05 relative to cortical hemisphere. 59

78 Table 3.3 Macronutrients of the standard (STD) diet and ketogenic (KG) diet MacroNutrients Starch % by weight D glucose % by weight Fructose % by weight Sucrose % by weight Lactose % Saturated % by weight Monounsaturated % by weight Polyunsaturated % Protein % by weight Fat % by weight Carbohydrates % by weight Protein % by energy Fat % by energy carbohydrates % by energy STD diet (Labdiet 5ANE RMH3000) KG diet (Research Diets D12369B) <1% <1%

79 STD diet KG diet RMH3000) (Research Diets D12369B) Minerals Ca % P % Mg % K % S % Na % Cl % Cr ppm (STD) or %(KG) * Cu ppm (STD) or %(KG) * I ppm (STD) or % (KG) * Fe ppm (STD) or % (KG) * Mn ppm (STD) or % (KG) * Se ppm (STD) or % (KG) * Zn ppm (STD) or % (KG) * F ppm Co ppm Vitamins Carotene ppm Vitamin A IU/g ** Vitamin D3 IU/g ** Vitamin E IU/g ** Vitamine K ppm ** Thiamin Hydrochloride ppm ** Riboflavin ppm ** Niacin ppm ** Pantothenic Acid ppm (calcium salt) ** Folic acid ppm ** Pyridoxine ppm ** Biotin ppm ** Vitamin B12 mcg/kg ** Choline Chloride ppm Ascorbic Acid ppm

80 Table 3.4 Micronutrients of the STD diet and KG diet *the STD diet nutrients are expressed in units of ppm, KG diet nutrients are expressed in % weight. ** The KG diet nutrients (Vitamins) are shown as their respective units in 1gram vitamin mix, which further mixes with the diet that contains 3917kcal. Refer to supplementary material 3, D12369B. The STD diet datasheet was obtained from Cincinnati lab (cincinnatilab.com) The KG diet datasheet was obtained from Research Diets (ResearchDiets.com) 62

81 Chapter 4 Contributions of Brain Glucose and Ketone Bodies to Oxidative Metabolism (This chapter has been published as an article : Zhang, Y., Kuang, Y., LaManna, J. C., & Puchowicz, M. A. (2013). Contribution of Brain Glucose and Ketone Bodies to Oxidative Metabolism. In Oxygen Transport to Tissue XXXIV (pp ). Springer New York ) 63

82 4.1 Abstract Ketone bodies are an alternative energy substrate to glucose in brain. Under conditions of oxidative stress, we hypothesize that ketosis stabilizes glucose metabolism by partitioning glucose away from oxidative metabolism towards ketone body oxidation. In this study we assessed oxidative metabolism in ketotic rat brain using stable isotope mass spectrometry analysis. The contribution of glucose and ketone bodies to oxidative metabolism was studied in cortical brain homogenates isolated from anesthetized ketotic rats. To induce chronic ketosis, rats were fed either a ketogenic (high-fat, carbohydrate restricted) or standard rodent chow for 3 weeks and then infused intravenously with tracers of [U-13C] glucose or [U-13C] acetoacetate for 60 min. The measured percent contribution of glucose or ketone bodies to oxidative metabolism was analyzed by measuring the 13 C- label incorpora- tion into acetyl-coa. Using mass spectrometry (gas-chromatography; GC-MS, and liquid-chromatography; LCMS) and isotopomer analysis, the fractional amount of substrate oxidation was measured as the M + 2 enrichment (%) of acetyl-coa rela- tive to the achieved enrichment of the infused precursors, [U-13C] glucose or [U-13C] acetoacetate. Results: the percent contribution of glucose oxidation in cortical brain in rats fed the ketogenic diet was 71.2 ± 16.8 (mean% ± SD) compared to the standard chow, 89.0 ± Acetoacetate oxidation was significantly higher with ketosis compared to standard chow, 41.7 ± 9.4 vs ± These data confer the high oxidative capacity for glucose irrespective of ketotic or non-ketotic states. With ketosis induced by 3 weeks of diet, cortical brain utilizes twice as much acetoacetate compared to non-ketosis. 64

83 4.2 Introduction This study was developed on the basis that ketones are effective against pathology associated with altered glucose metabolism, such as with ischemia reperfusion injury and seizure disorders. Ketosis can be induced by prolonged fasting or ketogenic (KG) diet. We had previously reported neuroprotection by ketosis following recovery from transient focal ischemia (14). Using a rat model of ketosis, we hypothesized that the cerebral metabolic rate for glucose (CMRglc) decreases with increasing ketosis. Thus, a shift of oxidative metabolism away from glucose towards ketone bodies may result in neuroprotection, irrespective of the mechanistic link. It has been described in humans and rodents that CMRglc decreases with ketosis (16, 55). To show the partitioning of glucose metabolism towards ketone oxidation, one would need to simultaneously measure CMRglc under ketotic and non-ketotic conditions. CMRglc can be readily measured by a metabolic trapping mechanism using 2-[18F]-Fluorodeoxyglucose (18FDG) tracer and Positron Emission Tomography (PET) imaging system (16). However, the CMRket cannot be reliably measured due to various constraints. These include (i) costly use of PET imaging systems, (ii) lack of a trapping tracer for accurate measurements of ketone body utilization, (iii) short half- life of the currently available tracers, and (iv) relatively low sensitivity of nuclear magnetic resonance (NMR) (78, 106, 107). To test in the cortical brain the partitioning of ketone utilization during ketosis, we designed a study using stable isotope tracers and mass spectrometry to estimate the 65

84 fractional contribution of glucose or ketone bodies (acetoacetate) to oxidative metabolism. This enabled the implementation of a relatively inexpensive method (compared to PET), using an in vivo rat model of ketosis to study cortical brain glucose and ketone body metabolism. GC-MS and LC-MS systems are used for investigating intermediary metabolism, as they have high sensitivities to many analytes and metabolites. Compared to NMR methods, the use of small sample size therefore allows a smaller blood volume to sample. Stable isotopes of 13 C-labeled tracers were infused into anesthetized rats and assayed by mass spectrometry. This approach assumes that the 13 C-label incorporated acetyl-coa is from the oxidation of the precursors, 13 C- glucose or 13 C-acetoacetate. Using isotopomer analysis, the M + 2 enrichment of acetyl-coa was measured and the fractional percent contribution of substrates (glucose or ketone bodies) to oxidative metabolism was calculated as the mole percent enrichment (MPE) of acetyl-coa (107, 108). 66

85 4.3 Methods The experimental protocol employed in this study was approved by the Institutional Animal Care and Use Committee (IACUC) at Case Western Reserve University Animal Preparation and Diets Adult male Wistar rats (final weight: g; n = 20) were purchased from Charles River and were allowed to acclimatize in the CWRU Animal Resource Center (ARC) for at least 1 week before feeding their respective diets. Rats were then fed either the ketogenic (KG) or Standard (STD) diets for 3 weeks prior to the experimental day (14, 16). The KG diet was purchased from Research Diet (New Brunswick, NJ, USA) and the standard rodent chow (Teklad 8664) was provided by CWRU ARC. All procedures were performed with approval from the Case Western Reserve University IACUC. On the experimental day, both diet groups (KG and STD) underwent the same surgical procedures for the placement of jugular and arte- rial catheters and tracer infusions (16). Rats were morning fasted for 4 h prior to tracer infusions prior to infusions. Anesthesia was induced with isoflurane balanced with a mixture of N 2 /O 2 and the rats were maintained under light anesthesia during the tracer infusions. The flow rates of the gases were manually adjusted to maintain breath rates (60 80 breath/min). Arterial blood gases were measured (ABL-5, Radiometer, 67

86 Copenhagen) to confirm stable arterial blood ph Experimental Design, Tracer Preparation, and Infusions Four study groups were implemented: rats were infused with tracers of [U-13C]glu- cose or [U-13C]AcAc and fed either standard (STD) or ketogenic (KG) diets. [U-13C] glucose (99.8 %) was solved in 0.9 % NaCl solution to a final concentration of 38.7 mm. [U- 13 C]AcAc was derived from [U- 13C] ethyl-acetoacetate, as previously described (109) and concentrated to 137 mm. All chemicals were purchased from Sigma-Aldrich. Tracers were infused via the jugular vein catheter (0.50 or 1.0 mmol/ kg/h) (Harvard Apparatus syringe pump-11 Plus) for 60 min. To verify steady-state conditions, blood samples ( ml) were taken from the tail artery at time point 0 (pre-infusion), and at 15, 30, 40, 50, and 60 min (during infusion), immediately centrifuged and the plasma frozen for GC-MS analysis of the [U-13C] precursor enrichments and concentrations of glucose and acetoacetate. At the end of infusion, the rats were decapitated; the brains were dissected immediately, frozen in liquid nitrogen, and stored at 80 C. Cortical sections (~200 mg tissue) were then dissected under frozen conditions and homogenized using a specific organic solvent mixture designed for isolation of acyl-coas and related metabolites (107, 108). 68

87 4.3.3 Estimation of the Contribution of Acetoacetate and Glucose to Oxidative Metabolism Cortical brains were processed for 13 C-acetyl-CoA (M + 2) enrichments (MPEs) using LC-MS, a similar method as previously described (107, 108). The plasma MPE of 13 C-glucose and 13 C-AcAc was measured using GC-MS methods (107, 108). After back- ground correction, the MPEs of the precursor 13 C-substrates and the oxidative prod- uct (acetyl-coa), were calculated from the measured ion masses (M + 4, [U- 13 C] AcAc; M + 6, [U- 13 C]glucose; M + 2, [U- 13 C]acetyl-CoA) to the unlabeled (MO, endogenous intermediate); e.g., acetyl-coa (M + 2) MPE = [M2/(M2 + M0) 100]. The percent fractional contribution of glucose or AcAc to oxidative metabolism in cortical brain was estimated from the MPE of acetyl-coa relative to the plasma MPE of the 13 C-infused substrates and calculated: Substrate contribution to oxidative metabolism (%) = [(brain acetylcoa MPE 2)/(plasma glucose or AcAc MPE)] 100. All data are expressed as mean ± SD. Statistical analyses were performed using a two sample t-test. Significance was considered at the level of p <

88 4.4 Results and Discussions The fractional contribution of glucose or AcAc to cortical brain oxidative metabo- lism was estimated in anesthetized ketotic rats using stable isotope mass spectrom- etry analysis. The plasma MPE tracer dilution profiles of 13 C-glucose and 13 C-AcAc reached steady-state conditions by 50 min (time course not shown). Ketosis induced by KG diet did not significantly affect plasma 13 C-glucose or 13 C-AcAc MPE compared to STD groups (9.8 ± 1.0 % vs. 9.2 ± 0.5 % and 20.9 ± 5.5 vs ± 3.3, respectively) (Fig. 4.1). Cortical oxidative metabolism was significantly altered by ketosis (Figs. 4.2 and 4.3). With glucose oxidation, a 30 % decrease in acetyl- CoA MPE was observed (Fig. 4.2, see STD and KG groups given tracer infusions of [U- 13 C]glucose), whereas with AcAc oxidation (see [U- 13 C]AcAc), acetyl-coa MPE increased about 40 % with ketosis (STD vs. KG groups). These data show a partitioning of brain glucose oxidation towards ketone body oxidation with chronic ketosis. When estimating the percent contribution of glucose to oxidative metabolism, ketosis (KG) resulted in a decrease in glucose oxidation which was not significantly different from the STD diet group (Fig 4.3). Data confirm the high oxidative capacity of glucose in cortical brain, irrespective of ketosis. With respect to percent contribution of ketone body oxidation, ketosis resulted in an increase in oxidative metabolism, as shown by the twofold increase in AcAc percent contribution compared to STD diet (Fig. 4.3). Consistent with our hypothesis, ketosis induced by diet plays a role in cortical brain utilization of AcAc. These findings demonstrate the ability of brain to switch towards ketone body oxidation with ketosis 70

89 (Figs. 4.2 and 4.3) (78, 106). This model appears to overestimate oxidative metabolism by about 15 %. The sum of the percent contribution of glucose and AcAc to oxidative metabolism exceeds 100 % (Fig. 4.3). Indeed, in healthy non-ketotic mammals, glucose contribution to oxidative metabolism in brain is about 90 %. So we suspect that the fraction of ketone contribution to oxida- tion metabolism is overestimated by about 15 %. The reason for this overestimation remains to be determined. One explanation is the precursor pool of 13 C-AcAc enrichment in brain tissue differs from plasma; an underestimation of the precursor MPE could account for this discrepancy. 71

90 4.5 Acknowledgments The authors would like to thank the CASE MMPC, affiliated staff and faculty, for their technical assistance and helpful discussions on mass isotopomer analysis. This research was supported by the National Institutes of Health, R01 HL , R21 NS and Mouse Metabolic Phenotyping Center, MMPC U24 DK

91 4.6 Figures and tables Figure 4.1 Plasma molar enrichment (MPE %) at t = 50 min. Tracers of [U- 13 C]glucose and [U- 13 C] AcAc (acetoacetate) were infused in two diet groups, standard (STD) and ketogenic (KG). Steady- state MPE was achieved by t = 50 min (time course not shown). 13 C-glucose infusions resulted in a 10 % plasma MPE in both diet groups. As a result of an increase in infusion rate of [U- 13 C]AcAc in the KG diet group compared to STD diet, a two fold increase in the 13 C-AcAc plasma MPE was observed (mean ± SD; *p < 0.05) 73

92 Figure 4.2 Acetyl-CoA MPE in cortical brain. Rats fed STD or KG diets were infused with either [U- 13 C]glucose or [U- 13 C]AcAc tracers. Ketosis resulted in decreased glucose MPE with a parallel increase in AcAc MPE 74

93 Figure 4.3 Contributions of glucose and AcAc to oxidative metabolism. Percent contribution of glucose oxidation in cortical brain decreased with ketosis. A significant increase in percent contri- bution to AcAc oxidation with ketosis was also observed 75

94 Chapter 5 Ketone bodies spares glucose oxidative metabolism in dietinduced ketosis in rat brain (This chapter is to be submitted to the Journal of Neurochemistry as a manuscript in June, 2013) 5.1 Abstract It is known that ketosis is neuroprotective to the brain. The mechanistic links from ketone bodies and glucose oxidations in the citric acid cycle (CAC) to neuroprotection remains to be explored. We hypothesized that ketone bodies serves the neuroprotective roles through sparing of the glucose carbon shunting to CAC intermediates and neurotransmitters. Rats were fed with either standard (STD) or ketogenic (KG) for 3-4 weeks and then infused with either [U 13 C]-glucose or [U 13 C]-acetoacetate to study glucose and ketone bodies fluxes toward oxidative metabolism. The plasma and brain homogenates were analyzed by gas-chromatography and mass spectrometry (GC-MS) for the isotopic fluxes. Results: 1) Brain [U 13 C]-glucose fluxes to CAC intermediates and neurotransmitters are reduced in ketosis; brain [U 13 C]-acetoacetate fluxes to CAC intermediates and neurotransmitters are increased in ketosis. 2) KG rat brains have significantly increased [U 13 C]-acetoacetate fluxes to GABA comparing with those in STD brains. 3) During ketosis, [U 13 C]-glucose infusion increases brain glutamine and 76

95 glutamate concentrations, while the uptake of [U 13 C]-acetoacetate decreased brain glutamate and glutamine concentrations. It can be concluded that diet-induced ketosis spares brain glucose oxidations with ketone bodies. Ketosis may protect the brain through reduction of the glutamate and glutamine and increasing GABA concentrations. 77

96 5.2 Introduction Chronic feeding of ketogenic diet had long been demonstrated to be neuroprotective. In humans, ketogenic diet pre-conditioning is known to reduce epilepsy occurrence (1, 8, 10). In animal studies, ketogenic diet is shown to be both protective in many injury models, including epilepsy, ischemia, traumatic brain injuries and hypoxia (13, 14, 63, 110). The hypothetical interpretations of the mechanism to neuroprotection from ketone bodies come from i) Ketosis increases consumptions of ketone bodies and decreases consumptions of glucose in the brain (3, 13, 15-17). ii) Ketosis alters availabilities of brain neurotransmitters, such as glutamate, and GABA, either in neurons or astrocytes. (18, 53, 111) iii) Adaptation to ketosis shifts important molecular regulator proteins and transporters (14, 21, 40, 41, 65) in the brain. iv) Ketosis reduces Reactive Oxygen Species (ROS) productions (21-23) and glutamate toxicity (24) to the brain. Our lab approached this mechanistic problem through the first two ideas. We deem that the biochemical pathways of the glucose and ketones are responsible for changes of the molecular regulators and the intracellular chemical milieu. First, changes of glucose and ketone bodies fates in utilization and oxidations directly leads to changes of downstream metabolites concentration and fluxes, therefore triggers altered enzymatic equilibriums. Chronic adaptation to the altered biochemical equilibriums leads to changes of protein expressions (32, 33, 41). Thus the third idea for explanation of neuroprotection is dependent on the first two ideas. Secondly, the utilization and oxidation of the ketone bodies and glucose are related to ATP generations through the electron transport chain in 78

97 the mitochondria (22, 66, 112). To this sense, the fourth idea can also be partly attributed to the changes in the utilization and oxidations of glucose and ketone bodies. Lastly, the citric acid cycling activities is linked together with the glutamate-glutamine cycling activities, which is responsible for generation of the glutamate and GABA. Ketosis was believed to increase ketone bodie s shunts to glutamine (19), similar with acetate in the astrocytes (113) during traumatic injury, that can generate specific pool of glutamate that ultimately turns to GABA. Conversion of glutamate to glutamine also reduces cytotoxicity (24). GABA, as the major inhibitory neurotransmitter, is also believed to be associated with anti-epileptic effect (66, 114), as well as significant contributions to the glutamate-glutamine recycling (115). Therefore, it is important to trace the ketone and glucose utilization and oxidations, as well as their contributions to neurotransmitters. We had recently shown that diet-induced ketosis suppresses the cerebral metabolic rate of glucose (CMRglc) in adult rats (16). Assuming the cerebral oxygen metabolic rate (CMR o2 ) stays relatively constant in ketosis, the reported reduction of CMR glc (essentially the steady state phosphorylation rate of glucose) seemed directly translates to ketone bodies sparing of glucose oxidative metabolism. However, the phosphorylation accounts only for the first step of glycolysis. The pathways of complete oxidation of glucose and ketone bodies converge at the entrance of citric acid cycle (CAC), where acetyl-coa was used to generate citrate. Furthermore, as the turning of the CAC proceeds, carbons from glucose or ketone bodies continue to be shunted towards glutamate through α-ketoglutarate- glutamate transferase, and further complicated cycling 79

98 between neuron and astroglial cells in the brain are reported (116, 117). To clarify whether ketone bodies spares the glucose oxidative metabolism, it is imperative to investigate the CAC intermediates and neurotransmitters fluxes and concentrations in the working brain. Previous works on the brain ketone body metabolic rates in humans and rats had generated very different results (2, 3, 55-57, 77, 78, 106). The CMR ket reported in ketosis had varied between 2-8μmol/100g/min (55, 56, 78) to about 20μmol/100g/min (2, 15, 57) humans during different models of ketosis. No evidence can be shown that the differences are solely due to species differences. We speculate two possible reasons. First, unlike glucose metabolism studies (13, 15-17), the investigation of ketone bodies lacks a trapping radiotracer (77, 78, 106). Usually, stable isotopes, with much less sensitivity, are applied to the subject or animals at orders of magnitude higher (28). However, the brain presents very low concentration of ketone bodies, even during ketosis (3, 38). The infusion or injection of exogenous ketone bodies are reported to increase the cerebral blood flow (CBF), which directly causes uncoupling of the metabolism and shifts the baseline level of metabolism in ketosis (56). This idea can also be supported by the observation that the studies with radiotracers with low amount of infusion (77, 78, 106) often yield lower ketone utilization or oxidations rate than when high amount infusion of tracers were applied (57). Secondly, the adaptation and stabilization of ketosis, as a necessary step for neuroprotection, requires increased regulation of molecular mechanisms is age dependent (33, 41) and ketotic-duration-dependent (13, 21, 80

99 40), which may not present with some acute high-infusion studies. The presence or absence of the adaptation phenomena in the studies may lead to the variations of the results. For these two reasons, one must carefully design the ketosis induction method to study the biochemistry underlying neuroprotection from ketosis. To address the ketone bodies neuroprotection in terms of oxidative metabolism and fluxes to neurotransmitters, we infused [U 13 C]-glucose and [U- 13 C]-acetoacetate (AcAc) in chronically diet-induced ketotic rats. We intentionally infused significantly lesser amount than what had been reported in literature (18, 53, 55-57, 106). Highly sensitive Mass Spectrometry was used to analyze the labeled metabolites in the brain as well as the plasma. 81

100 5.3 Methods Animal model and diets Young adult male Wistar rats were purchased from Charles River (Wilmington, MA, USA), 40 days old and weighing ~150 grams. All procedures were performed in strict accordance with the National Institutes of Health Guide for Care and were approved by Institutional Animal Care and Use Committee of Case Western Reserve University. Body weights were measured upon arrival and on the experimental day (Table 1). Littermates were housed in the Case Western Reserve University Animal Resource Center with 12h- 12h light-dark cycle. All rats were allowed to acclimate for 1 week prior to initiating dietary protocols. Standard rodent diet (STD) was fed to all rats during the acclimation period (Labdiet Cincinnati, OH, USA, Prolab RMH3000 5ANE) ad libitum. One week after their arrival, all rats were fasted overnight for 16 hours to deplete the liver glycogen stores and initiate ketosis. Rats were then randomly assigned to two diets, STD or Ketogenic diet (ketogenic, KG; Research Diet, New Brunswick, NJ, USA, D12369b) and fed for three weeks ad libitum until experiment day Tracer Infusion and tissue collection [U- 13 C] glucose (99.8 %) was purchased from Sigma-Isotec (St. Louis, MO, USA, Cat#389374) and solved in 0.9 % NaCl solution, with a concentration of 38.7 mm. [U 13 C]-AcAc was derived from [U 13 C] ethyl-acetoacetate, also purchased from Sigma- 82

101 Aldrich. (St. Louis, MO, USA, Cat# CX1474), as previously described (118) and concentrated to 137 mm (107). All other reagent chemicals were purchased from Sigma- Aldrich. The animals were divided into four groups: 1) Standard Chow (STD) diet, infused with [U 13 -C] glucose at 0.5mmol/kg/hr 2) STD diet, infused with [U 13 C]-acetoacetate at 0.5mmol/kg/hr 3) KG diet, infused with [U 13 -C] glucose at 0.5mmol/kg/hr 4) KG diet, infused with [U- 13 C]-acetoacetate at 1mmol/kg/hr. We had previously reported that the two groups with [U- 13 C]-Glucose infusion for 50 minutes both achieved ~10% plasma glucose M+6 enrichment, while the other two groups with [U- 13 C] acetoacetate infusion for 50 minutes both achieved ~ 20% plasma AcAc M+4 enrichment (28). On the experimental day (3-4 weeks of diets) rats were morning fasted for 6 hours prior to infusion of the stable isotopes. Rats were then anesthetized with vaporized 1.5% isoflurane balanced with pure oxygen delivered through a nose cone during the surgical placement of arterial and venous catheters: right jugular catheter (MRE, mm ID and 0.084mm OD, Braintree Scientific Inc, Braintree, MA, USA) was advanced towards the atrium for isotope infusion and the tail artery was cannulated with the same type of catheter for blood sampling during the experiment period. [U 13 C]-tracers were constantly infused via the jugular vein catheter (Harvard Apparatus syringe pump-11 Plus) for 50 min. Anesthesia level (1-2%), oxygen flow rate ( liters per minute) and air flow rate ( liters per minute) were adjusted to achieve a consistent physiological status across animals. Absence of hind-leg pinch reflex was monitored throughout the 83

102 experiment to ensure depth of anesthesia. Heart rate, respiratory rate (breaths/min), plethysmography and arterial oxygen saturation (%) were monitored (via hind leg sensor) and recorded throughout the experiment using a pulse oximeter system (MouseOx, Starr life sciences, Oakmont, PA, USA) (Table 1). To maintain breath rates (~70 per minute) and normal blood gases throughout the 50 minutes infusion process, isoflurane was adjusted, as well as the oxygen percentage and flow rates. The breath and heart rates were also recorded throughout the experimental process and were used as indicators for physiological status. Arterial blood ph were measured at t=0, 45 min (ABL5 Radiometer, Copenhagen, Denmark to ensure the absence of respiratory acidosis. To verify that glucose or ketone bodies are at steady-state conditions, blood samples ( m L) were taken from the tail artery at time point 0 (pre-infusion), and at 15, 30, 40, 50 minutes, immediately centrifuged and the plasma frozen for GC-MS analysis of the [U- 13 C] tracer enrichments and concentrations of glucose and AcAc. In addition to the verification of the tracer enrichment steady state, plasma D-glucose and L-lactate were also measured by YSI 2700 Biochemistry Analyzer (YSI Inc., Yellow Springs, OH, USA) at 45 minutes post infusion. The total amount of blood drawn from each animal during the infusion is less than 1.5ml. At the end of infusion, the rats were decapitated; the brains were dissected immediately, frozen in liquid nitrogen, and stored at 80 C. Cortical sections (~200 mg tissue) were then dissected under frozen conditions and homogenized using a specific organic solvent mixture designed for isolation of acyl-coas. Briefly, the homogenates were mixed with 84

103 CAC internal standards, homogenized with 3 ml of methanol and 3ml of methanol/water 1:1 containing 5% acetic acid using a polytron homogenizer, then centrifuged for 30 minutes at 3400 rpm Analytical method and theory of flux analysis The brain sample pellets were extracted by a mixture of Acetonitrile and 2-Propanol (3:1) and then centrifuged. Then the extracts were dried by nitrogen for 1-2 hours. The extracted pellets were derivatized by reagent TBDMCS (Regis Technologies, Inc. Morton Grove, IL, USA) by incubating at 70 C for 30 minutes, similar with previously described (Kombu et al). The derivatized products were measured under Gas- Chromatography Mass Spectrometry (GC-MS). The maximum oven temperature was set to 320 C, the pressure was psi, and the flow velocity was 45cm/sec. CAC intermediates, including citrate (m/z 459), succinate (m/z 289), fumarate (m/z 287), and malate (m/z 419) were ran under scan mode. Other intermediates and neurotransmitter, including aspartate (m/z 418), glutamate (m/z 432), glutamine (m/z 431) and GABA (m/z 274) were also measured. Internal standards of BHB D6, 2-oxohydroxyglutarate (2-OHG, m/z 433) D4, Succinate D4 and glutamate D4 were added to help determine the concentrations. GABA and fumarate concentrations were cross-corrected by succinate D4 internal standards; malate and citrate concentrations were cross-corrected by 3-OHG D4 internal standards. 85

104 To determine the isotopic fluxes of the intermediates, Molar Percent Enrichment (MPE) was determined by taking the ratio of isotopic abundance/sum of all isotopic abundances. The MPEs for all measured metabolites were further corrected for natural abundance and background by applying matrix method, as previously described (119). Briefly, each metabolite M0 were ran in a separate GC-MS experiment, and the fractions of the M+1 through the M+N (N being the highest detectable labeled m/z shift from M0) were recorded as a correction matrix. The raw data are then organized in a diagonal matrix is then multiplied by the inverse of the correction matrix to subtract the background MPE. Because both the [U 13 C]-glucose and [U 13 C-AcAc] enters the CAC as two acetyl-coa, we interpret the dominant of the labeling pattern of the intermediates to be from pyruvate dehydrogenase (PDH) activities, as shown in M+2 (%). Malate M+3 (%) would be directly from pyruvate carboxylase (PC) activities, derived from [U 13 C]-glucose infusion groups only. Other labeling patterns that come from pyruvate recycling are interpreted as non-dominant pathways and considered minor contributions to oxidative metabolism (See Figure 2). 86

105 5.4 Results Physiological parameters The rats in all four study groups had very similar weight, age, as well as the anesthesia levels during the study. The KG animals infused with [U 13 C]-AcAc were approximately 1 week older than the rest groups. The ketogenic rats infused with [U 13 C]-AcAc had showed an increased hematocrit from the other groups, but are still in a physiological range. All four groups of rats had the similar plasma glucose levels, and showed no signs of hypoglycemia or hyperglycemia during anesthesia and tracer infusion. Anesthesia level of breaths per minutes (awake rats have > 100 breaths per minute) by isoflurane indicates low suppression of brain metabolism. The STD diet animal groups had less than 0.75mM of total ketone body (BHB+AcAc) concentrations. The KG diet animal groups had mM of total ketone body concentrations. The plasma lactate levels were always higher in the STD diet groups comparing with the KG groups. In all 4 groups, the lactate levels were below 2mM, which indicates dominant aerobic respiration. Animal groups with the same diet conditions but different infusion ([U 13 C]-glucose or [U 13 C]-AcAc) did not show difference in physiological parameters except BHB/AcAc redox ratios, which are all different in the four study groups. The group with STD diet and [U 13 C]-AcAc infusion had the lowest BHB/AcAc redox, while the KG diet group with [U 13 C]-glucose infusion had the highest BHB/AcAc redox. The [U 13 C]-glucose 87

106 tracer infusion resulted in higher BHB/AcAc in redoxes in both diet groups. Arguably, the redox state is the most sensitive physiological parameters, and this indicates the amount of the different tracers indeed had some impact, though not high, to the energy balance states Plasma and BHB tracer enrichments As are presented in table 2, infusion of [U 13 C]-glucose tracers resulted in ~10% enrichment of brain glucose M+6 in both diet groups. In the plasma, the enrichment of glucose M+6 was also ~10% in both diet groups. This establishes the same glucose tracer pool availability to both STD and KG animals. Infusion of [U 13 C]-AcAc tracer resulted in ~95% of enrichment of brain AcAc M+4 in both dietary groups. In the plasma, the enrichment of AcAc M+4 was ~25% in both diet groups. Unlike glucose tracer, the brain in both diet groups showed higher tracer appearance than in the blood pool. The major ketone body, BHB, showed different labeling patterns in the brain and in the plasma. In the plasma, BHB M+4 enrichment in the STD and KG infused with [U 13 C]- glucose were both below 4% and showed no difference. However, when [U 13 C]-AcAc is given to both groups and that yields similar AcAc M+4 percent of enrichment, the KG group showed higher BHB M+4 enrichment. In the cortical brain tissue, the lowest BHB M+4 enrichment was observed in the KG rats infused with the [U 13 C]- glucose. The highest BHB M+4 enrichment was observed in the STD rats infused with [U 13 C]-AcAc. 88

107 It is important to note that the current GC-MS method does not allow separation of M+1, M+2 and M+3 signals from AcAc and BHB labels. This is because the TBDMCS method cleaves the four carbons chains to two double-carbon chains. As a result, we simplified the interpretations of the actual M+2 of the derivatized ketone bodies to M+4, as the dominant labeled ketone bodies First turn of CAC metabolites fluxes In this case, we only consider the first turn of CAC metabolites and neurotransmitter M+2 fluxes only, without the complex label exchanges from pyruvate recycling and the second turn of CAC. M+2 was used as the primary indicator of oxidative metabolism in the CAC, because both the U 13 C-glucose and U 13 C-AcAc can only label two of the acetyl-coa carbons in the first turn. When U 13 C-glucose was infused, we detected that 1) Acetyl coa M+2 had decreased by ~50%. 2) M+2 fluxes from succinate had decreased by 45%. Fumarate, citrate and malate are unchanged. 3) Neurotransmitter and glutamine. Aspartate, glutamate, glutamine and GABA all decreased ~35%. No change of citrate flux had been observed. We had also detected malate M+3 in the U 13 C-glucose infused brains, values are 0.75±0.23% for STD and 0.51±0.12%, with no statistical significance of difference (P=0.08, data not shown in 89

108 figures). Finally, 2-oxoglutarate, a convertible form of α-ketoglutarate, produced undetectable amount of M+2. (Figure 3A) When [U 13 C]-AcAc was infused, we detected that 1) Acetyl coa M+2 had increased to ~260%. 2) M+2 fluxes from citrate, succinate, malate, all had increased; the increment were approximately 11, 2, 9 folds. 3) Neurotransmitters, aspartate, glutamate M+2 fluxes had increased by approximately 7 and 10 folds. 4) M+2 fluxes towards fumarate and GABA, which showed undetectable amount in STD rat brain, showed observable and significant amount in KG rats. In short, the KG rat showed high increase of contribution of ketone body carbons towards all measured CAC intermediates, as well as neurotransmitters Pyruvate recycling and 2 nd turns of CAC In this case, we only consider the metabolites M+1. M+1 labeling patterns come from either the second turn of CAC (see figure 1D) or the activities of pyruvate recycling from malate (figure 5A and 5B). For GABA and succinate, the M+1 account for all pyruvate recycling activities from malate plus all the second turn of CAC. For citrate, the M+1 only comes from pyruvate recycling, not the second turn of CAC (Figure 5). For other measured metabolites, the pyruvate recycling from malate and second turn of CAC together contribute to the majority of M+1, but some small portion of their M+2 may also be from the pyruvate recycling from malate and the second turn of CAC. Interestingly, 90

109 except for U 13 C-glucose infusion groups, where we detected malate M+3 as mentioned, no M+3, M+4 or M+5 were observed in any metabolites in any study groups. This essentially rules out the labeled oxaloacetate combining with labeled acetyl coa scenario since no citrate M+3 could be detected. For [U 13 C]- glucose infused rats, KG group showed significantly reduced M+ 1 flux for citrate (~50%) as well as malate (~30%). M+ 1 flux from other CAC intermediates did not show significant differences. In addition, glutamine M+1 decreased ~50%. When citrate M+1 (Figure 3) were used to compare with their respective M+2 (figure 2), which reflect the pyruvate recycling versus combined PC and PDH activities, the M+1/M+2 ratios are 43% for STD rat and 21% for KG brains. In addition, glutamate M+1/M+2 ratios dropped from 54% for the STD rat to 41% in KG rats, indicating recycling of glucose that generates glutamate had decreased. For all other metabolites, the M+1/M+2 ratio were not significant different between STD and KG animals. No aspartate M+1 had been detected. 2-Oxoglutarate M+1 remained unchanged in both diet conditions. For [U 13 C]-AcAc infused rats, KG group showed increased M+1 fluxes for citrate (~2 folds), succinate (~2.5 folds), fumarate (~3 folds), malate(~2.5 folds), glutamate(~2 folds), glutamine(2 folds), and GABA(~7 folds). For citrate, the M+1 were lower than M+2 in STD rat brains, indicating that the exogenous ketone bodies are highly recycled. The M+1/M+2 ratios for citrate decreased from 3.9 for STD rats to 0.7 for KG rats, suggesting that ketone bodies are significantly used in oxidative metabolism rather than being recycled. Similar observations were found in fumarate M+1/M+2 ratios 91

110 (M+1=0.8%, M+2 not detected in STD, M+1/M+2= 1.6 for KG), malate M+1/M+2 ratios (M+1/M+2= 4.8 for STD and 1.4 for KG), glutamate M+1/M+2 ratios (M+1/M+2=4.7 for STD, 0.9 for KG), glutamine M+1/M+2 ratios (M+1/M+2=1.74 for STD, 0.6 for KG), GABA M+1/M+2 ratios (M+1=0.3% for STD, M+2 no detected in STD. M+1/M+2=0.8 for KG). The data suggest that ketosis actively shunts carbons from ketone bodies to all neurotransmitters and significantly decreased the amount of recycling. Succinate did not show any significant change of M+1/M+2 ratios. Similar with [U 13 C]- glucose infusion studies, no aspartate M+1 had been detected. 2-Oxohydroxyglutarate M+1 remained unchanged in both diet conditions Metabolite concentrations The [U 13 C]-glucose infusion studies showed that the animals had increased glutamate (6.6μmol/g in STD, 10.7μmol/g in KG), glutamine (4.5μmol/g in STD, 6.6μmol/g in KG), malate(0.25μmol/g in STD, 0.35μmol/g in KG), citrate(0.17μmol/g in STD, 0.28μmol/g in KG), as well as the expected BHB (0.02μmol/g in STD, 0.19μmol/g in KG) and AcAc (5nmol/g in STD, 28nmol/g in KG) in the cortical brains in ketosis. The data suggest that glucose augment glutamate synthesis during ketosis, and increases glucose oxidative metabolism. Interestingly, succinate, the key intermediate for oxidation and electron transport chain, did not show significant change of concentrations. GABA, aspartate, fumarate concentrations did not change, either (Figure 4A, 4B). 92

111 The [U 13 C]-AcAc infusion studies showed that the animals had decreased glutamate (6.9μmol/g in STD, 3.79μmol/g in KG) and glutamine (4.7μmol/g in STD, 2.7μmol/g in KG) in the cortical brains in ketosis. All measured CAC intermediates, as well as GABA and aspartate, did not show changes in concentrations. The BHB (0.02μmol/g in STD, 0.18μmol/g in KG) and AcAc (7nmol/g in STD, 28nmol/g in KG) concentration in each diet group was similar with the respective groups in [U 13 C]-glucose infusion studies. Those data confer that ketone bodies are able to reduce the glutamate and glutamine pool in ketosis, while maintaining the pools in the CAC intermediates (Figure 5.4C,5.4D). 93

112 5.5 Discussion Changes of oxidative metabolism in ketosis In this study, we have first demonstrated that in the rats with chronic diet-induced ketosis, glucose contributions to cortical citric acid cycle flux were spared by the ketone bodies, in consistent with our previous report with ketone bodies sparing effect on glucose phosphorylation rates (16). Unlike the glucose phosphorylation, the oxidative metabolism involves further downstream biochemistry (as the carbon contributions to acetyl coa generation) at the entrance of CAC, as well as the turning of CAC and generation of neurotransmitters, which may or may not theoretically generate consistent results. Our study had proved that the ketone bodies ability to suppress glucose metabolism is not limited at the first step of phosphorylation. Secondly, the carbon shunting switch from glucose to ketone bodies towards succinate may suggest more succinate participation in the respiration and electron transport chain (14). We had previously shown that rats infused with BHB had increased succinate content in the brain, and it appeared that the succinate concentration increase may be accountable for stabilizing HIF, which may explain the neuroprotection of ketosis from angiogenesis. Our new data suggests that the ketosis does not increase the unlabeled succinate content, but rather worked to increase the flux from ketone bodies to generate succinate while reducing the glucose contribution to succinate. To this sense, ketotic subjects would have increased brain succinate if either extra glucose or ketone bodies are 94

113 given; however, ketone bodies are more effective in generating succinate in this state (Figure 4). Furthermore, although no statistical significance was found in fumarate and malate in the brains infused with [U 13 C]-glucose, the data trends to suggest that the glucose fluxes in these two intermediates also decrease in ketosis. Citrate flux, which serves as converging point neurotransmitters (glutamate, aspartate, GABA) as well as CAC selfturning, appeared to be unchanged for sources from glucose fuels, but significantly increased for ketone body fuels. In diet-induced ketotic rat brains where no tracer is given, we can thus expect a net increase of citrate appearance comparing with the normal unketotic conditions. Meanwhile, the total carbon supply to the CAC intermediates (except for malate) may be maintained in ketosis, as are shown in figure 4. The sparing of glucose oxidation from ketosis in the CAC intermediates is an important step for verification of our hypothesis that total energy demand (glucose + ketone bodies) stays constant during diet-induced ketosis. The generation of ATP is fundamentally dependent on the functionalities of the electron transport chain activities, as well as the citric acid cycle intermediate balances. Although it is not certain whether brain energy balance is fundamentally a carbon molar balance from the total energy supply, our data suggests that cerebral metabolic rate (CMR) total may remain relatively constant, if the carbon loss in respiration and pyruvate-lactate interconventions are relatively small. 95

114 5.5.2 Shunts to neurotransmitters A metabolic fuel switch towards neurotransmitters is also observed in ketosis. The drastic reduction of glucose contributions to neurotransmitters and glutamine M+2 was in consistent with our previous report of CMR glc reduction (phosphorylation rate) in dietinduced ketotic rats, where we showed approximately 9% decrease of CMR glc for each 1mM of total ketone body increaseand a maximum of ~35% of CMR glc was reported in rats with plasma total ketone bodies ~4mM (16),. In this study, the glutamate, glutamine and GABA from glucose tracer are both reduced by ~35% in rats with ketosis, although the total plasma ketone bodies were only 2.5mM in ketosis, which corresponds to ~25% decrease in phosphorylation rate. The artifact from changes and uncoupling of metabolism from blood flow is not very likely, because the infusion amount set in this study is ~1/30 of that used by Hasselbach et al 1996 (56), where 25% increase of CBF was reported. Such difference in reductions from phosphorylation (prediction, -25%) and neurotransmitter (measured, -35%) generations may be from less reduction of carbons shuntings of glucose source in the CAC, where citrate was unchanged. A likable explanation would be a less responsive decrease of recycling from glucose in the neurons (120). Our studies showed that ketone bodies, which the brain normally rarely uses to generate GABA, can n be utilized to generate GABA in ketosis (Fig 3). As previously 96

115 shown in cultured neurons (58), ketone bodies are effectively shunted towards GABA. The current study suggests that the ketone bodies are very potent precursors for GABA synthesis when pyruvate recycling partitioning part is fulfilled (see next section). Similar phenomenon (absence in STD diet but presence in KG) was found in fumarate M+2, which reflect backfluxed malate M+2 (see next section). We speculate that the backflux from malate to fumarate increases in ketosis. Future work would be needed (121). Finally, it appears that the neurotransmitter concentrations in the brains are all highly responsive to the tracer infusion, as observed by the clear glutamate and glutamine concentration changes in all study groups (Fig 4). However, GABA and aspartate are less responsive to infusion. It could well be a sensitivity issue. Indeed, many literatures on compartmentation in the neuron-astroglial interactions in normal and ketotic brains had focused on the GLU-GLN cycling and deeming its key contribution to oxidative metabolism (55, 57, 116, 122). While recognizing the important roles of the cycling from GLU-GLN, we deem that GABA and aspartate, especially GABA, deserves more scrutiny in explaining the biochemical mechanisms underlying neuroprotection from ketosis (58, 115). Whether the GABA pool changes in ketotic rat brain in vivo in tracer conditions deserves more investigation Alterations of pyruvate recycling Pyruvate recycling phenomena are present in the brain, in which the carbons from malate are taken back to pyruvate and re-enters the CAC from PC or PDH. This 97

116 seemingly futile cycle is important for complete oxidation of carbon-fuels. Pyruvate recycling is an alternative explanation that accounts for the substrate partitioning, in addition to the neuronal-glial cyclings of glu-gln. NMR techniques that can effectively detect different labeling patterns of C-4 Glutamate and glutamine (117, 123), suggesting that a different partition, or pathways, of the fuels in oxidative metabolism. Ketones bodies are shown to be recycled in the brain (18, 58, 117), particularly in astrocytes (18, 112, 124, 125), when studied by a convenient acetate tracers. Whether the recycling occurs in neurons in vivo is under investigation (58, 120, 126, 127). On the other hand, one can also assess the PC/PDH contribution ratios to glutamate, glutamine and other detectable amino acids by NMR (18, 117, 128). However, our study tool, the GC-MS cannot allow positional tracing, thus cannot distinguish signals from the different positioning of M+1 or M+2 s from PC and PDH. Considering these two aspects, a better interpretation of the labeled metabolites would be to directly distinguish the M+2 and M+1. M+2 were naturally the dominant labeling pattern from [U 13 C]-tracers, while M+1 can only occur after the first turn of CAC, which includes pyruvate recycling. Most of our measured metabolites M+1 are theoretically from either pyruvate recycling or the second turn of the CAC (fig 5). Because citrate M+3 was never observed in any study groups, it is reasonable to assume that the recombination of labeled oxaloacetate and labeled acetyl-coa was negligible. It is therefore reasonable to assume that pyruvate recycling would be able to explain the majority labeling patterns of M+1 we observed. 98

117 First, in brains of rats infused with [U 13 C]-glucose, we showed that M+1 was no more than that of M+2 in all measured CAC intermediates; in brains of rats infused with U 13 C- AcAc, we showed that M+1 was no less than that of M+2 in all measured CAC intermediates. We interpret this by assuming two partitioning pathways for ketone and glucose to enter the CAC and neurotransmitters. If pyruvate recycling step is always preferred and prioritizes in the brain when exogenous ketones were present, the distinct labeling patterns of GABA and fumarate can clearly be explained. On the other hand, for exogenous glucose, the brain prioritizes it to oxidation instead of recycling in ketosis. This idea shares the same principles with the non-stoichimtric partitioning of glutamateglutamine (112) and the literature data where ketosis was reported to increase the pyruvate recycling (Melo et al, Ostad et al), although no clear distinctions from the carbon source for recycling was made. Our data suggest that ketosis decreases the carbons from pyruvate-recycled glucose to the neurotransmitters, while promotes more ketone carbons to be recycled, generate glutamate, glutamine and GABA. Our interpretation can also explain Yudkoff s theory (19), where ketosis reserves a pool of carbons at glutamine and releases upon energy needs. Pyruvate recycling may serve as a potential important reservoir during ketosis. Secondly, the aspartate M+1 was never observed in our study, indicating that pyruvate recycling does not come with a commensurate, though futile cycle of pyruvate carboxylation. This is further verified by the absence of citrate M+3. 99

118 Finally, 2-Oxohydroxyglutarate (OHG) M+1 was present in all groups; though no statistical significance was seem in any group-group comparison. The α-ketogluorate M+1 thus will probably present, although we cannot detect it by current GC-MS derivatization method. If it were present, then it should serve the source for glutamate and glutamine, as well as GABA (fig 5). Considering the neuronal-glial compartmentation of neuron-glial cells (116, 117, 122, 123, 128), if M+1 of GABA is solely synthesized in neurons, our data confers that the reserved carbons to synthesize the inhibitory neurotransmitters are present for ketone bodies anytime in ketosis. Unfortunately, we did not acquire multi-time point data for the 2-OHG and GABA M+1, so that some compartmental model could be developed to estimate the flux from α-keotglutarate to the GABA synthesis. Future work on this would shield light to the carbon reserves impact towards neuroprotection. 100

119 5.6 Acknowledgment We would like to thank Donald Harris for assisting with some tissue processing work. 101

120 5.7 Figures and tables Table 5.1 Physiological parameters of the rats. STD/U 13 C- Glucose(n=4) KG/U 13 C- Glucose(n=6) STD/U 13 C- AcAc(n=7) KG/U 13 C- AcAc(n=3) Age 81 ± ± 7 68 ± 6 79 ± 1 ǂ Weight (g) 363 ± ± ± ± 17ǂ ph 7.35 ± ± ± ± 0.02 Physiological Parameters Breath Rate (/min) 62 ± 3 67 ± 5 68 ± 4 70 ± 1 Hematocrit (%) 45 ± 1 44 ± 2 42 ± 2 48 ± 2ǂ Plasma Parameters BHB (mm) 0.29 ± ± 0.54 * 0.25 ± ± 0.81ǂ AcAc (mm) 0.17 ± ± 0.07 * 0.26 ± ± 0.29ǂ BHB+AcAc (mm) 0.45 ± ± 0.35 * 0.51 ± ± 1.10ǂ BHB/AcAc ratio 1.64 ± ± 0.74 * 1.03 ± ± 0.32ǂ L-Lactate (mm) 1.15 ± ± 0.12 * 1.22 ± ± 0.20ǂ D-Glucose (mm) 10.1 ± ± ± ±

121 Rats were divided into four study groups. STD: standard chow diet. KG: ketogenic diet. Rats were constantly infused with either [U 13 C]-glucose or [U 13 C] - acetoacetate (AcAc) for 50 minutes. All data presented are Mean ± SD. * P<0.05 in student 2-t test, when comparing the KG rats with STD rats (both infused with [U 13 C]-glucose). ǂ: P<0.05 in student 2-t test, when comparing the KG rats with STD rats (both infused with [U 13 C]- AcAc). : P<0.05 in student 2-t test, when comparing the KG rat with the STD rat (both infused with the same [U 13 C] tracer). 103

122 Table 5.2 Plasma and brain enrichments of glucose M+6 and ketone bodies M+4. STD/U 13 C- Glucose(n=4) KG/U 13 C- Glucose(n=6) STD/U 13 C- AcAc(n=7) KG/U 13 C- AcAc(n=3) Brain Glucose M6 % 10.6 ± ± Brain AcAc M4 % 97.7 ± ± ± ± 0.4 Brain BHB M4 % 19.8 ± ± 0.3* 50.7 ± ± 3.2ǂ Plasma Glucose M6 % 9.1 ± ± Plasma AcAc M4% 5.2 ± ± 0.7* 26.3 ± ± 3.7 plasma BHB M4 % 3.9 ± ± ± ± 1.1ǂ Tracer Infusion Rate (mmol/kg/hr) *: P<0.05 in student 2-t test, when comparing the ketogenic (KG) rats with standard diet (STD) (both infused with [U 13 C]-glucose). All data presented are Mean ± SD. ǂ: P<0.05 in student 2-t test, when comparing the KG rats with STD rats (both infused with [U 13 C]- Acetoacetate). - : measurement was not performed. AcAc: acetoacetate. BHB: β- hydroxybutyrate. 104

123 105

124 Figure 5.1 Simplified schematics of metabolite labeling patterns with [U 13 C]-Glucose or [U 13 C]-Acetoacetate (AcAc) infusion. All schematics do not account for pyruvate recycling and glutamine-glutamate cycling between astroglial cells and neurons. Backfluxes are indicated by by-directional half arrows. All positional carbons are noted from left to right (C1-C5). Labeled C 13 are presented by filled circles. Panel A: Brain metabolites labeling pattern from [U 13 C]- glucose, only considering pyruvate dehydrogenase (PDH) activities and the first turn of Citric Acid Cycle (CAC) from citrate to oxaloacetate. Panel B: [U 13 C]-glucose tracer or [U 13 C]-AcAc infusion, only considering PDH activities and the second turn of CAC. Panel C: [U 13 C]-glucose tracer, only considering pyruvate carboxylase (PC) activities and the second turn of CAC, from oxaloacetate to malate. Panel D: [U 13 C]-AcAc tracer, considering both PDH and PC activities and the first turn of the citric acid cycle. The second turn of CAC will be the same as shown in panel C. ASP: aspartate; AAT, aspartate aminotransferase; Alpha-KG: α-ketoglutarate; BHB, β-hydroxybutyrate; GAD: Glutamate acid decarboxylase; GLU: glutamate; GLN: glutamine. 106

125 107

126 Figure 5.2 Brain metabolite M2 enrichment from [U 13 C]-glucose studies (Panel A) and [U 13 C]-Acetoacetate studies (Panel B). All data are presented as Mean ± SD. *: P<0.05 in student 2-t test, when comparing metabolites M2 enrichment from rats fed with ketogenic (KG) vs. standard (STD) diets. 2OHG: 2-oxoglutarate. ASP: aspartate. 108

127 109

128 Figure 5.3 Brain metabolite M1 enrichment from [U 13 C]-glucose studies (Panel A) and [U 13 C]-acetoacetate studies (Panel B). All data are presented as Mean ± SD. *: P<0.05 in student 2-t test, when comparing metabolites M1 enrichment from rats fed with ketogenic (KG) vs. standard (STD) diets. 2OHG: 2-oxohydroxyglutarate. ASP: aspartate. 110

129 Figure 5.4 Brain metabolite concentrations in rats infused with [U 13 C]-glucose (Panel A and B) or [U 13 C]- acetoacetate (Panel C and D). Data bar graphs are presented as mean ± SD. *P<0.05 in student 2-t test, when comparing metabolites concentrations between standard diet (STD) vs. ketogenic diet (KG) rat brains. AcAc: Acetoacetate. ASP: aspartate. BHB: β-hydroxybutyrate. 111

130 Figure 5.5 Theoretical schemes for M+1 metabolites generation. Filled circles indicate 13 C labeling. M+1 metabolites that come from malate M+2 after the first turn of citric acid cycle (CAC) are indicated in square circles. Panel A shows the M+1 metabolites from the combination of unlabeled oxaloacetate and acetyl-coa M+1. Panel B shows the M+1 metabolites from the combination of labeled oxaloacetate and unlabeled acetyl-coa. Pyruvate recycling was considered, but not distinguished between neurons and astroglial compartments. Backfluxes are indicated by by-directional arrows. Malate M+3 from [U 13 C]-glucose infusion scenario are ignored due to its low enrichment (<10%) relative to M+2. All carbon positions are noted from left to right (C1-C5). ASP: aspartate; AAT, aspartate aminotransferase; Alpha-KG: α-ketoglutarate; BHB, β- hydroxybutyrate; Fum: fumarate; GAD: Glutamate acid decarboxylase; GLU: glutamate; GLN: glutamine; PC: pyruvate carboxylase; PDH: pyruvate dehydrogenase. 112

131 Figure 5.6 Chromatogram of the Citric Acid Cycle intermediates and neurotransmitters 113

132 Chapter 6 Conclusions & Future works 6.1 Introduction In the previous chapters, we had presented that diet-induced ketosis can i) spare glucose phosphorylation in the brain (chapter 3) ii) decreases the acetyl-coa synthesis from glucose while increases acetyl-coa synthesis from ketone bodies (chapter 4) iii) switch the fuel source for oxidative metabolism and neurotransmitters from glucose to ketone bodies (chapter 5). All the evidence seemed to imply that brain energy balances are essentially the fuel demand balance: the total energy demand of glucose and ketone bodies (CMR o2 that come from both CMR glc and CMR ket ), stays constant. Our future goal, investigation of ketone bodies neuroprotective mechanism, lies on validation of this hypothesis. Future works should be done with these guidelines: On energy metabolism: 1) Studies of the energy fuel utilization and oxidations in humans and animals vary with experimental conditions. It is important to perform meta-analysis of the data (see figure 3.2), apply appropriate normalization to eliminate inconsistencies of the absolute values due to anesthesia, physiological state and species differences. It is also known that the brain energy metabolism has ~20% of it as house-keeping portion during isoelectric state (98), estimated when overdose pentobarbital was applied to animals (no EEG 114

133 signal were detected). Our data can only be compared with the non-isoelectric or awake subjects data. 2) Different methods of ketosis induction play key roles in the stability and effect of ketosis (See section 2.4 and 5.1). The focus of this dissertation is on the explanations of the diet-induced ketosis, with known stable up-regulation of MCT transporters (40), and assumed no change of CBF. The level of the ketosis we observed was comparatively higher than many literature values, as reported by ketosis induced by fasting or 2-3 days feeding of ketogenic diet (see chapter 3 table and figures). It is also higher comparing with previously reported diet-induced ketosis for 3-4 weeks but with calorie restrictions (51, 52), as well as shorter-term diet-induced ketosis in mice (18, 53). The only studies that yield higher mean level of blood ketones were reported are either from infusion of ketones + starvation for 1.5 days (57) and chronic fasting (2) cases. It is important to understand that levels of ketosis, as indicated by both the redox of the BHB/AcAc and total ketone body concentrations, are keys to compare the studies by categories. 3) Age differences and the implications to metabolic energy balances. The current studies of the rats under ketosis were 3 months old adults. Considering the fact the experimental rat life span (2-2.5 years), the animals we used are healthy young adults. In the aged rats, the ketone and glucose metabolism were significantly different. Hence, the translation of our studies to older rats or human subjects requires caution. 115

134 Ketone bodies were to more extent used in developing rat and human brains (32, 43). This had been supported by several experimental reports. In humans, the ketogenic diet, as a regiment to contain epilepsy occurrences, was more efficient when applied to children than adults (129); in suckling rats, the brain A-V differences of the ketone bodies were 3-4 times greater than young adults, suggesting higher ketone body uptake. Furthermore, in animals with brain injuries (contusion) and ketogenic diet applied as a treatment, young rats with ~1 month old exhibited more reduction of contusion volumes compared with ~2.5-months old adults (13). All these evidence implies that the ketone bodies utilization rates and neuroprotective roles were weakened as subject ages. Lastly, the (85) the aging brain volume (Cerebral Blood Volume) has been reported to decrease. There are controversies as to whether the CMR glc decreases in aged human and animals (100, 130), with more recent publications in 2012 (85) pointing out that the global glucose utilization may decrease (CMR glc multiplied by volume). The lesson from the controversies of the energy balances with aging is that the CMR estimations are very dependent on the volume and flow of the system. Age-related alterations of the shift of energy demand (development, or revolution related) (32, 60), as well as the alterations of vasculature volumes cannot be ignored. 4) Clarifications of neuronal and glial metabolism from ketone bodies. Our studies on the oxidative metabolism from ketone and glucose tracers were performed by GC-MS method. Comparing with the NMR method, GC-MS had 116

135 advantages on 1) higher sensitivity, able to measure brain metabolites with nmol/g tissue, whereas NMR method usually can only detect μmol/g concentrations of metabolites in brain. 2) Metabolites with similar structures but different molecular weight can be readily separated, even in if the concentrations are low. For example, glutamate (GLU) and glutamine (GLN) signals are clearly separated in GC-MS chromatogram by the TBDMCS derivatization method (118, 131), whereas NMR method can only distinguish the GLU and GLN peak with C-4 with high sensitivity. Other positional GLU and GLN signals were often overlapped. Separations of the signals were done offline with additional assumptions (116, 132). NMR method, had the advantages on 1) Positional labeling identification. For example, [U- 13 C]-glucose infusion can generate [1,2-13 C 2 ]GABA from PDH activities and [3,4-13 C 2 ]GABA from PC activities (133). In the GC-MS chromatogram, both signals would be GABA M+2 and not identifiable. The NMR spectrum can separate those positional carbon labeling patterns. 2) The acquisition of the signals can be done in vivo instead of ex vivo. Dedicated NMR machine with dedicated animal or human coil can be used to obtain metabolite time activity curve, which can be used to generate ordinary differential equations from isotopic mass balances (116). Considering the trade-offs of using NMR and mass spectrometry, we now propose working in the future with NMR to study the neuronal-glial interactions in ketosis. This will be presented in section 6.3. On the measurement and estimation techniques, 117

136 1) Validations of constant CBF or changed CBF values across study groups. Tracer infusion may or may not perturb the brain physiology, in a dose dependent manner; it is also depending on the baseline state of the investigation. (Implications of infusions that may perturb the systems were discussed in section 2.4.1, and 5.2). As were discussed, the CMR of any metabolite that mainly gets utilize through blood flow (CBF) and reactions can be either measured by Kety-Schmidt method, or compartmental modeling and tracer infusion. The universal assumptions were that diffusions were negligible, and hence either the uptake or reaction rate would be representative of metabolic rate. However, careful examination of the 2-Tissue compartmental model and other multiple tissue compartmental models (84) implies that the assumptions for reaction rates were always dependent on known CBF that is not different than when determined in a separate experiment in literature. High exogenous stable tracers may shift the CBF, and lead to suspicious conclusions. For example, many NMR rat studies on the glutamateglutamine cycling were based on more than 3mmol/kg/hr infusion of the ketone or glucose tracers (18, 39, 53, 56, 103, 117, 123, 134). Assuming a 300g rat has ~7% of the body weight as blood with normglycemia (glucose) at 10mM and mild ketosis at 2mM, it would only have ~0.2mmol of glucose and 0.04mmol of ketone bodies in its body. The loading of the exogenous glutamate was interpreted (30, 112) as exogenous glutamate regulates endogenous metabolism. While reasonable, the changes of the physiological system may well undermine the conclusions. 118

137 2) Validations of the lumped constant (LC) in the FDG-PET experiments. This topic is discussed in section 6.2. Briefly, the FDG-PET and 2-DG methods both require compartmental modeling. The lumped constant is a practical conversion factor that relates tracer (FDG or DG) phosphorylation rate to that of real glucose phosphorylation rate. This number has been reported to shift significantly with age (100), insulin infusion (135), and slightly with hyperglycemia (29, 136). One must carefully examine the LC to validate that what we used in chapter 3.2, LC=0.71, is held true (105) in ketosis. 119

138 6.2 Estimation of the Lumped Constant in ketotic rat brains Objective and specific aims To investigate the Lumped Constant (LC) values of 2-[ 18 ]Fluoro-2-Deoxy-Glucose (FDG) in the young adult wistar rat brain, during fast-induced and diet-induced ketosis, using 1) estimation of brain glucose phosphorylation rate studies by FDG-PET 2) estimation of brain glucose uptake rate with 133 Xenon Infusion and measurement of the cerebral blood flow (CBF). The values of LC obtained at different states of ketosis and different methods will offer us the validation of the Cerebral Metabolic Rate of Glucose (CMR glc ) at steady state measurement by FDG-PET method. We have recently reported that the diet-induced ketosis reduces the CMR glc in a rat model, using 2-[ 18 ] Fluoro-2-Deoxy-Glucose (FDG) and Positron Emission Tomography (PET) technique (chapter 3). The FDG-PET method requires using a correction factor, the Lumped Constant (LC), to estimate the CMR glc. In that study, we assumed a constant LC in the plasma BHB + AcAc levels 0-6mM in the diet-induced ketotic rats. However, insofar we do not have the proof that the LC is real constant across this range and in other ketosis models. Any variations of the LC may change the CMR glc estimation significantly and undermine our interpretations. As of 2013, the LC had never been reported in dietinduced ketotic rats. To validate our previous CMR glc estimation in diet-induced ketotic rats, we propose to investigate the LC in the diet-induced ketosis. We also propose to compare LC values in different models of ketosis. The results of the LC values obtained 120

139 will allow future researchers to study the CMR glc in rats with FDG-PET technique, with better and more confident understanding of the ketotic process and energy balance of glucose and ketones in ketosis Background of the lumped constants To obtain CMR glc by FDG-PET, one would need to assume an appropriate value of CMR FDG to CMR glc ratio, which requires a Lumped Constant (LC; CMR glc =CMR FDG /LC. See appendix I for derivations) that involves FDG and glucose Michaelis-Menten constants (26, 27). The LC value plays vital role in estimating the CMR glc and any change would undermine the data interpretation. We currently assumed LC to be constant in rats with plasma total ketone bodies ranging 0-6mM (Chapter 3 article), and from that we reported the CMR glc decreases during diet-induced ketosis. However, due to the lack of literature data on LC in diet-induced ketotic rats, we will need to obtain the data by ourselves. If we find that LC in ketogenic diet group really did not change compare with the LC in rats fed with standard diet (STD), then our conclusion would be that CMR glc indeed decreases during diet-induced ketosis; if LC increases, then our previous CMR glc data would be an overestimation, i.e, diet-induced ketosis reduce the CMR glc more; lastly, if LC decreases, then our previous CMR glc overestimates, and thus our thought that CMR glc decreases in diet-induced ketosis may not be valid. 121

140 Due to the different physiological and pathological conditions, as well as the method to estimate the LC, and the species, the LC values reported in literature have been very different. Some reported increase of LC by fasting, insulin infusion or ketone infusions while others reported decrease of LC in fasting (15, 17, 56, 103, 135, 136). It is worthwhile to note three important phenomena in investigating the LC in ketosis, with FDG-PET or DG-Autoradiography methods. (i) Both the DG and FDG trap in the brain, due to the 2- position deoxy group on the carbon chains. However, the LC values are different since DG and FDG have different pharmacokinetics (135, 137). It had been shown that the LC for DG and FDG are held a proportionality relationship, so investigating one may lead to understanding of another (93). (ii) Fasting or starvation induced ketosis usually accompanies hypoglycemia (low blood glucose levels than normal), while diet-induced ketosis does not induce hypoglycemia (high blood glucose levels than normal) (16, 51). Ketosis induced by infusion may not cause hypoglycemia (56, 59). In addition, it is unknown whether a classic study of LC(93) had overlooked possible complications from hypoglycemia by infusion insulin(138). (iii) Fasting and diet-induced ketosis do not change the CBF(15, 17, 62), however infusion may increase the CBF(56, 59) Technical and scientific Challenges The possible challenges of the studies are 1) Maintaining the steady physiological states for the animals. The study lasts more than 105 minutes, in which the rat is anesthetized and several injections made. 2) Brain 122

141 surgery for the venous blood (confluence sinus) may be challenging. 3) Parameter estimations of the rate constants may or may not be identifiable for all studies. Different methods of estimation may be needed (29, 81-84). 123

142 6.3 Optimizing the stable isotope studies on oxidative metabolism in ketosis Our previous work (chapter 5) on the oxidative metabolism was performed using stable isotopes and Mass Spectrometry. As discussed, the disadvantages of the GC-MS and LC- MS comparing with NMR methods are 1) lack of distinction for positional carbon labeling patterns 2) cannot measure in vivo brain metabolites. Due to these limitations, we were unable to address two important issues underlying oxidative metabolism in brain during ketosis. The expanded compartmental model would be similar with what had presented by McKenna s review (Figure 6.2). This scheme is favored due to three reasons. First, scheme includes the key glutamate-glutamine cycling pattern from neuron and astrocytes. Often, studies with C1 or C6 labeled glucose tracer, or C2 or C4 labeled ketone tracers were used and glutamate C-4 signals (directly from 1 st turn of CAC) were compared with C3 (from CAC exchange and second turns) signals (31, 116, 122, 132). Second, this scheme includes important pyruvate recycling from both the astrocytes and neurons. We deem that astrocytes may present a large pool of glutamine reserve(19, 58), and astrocytes were shown to present with significant pyruvate recycling activities (18, 117, 120), which matches with what we reported in chapter 5. Although it is not clear whether the neurons have the recycling in vivo (58, 120, 126), it is worthwhile to assume that it did exist in neurons. 124

143 Finally, the GABA synthesis from glutamate is largely neuronal. The astrocytes provide extra reserves of the carbons from glutamine (124). This is important because our reported findings (chapter 5.4) regarding the absence of GABA M+2 from ketone bodies would need this re-examination. Whether one can really distinguish GABAergic versus glutamatergic labeling patterns will need more solid verifications (115). 125

144 6.4 Conclusions In this dissertation, we have examined the effects that diet-induced ketosis i) suppresses the glucose phosphorylation, ii) switches the acetyl-coa synthesis from glucose to ketone bodies, iii) spares glucose shunts to citric acid cycle intermediates, and iv) spares glucose shunts to glutamate and GABA generations. All those evidences imply that diet-induced ketosis readily changes the energy balance of the fuels in rat brain. The findings are crucial in understanding the neuroprotection from ketosis for several reasons. First, the diet-induced ketosis we investigated is the most prevalent therapeutic scheme used in treatment of human epilepsy, which affects more than 1% of the whole population (9). Secondly, the animal model we had used and reported had the similar levels of ketosis comparing with real human studies (8). We also took careful effort to ensure that the physiological parameters matches closely to those reported in normal humans. Thirdly, the findings on the biochemical and metabolic pathways of ketone bodies and glucose are directly linked with the mitochondria respiration and ATP synthesis, as succinate, a key citric acid cycle intermediate, participates readily with the electron transport chain. In addition, the acetyl-coa, which we measured in ketosis, directly reflects the converging point of ketone and glucose metabolism, which is not often reported elsewhere. Lastly, our data directly support the hypothesis that ketone bodies spares glucose shunts to glutamate and GABA, which are key neurotransmitters responsible for neuronal transduction, glutamate toxicity and ictal event in seizures. Our data not only offers reasonable explanations to the neuroprotection from ketosis, but also 126

145 explains the reversibilities of neuroprotection when glucose is re-consumed after ketosis is established (69). Future work on the metabolic explanations of neuroprotections from ketosis would require more validation. It is important to verify that the methodologies used in examining the cerebral metabolism are valid, and the key assumptions are held true when ketosis is introduced (for example, the lumped constant, and the cerebral blood flow). In-depth examinations of neuronal and astroglial partitioning effect of ketosis are also required. Particularly, recent advances in astrocytes researches suggest that the supporting cells are crucial in the maintenance of organisms (139). Our current work attributes the ketone bodies sparing effects to pyruvate recycling. More complex labeling patterns may be elucidated when one can combine the GC-MS analytical expertise with the NMR analysis for the dynamic fuel utilization rates. 127

146 6.5 Figures and tables Table 6.1 Literature Lumped Constant (LC) numbers for 2-Deoxyglucose(DG) and 18 FDG. Referenc e No. Year Subject Phy/Path models Induction Method Trace r Blood Ketone range Baselin e LC studied LC Values (26) 1977 Rat Normal N/A DG Unknown 0.48 N/A (27) 1979 Human Normal N/A FDG Unknown 0.42 N/A (100) 1983 Rat month old N/A DG Unknown 0.50 decreased to 0.42 (137) 1988 Rat Glioma N/A DG Unknown (17) 1989 Human Ketosis 3 week fast FDG Up to 4.3mM 0.57 Decrease d to 0.43 (135) 1990 Rat Hypoglycemia Insulin infusion DG Unknown 0.48 Increased to 1.20 (136) 1990 Rat Hyperglycemi a Glucose infusion DG Unknown 0.48 decreased to 0.36 (15) 1994 Human Ketosis Fasted 3.5 days FDG Total up to 3.2mM 0.70 No change (56) 1996 Human Ketosis Ketone infusion FDG Total up to 2.4mM N/A N/A (140) 1998 Human Malignant tumor N/A FDG Unknown 0.86 increased to 1.40 (105) 2006 Rat Normal N/A FDG Unknown 0.71 N/A 128

147 Figure 6.1 Proposed Neuron-Glial Compartmentation models for ketone metabolism studies. Figure reference from McKenna review 2007, JNR (112). 129

148 Appendix Appendix I Sample Files for PET plasma input functions (.crv) and time activity curves(.tac) Sample.crv file sample-time[seconds] plasma[nci/cc]

149 Sample_cerebellum.tac file start[seconds] end[nci/cc] value[nci/cc]

150

151 Appendix II Matlab code for Gjedde-Patlak analysis % suppose you have three vectors % % vector 1 is the time t % vector 2 is the plasma activity nci/cc data_input % vector 3 is the ROI activity nci/cc ROI_output % You want to plot the gjedde-patlak graph and find slope K %% Getting input function fname=uigetfile('*.crv'); % input function has suffix crv inputfun=dlmread(fname,'\t',1, 0); t_input=inputfun(:,1); data_input=inputfun(:,2); clear ans fname % Getting ROI fname=uigetfile('*.tac'); % output function; i.e time activity curve outputfun=dlmread(fname,'\t',1, 0); t_output= (outputfun(:,1)+outputfun(:,2))./2 ; data_output=outputfun(:,3); clear ans fname %% Now integrate the plasma input function with time 133

152 % first interpolate and make input function smooth t_fine=150:0.1:max(t_input); figure (1) plot (t_input, data_input) % Use spline or interp1 data_fine=interp1(t_input(1499:end), data_input(1499:end), t_fine); figure (2) plot (t_fine, data_fine); data_fine=data_fine'; t_fine=t_fine'; figure (3) t_input1=[t_input(1:1500);t_fine]; data_input1=[data_input(1:1500);data_fine]; plot (t_input1,data_input1, '.-'); % input1 are the fine input function time and curve % Then find the indices of the t_input1 that matches t_output ind=zeros(1, length(t_output)); for i=1:(length(t_output)) temp=abs( t_input1-t_output(i) ); %NOTE : must satisfy max(t_input)> max ( t_output) to have this code work ind(i)=min ( find(temp<0.11) ); % find all matching indices, 134

153 % pick the closest index from t, so we can find the output function % time point that matches the input function time point end % Then integrate the input function with repect to time % % Gjedde Patlak Theory states that the slope is LHS / RHS % left hand side is TAC(t) / PlasmaInputFunction(t) % right hand side is % TimeIntegratedPlasmaInputFunction(t)/PlasmaInputFunction(t) for i=1:(length(t_output)) LHS(i)=data_output(i)/data_input1(ind(i)); RHS(i)= ( trapz(t_input1(1:ind(i)),data_input1(1:ind(i))) )/ data_input1(ind(i)) ; end %% add your manual code over here stem (RHS (end-15:end), LHS(end-15:end),'o');grid on % this plots the patlak graph. we need to find the slope for this one. %% Now finding out the slope p=polyfit ( RHS(end-6:end), LHS(end-6:end),1); % only fit the last time points p(1)*60/10*

154 plasmaglc=6.5*1.2; % steady state plasma glucose level. It is measured. Needs to manually change with different studies. LC=0.71 % Lumped constant is assumed See J Nucl Med January 2007 vol. 48 no %% CMRglc= plasmaglc/lc * p(1) *60/10*1000 % x60 to convert sec^1 to min^1 % /10 to convert water 1L to 100g water tissue gram % x1000 to convert mmole to micromole % final unit for CMRglc umole/100g/min 136

155 Appendix III FDG-PET model and LC measurement 1. Model development Glucose and its radiolabeled derivative 18 FDG which enter the plasma can be transported into interstitial fluid and then into tissue cells. Within the cells, these substrates are phosphorylated to form glucose-6-p and 18 FDG-6-P. Whereas the cellular glucose-6-p can be dephosphorylated, the 18 FDG-6-P cannot be dephosphorylated. (26, 27) Also, both glucose-6-p and 18 FDG-6-P remain within the cells. See figure 2.3. After the phosphorylation process, glucose-6-p undergoes further glycolytic steps in the brain and eventually lost to CO FDG-6-P does not undergo further glycolytic steps. Therefore, at steady state, plasma glucose and 18 FDG-6-P concentrations are not changed; in the brain, glucose, 18 FDG, and 18 FDG-6-P concentrations are also staying constant. However, the glucose-6-p concentration does not reach steady state when other aforementioned metabolites are constant. The rate of glucose-6-p concentration change in the brain is defined as Cerebral Metabolic Rate of Glucose (CMRglc). State Variables C p,c e : Plasma and intracellular glucose concentrations (mm) C m : Glucose-6-P concentration (mm) 137

156 C p *,C e * : Plasma and intracellular labeled FDG concentrations (nci/ml) C m * : FDG-6-P concentration (nci/ml) Transport and metabolic processes Dynamic molar balances of intracellular endogenous glucose and glucose-6-p lead to dce * (1.1) = kc 1 kc 2 e R( C, C ) + kc 4 dt p e e m (1.2) dc m dt = R * ( Ce, Ce ) k4cm k5cm Where R(C e,c e * ) characterizes the forward competitive reaction rate to form glucose-6-p. The k s are first-order rate constants for chemical reaction and transport. From dynamic balances of the 18 FDG tracer, the concentrations change according to (1.3) dc e * dt = k 1 * C p * k 2 * C e * R * (C e,c e * ) (1.4) dc m * dt = R * (C e,c e * ) 138

157 Where R * (C e,c e * ) characterizes the forward competitive reaction rate to form 18 FDG-6-P: The k s are first-order rate constants for chemical reaction and transport of 18 FDG and 18 FDG-6-P. The tissue radioactivities correspond to the measureable output: (1.5) 2 Derivation of the competitive reactions of glucose and 18 FDG In this case, both glucose and 18 FDG can be phosphorylated by the same enzyme (E), hexokinase. The phosphorylation of 18 FDG is inhibitive to the phosphorylation of glucose and vice versa. Here, we define glucose as substrate (S) and 18 FDG as Inhibitor(I). The chemical reaction processes are (For equation , see reference link below: k1 (2.1) [S]+ [E] k [SE] [E]+ [P S ] k 2 k 3 (2.2) [I]+ [E] [P I ] k 4 Note that we assume that only the forward phosphorylation reaction processes are competitive. The de-phosphorylation process is not thought to be competitive in the discussion. Therefore equation (2.1) has unidirectional reaction to form P s. The reaction rate equations are 139

158 ds [ ] = k1[ S][ E] + k2[ SE]; dt (2.3) dp [ S ] == k[ SE] dt di [ ] dp [ I ] = k3[ I][ E] + k4[ PI ] = dt dt For the substrate and the inhibitor at equilibrium, we set the derivatives equal to zero and obtain the equilibrium constants: (2.4) (2.5) K K m i [ E][ S] k [ E][ S] [ ] [ SE] k K 2 = = SE = 4 = = PI = [ I] 3 1 [ I][ E] k [ E][ I] [ ] P k K i m Therefore the total concentration of compounds that contains enzyme would be [ S] [ I] (2.6) [ E0] = [ E] + [ SE] + [ PI ] = [ E](1 + + ) K K from (2.4) and (2.6) we find the intermediate concentration : m i (2.7) [ S] [ E0] [ SE] = K (1 + [ S]/ K + [ I]/ K ) m m i From (2.7) and (1.3) we arrive the product (P s ) production rate (2.8) dp [ s ] [ S] [ E0] ke [ 0][ S] = k[ SE] = k = dt K (1 + [ S]/ K + [ I]/ K ) K + K [ I]/ K + [ S] m m i m m i 140

159 The maximum reaction rate occurs when gets very large: dp [ s ] dt max ke [ ] V 0 m so that (2.9) dp [ s] Vm[ S] = dt K + K [ I]/ K + [ S] m m i 3. Finding the phosphorylation rate of glucose and 18 FDG From equation (2.8), we can write the brain glucose (substrate) phosphorylation rate R and brain 18 FDG (inhibitor) phosphorylation rate R * as follows: (3.1) R(C e,c e * ) = C e V m K m + C * e K m / K * m + C e (3.2) R * (C e,c e * ) = C * * e V m K * m + C e K * * m / K m + C e Here K m and K m * are half-maximum rate concentrations for glucose and 18 FDG, respectively. The V m and V m * are the maximum phosphorylation rates for glucose and 18 FDG, respectively. 141

160 For glucose, if K m >> C e * K m / K m * + C e (meaning the enzyme affinity is very low), then the reaction is approximated as first order: (3.1a) Here we have defined a first order rate constant k 3 to relate the glucose concentration to the phosphorylation rate. Similarly, for 18 FDG when K m * >> C e K m * / K m + C e *, (3.2a) R C V R ( C, C ) V * * * * * * * e m * e e m ( Ce, Ce ) k * 3 = * * Km Ce Km We here define k 3 * to relate the 18 FDG phosphorylation rate to a first order constant. Consequently, the model equations for endogenous glucose (1.1) (1.2) simplify as: (3.3) dc e dt = k 1 C p k 2 C e k 3 C e + k 4 C m (3.4) Similarly, (1.3) and (1.4) leads to (3.5) dc e * dt = k 1 * C p * k 2 * C e * k 3 * C e * and the approximation for the cerebral metabolic rate of FDG: 142

161 (3.6) CMR ; dc * m FDG dt = k 3 * C e * If the FDG is constantly infused for a sufficient long period of time, then becomes constant: dc dt * p = C * 0 p ( ) The Laplace transform C * (s) = L{ C * (t)} applied to the variables of Eq.(3.5) yields (3.7) sc e * (s) = k 1 * C p * (s) (k 2 * + k 3 * )C e * (s) Omit the following equation, which is not needed: sc m * (s) = k 3 * C e * (s) From equation (3.7), when s ==> 0 corresponding to t, we find 0 = kc () s ( kc() s + kc()) s * * * * * * 1 p 2 e 3 e so that (3.8) lim s 0 sc * e (s) = * k 1 k * 2 + k sc * (s) * p 3 C * e ( ) = k 1 * k * 2 + k C * ( ) * p 3 143

162 4. Linking the phosphorylation rate of glucose and 18 FDG concentration: If the dephosphorylation rate is much smaller than phosphorylation rate, then we can relate the cerebral metabolic rate of glucose to brain glucose dcm (4.1) CMRglc = Cek3 Cmk4 = Cek3φ dt where φ = 1 C m k 4 C e k 3. If the rate of loss of the phosphorylated glucose is sufficiently small, then φ 1 or approximately a constant close to one. For the endogenous glucose at steady state (equation 3.3 equal to zero) k 1 C p k 2 C e = k 3 C e k 4 C m = C e k 3 φ At steady state (ss), we can relate brain glucose concentration to plasma concentration as: (4.3) C ( ss) = kc 1 ( ss) /( k2 +φk3) e p From (4.2) & (4.3), (4.4) CMRglc = k 1 k 3 φ k 2 +φk 3 C p (ss) Because the glucose rate constants are hard to estimate, we relate glucose rate constants to 18 FDG kinetic constants, which can be estimated because the radioactivities 144

163 can be measured in the brain non-invasively. The ratio of the FDG phosphorylation kinetic constant from Eq. (3.2a) and the glucose rate constant and (3.1a) is (4.5) If we define a ratio λ of FDG kinetic coefficients to glucose rate coefficients as: (4.6) λ k 1 * k 2 * + k 3 * / k 1 φ k 2 +φk 3 then substitution of Eqs. (4.5) and (4.6) into 4.4) yields (4.7) * * * * * k1φ k 3 kk φc 1 3 p ( ss) kk 1 3 CMRglc= Cp( ss) = = C ( ) * * * * p ss k2 + φk3 f k2 + k3 λf ( k2 + k3) LC where (LC) is assumed to be a constant: (4.8) λ f LC φ and is called a Lumped Constant. Equation (4.7) is used as an operational equation for the estimation of CMR glc. 5. Estimation of 18FDG kinetic constants To evaluate according to Eq. 4.7, the kinetic parameters of the tagged FDG reactions must be estimated. The optimal parameter estimates are those for which 145