Discoveries on the storage of red blood cells and the exposure of cells in culture to xenobiotics

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations 2013 Discoveries on the storage of red blood cells and the exposure of cells in culture to xenobiotics Thomas Joost van 't Erve University of Iowa Copyright 2013 Thomas Joost van 't Erve This dissertation is available at Iowa Research Online: Recommended Citation van 't Erve, Thomas Joost. "Discoveries on the storage of red blood cells and the exposure of cells in culture to xenobiotics." PhD (Doctor of Philosophy) thesis, University of Iowa, Follow this and additional works at: Part of the Toxicology Commons

2 DISCOVERIES ON THE STORAGE OF RED BLOOD CELLS AND THE EXPOSURE OF CELLS IN CULTURE TO XENOBIOTICS by Thomas Joost van t Erve An Abstract Of a thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Human Toxicology in the Graduate College of The University of Iowa May 2013 Thesis Supervisor: Professor Garry R. Buettner

3 1 New medical treatments, compounds that affect human health, nutritional supplements, and other substances, are introduced to society every day. The accurate determination of the potential toxicity from these substances is of critical importance to our society. Goals of the modern toxicologist not only involve the determination of the toxic potential of new substances but also: the elucidation of mechanisms; improving existing assays; and developing new assays to study toxicity. This thesis addresses these goals in two topics fundamental to toxicology. Re-evaluating the expression of dose and susceptibility of cells in culture The exposure of cells in culture to drugs, xenobiotics, and other compounds is one of the first tools used to determine the potential for toxicity. Problems can arise when results of these experiments are translated to next-level toxicity experiments (e.g. animals and humans). I hypothesized that dose in cell culture can be improved by designing and reporting experiments based on dose in moles per cell. When experiments were compared on an extracellular concentration basis, a large apparent variability in toxicity was observed. However, if these same exposures were expressed as moles per cell, all experiments yielded the same toxicity. In addition to the evaluation of mole per cell, I investigated the susceptibility of various cells to 1,4-benzoquinone. I hypothesized that upon exposure to toxins that bind covalently, larger cells would require more molecules per cell of toxin versus a smaller cell to achieve identical toxicities. I found a linear correlation between cell volume (pl) and ED 50 (mole per cell where 50 % cell viability is lost), supporting my hypothesis. This work could improve current cell culture protocols and allow for better and less expensive determination of toxicities.

4 2 Heritability of the red blood cell storage lesion Blood transfusions are an integral part of modern medicine with 5 million people receiving blood each year in the United States. There is growing evidence that red blood cells (RBCs) stored for longer periods are less therapeutically beneficial and could even be harmful to patients. This phenomenon of diminished RBC function with increased time in storage is called the storage lesion. However, there is great variation between different donors in the severity of the storage lesion in their donated RBCs. I hypothesized that part of this variability in the RBC storage lesion is determined by heritable genetic differences. To test this hypothesis, a study using mono- and dizygotic twins was performed to determine the heritability of adenosine triphosphate (ATP), glutathione (GSH), glutathione disulfide (GSSG) and hemolysis in stored blood. Major discoveries in this study include: GSH, GSSG, and the half-cell reduction potential (E hc ) are heritable (57 %, 51 %, and 70 %, respectively) in non-stored RBCs. In addition, ATP was found to be heritable in two different storage solutions (62 % in AS-3, 71 % in CP2D); as well as GSH, GSSG, E hc and hemolysis (59 %, 48 %, 64 %, and 53 %, respectively). These discoveries could eventually be used to develop new genetic tests that would predict the rate of deterioration in stored blood quality on an individual basis. Abstract Approved: Thesis Supervisor Title and Department Date

5 DISCOVERIES ON THE STORAGE OF RED BLOOD CELLS AND THE EXPOSURE OF CELLS IN CULTURE TO XENOBIOTICS by Thomas Joost van t Erve A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Human Toxicology in the Graduate College of The University of Iowa May 2013 Thesis Supervisor: Professor Garry R. Buettner

6 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH.D. THESIS This is to certify that the Ph.D. thesis of Thomas Joost van t Erve has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Human Toxicology at the May 2013 graduation. Thesis Committee: Garry R. Buettner, Thesis Supervisor Larry W. Robertson Douglas R. Spitz Thomas J. Raife Michael W. Duffel

7 ACKNOWLEDGEMENTS I am extremely thankful to Dr. Garry Buettner for being an amazing mentor to me. His guidance and knowledge on all aspects of science have been invaluable. I could not imagine spending my Ph.D. career in a better lab or with a better advisor. Whenever I needed help on practical issues or a good laugh, Brett Wagner was always there to help me. He has made the time in the lab very enjoyable. I am eternally in his debt for teaching me all he knows about HPLC, spectrophotometry and all the other random facts a person could ever need in the future. I would like to thank my Ph.D. committee members, Larry Robertson, Michael Duffel, Thomas Raife, and Douglas Spitz for their tireless support in helping my research. Special thanks to Dr. Raife, for his many insightful discussions, and incredible knowledge, on blood storage and for fully including me in his blood storage research program. I could not have completed my projects if it was not for the incredible assistance from Claire Doskey. Thank you for helping this unorganized person, and I wish her the best of luck in continuing this project Thank you to all the current and former lab members: Jordan Witmer, Cameron Cushing, Justin Moser, Sam Schroeder, Zita Sibenaller, Jessemae Welsh, Malvika Rawal, Dr. Joe Cullen, and Weipeng Bian for their assistance in the lab and outside. It was a true joy to collaborate with all of them on a daily basis. Thank you to my friends: Kranti Mapuskar, Jim Jacobus, and Ian Lai for making Iowa City the best place to live. ii

8 I wish to thank Dr. Prabhat Goswami for his many insights into science and free radical biology, and life stories. I would like to acknowledge the Interdisciplinary Graduate Program in Human Toxicology and Free Radical and Radiation Biology Programs, especially, Patricia Ramstad Buettner, Jennifer DeWitte, and Laura Hefley for being there to answer and help me with all my questions. A special thanks to Monali Goswami for her love and incredible support during my studies. And finally I could never have made it this far in life without the invaluable support of my family: my parents Marlies, and Tom and my brothers, Jelle and Kjeld. iii

9 TABLE OF CONTENTS TABLE OF CONTENTS... iv LIST OF TABLES... vii LIST OF FIGURES... ix LIST OF ABBREVIATIONS... xii CHAPTER I : THEME, BACKGROUND, AND SIGNIFICANCE... 1 Theme... 2 Redox Biology... 3 Introduction... 3 Reactive oxygen species affect cell biology... 4 Glutathione and the GSSG/2GSH redox couple... 5 Quantitative redox biology (QRB)... 6 Re-evaluating Exposure to Toxins in Cell Culture... 7 Introduction... 7 Target theory... 8 Toxicity of 1,4-benzoquinone (1,4-BQ)... 9 Red Blood Cell Storage and the Storage Lesion Introduction Blood and its components RBC collection and storage procedures The storage lesion Reactive oxygen species in RBCs Significance CHAPTER II : MATERIALS AND METHODS Re-evaluating Exposure to Toxins in Cell Culture Materials Exposure methods Clonogenic survival assay Trypan blue ATP measurement Seahorse XF Sample preparation for HPLC-BDD with cultured cells GSH and GSSG in cultured cells with HPLC-BDD Calculation of intracellular concentration Heritability of Substances in Red Blood Cells Study design Recruitment Zygosity testing Materials iv

10 Sample preparation GSH and GSSG in RBCs with HPLC-BDD Calculation of intracellular concentration Calculation of ICC and heritability Hemolysis assay ATP sample preparation ATP assay Metabolon Mathematical Models of the Interaction of Cells and ROS CellDesigner COPASI LabVIEW CHAPTER III : RE-EVALUATING DOSE IN THE DETERMINATION OF CYTOTOXICITY IN CELL CULTURE BASED MODELS Introduction Results and Discussion Expression of dose in moles per cell Cell volume versus susceptibility Seahorse XF Depletion of GSH upon exposure to 1,4-BQ Conclusions Future Directions CHAPTER IV : THE HERITABILITY OF THE RED BLOOD CELL STORAGE LESION Introduction Results and Discussion ATP levels under storage conditions Hemolysis under storage conditions GSH, GSSG and E hc levels under storage conditions Mechanism of total GSH loss under storage conditions Heritability of analytes under storage conditions GSH, GSSG and E hc Known heritable traits in this study population Metabolomics Conclusions Future Directions CHAPTER V : GLUTATHIONE LEVEL IN HUMAN ERYTHROCYTES IS A HERITABLE TRAIT Introduction Results Characterization of study population v

11 GSH, GSSG and E hc in MZ and DZ twins Intra-class correlation of GSH, GSSG, and E hc Heritability of GSH, GSSG, and E hc in RBCs Study population compared to known heritable traits Discussion Zygosity testing The GSSG/2GSH redox couple in RBCs Human population compared to research animals Support for heritability of GSH in RBCs Limitations Implications Future Directions CHAPTER VI : STRATEGIES FOR MODELING THE INTRACELLULAR REDOX BUFFER UPON EXPOSURE TO OXIDATIVE STRESS Introduction Theory Random collision model Kinetic equation Model Results Random collision model Kinetic equation model Discussion Random collision model Kinetic equation model Conclusion CHAPTER VII : SUMMARY Chapter III Chapter IV and V Chapter VI APPENDIX A PHYSICAL AND CHEMICAL PARAMETERS OF RED BLOOD CELLS AND PLASMA APPENDIX B: PROTOCOL FOR GPx ASSAYS APPENDIX C: SAMPLE PREPERATION FOR HPLC-BDD APPENDIX D: BIOGRAPHY REFERENCES vi

12 LIST OF TABLES Table I-1: Anticoagulant-preservation solutions before addition to blood...17 Table I-2: Content of additive solutions before addition to blood...17 Table I-3: Table I-4: Storage expiration regulations for packed RBCs as dictated by the FDA...18 Values of biochemical substances in RBCs after storage in various preservation and additive solutions Table III-1: Cell lines used in toxicity experiments and their associated volumes Table IV-1: Gender distribution of final study participants...73 Table IV-2: Table IV-3: Comparison between the mono- and di-zygotic twin populations in this study...73 Comparison between measured values for both mono- and di-zygotic twin populations...74 Table IV-4: Top 25 heritable metabolites in RBCs stored for 29 days in AS Table IV-5: Table IV-6: Metabolites that have a significant inverse correlation with ATP levels...76 Metabolites that have a significant positive correlation with ATP levels...76 Table IV-7: Some purine nucleotides are heritable in stored RBCs Table IV-8: Poly unsaturated fatty acids in stored RBCs Table IV-9: Thiols and related metabolites in stored RBCs Table V-1: Observed ranges of glutathione in human erythrocytes...97 Table V-2: Characterization of study population...99 Table VI-1: Encounter probability calculations for each protein species Table VI-2: Superoxide and hydrogen peroxide have very short lifespan and diffusion distance within RBCs Table VI-3: Description of the constants used in equations 9 and Table VI-4: Description of the constants used in equation vii

13 Table VI-5: Description of the constants used in equation Table VI-6: Parameters for the glutathione system Table VI-7: Parameters for H 2 O 2 diffusion viii

14 LIST OF FIGURES Figure I-1: The network of reactions central to the removal of O 2 - and H 2 O Figure I-2: Figure I-3: Figure I-4: Figure I-5: Figure I-6: Figure I-7: The low flux and high flux electron circuits of the mitochondrial electron transport chain The intracellular status of the redox buffer indicates the biological state of the cell Theoretical survival curves demonstrating different shapes from which associated mechanisms can be deduced Exposure to xenobiotics can lead to cellular toxicity mediated by H 2 O 2 and subsequent oxidation of the redox buffer Red blood cell glycolysis with the Rapoport shunt and pentose phosphate pathway The relationship between hemoglobin autoxidation and glucose metabolism Figure II-1: Study design for blood storage twin study Figure II-2: Flow chart of the HPLC system with parts and solutions Figure III-1: Figure III-2: Reporting of mole per cell as the measure of dose allows for the comparison between different physical setups Exposures reported in extracellular concentration can be misleading in different physical setups Figure III-3: ATP per cell is decrease with increase dose to 1,4-BQ Figure III-4: 1,4-Benzoquinone reduces cell variability in a dose-dependent manner Figure III-5: Cell volume greatly affects susceptibility to 1,4-BQ...55 Figure III-6: Clonogenic Survival of A549 vs. OCR and ECAR Figure III-7: Glutathione is depleted with a small dose of 1,4-BQ Figure III-8: Loss of GSH upon exposure to 1,4-BQ is instant and does not recover quickly ix

15 Figure III-9: Figure III-10: Figure IV-1: Figure IV-2: Figure IV-3: Figure IV-4: Figure IV-5: Figure IV-6: Figure IV-7: Figure IV-8: Half-cell reduction potential of the GSSG/2GSH redox couple correlates well with clonogenic cell survival upon exposure to 1,4- BQ GSH concentration poorly correlates with clonogenic cell survival upon exposure to 1,4-BQ ATP concentrations decline differently in two distinct storage solutions Concentration of hemoglobin in storage media increases with time in storage GSH and GSSG concentration in RBCs decline with increasing time in storage; the half-cell reduction potential (E hc ) does not significantly change over the storage period Comparison of known heritability constants for weight, height, and BMI to those calculated for the population in this study The intra-class correlation coefficient of ATP and hemolysis are higher for MZ twins then for DZ twins indicating heritability...82 The intra-class correlation coefficient of GSH, GSSG, E hc, and tgsh are higher for MZ twins then for DZ twins indicating heritability...83 ATP and hemolysis are genetically regulated traits under storage conditions the behavior of GSH, GSSG, E hc, and tgsh are partially genetically determined Figure IV-9: The 15 most abundant metabolites in RBCs after 29 days in storage Figure IV-10: Figure V-1: Figure V-2: Figure V-3: Figure V-4: There is a good correlation between protein normalized ion counts and literature values for RBC amino acid content The distributions of GSH, GSSG, and E hc are very similar between MZ and DZ twins The intra-class correlation between MZ twins is greater than between DZ for GSH, GSSG, and E hc The concentrations of GSH, GSSG, and E hc within RBCs behave as heritable traits Distribution of GSH is between two ethnic groups due to prevalence of polymorphisms x

16 Figure VI-1: The removal of intracellular and extracellular H 2 O 2 by cells Figure VI-2: Flow diagram of a single simulation calculation Figure VI-3: Methods depicting spatial calculation methods Figure VI-4: Reactions considered in the random collision model Figure VI-5: Figure VI-6: Figure VI-7: Figure VI-8: General biochemical network as created in CellDesigner for removal of H 2 O 2 by cells Example of simulated motion for an O 2 - and H 2 O 2 molecule in an RBC Iteration steps for collision theory simulator needs to 1 x 10 7 seconds or bellow for accurate results Superoxide or H 2 O 2 generated inside an RBC have next to no change to reach the membrane and potentially escape Figure VI-9: A presentation of the transient levels of simulated species in a RBC Figure VI-10: 2D and 3D pictures of a single instance of a simulator with particle tracking enabled Figure VI-11: Time course of removal of O 2 - inside the RBC Figure VI-12: Traditional kinetic simulation performed by Copasi Figure VI-13: Figure VI-14: Figure VI-15: Simulation of the concentration of extracellular H 2 O 2 after addition to a cell suspension plotted over time Simulation of the intracellular glutathione redox couple upon addition of H 2 O 2 to a cell suspension Simulation of the predicted time to death for six cancer cell lines exposed to fluxes of H 2 O xi

17 LIST OF ABBREVIATIONS 1,4-BQ 1,4-Benzoquinone 2,3-BPG 2,3-Diphosphoglycerate 3-PGA 3-Phosphoglycerate ACD-A Acid citrate dextrose + adenine ADP Adenosine-diphosphate AS-3 Additive solution 3 storage medium ATP Adenosine-triphosphate BDD Boron doped diamond CBC Complete blood count CoQ 10 Co-enzyme Q 10 CP2D Citrate phosphate, double dextrose CPD Citrate phosphate dextrose CPDA-1 Citrate phosphate dextrose adenine CYP P450 Cytochrome P450 DETAPAC Diethylenetriaminepentaacetic acid DZ Dizygotic EDTA Ethylenediaminetetraacetic acid E hc Half-cell reduction potential GPx Glutathione peroxidase GR Glutathione disulfide reductase GSH Glutathione GSSG Glutathione disulfide Hct Hematocrit Hb Hemoglobin ICC Intraclass correlation coefficient MCH Mean corpuscular hemoglobin MCV Mean corpuscular volume MZ Monozygotic NADPH Nicotinamide adenine dinucleotide phosphate Prdx Peroxiredoxin PUFA Poly unsaturated fatty acids QRB Quantitative redox biology R 2 Coefficient of determination RBC Red blood cell SOD Superoxide dismutase WBC White blood cell xii

18 CHAPTER I : THEME, BACKGROUND, AND SIGNIFICANCE 1

19 2 Theme This thesis focuses on the need for quantitation and individualization in many aspects of chemistry, biology and toxicology. With absolute quantitation as an alternative to relative information, data can be interpreted in a more comprehensive manner. Comparisons across biology (i.e. absolute amounts of chemicals) and toxicology (i.e. amount of toxin vs. target) can be made; and data are directly exchangeable and comparable among different physical assay setups and laboratories. Quantitation can also lead to individualization. Individualization challenges current research philosophies, be it with cells, animals, or humans, where subjects are classified and analyzed in groups. Examples are smokers, diabetics, gene knockout animals, etc. By grouping individuals together, valuable information about biological variability can be lost. This problem can be overcome by quantitation where no one group needs to be treated as a control and all individuals can be compared to each other. The themes of this thesis are investigated in three interrelated projects that involve: 1. The determination of the heritability and influence of redox biology in the RBC storage lesion, 2. Improving the data gained from toxicity determinations in cell culture by altering the expression of dose. I hypothesize that altering experimental approaches that include quantitation and individualization could greatly improve and increase the information obtained. These new approaches would aid in designing new therapies, increase our knowledge of redox biology, and improve the determination of toxicities for compounds and identify individuals at risk for adverse effects.

20 3 Redox Biology The redox biology of cells is now recognized as an integral component to cellular health and regulation. Redox biology studies the interaction of oxidants and anti-oxidants with cellular processes e.g. proliferation, signaling, gene-expression and metabolism [1]. Aspects of redox biology are investigated further in chapters, IV, V, and VI in this thesis. Introduction It is now being recognized by the greater research community that redox active molecules (glutathione, thioredoxin, co-enzyme Q 10, etc.) in conjunction with reactive oxygen (ROS), nitrogen (RNS) and sulfur species (RSS) are integral to the regulation of biological processes. These reactive species affect the redox state of the many redox couples, proteins, and pathways. Altering the intracellular redox state can lead to changes in the biology of cells (proliferation or apoptosis) and organisms (health or disease). To investigate the exact contribution of each pathway, quantitative information on the redox active species and reversible redox couples that provide the connection among metabolism, protein function, and gene expression is needed. This includes the concentration of ROS, RNS and RSS; the actual concentrations of the antioxidant enzymes (superoxide dismutases (SOD), glutathione peroxidase (GPx), catalase, etc.); as well as the redox state and concentrations of many protein and small molecule thiols (glutathione, thioredoxin, glutaredoxin, peroxiredoxins etc.).

21 4 Reactive oxygen species affect cell biology To begin to understand redox biology, the interaction of superoxide (O - 2 ) and hydrogen peroxide (H 2 O 2 ) with cellular function must first be understood. It is these species in conjunction with SOD and the peroxide-removal system that set the redox environment of cells and tissues (Figure I-1). The superoxide dismutases are a family of enzymes that catalyze the dismutation of O - 2 to form H 2 O 2 and dioxygen (O 2 ) (Equation 1). O O H + H 2 O 2 + O 2 (1) Mitochondria are thought of as a source of O - 2 from oxidative phosphorylation [2]. The vast majority of electrons flowing through the electron transport chain go through the high flux circuit where they bring about the four-electron reduction of dioxygen to produce water; the energy is captured in ATP (Figure I-2). However, a very small percentage of these electrons (low flux circuit) go to dioxygen to produce O - 2, a - one-electron reduction. Under healthy conditions, the O 2 produced is quickly converted to H 2 O 2 by manganese superoxide dismutase (MnSOD), followed by removal by a - peroxide-removing enzyme. Under pathological conditions, the steady-state level of O 2 can increase, leading to reactions with iron-centers in enzymes and proteins that are sensitive to oxidation/reduction [3]. This is referred to as one-electron signaling, and could lead to the modulation of electron flow, metabolism, and gene expression [4,26]. Increasing the flux of O - 2 will also increase the flux of H 2 O 2 [4]. This could lead to modification of protein and enzyme by reactions with redox switches (cysteine residues) regulating their functions. It is proposed that these signaling processes affect cellular processes, for example: rate of metabolism, changes in gene expression, proteins

22 5 synthesis, cell cycle control, etc. By regulating these processes the biological state of cells and tissues is dictated. Glutathione and the GSSG/2GSH redox couple Glutathione is thought of as the major redox buffer in cells. The function of the buffer is to mitigate the effects of oxidative insults on proteins, enzymes and other biomolecules [53]. This is hypothesized as a protection mechanism, analogous to ph buffers, which stabilize the ph in a system. The status of the redox buffer is described by the half-cell reduction potential of the GSSG/2GSH couple (E hc ). The status of GSSG/2GSH couple is proposed as a marker for the biological state of a cell [53]. The status of GSSG/2GSH couple is given in V, or mv, and is calculated from the absolute molar concentrations of GSH and GSSG in the cell using the Nernst equation (Equation 2) [53]. Figure I-3 describes the hypothesis where proliferating cells are the most reduced with an GSSG/2GSH couple around -240 mv [53]. Oxidation of this couple can lead to different biological state e.g. quiescence, differentiation, apoptosis and necrosis. The absolute numbers in this scheme might not apply to all cell lines; the relative 30 mv difference between the reduction potential might be a better indicator for the prediction of cellular states. E hc = * Log ([GSH] 2 / [GSSG]) in mv ph = 7.4 (2) The status of the GSSG/2GSH couple is directly affected by ROS through the glutathione peroxidase/reductase (GPx, and GR respectively) systems [5]. The selenocysteine in GPx reacts with H 2 O 2 forming an oxidized form of GPx (selenic acid) [116,6]. This oxidized form is recycled by two sequential reactions with GSH; these

23 6 reactions lead to the formation of GSSG from the two GSH. The GSSG can be recycled by the enzyme GR which utilizes NADPH to reduce GSSG to two GSH [5,7]. GPx has several other isoforms most notably, GPx 4 which reacts with organic hydroperoxides and lipid hydroperoxides to protect cell membrane [8]. The utilization of GSH and oxidation to GSSG oxidizes the redox buffer. Quantitative redox biology (QRB) There is a wealth of data on antioxidants and redox active species in the literature, with preliminary attempts to create databases to organize this information being undertaken in our lab 1. However, much of the data is difficult to include in a widely usable format for multiple reasons; for instance, relative measurements that are not comparable because they are measured with different assays, physical conditions and non-uniform units. Other values are difficult to include because the normalization among different cells cannot be compared (i.e. amount of protein per cell is not the same; therefore comparisons between cells could be misleading). The goal of quantitative redox biology is to develop new and adjust existing methods to make comparison among different physical setups, organisms, and cell lines possible [1]. And important adjustment would be to change the reporting of the levels of enzymes, small molecules, and other substance levels in either number of copies or mole per cell. Distinctions should be made between active copies of an enzyme and total 1 Accessed

24 7 copies. Next generation reporting would be fraction with specific post-translational modifications To account for physical differences between cells, units similar to number of copies per liter of intracellular volume or intracellular concentration (mole per liter) should be reported. This later unit could allow for direct comparison between cell lines. Quantitative redox biology in many instances deals with fluxes of reactants, for example flux of ROS or rate of oxygen consumption. Some of these species can only be analyzed through kinetic models. Here using SI compliant units e.g. mole per cell per second will be very important. These units can be easily transferable to computer models and other kinetic calculations. Adjusting current protocols to report units based on QRB guidelines could lead to: simplification of comparison among experiments; more and improved information being generated from experiments; and the drawing of better experimental conclusions. Re-evaluating Exposure to Toxins in Cell Culture This section describes new approaches to the exposure of cells in culture. These approaches are further investigated in chapter III of this thesis. Introduction Cells in culture can be used to screen for the potential toxicity of substances including: metals, nutrients, medications, and environmental pollutants. General toxicity tests can be carried out on many cell types (e.g. fibroblasts, epithelial cells, or organspecific cells like hepatocytes or neurons). A number of parameters including: metabolic activity, cytosolic enzyme release, cell growth and cloning efficiency are used as end-

25 8 points to measure toxicity. Organ-specific toxicity can be tested by measuring alterations in specific cell functions (e.g. glycogen metabolism in primary hepatocytes, beating rate in myocytes, and phagocytosis in macrophages). Major problems in the interpretation of results obtained from cell experiments still remain since the biological effects observed may change depending on the exposure conditions (e.g. incubation time and amount of toxicant). Also, other factors, related to the chemical kinetics of a toxin such as rates of absorption, biotransformation, distribution and excretion, which influence the dose of cells in vivo can be difficult to duplicate in vitro. Even when the appropriate cell type is used, intrinsic cell sensitivity depends on a number of cell characteristics that are likely only partially preserved in vitro; these can include chemical biotransformation, binding, membrane permeability, surface determinants, intracellular synthetic pathways, and adaptive and recovery mechanisms. Therefore, unless a well-planned experimental design is used, it may be difficult to translate results from in vitro experiments to in vivo models. Target theory Target theory is used in the field of radiation biology to describe the relationship between radiation dose and biological effect e.g. protein damage, cell death. The theory was first described by Lea in 1946 [9]. Lea hypothesized that there are a given number of targets in cells that are sensitive points. When enough of these sensitive points are damaged, the cell undergoes some kind of biological effect e.g. apoptosis, necrosis, or senescence. By plotting the radiation dose vs. response of cells on a log-linear scale, mechanistic predictions based on the shape of the curve can be made (Figure I-4).

26 9 It is undeniable that radiation is a toxin [10]; it is believed that radiation toxicity is mostly due to the generation of hydroxyl radicals through the radiolysis of water in the cell. These radicals will covalently bind to biomolecules and possibly alter their function [11]. Therefore, I hypothesize that the relationship of dose and effect for other toxins that exert their toxicity via covalent binding can be described in part by target theory. Recent work by Gülden et al. describes the toxicity of bolus additions of H 2 O 2 to cells in culture on the basis of a per cell dose, i.e. moles of H 2 O 2 per cell [12]. This paper was based on earlier work by Spitz et al. [13] and others. The authors concluded that the primary descriptor of toxicity is nmol / mg cell protein [12]. The initial dose concentration, i.e. the extracellular concentration at the beginning of the experiment, did not appear to correlate with the observed toxicity. The mechanism behind H 2 O 2 as a cellular toxin involves irreversible oxidation of proteins, lipids, DNA, and the redox buffer [14,15,16]. All these mechanism have different degrees of importance in the overall toxicity of H 2 O 2. Target theory and mole per cell doses could be great tools to improve the results obtained from cell culture based toxicity experiments. Toxicity of 1,4-benzoquinone (1,4-BQ) 1,4-Benzoquinone (1,4-BQ) is the simplest of quinone structures. Quinone structures are found in many natural products as electron carriers (e.g. CoQ 10 ) and poisons [17]. Not all quinones are present by intent; most can be introduced by metabolism of aromatic rings through multiple oxidation steps or by oxidation of hydroquinones (Figure I-5). In this fashion, 1,4-benzoquinone is a metabolite of benzene formed through two rounds of cytochrome P450 oxidation in the liver. The toxicity of

27 10 1,4-BQ is believed to be related to its ability to act as an electrophilic Michael acceptor, which gives it the ability to make covalent bonds [18, 19]. Quinones can covalently bind to biomolecules containing free thiol and amine groups acting as nucleophiles. Covalently bound quinones can disrupt many biochemical processes as well as damage the cellular structure (Figure I-5) [20]. Since 1,4-BQ binds covalently to biomolecules, the dose will greatly depend on the absolute amount (mole or gram) present. The concentration of the compound will probably poorly describe the dose dependence since mass-action or kinetics do not directly drive 1,4-BQ toxicity. Therefore, 1,4-BQ is an ideal candidate to study with respect to target theory in toxicology. Red Blood Cell Storage and the Storage Lesion This section describes the red blood cell (RBC) and the processes used to optimize their storage for transfusion into humans. These problems are further investigated in chapters IV and V of this thesis. Introduction The successful collection and storage of human blood for transfusion was achieved almost 60 years ago. Many challenges had to be overcome such as how to keep the blood from clotting and how to keep bacteria and fungi from growing. The risks associated with blood transfusions have always been linked to the spread of diseases for example HIV and hepatitis. Through modern screening techniques, this risk has been significantly reduced. With these problems addressed, the next challenges revolve around

28 11 the quality of the blood during prolonged storage and the therapeutic efficacy of transfused units. The current benchmark for the quality of storage is how many cells lyse during the storage period (i.e. percent hemolysis) and how many cells remain in circulation 24 hours post-transfusion. These parameters are measured during the development of a storage solution. Limits for these parameters are mandated and regulated by the FDA in the United States. It has become clear however, that the transfusion of old blood can increase morbidity and mortality [21]. There appears a correlation with these adverse reactions and storage duration of the blood unit. This leads to the conclusion that, during storage, the RBCs lose part of their functionality and could be detrimental to a patient s health. The loss of function of RBCs during storage has collectively been referred to as the storage lesion. Some of the biochemical changes involved in the storage lesion have been well characterized and appear to relate to decreases in ATP and 2,3-bisphosphoglyceric acid (2,3-BPG an allosteric modulator of hemoglobin oxygen affinity). Other aspects of the storage lesion include the creation of membrane micro vesicles, change in RBC shape, and the RBC becoming more rigid. These biochemical changes lead to increased hemolysis and less effective RBCs. I hypothesize that the change in these known biochemical markers, as well as markers of oxidative damage, will dictate the severity of the storage lesion. The severity of the RBC storage lesion will depend mostly on the donor s genotype and much less on environmental influences like donor diet or life style. This could lead to new screening methods to determine the shelf life of blood products from individual donors.

29 12 Blood and its components Red Blood Cells: Erythrocytes (RBCs) are biconcave disk shaped cells of 8 µm by 2 µm [22] with an intracellular volume of 90 fl [23]. RBCs are found in the bloodstream at a density of x 10 6 cell µl -1 in whole blood from males and x 10 6 µl -1, in whole blood from females [22]. RBCs contain approximately pg of tetrameric hemoglobin per cell [22], which is the primary protein responsible for storage of molecular oxygen. Another notable characteristic about these cells is the lack of a nucleus or other organelles. The primary site of RBC production is the bone marrow, where they are generated from hematopoietic stem cells. See Appendix A1, for a table with RBC proteins and their intracellular concentrations. Plasma: Plasma is the liquid component of whole blood. It composes about 55% of whole blood. It is mostly comprised of water with a variety of hormones, glucose, proteins and ions. See Appendix A2, for a table listing plasma anti-oxidants and their concentrations. Platelets: Platelets or thrombocytes are the fragments of megakaryocytes and circulate in the blood of mammals to produce growth factors and aid in clotting. Platelets are nonnucleus-containing cells and are 2-3 µm in diameter. In a healthy person, there are x 10 9 platelets per liter.

30 13 RBC collection and storage procedures Storage of red blood cells for transfusion has become a routine practice. In the United States, blood is collected from donors either by collecting whole blood or by a process called apheresis. During apheresis the blood is passed through a machine that separates the RBCs from the other blood components. The RBC-depleted blood is then transfused back into the donor, whereas the RBCs themselves remain in the machine for further processing. RBCs can also be collected after the donation of whole blood. Separation is achieved by centrifuging the whole blood; the heavier RBCs form a pellet at the bottom and can be isolated from the other blood components. After isolation of the RBCs, a preservation solution is added to prevent coagulation and hemolysis (7:1 volume ratio of blood to preservation solution). There are several kinds of preservation and additive solutions in use today (Table I-1). Each preservation solution can be used individually (ACD-A = acid citrate dextrose + adenine; CPD = citrate phosphate dextrose; CP2D = citrate phosphate double dextrose; CPDA-1 = citrate phosphate dextrose adenine); after removal of plasma this method allows for storage up to 35 days depending on the formulation (Hct 75%) (Table I-2). An additional procedure can be performed where the plasma is replaced with 100 ml of one of three additive solutions (AS) (Table I-2). This procedure extends the storage time for all preservation solutions to 42 days (Hct 60% (Table I-3). Which AS is selected depends on the preservation solution utilized.

31 14 The storage lesion As red blood cells age in storage, they undergo biochemical and physical changes collectively referred to as the storage lesion. The reason why these changes occur is largely unknown; however, it is known that the adverse effects seen in patients correlate with excessive hemolysis of RBCs. Storage related hemolysis is linked to an inability of the RBCs to deform to their environment when they pass through narrow capillaries. The biochemical changes that occur are roughly summarized as a loss of ATP and 2,3 BPG and an increase in free hemoglobin (Table I-4). The loss of ATP is associated with the decrease in deformability and the inability to maintain optimal osmolarity inside the cell due to the reduced function of sodium and potassium pumps on the RBC membrane. The loss of 2,3 BPG results in a reduced capacity for RBCs to deliver oxygen. Due to the absence of 2,3 BPG, molecular oxygen has a greater affinity for hemoglobin. This makes the RBC more of an oxygen sink then a delivery vessel. The mechanism behind the loss of ATP and 2,3 BPG is believed to be associated with a more acidic intracellular ph. Since the storage solution does not contain mechanisms to break down the acidic pyruvate and lactate that are produced through glycolysis, the ph of both the extracellular media and the RBC cytosol become more acidic over time ( 0.02 ph units per day). This decrease in intracellular ph accelerates the conversion of 2,3 BPG to 3 PGA. The RBC tries to recover the 2,3 BPG levels to baseline by diverting glycolytic intermediates through the Rapoport-Leubering shunt to synthesize new 2,3 BPG. The downfall of going through this shunt is that there is no net gain in ATP (Figure I-6). These biochemical changes and others, lead to less functional

32 15 RBCs and decreased retention of cells upon transfusion, resulting in potentially poorer therapeutic efficacy and increased body burden. Reactive oxygen species in RBCs In normal circulation as well as in storage, oxyhemoglobin is not a stable form of hemoglobin. Once in a while hemoglobin donates an electron to the bound oxygen; this releases the oxygen in the form of O - 2. This reaction occurs in human RBCs at a rate of 1 nm s -1 or about 50 molecules of O - 2 per cell-second [24]. The O - 2 that is produced is dismuted by the enzyme superoxide dismutase (SOD). This reaction produces hydrogen peroxide (H 2 O 2 ) and molecular oxygen. The total flux of H 2 O 2, if all O - 2 is dismuted via this enzyme, would behalf the flux of superoxide which calculates to 0.5 nm s -1 or 43 µm day -1. H 2 O 2 is detoxified principally by three enzymes namely, catalase, glutathione peroxidase 1 (GPx) and peroxiredoxin 2 (Prdx) (major isoforms in RBC cytosol). Glutathione peroxidase and Prdx need to be recycled by intracellular reducing equivalents in the form of GSH and thioredoxin to preserve their H 2 O 2 removal capacity. Thioredoxin and GSH are in turn recycled by NADPH, which is produced by the metabolism of glucose (Figure I-7). The H 2 O 2 and O - 2 produced could be a driving component of the RBC storage lesion, especially after loss of reducing equivalents during prolonged storage. The potential role of reactive oxygen species in stored RBCs was therefore investigated.

33 16 Significance I think my discoveries could impact the approaches used to determine and analyze drug toxicity, which could be significant for pharmaceutical companies and other institutions performing toxicity determinations. My projects on the heritability of metabolites could revolutionize the way people approach the field of redox biology. It would most likely impact the development of therapies based on pro-oxidants and anti-oxidants. The future methods, storage solutions and management of blood storage could be greatly impacted by my results on the heritability of metabolites in stored RBCs. These findings can lead to improving the outcomes of thousands of transfusions each year and alter the world of blood banking significantly. The mathematical kinetic models could be used as first tools to better our understanding of ROS and their behavior in cellular systems in ways never imagined before. Finally, developing and applying quantitative redox biology in assays will greatly increase the information gained from experiments and improve the use and translatability of their results.

34 17 Table I-1: Anticoagulant-preservation solutions before addition to blood Component ACD-A CPD CP2D a CPDA-1 (g L -1 ) (g L -1 ) (g L -1 ) (g L -1 ) Trisodium citrate Citric acid Dextrose Monobasic sodium phosphate Adenine 0.28 a Used as initial collection solution in UIHC DeGowin Blood Center. From reference [25]. Table I-2: Content of additive solutions before addition to blood Component a AS-1 AS-3 b AS-5 (Optisol) (Nutricel) (AS-3) Dextrose Adenine Monobasic sodium phosphate Mannitol Sodium chloride a All values are in mm. b Used as extended storage medium at UIHC DeGowin Blood Center. From reference [25].

35 18 Table I-3: Storage expiration regulations for packed RBCs as dictated by the FDA Storage solution Storage duration a ACD 21 days CPD 21 days CP2D 21 days CPDA 35 days AS-1 AS-3 AS-5 42 days 42 days 42 days No anticoagulant 24 h a Durations are for storage at 1-6 C. From reference [25]. Table I-4: Values of biochemical substances in RBCs after storage in various preservation and additive solutions. Variable CPDA-1 AS-1 AS-3 AS-5 Day 0 Day 35 Day 0 Day 41 Day 0 Day 41 Day 0 Day 41 2,3 BPG (%) 100 < <5 100 < <5 ATP (%) Potassium ions (mm) Free Hemoglobin (mg/l) Ammonium (µm) 470 Sodium ions (mm) ph From reference [25].

36 19 Figure I-1: The network of reactions central to the removal of O 2 - and H 2 O 2. SOD is central as it controls the steady-state level of superoxide and can contribute to the flux of H 2 O 2.from processes like mitochondrial metabolism. The flux of H 2 O 2 affects both nodes of the peroxide removal/redox system, i.e. both the GSH and thioredoxin nodes. Figure from reference [5].

37 20 Figure I-2: The low flux and high flux electron circuits of the mitochondrial electron transport chain. The low flux circuit has a feedback loop, i.e. O 2 - inactivates aconitase, a key enzyme in the citric acid cycle. Thus, O 2 - can act as a governor for the high flux circuit. Superoxide will activate the HIF system, leading to expression of a large number of genes. These genes affect metabolism, angiogenesis, and induce cell proliferation, to name a few. Manganese superoxide dismutase (MnSOD), an enzyme in the mitochondria, controls the state level of O 2 -, converting it to H 2 O 2. The flux of H 2 O 2 is a key element in setting the redox state of glutathione (GSH), the cellular redox buffer. MnSOD is at the fulcrum between one-electron signaling pathways and two-electron signaling pathways. Figure from reference [26].

38 21 Necrosis & Apoptosis Figure I-3: The intracellular status of the redox buffer indicates the biological state of the cell. Cells that are actively dividing, or functioning normally, have a redox buffer that is in the reduced state. Cells that have a more oxidized buffer could become quiescent, undergo differentiation, or could undergo apoptosis or necrosis depending on both time and severity of over-oxidation. Adapted from [53].

39 Surviving fraction / % Dose / A.U. Figure I-4: Theoretical survival curves demonstrating different shapes from which associated mechanisms can be deduced. (Long dashed line) Single-hit, multi-target model. This model assumes that there are many independent targets per cell that must be inactivated before biological effects are observed. (Solid line) Linear-quadratic model. This model is described as having many targets of which only 1 has to be inactivated for biological effects to be observed. (Short dashed line) Single-target model. This model assumes that there is only one critical target per cell that must be inactivated before biological effects are observed.

40 23 Hydroquinone OH X Cytochome P 450 PAH, PCB X OH NQO1 GSH Cys Figure I-5: Fe III, Cu II X O O Quinone O 2 H 2 O 2 Cellular toxicity (Oxidation of the redox buffer; GSH) Cellular toxicity (Michael addition to biomolecules) Exposure to xenobiotics can lead to cellular toxicity mediated by H 2 O 2 and subsequent oxidation of the redox buffer. Formation of H 2 O 2 by hydroquinones is achieved by transfer of electrons to molecular oxygen; this is accelerated by metals. The oxidation products can be recycled; this leads to the non-stoichiometric generation of H 2 O 2.

41 24 Figure I-6: Red blood cell glycolysis with the Rapoport shunt and pentose phosphate pathway. The RBC uses this pathway to produce ATP and NADH to maintain cellular homeostasis and NADPH as reducing equivalents. The Rapoport shunt produces 2,3 BPG as an allosteric effector to decrease hemoglobin oxygen affinity. Going through the shunt does not allow for the production of ATP. The dephosphorylation of 2,3 BPG to 3 PGA is accelerated by increase in the intracellular ph as is seen in long term blood storage scenarios.

42 25 Figure I-7: The relationship between hemoglobin autoxidation and glucose metabolism. The oxidation of oxyhemoglobin to methemoglobin results in the formation of O 2 -. Superoxide is converted to H 2 O 2 by superoxide dismutase. Hydrogen peroxide is either removed by catalase, which does not require any co-factors; or it can be removed by the GPx or Prdx systems, which require NADPH for enzyme regeneration. The NADPH is generated from glucose in the pentose phosphate cycle. All species with the dashed red ovals are capable of crossing the RBC membrane. Hx is hypoxanthine.

43 CHAPTER II : MATERIALS AND METHODS 26

44 27 Re-evaluating Exposure to Toxins in Cell Culture In this section, the materials and methods used in the study of toxicity are described. These methods are used in chapter III. Materials MIA PaCa-2, C6, HepG2, MDA-MB-231, A549, and v79 cells were purchased from American Type Culture Collection (Manassas, VA). DMEM high glucose media were from Invitrogen supplemented with 10 % fetal bovine serum (FBS) and Penicillin Streptomycin (100 units per ml). Sufficient media is prepared to complete an experiment and its replicates. All media contains FBS from the same bottle; this is done to remove variations in the FBS from the experimental conditions. Exposure methods Cells were plated and allowed to grow in either 25 cm 2 or 75 cm 2 flasks. Exposure was started when cells reached 80 % or more confluence. 1,4-Benzoquinone was prepared in different concentration in DMSO. From these stock solutions 200 μl was added to fresh media in a 50 ml conical tube (exposure media). Prior to exposure the growth media was removed from the cells and replaced with the exposure media. Cells were exposed for 4 hours. After exposure, cells were trypsonized and washed with 25 ml of sterile PBS. Cells were pelleted and resuspended in cold 1 ml of sterile PBS. Counting was performed by hemocytometer on a 20-fold dilution of the original stock

45 28 Clonogenic survival assay From the 1 ml cell suspension, 50 L was pipetted in 2 ml of fresh media. This 2 ml aliquot was further diluted to 200 and 400 cell/ml for: MIA PaCa-2, MB-231, v79, C6 and A549. Dilution of 2000 and 5000 cell/ml were used for HepG2. 3 ml of diluted cell suspension were plated in 60 mm dishes. For each exposure, 3 dishes were plated for each density. Cells were allowed to grow for 7 to 14 days in a 5 % CO 2 incubator at 37 C. After the growth period, media was removed, cells fixed with 70 % ethanol and stained with Coomassi blue. Plates are allowed to dry and are then counted under the microscope. A colony is determined to be a grouping of 50 or more cells. The plating efficiency and surviving fraction are calculated using equations 3 and 4 respectively [27]. Plating Efficiency (PE) = number of colonies / number of cells plated (3) Survival Fraction (SF) = PE of treated sample / PE of Control 100 (4) From the plots of clonogenic survival fraction vs.1,4-bq dose, the dose of 1,4-BQ at which 50 % clonogenic survival is observed is determined. This dose is referred to as the effective Dose 50 (ED 50 ). Trypan blue Trypan blue is a dye-exclusion method (dead cells are unable to export the dye whereas live cell can). Trypan blue is considered a measure of membrane integrity. For the assay, 10 L 0.4 % trypan blue is added to 10 L of trypsonized cells. The sample is then mixed and cells are count on a hemocytometer. Distinction is made between stained cells and unstained cells. Cell viability is calculated using equation 5.

46 29 Cell Viability = unstained cells / total 100 (5) ATP measurement 50 L of cell suspension (10,000-40,000 cells) was added to each well in a 96 well plate (black). To this, 50 L of solution from an ATP kit (Promega, CellTiterGlo) was added to lyse the cells and start the luminescence reaction. Luminescence was measured after 10 minutes on Tecan microplate reader. ATP signal was normalized to total cells present to give a measure of ATP per cell. Seahorse XF96 Cells were seeded into XF96 cell culture plates 24 or 48 h before experiments. Cells were exposed for 4 hours in 200 L fresh DMEM media with 10 % FBS in the seahorse plate. After exposure, media was exchanged with Seahorse MEM medium with 25 mm glucose and 1 mm sodium pyruvate. OCR and ECAR was determined using standard approaches for this technology [28,29,30], using XF96 FluxPaks (37 C) from Seahorse Bioscience. Sample preparation for HPLC-BDD with cultured cells Cells were removed from the dish with trypsin/edta. After washing with 10 ml of PBS, cells are transferred to a 15 ml conical tube. The sample is centrifuged and the supernatant removed. Pellets are resuspended in 1 ml PBS and transferred to a 2 ml eppendorf tubes. An aliquot is removed for cell counting or protein assays. The sample is centrifuged again and the supernatant removed. The pellet is resuspended in 100 L 5% PCA/ 100 M diethylenetriaminepentaacetic acid (DETAPAC). The sample is stored at -

47 30 80 C until the day of analysis. Prior to analysis the sample is thawed and centrifuged to remove the precipitated proteins. GSH and GSSG in cultured cells with HPLC-BDD The method for determining GSH and GSSG is described in GSG and GSSG in RBCs with HPLC-BDD on page 34 of this thesis. Calculation of intracellular concentration For each sample, the amount (mole) of GSH and GSSG per RBC was determined via the standard curve specific to each run. First, the GSH amount was divided by the number of cells associated with each sample (number obtained by coulter counter, hemocytometer or Moxi Z mini automated cell counter) giving a value of mole per cell. The mole per cell value was then divided by the cell volume (value obtained from coulter counter, literature reference or Moxi Z mini automated cell counter) resulting in an intracellular concentration of mole per L for GSH and GSSG. From these molar concentrations the status of the GSSG/2GSH couple was calculated using the Nernst equation assuming an intracellular ph of 7.4 and 37 C, E hc = * Log ([GSH] 2 / [GSSG]) in mv [53]. Heritability of Substances in Red Blood Cells In this section, the study design, and recruitment method for the twin study are described along with the methods for biochemical measurements and statistical analysis. These methods are used in chapters IV and V.

48 31 Study design To investigate the influences of genetics on the storage lesion, mono (MZ) and dizygotic (DZ) twins were recruited. Participants were recruited over an 8 month period. On a given week, a maximum of three people were scheduled to donate one unit of whole blood at the DeGowin Blood Donor Center at The University of Iowa (Figure II-1). In conjunction with the standard unit of blood, a sample of blood was collected in a Vacutainer purple top tube; this sample represents day 0 and is only analyzed for GSH and GSSG concentrations. The day 0 sample is collected and processed within 1 hour after being drawn. The whole unit of blood is processed overnight according to standard blood bank procedures for extended storage in the AS-3 storage solution. After processing two samples are prepared from the bag: 1) a segment of tubing is sterilely filled with RBCs preserved in AS-3; 2) a segment of tubing with RBCs preserved in CP2D. These samples represent day 1. Subsequently on days 8, 15, 29, 43, and 57 samples are drawn in a similar fashion from the stored units. Recruitment Twins (19 pairs; 14 MZ and 5 DZ pairs) were recruited to donate blood at the DeGowin Blood Center at the University of Iowa Hospitals and Clinics. Prior to standard blood donation, samples of whole blood for this study were collected. Twins were not required to come in as a pair; nor were there restrictions on the time of day for donations. This resulted in individuals within pairs to donate weeks apart and at different times of the day.

49 32 Zygosity testing The zygosity of twin pairs was determined by isolating DNA from white blood cells (WBCs). White blood cells were obtained from leuko-reduction filters, utilized during the processing of the whole blood donation into components. Dulbecco's Phosphate-Buffered Saline (DPBS, 15 ml) was pushed through the filter to extract the WBCs. The DPBS with WBCs was collected into a 50 ml centrifuge tube. The 50 ml tube was centrifuged at 500 g for 10 min. WBCs are resuspended in 2 ml of DPBS. DNA was extracted using the AutoGen (Holliston, MA) QuickGene-610L nucleic acid extraction machine with the Fuji QuickGene DNA Whole Blood Kit (AutoGen), following manufacturer s instructions. Genotyping was performed with 24 single nucleotide polymorphisms (SNPs) chosen for an unrelated project. SNP genotyping was performed with TaqMan assays (Applied Biosystems, Foster City, CA) on the EP1 SNP Genotyping System and GT48.48 Dynamic Array Integrated Fluidic Circuits (Fluidigm, San Francisco, CA). MZ twins had 90 % or greater genotype concordance out of 24 total SNPs. All other twin pairs were considered DZ. Materials Perchloric acid, glutathione, glutathione disulfide, 1-octanesulfonic acid, sodium mono-phosphate, and DETAPAC were obtained from Sigma Aldrich (St. Louis, MO, USA). CELLPACK analyzer solution was obtained from Sysmex Corporation, Kobe, Japan.

50 33 Sample preparation Day 0 samples (fresh blood) Whole blood (EDTA, Vacutainer purple top blood collection tube, 8 ml) was collected from donors prior to blood donation. The sample was centrifuged at 500 g for 5 min, followed by removal of the plasma and buffy coat. RBCs were washed twice with cold isotonic saline solution. After washing, a 30 µl aliquot of the packed red blood cells (prbcs) was removed for complete blood count (CBC) analysis (Sysmex XE-2100 Automated Hematology System). Three, 400 µl aliquots of prbcs were each lysed with 500 µl of a 5 % perchloric acid/100 µm DETAPAC solution; this precipitates the protein and preserves glutathione (GSH) and glutathione disulfide (GSSG). Samples were thoroughly mixed and stored at -80 C for a minimum of 1 day and a maximum of 2 months prior to HPLC analysis. On the day of analysis, the three samples were thawed at room temperature (30 min) and centrifuged to pellet the protein (4000 g, 5 min). The supernatant ( 800 µl) for each sample was transferred to a separate Eppendorf tube. After an aliquot was removed for analysis, the remaining sample was stored at -80 C. Day 1 to 57 samples (stored blood) RBC segments were collected sterilely from the blood units. The segment was drained into a 5 ml eppendorf tube and centrifuged at 500 g for 5 min, after which the RBC-free storage media (AS-3) was removed. RBCs were washed twice with cold isotonic saline solution. After washing, an aliquot was taken for CBC analysis (Sysmex XE-2100 Automated Hematology System). 300 µl of RBCs was lysed with 400 µl 5 % perchloric acid/100 µm DETAPAC; this precipitates the protein and preserves

51 34 glutathione (GSH) and glutathione disulfide (GSSG). The sample was centrifuged to pellet the protein (4000 g, 5 min). The clean supernatant was stored at -80 C or immediately analyzed using HPLC. GSH and GSSG in RBCs with HPLC-BDD To determine the amount of GSH and GSSG in RBCs, HPLC with electrochemical detection (ESA Coularray) was used following the protocol outlined by Park et al. [31] (Figure II-2). The method is based on an electrochemical detection (ECD) system using a boron-doped diamond disc (BDD) electrode (Model 5040, ESA Biosciences, Chelmsford, MA, USA). Samples (65 µl) were loaded into auto sampler vials with a 100 µl glass insert. Samples were stored in the temperature controlled (+4 C) auto sampler until analysis. At time of analysis the sample (10 µl) was loaded on the column and eluted for 60 min (number of RBCs on column = million). With each set of samples, four standards containing GSH ( pmole on column) and GSSG ( pmole on column) were included; this standard curve was used to quantify the analytes for that particular sample set. Quantitation was performed by integrating the GSH and GSSG peaks in the BDD electrode channel with ESA Coularray for Windows version Calculation of intracellular concentration For each sample, the amount (mole) of GSH and GSSG per RBC was determined via the standard curve specific to each run. First, the GSH amount was divided by the number of RBCs associated with each sample (number obtained from CBC) giving a value of mole cell -1. The mole cell -1 value was then divided by the mean cell volume

52 35 (MCV; value obtained from CBC) resulting in an intracellular concentration of mole L -1 for GSH and GSSG. The median of the three independently processed samples was taken to reflect the intracellular GSH and GSSG concentration of an individual. From these molar concentrations the status of the GSSG/2GSH couple was calculated using the Nernst equation assuming an intracellular ph of 7.4 and 37 C, E hc = * Log ([GSH] 2 / [GSSG]) in mv [53]. Calculation of ICC and heritability For twin studies the one-way model of ICC is used to determine the similarity of a measure in a twin pair, ICC = (MS between - MS within ) / (MS between + MS within ), where MS between is the estimate of the mean-square variance between all twin-pairs and MS within is the estimate of the mean-square variance within the sets of pairs in that group [77]. The ICC is a variant of ANOVA analysis and is described on a scale of +1 to -1. An ICC that approaches +1 would be a near perfect one-to-one correlation within twin pairs and a large variation between twin pairs; an ICC of 0 would result from a set of data-points spread far apart among twin pairs with no one-to-one correlation within pairs; and an ICC of -1 would result from a data set with no one-to-one correlation within twin pairs and a small variation between twin pairs. A correlation coefficient approaching +1 is expected in MZ twins for a strongly heritable trait. The ICC of MZ and DZ pairs for GSH, GSSG and E hc were calculated using IBM SPPS Statistics version 20. From the ICC values heritability was estimated using the method derived by Newman, Freeman, and Holzinger, h 2 = (ICC mz - ICC DZ ) / (1 - ICC DZ ) [32].

53 36 Hemolysis assay The collected RBC-free storage medium (AS-3) was centrifuged and the supernatant was diluted with isotonic saline 4.5-fold (900 µl total volume). This diluted AS-3 solution was analyzed for free hemoglobin with UV/VIS spectroscopy (HP 8453 diode array spectrophotometer) in a 1 cm quartz cuvette using absorbance at 415 nm ( 415 = 128,000 cm -1 M -1 ). A single reference wavelength of 700 nm was used to correct for baseline drift. The molar amount was converted to grams of hemoglobin assuming a molecular mass of 64,500 g mol -1. To determine the percentage of hemolysis samples were compared to total hemoglobin as measured from day 1 CBCs. ATP sample preparation Segments of stored blood in both AS-3 and CP2D storage media where collected at day 1 of the storage period. Following the initial sampling, segments were collected and processed every two weeks until 57 days post-donation. The RBC segments where processed by draining into 15 ml conical tubes. From a tube, 250 µl of blood was diluted with 365 µl of isotonic saline and 35 µl 70% perchloric acid (PCA). After addition of PCA, the protein precipitate s and the sample centrifuged at 4000 g for 5 min. The clear supernatant (415 µl) was transferred to a new 1.5 ml eppendorf and 25 µl of 3 M potassium carbonate solution. The sample was centrifuged again and the clear supernatant stored at -80 C.

54 37 ATP assay The ATP concentration was determined with a kit from DiaSys Diagnostic Systems GmbH (ATP Hexokinase FS cat # ) based on NADPH production. The original protocol was adapted to work with 96-well microplates. Absorbance was measured at 340 nm with a Tecan SPECTRAFluor Plus microplate reader. A standard curve of ATP from 0 to 1 mm was made for ATP quantitation in a 96- well format. Metabolon Sample Preparation: The sample preparation process was carried out using the automated MicroLab STAR system from Hamilton Company. Recovery standards were added prior to the first step in the extraction process for QC purposes. Sample preparation was conducted using a proprietary series of organic and aqueous extractions to remove the protein fraction while allowing maximum recovery of small molecules. The resulting extract was divided into two fractions; one for analysis by LC and one for analysis by GC. Samples were placed briefly on a TurboVap (Zymark) to remove the organic solvent. Each sample was then frozen and dried under vacuum. Samples were then prepared for the appropriate instrument, either LC/MS or GC/MS. Liquid chromatography/mass Spectrometry The LC/MS portion of the platform was based on a Waters ACQUITY UPLC and a Thermo-Finnigan LTQ mass spectrometer, which consisted of an electrospray

55 38 ionization (ESI) source and linear ion-trap (LIT) mass analyzer. The sample extract was split into two aliquots, dried, then reconstituted in acidic or basic LC-compatible solvents, each of which contained 11 or more injection standards at fixed concentrations. One aliquot was analyzed using acidic positive ion optimized conditions and the other using basic negative ion optimized conditions in two independent injections using separate dedicated columns. Extracts reconstituted in acidic conditions were gradient eluted using water and methanol both containing 0.1 % formic acid, while the basic extracts, which also used water/methanol, contained 6.5 mm ammonium bicarbonate. The MS analysis alternated between MS and data-dependent MS 2 scans using dynamic exclusion. Gas chromatography/mass Spectrometry The samples destined for GC/MS analysis were re-dried under vacuum desiccation for a minimum of 24 hours prior to being derivatized under dried nitrogen using bistrimethyl-silyl-triflouroacetamide (BSTFA). The GC column was 5 % phenyl and the temperature ramp is from 40 C to 300 C in a 16-min period. Samples were analyzed on a Thermo-Finnigan Trace DSQ fast-scanning single-quadrupole mass spectrometer using electron impact ionization. The instrument was tuned and calibrated for mass resolution and mass accuracy on a daily basis. The information output from the raw data files was automatically extracted as discussed below. Mass Determination and MS/MS fragmentation The LC/MS portion of the platform was based on a Waters ACQUITY UPLC and a Thermo-Finnigan LTQ-FT mass spectrometer, which had a linear ion-trap (LIT) front

56 39 end and a Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometer backend. For ions with counts greater than 2 million, an accurate mass measurement could be performed. Accurate mass measurements could be made on the parent ion as well as fragments. The typical mass error was less than 5 parts per million. Ions with less than two million counts require a greater amount of effort to characterize. Fragmentation spectra (MS/MS) were typically generated in data dependent manner, but if necessary, targeted MS/MS could be employed, such as in the case of lower level signals. Mathematical Models of the Interaction of Cells and ROS In this section, the software packages used in the various simulation scenarios are described. These programs are used in chapter VI. CellDesigner 2 CellDesigner is a structured diagram editor for drawing gene-regulatory and biochemical networks [33]. Networks are drawn based on the process diagram, with graphical notation system proposed by [34] and are stored using the Systems Biology Markup Language (SBML), a standard for representing models of biochemical and generegulatory networks. Networks are able to link with simulation and other analysis packages through Systems Biology Workbench (SBW) [35]. 2 accessed

57 40 COPASI 3 COPASI is a software application for simulation and analysis of biochemical networks and their dynamics [36]. COPASI is a stand-alone program that supports models in the SBML standard and can simulate their behavior using ODEs or Gillespie's stochastic simulation algorithm; arbitrary discrete events can be included in such simulations. COPASI carries out several analyses of the network and its dynamics and has extensive support for parameter estimation and optimization. COPASI provides means to visualize data in customizable plots, histograms and animations of network diagrams [36]. LabVIEW LabVIEW is a graphical programming environment used by millions of engineers and scientists to develop sophisticated measurement, test, and control systems using intuitive graphical icons and wires that resemble a flowchart. LabVIEW offers unrivaled integration with thousands of hardware devices and provides hundreds of builtin libraries for advanced analysis and data visualization all for creating virtual instrumentation accessed accessed

58 41 Figure II-1: Study design for blood storage twin study.

59 42 Figure II-2: Flow chart of the HPLC system with parts and solutions. The scheme represents the HPLC-BDD system and method for preparing the sample, along with part names and solution compositions

60 43 CHAPTER III : RE-EVALUATING DOSE IN THE DETERMINATION OF CYTOTOXICITY IN CELL CULTURE BASED MODELS

61 44 Introduction Historically, the toxicologic study of compounds has been performed using animal models. Today, NIH, EPA and other government regulatory and funding agencies actively encourage the development and implementation of non-animal based studies (NIH revitalization act of 1993 sec.404c). One such approach uses cells or tissues in culture. These approaches have many advantages over the traditional animal-based studies; more experiments can be performed with various endpoints, and experiments are often more economic than typical animal experiments. The difficulty with cell culture experiments is poor translatability to whole organisms [37]. There are many cell lines to choose from, and it is often at the discretion of the investigator what cell line to use. Therefore, results from in vitro studies may reflect only a subset of what may occur in whole organisms. Another shortcoming in most tissue culture experiments is the normalization for variations in the population. In animal experiments, compounds are administered as a specific amount of compound per gram body weight. This normalization for the weight of the animal is necessary to equalize dose in different size animals (e.g. compare dogs to mice) and equalize differences between animals of the same species. However, in tissue culture experiments this normalization is sometimes overlooked; the total mass of cells or cell density exposed is rarely fully accounted for. The potential toxicity of compounds to cells in culture is, in most modern publications, expressed by the EC 50 value. The EC 50 value represents the half-maximal dose of a compound that elicits some biological response. The EC 50 is mostly expressed as an extracellular concentration. Because this value is given as a concentration,

62 45 duplication of the experiment can be challenging unless very specific physical conditions are provided. The problem with expressing dose as an extracellular concentration is that it is dependent on two variables: mass of compound present, and the volume of the exposure solution (e.g. cell culture media). I hypothesize that it is the absolute amount of compound (number of molecules of toxin per cell) that will best describe the dose of toxins that make or change covalent bonds. A second problem with expressing dose in concentration is that it is not immediately obvious to what number of cells the toxin was exposed. This omission can create confusion when the experiment is duplicated; it can even lead to misinterpretation of the data when conclusions are drawn with regard to susceptibility of a cell line to a toxin. With these problems in current cell culture experiments I hypothesize that toxicity estimates in cell culture models can be improved by designing and reporting dose in moles per cell. I also hypothesize that upon exposure to toxins that bind covalently, larger cells would require more molecules per cell of toxin versus a smaller cell to achieve identical toxicities. This study can serve as an example for new approaches to increase and improve the information gained from cell culture toxicity experiments. This can lead to the use of fewer animals in the determination of toxicities and better translation of results to toxicity in humans.

63 46 Results and Discussion Expression of dose in moles per cell To examine the use of moles per cell as a better method of expressing dose, two different experiments were performed: 1) treatment of A549 cells with 1,4-BQ in different physical setups (e.g. cell density and media volume); 2) exposure of MIA PaCa- 2 cells to identical extracellular concentrations of 1,4-BQ and varying volumes of media (5 ml to 40 ml). In both experiments, great variability in the clonogenic survival was observed when the dose of 1,4-BQ was reported in extracellular concentrations (Figure III-1). This variability was greatly reduced when dose was reported in moles per cell (Figure III-2). Other endpoints of toxicity like ATP per cell (Figure III-3) and trypan blue staining followed the results obtained for the clonogenic assays. Cell volume versus susceptibility To examine the correlation between cell volume and susceptibility to a toxin, six different cell lines with distinct volumes were exposed (Table III-1). Cells were exposed to different mole per cell amounts of 1,4-BQ to create dose response curves (Figure III-4). The dose resulting in 50 % loss of viability (ED 50 ) was calculated and plotted vs. cell volume (Figure III-5). A linear correlation (R 2 = 0.89) was observed between the two parameters suggesting cell susceptibility to 1,4-BQ is related to intracellular volume. Cell volume is thought of as a surrogate marker for the amount of targets present per cell. This relationship should hold if all cells are examined in similar media formulations. When a cell is grown in a different media formulation, the target to water ratio might have changed, disrupting the ability of intracellular volume to act as a

64 47 surrogate for number of targets. It should therefore be investigated if amount of protein per cell is a more stable surrogate marker for targets per cell. Seahorse XF96 To investigate the potential influence of 1,4-BQ on mitochondrial respiration and glycolysis, oxygen consumption rate (OCR) and the extracellular acidification rates (ECAR) where measure with a Seahorse XF96 analyzer (Seahorse bioscience, North Billerica, MA). A549 cells were exposed to 1,4-BQ in a 96 well XF flux plate for 4 hours. Immediately after exposure the plate was read. No statistical difference in the oxygen consumption rate or extracellular acidification rate between cells exposed to DMSO and different exposures of 1,4-BQ was observed (Figure III-6). This suggest that mitochondrial function is not affected by exposure to 1,4-BQ under these experimental conditions, and can thus not explain the toxicity observed. Depletion of GSH upon exposure to 1,4-BQ Since 1,4-BQ is known to react with thiols through a Michael addition reaction, the depletion of intracellular GSH was investigated. There is a dose-dependent depletion of intracellular GSH upon exposure to 1,4-BQ (Figure III-7). The dose required to achieve complete depletion is about 10X as much 1,4-BQ as there is GSH per cell. There is also no accumulation of GSSG, meaning that there is little to no production of H 2 O 2 from 1,4-BQ. The kinetics of this reaction appears to be very fast (Figure III-8); also no recovery of GSH levels upon exposure is observed with 24 hours. When comparing GSH levels and the half-cell reduction potential (E hc ) to determine clonogenic cell survival, a

65 48 much better correlation with E hc is found (R 2 = 0.7) (Figure III-9) than with GSH concentration (R 2 = 0.1) (Figure III-10). Conclusions The expression of dose for 1,4-BQ in cell culture is best described on a mole per cell basis. Reporting dose in this manner will allow for: a comparison between different physical setup s, allow for direct comparison with intracellular molecules, and improve the translation of cell culture experiments to animals and humans by including cell mass. The susceptibly of cells to 1,4-BQ toxicity is strongly linked to the intracellular volume, with bigger cells being more resistant to 1,4-BQ induced toxicity vs. smaller cells. This observation can be explained by target theory. There was no immediate observed change in mitochondrial function as measured by the OCR or glycolysis as measured by ECAR upon exposure of 1,4-BQ to A549 cells. This is an unexpected result since a significant decrease in clonogenic survival is observed. This leads us to conclude that 1,4-BQ induced toxicity is unrelated to metabolic functions. The focus should be on targets of cellular replication like cell cycle proteins or proteins involved in DNA replication and protein biosynthesis. Surprisingly, 1,4-BQ is able to significantly decrease the intracellular concentration of ATP. With the knowledge that ATP generating capacity is intact (conclusion from unchanged OCR upon exposure), it is hypothesized that 1,4-Q exposure leads to an increased utilization of ATP, possible due to ATP being involved in the mitigation of 1,4-BQ induced toxicity. Lastly, one of the first intracellular molecules to react with 1,4-BQ upon exposure is GSH. This rapid reaction is evident by the rapid depletion of GSH after exposure of

66 49 even small doses of 1,4-BQ. Of note is that there is no generation of GSSG during the exposure, suggesting the generation of ROS is a small contributor to 1,4-BQ induced toxicity. Glutathione depletion is however not a driver of toxicity since GSH depletion upon exposure does not correlate to observed clonogenic survival. Future Directions Firstly, I believe that the observations made for this toxin might not apply to other toxins. Therefore, more experiments should be done on a wide variety of toxins to determine which behave in a manner similar to 1,4-BQ and which will behave differently with regards to mole per cell as the describer of toxicity. This should be done both on the dosing scheme and the susceptibility of cells based on physical parameters i.e. cell size. To better apply target theory in toxicology, a quantitative method should be developed to determine the exact amount of total targets and possibly critical targets. At the moment, cell size is a plausible descriptor but could be heavily influenced from experiment to experiments by things like media osmolarity. Lastly, continued research should be focused on identifying the exact biological targets of toxins like 1,4-BQ which enact its toxicity. This will be critical in the development of therapies and interventions to mitigate 1,4-BQ toxicity.

67 50 Table III-1: Cell lines used in toxicity experiments and their associated volumes. Cell Line Type Volume (pl) Reference v79 Chinese hamster lung fibroblast 0.8 [38] C6 Rat glioma 1.08 [39] MDA- Human mammary 1.53 [40] MB231 adenocarcinoma A549 Adenocarcinoma alveolar 1.76 [39] epithelial MIA PaCa-2 Human pancreatic carcinoma 2.03 [40] HepG2 Human hepatoma cells 2.54 [41]

68 Clonogenic Survival / % Clonogenic Survival / % ml of media 2 x 10 6 cells ml of media 1 x 10 6 cells ,4-BQ extracellular concentration / μm ,4-BQ dose / fmol per cell Figure III-1: Reporting of mole per cell as the measure of dose allows for the comparison between different physical setups. Each clonogenic survival curve is one biological replicate in MIA PaCa-2 cells that was performed with a different physical step. The survival estimates of each dose are determined on three plates. Experiments where performed in collaboration with Claire M. Doskey MS.

69 Clonogenic Survival / % Clonogenic Survival / % Clonogenic Survival / % Clonogenic Survival / % Control 5 ml 10 ml 20 ml Control 10 ml 20 ml 40 ml ml ml 80 ml ,4-BQ extracellular concentration / μm ,4-BQ extracellular concentration / μm ,4-BQ dose / fmol per cell ,4-BQ dose / fmol per cell Figure III-2: Exposures reported in extracellular concentration can be misleading in different physical setups. Clonogenic survival of MIA PaCa-2 cells was measured in triplicate for each exposure. Each set of graphs (vertically) is one biological replicate. Date shown in the two sets of vertical graphs are representations of the same experiments where units of dose are reported in extracellular concentration in the upper panels with the lower panel have doses reported as mole per cell. Experiments where performed in collaboration with Claire M. Doskey MS.

70 ATP / RLU per cell E E E E E E E+00 Figure III-3: ,4-BQ dose / fmol per cell ATP per cell is decrease with increase dose to 1,4-BQ. ATP depletion is a previously used marker of cellular toxicity [42]. Each sample was measured in duplicate in MIA PaCa-2 cells. Experiments where performed in collaboration with Claire M. Doskey MS.

71 ED 50 / fmol per cell v79 MIA PaCa 2 HepG2 A549 MB231 C6 Figure III-4: 1,4-Benzoquinone reduces cell variability in a dose-dependent manner. Clonogenic survival of MIA Paca-2, A549, C6, HepG2, MB231, and v79 cell lines after 4 h exposure to 1,4-BQ. C6 cells showed the greatest sensitivity to 1,4-BQ, while v79 showed the least sensitivity. Each cell line has n = 2 for biological replicates. Experiments where performed in collaboration with Claire M. Doskey MS.

72 ED 50 / fmol per cell ED 50 / fmol per cell R² = 0.05 V79 MIA PaCa-2 HepG MDA-MB-231 A C Intracellular Volume (pl) R² = 0.89 MIA PaCa-2 HepG MDA-MB-231 C6 A Intracellular Volume (pl) Figure III-5: Cell volume greatly affects susceptibility to 1,4-BQ The 1,4-BQ exposure at which 50% clonogenic survival was observed for v79 (orange), C6 (green), MB231 (teal), A549 (red), MIA Paca-2 (blue), and HepG2 (purple) cells lines and plotted versus intracellular volume. While the correlation coefficient (R 2 = 0.06) indicates a weak correlation between 1,4-BQ ED 50 and intracellular volume, a correlation can be visually observed with the v79 (orange) as a clear outlier in the trend. When the v79 cell line is removed, the R 2 greatly improves to Experiments where performed in collaboration with Claire M. Doskey MS.

73 Clonogenic Survival / % OCR / amol cell-1 s-1 ECAR / nph cell-1 s OCR ECAR Survival 10 0 Figure III-6: ,4-BQ dose / fmol per cell Clonogenic Survival of A549 vs. OCR and ECAR. Oxygen consumption and extracellular acidification rate where measured on A549 cells following 4-h exposure to 1,4-BQ to determine the rate of oxygen consumption (OCR) and the rate of acid efflux (ECAR) of these cells. Even at the higher exposures of 1,4-BQ as the clonogenic survival approached zero, there was no change that could be detected in their mitochondrial function in terms of OCR and ECAR. Experiments where performed in collaboration with Claire M. Doskey MS.

74 Concentration / mm E hc / mv GSH MIA PaCa-2 = 0.12 fmol per cell E hc GSSG GSH ,4-BQ dose / fmol per cell Figure III-7: Glutathione is depleted with a small dose of 1,4-BQ. Exposure of MIA PaCa-2 cells to 1,4-BQ almost completely depletes (90 %) of the intracellular glutathione in. This depletion is plausible since there is only 0.12 fmol per cell of GSH in MIA PaCa-2 cells and the amounts exposed vastly exceed the intracellular amount.

75 Concentration / M E hc / mv Concentration / M E hc / mv E E E E E E E fmol per cell of 1,4-BQ GSH E hc E E E E E E E E E E E E E+00 GSSG Time / min -150 GSH DMSO Control E hc GSSG Time / min Figure III-8: Loss of GSH upon exposure to 1,4-BQ is instant and does not recover quickly. Following exposure of MIA PaCa cells to 1,4-BQ, there is no generation of GSSG indicating a negligible generation of H 2 O 2. On a mole per cell basis, it can be determined that only 1/100 of the molecules of 1,4-BQ that were present at the start of the exposure reacted with GSH. This experiment was only performed once.

76 Clonogenic survival / % R² = E hc / mv Figure III-9: Half-cell reduction potential of the GSSG/2GSH redox couple correlates well with clonogenic cell survival upon exposure to 1,4- BQ. Experiments were performed with MIA PaCa-2 cells. The results are obtained from one experiment where, after exposure, a fraction of the cells are plated for clonogenic cell survival in the normal fashion. The remainder of the cells were utilized for the measurement of GSH and GSSG with HPLC-BDD.

77 Clonogenic survival / % R² = GSH intracellular concentration / mm Figure III-10: GSH concentration poorly correlates with clonogenic cell survival upon exposure to 1,4-BQ. Experiments were performed with MIA PaCa-2 cells. The results are obtained from one experiment where, after exposure, a fraction of the cells are plated for clonogenic cell survival in the normal fashion. The remainder of the cells were utilized for the measurement of GSH and GSSG with HPLC-BDD.

78 61 CHAPTER IV : THE HERITABILITY OF THE RED BLOOD CELL STORAGE LESION

79 62 Introduction The storage of red blood cells (RBCs) for use in transfusions is one of the most significant medical breakthroughs in the history of man. The ability to provide a safe and constant blood supply has been a hard fought battle for many decades but most of the problems have been addressed. With this achievement, a new problem has emerged, called the storage lesion [43,44,45]. This term refers to the deterioration of RBC function with increased time in storage. These damaged RBCs are linked to increased adverse events after transfusions. The biochemical changes of this storage lesion have been characterized and appear to relate to decreases in ATP (59 % of initial in AS-3 at day 42) and 2,3-BPG (<1 % of initial in AS-3 at day 42) [46]. Besides these general biochemical changes, oxidative damage is believed to play a role in the RBC storage lesion. The principal source of oxidants in RBCs is the autoxidation of oxyhemoglobin [47]. At a slow rate (k = 4.5 x 10-7 s -1 [24]), oxyhemoglobin will donate an electron to the bound oxygen; the oxygen is subsequently released from the hemoglobin protein as superoxide (O - 2 ) (Equation 6). The hemoglobin is oxidized to form methemoglobin. HbFe 2+ + O 2 O HbFe 3+ (6) This significant flux of oxidants could lead to oxidation of the intracellular redox buffer (GSSG/2GSH), proteins and lipid membranes. It is known that glutathione (GSH) which is part of the intracellular redox buffer decreases in concentration with increasing time in storage [48]. Also, anti-oxidant enzymes such as SOD decrease in activity with

80 63 time in storage. This leads us to conclude that the intracellular redox buffer in RBCs could oxidize with increased time in storage and play a role in the storage lesion. From our understanding of the literature, the severity of the storage lesion is not a general phenotype, e.g. the adverse effects of RBC units from different unique donors are variable [49]. This leads us to hypothesize that a genetic component might be involved in the development of the biochemical and oxidative changes of the storage lesion, and the subsequent presentation of adverse effects. The goal of this project is to elucidate if the genotype of donors has an influence on the starting concentrations of intracellular metabolites related to the RBC storage lesion. Their genotype should also dictate the behavior of these metabolites under storage conditions. This goal will be investigated by utilizing the classical twin study design. Measurements were made on a set of monozygotic (MZ) and dizygotic (DZ) twin pairs recruited to donate blood at The University of Iowa DeGowin Blood Center (Table IV-2). By determining the similarity in MZ twin and the dissimilarity in DZ twin of the behavior and concentrations of various metabolites in RBCs, conclusions can be drawn on the influence of donor genotypes on these factors. I hypothesize that: 1) Glutathione (GSH) and glutathione disulfide (GSSG) concentrations and patterns of change in RBCs are more similar in MZ twins and less similar in DZ twins when measured over the storage period. 2) The ATP concentration and rates of decrease in RBCs is the same in MZ twins and different in DZ twins when measured at different time points during storage.

81 64 3) The extent of RBC hemolysis is the same in MZ twins and different in DZ twins when measured at different time points during storage. 4) More mechanistic insights and heritable metabolites can be discovered with metabolomic scans. By elucidating the influence of the donor genotype on the development of the storage lesion, individualized approaches can be explored to prevent or ameliorate the adverse effects associated with the transfusion of old units of blood. This might specifically lead to the development of individual expiration dates for stored blood units. Results and Discussion ATP levels under storage conditions ATP was measured in RBCs stored in two different media formulations. Mean ATP values over the 57 day storage period decreased in both CP2D and AS-3. However, RBCs stored in AS-3 retained their ATP levels longer as is reflected in the mean ATP levels over time (CP2D is 2.4 ± 1.2 mole per g Hb; AS-3 is 4.0 ± 1.3 mole per g Hb) (Figure IV-1). An apparent lag period of 21 days is observed where ATP levels remain stable followed by a steady decrease where ATP levels in AS-3 stored RBCs decrease to 60 % of initial over 57 days in storage. ATP levels in CP2D start decreasing immediately to 20 % of initial over 57 days. These values strongly correlate with literature values (Table I-4) The inter-individual CV for CP2D ATP is 28 %, and AS-3 is 29 %. When comparing mean values between MZ and DZ twin groups for ATP levels no statistical differences were observed (Table IV-3).

82 65 Hemolysis under storage conditions The hemolysis of RBCs follows a linear pattern ( 1300 RBCs or % of total RBCs per day) (Figure IV-2). At day 42 of storage 1.5 % of the RBCs present originally were hemolysed. This is a slight increase over the 1% hemolysis as regulated by the FDA in the United States. This increase in hemolysis may be due to repeated handling of RBC units for sample collection during the storage period. When comparing mean values between MZ and DZ twin groups for hemolysis, no statistical differences were observed (Table IV-3). GSH, GSSG and E hc levels under storage conditions Under processing and storage conditions used by the DeGowin Blood donor center, the intracellular concentration of GSH decreased over the 57-day sampling period (Figure IV-3; GSH). On average, -0.5 mm of intracellular GSH was lost, a rate of -10 µm per day. GSSG also declines under storage conditions at a rate of -2 µm per day (Figure IV-3; GSSG). This is a surprising finding; I expected GSSG levels to increase due to oxidation of GSH leading to the formation of GSSG during the storage period. The observed result indicates that total GSH (tgsh) equivalents are lost (Equation 7). tgsh = 2 * [GSSG] + [GSH] (7) Even though both GSH and GSSG levels are decreased, the reduction potential within the RBC became more oxidized during the entire storage period by +4 mv, a rate of +0.1 mv per day (Figure IV-3; E hc ). When comparing mean values between MZ

83 66 and DZ twin groups for GSH, GSSG and E hc levels no statistical differences were observed (Table IV-3). Mechanism of total GSH loss under storage conditions During storage, there is a loss of tgsh. Two pathways were investigated to elucidate the mechanism behind this phenomenon: 1) GSSG is exported from the cell; 2) GSSG reacts with protein cysteine to form protein mixed disulfides. The first mechanism was investigated by analyzing the AS-3 storage medium after 57 days in storage with HPLC-BDD in the normal manner. GSSG was below the limit of detection for 3 samples analyzed. I therefore conclude that GSSG export is probably not the mechanism behind loss of total GSH. The second mechanism was investigated by treating precipitated protein pellets with 1 mm dithiotreitol (DTT) for 1 h at 40 C. The supernatant was analyzed for released GSH with HPLC-BDD in the normal manner. There is no difference in the released GSH content between day 0 and day 57 samples in the one sample analyzed. I therefore conclude that protein glutathiylation is not the mechanism for loss of tgsh. Heritability of analytes under storage conditions ATP and hemolysis There appears to be a strong mean intra class correlation coefficient (ICC) between the MZ twins for hemolysis and ATP in two distinct storage media formulations AS-3 and CP2D (mean ICC values are 0.75, 0.63, and 0.92 respectively). The ICC is weaker in DZ twins (0.47, -0.10, and 0.66) (Figure IV-5). This stronger correlation for

84 67 MZ twins vs. DZ twins results in estimated heritabilities for hemolysis and ATP in AS-3 and CP2D of 53 %, 77 %, and 66 % respectively (Figure IV-7). GSH, GSSG and E hc There appears to be a strong mean intra class correlation coefficient (ICC) between the MZ twins for GSH, GSSG, and E hc (0.64, 0.56, 0.65 and0.70 respectively); this correlation is weaker in DZ twins (-0.26, 0.15, 0.04 and respectively) (Figure IV-6). This stronger correlation for MZ twins vs. DZ twins results in estimated heritabilities for GSH, GSSG E hc and tgsh, of 71 %, 48 %, 64 % and 76 % respectively (Figure IV-8). Known heritable traits in this study population In order to determine if the subjects in our study exhibited features of an informative twin study population, the heritability of three well-studied heritable traits namely height, weight, and BMI were analyzed. The ICC values for height, weight and BMI in our study population were MZ twins = 0.94, 0.98, and 0.98; for DZ twins = -0.37, 0.31 and 0.92 respectively. This results in an estimated heritability of 96 % for height, 97% for weight and 69 % for BMI in our study population (Figure IV-4). This gives us great confidence that estimated heritability for unknown traits in this study population are informative and did not manifest randomly. Metabolomics Metabolomic scans were performed by Metabolon (Durham, NC) on RBCs stored 29 days under standard conditions. The most abundant metabolites found are

85 68 shown in Figure IV-9. To demonstrate if normalized ion counts represent actual levels within RBCs, literature values for amino acids were correlated with the normalized ion counts (Figure IV-10). The strong correlation seen indicates that normalized ion counts reflect actual concentrations seen within the RBC. In total, the RBCs of 18 twin pairs were analyzed by Metabolon and heritability was calculated together with correlations between biochemically measured analytes (ATP, GSH, GSSG, Ehc, tgsh, and hemolysis) and normalized ion counts for each metabolite. Correlation of metabolite levels with ATP A number of metabolites demonstrated a significant (p < 0.1) correlation with ATP levels, the vast majority of which were inversely correlated (Table IV-5). These many inverse correlations may reflect a general degradation phenotype driven by low ATP levels. In contrast, some metabolites have positive correlations with ATP (Table IV-6); the most notable is ADP. This observation suggests that higher ATP levels in stored RBCs reflect a greater capacity to stockpile ATP, as opposed to increased ATP synthesis or reduced ATP utilization. However, this observation may also be due to the abundance of adenine (pre-cursor to both ADP and ATP). This correlation could thus be the effect of the artificial storage environment and not normal biology. Additional energy-related metabolites that showed a negative correlation with ATP levels were NAD +, a key cofactor in energy production, and panothenate, the precursor to coenzyme A whose synthesis is ATP dependent. An inverse correlation with

86 69 NAD + suggests that maintaining high ATP levels is crucial to keep the levels of NADH and possibly NADPH high to maintain redox homeostasis in RBCs during storage. Heritability of purine nucleotide metabolism Similar to ATP, ADP levels were highly heritable (74%), while the ATP precursors/breakdown products adenine, adenosine, and AMP demonstrated varying degrees of heritability. Although adenine and adenosine levels did not correlate significantly with ATP, levels of the purine catabolites hypoxanthine and allantoin did demonstrate a significant association with ATP levels. In fact, hypoxanthine was inversely correlated with ATP while its downstream metabolite allantoin demonstrated a positive correlation with ATP (Table IV-7). As red blood cells do not contain xanthine oxidase, decreased levels of hypoxanthine and increased levels of allantoin in stored RBCs that have high ATP levels may reflect higher levels of circulating xanthine oxidase in those donors. As urate produced from xanthine may be taken up by RBCs and act as a scavenger for reactive oxygen species, higher levels of circulating xanthine oxidase may be beneficial in reducing the accumulation of oxidative damage during RBC storage. Furthermore, higher levels of hypoxanthine in RBCs during storage may be detrimental, as metabolism of hypoxanthine released from donor RBCs upon transfusion generates H 2 O 2. Hypoxanthine levels were found to be somewhat heritable (17 %). Allantoin levels were determined to be heritable (46 %). Together these results highlight purine metabolism in general and xanthine oxidase specifically as potential genetic traits that contribute to the variation in RBC quality upon storage.

87 70 Heritability of redox homeostasis metabolites Accumulation of oxidative damage and a decline in cellular antioxidant defenses is associated with reduced function of RBCs in storage. Consistent with the hypothesis that low ATP levels in stored RBCs are indicative of reduced transfusion efficiency, levels of the oxidative stress markers cysteine-glutathione disulfide and oxidized glutathione were found to be negatively correlated with ATP levels (Table IV-9). In addition, levels of cysteine-glutathione disulfide were found to be 12 % heritable while oxidized glutathione levels demonstrated 33 % heritability. These results suggest that the mechanisms responsible for differential ATP levels between donors after storage of RBCs may also play a role in the accumulation of oxidative damage during storage. Further supporting a genetic component in the maintenance of redox homeostasis were varying degrees of heritability in levels of glutathione (in the reduced state), the glutathione turnover product 5-oxoproline, and the oxidized cholesterol species 7-αhydroxycholesterol, 7-β-hydroxycholesterol, and 7-ketocholesterol. Phospholipid metabolites correlate with ATP levels Lysolipids, single chain glycerophospholipids, are generated by the action of phospholipase A (PLA) on membrane phospholipids. Several lysolipids containing ethanolamine as a head group were found to be significantly correlated with ATP levels such that lower levels of these species were present in RBC samples with high ATP levels. This phenomenon was not observed for lysolipids containing choline as the head group, but was observed for the single inositol containing lysolipid detected, 1- stearoylglycerophosphoinoisitol. As choline-conjugated phospholipids are most prevalent on the outer leaflet of cellular membranes while ethanolamine and inositol conjugated

88 71 phospholipids are enriched in the inner leaflet of membranes, these results may reflect increased degradation/remodeling of intracellular membranes in RBCs with depleted ATP stores. PUFA and eicosanoid biosynthesis during storage In addition to phospholipid metabolites, levels of several polyunsaturated fatty acids (PUFAs) were found to be both heritable and associated with ATP levels in stored RBCs (Table IV-8). These fatty acids include the ω-6 fatty acid linoleate (18:2n6) and its derivatives dihomo-linolenate (represented by an isobar of 20:3n3 and DHGLA, 20:3n6) and arachidonate (AA, 20:4n6), as well as the ω-3 PUFA docsoapentaenoate (DPA; 22:5n3). The increase in PUFA s in low ATP conditions might be a mechanism to increase the flexibility of RBC membranes, but can also make them more susceptible to oxidative damage. Conclusions During the storage period, the intracellular concentration of GSH and GSSG declines. This leads to a loss of tgsh equivalents from the RBCs. Many possibilities were investigated to elucidate the fate of these removed GSH equivalents however to date; none have proven to be the route of GSH removal. The decrease of GSH and GSSG does not lead to an oxidation or reduction of the mean E hc observed for the population, suggesting that the redox buffer status is not changed. However, since a loss of total GSH is observed, the capacity of the buffer to respond to an oxidative insult is diminished. During the study, we discovered the heritability of GSH, GSSG, E hc and tgsh under storage conditions. Together with the confirmed heritability of ATP in two storage solutions, this will revolutionize the blood banking industry.

89 72 Metabolomic scans from stored RBC samples revealed that many more metabolites have heritable components to their expression. In addition to heritability, we uncovered many more metabolites that correlate with ATP (a marker of RBC transfusion efficacy). This has provided us great insight into the molecular mechanisms and pathways involved in the RBC storage lesion. Future Directions Next in this project should be the identification of the genes involved in the regulation of the various metabolites. This work can be guided by additional metabolomic scans to, more closely, identify specific pathways that are involved in regulation of the heritable factors in RBCs. Since ATP is used as a surrogate marker for 24-h post transfusion recovery of RBCs (The regulatory standard for storage solutions); future research should determine the exact relationship of all heritable factor to this marker of RCB health, and the heritability of the 24-post transfusion recovery. This work should ultimately lead to the development of screening kits that can identify the ideal storage time for a specific donor. However, the heritability was estimated in an affluent population i.e. our result may not hold in population where diet is a limiting factor. The contribution of limiting diet to heritability this should be determined before genetic screening of storage time can be universally applied.

90 73 Table IV-1: Gender distribution of final study participants MZ twins DZ twins Female pairs 11 2 Male pairs 2 2 Male/female pairs - 1 Total pairs 13 5 Table IV-2: Comparison between the mono- and di-zygotic twin populations in this study Trait Monozygotic (MZ) Dizygotic (DZ) P value a Age / Years 24 ± 8 22 ± Weight / kg 64 ± ± Height / m 1.68 ± ± BMI 24 ± ± a Dizygotic versus monozygotic b Median ± SEM

91 74 Table IV-3: Trait a Comparison between measured values for both mono- and dizygotic twin populations Monozygotic Dizygotic P value b (MZ) (DZ) [ATP] AS-3 / µmole g Hb ± ± [ATP] CP2D / µmole g Hb ± ± Hemolysis / % 1.2 ± ± [GSH] / mm 1.3 ± ± [GSSG] / µm 151 ± ± E hc / mv -207 ± ± a Measured over entire storage period b One way ANOVA DZ versus MZ c Median ± SD

92 75 Table IV-4: Top 25 heritable metabolites in RBCs stored for 29 days in AS-3 Biochemical Name Heritability DZ ICC MZ ICC pyroglutamine phenylacetylglutamine phenol red adenosine 5'-diphosphate (ADP) pipecolate isobutyrylcarnitine isovalerylcarnitine propionylcarnitine tryptophan betaine mannitol beta-hydroxycholesterol sorbitol myo-inositol ergothioneine indoleacrylate methyl-2-oxopentanoate alpha-hydroxycholesterol gulono-1,4-lactone methyl-2-oxovalerate choline dihydroxyacetone phosphate (DHAP) N-acetylmannosamine oleoylglycerophosphocholine* palmitoylcarnitine erythronate

93 76 Table IV-5: Biochemical Name Metabolites that have a significant inverse correlation with ATP levels docosapentaenoate (n3 DPA; 22:5n3) glutathione, oxidized (GSSG) oleoylglycerophosphocholine betaine stearoylglycerophosphoethanolamine palmitoylplasmenylethanolamine guanidinobutanoate dihomo-linolenate (20:3n3 or n6) linoleate (18:2n6) guanosine 5'-diphospho-fucose phenylacetylglutamine proline arachidonate (20:4n6) adenine indoxyl sulfate aspartate uridine tryptophan methylthioadenosine (MTA) alpha-hydroxyisocaproate Correlation with ATP Table IV-6: Metabolites that have a significant positive correlation with ATP levels Biochemical Name Correlation with ATP 3-methylhistidine 0.33 Isobar: fructose 1,6-diphosphate, glucose 1,6-diphosphate, myoinositol 1,4 or 1,3-diphosphate 0.33 allantoin 0.36 N-acetylmannosamine 0.41 adenosine 5'-diphosphate (ADP) 0.41

94 77 Table IV-7: Some purine nucleotides are heritable in stored RBCs. Biochemical Name Heritability Correlation adenine adenosine adenosine 5'-monophosphate (AMP) adenosine 5'-diphosphate (ADP) hypoxanthine xanthine urate allantoin Table IV-8: Poly unsaturated fatty acids in stored RBCs. Biochemical Name Heritability linoleate (18:2n6) 26 dihomo-linolenate (20:3n3 or n6) 45 arachidonate (20:4n6) 22 docosapentaenoate (n3 DPA; 22:5n3) 28 docosahexaenoate (DHA; 22:6n3) -47 Table IV-9: Thiols and related metabolites in stored RBCs. Biochemical Name Heritability glutathione, reduced (GSH) 19 5-oxoproline 23 glutathione, oxidized (GSSG) 33 cysteine-glutathione disulfide 12 ophthalmate 4

95 ATP / mole g Hb AS-3 10 CP2D Time in storage / days Figure IV-1: ATP concentrations decline differently in two distinct storage solutions. Segments of tubing from AS-3 and CP2D stored RBCs were collected at various time points during the storage period. Independent duplicate samples from each time point were generated. Each independent sample (n = 2) was measured twice in the AP assay. This resulted in a total n = 4 for each measurement in the graph.

96 Median lysis (%) R² = Time in storage (Days) Figure IV-2: Concentration of hemoglobin in storage media increases with time in storage. Hemolysis was measured by spectrophotometer once for each sample. The slope of the line is 0.02% per day. Plotted are the median and standard deviation for the population (n = 36) at each time point.

97 E hc / mv GSSG / µm GSH / mm Figure IV-3: Time in storage / days GSH and GSSG concentration in RBCs decline with increasing time in storage; the half-cell reduction potential (E hc ) does not significantly change over the storage period. Blood was collected from 36 individuals, processed and [GSH] and [GSSG] were measured using HPLC-BDD. With these values the half-cell reduction potential (E hc ) of the GSSG/2GSH couple was calculated with the Nernst equation. The line in each point represents the median values of all individuals at that time point.

98 Heritability / % BMI Height Weight Figure IV-4: Comparison of known heritability constants for weight, height, and BMI to those calculated for the population in this study. Heritability was estimated using the method of Newman, Freeman, and Holzinger. Estimated heritabilities for these traits match with previously published literature values [78,80,81]. This gives us great confidence that estimated heritability for unknown traits in this study population are informative and did not manifest randomly.

99 ATP AS-3 ATP CP2D Hemolysis ATP AS-3 ATP CP2D Hemolysis ICC Median ICC MZ DZ ATP AS-3 ATP CP2D Hemolysis MZ twins DZ twins Figure IV-5: The intra-class correlation coefficient of ATP and hemolysis are higher for MZ twins then for DZ twins indicating heritability The interclass correlation coefficient of the analyzed traits is measured over the storage period for each pair (MZ = 13 pairs, DZ = 5 pairs). The top panel represents the median ICC values with standard deviation between the twin pairs for each trait. The lower panel displays the range in ICCs. Each data point is one twin pair in their respective categories.

100 ICC ICC Median ICC 83 MZ DZ GSH GSSG Ehc tgsh MZ DZ Ehc Ehc GSH GSSG tgsh GSH GSSG tgsh Figure IV-6: The intra-class correlation coefficient of GSH, GSSG, E hc, and tgsh are higher for MZ twins then for DZ twins indicating heritability The interclass correlation coefficient of the analyzed traits is measured over the storage period for each pair (MZ = 13 pairs, DZ = 5 pairs). The top panel represents the median ICC values with standard deviation between the twin pairs for each trait. The lower panel displays the range in ICCs. Each data point is one twin pair in their respective categories.

101 Heritability / % ATP AS-3 ATP CP2D Hemolysis Figure IV-7: ATP and hemolysis are genetically regulated traits under storage conditions. Heritability was estimated using the method of Newman, Freeman, and Holzinger from the mean intra-class correlation coefficients as shown in Figure IV-5. The estimated heritability value for ATP in CP2D is 77 %, for ATP in AS-3 66 % and hemolysis 53 %.

102 Heritability / % GSH GSSG Ehc tgsh Figure IV-8: The behavior of GSH, GSSG, E hc, and tgsh are partially genetically determined. Heritability was estimated using the method of Newman, Freeman, and Holzinger from the mean intra-class correlation coefficients as shown in Figure IV-6. The estimated heritability value for GSH is 71 %, for GSSG 48 %, E hc 64 % and tgsh = 76 %.

103 Abundance / % Figure IV-9: The 15 most abundant metabolites in RBCs after 29 days in storage. Abundance was determined by dividing the ion count intensity of each metabolite by the total ion count of all metabolites measured for each individual. This value was multiplied by 100 to derive at a percentage of the total ions measured.

104 Amount / mg per 100 ml R² = Ions / mg protein Thousands Figure IV-10: There is a good correlation between protein normalized ion counts and literature values for RBC amino acid content. The normalized signal intensities of 12 amino acids where compared with literature values to determine the usability of the ion counts of a metabolite as a quantitative measure. The literature values are obtained from [50].

105 88 CHAPTER V : GLUTATHIONE LEVEL IN HUMAN ERYTHROCYTES IS A HERITABLE TRAIT

106 89 Introduction Glutathione (GSH) is an important biomolecule ubiquitously present in cells and tissues. One function of GSH is to assist in the detoxification of excess peroxides, facilitated by enzymes like glutathione peroxidase [26] and peroxiredoxins [51]. Glutathione also adducts to electrophilic xenobiotics via glutathione-s-transferases [52]. This addition generally increases the rate of excretion of the xenobiotic from cells and organisms, leading to detoxification. In addition to serving as a co-factor in detoxification reactions, GSH serves as a redox buffer, maintaining the redox status of cells by absorbing oxidative insults. The glutathione disulfide/glutathione redox couple (GSSG,2H + /2GSH) is a major contributor to the intracellular redox buffer-system of cells and tissues [53]. The status of a redox buffer is represented by its half-cell reduction potential (E hc ), determined using the Nernst equation [53,54]. The status of redox buffers can change by altering the balance between the availability of buffer components (i.e. GSH) and demand for reducing equivalents. Changing the status of redox buffers can alter cellular redox circuits and signaling cascades, affecting the functions of enzymes and proteins (e.g. hemoglobin in red blood cells (RBCs, erythrocytes) [55]. Depending on the magnitude, changes in the status of redox buffers can lead to adjustments in cell metabolism and gene expression that ultimately affect the biological state and/or fate (e.g. proliferation vs. apoptosis) of cells [53,54,55]. On the organism level, availability of GSH can: modify risk for a wide range of human health issues; be an indicator of poor overall health; and affect disease severity

107 90 [56]. Statistically significant lower levels of GSH in human RBCs have been observed in many disease states including: acute exposure to drugs and toxins that deplete GSH [57,58], protein-malnutrition [59], hormonal imbalance [60], genitourinary disease [61], gastrointestinal disease [61], cancer [61], cardiovascular disease [61], musculoskeletal disease [61], Parkinson s disease [62], adult respiratory distress syndrome [63], diabetes mellitus [64,65], liver disease [66], AIDS [67],cataracts [68] and aging [69,70]. On the other hand, high availability of GSH in RBCs is correlated with longevity in mosquitos [71] and mice [72], and good health in elderly humans [73]. GSH is an important biomolecule in human health; therefore one would expect a tight range of GSH levels between healthy individuals (i.e. inter-individual). However, data from multiple studies observe a wide inter-individual range (0.4 to 3.0 mm (Table V-1)). This large inter-individual range (coefficient of variation (CV) 36 % for the data in Table V-1) is present regardless of age or method of detection. In contrast, intraindividual (i.e. within an individual) levels of GSH in RBCs are relatively stable over time (<10 % CV over a period of several months, 15 % over several years) [74,79]. This suggests that GSH levels are maintained at an innate level that is distinct among individuals. Here the mechanism behind the large inter-individual variation for GSH levels in human adult RBCs is examined. I hypothesized that there are heritable genotypic differences among individuals that determine their GSH, glutathione disulfide (GSSG) concentrations, and E hc status in RBCs. To test this hypothesis a classic twin study was utilized. Classic twin studies compare the similarity of a trait in mono-zygotic (MZ) and di-zygotic (DZ) twins [75,76]. For twin studies, this measure of similarity has been the

108 91 intra-class correlation (ICC) [77]. With the ICC values for MZ and DZ twins for a given trait, the proportion of the observable inter-individual variation attributed to genotypic differences (i.e. heritability) can be estimated [78]. Results Characterization of study population Adult twins (aged y) were recruited to The University of Iowa DeGowin Blood Center (Table V-2). To be included, volunteers had to meet the requirements for donation of a standard unit of whole blood. Prior to donation, a blood sample was drawn as described in Materials and Methods. Fourteen MZ twin pairs (median age 24 ± 8 y; range, y) and 5 DZ twin pairs (median age 22 ± 6 y; range, y) were included in the study. There were no statistically significant differences in the mean height, weight or age between the MZ and DZ groups (Table V-2). Zygosity was determined by DNA-based testing (See Materials and Methods.). GSH, GSSG and E hc in MZ and DZ twins The GSH and GSSG levels in freshly collected RBCs from MZ and DZ twins were determined using high performance liquid chromatography with electrochemical detection. The median intracellular concentration of GSH in RBCs was 1.4 ± 0.7 mm with a range of mm in the study population, (Figure V-1). This corresponds well with previous observations in healthy human RBCs (Table V-1; intracellular GSH = 1.4 ± 0.5 mm, range = mm). The median concentration of GSSG was 214 ± 114 µm with a range of µm, (Figure V-1). These values are also within the range

109 92 previously reported, 77 ± 38 µm [79]. The median value for E hc in RBCs was -208 ± 11 mv with a range of -182 to -232 mv, Figure V-1. There were no statistically significant differences between the mean GSH, GSSG, and E hc values for MZ and DZ twin groups. Intra-class correlation of GSH, GSSG, and E hc The ICC values for MZ twins were: 0.56 for GSH, 0.36 for GSSG, and 0.63 for E hc, while the ICC values for DZ twins were: for GSH, -0.3 for GSSG, and for E hc, (Figure V-2). The consistently stronger correlation for MZ twins over DZ twins indicates a heritable trait. The ICC values for GSH, GSSG, and E hc of MZ twins correspond well with previously observed ranges for heritable traits. For example, results from many studies on the heritability of body mass index yield a range of ICC from 0.39 to 0.91 in MZ twins with a weighted mean of 0.74 [80]. Heritability of GSH, GSSG, and E hc in RBCs Heritability was estimated using the ICC values determined for GSH, GSSG, and E hc (See Materials and Methods.). Figure V-3 depicts the estimated heritabilities: GSH, 57 %; GSSG, 51 %; and E hc, 70 %. These values for heritability indicate that a significant portion of the inter-individual variation can be explained by genotypic differences. For example, the range for inter-individual variation in GSH levels is 2.6 mm, Table V-1. With heritability of GSH = 57 %, then 1.5 mm of the 2.6 mm range in inter-individual variation is due to genetic differences. The remainder is due to differences in environmental influences, (e.g. diet and life style).

110 93 Study population compared to known heritable traits In order to determine if the subjects in our study exhibited features of an informative twin study population, the heritability of height (a well-studied heritable trait) was analyzed. The ICC values for height in our study population were MZ = 0.94 and for DZ = This results in an estimated heritability of 96 % for height in our study population (Figure V-3). An estimated heritability of 96 % corresponds well with previous reports (87-93 %) in the affluent societies of Western Europe [81]. Discussion Zygosity testing In contrast to DNA-based testing, self-reported zygosity resulted in 20 % error in our study population, much more than the 6 % expected [82]. This demonstrates the need to perform DNA-based zygosity testing, especially when the sample size is small. The GSSG/2GSH redox couple in RBCs The value of E hc has been proposed as a key parameter associated with the fundamental biology of cells and tissues. For proliferating cells E hc is approximately -240 mv (or more negative). Quiescent or differentiated cells have an E hc -200 mv; cells undergoing apoptosis will have an E hc -170 mv. If E hc is more positive (more oxidized) than -160 mv, cells are considered necrotic [53]. These values for E hc and associated biology are conserved across many forms of life: from yeast [83]; to plants (including seeds) [84,85,86,87,88]; to mammals [89,90]. The measured median value of -208 mv for E hc is consistent with RBCs being a terminally differentiated cell.

111 94 Human population compared to research animals Since errors of measurement in samples from individuals are relatively small (CV = 13 % across triplicate, independently processed samples), the large observed interindividual range (GSH, CV = 37 %; GSSG, CV = 51 %) reflects the genetic diversity in this study population. Observed intra-individual variations in GSH levels are typically much smaller in animals utilized in laboratory research (CV =13 % in mosquitos and 8 % in mice) [69,70,72]. This smaller CV can be explained by the genetic similarity desired in the various strains of research animals [91]. Support for heritability of GSH in RBCs There is support for our observations on heritability in published studies that determined GSH levels in RBCs. For instance, in a study of GSH in whole blood from two distinct Jewish populations, the distributions and ranges of GSH observed were quite different between the two populations but similar within the populations; this is consistent with a heritable trait [92]. In a twin study of toddlers ( 4 years old) Lang et al. estimated total GSH (GSH + GSSG) in RBCs to be 23 % heritable [93]. In a study of newborn infants, Küster et al. found an excellent correlation (R 2 = 0.62) between the GSH concentration in RBCs from cord blood of newborns with the GSH concentration in the RBCs from their mothers; again, suggesting heritability of GSH in RBCs [94]. Limitations The estimates of heritability found here may not be reflected in all population groups. In addition, our estimates have been made on a relatively small sample size.

112 95 However, estimates of heritability of two well-known heritable traits in this study match exceptionally well published estimates for these traits: height (estimated heritability of 96 % in this study compared to the published range of % [81] and BMI (69 %, published range % [80]). These two results engender considerable confidence in our estimates on the heritability of GSH, GSSG, and E hc in RBCs. Implications The heritability of GSH, GSSG and E hc levels in RBCs could evolve the understanding of many issues in human health. The observed heritabilities could in part explain: Why individuals are affected differently by aging, disease and other daily insults to our bodies; Mechanisms of drug efficacy and adverse reactions on an individual basis (pharmacogenomics); and Individual outcomes upon exposure to toxins and xenobiotics (toxicogenomics). With this new understanding, and by identifying the specific genes responsible for individual GSH phenotypes, novel personalized therapeutic strategies to: effectively combat disease, increase longevity, and reduce xenobiotic toxicity, can be designed. I hypothesize that the variance between the innate (i.e. genotypically determined) GSH level and the phenotypic GSH level can serve as a biomarker for health and risk for disease.

113 96 Future Directions With the discovery of a genetic component in the regulation of GSH, GSSG, E hc and tgsh levels in human RBCs, one can ask the question which gene or genes are involved in the regulation of these factors. This could lead to the development of new genetic screens to identify the expected values of these factors. Also one could triage individuals for care based on their genetic GSH profiles. The study as performed here represents only a very small percentage of population. Therefore, this study should be performed in different population groups and ethnicities to determine the generality of these results (very important in light of Figure V-4). Also the genetic influence could be investigated in populations where diets are more calorie and protein restricted. In these populations the environment might play a larger role compared to genetics. GSH in RBCs is an easily accessible biological sample. Future studies should determine if the genetic component seen in RBC GSH is present in other tissues as well e.g. liver, lung, and kidney. Knowing this relationship can establish RBC GSH as a marker of whole organism GSH and be used to predict reactions in specific tissues without taking direct biopsy samples from them.

114 97 Table V-1: Observed ranges of glutathione in human erythrocytes GSH a GSH / mm b Population Method c Ref Mean SD Range Units Mean SD Range mg / cc prbcs 1.8 e Healthy adults I 95 d (n=117) mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs mg / 100 cc prbcs µmole/l whole blood µmole/l whole blood 1.8 e Non-Ashkenazi subjects I e Ashkenazi subjects I e e e e e e e e e g g µm/ g Hb 1.2 h µm/ g Hb 0.8 h µmole/l RBCs meq / L RBCs meq / L RBCs mg / 100 cc prbcs pmole / 10 7 cells mmol/l RBCs µmole/l RBCs µmole/l RBCs e i G6PD normal patients (n=25) Sensitive Iranian males(n=35) Non-sensitive Iranian males(n=257) Sensitive Iranian females(n=12) Non-sensitive Iranian females(n=132) Random samples (n=15) Random samples (n=15) Normal blood donors (n=85) Normal infants (n=253) Healthy adults (n=47) Elderly subjects (n=64) Young (31 ±10 years, n=33) Aged (69±11 years, n=28) Healthy adults (18-73 years, n=107 men, 94 woman) Healthy men (n=484) Healthy woman (n=231) Healthy adults (n=30) Unknown population Healthy adults (n=10) Healthy individuals aged years (n=25) Healthy individuals aged 2-11 years (n=28) I 97 I 98 I 98 I 98 I 98 II 99 III 99 III 99 III 99 IV 100 IV 100 I 101 I 101 V 102 VI 103 VI 103 I 104 VII 105 VI 106 VI 107 VI 107

115 98 Table V-1 : Continued GSH a GSH / mm b Population Method c Ref Mean SD Range Units Mean SD Range µmole/l RBCs µmole/l RBCs µmole/l RBCs Overall mean, SD and range 1.4 j a Values as reported in the units provided. Healthy individuals aged years (n=23) Healthy individuals aged years (n=40) Healthy individuals aged years (n=60) VI 107 VI 107 VI 107 b Reported values converted to an estimated intracellular concentration (mm). c Legend of methods used to measure GSH: I: Beuttler [108]; II: Chlorpromazine displacement; III: Nitroprusside; IV: Capillary electrophoresis; V: HPLC with post column derivatization and fluorometric detection; VI: DTNB; VII: Isotope dilution liquid chromatography-tandem mass spectrometry. d prbcs = packed red blood cells. e Assuming packed RBCs hematocrit (Hct) is 80 %. f Assuming an average hematocrit (Hct) of 40 %. h Assuming average hemoglobin (Hb) of 140 g/l in whole blood and an Hct of 40%. i Assuming a mean corpuscular volume (MCV) of 90 fl j Mean is not weighted.

116 99 Table V-2: Characterization of study population Trait Monozygotic (MZ) Dizygotic (DZ) p value a Age / y 24 ± 8 b 22 ± Weight / kg 64 ± ± Height / m 1.68 ± ± Female pairs Male pairs Male/female pairs a Dizygotic versus monozygotic b Median ± standard deviation

117 100 Figure V-1: The distributions of GSH, GSSG, and E hc are very similar between MZ and DZ twins. The intracellular concentration of GSH for MZ and DZ twins combined is: median = 1.4 mm; mean = 1.5 mm; standard deviation = 0.7 mm; and range = mm GSSG is: median = 214 µm; mean = 224 µm; standard deviation = 114 µm; and range = µm. The half-cell reduction potential is: median = -208 mv; mean = -204 mv; standard deviation = 11 mv; and range = -182 to -232 mv. Bins were equally distributed between the highest and lowest values of the whole study population (MZ + DZ). The difference was divided by 5 to assign the size of the bins (e.g. GSH bin 0.6 mm contains those individuals with a GSH concentration between 0.6 mm and 1.2 mm). There is no statistically significant difference (Students t-test) between the means of the two groups for all three analytes. This indicates that the only difference between the MZ and DZ groups is the genetic similarity of pairs within the study population.

118 101 Figure V-2: The intra-class correlation between MZ twins is greater than between DZ for GSH, GSSG, and E hc. ( ) are the measured values for GSH, GSSG or E hc. (-) is the mean value for a twin pair. In our population, the ICC values for MZ twins were: 0.56 for GSH, 0.36 for GSSG, and 0.63 for E hc. The ICC values for DZ twins were: for GSH, -0.3 for GSSG, and for E hc. The consistently stronger correlation for MZ twins over DZ twins is consistent with a heritable trait. Measured values are the median of independently processed samples (n = 3)

119 Heritability / % GSH GSSG Ehc Height Figure V-3: The concentrations of GSH, GSSG, and E hc within RBCs behave as heritable traits. Heritability was estimated using the method of Newman, Freeman, and Holzinger [75]. The estimated heritability value for GSH is 57 %, for GSSG 51 % and E hc 70 %. The heritability for height (96 %) was used as a marker for an informative twin study to compare our study population to other published reports on this known heritable trait (87-93 %) [81].

120 Number of subjects Random population n = 280 Mean = 71 ± 17 Ashkenzic jews n = 310 Mean = 72 ± 12 GSH/ mg 100cc RBCs Figure V-4: Distribution of GSH is between two ethnic groups due to prevalence of polymorphisms. There are certain individuals in a random population with a specific polymorphism resulting in very low GSH levels in their RBCs cc ( 45 mg per 100 cc of RBCs). This polymorphism is not present in the Ashkenazi Jew population. Extracted from [92].

121 104 CHAPTER VI : STRATEGIES FOR MODELING THE INTRACELLULAR REDOX BUFFER UPON EXPOSURE TO OXIDATIVE STRESS

122 105 Introduction Mathematical simulations are a great tool in the field of redox biology and other areas. People rely on models daily for the prediction of the weather and other applications. The goal of this study was to develop simulators that could be used to predict the outcomes of experiments; as well as predict toxicities based on oxidation of the GSSG/2GSH redox couple. Secondly, the simulators were designed to identify key parameters that play a major role in processes like oxidation of the redox buffer, removal of oxidative species, and the maintenance steady-state levels of ROS levels. In this chapter the construction of the models and the theories behind them are described. For this project three different programs were used to construct the models and simulate various aspects of metabolism, antioxidant defenses, and oxidant generation within cells (Figure VI-1). Theory The first model is designed to give insight into the lifespan and diffusion distance of O - 2 and H 2 O 2. This model is based on the RBC, but can be translated to any cell. The simulator was custom-built in LabVIEW with parameters from the literature to simulate individual particles and measure their travel distance and lifespan.

123 106 Random collision model Single particle tracking: This simulator is based on the collision theory. At the beginning of each calculation one molecule representing O - 2 is formed and begins its motion (Figure VI-6). The simulator calculates the distance traveled and volume traversed (Figure VI-3) after a user-adjustable predetermined time has expired (iteration time). At this point the total number of encounters is calculated and this number is used to calculate the probability of an encounter with a specific protein and the possibility of a reaction occurring. If no reaction occurs, the molecule moves along, and the cycle repeats itself; if there is a reaction, the molecule is removed, and the simulation is stopped. Encounter statistics are based on total reaction volume occupied by each specific protein. Total volume occupied is calculated using the specific volume for each protein. This value is multiplied by the total number of molecules of each specific protein to give the occupied volume per RBC. The total volume occupied is divided by the RBC volume (90 fl) to calculate the fraction the protein occupies. This number is used to calculate the probability of encountering each protein (Table VI-1). Reaction probability is based on known rate constants for each reaction simulated (Figure VI-2). The diffusion limit (2 x M -1 s -1 ) represents the situation where every encounter results in a reaction. Using this limit a probability of reaction can be calculated for each individual reaction. Cartesian coordinates (x,y,z) are generated using a Gaussian signal generator, which has an offset of 1 nm and a standard deviation of 1. This provides a random speed and travel distance per time step (Figure VI-3). With this speed a total swept-volume can be

124 107 calculated (Figure VI-3). Dividing this number by the volume of our target species (O 2 - = nm 3, H 2 O 2 = 0.2 nm 3 ) gives the number of encounters for each time step 5. Multiple particle tracking and flux: To model the system as a whole the simulator designed for tracking a single particle was modified. At the start of the simulation the desired numbers of initial particles are placed randomly in the space representing the RBC. For this simulator, this space represents the inner layer or cytosol of the RBC. The volume of the RBC is known to be on average 90 fl, this volume is approximated by a cube with equal sides of 15 µm. The sides of the cube are the boundaries of the simulation. After the initial particles are created they are allowed to move randomly through the space in a manner identical to the single particle-tracking model. For each movement a certain distance has been traveled. To simplify the reaction prediction and encounter frequencies the iteration time has been fixed to 0.2 nanoseconds (ns). In this period of time a particle will on average encounter one other species. To speed up the simulation a time warp function has been added to increase the iteration time. For a time warp of 1X, each iteration is 0.2 ns, encounters is 1, and the traveled distance is 0.2 nm. When the time warp is increased, the factors are multiplied by the value as specified. This allows time to be sped up or slowed down. To account for the creating of new O - 2 molecules from hemoglobin autoxidation, a new flux element was created and introduced to the simulator. Since the rate of 5 accessed on

125 108 hemoglobin autoxidation is known, the flux of new molecules can be calculated. This process was estimated to lead to the formation of an O - 2 every 2 milliseconds. To model hemoglobin autoxidation, a random number was used that would generate a number within 0.1 standard deviations of 2. A timer was created to keep track of the amount of time that had expired in the simulator, when the time-passed matches the generation time a new O - 2 would be created randomly in the RBC volume. When simulating at higher time warps (iterations are more than 2 ms), creating one O - 2 at a time is insufficient for the amount of time that has passed in the simulator. To account for this phenomenon the program was altered. The simulation determines if the elapsed time has exceeded the generation interval; if this is true, a second formation time will be randomly generated as described above. This new value is added to the previous creation time and compared against the passed time. If the new formation time has exceeded the past time the simulation is allowed to continue; if this has not occurred new formation times are added to the old and compared to the passed time until it does exceed. The number of O - 2 molecules generated is proportional to the number of times a new formation time was added. The flux is modifiable in a manner similar to the time warp feature. The normal autoxidation rate (2 ms/molecule) is 1X and can be increased as desired. Increasing the modifier decreases the amount of time it takes to generate an O - 2.This results in a higher flux of molecules into the system.

126 109 Kinetic equation Model To simulate more complex models, collision theory is computationally intensive and therefore impractical for this project. To overcome this problem, kinetic equation models were constructed. The first generation model is built on the works of Adimora et al. [109], Antunes et al. [110], and Makino et al. [111]. These papers describe the kinetic equations that govern the uptake of H 2 O 2 by intact cells, as well as its removal by the three main H 2 O 2 removing enzymes (Figure VI-4). These equations, together with unique quantitative data on intracellular enzyme concentration, as measured in the lab, are the basis of our first generation model (Figure VI-5). The first generation model includes the removal of H 2 O 2 by GPx and catalase. The Prdx system will be implemented in the future. CellDesigner Inter-compartmental diffusion The diffusion of compounds between compartments is described by Fick s first law of diffusion (Equation 8). The law states that the rate of diffusion is depended on four variables: the H 2 O 2 concentration gradient between the extracellular space and the cytosol (Δ[H 2 O 2 ]), the permeability coefficient (P), the surface area of a single cell (A), and the number of cells (N cell ). The permeability coefficient is a parameter that describes the ease of a compound to penetrate a membrane. This parameter is described in centimeter per second. ([H 2 O 2 ])/ t = ( [H 2 O 2 ] P A) N cell A (8) Since the rate of diffusion between the media and cytosol is known, it is possible to set limits to the permeability coefficient (for H 2 O 2 through biological membrane this

127 110 value is in the range of 1 x 10-2 and 1 x 10-5 cm s -1 ). The surface area of the cell is a measurable using a Coulter Counter. The concentration outside and inside the cell are known or calculated during the course of the simulation. Removal of H 2 O 2 The kinetics of the removal of H 2 O 2 by GPx and catalase has been extensively described. The current model incorporates measurements performed by our lab that determined the absolute initial quantity of enzyme present ([GPx], [catalase], [compound I]) for each cell line simulated. This number is entered into a differential equation describing the enzyme kinetics (Equations 9, 10.1, 10.2 and Table VI-3). [GPx] / (1 / (V perox [H 2 O 2 ]) + 1 / (V GSH [GSH]) N cells (9) V cat [catalase] [H 2 O 2 ] N cells (10.1) V comp [compound I] [H 2 O 2 ] N cells (10.2) Regeneration of GSH and NADPH Currently, the rate of NADPH regeneration is based on kinetics as described in Adimora et al. [109]. This publication utilizes the gradient between initial [NADPi] and current NADP + [NADP] to establish a rate of NADPH formation. Two rate terms define the maximum formation rate (Equation 11, Table VI-4). V recycle ([NADP] - [NADPi]) / (k recycle + [NADP]) (11) The regeneration of GSH is defined by the kinetics of the enzyme glutathione disulfide reductase (GR). This enzyme reduces the GSSG that is formed by GPx back to GSH and uses NADPH in the process. The rate equation is provided by Makino et al. [112] (Equation 12, Table VI-5).

128 111 (V gr N cells (k 0 [GSSG] [NADPH] + [GSSG] 2 [NADPH) / (k 1 [GSSG] + k 2 [NADPH] + k 3 [GSSG] 2 + k 4 [GSSG] NAPDH + ([GSSG] 2 [NADPH]) (12) The redox buffer To determine the state of the glutathione redox buffer, the following equation is used (Equation 13) [53]. The half-cell reduction potential that is calculated with this equation is represented as a separate species in COPASI. It is determined for every time point simulated. The ph is assumed as 7.4 which is why the E 0 = E hc (mv) = -264 (59.1 / 2) log ([GSH] 2 / [GSSG]) (13) The status of the redox buffer correlates with the biological state of the cell. A glutathione redox state more positive than -160 mv is representative of a cell that is undergoing apoptosis or necrosis. These values were utilized as the marker of toxicity for our H 2 O 2 exposures in this model. Determination of Area Since the model needs to be adjusted for various cell densities, and cell death during the simulation; terms for area and volume are needed in the model that can account for the change in compartment volume and surface area (i.e. loss or gain of cell number). The volume of the simulated cytosol volume will increase linearly with increasing cell number. Cell area however does not increase in a linear fashion; therefore, Equation 14 is used as the substitute for A in Equation 8. This equation determines the area (A) based on the individual cell volume (V), the individual cells radius (r). To model peroxisomal diffusion Equation 15 is used. This equation is based on Equation

129 112 14; however it uses the volume and radius of a single peroxisome to determine area. Also, there could be multiple peroxisomes per cell; therefore, an additional multiplication term (N p ) is added that could be utilized if this number is known. A = (3 V) / r (14) A = (3 V) / r N p (15) Model Parameters The first generation model is constructed for 6 cancer cell lines that are commonly used in the lab. Table VI-6 and Table VI-7 describe the values for each cell line in detail. Quantitative data for catalase has not been entered into the current model. All cells have an estimated 100 µm of catalase in their peroxisomes. Results Random collision model Single particle tracking A basic simulator has been created using LabVIEW. The simulator is capable of representing single molecules (Figure VI-6) and calculating their predicted life time and diffusion distance (Table VI-2). To determine the accuracy of the simulator, a series of calculations was started that varied the duration of each time step (Δt). This parameter determines how far a target molecule can travel and how many encounters it has per time step. At high t values great theoretical distances are traveled during a time step, which results in a large lifetime and travel distance. However, this is a gross overestimation of the calculated values due

130 113 to the high t of the simulation. The calculation assumes that the molecule has existed for the entire time step whereas in reality, it was removed early in the step. The model does not take this phenomenon into account. Therefore it is advisable to minimize the duration of a time step leading to a more accurate result. From a series of calculations (Figure VI-7) it was determined that a t below 1/10 of a microsecond provides accurate results. Care has to be taken when choosing a t since large time-steps result in overestimated values, but short time-steps require many more cycles and therefore it takes longer for a calculation to finish without gaining more accuracy. From the simulated travel distances I find that O - 2 is capable of reaching the plasma membrane if generated in 0.06 % of the RBC volume. Current literature suggests that O - 2 is unable to effectively cross the membrane boundary [114]. In contrast, H 2 O 2 has a far greater potential for reaching the plasma membrane (1.5 % of the volume if generated from O - 2 and 0.4 % of the volume if generated as H 2 O 2 ) (Figure VI-8). This is due to its greater lifespan. The current simulator does not take into account the actual event of escaping since the mechanism for this is not fully understood at this point; however, the first step in escaping is reaching the barrier of the plasma membrane. Multiple particle tracking and flux The first experiments performed with the model excluded the particle tracking ability and focused on the number of ROS molecules present over time at different fluxes. As seen in (Figure VI-9); an increase in the flux of O - 2 changes the number of molecules present at any given time. When the tracking of individual particles is activated, a true steady-state can be observed. An example of a continuous simulation is

131 114 shown in Figure VI-10. The picture displays the steady-state of O - 2 and H 2 O 2 at an increased flux of hemoglobin autoxidation. It is also possible to simulate the removal of ROS species by an RBC that is given a bolus addition. In Figure VI-11, a virtual RBC is exposed to 1000 molecules (18 nm) of O - 2. To focus solely on the removal ability, generation of new O - 2 molecules was disabled. In this simulation all the O - 2 molecules were removed 0.12 ms after addition of the bolus. A maximum amount of 340 molecules of H 2 O 2 were present 0.03 ms after the start of the simulation. All ROS molecules were removed ms into the simulation. To confirm the results as generated by the random collision simulator, the same exact experiment was performed in the CellDesigner kinetic simulator. As seen in (Figure VI-12), the results are nearly identical. There is a lower accumulation of H 2 O 2 in this simulator; this is because the Copasi model has more routes for peroxide removal then random collision model, therefore there should be less accumulation. Assuming that the normal flux of O - 2 due to hemoglobin autoxidation is 50 molecules per second per cell; a maximum of 1 molecule per cell could escape from an individual cell. The average human male possesses 2.5 x RBCs in 5 liters of whole blood. When utilizing the flux of 1 molecule per second per RBC and converting this to the total contribution of RBCs to H 2 O 2 steady-state levels in plasma; RBCs produce, on average, 83 pm H 2 O 2 in the plasma per second. Kinetic equation model To determine the sensitivity of six cancer cell lines to H 2 O 2, a simulation was performed where cells at a constant density were subjected to different fluxes or bolus

132 115 additions of H 2 O 2. From these experiments the extracellular concentration of H 2 O 2 (Figure VI-13) and the status of the glutathione redox buffer (Figure VI-14) are plotted over time. From these data, a first prediction can be made as to when the various cell lines will start dying (Figure VI-15). Discussion Random collision model Care must be taken when drawing conclusion for travel distances and life span from simulations with a large time warp. At large time warps the observed species may have been generated and removed in an iteration; however, no distinction is made as to, where in the span of an iteration, the molecule was removed. The model assumes that the molecule was removed at the end of the iteration, therefore it overestimated the life span and distance it traveled. The most accurate representations of the system are those made with low time warps; this however leads to a great increase in the simulation time. The random collision model shows that under normal circumstances it is difficult to define a steady-state; most of the time (99%) there are no molecules of either H 2 O 2 or O - 2 present. The behavior of O - 2 is more defined by spikes of presence, this phenomenon makes the lifetime of these molecules very important since this is the window were they can damage the cell. When the flux is increased due to a disease state or presence of a xenobiotic producing ROS, the lifetime of the molecules becomes less important and the steady-state more important.

133 116 Kinetic equation model From the various simulations, it is apparent that there is a significant difference between the methods of exposure (bolus vs. constant flux), and the dose associated with this method. As seen in Figure VI-13, most cells investigated consume the bolus of H 2 O 2 within minutes. This results in a small area under the curve (AUC). The AUC is a measure of the time and dose of exposure that is commonly used in pharmacology but has also been used to relate to toxicity [12]. In the constant flux method of exposure, the levels of peroxide that are seen by the cell are lower (no more than single digit µm); however, these low concentrations are sustained for long periods of time leading to a greater AUC then a bolus addition. There are also great differences in the state of the redox buffer; our marker for toxicity. In the constant flux method, the cell slowly oxidizes its redox buffer to remove the peroxide that enters the cell. The rate at which this oxidation occurs is cell type dependent; this leads to different times for when the state of the redox buffer is greater than -160 mv. In contrast to the bolus addition were some cells reach -160 mv, and others do not. Also, the duration that the cells remain above -160 mv is important. Since cells have GR, they can reduce the GSSG formed from the removal of H 2 O 2. This leads to more reduced intracellular environment. Since cells remove H 2 O 2 in the extracellular space within minutes [123], the cell can try and restore the redox buffer before necrosis or apoptosis occurs. This phenomenon has to be investigated further. A good example of this is the MIA PaCa-2 cell line. In Figure VI-14 the bolus addition results in a short period were the redox buffer is > -160 mv. The cell slowly restores the buffer over time. In a constant addition, the cell remains < -160 mv for a

134 117 long time. This difference between these two methods could affect the cell killing effectiveness of H 2 O 2. Conclusion In conclusion, this project needs a significant amount of work, especially in the verification of simulator performance with experimental results. Also, several key parameters and pathways have not been determined for each cell line. Therefore, the simulations could be inaccurate by design. However there is great promise in this approach since the program around the simulations works well, and can be adjusted to monitor other endpoints. The motion analyzer is currently in a prototype stage of development. The model performs in a manner similar to solver simulators like Copasi, which uses differential equations to solve for the metabolite concentration over time. This is with regards to removal kinetics of O - 2 and H 2 O 2.

135 118 Table VI-1: Species Encounter probability calculations for each protein species. Specific MW Volume volume (g/mole) occupied (cm 3 /g) (10 3 ) (10 6 nm 3 /RBC) Molecules /RBC (10 5 ) Fraction volume occupied (10-4 ) Reaction probability (10-2 %) SOD b Hemoglobin c GPx d Prdx a Catalase d a Estimated based on molecular weight. b From [113]. c From [114]. d From [115]. Table VI-2: Superoxide and hydrogen peroxide have very short lifespan and diffusion distance within RBCs. - O 2 Lifespan b (µs) O - 2 diffusion distance (nm) H 2 O 2 lifespan (µs) H 2 O 2 diffusion distance (nm) Mean a, c Median Mode Maximum Stdev a t during calculations was 1 nanosecond. b Lifespan = time from creation of molecule to reaction to something different. c Result of 1600 simulations.

136 119 Table VI-3: Description of the constants used in equations 9 and 10. Reaction Value constant (M -1 s -1 Description ) V perox 2 x 10 7 a Reaction velocity of Human GPx 1 with H 2 O 2 V GSH 4000 a Reaction velocity of Human GPx 1 with GSH V cat 1.6 x 10 7 b Reaction velocity of Human catalase with H 2 O 2 V comp 2.7 x 10 7 b Reaction velocity of Human catalase compound I with H 2 O 2 From [116]. b From [117]. Table VI-4: Description of the constants used in equation 11. Reaction Value Description constant V recycle 3.75 x 10 4 M s -1 Maximum reaction rate constant for NADPH recycling k recycle 5.7 x 10 5 M Binding constant for NADP + Table VI-5: Description of the constants used in equation 12. Reaction constant Value Description k M Binding constant k M 2 Binding constant k M 2 Binding constant k M Binding constant k M Binding constant V gr Variable (M -1 s -1 Cell specific activity for glutathione reductase ) See Table VI-6 for cell specific numbers

137 120 Table VI-6: Parameters for the glutathione system. Cells GPx (µm) a Vgr / (10-13 M s-1) GSH(mM) GSSG b (µm) HL c 4.8 [118] 150 U [119] 290 MB c 1.2 [120] 9 MCF c 2.3 [120] 33 MIA PaCa [121] 14 PC c 4.8 [122] 150 a Concentration determined by TJ van t Erve in unpublished work. b Concentration that assumes E hc = -240 mv as calculated using equation 13. c These values have not been measured; therefore, a default value has been assigned to them. Table VI-7: Parameters for H 2 O 2 diffusion. Cells P a plasma / (10-4 cm s -1 ) P b peroxisome / (10-3 cm s -1 ) HL U MB MCF MIA PaCa PC a Permeability coefficient as determined by fitting H 2 O 2 consumption rate from [123]. b Assumed from the literature [111].

138 121 Xenobiotic + O 2 (1) H 2 O 2 Extracellular space (3) Prdx(SH) 2 (4) Prdx(S 2 ) Trx(SH) 2 Trx(S 2 ) TrdxR H 2 O 2 (2) (5) GPx GSSG 2 GSH GR Xenobiotic + (6) O 2 Peroxisome (7) 2 H 2 O 2 Catalase NADPH NADP + NADPH Cytoplasm Figure VI-1: The removal of intracellular and extracellular H 2 O 2 by cells. (1) Oxidation of xenobiotics in the presence of molecular oxygen forming H 2 O 2 extracellulary. (2) Formation of H 2 O 2 by oxidation of a xenobiotics intracellularly. (3) Diffusion of H 2 O 2 through the plasma membrane into the cytoplasm. (4) Removal of H 2 O 2 by the peroxiredoxin/thioredoxin/thioredoxin reductase system. (5) Removal of H 2 O 2 by glutathione peroxidase and recycling of glutathione. (6) Diffusion of H 2 O 2 into the peroxisome. (7) Removal of H 2 O 2 by peroxisomal catalase.

139 122 Figure VI-2: Flow diagram of a single simulation calculation.

140 123 Figure VI-3: Methods depicting spatial calculation methods. A: Method to calculate the traveled distance and speed of the target species. B: Method to calculate the total swept volume and number of encounters in that volume for the target species. D = effective traveled distance t = iteration time U = traveled distance per iteration o = radius of molecule V = swept volume

141 124 Figure VI-4: Reactions considered in the random collision model. Second-order rate constants are provided below the reaction arrow.

142 125 Peroxisome Cytosol Figure VI-5: General biochemical network as created in CellDesigner for removal of H 2 O 2 by cells. The network describes the removal of H 2 O 2 by GPx and catalase, as well as the recycling of GSH by GR, and the recycling of NADPH by G6PD. The diffusion of H 2 O 2 between compartments is also represented.

143 126 Figure VI-6: Example of simulated motion for an O 2 - and H 2 O 2 molecule in an RBC. The random motion of a single O 2 - molecule that is dismuted into H 2 O 2 and lives until it reacts with a peroxide-removing enzyme. The molecule is created at the green dot, follows the red path and is dismuted at the dark blue dot. It then travels as H 2 O 2 along the blue path, and is removed at the purple dot. The effective traveled distance is portrayed by the yellow lines in 3D space.

144 Lifespan / µs E+18 1.E+16 1.E+14 1.E+12 1.E+10 1.E+08 1.E+06 1.E+04 1.E+02 1.E Iteration duration / s 1E-08 1E-10 Figure VI-7: Iteration steps for collision theory simulator needs to 1 x 10 7 seconds or bellow for accurate results. Due to the design of the simulator at iterations longer than 1/10 of a microsecond, the outcome of the calculations is overestimated. Therefore, a t below 1/10 of a microsecond must be chosen to generate accurate results. The lifespan of superoxide when the duration of an iteration is 10-8 s is 16 μs in this simulation

145 128 Figure VI-8: Superoxide or H 2 O 2 generated inside an RBC have next to no change to reach the membrane and potentially escape. The figure displays the volumes in which an O 2 - or H 2 O 2 can be generated and reach the plasma membrane before it reacts. 1.5 % of O 2 - molecules generated are capable of reaching the plasma membrane as H 2 O 2. These results are based on 1000 simulations.

146 Number of molecules Figure VI-9: Time / milliseconds A presentation of the transient levels of simulated species in a RBC. The number of molecules (red = O 2 -, blue = H 2 O 2 ) that are present at a given iteration versus expired time. The rate of hemoglobin autoxidation was simulated at the normal rate ( 2 ms/molecule) (left) and 200 X normal rate ( 0.01 ms/molecule) (right). O 2 - was present 0.7% of the time in the normal flux and 85% in the increased flux. Total simulation time was 50 ms. Total number of O 2 - generated was 24 in the normal and 4543 in the increased flux. Resolution for this experiment was 2 µs.

147 130 Figure VI-10: 2D and 3D pictures of a single instance of a simulator with particle tracking enabled. This figure demonstrates the steady-state of ROS within and RBC that has 1000X the O 2 - production of a normal cell. This picture was taken 0.43 ms into the simulation. At this instant there are 7 O 2 - molecules (red dots) and 8 H 2 O 2 molecules present.

148 131 Superoxide Hydrogen peroxide Time / s Time / s A B C D A B C D A B C D Figure VI-11: Time course of removal of O 2 - inside the RBC. The RBC was exposed to 1000 molecules (18 nm; intracellular concentration) of O 2 - at the start of the simulation. The simulation resolution was set at ms/iteration. Only H 2 O 2 was allowed to escape from the system; O 2 - was confined to the predefined volume. Simulation was ended at 0.3 ms, at this time all initial molecules have disappeared. The 3D picture display the particle distribution at various time points as indicated on the time course graphs. Red dots represent O 2 - and blue dots represent H 2 O 2.

149 Amount / molecules 132 Superoxide Hydrogen peroxide Time / h Figure VI-12: Traditional kinetic simulation performed by Copasi 4.5. The differential equations were generated using cell designer. The graph shows the removal of a bolus of 18 nm (1000 molecules) O 2 - (red line). The removal of O 2 - leads to the formation of H 2 O 2 (blue line).

150 Extracellular H2O2 / M Extracellular H2O2 / M 133 MB231 HL60 MCF7 MiaPaCa 2 U937 PC3 1.2E-06 Flux = 3 nm s E E E E E E Time / s 1.2E-05 MB231 HL-60 MCF7 MiaPaCa 2 U937 PC3 1.0E E-06 Bolus = 10 µm 6.0E E E E Time / s Figure VI-13: Simulation of the concentration of extracellular H 2 O 2 after addition to a cell suspension plotted over time Top: Simulation of 5 million cell ml -1 exposed to a constant flux of 3 nm s -1 H 2 O 2 Bottom: Simulation of 5 million cell ml -1 exposed to a bolus of 10 µm H 2 O 2. Both figures plot the extracellular concentration of H 2 O 2, which is assumed to be uniformly distributed throughout the extracellular environment. Note the differences in the time scales.

151 Ehc / mv Ehc / mv 134 MB231 MCF7 U937 HL60 MiaPaCa PC3 MB231 HL-60 MCF7 MiaPaCa 2 U937 PC Flux = 3 nm s Bolus = 10 µm Time / s Thousands Time / s Thousands Figure VI-14: Simulation of the intracellular glutathione redox couple upon addition of H 2 O 2 to a cell suspension. Left: Simulation of 5 million cell ml -1 exposed to a steady flux of 3 nm s -1 H 2 O 2 Right: Simulation of 5 million cell ml -1 exposed to a bolus of 10 µm H 2 O 2. Both show the state of the intracellular redox buffer (GSSG/2GSH). The state of the buffer is calculated for each time point by utilizing the concentration of GSH and GSSG and determining E hc via the Nernst equation.

152 Time to predicted cell death (min) MB231 HL60 MCF7 MiaPaCa U937 PC H2O2 Flux (nm/s) Figure VI-15: Simulation of the predicted time to death for six cancer cell lines exposed to fluxes of H 2 O 2. The models of six cancer cell lines were subject to varying fluxes of H 2 O 2 (1 to 30 nm per s). The time point were the redox buffer of the cell exceeds -160 mv is taken as the predicted time of death for each individual cell line.

153 CHAPTER VII : SUMMARY 136

154 137 Chapter III In this chapter, dosing in cell culture experiments was examined. Experiments done with the model toxin 1,4-BQ (an oxidation product of benzene) showed that dose is best expressed on a mole per cell basis to accurately describe the exposure. Using this quantitative approach I believe the toxicity of covalent binding toxins should be reported. Dosing on a mole per cell basis also allows for quantitative comparisons with intracellular targets. This concept was demonstrated with the depletion of intracellular GSH levels and direct comparison with the moles of 1,4-BQ exposed. The second discovery in this chapter involves susceptibility of cell to 1,4-BQ exposure. By taking into account the physical properties of individual cell types i.e. cell size we were able to determine that cells with a large intracellular volume are more resistant to 1,4-BQ induced decrease in clonogenic cell survival. This result supports our interpretation of covalent biding and target theory in cellular toxicity. Chapter IV and V In these chapters, some of the factors of the RBC storage lesion are explored looking specifically at the presence of heritable genetic regulation. In this study, the individual variation of GSH, GSSG, E hc, tgsh, ATP, and hemolysis was investigated and the heritable component for each factor determined in a study population. All of these species and associated properties were found to a have a heritable component to their regulation. This suggests that there are heritable differences between individuals that affect the quality of their RBCs under storage condition.

155 138 These findings could lead to individualized approaches with regards to the storing of blood. This is comparable to taking into account physical differences in individual cells to explain their susceptibility to 1,4-BQ toxicity. I believe that focusing on specific patient or donor differences in pathologies like the RBC storage lesion will greatly improve transfusion medicine and reduce the negative side-effects from transfusion of poorly storing old units. This phenomenon of heritability is also observed in fresh RBCs for GSH, GSSG, E hc and tgsh. This suggests that there are healthy people around with a large range of the levels of these factors in their RBCs. These individual differences can be used to direct personalized therapies, and provide targeted interventions to ameliorate potential toxicities from such compounds as 1,4-BQ based on GSH genetic profiles. Chapter VI In this chapter, an attempt was made to build quantitative kinetic models that can simulate the complex world of redox biology. These models could ultimately provide great insight into short lived ROS and other fast reactions. With kinetic models, the involvement of ROS in pathologies like the RBC storage lesion can be studied with great precision, leading to predictions and hypotheses of which processes are the most important in the disease. The models can then be used to design better experiments. The model cannot accurately accomplish these goals without taking into account the quantitative nature of all species involved, and the individuality of the system. The models build here incorporate flexibility and individuality to better study problems like dosing of toxins leading to a biological response. Models can lead to advances in

156 139 toxicology where can reduce the need for animal experiments and costly clinical trials to determine drug safety.

157 140 APPENDIX A PHYSICAL AND CHEMICAL PARAMETERS OF RED BLOOD CELLS AND PLASMA

158 141 ADULT RED BLOOD CELL (RBC) PHYSICAL PARAMETERS Parameter Value Units Ref Volume fl fl (male) fl (female) [22] Diameter ± ± 0.28 µm µm µm (disk thickness) [124] [23] [23] Surface area ± 13.5 µm 2 [23] Protein concentration 330 g L -1 (whole blood) [125] H 2 O 2 Permeability (plasma membrane) 6 x10-4 cm s -1 [110] ANTIOXIDANT ENZYMES Enzyme Activity Units Ref Superoxide Dismutase CuZn-SOD 2350 U g Hb -1 [126] Catalase IU g Hb -1 [126] Glutathione Peroxidases GPx ± ± ± 1.7 IU g Hb -1 µmole min -1 g Hb -1 U g Hb -1 nmole NADPH min -1 mg Hb -1 [126] [127] [128] [129] Thioredoxin 1.29 ± 0.15 nmole mg protein -1 [130] Peroxiredoxin Prdx [130] [126] [126] µm million copies cell -1 [130] Glutathione-S-Transferase 6.81 IU g Hb ± 2.1 µmole min -1 g Hb -1 [127] Glutathione Reductase 7.79 IU g Hb ± 1.2 µmole min -1 g Hb -1 [127] SMALL MOLECULE ANTIOXIDANTS Molecule Concentration Units Ref Glutathione ± ± 0.15 µm g Hb -1 µm mm [126] [131] [132] Glutathione Disulfide 60 ± 34 µm [131] Ascorbic acid 55 ± 10 µm [132] α-tocopherol 3.5 ± 0.5 µm [132] Protein sulfhydryl s 44.9 ± 1.8 nmole mg protein -1 [131] Protein glutathione mixed disulfides 0.49 ± 0.10 nmole mg protein -1 [131]

159 142 OTHER MOLECULES Molecule Concentration Units Ref Hemoglobin pg cell -1 [22] NADPH 27 ± 4 µm [132] NADH 35 ± 6 µm [132] ADULT PLASMA SMALL MOLECULE ANTIOXIDANTS Molecule Concentration Units Ref Glutathione 1.92 ± ± 0.87 µm µg/µl [133] [134] Glutathione Disulfide 0.05 ± ± 0.05 µm µg/µl [133] [134] Cysteine 8.6 ± ± 3.6 µm µg/µl [133] [134] Cystine 47 ± ± 2.25 µm µg/µl [133] [134] Ascorbic acid 4.29 ± 0.36 µg/µl [134] N-acetyl cysteine 0.73 ± 0.06 µg/µl [134] Homocysteine 1.31 ± 0.62 µg/µl [134] Methionine 5.74 ± 0.93 µg/µl [134] OTHER MOLECULES Molecule Concentration Units Ref Hemoglobin 300 nm [22] Glucose 140 mg / 100 ml blood [135]

160 APPENDIX B: PROTOCOL FOR GPX ASSAYS 143

161 144 Introduction Glutathione peroxidase (GPx is a cytosolic enzyme that detoxifies hydrogen peroxide (H 2 O 2 ). The assay works by coupling the formation of GSSG by GPx to the regeneration of this compound by the enzyme glutathione disulfide reductase (GR). This reaction uses NADPH as the reductant. The product of the reaction is 2 glutathione molecules and NADP +. The substrate NADPH absorbs light at 340 nm whereas the product NADP + does not. This allows for the loss of NADPH over time to be monitored at 340 nm [136]. The assay was adapted for use with a plate reader. This is achieved by utilizing a correction factor for the difficulty in NADPH quantitation due to path length. Solutions and Buffers: Name Preparation The following solution can be made before the assay, and stored for 1 year at 25 C 7.56 g Potassium phosphate monobasic 0.41 g EDTA Buffer #1A g NaN 3 1 L H 2 O 9.68 g Potassium phosphate dibasic 0.41 g EDTA Buffer #1B g NaN 3 1 L H 2 O Buffer #1 Add buffer 1B to Buffer 1A until ph is g GSH ml GR (100 U/mL) in 50 ml Buffer 2 buffer # g Potassium phosphate dibasic g Buffer 3 EDTA in 1 L H 2 O, add NaOH to ph 7.8 The reagents below need to be made on the day of the experiment, and kept on ice 4 mm NADPH g NADPH in 3 ml buffer #1 2.5 mm H 2 O 2 20 µl 30% H 2 O 2 in 50 ml H 2 O, adjust by dilution with H 2 O until Abs = at 240 nm Validity of Assay linearity To verify that GPx is the limiting reagent in the assay, a validation with commercially available GPx should be performed. This establishes the range of concentration in which unknown samples are valid. This step should be repeated until operators are proficient in the assay and have reproducible intra-day variability.

162 145 Commercial source Bovine GPx is obtained from Sigma Aldrich as lyophilized powder (G6137). Keep GPx stock #1 and #2 at -20 ⁰C. Keep other standards on ice until use. Name Preparation 10 U/mL GPx stock #1 Dissolve contents of a vial in buffer #1 to a concentration of 10 U/mL, prepare 45 µl aliquots and freeze. 1 U/mL GPx stock #2 Add 40 µl, GPx stock #1 to 0,360 ml buffer #1 1. Add the following concentrations of GPx stock #2 to an empty tube, followed by addition of the desired volume of buffer #1. 2. Do not re-freeze aliquot, use fresh aliquots for each day the experiment is performed. Concentration (U/mL) Volume Stock #2 (µl) Volume buffer #1 (µl) Sample preparation: Cultured cells (attached) 1. Remove media from flask and wash twice with cold 0.9 % saline solution. 2. Cover growing surface with 0.9 % saline solution and collect cells with a cell scraper. 3. Collect cells in 15 ml centrifuge tube; wash once more with NaCl buffer. 4. Centrifuge to form pellet (5 min, 500 g). 5. Remove buffer, and resuspend in 500 µl buffer. 6. Transfer suspension to 2 ml eppendorf tube, and centrifuge at +4⁰C for 5 min at 0.3 RCF. 7. Remove buffer and resuspend in 200 µl buffer. (this suspension can be stored frozen) 8. Thaw the cell suspension, and sonicate on 20% intensity for several seconds to lyse. 9. Centrifuge at +4 ⁰C for 5 min at 500 g. 10. Collect supernatant without disturbing the pellet, and transfer to separate 2 ml eppendorf tube.

163 146 Homogenized tissue 1. Place tissue in glass manual homogenizer. 2. Fill glass tube with 200 µl buffer #1. 3. Homogenize for 1 min. 4. Place homogenate in 2 ml eppendorf tube. 5. Centrifuge at +4⁰C for 5 min at 500 g. 6. Collect supernatant without disturbing the pellet, and transfer to separate 2 ml eppendorf tube Red Blood Cells 1. Separate RBCs from plasma by centrifuge at 1500 g for 5 min 2. Remove plasma and add 5 ml of isotonic saline. 3. Centrifuge again and repeat steps 2-3 three times. 4. Dilute cells to desired density (3 million cells/µl) 5. Lyse cells in nanopure water to a density of 40,000 cells/µl 6. Use in the assay or freeze at -80ºC for future use. Assay procedure: Equipment UV-Vis spectrophotometer and/or microplate reader Plastic 96-well plate 1 cm quartz cuvette Standard pipettes Cuvette UV-Vis spectrometer 1. Add to 1 ml cuvette a. 750 µl Buffer #2 b. 50 µl 4mM NADPH solution c. 100 µl sample or standard (buffer can be used for blanks) 2. Cover with film and mix by inverting 3. Incubate at 30 ⁰C for 5 min 4. Add to the cuvette a. 100 µl H 2 O 2 or CuOOH to determine selenium depended and total activity respectively 5. Cover with film and mix by inverting 6. Follow reaction kinetics for 5 min at 340 nm

164 147 Plate reader 1. Add to each well (one row at a time) a µl buffer #2 b µl 4mM NADPH solution c. 25 µl sample or standard (buffer can be used for blanks) 2. Incubate at room temperature for 5 min 3. Add to each well (multipipetter) a. 25 µl H 2 O 2 or CuOOH to determine selenium-dependent and total activity respectively 4. Start measurement Plate reader settings: 1. Absorbance : 340 nm 2. Plate : GRE96 3. # Measurements : 1 4. Kinetics : 45 cycles, minimal interval Assay calculations, GPx copy number: Standards: 1. Determine slope for first 200 seconds (Abs/s) 2. Correct for background loss of NADPH (subtract blank average from each standard) 3. Determine ng GPx for each standard (amount of GPx in standard / 51 u / mg * 1 x 10 6 ) 4. Determine slope and y-intercept for best fit line (linest X=ng GPx; Y=corrected rate (Abs/s)) Samples: 1. Determine slope for first 200 seconds (Abs/s) 2. Correct for background loss of NADPH (subtract standards blank average from each sample) 3. Determine ng GPx using the calibration curve from standards 4. Convert ng GPx to grams and divide by molecular weight (22 kda) (mole GPx) 5. Multiply by Avogadro s number (6.022e 23 ) to obtain # GPx 6. Divide by protein conc. (mg/25 µl) to calculate #GPx /mg protein/sample

165 148 Unit ( mole/min): Standards: 1. Determine slope for first 200 seconds (Abs/s). 2. Correct for background loss of NADPH (subtract blank average from each standard). 3. Convert Abs/s to µmole/s using the extinction coefficient of NADPH (6.2e-3 µm/cm) and the total volume in the cuvette. 4. Multiply the value by 60 to get μmole/min or a Unit. Samples: 1. Determine slope for first 200 seconds (Abs/s). 2. Correct for background loss of NADPH (subtract standards blank average from each sample). 3. Convert Abs/s to μmole/s using the correction factor as determined previously (for this assay and filter setup = μmole/abs) 4. Multiply the value by 60 to get μmole/min or a Unit. 5. Divide by the amount of protein in the sample to obtain Unit/mg protein.

166 APPENDIX C: SAMPLE PREPERATION FOR HPLC-BDD 149

167 150 Protocol for the quantitative measurement of GSH and GSSG in red blood cells Developed by: Thomas Joost van t Erve V Introduction This protocol describes the determination of the concentration of GSH and GSSG in red blood cells (RBCs) using an HPLC-BDD approach. The general steps in the protocol are: 1. (Steps 1-6): Collect blood, remove the plasma, and wash the RBCs. This concentrates the RBCs and removes possible interfering substances. It also allows for the measurement of plasma-related compounds, such as ascorbate. 2. (Step 9): The number of RBCs needs to be determined by counting an aliquot using a Z2 coulter counter or other methods for counting particles in solution. This step is crucial since it is the normalizing factor (both volume and cell number/density are determined). 3. (Step 10): Remove interfering proteins by perchloric acid (PCA) precipitation and centrifugation. This step also acidifies the sample, ensuring that no oxidation occurs to alter the results. 4. (Optional, Step 11): A solution of lysed RBCs in Nanopure water can be prepared and stored for future studies. 2.0 Materials All materials are listed to process 1 sample in triplicate: three 1.5 ml conical tubes labeled for PCA supernatant three 1.5 ml conical tubes labeled for RBC lysates three 1.5 ml conical tubes labeled for protein precipitate one tube for RBCs in isotonic saline one tube for RBCs in CellPack counting solution one glass or conical tube that can hold 5 ml or more three 1.5 ml conical tubes for plasma (optional)

168 Stock Solutions PCA solution: The PCA solution is made by weighing out enough DETAPAC for 500 ml of a 100 µm solution. To this is added, water and PCA (75 % solution as bought from sigma). PCA is added to a final concentration of 5 % PCA in the 500 ml stock. This solution needs to be refrigerated and can be used for up to 1 year after preparation. 3.0 Experimental Procedure 3.1 Tube with anti-coagulant 1. Collect 1 tube of blood from donor with EDTA anti-coagulant (purple top / 4 ml); put on ice and process as below ASAP. 2. Aliquot the sample into three 1.5 ml Eppendorf tubes. 3. Centrifuge the sample for 5 min at 500 g (centrifuge in cell culture room / 2000 RPM). 4. Harvest clean plasma from RBCs using a pipette, aliquot and store the plasma at - 80 C for ascorbate analysis (avoid disturbing the pellet). 5. Remove the remaining plasma and buffy coat and discard. 6. Wash RBCs by adding 3 ml 0.9 % cold saline followed by centrifugation at 500 g for 5 min. 7. Remove the saline solution (if a white layer persists on top of the pellet, remove this during the wash). a. Repeat steps 5 and 6 three times. 8. Pellet the RBCs to produce a packed RBC solution. 9. Remove aliquot of RBCs for cell counting (see chapter 4 for counting methods and procedure) 10. Remove aliquot of RBCs for osmotic fragility (Pipette 10.0 µl packed RBCs in 1000 µl 0.9 % saline) 11. Pipette µl packed RBCs into 1.5 ml eppendorf and add µl of 5 % PCA/100 µm DETAPAC. a. Vortex for 1 min or until solution is homogeneous and brown. b. Centrifuge at 4,000 g for 5 min (Eppendorf microfuge / 10,000 RPM). c. Remove the clear supernatant taking extra care not to disturb the fragile pellet. i. If needed the supernatant can be transferred to a clean 1.5 ml Eppendorf and re-centrifuged to remove more protein (sample should be completely clear). d. (store the aliquot at -80 C until it can be assayed)

169 152 e. Transfer 50 µl of the clear pellet to a 2 ml auto-sampler vial with 50 µl insert. f. Aliquot and store the remaining supernatant at -80 C. g. Repeat for all three samples. 12. Pipette µl packed RBCs into 1.5 ml eppendorf and add 900 µl Nanopure water. a. Vortex for 10 seconds b. Store the aliquots at -80 C. c. Repeat three times. 3.2 Segments of blood from donor bag 1. Drain sample in 5 ml conical tube by cutting both sides of sealed tube. 2. Aliquot the sample into three 1.5 ml Eppendorf tubes. 3. Centrifuge the sample for 5 min at 500 g. 4. Remove storage solution and collect in 1.5 ml Eppendorf tube. 5. Wash RBCs by adding 3 ml 0.9 % cold saline followed by centrifugation at 500 g for 5 min. 6. Remove the saline a. Repeat steps 5 and 6 three times. 7. Pellet the RBCs to produce a packed RBC solution. 8. Remove aliquot of RBCs for cell counting (see chapter 4 for counting methods and procedure) 9. Remove aliquot of RBCs for osmotic fragility (Pipette 10.0 µl packed RBCs in 1000 µl 0.9 % saline) 10. Pipette µl packed RBCs into 1.5 ml eppendorf and add µl of 5% PCA/100 µm DETAPAC. a. Vortex for 1 min or until solution is homogeneous and brown. b. Centrifuge at 4,000 g for 5 min (Eppendorf microfuge / 10,000 RPM). c. Remove the clear supernatant taking extra care not to disturb the fragile pellet. i. If needed the supernatant can be transferred to a clean 1.5 ml Eppendorf and re-centrifuged to remove more protein (sample should be completely clear). d. (store the aliquot at -80 C until it can be assayed) e. Transfer 50 µl of the clear pellet to a 2 ml auto-sampler vial with 50 µl insert. f. Aliquot and store the remaining supernatant at -80 C. g. Repeat for all three samples. 11. Pipette µl packed RBCs into 1.5 ml Eppendorf and add 900 µl Nanopure water. a. Vortex for 10 seconds

170 153 b. Store the aliquots at -80 C. c. Repeat 3 times. 4.0 Counting RBCs 4.1 Using the Z2 coulter counter a. Pipette 10.0 µl packed RBCs in 1000 µl 0.9 % saline. b. Pipette 40.0 µl of RBC counting aliquot into 20.0 ml of isotonic counting solution. c. Count sample using coulter counter, multiply value by 10,000 and take median value as cell density in million cells / µl. Repeat step c three times and calculate median value of the three densities. 4.2 Using the Sysmex RBC flow cytometer 5.0 Important notes a. Pipette 30.0 µl packed RBCs in 120 µl CellPack solution. b. Program Sysmex to Capillary, and method 5 c. Introduce the sample and obtain results Packed red blood cells need to be extremely well mixed before pipetting. This is best done with a vortexer, followed by repeated up at down pipetting of the sample. Since packed red blood cells are very concentrated, 1 L extra in a sample can add millions of cells. It is therefore imperative that only the tip of the pipette is submerged when transferring RBCs to a new solution. PCA might not be the best acid to precipitate the proteins due to supposed generation of H 2 O 2. A better alternative might be meta-phosphoric acid.

171 APPENDIX D: BIOGRAPHY 154

172 155 Curriculum Vitae Education: May 2013 Ph.D., Human Toxicology, The University of Iowa, Iowa City, IA, B.A.Sc, Chemistry, Saxion University of Applied Sciences, Enschede, The Netherlands Employment History: Visiting Fellow, NIEHS, Research Triangle Park, North Carolina Graduate Research Assistant, The University of Iowa, Iowa City, Iowa Research Asst I Nat/Hlth Sci, The University of Iowa, Iowa City, Iowa Awards: April , Research Interests: Health Sciences Research Week, The University of Iowa, Iowa City, IA: Best poster, T2 translational category. Identification of biomarkers for use in disease and toxicity prediction Develop novel methods to measure biomarkers Create assays to improve the determination of toxicity upon exposure Improving translation of toxicity studies from cell culture models to animals and humans Experience: Graduate Thesis Research: The University of Iowa, (adviser: Dr. Garry R. Buettner): Cell culture toxicity assays (ATP per cell, clonogenic survival, trypan blue) Exposure of xenobiotics to cells in culture Quantitation of GSH and GSSG in blood, cells, and tumors with HPLC-BDD ESR detection of spin trapped radicals with DMPO and POBN Incubation of compounds with CYP P450 supersomes for ESR spin trap Microplate reader assay for ATP in red blood cells Quantitation of GSH adducted to protein with HPLC-BDD ESR detection of adducted benzoquinones to intact cells Mathematical modeling with CellDesigner and COPASI Microplate absolute quantitation glutathione peroxidase Laboratory Rotations: The University of Iowa, 2009 (adviser: Dr. Jerry Schnoor): Growth of bacteria from soil polluted with explosives Isolation of DNA from bacteria

173 156 PCR amplification with universal primers DNA detection via gel-electrophoresis and ethidium bromide staining Cloning of isolated DNA Sequencing of DNA to identify Laboratory Rotations The University of Iowa, (adviser: Dr. Garry Buettner): Synthesis of oxidized hydroxytyrosol Synthesis of a redox active iron chelator (DP44MT) Chemical characterization with HPLC-MS, GC-MS and UV-VIS ESR of PCB-quinones in fetal bovine serum Laboratory Rotations The University of Iowa, 2008 (adviser: Dr. Michael Duffel): In vitro inhibition studies of human-sulfotransferase 2A1 with mono-hydroxy-f- PCBs (OH-F-PCBs) Mathematical modeling of OH-F-PCB for QSAR studies. In vitro binding assays of OH-F-PCBs to human sulfotransferase 2A1. Undergraduate Thesis Research: Saxion University of Applied Sciences / The University of Iowa, (advisers: Dr. Gregor Luthe & Dr. Larry Robertson): Isolation of CYP P450 induced microsomes from rats Characterization of CYP P450 induced microsomes by enzyme activity assays In vitro metabolism of 3-Chloro biphenyl with rat microsomes Synthesis of 6 fluorinated mono-hydroxy-pcb standards (OH-F-PCBs) Chemical characterization of F-PCB by GC-FID, GC-MS, NMR Derivatization of mono-hydroxy-pcb metabolites (OH-PCBs) Analysis of OH-PCBs metabolites from in vitro metabolism studies with GC-MS Undergraduate Research: Saxion University of Applied Sciences, (adviser: Drs. Jan Scharp): Student Leader of organic synthesis group Organic synthesis of various chemical compounds o 4-keto-1,2,3,4 tetrahydrodibenzothiophene (precursor to internal standard for petroleum analysis) o 1-(trifluoromethyl)dibenzo[b,d]thiophene (internal standard for petroleum analysis) o 2,2',6,6'-tetrachloro-3,3'-dimethoxybiphenyl (environmental pollutant (PCB)) Chemical analysis of prepared compounds by GC-FID, NMR Publications: 1. van t Erve TJ, Wagner BA, Ryckman KK, Raife TJ, Buettner GR. (2013) The concentration of glutathione in human erythrocytes is a heritable trait. (in prep) 2. Wagner BA, Witmer, JR, van t Erve, TJ. Buettner GR. (2013) An assay for the rate of removal of extracellular hydrogen peroxide by cells. Redox Biology. 1(1): DOI: /j.redox Open access.

174 Welsh JL, Wagner BA, Van't Erve TJ, Zehr PS, Berg DJ, Halfdanarson TR, Yee NS, Bodeker KL, Du J, Roberts LJ 2nd, Drisko J, Levine M, Buettner GR, Cullen JJ. (2013) Pharmacological ascorbate with gemcitabine for the control of metastatic and node-positive pancreatic cancer (PACMAN): results from a phase I clinical trial. Cancer Chemother Pharmacol. 71(3): PMCID:PMC DOI: /s Olney KE, Du J, van 't Erve TJ, Witmer JR, Sibenaller ZA, Wagner BA, Buettner GR, Cullen JJ. (2013) Inhibitors of hydroperoxide metabolism enhance ascorbate-induced cytotoxicity. Free Radic Res. 47(3): PMID: DOI: / van 't Erve TJ, Rautiainen RH, Robertson LW, Luthe G. (2010) Trimethylsilyldiazomethane: A safe non-explosive, cost effective and less-toxic reagent for phenol derivatization in GC applications. Environ Int. 36(8): PMID: DOI: /j.envint PMCID: NIHMSID Oral Presentations: Date Title Location Forecasting the Cellular Response to Hydrogen Peroxide: Considerations for Anti-Cancer Therapies and Xenobiotic Toxicity September 2, 2010 November 17-21, 2010 June 1, 2011 October 20, 2011 May 1, 2012 October 5, 2012 Predicting Cell Vulnerability to Hydrogen Peroxide Generating Compounds: Possible Use in Anti-Cancer Therapies The Intracellular Redox Buffer: Applications in Toxicology Alterations in the Intracellular Redox Buffer: First Observations on RBC Storage and Exposure to Xenobiotics Rethinking Exposures in Cell Culture: Mol per Cell and beyond Heritability of thiol and ATP status during the storage of RBCs at the DeGowin Blood center Free Radical and Radiation Biology Program seminar, The University of Iowa, Iowa City, IA Society for Free Radical Biology and Medicine (SFRBM) Annual Meeting, Orlando, FL Iowa Superfund Research Program (ISRP) monthly meeting, Iowa City, IA Free Radical Radiation Biology Program seminar The University of Iowa, Iowa City, IA Iowa Superfund Research Program (ISRP) monthly meeting, Iowa City, IA EHSRC/Superfund/Human Toxicology Seminar Series, Iowa City, IA

175 158 November 29, 2012 January 31, 2013 May 2, 2013 Poster Presentations: A Twin Study of the Genetic Determinants of Blood Storage Glutathione concentration in human erythrocytes is a heritable trait Project 1 Specific aim 3, overview. Pathology Grand Rounds, The University of Iowa Carver College of Medicine, Iowa City, IA Free Radical Radiation Biology Program seminar The University of Iowa, Iowa City, IA Iowa Superfund Research Program (ISRP) monthly meeting, Iowa City, IA Date Title Location February 7-12, 2010 April 13, 2010 November 5, 2010 April 12, 2011 November 3, 2011 November 16-20, 2011 April 3, 2012 Simulation of Lifespan and Diffusion Distance for Superoxide and Hydrogen Peroxide in Human Erythrocytes Simulation of Lifespan and Diffusion Distance for Superoxide and Hydrogen Peroxide in Human Erythrocytes Predicting the Vulnerability of Cells to Hydrogen Peroxide: Consequences for Xenobiotic Toxicity Oxidation of the Intracellular Redox Buffer in Human RBCs by Hydrogen Peroxide: Absolute Quantitation using a new HPLC Technique. Quinone-Containing Xenobiotics Oxidize the Intracellular Redox Buffer: The Chemistry and Biology of a Redox Poison. Quinone-Containing Xenobiotics Oxidize the Intracellular Redox Buffer: The Chemistry and Biology of a Redox Poison Concordance of Glutathione Metabolites in the Red Blood Cells of Mono-zygotic Twin Pairs: Implications for the RBC Storage Lesion. Gordon Research Conference, Ventura, CA University of Iowa Health Sciences Research Week, Iowa City, IA Central States Society of toxicology (CS-SOT) regional meeting, the University of Iowa, Iowa City, IA University of Iowa Health Sciences Research Week, Iowa City, IA Iowa Superfund Research Program (ISRP) Meeting. Iowa city, IA Society for Free Radical Biology and Medicine (SFRBM) Annual Meeting, Atlanta, GA University of Iowa Health Sciences Research Week, Iowa City, IA

176 159 April 16, 2013 Date Title Location The Concentration of Glutathione in Human Erythrocytes is a Heritable Trait University of Iowa Health Sciences Research Week, Iowa City, IA April 16, 2013 April 23, 2013 The Heritability of Red Blood Cell Metabolites Influencing the Storage Lesion The Heritability of Red Blood Cell Metabolites Influencing the Storage Lesion University of Iowa Health Sciences Research Week, Iowa City, IA 14th Annual Student Interdisciplinary Health Research Poster Session, Iowa city, IA