Structure-Function Studies of Xanthine Oxidoreductase

STRUCTURE-FUNCTION STUDIES OF XANTHINE OXIDOREDUCTASE

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree of Doctor of

Philosophy in the Graduate School of The Ohio State University

By

James Michael Pauff

The Ohio State University

2008

Dissertation Committee:

Professor Virginia Sanders, Advisor on Record Approved By Professor Russ Hille, Graduate Advisor

Associate Professor Charles E. Bell ______Professor Allan Yates Advisor

Integrated Biomedical Science Graduate Program

ABSTRACT

Xanthine oxidoreductase (XOR) is a 290 kDa molybdenum-containing enzyme that catalyzes the final two steps in human purine catabolism, taking hypoxanthine to xanthine and then on to uric acid. The enzyme exists as a homodimer, with each monomer possessing a catalytic active site that contains one molybdenum atom coordinated to a pterin ring via an enedithiolate side chain. Each monomer also contains two [2Fe2S] clusters and one equivalent of FAD. These centers form an electron transfer chain as electrons are passed sequentially from the molybdenum to the FAD via the iron-sulfur clusters. The molybdenum center cycles from MoVI to

MoIV and back during catalysis, passing through an occasional MoV intermediate state.

Electrons are passed from the flavin site out of the enzyme to either NAD+ or molecular oxygen. XOR is initially expressed as a dehydrogenase (xanthine dehydrogenase, XDH) with a strong preference for reducing NAD+ to NADH,

+ although it can reduce O2 under conditions of low NAD concentrations. Under certain conditions, XDH can be converted by oxidation and/or limited proteolysis to an oxidase form (xanthine oxidase, XO) that utilizes O2 exclusively as the terminal electron acceptor, thereby generating superoxide and other reactive oxygen species.

The oxidase is formed primarily during hypoxia or ischemia, and the corresponding increase in reactive oxygen species has lead to investigation of the enzyme’s role in the pathophysiological mechanism of ischemia-reperfusion. The production of uric acid by XOR makes the enzyme a primary target in hyperuricemia, and it is this ii therapeutic intervention for which allopurinol has been used for over 60 years. The prevalence of XOR in mechanisms leading to some human diseases and the unique chemistry by which the enzyme catalyzes the production of uric acid makes this enzyme system an interesting subject of biochemical investigation. We have sought here to understand the structure and function of xanthine oxidoreductase, focusing on the nature of the molybdenum-containing active site as well as innate functional differences between the two forms of the enzyme. Our crystallographic and spectroscopic studies provide insight into the mechanism of XOR catalysis as well as the roles of XOR in human pathology.

iii

DEDICATION

To my parents, Neil and Deborah Pauff, my brother Frank, to my fiancé Beth, and to

my entire family

iv

ACKNOWLEDGMENTS

I would first like to thank my parents for countless hours (and dollars) spent raising, teaching, correcting, guiding, and being patient with me through the years.

Many lessons that I only realized much later have helped carry me and sustain me throughout my educational training.

I would like to thank Dr. Russ Hille for his support, guidance, and patience during my research training in his laboratories. My sometimes sporadic ideas were met with encouragement and a willingness to let me go my own way, even though in some instances that meant a waste of resources. Thank you for allowing me the independence to learn from my own successes and mistakes, and to grow as a researcher along the way. I would also like to thank Craig Hemann for his efforts, consultation, and advice, particularly in the early stages of my dissertation work. You remain a tremendous resource. To Dr. Silke Leimkühler, thank you for the collaboration and efforts with the xanthine dehydrogenase system. To Dr. Charles

Bell, thank you for the many hours of guidance and assistance with the crystallographic work. Without your advice we would not have any structures. I would also like to thank Dr. Allan Yates for his guidance and critical evaluation of my research and training as an MD/PhD student.

To the entire Hille lab past and present, thank you for the consultations and the laughter as we worked alongside one another.

v

VITA

1981 Born – Bluffton, OH

2004 B.S. Chemistry, B.S. Biochemistry, B.A. Zoology Miami University Oxford, OH

2004 – Present Graduate Fellow, Graduate Research Associate Medical Scientist Program The Ohio State University Columbus, OH

2007 – 2008 Junior Specialist Department of Biochemistry University of California at Riverside Riverside, CA

PUBLICATIONS

Pauff, J.M., Hemann, C.F., Jünemann, N., Leimkühler, S., and Hille, R. (2007) The

Role of Arginine 310 in Catalysis and Substrate Specificity in Xanthine

Dehydrogenase from Rhodobacter capsulatus. J. Biol. Chem. 282, 12785 –

12790.

Pauff, J.M., Zhang, J., Bell, C.E., and Hille, R. (2008) Substrate Orientation in

Xanthine Oxidase: CRYSTAL STRUCTURE OF ENZYME IN REACTION

WITH 2-HYDROXY-6-METHYLPURINE. J. Biol. Chem. 283, 4818 – 4824.

Fields of Study: Biochemistry, Enzymology, Macromolecular X-ray Crystallography

Major Field: Integrated Biomedical Science Graduate Program vi

TABLE OF CONTENTS

Abstract………………………………………………………………………………..ii

Dedication…………………………………………………………………………….iv

Acknowledgments……………………………..………………………………………v

Vita……………………………………………………………………………………vi

List of Figures…………………………………………………………………………x

List of Tables…………………………………………………………………………xii

Abbreviations………………………………………………………………………..xiii

References…………………………………………………………………………..154

Chapters:

1. Introduction………………………………………………………………………..1

1.1. General Background………………………………………………………….1 1.1.1. Molybdenum…………………………………………………………..1 1.1.2. Molybdenum in biology and molybdenum-containing enzymes…...... 2 1.1.3. Xanthine oxidoreductase………………………………………………4

1.2. Xanthine oxidoreductase in physiology………………………………………5

1.3. Xanthine oxidoreductase in pathology………………………………………..7 1.3.1. Inhibition of XOR……………………………………………………..9

1.4. Structure of xanthine oxidoreductase………………………………………..12

1.5. Mechanism of catalysis and the oxidation-reduction chemistry chemistry of XOR…………………………………………………………………………15 1.5.1. The molybdenum-containing active site……………………………..17 1.5.2. Oxidative and reductive half-reactions of XOR catalysis……………22 1.5.3. Active site residues in the molybdenum site…………………………25 1.5.4. The pterin cofactor, iron-sulfur clusters, and the flavin site…………29 vii

1.6. Preparation and Isolation of Xanthine Oxidoreductase in the Present Studies 32 1.6.1. Bovine xanthine oxidase and xanthine dehydrogenase………………32 1.6.2. Xanthine dehydrogenase from Rhodobacter capsulatus……………..34 1.6.3. Activity-to-flavin determination and percentage of active enzyme….34 1.6.4. Inactive and inactivating XOR……………………………………….35

2. The Role of Arginine 310 in the Active Site of Xanthine Dehydrogenase from Rhodobacter capsulatus………………………………………………………….37

2.1. Introduction………………………………………………………………….37 2.2. Materials and Methods………………………………………………………42 2.2.1. Preparation of wild-type and mutant xanthine dehydrogenase from Rhodobacter capsulatus………………………………………………...42 2.2.2. Rapid-reaction experiments…………………………………………..43 2.2.3. Data acquisition and analysis………………………………………...43 2.3. Results and Discussion………………………………………………………44

3. Substrate Orientation in Xanthine Oxidase from Bos taurus, Crystal Structure with 2-hydroxy-6-methylpurine………………………………………………….50 3.1. Introduction………………………………………………………………….50 3.2. Materials and Methods………………………………………………………51 3.2.1. Materials……………………………………………………………...51 3.2.2. Preparation and isolation of bovine xanthine oxidase………………..51 3.2.3. Crystal growth, diffraction, and data acquisition…………………….52 3.2.4. Data processing and refinement……………………………………...55 3.3. Results and Discussion………………………………………………………56 3.3.1. Overall structure of xanthine oxidase with 2-hydroxy-6-methylpurine………………………………………...56 3.3.2. 2-hydroxy-6-methylpurine in the active sites of xanthine oxidase…..60

4. Xanthine and Lumazine in the Active Site of Xanthine Oxidase from Bos Taurus………………………………………………………………….64

4.1. Introduction………………………………………………………………….64 4.2. Materials and Methods………………………………………………………68 4.2.1. Materials……………………………………………………………...68 4.2.2. Preparation of bovine xanthine oxidase and the desulfo-form……….68 4.2.3. Crystal growth, diffraction, and data acquisition…………………….69 4.2.4. Data processing and refinement……………………………………...70 4.3. Results and Discussion………………………………………………………71 4.3.1. Xanthine oxidase with lumazine……………………………………..71 4.3.2. Desulfo-xanthine oxidase with xanthine……………………………..75

5. Activity of Xanthine Dehydrogenase versus Xanthine Oxidase in the Range of Physiological pH…………………………………………………………………91

5.1. Introduction………………………………………………………………….91 viii 5.2. Materials and Methods………………………………………………………96 5.2.1. Preparation of bovine xanthine oxidase and bovine xanthine dehydrogenase…………………………………………………………..96 5.2.2. Steady-state kinetics and construction of the pH profile for bovine XDH…………………………………………………………………….98 5.2.3. Consumption of oxygen by XDH in the absence of NAD+………….99 5.3. Results and Discussion……………………………………………………..100 5.3.1. pH profiles for the two forms of XOR……………………………...100 5.3.2. Oxygen consumption by XDH……………………………………...105

6. Inhibition Studies of Xanthine Oxidase………………………………………...109

6.1. Introduction………………………………………………………………...109 6.2. Materials and Methods……………………………………………………..119 6.2.1. Compounds and reagents……………………………………………119 6.2.2. Isolation of xanthine oxidase………………………………………..119 6.2.3. Steady-state kinetics and absorption spectra………………………..120 6.2.4. Mass spectrometry to formally assess the hydroxylation of coumarin……………………………………………………………121 6.3. Results and Discussion……………………………………………………..122 6.3.1. Steady-state inhibition studies with luteolin, quercetin, silibinin, and allopurinol………………………………………………………...122 6.3.2. Superoxide production in the presence of luteolin, quercetin, silibinin, curcumin, and allopurinol…………………………………...131 6.3.3. Role of luteolin, quercetin, and silibinin and the inhibition of XOR………………………………………………………………...134 6.3.4. Steady-state inhibition studies with coumarin, esculetin, and ferulic acid………………………………………………………...136 6.3.5. Potential reduction of the molybdenum center and mass spectrometry data…………………………………………145

7. Conclusions……………………………………………………………………..146 7.1. Summary…………………………………………………………………...146 7.1.1. The role of arginine 880/310 in catalysis…………………………...147 7.1.2. The two forms of xanthine oxidoreductase…………………………149 7.1.3. Inhibition of xanthine oxidoreductase………………………………150 7.2. Future Directions…………………………………………………………...152

ix

LIST OF FIGURES

Figure 1.1 Molybdenum cofactors in the mammalian molybdenum-containing enzymes………………………………………………………………………………..3

Figure 1.2 Structures of allopurinol, xanthine, and febuxostat………………………11

Figure 1.3 One monomer of xanthine oxidase……………………………………….14

Figure 1.4 The reaction mechanism of xanthine oxidase…………………………….19

Figure 1.5 Spectrum for reduction of xanthine oxidase by dithionite………………..24

Figure 1.6 Alloxanthine-inhibited active site of xanthine dehydrogenase…………...26

Figure 1.7 Relative positions of redox-active centers in bovine XO………………...30

Figure 2.1 Proposed method of transition state stabilization by Arg 880……………39

Figure 2.2 Homologous series of six substrates used in study of Arg 310…………..41

Figure 2.3 Proposed orientation of xanthine and 2-hydroxy-6-methylpurine in the active site of xanthine oxidase………………………………………………………..48

Figure 3.1 Sample screenshot of XO crystals on rotating anode source……………..54

Figure 3.2 Overall structure of XO with HMP……………………………………….59

Figure 3.3 Omit maps with final model of XO-HMP structure……………………...61

Figure 4.1 Structures of xanthine and lumazine……………………………………...65

Figure 4.2 Reaction mechanism of xanthine oxidase with inactivation by cyanide…67

Figure 4.3 Lumazine in the active sites of bovine xanthine oxidase…………………74

Figure 4.4 Two separate monomers of XO following incubation with cyanide……..77

Figure 4.5 The interface between monomers in the homodimer of XO……………...79

Figure 4.6 Final structural model of the desulfo enzyme with xanthine……………..83 x

Figure 4.7 Active sites of desulfo-XO with xanthine………………………………...86

Figure 4.8 Alloxanthine and uric acid in the active sites of XOR……………………88

Figure 5.1 Flow chart for the development of ischemia with potential reperfusion injury…………………………………………………………………………………92

Figure 5.2 pH profiles for bovine xanthine oxidase………………………………...101

Figure 5.3 pH profiles for bovine xanthine dehydrogenase………………………...103

Figure 6.1 The orientation of Febuxostat in xanthine oxidase……………………...111

Figure 6.2 Natural products proposed to reduce or suppress activity of XOR……..113

Figure 6.3 The cardiovascular druges ticlopidine and clopidogrel…………………115

Figure 6.4 Coumarins investigated as potential inhibitors of XOR activity………..118

Figure 6.5 Double-reciprocal plots of [XO]/vobs vs. 1/[xanthine] for luteolin and quercetin…………………………………………………………………………….127

Figure 6.6 Time course of the inhibition of XO by allopurinol, luteolin, silibinin, and quercetin…………………………………………………………………………….129

Figure 6.7 Double-reciprocal plots of [XO]/vobs vs. 1/[xanthine] for esculetin and ferulic acid………………………………………………………………………….139

Figure 6.8 15-minute reactions with 50 µM allopurinol, esculetin, or ferulic acid...140

Figure 6.9 Active site of bovine xanthine oxidase with bound HMP………………142

xi

LIST OF TABLES

Table 2.1 Kinetic constants for wild-type, R310K, and R310M mutant xanthine dehydrogenase from Rhodobacter capsulatus……………………………………….45

Table 3.1 Structure factors and refinement statistics for the XO-HMP structure……57

Table 4.1 Statistics and structure factors for the crystal structure of XO with bound lumazine……………………………………………………………………………...72

Table 4.2 Statistics and structure factors for the crystal structure of desulfo-XO with bound xanthine……………………………………………………………………….81

Table 5.1 Oxygen consumption kinetics for XDH and XO………………………...106

Table 6.1 Steady-state kinetic constants for luteolin, quercetin, curcumin, silibinin, and allopurinol………………………………………………………………………124

Table 6.2 Inhibition studies on the production of superoxide by XO………………132

Table 6.3 Kinetic constants for potential steady-state inhibition of xanthine oxidase by coumarin, esculetin, and ferulic acid…………………………………………….137

xii

ABBREVIATIONS

AFR – activity-to-flavin ratio DNA – deoxyribonucleic acid DTT – dithiothreitol EDTA – ethylene diamine tetra-acetic acid EPR – electron paramagnetic resonance ESI – electrospray ionization EXAFS – extended X-ray absorption fine structure FAD/FADH2 – flavin adenine dinucleotide (fully oxidized/reduced) FDA – Food and Drug Administration FPLC – fast protein liquid chromatography g – gram HAP – hydroxyapatite HMP – 2-hydroxy-6-methylpurine HPRT – hypoxanthine phosphoribosyltransferase L – liter LPO – lipid peroxidase µ – micro = x 10-6 m – milli = x 10-3 Mo - molybdenum mol – mole = 6.022142 x 1023 units M – molar (mol/L) n – nano = x 10-9 NAD+/NADH – nicotinamide adenine dinucleotide (oxidized/reduced) PDB – protein databank PEG – polyethylene glycol (of various average molecular weights) PMSF - phenylmethylsulphonyl fluoride RNA – ribonucleic acid ROS – reactive oxygen species σ – sigma (standard deviation from a mean) TLS – translation, libration, screw-motion XDH – xanthine dehydrogenase XO – xanthine oxidase XOR – xanthine oxidoreductase

xiii

CHAPTER 1

INTRODUCTION

1.1 General Background

1.1.1 Molybdenum

Molybdenum is found in the 5th row of the periodic table below chromium and above tungsten, at atomic number 42, near the center of the d-block of “transition metals”. It is the 42nd most abundant element in the universe, and 25th most abundant element in the earth’s oceans (1). Molybdenum was discovered in 1778 by the

Swedish chemist Carl Welhelm Scheele in the form of molybdenite (MoS2), and purified as an element in 1781 by Peter Jacob Hjelm (2, 3). Molybdenum is an extremely versatile element and can withstand extreme temperatures without significant volume expansion or retraction, which accounts for its use in many industrial applications such as in steel alloys, electrodes, and inorganic catalysts of industrial reactions (4). Richard R. Schrock shared the 2005 Nobel Prize in

Chemistry with Robert H. Grubbs and Yves Chauvin for engineering the process of olefin metathesis using a molybdenum-based catalyst. Molybdenum can exist in any oxidation state from -2 to +6, can coordinate from 4-8 inorganic or organic ligands in very stable configurations, has a melting point over 2600 oC, and is relatively

2- abundant on earth as MoS2 in the oceans as MoO4 (5).

1 1.1.2 Molybdenum in biology and molybdenum-containing enzymes

By far the most common use of molybdenum in life is as a cofactor for enzymes. Like iron, zinc, copper, manganese, and cobalt, molybdenum can be utilized as a stably bound, variably coordinated cofactor in proteins, and is found in more than 50 molybdenum-containing enzymes throughout biology (6, 7). Enzymes that possess molybdenum in their active sites catalyze biological processes that are essential to the organism; indeed neither plants nor animals can survive without molybdenum (8). In prokaryotic organisms, the four most prevalent molybdenum- containing enzymes are 1) dimethylsulfoxide reductase (DMSOR), which catalyzes the reduction of dimethylsulfoxide to dimethylsulfide, 2) carbon monoxide dehydrogenase (CODH), which takes carbon monoxide to CO2, 3) nitrogenase, with its unique MoFe7 center that catalyzes the all-important process of nitrogen fixation to form bioavailable ammonia, and 4) dissimilatory nitrate reductase, which catalyzes

- the reduction of NO3 to N2 in autotrophic bacteria such that nitrate can be used as a terminal electron acceptor rather than O2. The last group of respiratory enzymes involved in denitrification is found predominately in facultative anaerobes, and differ from the molybdenum-containing assimilatory nitrate reductases that catalyze the conversion of nitrate to ammonium found in bacteria, fungi, and plants (9, 10).

In mammals, molybdenum is found in three different enzymes: aldehyde oxidoreductase (AOR), sulfite oxidase (SOX), and xanthine oxidoreductase (XOR).

Humans possess each of these enzymes, which differ slightly in the coordination of the molybdenum-containing cofactor as shown in Figure 1.1.

2

Figure 1.1 The molybdenum cofactors in the three mammalian molybdenum-containing enzymes. The pterin moiety of the cofactor is abbreviated where the molybdenum is shown and indicated below.

3 The molybdenum-containing enzymes including those above can be characterized into three main families, based on the coordination of the molybdenum center and their mechanism of catalysis. Each family is named and represented by a prototypical enzyme; 1) xanthine oxidoreductase, which includes aldehyde oxidase

(the molybdenum hydroxylases), 2) sulfite oxidase (the eukaryotic oxotransferases), and 3) DMSO reductase (the bacterial oxotransferases). The first two groups include all four of the molybdo-enzymes expressed in the human body (6, 11).

Molybdenum by itself is relatively inert in biological processes, and requires an additional pterin cofactor, such as those shown coordinated to the atom in Figure

1.1, to be biologically active in molybdo-enzymes (6). The only known exception to this is the MoFe7 center of nitrogenase as mentioned above, although obviously here the molybdenum requires an additional cofactor(s) (12). The pterin cofactor has been proposed as assisting in electron transfer out of the molybdenum atom, a point that is discussed below.

1.1.3 Xanthine oxidoreductase

One of the most studied of the molybdenum-containing enzymes is xanthine oxidoreductase. Known initially as Schardinger’s enzyme, XOR was discovered in

1902 by Austrian biochemist Franz Schardinger who observed that methylene blue could be reduced in fresh milk with the addition of formaldehyde (13). The responsible enzyme was later isolated, purified, and studied by Malcolm Dixon and

Sylva Thurlow in the 1920’s, and in 1938 it was suggested by V.H. Booth that

Schardinger enzyme be thereafter referred to as xanthine oxidase (14, 15). XOR was found capable of oxidizing aldehydes and xanthine, and is now the prototypical member of the family of proteins known as the “molybdenum hydroxylases”. This 4 large group of enzymes includes the four that are present in the human body: XOR as well as three different aldehyde oxidases that appear to be expressed as tissue-specific isozymes (12). The molybdenum hydroxylases catalyze the hydroxylation of a carbon center by a fundamentally different mechanism than the monoxygenases such as the cytochrome-P450 class of enzymes, utilizing water rather than O2 as the source of oxygen and generating rather than consuming reducing equivalents (16). The mechanism of XOR has come to represent this unique method of catalysis.

XOR can be isolated from fresh, unpasteurized mammalian milk, most often utilizing a bovine source (Bos taurus) as was the case in most early studies of the enzyme. The bovine enzyme is nearly identical to the human form of XOR at all levels of structure and cofactor positions, and has 90% sequence identity with the human liver enzyme (17).

1.2 Xanthine oxidoreductase in physiology

Xanthine oxidoreductase catalyzes the final two steps in purine degradation, taking hypoxanthine to xanthine and then xanthine to uric acid by sequential hydroxylations at the C2 and C8 positions of the molecule, respectively (12). As mentioned, in doing so the enzyme generates rather than consumes reducing equivalents, which sets it apart from other enzyme systems such as the cytochrome

P450s that consume electrons (e.g. NADPH) and molecular oxygen (16). The name xanthine oxidoreductase describes an enzyme that exists in one of two forms, differing in respect to the preferred terminal electron acceptor(s) (17, 18). XOR is initially expressed as a dehydrogenase (XDH) that utilizes NAD+ to produce NADH, and it exists predominately in this form under physiological conditions (19, 20). The enzyme can be converted to an oxidase (xanthine oxidase, XO) by oxidation and/or 5 limited proteolysis, a process that appears to be most prevalent in various pathologic states (18). XO utilizes O2 exclusively as its terminal electron acceptor and produces superoxide (17).

Another physiological role for XOR is in mammalian milk secretion. The enzyme has been shown to be membrane-associated with milk fat globules as they are enveloped by apical cell membranes, and several studies suggest that its presence is necessary for proper secretion of milk during lactation (21). The specific role of XOR in regards to lactation remains unknown.

XOR is found in virtually all cells throughout the human body, although its expression levels are highest in the hepatocytes, small intestinal enterocytes, and vascular endothelium (22). It is a cytosolic enzyme, although some cells do exocytose the enzyme under certain physiological and pathological conditions as discussed below (18).

Aside from its role in normal purine catabolism, many other physiological functions have been proposed for the enzyme. One of these involves the production

- - of nitric oxide from nitrate, via reduction of NO3 to NO2 and then to NO. This hypothesis has received considerable attention and has been reviewed and supported by several studies (23, 24). However, the involvement of XOR in NO production may be relevant in human pathology as mentioned below (24).

Another contested role for XOR is in antimicrobial immune defense (25), specifically via XOR-derived reactive oxygen species (e.g. hydrogen peroxide) to recruit neutrophils in the inflammatory response. Another involves possible bacteriocidal action of XOR-generated ROS as in milk, although it is not clear what the reducing substrate might be in such a case (25, 26). This topic is still an active area of investigation. 6

1.3 Xanthine oxidoreductase in pathology

Hyperuricemia is the most cited pathology involving XOR. It is a pathological state that arises from overproduction (by XOR) or underexcretion (renal tubule disorders) of uric acid (27). This condition is often clinically expressed as gout, with the deposition of uric acid crystals, particularly in the joints, giving rise to rheumatic problems such as dysarthria. A hyperuricemic state may also develop following tumor cell necrosis (producing the so called “tumor lysis syndrome”) as necrotic neoplastic cells spill purines into the interstitial fluid and plasma, potentially leading to renal failure (27, 28). Congenital diseases may also give rise to hyperuricemia, the two most common are Lesch-Nyhan and Kelley Seegmiller

Syndromes – X-linked recessive disorders involving the overproduction of uric acid due to complete or partial lack of hypoxanthine phosphoribosyltransferase (HPRT)

(27). HPRT acts to salvage purines from degraded DNA, taking intracellular hypoxanthine to inosine monophosphate (IMP) and xanthine to xanthine monophosphate (certain isozymes), and a deficieny or absence of this enzyme results in elevated concentrations of XOR substrates in the cell (27, 29, 30). The most profound manifestations of this pathology are in the central nervous system, where the accumulation of uric acid leads to clinical symptoms such as spasticity, hyperreflexia, choreoathetosis (involuntary movements ranging from flailing limbs to slow, writhing movements of the distal extremeties), mental retardation, and self-mutilation (27, 30).

The most used treatment for hyperuricemia of any cause is allopurinol, which has been a mainstay of clinical therapeutics for over 60 years. Allopurinol is converted to oxypurinol (alloxathine) in the process of inhibiting XOR, and is a

7 mechanism based inhibitor that actually inactivates the enzyme. It remains the only commercially available inhibitor of XOR in the United States (31).

The broader consideration of XOR in human pathology involves the ability of the enzyme to generate reactive oxygen species, via both the production of superoxide and peroxide, when using O2 as its terminal electron acceptor (18). As mentioned previously, the enzyme exists primarily as a dehydrogenase under physiological conditions, but XDH maintains the capacity to produce superoxide, particularly in the absence of NAD+ (32, 33). More relevant to human pathology is the conversion of

XDH to XO by thiol oxidation or limited proteolysis, yielding an oxidase form that is

+ + unable to reduce NAD . The ability of XDH to utilize NAD or O2 while XO is only able to utilize O2 stems from modifications at the flavin site the conversion of XDH to

XO (17).

Conversion of XDH to XO, with the subsequent increase in superoxide and peroxide, has been cited as a primary contributor to tissue damage in ischemia- reperfusion injury (31, 34). Ischemia-reperfusion injury refers to the magnification of necrosis and other tissue damage that occurs after the initial hypoxic/ischemic insult with subsequent reperfusion by oxygenated blood. In essence, the magnitude of recovery expected from reperfusion is significantly diminished, and the increased tissue damage/reduced recovery may be due to reactive oxygen species produced by

XOR (18, 35). The theory is that during hypoxia or ischemia, XDH is gradually converted to XO, and upon reperfusion the sudden availability of oxygen leads to a significant production of superoxide and peroxide by the oxidase form of XOR.

Also of note in vascular pathology, the exocytosis of XOR upon cellular insults/stress such as ischemia has been suggested to be a component in the pathogenesis of reperfusion injury (24, 36). In addition, several studies have 8 suggested that circulating XOR can bind vascular endothelial cells, thereby providing a focus for oxidative stress/damage at sites far removed from the initial injury (24, 37).

Another aspect of XOR activity that is often contrary to human health involves the enzyme’s wide substrate specificity. XOR is able to catalyze the hydroxylation of many nonphysiological substrates, including several therapeutic compounds (18, 38).

The chemotherapeutic agent 6-mercaptopurine is metabolized to an inactive compound by xanthine oxidoreductase, and one of the first applications of inhibitors directed at XOR involved the use of allpurinol in tandem with the drug (31).

Allopurinol was shown to prolong the efficacy of 6-mercaptopurine, and it was for this use of allopurinol to inhibit XOR that Gertrude B. Elion and George H. Hitchings shared the 1988 Nobel Prize in Physiology or Medicine.

1.3.1 Inhibition of XOR

XOR has long been a therapeutic target, as decreasing the production of uric acid and/or inhibiting the catabolism of various xenobiotics are desirable under a variety of circumstances. In addition, the production of reactive oxygen species by

XOR has been implicated in many human processes both physiological and pathological, ranging from heart failure to normal aging (39-41). Given the ability of

XOR to produce ROS in either a dehydrogenase or oxidase form, a reduction in enzymatic activity may be desirable even in physiological states prior to symptomatic hyperuricemia or tissue ischemia. To this end, many recent studies have looked for natural and synthetic inhibitors of XOR that could be used as part of a daily regimen, or be used to replace or synergize with allopurinol (31, 41, 42). Perhaps the most promising synthetic pharmaceutical is known as TEI-6720 or Febuxostat (31, 44-46).

This compound does not take advantage of the mechanism of XOR as does 9 allopurinol, but instead intervenes in the active site as a stably bound/docked molecule. As may be expected, Febuxostat bears no resemblance to xanthine or allopurinol, as shown in Figure 1.2.

10

Figure 1.2. Structures of allopurinol, xanthine, and Febuxostat.

11 The rational design of molecules that inhibit XOR must take into account the nature of molybdenum-containing active site, as well as the catalytic mechanism of the enzyme. Such considerations are also important in understanding the rate of and potential for hydroxylation of compounds other than hypoxanthine and xanthine, such as 6-mercaptopurine discussed above. Detailed structural knowledge of the enzyme may also highlight the basis for its specific functions in human physiology and pathology.

1.4 Structure of xanthine oxidoreductase

Much of the early work on the structure of XOR was conducted by biophysical, chemical, and spectroscopic methods, utilizing ultraviolet-visible absorbance, electrophoresis, and paramagnetic resonance techniques among others to elucidate the nature of the enzyme’s structure and mechanism. In 1939, Professor

E.G. Ball showed that the yellow substance obtained following denaturation of isolated XOR was a flavin adenine dinucleotide similar to that isolated by Warburg and Christian from amino acid oxidase, thus indicating that the enzyme possessed an

FAD cofactor (47). This was the first of the redox-active centers to be identified in the enzyme, although it was uncertain as to what role the FAD played in the function of XOR.

Another redox-active component of native XOR was identified in 1954 by

D.A. Richert and W.W. Westerfeld, who demonstrated the requirement of iron in the functional enzyme, thus making XOR an iron-sulfur protein (48). The location of the iron-sulfur clusters and their part in electron transfer within XOR would have to wait for more structural work, but the involvement of these centers in intramolecular electron transfer was recognized with their discovery. 12 That xanthine oxidoreductase contained molybdenum was elegantly and unequivocally demonstrated by Dr. R.C. Bray in 1966 with the observation of a paramagnetic MoV species in the enzyme (49). Prior to this, the presence of molybdenum in aldehyde oxidase (another molybdenum hydroxylase) had been established and its role in this enzyme, as well as in XOR, was proposed to be that of aiding the catalytic hydroxylation of substrates (50).

Having thus established the presence of multiple redox-active centers in the functional enzyme, and having demonstrated a complex process of electron transfer through these centers with catalysis at the molybdenum site, further interpretation determined that the enzyme existed as a homodimer in solution (51). Each respective monomer was shown to contain a molybdenum-containing active site as well as the additional redox-active centers.

Crystal structures of the bovine enzyme were first determined separately by

Pai, Nishino, and coworkers in 2000 (17, 52). These studies provided direct assessment of the redox-active centers in each monomer of the homodimeric protein, confirming previous conclusions based on non-crystallographic techniques. The bovine enzyme is a 290 kDa homodimer, similar in molecular weight to the 275 kDa homodimeric human enzyme. XOR from both sources contains one molybdenum atom site, two [2Fe-2S] clusters positioned approximately 10 Å away, and an FAD site where electrons are finally transferred out of the protein to a terminal electron acceptor. The position of these centers is shown in Figure 1.3.

13

Figure 1.3. One monomer of xanthine oxidase. Redox-active centers are shown as spheres. From bottom-to-top: molybdenum-containing active site with coordinated molybdenum and pterin ring; 2Fe-2S cluster; 2Fe-2S cluster; flavin site with FAD. (adapted from 17, PDB code 1FIQ)

14 As mentioned, XOR exists as either XDH or XO, which preferentially utilize

+ NAD or O2 respectively as the terminal electron acceptor from the flavin site. The specific preference of oxidizing substrate derives from the nature of the FAD site of the enzyme. XOR is initially expressed as XDH, but can undergo conversion to XO by oxidation of sulfhydryl residues or by limited proteolysis around the flavin site, closing access to the site such that XO can utilize only molecular oxygen as its final electron acceptor (17). Prior to this modification, XDH can use either electron acceptor, although it maintains a strong preference for NAD+ as given by the much

+ + lower Kd for binding NAD over O2 and a rate constant of NAD reduction that is 5 to

10-fold greater than that for O2 reduction (33).

The catalytic sequence of XOR begins at the molybdenum site where electrons are introduced into the molybdenum cofactor, which is reduced from MoVI to MoIV

(53). From there, electrons are transferred one at a time to the more proximal of two

[2Fe-2S] clusters, positioned approximately 10-15 Å away. Each electron then moves approximately 8-10 Å to the second cluster, and finally each is passed 10-12 Å to the

FAD, forming FADH and then FADH2 prior to reduction of the final exogenous oxidant. This sequence of intramolecular electron transfer occurs independently in each respective monomer, and is not rate limiting in catalysis by the enzyme (54).

1.5 Mechanism of catalysis and the oxidation-reduction chemistry of XOR

The catalytic mechanism at the molybdenum-containing active site of XOR, as well as the nature of electron transfer within and out of the enzyme, have been the subject of investigation since the first isolation of the enzyme. Many of the first mechanistic studies utilized electron paramagnetic resonance spectroscopy (EPR) and anaerobic reductive titrations with sodium dithionite, given what is now known that 15 all four redox active centers in each monomer of XOR can exist in a paramagnetic form. Kinetic studies were also employed with rapid freeze-quenching to prepare samples at different points in the reaction mechanism, for study by EPR. Two such studies in 1964 by Bray, Palmer, and Beinert probed the ligand environment of the molybdenum and MoV species during catalysis by the enzyme with xanthine, comparing this to the molybdenum centers of inorganic molybdenum complexes.

These same studies investigated the kinetics of electron transfer between catalytic components of the enzyme, as well as noting the presence of nonheme iron in the enzyme (55, 56). It was concluded that the molybdenum center, iron centers, and

FAD center are all involved in electron transfer during catalysis and cycle between different redox-states. In a 1969 study by Orme-Johnson and Beinert, EPR was used to suggest the transfer of electrons between 2Fe-2S clusters in a direction “away” from the molybdenum center of the enzyme. Shortly after this finding, R.C. Bray, V.

Massey, and others subsequently built on previous EPR work and g-values for the redox centers in XOR to investigate interactions between the Mo and 2Fe-2S clusters, and a 1974 study by J.S. Olson and coworkers continued to probe the sequence of electron transfer in the enzyme (57-59). Expounding on the details of these studies, the ordered sequence of electron transfer from the molybdenum (cycling between

MoVI-MoIV-MoVI) to 2Fe-2SI, to 2Fe-2SII, and finally on to the FAD in the flavin site of the enzyme eventually became apparent (12). The order of electron transfer coupled with the known involvement of molybdenum as essential to the function of

XOR indicates that electrons must be introduced into the molybdenum at the beginning of the redox sequence of the enzyme, and it is the flavin site where electrons are finally transferred out of the enzyme.

16 1.5.1 The molybdenum-containing active site

The exact manner by which substrate introduces electrons into the molybdenum cofactor has been the subject of much work. Mechanisms based on kinetic assays, spectroscopic techniques, and the known chemistry of model inorganic molybdenum complexes were proposed early in investigations of the enzyme, ranging from oxidation by atmospheric oxygen to the (correct) hypothesis of oxygen transfer from the molybdenum coordination sphere to the carbon of substrate. In their 1964 studies, Beinert et al. utilized EPR to propose three nonequivalent pairs of ligands around the Mo, based on an octahedral coordination sphere given this preference by the atom in inorganic Mo-complexes (55). Structural work would eventually show that the Mo adopts a square pyramidal configuration in the enzyme as discussed below. In 1966, K.N. Murray and others demonstrated that the source of the oxo group transferred to xanthine was ultimately solvent water and not molecular oxygen

(60). This same study confirmed that molecular oxygen functions only as an electron acceptor in the mechanism of XOR. Further support for the oxo-transfer mechanism of catalysis, with the catalytic Mo-oxygen regenerated by solvent water, was obtained in 1987 by R. Hille and H. Sprecher using mass spectrometry (61). In 1999, Xia et al. concluded, based on kinetic and EPR experiments, that the catalytically labile molybdenum-oxygen incorportated into substrate was a Mo-OH ligand rather than

Mo=O (62). In addition to the oxygen ligands, an earlier structural characterization of the molybdenum site in 1979 by Tullius and coworkers utilizing EXAFS demonstrated the presence of oxo, sulfido, and thiolate groups in the molybdenum coordination sphere as in Figure 1.1 (63). The first X-ray crystal structure of a molybdenum hydroxylase was reported in 1996 by Huber et al., who worked with aldehyde oxidoreductase from Desulfovibrio gigas, a member of the xanthine 17 oxidoreductase family of enzymes (64). Based on that structure, the molybdenum center of xanthine oxidase had a square-pyramidal coordination geometry, with two dithiolene sulfurs from the pterin (discussed below), one oxo, one sulfido, and a hydroxyl/water ligand. The exact position of the sulfido and oxo ligands would be determined later to be that shown in Figure 1.1, with an oxo group in the apical position and the sulfido ligand in the same plane as the hydroxyl (Mo-OH) group introduced into substrate as discussed below. The mechanistic study by Huber proposed attack on the carbon center to be hydroxylated by a molybdenum- coordinated hydroxyl group with concomittant hydride transfer from this same carbon center to the Mo=S, thereby reducing the MoVI to MoIV with the new Mo-SH ligand.

Regarding the chemistry by which substrate became hydroxylated and the molybdenum reduced, also given that the MoV state was known to exist at least transiently during catalysis, it was suggested in 1999 that electron transfer from substrate might occur by two one-electron steps, taking the MoVI to MoV and then to

MoIV (65). Subsequent work in 2002 by A.L. Stockert et al. demonstrated that there was no linear relationship between the kred or kred/Kd of substrate oxidation relative to the reduction potentials for several purine substrates, a relationship that would be expected if sequential one-electron steps occurred in reducing the Mo-center of the enzyme. Thus a one-electron transfer mechanism was not plausible, and the two- electron reduction of the molybdenum was upheld (53). The most recent formulation of the mechanism of xanthine oxidoreductase is shown for the bovine enzyme in

Figure 1.4.

18

Figure 1.4. The reaction mechanism of xanthine oxidoreductase with xanthine at the molybdenum- containing active site. Also shown is the proposed role of Glu1261 (numbering convention for the bovine enzyme; Glu1260 in the human enzyme) as an active site base.

19 As shown in Figure 1.4, the mechanism is initiated by base-assisted nucleophilic attack by the planar Mo-O(H) ligand in the molybdenum coordination sphere, assisted by a highly conserved active site glutamate (as discussed below and in references 12, 64). With the concomitant hydride transfer from the substrate carbon to the planar Mo=S, the molybdenum center is reduced directly to the MoIV state (12). The MoIV species thus has what is now bound product coordinated to the

Mo via a bridging oxygen, with a Mo-SH ligand replacing the Mo=S of the oxidized center (Figure 1.4). Once reduced, the molybdenum center must be reoxidized by transferring electrons to the other redox centers of the enzyme. This occurs by one- electron transfer out of the MoIV through the bonds of a pterin cofactor, although the rate at which each successive electron is transferred out in yielding back the MoVI state appears variable. The transfer of a single electron gives a transient paramagnetic

MoV species that exists until the second one-electron transfer to return to the MoVI state (12, 64-68). It is this +5 species that is amenable to study by EPR and other spectroscopic means of analyzing paramagnetic atoms. The course for the existence/accumulation of this paramagnetic molybdenum species is dependent on the rate of electron transfer out of the Mo, and appears to vary based on the particular substrate and even reaction conditions used in a study. Some substrate analogs (e.g.

2-hydroxy-6-methylpurine) give rise to a significantly higher percentage of the MoIV atoms that pass through the MoV state prior to the dissociation of product rather than directly reoxidizing to MoVI with concurrent product release, although the fraction of a given number of molybdenum centers that pass through the “very rapid” MoV state for any substrate is actually rather small (12). In the reaction with xanthine at a pH<8.5, for instance, very little of this MoV state is observed as the molybdenum

20 center instead undergoes two rapidly successive electron transfers back to MoVI with release of product (66, 68).

It is also worth noting again that the rate-limiting step in the catalytic mechanism of converting xanthine to uric acid is not any process of the electron transfer(s) or oxidation/reduction, but instead is product release in going from the oxidized (MoVI) to reduced (MoIV) state of the molybdenum center as discussed below

(12, 69). Bound product is formed upon the initial two-electron reduction of the molybdenum (noted Ered·P where the molybdenum is MoIV) (66). It is specifically the dissociation of uric acid (hydroxylated xanthine) from the molybdenum center, with the regeneration of the Mo-OH ligand from solvent water, that is rate-limiting in catalytic hydroxylation by the enzyme (69).

This mechanism of hydroxylation presumably holds for hydroxylation of hypoxanthine as well as xanthine, as these differ only in the particular carbon being hydroxylated (12). The common nature of this reaction in and of itself is worth noting, as the enzyme is known to hydroxylate a wide variety of aromatic heterocycles, and even simple aldehydes (17, 31).

With the establishment of the catalytic mechanism of XOR, more recent attention has turned to the role of specific active site residues in the mechanism of the enzyme. These studies have utilized steady-state and rapid-reaction kinetics to probe the effect(s) of active site residue mutations on catalysis at the molybdenum site, which can be considered as reductive (MoVI to MoIV) and oxidative (MoIV to MoVI) half-reactions.

21 1.5.2 The oxidative and reductive half-reactions of XOR catalysis

Kinetic constants can give a wealth of useful information on enzyme catalysis and mechanism. In the steady-state, obtaining kcat, Km, and kcat/Km gives useful information on overall enzyme turnover at low and high substrate concentrations (70).

By following the reaction aerobically in the steady-state one can assess the effect(s) on overall catalysis, for instance assessing a change in rate-limiting step from one of product release to one involving the loss of a rate-accelerating contribution by an active site residue.

Obtaining kred, Kd, and kred/Kd for the reductive half-reaction with fully oxidized enzyme provides complementary information on substrate binding as enzyme meets substrate. This is especially true and useful in the case of xanthine oxidoreductase, whose molybdenum center cycles from MoVI + S in the fully oxidized enzyme to MoIV·P or MoIV + P in the fully reduced state, providing an observable reductive half-reaction for the enzyme wherein substrate is oxidized. Experimental determination of the aforementioned constants for wild-type enzyme and active site mutants can provide both a qualitative and quantitative assessment of the functional role of a particular active site residue in catalysis. By examining the reductive half- reaction under anaerobic conditions, whereby electrons are introduced into the enzyme (while there is no terminal electron acceptor to reoxidize the enzyme and the

MoIV state) one can investigate a residue’s kinetic and thermodynamic contributions to substrate binding and transition state stabilization. Changes in kred or Kd can be converted to thermodynamic contributions by a particular residue, and that residue’s role in substrate binding and transition state stabilization can be quantitatively assessed and qualitatively inferred. Such studies have been carried out for XOR using

UV-Visible absorbance spectroscopy, following the formation of uric acid at 295 nm 22 -1 -1 (ε295 = 9600 M cm ) or reduction/oxidation of the molybdenum center at 450 nm

-1 -1 (bovine enzyme, ε450 = 37.8 mM cm ) or 465 nm (Rhodobacter capsulatus enzyme,

-1 -1 ε465 = 31.6 mM cm ) (54, 71, 72). The UV-Visible spectra of oxidized and reduced enzyme are shown in Figure 1.5.

23

Figure 1.5. Reduction of oxidized xanthine oxidoreductase by sodium dithionite at pH 10.0, showing the decrease in absorbance from the peak at 450 nm. This absorbance change is due to reduction/oxidation of the flavin and molybdenum centers of the enzyme. Taken from a 1985 study by Hille et al. (71).

24 As mentioned above, kinetic studies of the reductive and oxidative half- reactions have proven invaluable in probing the role(s) of specific active site residues in XOR, and have been used extensively in the current study.

1.5.3 Active site residues in the molybdenum site

In addition to the structures of bovine XOR, the crystal structure of reduced xanthine dehydrogenase from Rhodobacter capsulatus has been determined in complex with the inhibitor alloxanthine (73). This compound is the product of enzyme action on the inhibitor allopurinol, where its tight binding forms the basis of effective enzyme inhibition. Alloxanthine coordinates directly to the reduced molybdenum atom by replacing the planar Mo-OH of native enzyme, and interacts with specific active site residues, which may be assumed to interact with bound substrate and/or product during catalysis. The structure of XDH with bound alloxanthine is shown in Figure 1.6.

25

(A)

(B)

Figure 1.6. The alloxanthine-inhibited active site of xanthine dehydrogenase from Rhodobacter capsulatus. Panel (B) is rotated clockwise 90 degrees relative to panel (A). Numbering is for the R. capsulatus enzyme, with bovine numbering in parentheses. (modified from 73, PDB 1JRP)

26 Glu 730 (Glu 1261 in the bovine enzyme) is universally conserved among the xanthine oxidizing molybdenum hydroxylases, and was proposed to be the active site base in the base-assisted nucleophilic attack on the carbon center of substrate (73).

Lending further support to this base-assisted mechanism, a pH-dependent kinetic study by Choi et al. demonstrated that the steady-state and reductive half-reaction kinetics of the enzyme exhibits a bell-shaped profile with pKas of 6.6 and 7.4. These were assignable respectively to an active site base and the first ionization of substrate

(74). Leimkühler et al. mutated Glu 730 to Ala and found that, whereas the reaction of wild-type XDH with xanthine goes to completion in a matter of 100 ms, the E730A mutant showed no appreciable activity in an overnight incubation with 100 µM xanthine. The results indicated a decrease of at least 107 of the catalytic effectiveness with the loss of Glu 730 (72). Furthermore, conducting the reaction at a pH of 10.0

(where the Mo-OH species spontaneously deprotonates) resulted in a partial recovery of activity by the E730A mutant. These results clearly support not only the base- assisted nucleophilic attack mechanism, but also that Glu 730 is the active site base.

Another glutamate conserved among xanthine oxidoreductases and implicated in catalysis based on the alloxanthine-inhibited structure is Glu 232 (Glu 802 in the bovine enzyme), which is positioned to interact with the six-membered ring of the inhibitor. An E232A mutant showed a 10-fold lower kred and 10-fold higher Kd relative to wild-type enzyme, indicating that this residue contributes approximately 1-

3 kcal/mol each to transition state stabilization and substrate binding (72). One possibility for the function of Glu 232 in transition state stabilization involves the tautomerization of substrate. A study in 1997 by P. Ilich and R. Hille suggested that for xanthine, N3→N9 tautomerization occurs (from the pyrimidine to the imidazole subnucleus) in the mechanism of hydroxylation at C8 (75). Such transfer was shown 27 to lower the enthalpy change in forming the transition state with hydroxylated xanthine coordinated to the molybdenum. As Glu 232 is positioned closest to N9 in one suggested orientation of xanthine, it may be facilitating such a process.

A third residue that appeared to be interacting with the inhibitor was Arg 310

(Arg 880 in the bovine enzyme) (73). This residue is conserved among xanthine oxidoreductases. That this arginine might be able to interact with substrates in a meaningful way was particularly interesting, given its position some 10 Å away from the molybdenum and carbon center being hydroxylated. This residue is a major focus of the present work.

Other residues that appeared potentially important to catalysis were Gln

197 (Gln 767 in the bovine enzyme), which hydrogen bonds to the apical oxygen in the molybdenum coordination sphere, and the Phe residues between which substrate/inhibitor binds. Phe 459 (Phe 1009) is edge-on and Phe 344 (914) face-on in respect to alloxanthine, indicating that π-stacking and hydrophobic interactions play a key role in substrate binding and orientation.

Taken together, there appear to be at least six residues in the active site of

XOR that are likely to be catalytically important. Foremost among these, the role of

Glu 730 (Glu 1261) has been suggested to be that of the active site base in assisting nucleophilic attack at the beginning of catalysis as in Figure 1.4 (72, 74, 76). The role(s) for the other five residues have yet to be conclusively shown, and in particular

Glu 232, Gln 197, and Arg 310 warrant further investigation in the R. capsulatus enzyme. The bovine enzyme, with its high yield preparation, provides a convenient system for structural work such as X-ray crystallography and for mechanistic/inhibition studies. In a clinical and pharmaceutical sense, further

28 investigations of the nature of the molybdenum site in the bovine enzyme are also warranted.

1.5.4 The pterin cofactor, iron-sulfur clusters, and the flavin site

Substrate hydroxylation occurs at the molybdenum site in XOR, after which electrons must be passed out of the molybdenum in order for catalysis to continue.

The relationship between the molybdenum cofactor and the other redox-active centers in the enzyme are shown in Figure 1.7.

29 (A)

(B)

Figure 1.7. Relative positions of redox-active centers in bovine XOR. Panel (A) shows the position of all four cofactors in one monomer of XOR. From bottom, the molybdopterin, [2Fe2S]II, [2Fe2S]I, and the FAD. Panel (B) is a closer view of the molybdopterin, with the pterin ring projecting back toward the iron-sulfur clusters. The positions of the sulfur and an oxygen ligand were switched from that in PDB code 1FIQ, given recent evidence that it is an oxo rather than sulfido group that occupies the apical position. (from reference 17)

30

As shown in Figure 1.7, the means by which electron transfer out of the molybdenum occurs involves the pterin ring coordinated by its two thiolates to the Mo atom. As mentioned, this additional ligand to the molybdenum center is widespread and apparently essential for the biological functionality of molybdenum (6, 12). The role of the pterin is in electron transfer out of the molybdenum and to the first of two [2Fe-

2S] clusters (77-79). This electron transfer has been discussed as occurring through the σ bonds (rather than the conjugated π-bonding system) of the pterin molecule, as the π-orbitals of the pterin are aligned out of plane with the dxy orbital of the Mo coordination sphere (80, 81).

Electron transfer proceeds via the pterin to the proximal [2Fe-2S] cluster, which has been designated Fe/S II by EPR . Although the pterin intervenes between the molybdenum and proximal iron-sulfur cluster, electron transfer is not exceptionally fast to Fe/S II (approximately 8,000 s-1). The distance here is approximately 12-14 Å, an allowable distance at < 14 Å for effective tunneling (65).

Tunneling has also been proposed as the means of transfer to the second [2Fe-2S] cluster, over approximately the same distance (78).

From the distal cluster (Fe/S I), electrons are transferred individually to the

FAD site. Once the FAD is sequentially reduced first to the semiquinone (FADH·) and then hydroquinone (FADH2) form, it is ready to reduce a terminal electron acceptor in transferring electrons out of the enzyme (82, 83). As indicated above, the preferred electron acceptor depends on the form of the enzyme at this site; either

+ NAD (generating usable NADH) or O2 (generating superoxide or peroxide) will be the preferred oxidant.

31 1.6 Preparation and Isolation of Xanthine Oxidoreductase in the Present Studies

1.6.1 Bovine xanthine oxidase and xanthine dehydrogenase

In this work, XOR was isolated as xanthine oxidase from bovine milk as follows, following to the protocol by Massey et al. (54). Fresh, unpasteurized bovine milk was obtained in one or two 20 L carboys, to which the following was added to each 20 liter portion in the order listed: 50 ml of 0.3 M EDTA pH 7.0; 20 ml of 1.0 M salicylate pH 7.0; 6.0 g cysteine·HCl -or- 1 mM DTT; 20 ml 0.1 M PMSF (in EtOH);

316 g sodium bicarbonate; 31.6 g pancreatin. The mixture was allowed to stir for 30 minutes then stirring was stopped, stirring rod removed, and the mixture allowed to sit overnight at 4 oC.

The next day, approximately 4 kg of (NH4)2SO4 was added and the mixture(s) stirred for 1 hour. 3.33 L of cold 1-butanol (at -20 oC) was then added over 5 minutes with vigorous stirring, then slower stirring for 30 minutes. The stirring rods were then removed and the solution allowed to sit overnight at 4 oC.

The next morning, the solution consisted of 3 layers: a topmost lipid phase, an intermediate lipid phase, and a lower brown aqueous phase that contained the XO.

This phase was transferred via the spigot on each carboy into a 20 L container (e.g. carboy with top removed). To each respective carboy’s contribution, 160 g/L ammonium sulfate was added, stirred for 30 minutes, and left to stand for 1 hour.

Crude XO coagulated and floated to the surface, and the lower green-tinted aqueous phase was discarded via spigot. The light brown precipitate was combined and the aggregate of orange-brown precipitant divided among centrifuge bottles, then centrifuged at 10,000 rpm for 20 minutes. This yielded an intermediate solid layer between and upper and lower liquid layer. Both liquid layers were removed by introducing a hole in the solid layer and pouring off the liquids. The solid was then 32 resuspended in 300 ml 0.1 M potassium phosphate, 0.3 mM EDTA, 1 mM salicylate,

10 µM PMSF, pH 6.0 and dialyzed against 20 liters of the resuspension buffer over three days, with buffer changes every 12 hours. The dialysis solution was stirred slowly following the first 12-hour interval.

Next, a hydroxyapatite (HAP) column was equilibrated with 0.1 M potassium phosphate, 1 mM salicylate, 10 µM PMSF, pH 6.0. The brown dialysate was centrifuged at 10,000 rpm for 30 minutes to remove any precipitant. The resulting supernatant was divided into two fractions, to be run separately on the HAP column.

Each fraction was loaded in turn onto the equilibrated column and washed with 0.5 to

1.0 L of the dialysis buffer above until the eluate was colorless. XO was eluted with a gradient of the dialysis buffer versus 0.1 M sodium pyrophosphate, 0.3 mM EDTA, pH 8.5.

The fractions coming off of the HAP column that were light/orange-brown in color were analyzed by spectrophotometry for a shoulder at 420 nm. Fractions that showed no contamination were pooled and concentrated separately from those showing LPO contamination. The latter fractions were concentrated separately and run back down the HAP column.

Concentrated fractions from above were loaded onto an S-300 column (Bio-

Rad Laboratories, Hercules, CA) equilibrated with 1 L of 0.1 M sodium pyrophosphate, 0.3 mM EDTA, pH 8.5 and eluted with the equilibration buffer in 2-3 ml fractions. These were assessed by spectrophotometry for the absorbance ratio of

A276 nm/A450 nm. Those fractions with a ratio of < 6 were concentrated, 500 mM salicylate added, and then frozen in N2(l). Those with ratio > 6 were concentrated and run back down the S-300. Activity of the isolated enzyme was assessed prior to addition of salicylate as described below. 33

XOR was isolated as XDH (rather than XO as above) from bovine milk by a slight modification of the above procedure, following the protocol of Hunt and

Massey (69). The essential steps required omission of pancreatin and the initial incubation with and addition of 2.5-5.0 mM DTT in all buffers and solutions throughout the entire procedure. Care was taken to avoid proteases in all solutions.

With these modifications, the enzyme was isolated as nearly 75% XDH, which stored

o in N2(l). Continuous addition of 1-5 mM DTT was required to solutions at 4 C, which retains its activity for < 2 weeks at this temperature.

1.6.2 Xanthine dehydrogenase from Rhodobacter capsulatus

Wild-type and mutant forms of XDH from R. capsulatus were generously provided by Dr. Silke Leimkühler from the University of Potsdam, Germany, prepared according to published protocols (72, 84). The xdhABC genes, which encode R. capsulatus XDH, were cloned into the NdeI and HindIII sites of the pTrcHis expression vector to create a resulting plasmid designated pSL207 for transformation and expression in the E. coli TP1000 mutant strain. For site-directed mutagenesis, the transformer site-directed mutagenesis kit (Clontech) was used (84).

1.6.3 Activity-to-flavin determination and the percentage of active enzyme

The percentage of active XO, having an intact (pterin)MoVIOS(OH) center, in a given preparation was determined as follows. Isolated XO was passed down an equilibrated Sephadex G-25 column in 0.1 M sodium pyrophosphate at pH 8.5 and eluted with the same buffer. An aliquot of the enzyme was diluted such that A450 nm =

0.3 (i.e. 8 µM XO). In a 4.0 ml quartz cuvette, 10 µl enzyme solution was added to 34 3.0 ml of the buffer, and used to obtain a blank spectrum. Then 10 µl of 33.3 mM xanthine (approximately 110 µM) was added, the cuvette solution mixed well, and the reaction monitored at 295 nm per second for 120 seconds. The ∆A295/minute was recorded, and an average of three assays was obtained.

The activity-to-flavin ratio (AFR) was calculated as [(∆A295/min) x 300 ÷ A450 nm] (85). Fully active solutions of bovine XO have an AFR of 210, and thus division of the above result by 210 gives the active fraction of a particular solution.

A second method of determining the fraction of a given XOR preparation that contains an active molybdenum center (as opposed to the inactive “desulfo” form) was also used in the present work, based on the fact that while xanthine can reduce only the functional form of the enzyme, sodium dithionite reduces both the functional and desulfo enzyme. Anaerobic reduction of the enzyme solution by xanthine, recording the change at 450 nm, was compared to any further decrease at A450nm upon addition of excess dithionite. Comparison of these values with the concentration of the enzyme in solution gives the percentage of active XOR (86, 87).

1.6.4 Inactive and inactivating XOR

In any given preparation of the XOR, a percentage of the isolated enzyme is inactive. In 1952, D.B. Morell first hypothesized the existence of inactive enzyme in preparations, and work in 1970 by Hart et al. suggested the existence of inactive enzyme that lacked molybdenum as well as an inactive form that had a full complement of redox-active centers (88, 89). Subsequent work showed that in addition to the demolybdo- form of XOR, some percentage of the enzyme is prepared in a desulfo- form where the Mo=S of the molybdenum center has been replaced by a 35 second Mo=O (90). Reactivation of desulfo-XOR can be accomplished by incubation with sulfide and dithionite (91).

The enzyme may also be made inactive (desulfo) by treatment with cyanide

(92). This method leads to the replacement by oxygen of the planar sulfido group, which is lost as thiocyanate, while the other redox-active centers in the enzyme remain functionally intact. Both the thiocyanate and any excess cyanide can be removed by a Sephadex G-25 or other size-exclusion column.

36

CHAPTER 2

THE ROLE OF ARG 310 IN THE ACTIVE SITE OF XANTHINE

DEHYDROGENASE FROM RHODOBACTER CAPSULATUS

2.1 Introduction

Based on the crystal structure of the alloxanthine-inhibited form of XDH from

R. capsulatus, several residues in the molybdenum-containing active site of XOR were identified as likely playing a role in substrate binding and catalysis (73). The proposed roles for these amino acids range from acting as the critical active site base initiating the nucleophilic attack on substrate (bovine Glu 1261) to potential interactions with the apical oxo ligand in the molybdenum coordination sphere

(bovine Gln 767). Among these residues, Arg 310 (Arg 880 in the bovine enzyme,

Arg 881 in the human enzyme) is unique by virtue of its considerable distance

(approximately 7-8 Å) from the site of substrate hydroxylation, as shown in Figure

1.6. It is also interesting that this residue is highly conserved among the molybdenum hydroxylases that use xanthine as a substrate, but not in the aldehyde-hydroxylating enzymes. Located approximately 10 Å from the molybdenum center, this arginine is too far removed to participate directly in catalysis, but is close enough for electrostatic

(hydrogen-bonding) interactions with the alloxanthine molecule in the R. capsulatus

XDH structure (73).

37 Based on these observations, it is possible that Arg 310 may interact with substrates in a manner that somehow facilitates catalysis, e.g. by stabilizing negative charge accumulation on the substrate heterocycle in the course of nucleophilic attack on a substrate as shown in Figure 2.1.

38

Figure 2.1 Proposed method of transition state stabilization by Arg 880 in xanthine oxidoreductase (Arg 310 in xanthine dehydrogenase from Rhodobacter capsulatus). (adapted from reference 93)

39 In the alternative orientation of substrate to that shown in Figure 2.1, Arg 880 would be interacting with the oxo group at the C2 position, which does not receive negative charge density by resonance as well as the C6=O. Thus the proposed role for Arg

310/880/881 in stabilizing negative charge density may lend specificity to the orientation of certain substrates in the XOR active site.

In order to characterize the nature of this residue’s interaction(s) with substrate molecules and its contribution to catalysis, we have conducted rapid-reaction kinetic experiments with a homologous series of six substrates. The substrate molecules in this study were chosen based on their functional groups at the C6 position, where proposed interactions with the arginine residue preferentially occur. If our hypothesis for the role of this active site arginine is correct, then transition state stabilization (and thus rate acceleration) occurs by electrostatic interactions at this C6 position of substrate. Those substrates that can direct negative charge accumulation out to a receptive C6 functional group can utilize this transition state stabilization, and will thus be turned over faster and more efficiently as the activation barrier is decreased in obtaining the slightly-lower energy transition state. Substrates that lack this ability to shunt accumulating negative charge to C6 will be manifestly poorer substrates with a larger energy barrier to reaction. The six substrates selected for use in this study are shown in Figure 2.2.

40

Figure 2.2 Homologous series of six substrates used in the current study. Numbering of the heterocyle is shown for xanthine.

41 Of these, 2-hydroxy-6-methylpurine and 2,6-diaminopurine have functional groups at

C6 that will not accept negative charge density, and thus cannot interact with the arginine as proposed here. Interestingly, these two substrates are known to be turned over more slowly (are less effective substrates) than the other four (66, 93).

By utilizing site-directed mutagenesis of the enzyme from R. capsulatus, we were able to generate arginine-to-lysine (R310K) and arginine-to-methionine (R310M) variants that allowed us to directly examine the role of this active site residue. The conservative mutation creating the R310K mutant enzyme should preserve the discrimination of “good” and “poor” substrates in this series. Mutation to methionine with the loss of positive charge and ability to stabilize the catalytic transition state should reduce the ability of the enzyme to discriminate between the “good” and

“poor” substrates, as shown by the rate of turnover (or reduction). Rapid-reaction experiments, following the reductive half-reaction of the enzyme’s catalytic sequence as substrate reduces the enzyme under anaerobic conditions, allows us to quantify the effect that the presence or absence of Arg 310 has on substrate binding and transition state stabilization.

2.2 Materials and Methods

2.2.1 Preparation of wild-type and mutant xanthine dehydrogenase from Rhodobacter capsulatus

Xanthine dehydrogenase from R. capsulatus was generously supplied by Dr.

Silke Leimkühler (University of Potsdam, Germany), having been prepared according

Leimkühler et al. and as mentioned in Chapter 1 (72, 73). The mutant enzyme was created using PCR mutagenesis and prepared following the same protocol as for wild-

- type (72). Enzyme concentrations were determined at 465 nm using ε465 = 31.6 mM 42 1cm-1 and activity of the preparations were assessed as in previous work (72, 84).

Enzyme was shipped on dry ice and stored at -70oC until use.

2.2.2 Rapid-reaction experiments

Rapid-reaction experiments were conducted using an Applied Photophysics

SX-18MV kinetic spectrophotometer with a 1 cm pathlength (Applied Photophysics,

England). The reductive half-reaction was examined by following reduction of the enzyme at 465 nm under anaerobic conditions as described previously (84). In brief, experiments were conducted using a sealed tonometer containing the enzyme solution, made anaerobic by the introduction of Ar (g) for 60 minutes with evacuation of the tonometer by vacuum at 15 minute intervals. All other solutions were made anaerobic by bubbling argon through the liquid for 30 minutes, then quickly transferring the solution to the sealed spectrophotometer chambers. All chambers of the spectrophotometer were rinsed with anaerobic buffer prior to addition of the experimental solutions.

Assays were conducted at pH 7.8 in 20 mM Tris, 0.2 mM EDTA, at 4°C. The concentration of enzyme in a given experiment was 12 µM, mixed with an equal volume of substrate solution to give 6-8 respective concentrations from 20 µM to 2.0 mM.

2.2.3 Data acquisition and analysis

Rapid-reaction transients for the reduction of the enzyme were acquired under anaerobic conditions on an Applied Photophysics SX-18MV kinetic spectrophotometer with a 1 cm observation path length, in triplicate at ranges of 10 –

3600 seconds and datapoints every 1.26 seconds. Transients were fit using the

43 program Pro-K (Applied Photophysics) to obtain the best fit values for that part of the transient comprising the majority of the absorbance change. Exponential fitting of the spectra at all wavelengths monitored (i.e. a global fit) was used. For those experiments that showed a dependence of kobs on substrate concentration, a plot was constructed to obtain kred (the rate constant for reduction, equal to kobs/[enzyme]rxn) and Kd (the dissociation constant, a reflection of substrate binding). For those experiments that did not demonstrate a dependence on substrate concentration, the average of kobs values was taken to be kred.

2.3 Results and Discussion

The kinetic constants for the reductive half-reaction with the series of substrates examined are given in Table 2.1.

44

wt R310K R310M

-1 -1 -1 Substrate kred (s ) Kd kred/Kd kred (s ) Kd kred/Kd kred (s ) Kd kred/Kd

(µM) (x 106) (µM) (x 106) (µM) (x 106)

xanthine (67.3) (33.6) (2.00) 1.63 ± 44.7 ± 0.0365 0.0022 ND ND 29.3 ± 24.9 ± 1.18 0.05 6.0 ± 1.0 4.0 0.0005 2-thioxanthine 14.5 ± 40.9 ± 0.354 0.037 ± 46.3 ± 0.0007 0.0017 ND ND 0.8 10.0 0.006 40.0 99 ± 0.0005 6-thioxanthine 22.4 ± 24.1 ± 0.928 1.26 ± 9.4 ± 0.134 0.0054 15.05 ± 0.0003 0.7 4.0 0.02 1.0 ± 3.0 59 0.0002 1-methylxanthine 2.65 ± 77.8 ± 0.0341 0.014 ± ND ND 0.00547 ND ND 0.07 8.0 0.005 ± 0.002 2,6-diaminopurine 0.0075 ND ND ± 0.001 ND ND 0.016 ± 31.9 ± 0.0005 0.00640 0.001 10.0 05 ± 0.002 2-hydroxy-6- 0.22 ± ND ND 0.837 ± 10.7 ± 0.0785 0.0018 ND ND 0.02 * 0.04 5.0 ± methylpurine 9.5 ± 0.0176 0.0006 0.168 ± 1.0 0.775 ± 27.3 ± 0.0284 0.003 0.04 7.4 ** * fit using a single exponential ** fit for a double exponential process

Table 2.1 Kinetic constants for wild-type, R310K and R310M mutant xanthine dehydrogenase from R. capsulatus with the series of 6 homologous substrates. Literature values for wild-type XDH with xanthine are given in parentheses. For those substrates that did not demonstrate concentration dependence of kobs, no value for Kd (and thus also no kred/Kd) could be determined. This is indicated by ‘ND’ for ‘not determined’ in the table. (table modified from reference 93)

45 As shown in Table 2.1, the substrates examined appear to fall into two distinct groups. The first group consists of xanthine, 2-thioxanthine, 6-thioxanthine, and 1- methylxanthine. These all are effective substrates for the wild-type enzyme, and all have functional groups at their C6 positions (oxo or sulfido, respectively) that can accept electron density with nucleophilic attack on C8 of their imidazole subnucleus.

The rate constants associated with the hydroxylation of this group by wild-type enzyme span an order of magnitude and are significantly affected by the conservative

R310K mutation (kred decreasing 10 to 100-fold). Most striking is the >1000-fold decrease in kred upon loss of the charged arginine in the R310M mutant.

The second group consists of 2,6-diaminopurine and 2-hydroxy-6- methylpurine, two substrates that lack functional groups capable of receiving negative charge density, and are both poor substrates for the wild-type enzyme. There is no change in the rate constants for turnover of these substrates with the R310K mutant; in fact there is an increase in kred. There is a 100-fold decline in kred with the R310M mutant, however, and this variant gives a rate constant for reduction that is nearly identical (of the same order of magnitude) to that of the four good substrates for wild- type XDH. The enzyme is no longer able to discriminate good from poor substrates with the loss of R310, and thus the R310M variant catalyzes the hydroxylation of any substrate at approximately the same slow rate as for any other.

There was a 4000-fold range of kred for these six substrates, a range that is reduced to fourfold with the loss of R310. The six substrates fall into two groups, consisting of four good and two poor, with wild-type enzyme, a distinction that is lost in the absence of R310. Without the charge complementation provided by this arginine in stabilizing the transition state for good substrates, XOR is no longer able to discriminate good from poor substrates. 46 From the 10,000-fold effect on kred, it is evident that R310 contributes approximately 4.5 kcal/mol toward transition state stabilization for xanthine, thereby providing a 2 x 104-fold increase in the catalytic rate. Substrates that are able to utilize this stabilization via their functional groups at C6 are good substrates, and are thus turned over by the R310M variant at rates comparable to those seen in the reaction of poor substrates with wild-type enzyme. Those substrates that are unable to utilize Arg 310 are poor substrates regardless of the presence or absence of this residue; they do not use it for transition state stabilization in the wild-type enzyme anyway and are much less affected by the mutation.

The above results suggest a role in substrate discrimination for Arg 310 (Arg

880/Arg 881) in xanthine oxidoreductase, providing a mechanism by which substrates can be classified as “good” and “poor”. A good substrate binds in the same orientation as xanthine in Figure 2.1 above, with its C6 functional groups directed at the arginine. This allows the negative charge accumulation at C6 in the transition state to be stabilized by charge complementation. Our results thus suggest that there is a preferred orientation of good substrates such as xanthine in the molybdenum- containing active site with respect to the arginine. Good substrates (e.g. xanthine) are positioned with their C6 functional groups directed toward Arg 880, in order to utilize its stabilizing contributions to the transition state. Poor substrates (e.g. 2-hydroxy-6- methylpurine) gain little or no advantage from this preferred orientation, and orient with the C6 position opposite this, projecting their C6 functional group into the hydrophobic pocket on that side of the active site. The suggested situation is shown in Figure 2.3.

47

Figure 2.3 Proposed orientations of xanthine (a “good” substrate) and 2-hydroxy-6-methylpurine (a “poor” substrate”) with respect to Arg 880 in the active site of bovine xanthine oxidoreductase.

48 With the accumulation of negative charge on xanthine as shown in Figure 2.1, the arginine is positioned in close proximity to the C6 position for stabilizing such negative charge density. The results of our kinetic study here support such a role for

Arg 310/880/881 in the determination of substrates which are effectively hydroxylated by XOR and those that are much less so.

It is worth briefly mentioning that in the course of our experiments, a small increase in absorbance at 465 nm was noted immediately following the initial sharp decrease associated with reduction of the enzyme. This small increase was never more than 5-10% of the total absorbance change at 465 nm, but nevertheless suggested the possibility of an intermediate with the mutant enzymes. We attempted to investigate this intermediate by examining the rapid-reaction kinetics with substrates at stoichiometric excess and sub-stoichiometric amounts in very short assays (e.g. 1 minute) and those over several hours, but no conclusive information was gained from these studies, and the transient phase was not investigated further.

49

CHAPTER 3

SUBSTRATE ORIENTATION IN XANTHINE OXIDASE FROM BOS TAURUS, CRYSTAL STRUCTURE WITH 2-HYDROXY-6-METHYLPURINE

3.1 Introduction

Our previous work in Chapter 2 provided insight into the possible role of Arg

310 in the reaction catalyzed by xanthine dehydrogenase from Rhodobacter capsulatus. This residue was found to contribute 4.5 kcal/mol to transition state stabilization via electrostatic interactions with substrates during catalysis, surprisingly high for a residue some 10 Å removed from the site of substrate hydroxylation (73,

93). The proposed interaction of this residue with the transition states suggests a means of discriminating good substrates from those that are more slowly hydroxylated. In our previous work, both 2-hydroxy-6-methylpurine (HMP) and 2,6- diaminopurine, which are poor substrates with wild-type XOR, were shown to be relatively unaffected by loss of Arg 310. This is expected if our hypothesis of transition state stabilization by Arg 310 holds true, as these two substrates are unable to utilize this residue with their respective functional groups at the C6 position. The other four good substrates examined were markedly affected by mutation of this arginine, with catalytic rate constants reduced by over three orders of magnitude (93).

These results lend insight not only into the role of Arg 310 in substrate discrimination, but also into the role of substrate orientation in catalysis (see Figure

2.3). Good substrates may orient with their C6 functional groups directed toward this 50 residue, in order to direct negative charge density out to this group and thereby utilize electrostatic interactions with Arg 310 for transition state stabilization. Poor substrates, unable to localize negative charge out to their C6 position, will orient with this group projecting opposite that preferred for good substrates; their group at C6 will project up into a hydrophobic pocket alongside Glu 232 (Figure 1.6).

To test the putative importance of substrate orientation in the active site and more directly confirm the role of Arg 310/880, we sought to obtain the crystal structure of bovine xanthine oxidase with the slow substrate HMP bound in the molybdenum-containing active site(s). We were successful in obtaining at 2.3 angstroms this structure with HMP in both active sites of the homodimer and, importantly, observe a catalytic intermediate of bound product in one of the two sites

(94).

3.2 Materials and Methods

3.2.1 Materials

Magnetic bases for crystallography mounts were obtained from MAR-USA

(now Rayonix, Evanston, IL). Mounting cryoloops, pins, magnetic cryovials, and various molecular weight PEG solutions were obtained from Hampton Research

(Aliso Viejo, CA). All other reagents were obtained from Sigma-Aldrich (St. Louis,

MO) or Thermo-Fisher Scientific (Waltham, MA) and used without further purification.

3.2.2 Preparation and isolation of bovine xanthine oxidase

Previous attempts at obtaining diffraction-quality crystals of bovine xanthine oxidase in our own laboratory had been unsuccessful. One potential problem that had 51 been suggested involved the heterogeneity of the source protein, as typical preparations of bovine XO utilize milk from multiple animals in a given dairy herd.

Given the requirements of protein crystallography for highly pure, homogeneous samples, this problem seemed plausible and was addressed by preparation of bovine xanthine oxidase from a single animal. 20 liters of fresh, unpasteurized bovine milk was obtained from a single cow, and the enzyme was prepared according to the method in Chapter 1.

3.2.3 Crystal growth, diffraction, and data acquisition

Crystals were grown via the batch method in 24-well trays (94). Microbridges

(Hampton Research) were placed in each well to hold the well solutions. Initial conditions that yielded diffraction-quality crystals were modified from previous work by B.T. Eger et al. (52). First attempts at crystallization produced stellate aggregates of protein, suggesting an overabundance of nucleation sites. Review of the literature led to further modifications being made to our methods, the most important being the pre-mixing of enzyme and precipitant solutions in an Eppendorf tube with very gentle agitation of less than 5 seconds (95). After this, the well solution was transferred to the microbridge well and the tray well itself was sealed.

The final batch solutions that yielded the best crystals contained 20 µl of enzyme solution mixed with 10 µl of the following precipitant solution: 12-14%

PEG8000, 0.1 M potassium phosphate at initial pH 6.5, and 0.2 mM EDTA. The enzyme solution contained 10 mg/ml single source XO in 40 mM TrisHCl initial pH

7.8, 20 mM pyrophosphate initial pH 8.5, 0.2 mM EDTA, and 5 mM DTT added last.

The enzyme solution was allowed to sit on ice for 1 hour following the addition of

DTT before construction of the batch solutions. The final pH of the well solutions 52 was approximately 7.25. Crystals were observed to grow after 2-3 days at 25oC in darkness. Substituting PEG8000 for PEG4000, while lowering the PEG concentration, was an important modification to the previous conditions (94). Our best XO crystals were in the form of rectangular plates, ranging from 0.1 mm to 1.0 mm in their longest dimension.

Standard cryoprotectants such as ethylene glycol or glycerol could not be used with our crystals of XO, as these intervene to block access to the molybdenum- containing active sites. We therefore utilized low molecular weight PEG solutions, settling on 40-45% PEG200 as the best cryoprotectant. This solution was made up with equal proportions of all other reagents (excepting of course the enzyme) in the batch solutions. Cryoprotectant was introduced by micropipette, using 2 µl exchanges until the original well volume was exceeded.

2-hydroxy-6-methylpurine was introduced into the cryoprotectant solution by soaking as follows. A 10 mM solution of HMP was made in the same cryoprotectant solution. This HMP stock was diluted 1:10 in the final well volume containing the

XO crystals, thus giving 1 mM HMP. Crystals were allowed to incubate in the presence of HMP for 1-5 minutes, mounted and flash frozen on N2 (l), and stored in cryovials under N2 (l) until mounted for data acquisition.

Initial crystals were screened on the home rotating anode source at the Ohio

State University. An example of our screens is shown in Figure 3.1.

53

Figure 3.1. Sample screenshot of our initial crystals on the rotating anode home X-ray source. As shown, XO crystals exhibited strong diffraction, but the power of the home source did not allow high enough resolution data to be obtained on our large protein.

54 The best datasets that could be acquired from the home source were at 2.9 Å, and thus we required a more powerful X-ray source (i.e. synchrotron radiation). Final diffraction data were collected at Argonne National Laboratory (Argonne, IL) on the

SGX Pharmaceuticals, Inc. beamline, using a wavelength of 0.9793 Å and a

MARCCD 165 detector. Data sets here were collected and processed at better than

2.3 Å resolution.

Use of the Advanced Photon Source was supported by the U. S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No.

DE-AC02-06CH11357. Use of the SGX Collaborative Access Team (SGX-CAT) beamline facilities at Sector 31 of the Advanced Photon Source was provided by SGX

Pharmaceuticals, Inc., who constructed and operates the facility.

3.3.4 Data processing and refinement

Data were processed at the synchrotron using the MOSFLM package of the

CCP4 program suite (96, 97). The structure of the enzyme was determined by molecular replacement using the MOLREP package of CCP4, with protein data bank file 1FIQ as the search model (17). Following the initial rigid body refinement in

REFMAC of the CCP4 suite, the molybdenum center, including the pyranopterin cofactor, was constructed using HETATM records, and the structure was refined using the restrained refinement protocol in REFMAC (98 – 102). Initial efforts to impose non-crystallographic symmetry on the two monomers of the homodimeric enzyme in the asymmetric unit were discarded after it was found that omission of these NCS restraints yielded a lower Rfree. Geometric restraints based on the weighting term for the reflections in the dataset were also adjusted to minimize Rfree

55 (94). The HMP molecule was constructed using the PRODRG2 server (103), and the resulting PDB file was built into the corresponding positive electron density observed in the XO active site using COOT (104). Water molecules were added using

REFMAC and COOT, and subsequently removed where in excess. The final R- factors for the XO-HMP structure were an Rcryst of 19.4 and an Rfree of 26.3 (Table

3.1). Images were rendered using PyMol (105).

3.3 Results and Discussion

3.3.1 Overall structure of xanthine oxidase with 2-hydroxy-6-methylpurine

Structure factors and refinement statistics are shown in Table 3.1 for structure

3B9J in the protein data bank (94).

56

Statistic XO with bound HMP PDB Code 3B9J Spacegroup P21 Resolution 33.6 – 2.3 Å Wavelength (Å) 0.9793 Unique Reflections (test set) 114,520 (5785) Completeness % (highest resolution shell 2.36-2.30 Å) 96.3 (90.5) I/σ (highest resolution shell) 13.5 (1.8) Rcryst (highest resolution shell) 19.4 (25.1) Rfree (highest resolution shell) 26.3 (32.5) Ramachandran Statistics (%) 87.2, 11.0, 1.0, 0.7 Mean coordinate error based on free R value (Å) 0.260 Mean coordinate error based on maximum likelihood 0.196 (Å) Rmsd bond length (Å) 0.017 Rmsd bond angles (o) 1.8 Average B factor (A2) 22.7 Number of non-hydrogen atoms in refinement 19,852 Number of Waters 861

Table 3.1. Structure factors and refinement statistics for the XO-HMP structure. Ramachandran statistics indicate the percentage of residues in the most favored, additionally allowed, generously allowed, and disallowed regions of the Ramachandran diagram as defined by the program PROCHECK (108). (taken from 94)

57 XO-HMP crystals were monoclinic in spacegroup P21. There was one XO homodimer in the asymmetric unit, with unit cell dimensions of a = 133.2 Å, b = 73.8

Å, and c = 146.5 Å with angles of 90, 98.9, and 90 degrees. The space group for our crystals differs from that seen in previous work with XO, although there is recent precedence for a P21 spacegroup in a structure by Yamaguchi and coworkers (106).

Another difference from previous structures was our observation of additional density at the C-terminus in a chain of one monomer. Thus residues 1316 – 1328 that were previously disordered in the search model were added to fit ordered electron density in one monomer, as was residue 1292 (glutamate) which was also previously disordered. All redox-active centers were accounted for by omit density in the final structure, and were positioned at comparable distances relative to previous structures

(17, 106). The rms displacements for the alpha-carbons of each monomer in our final

XO-HMP structure relative to that of 1FIQ were 0.345 and 0.340 Å, respectively (94).

Figure 3.2 shows the dimer of our structure here, with cofactors as space-filling models in the stick-rendered protein.

58

Figure 3.2 Overall structure of XO with HMP, showing relative cofactor locations. Cofactors are presented as spheres in the stick-rendered protein. From bottom in each monomer: molybopterin and molybdenum cofactor, 2Fe2S cluster, 2Fe2S cluster, flavin adenine dinucleotide.

59 3.3.2 2-hydroxy-6-methylpurine in the active sites of xanthine oxidase

The HMP molecule was fit into each active site of the unit cell to account for positive density in the initial structure following restrained refinement, as well as into omit maps constructed from the final refined structure. The latter maps were used to construct Figure 3.3.

60

Figure 3.3. Omit maps overlaid with the final model of the XO-HMP structure. (PDB code 3B9J) Panels (A) and (B) showing 2Fo-Fc maps of HMP in the two respective active sites of XOR, contoured at +1.0σ. Panels (C) and (D) show Fo-Fc maps of the two active sites contoured at +2.5σ. (modified from reference 94)

61 As shown in Figure 3.3, the HMP molecules are oriented as predicted from our earlier kinetic study in both active sites of xanthine oxidoreductase. The methyl groups at the C6 position of substrate are oriented opposite the side of the pyrimidine subnucleus from that facing Arg 880, such that it is the C2=O rather than the C6 position that is closest to the arginine. The C6-CH3 thus projects into a hydrophobic pocket beyond Glu 802. In the final refined structure, the HMP molecules are observed to be stacking in face-on and edge-on interactions with Phe 914 and Phe

1009, respectively, as seen earlier in the structure of XDH with alloxanthine (73).

The C8 positions of the HMP molecules are positioned 2.2 Å and 1.6 Å from the equatorial oxygen of the molybdenum center in the two respective active sites of the dimer.

It is evident that the two active sites in the asymmetric unit reflect two different points in the catalytic sequence of XOR. Whereas the first site in panels (A) and (C) show a Michaelis complex with bound substrate, in the second site shown in panels (B) and (D) there is obvious electron density between the C8 of substrate and the Mo-OH oxygen, thus indicating that catalysis has proceeded to the formation of bound product as Mo-OR. The Mo-S distances in both active sites are 2.0 and 2.1 Å

(with a mean coordinate error of 0.2 Å), suggesting a Mo=S bond and a MoV oxidation state for the metal, rather than MoIV with a Mo-SH ligand and its longer bond length. We conclude that we observe a MoVI state in one active site prior to the onset of catalysis, and that the second site is a true catalytic intermediate in the MoV state. This conclusion is made with consideration of the known difficulties in determining metal-ligand distances that are associated with Fourier truncation artifacts in the vicinity of a metal as heavy as molybdenum (78). This consideration aside, however, under reaction conditions very similar to those explored here catalysis 62 by XOR with HMP results in the accumulation of large amounts of the EPR active

MoV state (66, 107).

The C2=O of each HMP is positioned 3.0 ± 0.2 and 3.1 ± 0.2 Å from the closest point of Arg 880, certainly close enough for hydrogen bonding. The orientation of HMP in both active sites of wild-type XO is that predicted by our previous kinetic work, and supports our hypothesis regarding the role of Arg 880 in the active site of XOR (93, 94). HMP less effectively utilizes transition state stabilization by charge complementation at its C6 position by Arg 880, and is also unable to utilize Glu 802 (which is now positioned on the wrong side of the heterocycle). HMP is thus likely to be a poor substrate because it cannot utilize such stabilization. Xanthine and other substrates with oxo or thio groups at their C6 position are able to bind in the preferred orientation, thereby most effectively utilizing transition state stabilization as they direct negative charge accumulation introduced at their C8 position out to their respective functional groups at C6 as shown in Figure

2.3.

The next step in confirming the role of Arg 880 will be to obtain the crystal structure of bovine XO with a good substrate or substrate analog in the active site.

Such a structure should show that substrates with proper functional groups at C6 (or an analogous position) will orient opposite to that seen with HMP, with such groups directed toward Arg 880.

63

CHAPTER 4

STRUCTURES OF XANTHINE OXIDASE FROM BOS TAURUS WITH

XANTHINE AND LUMAZINE BOUND IN THE ACTIVE SITE

4.1 Introduction

Having obtained the crystal structure of xanthine oxidase with the poor substrate 2-hydroxy-6-methylpurine bound in the active site, the next logical step was to attempt crystallization with good substrates, i.e. those with functional groups at their C6 positions that allow for transition state stabilizing interactions with Arg 880.

Ideally, an X-ray crystal structure with xanthine bound in the active site would be obtained. However, the rate of xanthine turnover is on the order of tens of milliseconds when xanthine is in 10-fold stoichiometric excess (56). Given that enzymes in crystals remain catalytically active to a good degree, and that the time elapsed between cryoprotectant exchanges, substrate soaking, and then mounting/freezing the crystals was on the order of minutes, it was unlikely that we could obtain bound substrate or product with xanthine. Our attention thus first turned to the pteridine lumazine, which has long been studied as a fairly good substrate for

XOR and has the advantage of having the identical pyrimidine subnucleus as does xanthine (109, 110, 111).

64

Figure 4.1 Structures of xanthine and lumazine. Both molecules share the same pyrimidine subnucleus, and thus the orientation of lumazine should be analogous to that of xanthine in the XO active site. The C4 position of lumazine corresponds to the C6 of xanthine, and lumazine is hydroxylated on its C7 position rather than at C8 as is xanthine.

65 Lumazine (2,4-dihydroxypteridine) is a good substrate for XOR, and is hydroxylated at its C7 position to form violapterin (111). Catalytic hydroxylation by bovine XO occurs with a rate constant of 0.7 s-1 at pH 8.3 and 25oC, which is lower than for xanthine but far greater than the same constant for a poor substrate such as HMP (66,

85). The pyrimidine subnucleus of lumazine possesses carbonyls at the C4 and C2 positions (corresponding to C6 and C2 on xanthine), and thus lumazine should be able to utilize transition state stabilization by Arg 880 via its C4=O.

Previous work with pteridine substrate lumazine has focused on the mechanism of hydroxylation as well as the role of various active site residues in this mechanism (110, 111). The hydroxylation of lumazine to violapterin is amenable to study by resonance Raman spectroscopy, as the resulting 2,4,7-trihydroxypteridine exhibits several well-characterized vibrational modes (111, 112).

Obtaining a crystal structure with bound lumazine would test the proposed role of Arg 880 in discriminating good from poor XOR substrates, and the concept of a preferred orientation for good substrates. Nevertheless, although lumazine shares its pyrimidine ring with xanthine, it is still not a physiological substrate for XOR.

Crystallizing native XO with the introduction of xanthine would most likely result in its complete turnover to uric acid, making the observation of bound substrate or product improbable in such a structure. We therefore decided to crystallize an inactive form of bovine XO, subsequently introducing xanthine into an active site that is not catalytically active yet structurally equivalent to wild-type enzyme. The most obvious means of inactivating XO for this procedure is by incubation of the enzyme with cyanide (CN-) as has been noted in many previous studies (92, 113). This leads to replacement of the planar Mo=S with another oxo ligand, thereby abolishing any catalytic activity as shown in Figure 4.2. 66

Figure 4.2 The reaction mechanism of bovine xanthine oxidase showing the resulting inactivation from incubation with cyanide. This occurs at the first step in the catatlyic sequence, whereby the desulfo-enzyme is unable to accept the hydride transfer concurrent with nucleophilic attack by the Mo- O- group.

67 As shown, the replacement of the sulfido ligand with another oxo group produces a molybdenum center that is catalytically inactive. Thus the enzyme is able to bind substrate, but will be unable to catalyze hydroxylation to uric acid. Theoretically there should be no modifications to the enzyme aside from this at the molybdenum center, and thus crystallization of the desulfo-enzyme should follow the same protocol as for native enzyme. Following incubation with cyanide and removal by gel filtration, crystals can be grown and xanthine introduced by soaking as with any other substrate.

Our efforts here have produced the structure of wild-type bovine XO with bound lumazine, as well as desulfo-XO with bound xanthine. Both structures demonstrate clear electron density maps supporting our hypothesis regarding the role of Arg 880 and substrate orientation in the active site.

4.2 Materials and Methods

4.2.1 Materials

Materials were obtained as described in Chapter 3.

4.2.2 Preparation of bovine xanthine oxidase and the desulfo-form

Native bovine xanthine oxidase was prepared as before from a single animal according to the methods in Chapter 1.

Preparation of desulfo-XO initially followed the method of Massey and

Edmondson (92). Single-source XO was incubated on ice for 3-4 hours with 1 mM

NaCN. As discussed below, 10 mM oxidized glutathione was added at the end of the incubation with cyanide. This treated enzyme solution was then passed down a

Sephadex G25 column to remove the CN- and glutathione. Resulting solutions were 68 concentrated using a Centricon filter (Millipore, Billerica, MA) to 34.5 µM (10 mg/ml

XO).

4.2.3 Crystal growth, diffraction, and data acquisition

Crystals of both the active and inactive forms of the enzyme were grown via the batch solution as for the structure with 2-hydroxy-6-methylpurine, with some modifications (94). For the structure of wild-type XO with lumazine, final well solutions consisted of 10 µl enzyme solution with 5-6 µl of the precipitant solution.

PEG8000 was again used as the precipitant, in a 12% solution as before. The cryoprotectant solution was 42% PEG200 at pH 8.0, again with concentrations of all other reagents in the wells during crystal growth, and was introduced by buffer exchanges with a 2 µl micropipette. A 10 mM stock solution of lumazine (or a 33.3 mM stock solution of xanthine) was used to introduce the substrate at 1 mM (or 10 mM) into the cryoprotectant solution in each well.

The conditions yielding the best crystals of desulfo-XO were a 12% PEG8000 solution as before, with 10 µl enzyme solution and 5-6 µl precipitant solution. The crystals grew as for previous well solutions, with a mixture of large and small rectangular plates. It was the smaller crystals that yielded the best datasets for the desulfo form of the enzyme.

Crystals were not screened prior to shipment to a synchrotron radiation source.

Thanks to the staff of the SGX-collaborative access team (-CAT), all screening was done prior to full dataset acquisition at the synchrotron. Final diffraction data were then collected again at Argonne National Laboratory on the SGX Pharmaceuticals,

Inc. beamline, using a wavelength of 0.9793 Å and a MARCCD 165 detector.

69

4.2.4 Data Processing and Refinement

Data were processed using the MOSFLM package of the CCP4 program suite

(96, 97). The structures of both forms of the enzyme were determined by molecular replacement using the MOLREP package of CCP4, using the protein data bank file

1FIQ as the search model (17). Following the initial rigid body refinement in

REFMAC of the CCP4 suite, the structure was refined using the restrained refinement protocol in REFMAC (98-102). The weighting term for geometric restraints was adjusted in REFMAC to minimize Rcryst while at the same time minimizing the difference between Rcryst and Rfree, as was the case for the structure with HMP (94).

Non-crystallographic symmetry restraints (set on ‘tight’ in REFMAC) were used between the monomers of each respective homodimer in refining the structure of the desulfo enzyme, as this also minimized the divergence between Rcryst and Rfree.

The lumazine and xanthine molecules were constructed using the PRODRG2 server (103), and the respective PDB files were built into the corresponding 2Fo-Fc and Fo-Fc omit electron density maps observed in COOT (104). Following the merging of the substrate structures with their respective XO structures into one PDB file, the resulting structures were refined again using restrained refinement in

REFMAC. Water molecules were added following the XO-lumazine electron density maps, but were not added to the dsXO-xanthine structure as the resolution was only

2.6 angstroms.

70 4.3 Results and Discussion

4.3.1 Xanthine oxidase with lumazine

The overall structure of xanthine oxidase with lumazine was quite similar to that previously determined with the slow substrate HMP (94). The space group was again P21, the asymmetric unit contained one dimer, and the unit cell was monoclinic.

The overall dimensions of the unit cell were a = 133.2 Å, b = 73.5 Å, and c = 146.5 Å with angles of 90, 98.7, and 90 degrees. Also like our previous crystal form, specific residues that were disordered in the search model for molecular replacement became more apparent in the electron density maps of our final structure. This was true of residues 1316 – 1328 in one of the two monomers. Table 4.1 gives the statistics and structure factors for the structure with lumazine.

71

Statistic XO with lumazine PDB Code 3ETR Spacegroup P21 Resolution 26.4 – 2.2 Å Wavelength (Å) 0.9793 Unique Reflections (test set) 133,695 (7052) Completeness % (highest resolution 98.7 (96.1) shell, Å) I/σ (highest resolution shell) 5.0 (1.6) Rcryst (highest resolution shell) 19.7 (23.6) Rfree (highest resolution shell) 26.7 (34.2) Ramachandran Statistics (%) 88.2, 10.1, 1.0, 0.7 Mean coordinate error based on free 0.234 R value (Å) Mean coordinate error based on 0.176 maximum likelihood (Å) Rmsd bond length (Å) 0.020 Rmsd bond angles (o) 2.0 Average B factor (A2) 29.5 Number of non-hydrogen atoms in 20,307 refinement Number of Waters 1232

Table 4.1. Statistics and structure factors for the crystal structure of XO with lumazine. Ramachandran statistics indicate the percentage of residues in the most favored, additionally allowed, generously allowed, and disallowed regions of the Ramachandran diagram as defined by the program PROCHECK (108).

72 As seen in the structure with HMP, all cofactors in the structure with lumazine were accounted for by the electron density maps, showing clearly delineated positive density with omission of each cofactor from each monomer. The rms displacements for the alpha-carbons of each respective monomer in the final refined structure with lumazine, relative to that of the oxidized monomer used as the search model (PDB accession number 1FIQ), were 0.366 and 0.338 Å. All distances involving the cofactors were nearly identical to those reported previously (17, 94, 106).

As was the case for the previous structure with HMP, residues 1316-1328 in one monomer, which were disordered in the search model, were shown to be ordered as indicated by strong positive electron density in our maps. These residues were fit in COOT, and the resulting peptide chain contained 2-3 residues in disallowed orientations by Ramachandran statistics. Overall, the Ramachandran statistics of our structure here are very similar to those with HMP (94). Disallowed residues aside from the 2-3 in the newly fit region were largely scattered at random throughout the

>2500 residue dimer. Given the clear positive density that was present in the electron density maps, all of these residues were retained in our final structure.

As given in Table 4.1, the Rcryst and Rfree for the structure with lumazine were

19.7 % and 26.7 %, respectively. A MolProbity report on the structure yielded a score of 2.69 with a Clashscore of 17.59 following the addition of hydrogens (114).

This compares favorably to the results for our previous structure with HMP, which had values of 2.66 and 17.58 (94, PDB code 3B9J). Lumazine in both active sites of native xanthine oxidase, as fit into the 2Fo-Fc and Fo-Fc omit density maps, is show in

Figure 4.3.

73 (A)

(B)

(C)

(D) Figure 4.3 Lumazine in the active sites of bovine xanthine oxidase. Panels (A) and (B) show the 2Fo- Fc (contoured at 1.0σ) and Fo-Fc (contoured at 3.0σ) maps in one active site following omission of the lumazine molecule, both maps being overlayed on the final structure. Panels (C) and (D) show the second active site of the dimer. Note that it is the C4=O in both cases that is directed toward Arg 880.

74 As shown, the orientation of lumazine in the XO active site is such that its C4 carbonyl (analogous to the C6 position in purines such as xanthine) is directed toward

Arg 880, at distances of 5.9 ± 0.02 and 3.5 ± 0.02 Å at the nearest point, respectively, in the two active sites present in the asymmetric unit. This orientation is opposite that for HMP described previously, and supports our hypothesis that this arginine stabilizes the transition state electrostatically, as negative charge is directed out to the

C4=O with nucleophilic attack at C7 (93, 94, 111). Also, having demonstrated in the structure with HMP that our methods can capture substrates present as true catalytic intermediates, it is worth noting that electron density for catalysis is present/forming between the Mo-OH and C7 of lumazine in Figure 4.3 panel C as predicted by previous work (111). The C7 is positioned 2.3 ± 0.02 Å from the Mo-OH here, and is at a distance of 4.0 ± 0.02 Å in the other active site. The structure here thus truly represents the orientation of lumazine in the active site of xanthine oxidase prior to hydroxylation but at the beginning of the true catalytic cycle, and that this orientation is opposite that for HMP (94).

Attempts to fit the lumazine molecules in opposite orientations to those shown in Figure 4.3 resulted in several atoms on the pyrazine subnucleus being placed outside of any observable electron density.

4.3.2 Desulfo-xanthine oxidase with xanthine

The structure of the desulfo form of the enzyme with xanthine was much more of a challenge to obtain. Early attempts at growing crystals following incubation

(with subsequent removal) of CN- did yield crystals that were nearly identical to those of our previous structural work, following the same protocol as before. These early

75 crystals, however, did not diffract nearly as well, with a much lower percentage of a given set of crystals yielding datasets that could be processed at all. Furthermore, in the few datasets that were able to be collected, the spacegroup was unlike that of our previous structures, and upon molecular replacement with 1FIQ as the search model, the resulting asymmetric unit consisted of either 1) a dimer of the homodimeric enzyme or more likely 2) two separate monomers. An example of the repeatable structure as two monomers is shown in Figure 4.4.

76

Figure 4.4. Two separate monomers of xanthine oxidase following incubation with cyanide, following molecular replacement using the CCP4 Program Suite. The two respective molybdopterin centers are shown as spheres in each monomer. (Pauff and Cao, unpublished)

77 Cyanide is known to interfere with disulfide bridges as well as other ionic interactions within proteins, and it was possible that our prolonged incubation with

CN- resulted in such covalent modifications, thereby disrupting the structure of XO

(115). Our initial hypothesis was that excess cyanide was reacting with disulfides near the dimer interface, and by removing or reducing this excess we might be able to reduce these side effects of cyanide and oxidize enough disulfide bridges so as to prevent what appeared to be a division of the dimer. A comparison of our final structure with previous structures following molecular replacement, however, revealed that it is most likely not disulfide bridges, but rather several non-cysteine residues on the two monomers that may interact to hold the two together. The interactions appeared to be those of π-stacking, hydrophobic contacts, and salt-bridges rather than disulfide bonds, as the cysteine residues of each respective monomer near the dimer interface are not appreciably close to one another. Figure 4.5 shows the residues at the monomer-monomer interface that may be interacting to hold the enzyme together as a functional dimer. All residues are 2.5 – 4.8 Å apart.

78 (A)

(B) Figure 4.5. Panel (A) showing residues at the interface of the two monomers that may be responsible for formation of the dimer in bovine xanthine oxidase. Each pair of residues consists of one residue labeled in black and another in red, from each respective monomer. Panel (B) is a stereo-view representation of the interface.

79 Kinetic investigations suggested that our incubation protocol was necessary to completely inactive the enzyme (less time or lower concentrations of CN- were not effective). Therefore, we needed a means to quickly remove excess cyanide, in hope of minimizing other deleterious effects on the protein aside from inactivation at the

Mo center. Oxidized glutathione (10 mM) was added in excess of the cyanide immediately following incubations, and produced better datasets than could otherwise be obtained. Following this adjustment of our methods, one best dataset was obtained.

The refinement statistics for our final crystal structure of desulfo-XO with xanthine are given in Table 4.2.

80

Statistic Desulfo-XO with xanthine PDB Code 3EUB Spacegroup P1 Resolution 33.1 – 2.6 Å Wavelength (Å) 0.9793 Unique Reflections (test set) 119,503 (5403) Completeness % (highest resolution 72.7 (46.8) shell, Å) I/σ (highest resolution shell) 6.2 (5.0) Rcryst (highest resolution shell) 21.4 (26.5) Rfree (highest resolution shell) 26.8 (39.8) Ramachandran Statistics (%) 86.5, 12.0, 0.9, 0.6 Mean coordinate error based on free 0.457 R value (Å) Mean coordinate error based on 0.263 maximum likelihood (Å) Rmsd bond length (Å) 0.015 Rmsd bond angles (o) 1.7 Average B factor (A2) 13.9 Number of non-hydrogen atoms in 38,070 refinement Number of Waters 0

Table 4.2 Statistics and structure factors of the crystal structure of desulfo-XO with xanthine. Ramachandran statistics indicate the percentage of residues in the most favored, additionally allowed, generously allowed, and disallowed regions of the Ramachandran diagram as defined by the program PROCHECK (108).

81 The final structure of the desulfo enzyme with xanthine was obtained from a triclinic crystal of space group P1, with two dimers in the asymmetric unit, and is shown in Figure 4.6.

82 (A)

(B)

Figure 4.6. Final structural model of the desulfo-enzyme with xanthine (PDB code 3EUB). Shown are the two dimers present in the asymmetric unit. Panel (B) is rotated 90 degrees into the page followed by a 90 degree rotation about the vertical axis relative to panel (A).

83 Such an asymmetric unit containing two enzyme molecules has precedent in previous structures of XOR (73), had been observed often for our crystals here, and thus we processed the dataset from 33.7 to 2.6 Å despite having a somewhat low completeness of 73 % over that resolution range. As was the case for the structures with HMP and with lumazine, one monomer of each dimer contained positive density for residues that were disorded in the search model. These were residues 1316 – 1324 and 1316 – 1326 respectively in each dimer. The unit cell dimensions here were a =

73.30 Å, b = 133.18 Å, and c = 142.63 Å with angles of 96.9, 93.1, and 90.0.

Ramachandran statistics from PROCHECK were similar to our previous structures

(94). The quality of our final structure was also checked using MolProbity, which gave a score of 2.57, placing our structure in the 78th percentile among structures of comparable resolution by this method of validation (114). The Clashscore for all- atom contacts was 15.81 after the addition of hydrogens and allowing Asn/Gln/His flips, placing our structure in the 88th percentile by this measure.

Despite the acceptable MolProbity analysis as well as Ramachandran statistics to validate our refined model, the average B factor remained suspiciously low at 13.9

Å2. In the absence of employing TLS parameters in the refinement protocol, the most likely explanation for such a low value is the low overall completeness of the reflections used to construct our atomic model, in particular the significant drop in completeness vs. resolution at the higher end of our dataset (CCP4 bulletin board, personal communications). This is supported by the mean B value of 23.7 derived from the slope of the Wilson plot of ln(diffraction intensity) vs. sin2θ/λ2 from the initial dataset, which is comparable to that for our structure with HMP (94).

84 All cofactors in the two dimers were accounted for by both the refined density maps and those following omission of the cofactors. The alpha-carbon rms xyz- displacements of the four monomers in our asymmetric unit as compared to the 1FIQ search model were 0.314, 0.299, 0.291, and 0.300 Å. The overall structure of each dimer within the asymmetric unit resembled that of our previous structures.

The Rcryst and Rfree of the desulfo structure with xanthine were approximately

21% and 28%, respectively, following our initial round of restrained refinement without NCS restraints, and the subsequent Fo-Fc omit maps contoured at 3.0 σ clearly showed xanthine with identical orientation in each of the four active sites of the asymmetric unit.

The orientation of xanthine in the active sites of xanthine oxidase is shown in

Figure 4.7.

85

(A)

(B)

Figure 4.7. The active sites of desulfo-XO with xanthine. The active site in each dimer that was most occupied is shown. Fo-Fc maps were contoured at 3.0 σ and were constructed with omission of xanthine, and overlaid with the final model (PDB code 3EUB). The molybdenum atom is in teal, sulfurs in yellow, oxygens in red, nitrogens in blue.

86 As shown in Figure 4.7, xanthine orients with its C6=O directed toward Arg

880 as predicted by our previous kinetic and crystallographic studies and analogously to the orientation seen with lumazine. In the two most occupied sites of the four in the unit cell, shown in Figure 4.5, the C6=O is 2.9 ± 0.3 and 3.0 ± 0.3 Å from the guanidinium nitrogen of Arg 880. In these two sites, the C8 of xanthine is positioned

4.7 ± 0.3 Å and 4.3 ± 0.3 Å from the Mo and 2.8 ± 0.3 Å and 2.8 ± 0.3 Å from the

Mo-OH, creating angles of 29.9 ± 1.7 and 15.7 ± 1.7 degrees centered on the Mo atom.

No evidence for catalysis was seen in any of the four active sites, as would be expected for the inactive desulfo form of the enzyme (92).

The orientation of xanthine as with lumazine is again opposite that seen in the previously described structure with 2-hydroxy-6-methylpurine (94). It thus appears that Arg 880 indeed acts in a manner that discriminates good from poor substrates, stabilizing the transition state of good substrates such as xanthine and lumazine by stabilizing the accumulation of negative charge on the C6=O (or C4=O) on their pyrimidine ring.

The results here conflict with previous interpretations based primarily on the crystal structure of the reduced enzyme in complex with the inhibitor alloxanthine as shown in Figure 4.8 (73).

87 (A)

(B)

Figure 4.8. Alloxanthine and uric acid in the active sites of XOR. Panel A shows the orientation of alloxanthine in complex with the reduced molybdenum center of XDH from Rhodobacter capsulatus (adapated from 73, PDB code 1JRP). Numbering has been changed to that for the bovine enzyme. Panel B depicts uric acid bound in the active site of demolybdo-XO from bovine milk (adapted from 116, PDB code 2E3T).

88 As shown in Figure 4.8, the orientation of both alloxanthine and uric acid indicate that it is the C2 position on both molecules that is interacting with Arg 880 in these structures. However, both of these structures depict a bound molecule in an enzyme that lacks the catalytic Mo-OH, in which the heterocycle sits some 1-2 Å closer to the molybdenum than does substrate in the course of catalysis. The equatorial Mo-O- ligand is present in all three of our structures, and thus even in the case of the desulfo-enzyme our structures represent or closely mimic a catalytic active site as in the native enzyme.

The orientation of xanthine and lumazine opposite to that seen for 2-hydroxy-

6-methylpurine lends confirmation to the proposed role of Arg 880 in transition state stabilization and substrate discrimination in xanthine oxidoreductase. The poor substrate HMP is unable to utilize transition state stabilization by this arginine residue, and orients in the active site with its C6 functional group directed opposite Arg 880, thereby interacting with the arginine via its C2=O as seen in our previous structure

(94). HMP and other substrates lacking an oxo or thio group at C6 are thus poor substrates for wild type XOR. Xanthine and lumazine, however, are able to direct negative charge accumulation to their C6 (C4) position, and orient to utilize such electrostatic transition state stabilization by Arg 880 via their C6=O (C4=O). Thus these are consequently good substrates.

Our results here, particularly with the xanthine-bound desulfo enzyme, also lend further support to the role of Glu 802 (Glu 232 in the enzyme from R. capsulatus). This residue has been previously proposed to facilitate tautomerization of the substrate molecule and thereby effect rate acceleration; computational studies showing that the tautomer of purine substrate with protons on nitrogens 1, 7, and 9

89 was more stable relative to the predominant tautomer in solution, which has protons on nitrogens 1, 3, and 7, upon nucleophilic attack by the Mo-OH (75). The orientation of xanthine seen here with the bovine enzyme is consistent with such a role, with the unprotonated N9 of free substrate oriented “up” toward Glu 802/232.

An alternative proposal, with tautomerization facilitated by Glu 1261 (Glu 730) would require the opposite orientation for xanthine as discussed above (116).

90

CHAPTER 5

ACTIVITY OF THE TWO FORMS OF XANTHINE OXIDOREDUCTASE IN THE

PHYSIOLOGICAL pH RANGE

5.1 Introduction

Having established the role of Arg 880 in the catalytic mechanism of xanthine oxidoreductase, and having obtained several crystal structures of the bovine enzyme, our attention now turned to more clinical aspects of the enzyme’s structure and function. As mentioned earlier, the enzyme plays a significant role in several areas of human pathology. The most recently studied and perhaps the most clinically emergent aspect of pathological activity by XOR involves vascular pathology via the production of superoxide radicals. These are produced primarily by the final reduction of molecular oxygen by the oxidase form of the enzyme, although it has been shown that the dehydrogenase is also capable of reducing O2 to form superoxide as well as peroxide. Superoxide and other free radicals can go on to interfere with many cellular functions and processes as shown in Figure 5.1.

91

Figure 5.1. Flow chart for the development of ischemia with the possibility for reperfusion injury. Shown here for the case of cerebral ischemia, the resulting cascade of pathogenetic events following free radical formation is indicated. (from 27)

92 Examples from Figure 5.1 include a decrease in the structural integrity of cellular membrane lipids (i.e. lipid peroxidation and destruction of cellular membranes), and the interactions of superoxide with the vasodilator NO in circulation (18, 117).

Superoxide from xanthine oxidase has been shown to degrade S-nitrothiols, which are considered to be a storage form of NO, as well as to convert NO to the vaso-inactive compound peroxynitrite (118). Both of these actions lead to vasoconstriction and in some cases ischemic conditions. Superoxide also dismutes to form hydrogen peroxide and other reactive oxygen species, leading to further cellular destruction (18).

As mentioned in Chapter 1, XOR is initially expressed as a dehydrogenase

(xanthine dehydrogenase, XDH), and exists in this form under physiological conditions. XDH primarily utilizes NAD+ as its final electron acceptor and thereby produces NADH. XDH can also utilize O2, however, although this form of the enzyme prefers NAD+ nearly 600-fold over molecular oxygen as reflected by kinetic constants for the reaction of each species (33). It can be concluded that XDH will not

+ utilize O2 (and thus not produce superoxide) in the presence of NAD .

In pathological conditions such as those in ischemic/hypoxic tissue, it has been proposed that native XDH is gradually converted to xanthine oxidase (XO), which utilizes O2 exclusively. This conversion arises by modification of the access channel to the flavin site of the enzyme, which prevents binding of NAD+ (119). This conversion has both a slow and a fast component, occurring by oxidation of sulfhydryl groups and by limited proteolysis, respectively (119, 120).

Compounding the conversion of existing XDH to XO, conditions of low oxygen in tissue and vasculature have been shown to bring about an increase in XOR expression and enhance the deleterious conversion to XO (121, 122). Furthermore, the activity of XOR has been shown to increase during hypoxia, and one proposed 93 mechanism involves the phosphorylation of XO by enzymes such as p38 MAP kinase and/or casein kinase II (123). Although this and other post-translational modifications remain unproven, it is worth noting that the intrinsic in vitro activity of mammalian

(bovine) XO increases in going from a pH of 7.5 to 7.0, analogous to acidosis in ischemic or hypoxic tissue. In this situation the activity of the enzyme can be assessed by an in vitro study of kcat/Km versus pH (70), a study that has been done previously for bovine XO (76). Conversely, XO was found to operate most efficiently at higher pH (optimally at 8.5) under conditions of higher substrate concentrations, assessed by observing the dependence of kcat on pH (70). Thus, in the physiological pH range of 7.2 to 7.6, xanthine oxidase operates with greater or lesser efficiency depending on the specific pH relative to the concentration of its substrates.

In contrast to this behavior by bovine XO, little attention has been given to the pH-dependence of XDH activity despite this being the initial and predominate form of the enzyme. A preliminary study of XDH from Rhodobacter capsulatus showed that the activity of this dehydrogenase had an optimal pH in the range of 7.5 to 8.5 regardless of substrate concentration (Pauff and Capretta, unpublished). The same is true of the enzyme from other organisms as well, for instance the avian enzyme from chicken hepatocytes is maximally active at a pH near 8.0 (124, 125). Therefore, the intrinsic activity of XDH from these organisms declines as pH falls from 7.5 to 7.0, absent conversion to XO. Also to be considered here is that post-translational modifications, if they occur, are likely to take place not only on XO but also on XDH, the initially expressed form of XOR and the physiologically relevant species (18).

Although much work has been done to characterize XO in animal models of ischemia/hypoxia and other disease states, few studies have focused on any innate differences between mammalian XO as compared to XDH from the same mammalian 94 organism. While our laboratory focuses on studies of bovine XO, bovine XDH has been prepared previously by others, and can be studied in its own right (33). The bovine enzyme is very similar to human XOR (having 90% sequence identity), and thus characterizing the function of XO and XDH from this mammalian source should be directly applicable to the function of the human enzyme (17). Our efforts here focus on the dependence of activity on pH for both forms of the enzyme, under conditions of low and high substrate concentrations. Low substrate concentrations

(where [hypoxanthine] or [xanthine] is less than Km at a given pH) are analogous to physiological conditions, while conditions of high purine-substrate concentrations

(greater than Km) are analogous to pathological conditions such as those found following prolonged ischemia.

Explanations for the pH dependence of XOR activity have centered primarily on ionizable residues in the enzyme’s molybdenum-containing active site. Another explanation involves the nature of the flavin site, the site of final electron transfer out of the protein to form NADH or superoxide. The pH profile for both kcat and kcat/Km of bovine XO have been previously determined, and we have obtained analogous profiles for bovine XDH. Our findings show a significant difference between the pH- dependence of XDH versus XO activity, and suggest a mechanism by which

XDH/XO may contribute to ischemia-reperfusion injury.

Another area of investigation here involves the production of superoxide by

XDH. As mentioned, XDH is also capable of producing superoxide, and studies on the interaction of bovine XDH with oxygen in the presence and absence of its preferred oxidant NAD+ have been reported (32, 33). In the context of ischemia- reperfusion pathology, the question of ROS production by XDH in the presence of a smaller fraction of the enzyme existing as XO is quite relevant. The pool of NAD+ is 95 depleted under these ischemic conditions, and the fraction of XOR as XDH may well increasingly utilize O2 as an electron acceptor (126, 127, 128). Thus we have investigated the in vitro consumption of oxygen by XDH relative to that by XO in order to gain insight into the production of superoxide by XDH in the absence of its preferred electron acceptor NAD+.

5.2 Materials and Methods

5.2.1 Preparation of bovine xanthine oxidase and bovine xanthine dehydrogenase

Bovine xanthine oxidase was prepared as in Chapter 1.

Bovine xanthine dehydrogenase was purified according to the procedure of

Hunt and Massey (69, 129). The two most important modifications of previous methods, in order to isolate a predominance of xanthine dehydrogenase relative to the oxidase, involved omission of pancreatin in the first step of the purification and the addition of 2.5 mM fresh DTT to all solutions and reagents throughout. The enzyme was isolated from fresh, unpasteurized bovine milk from Scott Brothers Dairy (Chino,

CA) on a ÄKTApurifier FPLC (GE Healthcare, Piscataway, NJ) using hydroxyapatite followed by a Sephacryl S-300 column and otherwise according to previous methods

(but with the modifications above) (54).

The ratio of XDH to XO was determined by comparing xanthine turnover activity in the presence of 500 µM NAD+ under anaerobic conditions versus that in the presence of oxygen, utilizing the near absolute preference of XDH for NAD+ and the inability of XO to utilize NAD+ (33, 129). Purified stock solutions of XOR were determined as by previous methods to be 65% XDH of which 60% was active (i.e. containing the MoSOOH cofactor in the active site of the enzyme) (129). The percentage of a given enzyme solution of xanthine oxidoreductase in vitro that is 96 capable of catalysis, in which the labile sulfur on the molybdenum center is intact and has not been replaced by oxygen, can be assessed under anaerobic conditions by comparing reduction of the molybdenum center at 450 nm absorbance by the physiological substrate xanthine versus any further reduction by the nonphysiological reductant sodium dithionite.

The presence of xanthine oxidoreductase in its dehydrogenase form was further confirmed by observing the oxidative half-reaction of the enzyme stock with

NAD+, monitoring oxidation of the dithionite-reduced enzyme at 450 nm and the production of NADH at 340 nm. Our results were consistent with previous data, giving a rate constant of 175 s-1 for the oxidation of enzyme by NAD+ (33).

Given that XDH can be converted slowly to XO by oxidation of thiols, it was important that the stock solutions of XDH remain stable (65% XDH as isolated here) for the duration of our experiments. This was tested by repeat assays with exposure of the enzyme stock to normal oxygenated air over 10 hours. XDH was thawed and incubated for 1 hour with 5 mM DTT, then passed down a G-25 column prior to the test. The enzyme was made anaerobic with argon gas for 60 minutes following filtration, was assayed immediately upon exposure to oxygen, and then assayed at 60- minute intervals for 540 minutes. Potential conversion of XDH to XO was monitored by observing the production of NADH by 100 nM enzyme, as XO is incapable of catalyzing this reaction. Each reaction was conducted at pH 7.4 in 0.1 M MOPS, 0.1

M KCl, 0.2 mM EDTA and contained 35 µM xanthine, 1 mM β-NAD+, and 100 nM enzyme. No decrease in the rate of NADH production was observed over 10 hours, confirming that the enzyme was stable on the time scale of our experiments.

Finally, to further ensure stability of our stock solutions, fresh DTT was added at 5 mM daily to any thawed stock solutions of enzyme. 97

5.2.2 Steady-state kinetics and construction of the pH profile for bovine XDH

Enzyme stocks were incubated with 5 mM DTT for 1 hour prior to each assay.

The DTT was subsequently removed by filtration using a Sephadex G-25 column and

Centricon (Millipore, Billerica, MA) filtration to reconcentrate the stock.

Steady-state assays were conducted at pH values varying from 5.5 to 8.5 in increments of 0.5 pH units. Reactions were carried out anaerobically at 25 oC in a 1 ml quartz reaction cuvette. Each reaction vessel was made anaerobic by bubbling pure argon through the reaction mixture (minus the enzyme) in the cuvette for 30 minutes. The xanthine stock solution and all other buffered solutions were also bubbled with pure argon for at least 30 minutes. Enzyme stock solutions were made anaerobic by passing argon over the solution in a closed, vented cuvette on ice for one hour.

Each reaction contained 1 mM β-NAD+ in the reaction buffer prior to the addition of enzyme. Buffers were all used at 0.1 M: MES (for pH 5.0-5.5), CHES

(for pH 6.0-6.5), MOPS (for pH 7.0-7.5), and TrisHCl (for pH 8.0-8.5). At a given pH, assays were conducted by first blanking on the buffer plus β-NAD+ after addition of xanthine, then 50 nM enzyme was added using a Hamilton gastight syringe

(Hamilton Company, Reno, NV) to initiate the reaction. Reactions were followed at

295 nm for the appearance of uric acid, and also concurrently at 340 nM for the appearance of NADH. NADH also absorbs at 295 nm, and thus to accurately monitor the turnover of xanthine to uric acid (and thus calculate a true kcat and Km for the reaction at the molybdenum-site of the enzyme) we needed a correction for the data at

295 nm. The absorbance data at 295 nm (∆A295 per second) was corrected by subtracting 0.176 x ∆A340 per second to allow for the increase in absorbance at 295 98 nm due to NADH formation, a correction arrived at by measuring the absorbance change at 295 nm and 340 nm associated with the formation of a specific concentration of NADH, and then comparing this spectra to that for unreacted β-

NAD+ at the same concentration.

5.2.3 Consumption of oxygen by XDH in the absence of NAD+

Oxygen consumption by XDH was measured on a Hansatech oxygraph

(Hansatech Instruments, Norfolk, England). As mentioned, our XDH preparations yielded a mixture consisting of 65% XDH with 35% XO in the same solution. Thus we needed to determine a way to distinguish O2 consumption by XO from the lesser contribution by XDH. In order to account for O2 consumption by XO, two control assays were run to determine its given fraction of oxygen consumption in our assays.

The first was conducted with an enzyme stock of bovine XO isolated with no observable presence of XDH. The second control assessed the consumption of O2 by

+ XDH in the presence of 500 µM NAD , as any O2 consumption under these conditions can be taken as due to the 35% of the enzyme that is XO. This assumption

+ holds given that the preference of XDH for binding NAD is 570-times that for O2 (as measured by kinetic binding constants) and the rate constant is 8-times greater for

NAD+ reduction (33). All assays were conducted with 100 nM enzyme and xanthine concentrations ranging from 1 µM to 50 µM, at pH 7.4 in 0.1 M MOPS, 0.1 M KCl, and 0.2 mM EDTA. Comparison of these O2 consumption rates allowed determination of the rate for oxygen consumption in the steady-state by XDH. All assays were conducted in triplicate and final kinetic values were adjusted for the fraction of the enzyme stocks determined to be active as explained above. Data analysis and plot-construction was performed using SigmaPlot 10.0. 99

5.3 Results and Discussion

5.3.1 pH profiles for the two forms of bovine XOR

The pH profiles for steady-state activity of XO have been determined previously in our laboratory (76). Figure 5.2 shows the pH dependence for the kinetic parameter kcat/Km. The optimal pH for kcat of XO here is 8.5, while the optimal pH for kcat/Km is 7.0 for the oxidase form of XOR.

100

(A)

(B)

Figure 5.2. pH profiles for bovine xanthine oxidase. Panel (A) shows the pH dependence of the enzyme’s activity at low substrate concentrations as given by kcat/Km. Panel (B) shows the same at high substrate concentrations as given by kcat. (adapted from 76)

101 Figure 5.3 depicts our pH profiles for kcat of bovine xanthine dehydrogenase for the kcat associated with the formation of NADH (A), the absorption-corrected kcat of urate formation (B), and the kcat/Km of XDH for xanthine-to-urate catalysis (C). The optimal pH for both kcat and kcat/Km in these three cases for XDH is 7.0 ± 0.1. This value was independent of a Gaussian, Lorentzian, or Voigt fit to the data.

102 (A)

(B)

(C)

Figure 5.3. pH profiles of bovine xanthine dehydrogenase. Shown in panel (A) is the pH dependence for NAD+ to NADH turnover, showing an optimum at pH 7.0. Panel (B) shows the pH dependence of xanthine turnover at high concentrations of xanthine, while panel (C) shows the corresponding pH dependence at low concentrations of xanthine.

103 The pH profile for XDH indicates that the enzyme functions optimally at a pH of 7.0 regardless of substrate concentration. This is distinctly different from the oxidase form of the enzyme, which functions most efficiently at a pH of 7.0 when substrate concentrations are low (lower than the Km of the enzyme or less than approximately 5 µM), but functions more effectively at higher pH (with an optimum of pH 8.5) when substrate concentrations are high. Such differences between the two forms of XOR lend insight into the contributions of XOR to ischemia-reperfusion pathology. Under physiological conditions at a pH of 7.4, the enzyme exists as XDH and is functioning slightly below optimal activity. With ischemia/hypoxia/hypoxemia and subsequent metabolic acidosis, pH falls toward 7.0, the activity of XDH at this time increases relative to physiological conditions. At the same time, expression of the enzyme may be increasing and the pool of XOR is being converted to the oxidase form XO, which has the same optimal pH for activity as XDH at low substrate concentrations. Substrate concentrations for XOR in this situation will initially remain low, although as the hypoxic state continues and cellular necrosis begins to occur (along with degradation of NAD+), the degradation of cellular metabolites is thought to increase concentrations of XOR substrates. DNA and thus future purine substrates for XOR are released into the cell cytoplasm and into circulation and/or interstitial fluid, where XO (and theoretically XDH) can catalyze the production of superoxide radicals. Upon reperfusion by oxygenated blood, as the pH then begins to rise back toward an average of 7.4, the tissue and plasma environments are rich in substrates for XOR, which now exists in large part as xanthine oxidase with the capacity to produce reactive oxygen species such as superoxide and hydrogen peroxide. As the pH-dependent activity for both forms of the enzyme may now be inferred from the effective kcat at a given pH, XO becomes more active (by 104 approximately 15-20%) with increasing pH while XDH activity decreases (by approximately 25%). Therefore the stage is set for increased superoxide-generation.

5.3.2 Oxygen consumption by XDH

Having presented evidence for an increase in enzymatic activity of XO

(relative to a decline in XDH activity) at increasing pH based on our pH profiles, we decided next to examine if the fraction of XOR remaining as XDH would also be capable of producing superoxide under the same conditions. Such an investigation is particularly relevant given previous work by others indicating that XDH also reacts with molecular oxygen to produce superoxide, and given the fact that hypoxia leads to the depletion of intracellular NAD+ (126 - 128). Thus, the strong preference for

NAD+ as its final oxidant would theoretically be compromised under ischemic conditions or in the acute stages of reperfusion.

-1 XDH was shown to consume oxygen at a rate near 30 (µM O2) min (µM

XDH)-1. Kinetic values for the consumption of oxygen by XDH and XO in our stock solutions, with corresponding controls in the presence of NAD+ and absence of XDH, are shown in Table 5.1.

105

XO XDH/XO XDH/XO + 500 µM NAD+

-1 Vmax (µM O2 min ) 63 ± 3 16.5 ± 1.0 14.4 ± 0.6

Table 5.1. Oxygen consumption kinetics for XDH and XO. All numbers adjusted for enzymatic activity. XO stock was 30% active. XDH/XO was 65% XDH, 35% XO, and was 60% active.

106 Our results show an enhanced ability of XO over XDH to utilize oxygen and thus produce reactive oxygen species. They also indicate that in the presence of a smaller fraction of XO the vast majority of oxygen consumed (and thus reactive oxygen species produced) will be by the oxidase form of the enzyme. Furthermore,

+ given the strong preference by XDH for NAD over O2, in the presence of any appreciable NAD+ it can be assumed in this case that XO will operate as the principle producer of reactive oxygen species. Our results here are consistent with the strong

+ + preference of XDH for binding NAD over O2 and having a rate constant of NAD reduction that is 5 to 10-fold greater than that for O2 reduction. These two kinetic parameters both indicate negligible production of reactive oxygen species by XDH in the presence of any appreciable NAD+.

More importantly, our results here suggest that in an environment depleted of

NAD+ such as that following ischemia, XDH does possess the ability to generate reactive oxygen species at a measurable rate with sudden exposure to oxygen in the absence of its preferred oxidant. The reintroduction of oxygen with reperfusion following prolonged ischemia will provide a usable oxidant for both forms of the enzyme. XOR thus has the capacity to produce reactive oxygen species proportional to the fraction of the enzyme that exists as XO, and perhaps also to a lesser extent via the remaining XDH as NAD+ levels are depleted.

107 The conclusions of this study, especially when put into the context of other work demonstrating increased levels of XO(R) and increased XO activity (both from increased expression and an increase in enzymatic activity), suggest that there exists a fundamental functional difference in the pH-dependence of the two forms of XOR.

This difference must be considered when interpreting any post-translational modifications such as phosphorylation and the resulting effects on XOR activity. Our results also suggest that the xanthine oxidoreductase system functions in a manner that may exacerbate damage from ischemia with subsequent reperfusion.

108

CHAPTER 6

INHIBITION STUDIES OF XANTHINE OXIDASE

6.1 Introduction

There have been many recent reports of inhibitory molecules that may be novel therapeutics for human disease. As mentioned, XOR is a key agent in many pathological conditions. Most notably, hyperuricemia resulting from unchecked XOR activity is central to the pathogenesis of gout and gouty arthritis, as well as causing the high serum urate levels associated with (tumor) cell necrosis (28, 130). XOR in its oxidase form is considered to be a main source of oxidative stress and destructive free radicals in ischemia-reperfusion events such as transient ischemic attacks and stroke, as well as in myocardial or renal hypoxia and infarctions (31, 131, 132).

Inhibition of XOR is a primary objective in treating any case of hyperuricemia

(31, 133). Allopurinol was the first mechanism-based inhibitor of the enzyme to be developed, and is still the primary drug for treatment of hyperuricemia. Allopurinol is hydroxylated in the course of inhibition by the enzyme to alloxanthine (oxypurinol), which coordinates tightly to the now reduced molybdenum center (Figure 4.6 panel

(A), replacing the Mo-OH group of native enzyme) (134). Although allopurinol has longstanding use in pharmacotherapy, is efficacious in lowering urate levels in the body, and is generally well tolerated, some individuals have exhibited hypersensitivity to the drug with side effects such as vasculitis, especially in those with already 109 compromised renal function (31). The development of alternative XOR inhibitors is therefore desirable and has been the subject of several recent studies.

One alternative inhibitor that is currently in clinical trials is trade named

Febuxostat (135, 136, 137). The crystal structure of the inhibitor in complex with the enzyme has been determined (Figure 6.1), showing the (hydroxylated) inhibitor blocking the access route to reduction of the enzyme analogous to alloxanthine binding (44).

110 (A)

(B)

Figure 6.1. The orientation of febuxostat in the access channel to the Mo in xanthine oxidoreductase. Panel (A) shows specific residues, while panel (B) shows the structure of febuxostat relative to the mechanism-based inhibitor allopurinol. (adapted from 44)

111 Febuxostat was developed in Japan, is currently in Phase III clinical trials in the

United States, and has received many favorable reviews and recommendations (45,

138).

There have been many other reports suggesting that any number of naturally occurring molecules ranging from flavonoids to a host of other natural plant products inhibit XOR (139-146). Most of these studies have utilized in vivo data exclusively and focused on outcomes in murine models for various states of hyperuricemia.

Several of these studies have utilized the readily available enzyme from bovine milk, but little work has focused on the mechanism by which such inhibition may occur.

Based on previous work by others, we chose to characterize the mechanism of this proposed inhibition for four natural products; silibinin, quercetin, curcumin, and luteolin, which are shown in Figure 6.2.

112

Figure 6.2. Natural products proposed to reduce or suppress the activity of XOR. Also shown for comparison are allopurinol, xanthine, and lumazine.

113 In addition to these natural products, the identification of compounds already in clinical use as potential inhibitors of XOR would be valuable. The secondary application of FDA-approved drugs that are currently available for the treatment of human disease is a very desirable goal and has been an area of active research for many years. In fact, several drugs (e.g. statins) have recently received approval for the treatment of diseases for which they were not originally intended. Others, such as coumarin derivatives, have long been used for a variety of clinical applications, and novel applications of these compounds continue to be suggested by further research

(147, 148, 149).

Our own examination of existing drugs that may inhibit XOR began with two cardiovascular drugs and one parent compound: ticlopidine (5-(o-Chlorobenzyl)-

4,5,6,7-tetrahydrothieno(3,2-c)pyridine), clopidogrel (methyl (2-chlorophenyl)(6,7- dihydro-4H-thieno[3,2-c]pyridin-5-yl)acetate hydrogen sulfate), and the parent compound coumarin. The two antiplatelet compounds are shown in Figure 6.3.

114

Figure 6.3. The cardiovascular drugs ticlopidine and clopidogrel. Both inhibit platelet aggregation and are therefore used as antithrombotic agents. Note the respective purine-resembling moieties.

115 Our work initially focused on the turnover of these three compounds by XOR, as

XOR is know to catalyze the hydroxylation of various compounds and has been targeted to avoid the inactivation of drugs such as 6-mercaptopurine as mentioned in

Chapter 1. Mass spectrometry was employed to observe catalytic hydroxylation following incubation of 100 - 500 µM of each compound of interest with 50 nM enzyme. Our results, including attempts at increasingly longer timescales, various temperatures, and various enzyme concentrations, were uniformly negative and we conclude that none of these compounds (ticlopidine, clopidogrel, coumarin) were hydroxylated by XOR.

We then turned to inhibition of XOR by these compounds, particularly noting that ticlopidine, clopidogrel, and coumarin all bear some resemblance to a purine molecule. Steady-state assays were conducted in the presence of 10 – 100 µM of each compound, but no inhibitory effect was found for any of the three. Our work with ticlopidine, clopidogrel, and coumarin was thus abandoned.

116 In the context of the novel application of FDA-approved compounds for new treatments of human disease, coumarin-derived compounds have proven clinically effective (147, 150-152). Given our negative results with coumarin itself, it was a surprise to find several studies in the literature suggesting that coumarin itself inhibits

XOR, and that modified coumarins are even more potent inhibitors (147). These studies primarily utilized whole-cell extracts, assaying bulk properties of the lysate for the turnover of xanthine or hypoxanthine. In light of these suggestions that several coumarin derivatives inhibit XO and/or scavenge free radicals, we decided to undertake the study of coumarin and two of its derivatives, 6,7-dihydroxycoumarin

(esculetin) and trans-4-hydroxy-3-methoxycinnamic acid (ferulic acid) as potential inhibitors of xanthine oxidase, and compare any inhibitory activity to that of allopurinol (153, 154).

117

Figure 6.4. Coumarins investigated as potential inhibitors of XOR activity. Also shown are xanthine and uric acid.

118 We find that luteolin, silibinin, and quercetin are indeed inhibitors of XOR, as measured by their effect on the initial rate of catalysis of xanthine to urate, and that all three compounds also reduce the rate of superoxide production by XO with xanthine.

In contrast to these flavonoid compounds, curcumin does not inhibit the production of urate or superoxide by XO in vitro. Our results also support a role for coumarin derivatives in suppressing XO activity and suggest that the inhibition is purely competitive unlike allopurinol.

6.2 Materials and Methods

6.2.1 Compounds and Reagents

Silibinin, quercetin, luteolin, curcumin, coumarin, esculetin, ferulic acid, and allopurinol, xanthine were purchased from Sigma-Aldrich (St. Louis, MO). All were

>95% purity and were used without further purification. Stock solutions of each potential inhibitor were prepared at a concentration of 5 mM in 0.1-0.2 M aqueous potassium hydroxide, as were 33.3 mM and 1 mM xanthine stock solutions. All other reagents were purchased either from Aldrich or Fisher (Thermo Fisher Scientific,

Waltham, MA) and were of the highest purity commercially available, and were used without further purification.

6.2.2 Isolation of xanthine oxidase

Xanthine oxidase was isolated and purified from bovine milk as described in

Chapter 1. The enzyme as isolated was approximately 50 - 70% active based on the activity-to-flavin ratio as well as by comparing the extent of reduction of the enzyme by xanthine under anaerobic conditions, as monitored at 450 nm. Enzyme

-1 -1 concentrations were determined at 450 nm (ε450 = 37.8 mM s ) (155). 119

6.2.3 Steady-state kinetics and absorption spectra

For luteolin, quercetin, silibinin, and allopurinol, inhibitory activity at the molybdenum active site of XO was determined as follows. Each assay contained 50 nM XO, with seven to eight xanthine concentrations ranging from 1 µM to 800 µM in the presence or absence of a known concentration of each compound. Each of the four compounds in question was tested at concentrations of 10 µM, 25 µM, 50 µM, and 100 µM, monitoring the ∆A295 associated with the generation of uric acid (ε295 =

-1 -1 9600 M cm ). Kinetic constants kcat and Km were determined from a plot of the initial reaction velocity vobs versus [xanthine]. Values for Ki in competitive inhibition were determined from secondary plots of the varying slopes versus [inhibitor] using regression data generated from double-reciprocal plots of [enzyme]/vobs versus

1/[xanthine] at 10, 25, and 50 µM concentrations of inhibitor. All data analyses were performed on SigmaPlot 10 (Systat Inc., San Jose, CA).

For the coumarin derivatives, 50 nM XO was again used in each reaction.

Xanthine concentrations were varied from 2 µM to 50 µM, and assays were conducted with 10 µM, 25 µM, and 50 µM of each potential inhibitory compound.

In addition to the above, assays were conducted following 10-minute incubations of XO with 25 µM of each compound prior to the addition of xanthine to initiate the reaction. Controls were run in the absence of any inhibitor, before and after testing each respective compound in the assays above. The time-dependent nature of inhibition was also assessed by monitoring the production of uric acid over

15 minutes in the presence of 50 µM of each compound in question. Each assay was conducted in the same steady-state assay buffer and contained 5 nM XO, 100 µM

120 xanthine, and 50 µM inhibitor. Absorbance at 295 nm was measured at 30-second intervals for 900 seconds.

The capacity of luteolin, silibinin, quercetin, and allopurinol to influence superoxide production was determined by including 75 µM oxidized cytochrome c in

-1 -1 the reaction mix, monitoring the reaction at 550 nM (ε550 = 19.6 mM cm ) (156).

Oxidized cytochrome c was incubated for 10 minutes with 2-3 mM ferricyanide, passed down a Sephadex G-25 column, and the concentration was determined from

-1 -1 the absorbance at 410 nm using ε410 = 106 mM cm (157). Assays were conducted in the presence of 10 µM, 25 µM, and 50 µM of each compound tested with 50 nM

XO and 100 µM xanthine.

To assess whether each compound could coordinate to the molybdenum, 2 µM

XO was titrated under anaerobic conditions with 25 µM of each compound.

Reduction of the molybdenum center was monitored over 10 minutes at 450 nm as

-1 -1 described previously using ε450 = 37.8 mM cm .

Each reaction was run at 25oC in 0.1 M MOPS, 0.2 mM EDTA, with 0.1 M

KCl (for ionic strength) at a pH of 7.4, with a final reaction volume of 1 ml (3 ml in the case of anaerobic titrations). All assays were conducted using a Hewlett-Packard

8452A diode array spectrophotometer interfaced with the Hewlett-Packard

Chemstation (Palo Alto, CA). All assays were done in triplicate.

6.2.4 Mass spectrometry to formally assess the hydroxylation of coumarin

To confirm our previous, unpublished conclusion that xanthine oxidase did not hydroxylate coumarin, we incubated 100 µM coumarin with 100 nM XO for 24 hours in 0.1 M Tris hydrochloride, 0.1 M KCl, and 0.2 mM EDTA at pH 7.8 and then

121 analyzed for potential product(s) using mass spectrometry. The reaction mix was filtered using a Centricon filter to remove enzyme. All experiments were performed on a Micromass ESI-Tof™ II (Micromass, Wythenshawe, UK) mass spectrometer equipped with an orthogonal electrospray source (Z-spray) operated in positive ion mode. Sodium iodide was used for mass calibration for a calibration range of m/z 100

- 2000. Coumarin and the potential hydroxylated product(s) were each prepared in a solution containing acidified methanol and infused into the electrospray source at a rate of 5 - 10 µl min-1. Optimal ESI conditions were: capillary voltage 3000 V, source temperature 110oC and a cone voltage of 55 V. The ESI gas was nitrogen. Data was acquired in continuum mode until acceptable averaged data was obtained. The spectrum of the resulting compound(s) was compared to that of coumarin prepared alone in the reaction buffer.

6.3 Results and Discussion

6.3.1 Steady-state inhibition studies with luteolin, quercetin, silibinin, and allopurinol

Absorbance spectra were taken for all four compounds and peaks of maximal absorption were recorded for luteolin (210, 260, 320 nm), quercetin (230, 320 nm), curcumin (420 nm), and silibinin (230, 320 nm). Several absorbance spectra were taken each of 10 consecutive minutes to assess the stability of our stock solutions.

Each compound was also monitored at 295 nm with 50 nM XO in reaction buffer, and under the same conditions at 550 nm with 75 µM cytochrome c. No change in absorbance was observed at 295 nm for any compound, although a significant albeit slow increase in A550 was observed for quercetin and luteolin indicating very slow reduction of oxidized cytochrome c.

122 Table 6.1 gives the resulting steady-state kinetic constants for our inhibition studies with luteolin, quercetin, silibinin, curcumin, and allopurinol. Error from a hyperbolic fit of the data is given as one standard deviation.

123

-1 kcat (s ) Km (µM) kcat/Km (µM-1·s-1) Luteolin 10 µM 6.2 ± 0.2 31 ± 5 0.20 ± 0.04 25 µM 3.8 ± 0.2 127 ± 20 0.03 ± 0.006 50 µM 2.6 ± 0.2 164 ± 30 0.02 ± 0.004 Incubation 5.8 ± 0.5 126 ± 30 0.05 ± 0.01 Control 12.1 ± 0.6 3.5 ± 0.8 3.5 ± 0.7

Quercetin 10 µM 5.9 ± 0.2 3.7 ± 0.3 1.6 ± 0.2 25 µM 4.5 ± 0.4 43 ± 10 0.10 ± 0.02 50 µM 1.9 ± 0.2 106 ± 30 0.02 ± 0.006 Incubation 4.3 ± 0.8 104 ± 50 0.04 ± 0.02 Control 10.5 ± 0.4 1.2 ± 0.3 8.8 ± 2.0

Curcumin 10 µM 13.3 ± 0.6 2.8 ± 0.6 4.8 ± 1.0 25 µM 11.8 ± 0.4 1.8 ± 0.3 6.5 ± 1.0 50 µM 12.2 ± 0.5 2.4 ± 0.5 5.1 ± 1.0 Incubation 12.4 ± 0.8 3.2 ± 0.9 3.9 ± 1.0 Control 12.4 ± 0.4 2.1 ± 0.4 5.9 ± 1.0

Silibinin 10 µM 5.0 ± 0.3 36 ± 8 0.14 ± 0.03 25 µM 4.0 ± 0.4 30 ± 10 0.13 ± 0.04 50 µM 4.3 ± 0.3 39 ± 10 0.11 ± 0.03 Incubation 3.8 ± 0.1 53 ± 7 0.07 ± 0.007 Control 10.9 ± 0.4 1.9 ± 0.3 5.6 ± 1.0

Allopurinol 10 µM 8.1 ± 0.6 25 ± 7 0.32 ± 0.1 25 µM 6.5 ± 0.6 43 ± 20 0.15 ± 0.08 50 µM 6.9 ± 0.5 63 ± 20 0.11 ± 0.03 Incubation 1.3 ± 0.3 189 ± 90 0.007 ± 0.004 Control 10.1 ± 0.8 3.1 ± 1.0 3.3 ± 1.0

Table 6.1. Steady-state kinetic constants for our inhibition studies with luteolin, quercetin, curcumin, silibinin, and the control with allopurinol.

124 Allopurinol at 10, 25, and 50 µM was used as a control for comparison to any inhibition by each of the natural products. In our work here, allopurinol inhibited XO in a complex manner fully consistent with inhibition being due to binding of alloxanthine/oxypurinol to reduced (MoIV) enzyme (134). Allopurinol reduced kcat/Km for xanthine turnover, reflecting the reaction of free enzyme and free xanthine at low [xanthine], 30-fold at a concentration of 50 µM, and >400-fold at 25 µM following a 10 minute pre-incubation with the enzyme. Other kinetic data for the inhibition of XO by allopurinol are given in Table 6.1.

As shown in Table 6.1, luteolin was the best inhibitor among the natural products examined here, with a Ki of 1.9 ± 0.7 µM as shown by the double-reciprocal plot in Figure 6.5. At 10 µM, the inhibitory effect of luteolin, when presented to XO along with xanthine, resulted in a nearly 20-fold reduction in the overall catalytic power of the free enzyme as assessed by kcat/Km. There was further inhibition of turnover with the free enzyme at luteolin concentrations of 25 and 50 µM culminating in a 170-fold reduction in kcat/Km. The observed rate constants were reduced up to sixfold compared to control at these three concentrations. Any additional effect following incubation of the enzyme with luteolin prior to the introduction of xanthine was negligible compared to that when the enzyme was exposed to xanthine and luteolin simultaneously, suggesting a competitive mode of inhibition.

125 Inhibition of XO by quercetin was similar to that by luteolin, as may have been expected from their similar structures. The Ki for competitive inhibition by quercetin was determined to be 1.2 ± 0.7 µM from the plot in Figure 6.5. Pre- incubation of XO with 25 µM quercetin did not result in any additional decrease in turnover rate, although the Km for xanthine was slightly more than twice that seen without pre-incubation.

126

Figure 6.5. Double-reciprocal plots of [XO]/vobs versus 1/[xanthine] for each concentration of luteolin (panel A) and quercetin (panel B), with resulting plots for the determination of Ki also shown. The approximate intersection of the linear fits at x = 0 for each [inhibitor] indicates that the mode of inhibition is competitive.

127 Inhibition of XO by silibinin in the steady-state was independent of the silibinin concentration. At silibinin concentrations of 10, 25, and 50 µM, the kcat for xanthine turnover is reduced by approximately 50 % in each case, and the value of kcat/Km is reduced 50-fold. Pre-incubation of the enzyme with 25 µM silibinin resulted in no additional decrease in kcat, but rather a significantly increased Km, and the value for kcat/Km was further decreased from 5.63 to 0.07, reflecting an 80-fold reduction when compared to the control.

Curcumin did not demonstrate any appreciable inhibition of bovine xanthine oxidase at 10 µM, 25 µM, 50 µM, or 100 µM in steady-state assays. Neither did pre- incubation of XO with 25 µM curcumin for 10 minutes result in a significant change in either kcat or Km.

Figure 6.6 compares the ∆A295 over 10 minutes in reactions with 25 µM inhibitor, 100 µM xanthine, and 5 nM XO. As shown, time-dependent inhibition of

XO by allopurinol was observed in 10 minute trials, as noted previously by others

(158). Significant inhibition was also observed with luteolin, as noted by the linear increase in A295 over 10 minutes, although no time-dependent increase in inhibition was observed. The degree of inhibition by quercetin was very similar to that seen with luteolin, and also showed no indication of delayed onset of inhibition as is seen with allopurinol. Silibinin was a less potent inhibitor by this method.

128

Figure 6.6. Time course of the inhibition of XO by allopurinol, luteolin, silibinin, and quercetin. Reactions were monitored by A295 at 30 second intervals over 10 minutes. Each point represents an average of 3 trials.

129 Our investigations of luteolin, quercetin, silibinin, and curcumin were stimulated by various and sometimes conflicting reports on natural inhibitors of xanthine oxidase. Were we to confirm that these proposed inhibitors indeed reduced the activity of XO, it would suggest that these could be used as lead compounds for the development of new inhibitors of XO. The turnover of xanthine to uric acid by

XO is inhibited by luteolin, quercetin and, to a lesser degree silibinin. Inhibition by luteolin and quercetin was competitive in nature, influencing both kcat and Km of the reaction, with the stronger effect on Km. Inhibition by silibinin, however, showed no concentration dependence and may be taken as a mixed-type of inhibition. Such a case indicates that the inhibitory molecule(s) interacts not only with the same enzyme form as does xanthine (i.e. free enzyme with MoVI, denoted E), but also with a different form in the catalytic sequence of XO, such as the enzyme-substrate or enzyme-product complexes (E·S or E·P). The rate-limiting step in the reaction of XO with xanthine is the release of uric acid, and thus an inhibitor that either interferes with substrate binding (thereby interfering with formation of the E·S complex and altering Km) or blocks product release (thus decreasing kcat) can exert an inhibitory effect on the enzyme. Inhibition by luteolin and quercetin is not surprising when one considers that each molecule bears some resemblance to the known XO substrate lumazine (Figure 6.2); a planar molecule known to follow the same catalytic sequence in its hydroxylation to violapterin (7-hydroxylumazine) as does xanthine in its conversion to uric acid (111). The somewhat stronger inhibition by luteolin versus quercetin is somewhat surprising when one considers their structures, the two differing only by the additional hydroxyl group on the quercetin molecule. For both molecules, hydrogen bonding interactions with active site residues near the

130 molybdenum center of XO may afford inhibitory binding properties as discussed below.

The results for silibinin are consistent with those of luteolin and quercetin, despite the overall size of the molecule. This natural compound possesses a benzopyran ring, similar to the analogous moieties of quercetin and luteolin, which is planar and resembles lumazine. The solvent channel in approaching the molybdenum center of XO is fairly wide, and could easily accommodate such a molecule.

6.3.2 Superoxide production in the presence of luteolin, quercetin, silibinin, curcumin, and allopurinol

The reduction of cytochrome c by superoxide in the course of XO turnover was significantly decreased at each concentration of allopurinol as shown in Table 6.2.

The decrease was most pronounced following incubation of the enzyme with 25 µM allopurinol, as expected.

131

Rate of cyt c reduction (nM · s-1) Luteolin Control 160 ± 0.6 10 µM 100 ± 3 25 µM 82 ± 9 50 µM 71 ± 3 Incubation 110 ± 10 (Control) (170 ± 20)

Quercetin Control 170 ± 1 10 µM 140 ± 20 25 µM 100 ± 7 50 µM 80 ± 6 Incubation 110 ± 7 (Control) (170 ± 20)

Curcumin Control 160 ± 20 10 µM 160 ± 7 25 µM 150 ± 6 50 µM 150 ± 6 Incubation 150 ± 10 (Control) (170 ± 20)

Silibinin Control 160 ± 0.6 10 µM 160 ± 7 25 µM 150 ± 17 50 µM 140 ± 13 Incubation 120 ± 4 (Control) (170 ± 20)

Allopurinol Control 170 ± 10 10 µM 140 ± 10 25 µM 140 ± 6 50 µM 120 ± 9 Incubation 28 ± 1 (Control) (170 ± 20)

Table 6.2. Inhibition studies on the production of superoxide by xanthine oxidase with each natural compound. Superoxide production was monitored following the rate of cyt c reduction at 550 nm. Error is given as one standard deviation from the assays in triplicate.

132 The inhibitory effect of luteolin toward the production of superoxide by XO with xanthine was the most potent of the natural products examined here. The production of superoxide was also reduced at 25 µM and 50 µM quercetin, with resulting decreases analogous to those with luteolin and most significant at 50 µM. In the presence of 50 µM silibinin, the production of superoxide by XO was decreased by approximately 20 % relative to the control at 100 µM xanthine, but still less than with luteolin or quercetin. Superoxide production was not significantly altered by any concentration of curcumin, either initially or following pre-incubation.

Superoxide production by XO occurs by the transfer of reducing equivalents derived from xanthine to O2 at the enzyme’s flavin site. Three of the four compounds tested here appear to decrease the initial production of superoxide following the reaction of xanthine oxidase with xanthine, as followed by a ∆A550 associated with the reduction of cytochrome c. The declines in superoxide generation are proportional to the decreased turnover of xanthine at each concentration of these natural products, and for luteolin and quercetin the decreases are quite significant at 50 µM, at or near

50% of the control (Table 6.2). Considering the proportionality to xanthine turnover, the declines in superoxide production are most likely due to inhibition at the molybdenum-containing active site (which interacts with xanthine) rather than at the

FAD-containing flavin site (where reduced FAD interacts with O2) of the enzyme.

Thus, any contribution of the radical scavenging ability of the inhibitor itself

(inhibitor acting as an antioxidant) or the ability for direct interaction (i.e. interposition) in the flavin site of XO, to the decreased reduction of cytochrome c can be considered minimal on the basis of the present results. It is worth noting also that

133 luteolin, silibinin, and quercetin appear structurally analogous enough to xanthine, and especially lumazine, and are likely interact exclusively at the enzyme’s molybdenum-containing active site rather than at the flavin site. Thus a reduction in superoxide production that parallels the competitive decrease in catalytic reduction of the enzyme by xanthine seems most likely.

6.3.3 Role of luteolin, quercetin, and silibinin in the inhibition of XOR

Various in vivo studies have suggested that luteolin, quercetin, or silibinin may be as potent an inhibitor of xanthine oxidase as is allopurinol, and our inhibition studies were undertaken to address this point. We find that while this is true in vitro on a short timescale in the steady-state, there is no time-dependent increase seen with any of these compounds in inhibition analogous to that with allopurinol (which is the basis for its profound effect on the enzyme). Thus when compared to allopurinol/oxypurinol (and febuxostat as a rising and clinically promising therapeutic) the inhibition of XO by luteolin, silibinin, and quercetin is not nearly as effective as allopurinol. Given the pharmacokinetics of these natural products, it seems that a high sustained level of these compounds would be required for continued inhibition of

XO. Still, one advantage of these natural compounds is that the side effects associated with the intake of large amounts are reportedly minimal. Thus, luteolin, silibinin, and quercetin may serve as promising lead compounds for the development of novel therapeutics that inhibit xanthine oxidase.

Long-term intake of foods rich in luteolin, silibinin, or quercetin may indeed lower levels of xanthine oxidase activity. The many benefits of fruits and vegetables in the daily diet, and of moderate consumption of wine and other food sources of natural products such as flavonoids, has indeed been suggested by many previous 134 reports. These natural products are readily available in many plants and foods.

Nonetheless, our results here do support the role of luteolin, silibinin, and/or quercetin as natural products that inhibit xanthine oxidase and reduce the production of uric acid and of superoxide.

In considering the non-flavonoid curcumin, the lack of inhibition in this study is not surprising, considering the size and shape of the molecule, the size and steric constraints of the solvent access channel to the molybdenum site of the enzyme, and the accepted mechanism by which the enzyme catalyzes the production of uric acid.

Curcumin is not unreasonably large by molecular weight, but the molecule is not planar like xanthine, lumazine, or allopurinol and does not resemble any of these compounds. For steric reasons alone, with a channel that narrows to slightly less than

10 Å for substrate/inhibitor access to the active site, the sp3 hybridized carbon in the structural center of curcumin (Fig. 6.2) would create a roughly 12 Å long molecule with a 109.5o bend trying to move into and through the access channel. Other means of inhibition, e.g. the ability of Febuxostat to block the solvent access channel by intercalating between Phe914 and Phe1009 or of the other three compounds to utilize their benzopyran moieties in a presumably similar manner, is unlikely based on the same steric considerations as mentioned above. Thus with the lack of a plausible mechanistic basis for inhibition and these steric constraints of the molecule and enzyme, our results showing a lack of XO inhibition by curcumin are understandable.

Our efforts here demonstrate inhibition of xanthine oxidase by luteolin, silibinin, and quercetin, and thus suggest that these may be clinically useful inhibitors of XO, particularly as dietary consumables, supporting suggestions previously 135 proposed by others. The nature of this inhibition, particularly the somewhat stronger effect of quercetin and luteolin over that of silibinin, is interesting and worth further characterization. The multiple opportunities for hydrogen bonding on these three molecules undoubtedly facilitate binding to the enzyme. Also, the more bulky silibinin, which does not resemble xanthine, may compromise the effectiveness of this molecule relative to the other two. Structural modeling may lend insight into the potential orientations and interactions of these molecules with specific active site residues, but such software applications would likely require a limited model of the active site and need to consider the flexibility of the ligand and residues near the molybdenum center. Given recent success in our laboratory at obtaining crystal structures of bovine XO with substrate analogs bound in the enzyme’s active site (94), however, it may be possible to obtain structures of XO with luteolin, quercetin, and possibly silibinin bound in the active site. This would allow direct observation of each compound’s interaction(s) with active site residues, and provide further insight not only into the role that these residues may be playing in normal catalysis, but also into new means by which the enzyme may be inhibited.

6.3.4 Steady-state inhibition studies with coumarin, esculetin, and ferulic acid

Table 6.3 gives the kinetic data for the inhibition of XO (or lack thereof) by each compound.

136

-1 kcat (s ) Km (µM) kcat/Km (µM-1·s-1) Coumarin 10 µM 10.5 ± 0.5 1.4 ± 0.4 7.5 ± 2.0 25 µM 10.9 ± 0.7 1.8 ± 0.7 6.1 ± 2.0 50 µM 12.0 ± 0.7 1.3 ± 0.5 9.2 ± 4.0 Incubation 11.9 ± 0.5 1.2 ± 0.3 9.9 ± 2.0 Control 10.5 ± 0.5 1.7 ± 0.5 6.2 ± 2.0

Esculetin 10 µM 10.1 ± 0.4 6.2 ± 1.0 1.6 ± 0.3 25 µM 11.3 ± 0.8 27.0 ± 4.0 0.4 ± 0.04 50 µM 11.4 ± 2.0 44.0 ± 10 0.3 ± 0.06 Incubation 8.0 ± 1.0 9.6 ± 4.0 0.8 ± 0.3 Control 12.3 ± 0.6 1.9 ± 0.5 6.5 ± 2.0

Ferulic acid 10 µM 9.4 ± 0.5 2.4 ± 0.6 3.9 ± 1.0 25 µM 8.9 ± 0.6 2.0 ± 0.8 4.5 ± 2.0 50 µM 7.7 ± 0.4 2.5 ± 0.7 3.1 ± 1.0 Incubation 5.9 ± 0.6 3.1 ± 1.5 1.9 ± 1.0 Control 9.6 ± 0.4 1.5 ± 0.4 6.4 ± 2.0

Allopurinol 10 µM 5.0 ± 0.3 36 ± 8 0.14 ± 0.03 25 µM 4.0 ± 0.4 30 ± 10 0.13 ± 0.04 50 µM 4.3 ± 0.3 39 ± 10 0.11 ± 0.03 Incubation 3.8 ± 0.1 53 ± 7 0.07 ± 0.007 Control 10.9 ± 0.4 1.9 ± 0.3 5.6 ± 1.0

Table 6.3. Kinetic constants for potential steady-state inhibition of xanthine oxidase by coumarin, esculetin, and ferulic acid as contrasted with that by allopurinol. Error is expressed as one standard deviation from the hyperbolic fit to vobs vs. [xanthine].

137 Allopurinol demonstrated its previously established time-dependent inhibition; and incubation of the enzyme with 25 µM allopurinol increased the inhibitory effect as expected. Coumarin did not inhibit xanthine oxidase in vitro, either in ordinary assays or after pre-incubation with the enzyme. Analysis by mass spectrometry showed no change from starting material following overnight incubation with XO, indicating that xanthine oxidase does not hydroxylate or otherwise alter coumarin.

Esculetin demonstrated a competitive mode of inhibition with XO, and showed no time-dependence. The Ki was found to be 6.3 µM as determined from the results of the double-reciprocal plot in Figure 6.7. At 50 µM esculetin, the kcat/Km for xanthine turnover by XO was decreased by 95 %, with the effect entirely attributable to changes in the apparent Km.

Ferulic acid demonstrated weaker competitive inhibition with a Ki of 35 µM.

Interestingly, incubation of XO with 25 µM ferulic acid resulted in significantly greater inhibition, although the reaction of XO with xanthine over a longer time scale in the presence of ferulic acid did not significantly alter ∆A295 from the control. The

15-minute reactions are plotted in Figure 6.8, which further demonstrates the time- dependent nature of inhibition by allopurinol as compared to the lack of time- dependence for the efficacy of esculetin.

138

(A)

(B)

Figure 6.7. Double-reciprocal plots of [XO]/vobs versus 1/[xanthine] for esculetin (panel A) and ferulic acid (panel B). The approximate intersection of the linear fits at x=0 for each concentration of inhibitor indicates that the mode of inhibition is competitive.

139

Figure 6.8. 15-minute reactions with 100 µM xanthine, 5 nM XO, and 50 µM of each potential inhibitor.

140 Potential inhibitors of XOR must be considered in the context of the physical structure of the enzyme, particularly in the molybdenum-containing active site. The manner in which potential substrates and inhibitors may interact with specific residues are essential considerations in designing or proposing molecular binding and enzyme inhibition. As shown in Figure 1.6 and discussed in Chapter 1, the active site residues

Gln767, Glu802, Phe914, Phe1009, Arg880, and Glu1261 are near enough to interact with substrates or inhibitors and have been suggested to be involved in catalysis by

XOR. Previous studies have identified the role(s) played by several of these residues

(73, 93). Direct stacking interactions between Phe914 and Phe1009 seem to be a general feature in the binding of molecules in the active site of XOR, and this constitutes a good starting point for the consideration of competitive inhibition of the enzyme. The solvent access channel (lined by Phe649, Gln1122, Val1011, and

His875 as shown in Figure 6.9) to the molybdenum is quite large, and can accommodate a variety of potential substrates and inhibitory compounds, although planar compounds are expected to interpose between the phenylalanine residues most effectively.

141

(A)

(B)

Figure 6.9. Active site of xanthine oxidase with the bound catalytic intermediate for the slow-substrate 2-hydroxy-6-methylpurine (HMP) shown. Important active site residues are labeled, as well as representative residues around the solvent access channel (Phe649, His875, Gln1122, and Val1022) through which any substrate or inhibitor would pass to access the molybdenum-site. Panel (B) is looking down into the active site from the surface of the enzyme. Nitrogens are in blue, oxygens in red, molybdenum in teal, phosphorus in orange, and sulfur in gold. (Adapted from 94, PDB code 3B9J) 142 The mode of XO inhibition by esculetin and ferulic acid is competitive.

Although it is unlikely given the structure of these compounds and our mass spectrometry data on coumarin that XO acts to alter the structure of these compounds, it does seem likely that they interpose between Phe914 and Phe1009 in blocking the active site of xanthine oxidase. Once in this position, reasonable models of the complexes have the hydroxyl groups on both compounds hydrogen-bonding with active site amino acids such as glutamates 802 and 1261 and arginine 880 (bovine numbering). Such interactions may explain the discrepancy between the 15-minute assays and the 10-minute incubation assays with ferulic acid. Under conditions where xanthine and ferulic acid are presented simultaneously to the enzyme, the strong preference for xanthine overcomes any potential inhibitory binding by ferulic acid in the active site. This is the case of the 15-minute assays. In the other case with pre- incubation of enzyme with inhibitor in the absence of xanthine, ferulic acid may initially bind but be displaced by xanthine during subsequent turnover. Regardless, both esculetin and ferulic acid were shown to be competitive inhibitors of XO with esculetin being the much more potent inhibitory molecule. The extent of inhibition for either esculetin or ferulic acid in vitro is significant at low substrate concentrations

(e.g. physiological conditions) as assessed by the effect of each compound on kcat/Km

(70).

Our results demonstrate that at least some coumarin derivatives act as inhibitors of XOR. They also highlight the requirements for such inhibition given the nature of the molybdenum-containing active site. Compounds that are physically able to interpose between the two phenylalanine residues that bookend the approach to the molybdenum site are most likely to have the potential for inhibition there. Structural features of such compounds that may prevent access through this approximately 10- 143 angstrom channel compromise effective inhibition. Furthermore, the ability of a compound to interact by hydrogen-bonding or otherwise with active site residues

(particularly those indicated in Figure 6.9) must be taken into consideration when assessing potential inhibitors of XO or XDH. The structurally demonstrated mode of binding substrates and established inhibitors of the enzyme should also be taken into account when investigating potential inhibition of XOR. Esculetin (6,7- dihydroxycoumarin) demonstrated significant inhibition of the steady-state turnover of xanthine whereas coumarin itself did not, and presumably this is due to hydrogen- bonding interactions involving the additional hydroxyl groups of esculetin with active site residues. Ferulic acid, which may be considered as a ring-opened form of coumarin, is a weaker inhibitor due to the replacement of a hydroxyl with a methoxy group, and this may explain its decreased inhibitory properties.

Finally, our results suggest that the observed decrease in the production of superoxide in the presence of ferulic acid is due to the innate radical-scavenging ability of this compound and not via the inhibition of substrate hydroxylation by XO.

In cases where physiological substrates are present, inhibition by ferulic acid is negligible particularly over longer time scales. It thus remains important to investigate antioxidants proposed to inhibit xanthine oxidoreductase in order to elucidate whether the antioxidant properties are due to innate scavenging, inhibition of the enzyme, or perhaps both (as in the case of allopurinol).

Future structural work on the interactions of inhibitory compounds with XOR would be useful in elucidating further insights into the requirements for and design of pharmaceutical and non-pharmaceutical inhibitors of the enzyme. In addition to kinetic analyses, our own laboratory has obtained crystal structures of bovine XO, and may move to obtain structures with bound inhibitory compounds in the near future. 144

6.3.5 Potential reduction of the molybdenum center and mass spectrometry data

XO was not reduced by any of the compounds examined here when incubated under anaerobic conditions. The lack of any decrease in the absorbance at 450 nm exhibited by oxidized enzyme indicates that none of the compounds here are mechanism-based inhibitors, as is allopurinol. Similarly, incubation of XO with luteolin, quercetin, silibinin, curcumin, or the coumarin derivatives alone yielded no ∆A295 or ∆A550 for our two experiments. The compounds alone and with an equal concentration of xanthine without enzyme also demonstrated no spectral change at 295 nm. Although cytochrome c seemed to be reduced slowly in the presence of XO plus quercetin, the overall decrease in the initial rate of cytochrome c reduction relative to the control suggests that inhibition of superoxide production predominates over any other initial

∆A550 in the presence of the physiological substrate xanthine. Mass spectrometry following overnight incubation with coumarin showed no apparent change between the starting reactants and final molecules following the removal of enzyme.

This all indicates that none of these compounds are hydroxylated by the enzyme. Thus we conclude that luteolin, silibinin, and quercetin inhibit xanthine oxidase by competing with xanthine and interposing in a noncatalytic fashion in the active site, a manner of inhibition that is reminiscent of Febuxostat.

145

CHAPTER 7

CONCLUSIONS

7.1 Summary

The present thesis project has been aimed at elucidating the relationship between structure and function of the metalloenzyme xanthine oxidoreductase. We have utilized several techniques and approaches to investigate the manner in which structure relates to function, and to test specific hypotheses regarding the catalytically effective orientation of substrate in the active site. A kinetic analysis of an R310M mutant led to the proposal that this active site arginine (Arg880 in the bovine enzyme) stabilizes the transition state for those substrates that can accommodate negative charge density at their C6 or analogous positions, and that this residue allows discrimination between good and poor substrates by the enzyme. Subsequently, we obtained crystallographic evidence that directly supported our conclusions regarding such substrate orientation. Our efforts next involved a study of enzyme inhibition by various natural products based on our insights into the structure of the enzyme.

Xanthine oxidoreductase is an approximately 300 kDa molybdenum- containing protein that catalyzes the final two steps in purine catabolism in humans.

The enzyme is a constitutive cytosolic protein that exists functionally as a homodimer.

It is found in all human cells, although its concentration is highest in the hepatic and gastroinstestinal systems. In addition to catalyzing the conversion of hypoxanthine to 146 xanthine and then xanthine to uric acid, the enzyme is also able to catalyze the concurrent conversion of NAD+ to NADH and to form superoxide from molecular oxygen. The enzyme is also of great mechanistic interest, as the reaction mechanism involves hydroxylation of a carbon center in a manner that produces rather than consumes usable reducing equivalents.

XOR possesses four redox-active sites in each monomer. The substrate to be hydroxylated introduces reducing equivalents at the molybdenum center, and following reduction of the molybdenum electrons are passed sequentially to the first of two [2Fe2S] clusters and then to an FAD in a separate site. With the number of electron-accepting centers at four per monomer, it takes three equivalents of substrate molecules (for a total of six reducing equivalents) to fully reduce each monomer of the enzyme.

7.1.1 The role of arginine 880/310 in catalysis

Substrate is catalytically hydroxylated at the molybdenum center of XOR.

The generally accepted mechanism of this catalysis is depicted in Figure 1.4 and is proposed to involve base-assisted nucleophilic attack by the planar Mo-OH ligand with the help of Glu 1261 (bovine numbering). Several other active site residues are also thought to play integral roles in the enzyme’s mechanism during catalysis. Our work has focused on the role of Arg 880 (Arg 881 in the human enzyme), a residue that is positioned some 10 Å from the site of hydroxylation yet is likely to interact with the heterocycle of the alloxanthine-inhibited structure of XOR (73). We find that this residue contributes approximately 4.5 kcal/mol toward transition state stabilization, where product is coordinated to the Mo center and negative charge density is directed out toward the substrate’s C6 position. The ability of this arginine 147 residue to stabilize a transition state in this manner also allows the enzyme to discriminate good versus poor substrates. Good substrates for the wild-type enzyme are those that have functional groups at their C6 (purine) or C4 (lumazine) positions capable of accepting and stabilizing negative charge, and these substrates bind such that this group is in close proximity to Arg 880. Poor substrates lack such an electronegative functional group, and bind in an orientation opposite (i.e. upside- down) to that of good substrates.

Our work here has provided experimental support for this means of substrate discrimination. Our rapid-reaction kinetic studies indicate that good substrates such as xanthine and 6-thioxanthine are markedly affected by mutation of Arg 880 to

4 methionine (where kred is reduced by as much as 10 ), while the poor substrates 2- hydroxy-6-methylpurine and 2,6-diaminopurine are relatively unaffected (kred reduced

10-fold or less). Our crystallographic work has indeed demonstrated that one poor substrate (2-hydroxy-6-methylpurine) does indeed bind with its C6 position directed opposite Arg 880, whereas both xanthine and lumazine (which are good substrates) bind with their C6 (C4) carbonyl groups directed toward Arg 880.

Such results indicate that it is not the case that all substrates bind in the same orientation as HMP, a proposal initially based on the structure of the alloxanthine- inhibited enzyme from Rhodobacter capsulatus, which helped to identify the several active site residues that may interact with substrates, as well as the structure of mutant

(demolybdo) enzyme with uric acid (PDB codes 1JRP and 2E3T). In both of these structures, the heterocycle is oriented with its C6 or analogous position directed away from Arg 880. As shown in Figure 4.8, however, alloxanthine is coordinated directly to the Mo atom (with no bridging oxygen) and uric acid is present in a structure without the molybdenum cofactor, and it is apparent that both molecules are shifted 148 by nearly 0.5 to 1.0 Å further into the solvent access channel relative to substrates in the presence of a catalytically intact molybdenum cofactor. Our observation of a catalytic intermediate in the turnover of HMP is further confirmation that our structures represent a snapshot of the functional molybdenum center of XOR during true catalysis.

7.1.2 The two forms of xanthine oxidoreductase

The conversion of XDH to XO is higly relevant in the context of human pathology. The production of reactive oxygen species (ROS) such as superoxide and hydrogen peroxide by any system in the human body is deleterious in some sense, and there are many possible clinical manifestations of ROS production in human pathology. XOR is a primary producer of ROS in many settings, particularly in cases of ischemia-reperfusion injury. The enzyme exists initially and primarily as XDH in the human body, whereas XO is the predominant producer of ROS between the two forms of XOR and is the form implicated in many human disease processes. We have investigated the functional differences between the two forms of the enzyme, focusing on the pH dependence of catalysis for each respective form. Our findings provide insight into the nature of ROS production by XO, showing that the oxidase functions with increasing efficiency in going from pH 7.0 up to 7.5 in the presence of large amounts of substrate, whereas XDH activity is declining over this same pH range.

This suggests that following an ischemic insult where the pH falls from a physiological value of 7.4 down to 7.0, both XDH and XO share a common pH for optimal activity, meanwhile substrates for the enzyme are building up with continued hypoxia and cellular necrosis. Upon reperfusion as the pH rises in this environment now rich in XOR substrates, XO is expected to be more efficient than XDH at 149 catalyzing purine degradation, thus favoring the production of superoxide and peroxide over NADH.

Our results also suggest that as NAD+ is depleted in hypoxic cells, XDH might also reduce molecular oxygen in the absence of its preferred electron acceptor. Our investigations on oxygen consumption by XDH have shown that indeed XDH can utilize molecular oxygen and produce a significant amount of superoxide in the absence of NAD+. Thus in a case of ischemia with subsequent reperfusion, both forms of the enzyme may generate superoxide, with the reactivity of the oxidase form of the enzyme favored at increasing pH.

7.1.3 Inhibition of xanthine oxidoreductase

A brief search of the literature reveals a large number of compounds identified as putative inhibitors of xanthine oxidoreductase. These molecules range from those produced by plants to a host of novel synthetic compounds. Most of these studies were conducted entirely in an in vivo setting, however, with no in vitro confirmation using the native enzyme and very little mechanistic insight into the nature of the proposed inhibition. This also made it difficult to assess the degree of inhibition, particularly with respect to allopurinol, and made practical interpretation of the results difficult. We therefore looked at several of these compounds in in vitro assays with the wild-type enzyme, choosing to focus on two broader catagories of molecules: flavonoids and coumarin derivatives. Each of these groups contains several molecules proposed to be inhibitors of xanthine oxidoreductase.

Our results with the flavonoids luteolin, quercetin, and silibinin demonstrated these were indeed competitive inhibitors (with Ki values of 1.9 and 1.2 µM for luteolin and quercetin, respectively), in effective agreement with previous studies 150 (although the practical applications of these compounds themselves as inhibitors of

XOR do not seem to be as promising as would be inferred from previous work).

These compounds also suppressed the production the superoxide by xanthine oxidase, although the effect was largely the result of a reduced overall catalytic throughput.

The inhibition of XO by these compounds is reasonable on the basis of their general similarity to known substrates of XOR.

Perhaps the most immediate application of these results is the confirmation of dietary flavonoids such as luteolin as being of potential benefit. Our results suggest that consumption of vegetables and certain teas, all sources of flavonoids, may reduce the physiological production of ROS. Our results support this as well as the continual suppression of XOR activity, thus possibly befitting those suffering from chronic conditions such as gouty arthritis. Our results also suggest that these molecules may serve well as promising lead compounds for the derivitization of novel XOR inhibitors.

Our investigations of coumarin derivatives as inhibitors of XO were prompted by previous, unpublished work with ticlopidine, clopidogrel, and coumarin.

Importantly, our work has demonstrated that coumarin is not an effective inhibitor of bovine xathine oxidase. This result is at variance with reports in the literature based on in vivo studies.

The coumarin derivatives esculetin and ferulic acid were found to inhibit XO- catalyzed turnover of xanthine to uric acid. In line with previous in vivo studies, esculetin was the most effective inhibitor in our study (with a Ki of 6.3 µM), and ferulic acid was found to be a much less potent inhibitor of xanthine turnover (with a

Ki of 35 µM). Our results also highlight the importance of in vitro biochemical analysis in drug development. Any proposed modulatory compound should be 151 assayed for activity with the native target enzyme, in and apart from any in vivo experimentation.

7.2 Future Directions

With new insights into the structure of XOR as well as into its function in the human body, future directions could be aimed at understanding the interplay between

XOR and other proteins. Specifically, phosphorylation and dephosphorylation are perhaps the most common post-translational modifications of enzymes, and an enzyme the size of XOR will have many potential sequence sites for potential interaction with kinases and phosphatases. That XOR may be a phospho-enzyme has been suggested by several studies as discussed in Chapter 1, and future work could focus on methods to establish any site(s) of phosphorylation or dephosphorylation in the structure of XO. It is even possible, although not apparently suggested at present in the literature, that other post-translational modifications could take place with XDH and/or XO in or outside of the cell, and investigations as to the nature of these modifications, as well as the effect on enzymatic activity, may also be warranted.

Another avenue involves examining the possible interactions between XOR and other cellular constituents. The cytoplasmic enzyme is also known to be secreted under physiological conditions of milk secretion, and may also be secreted into circulation and interstitial fluid under conditions of pathologic stress. Once secreted, it has been suggested by others that XOR binds vascular endothelial cells with possible deleterious effects (37, 159). As a specific future aim, the characterization of

XOR binding to an extracellular scaffold could be pursued. If XOR does indeed bind to glycyosaminoglycans and/or other protein-based components of the extracellular matrix, the nature of these binding interactions and corresponding effect(s) on activity 152 and/or inhibition of the enzyme would be quite interesting. This area of research could be approached crystallographically, either by crystallization followed by soaking in (as before) or co-crystallizing with a soluble proteoglycan analog. Dextran sulfate has been used in previous non-crystallographic experiments to serve as such an analog, and has the advantage of being quite water soluble, which make it a logical choice here.

Future structural and functional work may follow from our investigations of active site residues in the molybdenum center, such as Glu802, Gln767, and others.

With a working method for substrate-soaking, determination of the crystal structure of

XO with bound hypoxanthine would be a realistic, worthwhile pursuit. Such a structure for a species hydroxylated at C2 (while only having a pre-exisiting functional group at C6) may reveal interactions with and different roles for residues from those for the hydroxylation of the C8 position of xanthine.

Crystallographic and kinetic work may also be warranted to probe the role(s) of residues surrounding the flavin site of the enzyme. Obtaining a structure of NAD+ or NADH, or any amendable electron acceptor, in or adjacent to the flavin site would provide direct insight into the nature of this electron transfer process, and identify the role of specific amino acids at this part of the protein.

153

LIST OF REFERENCES

1. Emsley, John (2001) Nature's Building Blocks. Oxford: Oxford University Press, 262-266.

2. Urdang, G. (1958) The Apothecary Chemist, Carl Wilhelm Scheele, 2nd ed.

3. Hjelm, P.J. (1788). "Versuche mit Molybdäna, und Reduction der selben Erde". svenska vetensk. Academ. Handlingar 49: 268.

4. Considine, Glenn D., ed. (2005) "Molybdenum", Van Nostrand's Encyclopedia of Chemistry, New York: Wylie-Interscience, 1038-1040.

5. Mitchell, P.C.H. (2003) “‘Molybdenum’ in ‘It’s Elemental: The Periodic Table’” Chemical & Engineering News

6. Mendel, R.R. and Bittner, F. (2006) Cell Biology of Molybdenum. Biochim. Biophys. Acta 1763, 621 – 635.

7. Hille, R. (1996) The Mononuclear Molybdenum Enzymes. Chem. Rev. 96, 2757 – 2816.

8. Mendel, R.R. and Hänsch, R. (2002) Molybdoenzymes and molybdenum cofactor in plants. Journal of Experimental Botany 53, 1689-1698.

9. Gangeswaran, R., Lowe, D.J., and Eady, R.R. (1993) Purification and characterization of the assimilatory nitrate reductase of Azotobacter vinelandii. Biochem. J. 289, 335 - 342.

10. Cabello, P., Rolda´n, M.D., and Moreno-Vivia´n, C. (2004) Nitrate reduction and the nitrogen cycle in archaea. Microbiology 150, 3527 – 3546.

11. Hille, R. (1999) Molybdenum Enzymes. Essays Biochem. 34, 125-137.

12. Hille, R. (2006) Structure and Function of Xanthine Oxidoreductase. Eur. J. Inorg. Chem. 2006, 1913–1926.

13. Schardinger, F. (1902) Z. Untersuch. Nahrungs- u. Genussmittel, 5, 1113.

14. Dixon, M. and Thurlow, S. (1924) Studies on Xanthine Oxidase. Biochem. J., 18, 976 - 988.

154 15. Booth, V.H. (1938) LXV. The Specificity of Xanthine Oxidase. Biochem. J. 32, 494 – 502.

16. Hille, R. (2005) Molybdenum-containing Hydroxylases. Arch. Biochem. Biophys. 433, 107 116.

17. Enroth, C., Eger, B.T., Okamoto, K., Nishino, To., Nishino, Ta., and Pai, E.F. (2000) Crystal structures of bovine milk xanthine dehydrogenase and xanthine oxidase: Structure-based mechanism of conversion. Proc. Nat. Acad. Sci. USA 97, 10723 – 10728.

18. Berry, C.E. and Hare, J.M. (2004) Xanthine oxidoreductase and cardiovascular disease: molecular mechanisms and pathophysiological implications. J. Physiol. 555.3, 589 – 606.

19. Delia Corte, E. and Stirpe, F. (1968) The regulation of xanthine oxidase in rat liver: Modification of the enzyme activity of rat liver supernatant on storage at – 20oC. Biochem. J. 108, 349 – 351.

20. Waud, W.R. and Rajagopalan, K.V. (1976) Purification and properties of the + NAD -dependent (Type D) and O2-dependent (Type O) forms of rat liver xanthine dehydrogenase. Arch. Biochem. Biophys. 172, 354 – 364.

21. Harrison, R. (2006) Milk xanthine oxidase: Properties and physiological roles. Int. Dairy J. 16, 546 – 554.

22. Linder N., Rapola J., Raivio K.O. (1999) Cellular expression of human xanthine oxidoreductase protein in normal human tissues. Lab. Invest. 79, 967 – 974.

23. Li, H., Samouilov, A., Liu, X., and Zweier, J.L. (2004) Characterization of the effects of oxygen on xanthine oxidase-mediated nitric oxide formation. J. Biol. Chem. 279, 16939 - 16946.

24. Boueiz, A., Damarla, M., and Hassoun, P.M. (2008) Xanthine oxidoreductase in respiratory and cardiovascular disorders. Am. J. Physiol. Lung Cell. Mol. Physiol. 294, L830 - L840.

25. Hannah M. Martin, John T. Hancock, Vyv Salisbury, and Roger Harrison (2004) Role of xanthine oxidoreductase as an antimicrobial agent. Infect. Immun. 72, 4933 - 4939.

26. Stevens, C. R., Millar, T.M., Clinch, J.G., Kanczler, J.M., Bodamyali, T., and Blake, D.R. (2000) Antibacterial properties of xanthine oxidase in human milk. Lancet 356:829–830.

27. Kasper, D.L., Braunwald, E., Fauci, A.S., Hauser, S.L., Longo, D.L., Jameson, J.L. (2008) Harrison's Principles of Internal Medicine. New York: McGraw- Hill. 155

28. Smalley, R.V., Guaspari, A., Haase-Statz, S., Anderson, S.A., Cederberg, D., and Hohneker, J.A. (2000) Allopurinol: Intravenous Use for Prevention and Treatment of Hyperuricemia. J. Clin. Oncol. 18, 1758 - 1763.

29. Puig, J.G., Torres, R.J., Mateos, F.A., Ramos, T.H., Arcas, J.M., Buño, A.S., and O'Neill, P. (2001) The spectrum of hypoxanthine-guanine phosphoribosyltransferase (HPRT) deficiency. Clinical experience based on 22 patients from 18 Spanish families. Medicine (Baltimore) 80, 102 - 112.

30. Stout, J.T. and Caskey, C.T. (1985) HPRT: GENE STRUCTURE, EXPRESSION, AND MUTATION Ann. Rev. Genet. 19, 127 – 151.

31. Pacher, P., Nivorozhkin, A., and Szabó, C. (2006) Therapeutic Effects of Xanthine Oxidase Inhibitors: Renaissance Half a Century after the Discovery of Allopurinol. Pharmacol. Rev. 58, 87 - 114.

32. Harris, C.M. and Massey, V. (1997) The Reaction of Reduced Xanthine Dehydrogenase with Molecular Oxygen. REACTION KINETICS AND MEASUREMENT OF SUPEROXIDE RADICAL. J. Biol. Chem. 272, 8370 - 8379.

33. Harris, C.M. and Massey, V. (1997) The Oxidative Half-reaction of Xanthine Dehydrogenase with NAD; Reaction Kinetics and Steady-state Mechanism J. Biol. Chem. 272, 28335 – 28341.

34. McCord, J.M. (1985) Oxygen-Derived Free Radicals in Post-Ischemic Tissue Injury. New Eng. J. Med. 312, 159 - 163.

35. Yellon, D.M. and Hausenloy, D.J. (2007) Myocardial Reperfusion Injury. New Engl. J. Med. 357, 1121 – 1135.

36. Weinbroum, A., Nielsen, V.G., Tan, S., Gelman, S., Matalon, S., Skinner, K.A., Bradley, Jr, E., and Parks, D.A. (1995) Liver ischemia-reperfusion increases pulmonary permeability in rat: role of circulating xanthine oxidase. Am. J. Physiol. Gastrointest. Liver Physiol. 268, 988 - 996.

37. Houston, M., Estevez, A., Chumley, P., Aslan, M., Marklund, S., Parks, D.A., and Freeman, B.A. (1999) Binding of Xanthine Oxidase to Vascular Endothelium. KINETIC CHARACTERIZATION AND OXIDATIVE IMPAIRMENT OF NITRIC OXIDE-DEPENDENT SIGNALING J. Biol. Chem. 274, 4985 - 4994.

38. Roy, S.K., Korzekwa, K.R., Gonzalez, F.J., Moschel, R.C., Dolan, M.E. (1995) Human liver oxidative metabolism of O6-benzylguanine. Biochemical Pharmacology 50, 1385 – 1389.

156 39. Baudry, N., Laemmel, E., and Vicaut, E. (2008) In vivo reactive oxygen species production induced by ischemia in muscle arterioles of mice: involvement of xanthine oxidase and mitochondria. Am. J. Physiol. Heart Circ. Physiol. 294, H821 - H828.

40. Jian-Mei Li and Ajay M Shah (2004) Endothelial cell superoxide generation: regulation and relevance for cardiovascular pathophysiology. Am. J. Physiol. Regulatory Integrative Comp. Physiol. 287, R1014 - R1030.

41. Cejkova, J., Vejrazkab, M., Platenıkb, J., Stıpekb, S. (2004) Age-related changes in superoxide dismutase, glutathione peroxidase, catalase and xanthine oxidoreductase/xanthine oxidase activities in the rabbit cornea. Experimental Gerontology 39, 1537 – 1543.

42. Lin, H.C., Tsai, S.H., Chen, C.S., Chang, Y.C., Lee, C.M., Lai, Z.Y., Lin, C.M. (2008) Structure–activity relationship of coumarin derivatives on xanthine oxidase-inhibiting and free radical-scavenging activities. Biochemical Pharmacology 75, 1416 - 1425.

43. Gupta, S., Rodrigues, L.M., Esteves, A.P., Oliveira-Campos, A.M.F., São José Nascimento, A., Nazareth, N., Cidade, H., Neves, M.P., Fernandes, E., Pinto, M., Cerqueira, N.M.F.S.A., Brás, N. (2008) Synthesis of N-aryl-5-amino-4- cyanopyrazole derivatives as potent xanthine oxidase inhibitors. European Journal of Medicinal Chemistry 43, 771 – 780.

44. Okamoto, K., Eger, B.T., Nishino, To., Kondo, S., Pai, E.F., and Nishino, Ta. (2003) An Extremely Potent Inhibitor of Xanthine Oxidoreductase. CRYSTAL STRUCTURE OF THE ENZYME-INHIBITOR COMPLEX AND MECHANISM OF INHIBITION. J. Biol. Chem. 278, 1848 – 1855.

45. Bruce, S.P. (2006) Febuxostat: A Selective Xanthine Oxidase Inhibitor for the Treatment of Hyperuricemia and Gout. Ann. Pharmacother. 40, 2187 - 2194.

46. Becker, M.A., Schumacher, Jr., H.R., Wortmann, R.L., MacDonald, P.A., Eustace, D., Palo, W.A., Streit, J., and Joseph-Ridge, N. (2005) Febuxostat Compared with Allopurinol in Patients with Hyperuricemia and Gout. N. Engl. J. Med. 353, 2450 - 2461.

47. Ball, E.G. (1939) XANTHINE OXIDASE: PURIFICATION AND PROPERTIES. J. Biol. Chem. 128, 51 – 67.

48. Richert, D.A. and Westerfeld, W.W. (1954) THE RELATIONSHIP OF IRON TO XANTHINE OXIDASE. J. Biol. Chem. 209, 179 – 189.

49. Bray, R.C. and Meriwether, L. (1966) Electron Spin Resonance of Xanthine Oxidase Substituted with Molybdenum-95. Nature 212, 467 - 469.

50. Rajagopalan, K.V. and Handler, P (1964) Hepatic Aldehyde Oxidase J. Biol. Chem. 239, 2027 – 2035. 157

51. Andrews, P., Bray, R.C., Edwards, P., and Shooter, K.V. (1964) The chemistry of xanthine oxidase. 11. Ultracentrifuge and gel-filtration studies on the milk enzyme. Biochem. J. 93, 627 - 632.

52. Eger, B.T., Okamoto, K., Enroth, C., Sato, M., Nishino, To., Pai, E.F., and Nishino, Ta. (2000) Purification, crystallization and preliminary X-ray diffraction studies of xanthine dehydrogenase and xanthine oxidase isolated from bovine milk. Acta Crystallographica D56, 1656 – 1658.

53. Stockert, A.L., Shinde, S.S., Anderson, R.F., and Hille, R. (2002) The Reaction Mechanism of Xanthine Oxidase: Evidence for Two-Electron Chemistry Rather Than Sequential One-Electron Steps. J. Am. Chem. Soc., 124, 14554 - 14555.

54. Massey, V., Brumby, P.E., Komai, H., Palmer, G. (1969) Studies on milk xanthine oxidase. Some spectral and kinetic properties. J. Biol. Chem. 244, 1682 - 1691.

55. Palmer, G., Bray, R.C., Beinert, H. (1964) Direct Studies on the Electron Transfer Sequence in Xanthine Oxidase by Electron Paramagnetic Resonance Spectroscopy. I. Techniques and Description of Spectra J. Biol. Chem. 239, 2657 – 2666.

56. Palmer, G., Bray, R.C., Beinert, H. (1964) Direct Studies on the Electron Transfer Sequence in Xanthine Oxidase by Electron Paramagnetic Resonance Spectroscopy. II. Kinetic Studies Employing Rapid Freezing J. Biol. Chem. 239, 2667 – 2676.

57. Orme-Johnson, W.H. and Beinert, H. (1969) Heterogeneity of paramagnetic species in two iron-sulfur proteins: Clostridium pasteurianum ferredoxin and milk xanthine oxidase. Biochem. Biophys. Res. Comm. 36, 337 – 344.

58. Lowe, D.J., Lynden-Bell, R.M. and Bray, R.C. (1972) Spin-Spin Interaction between Molybdenum and One of the Iron-Sulphur Systems of Xanthine Oxidase and its Relevance to the Enzymic Mechanism. Biochem. J. (1972) 130, 239-249.

59. Olson, J.S., Ballou, D.P., Palmer, G., and Massey, V. (1974) The Mechanism of Action of Xanthine Oxidase. J. Biol. Chem. 249, 4363 – 4382.

60. Keith N. Murray, J. Gary Watson, and Sterling Chaykin (1966) Catalysis of the Direct Transfer of Oxygen from Nicotinamide N-Oxide to Xanthine by Xanthine Oxidase. J. Biol. Chem. 241, 4798 – 4801.

61. Xia, M., Dempski, R., and Hille, R. (1999) The Reductive Half-reaction of Xanthine Oxidase. J. Biol. Chem. 274, 3323 – 3330.

158 62. Hille, R. and Sprecher, H. (1987) On the mechanism of action of xanthine oxidase. Evidence in support of an oxo transfer mechanism in the molybdenum-containing hydroxylases. J. Biol. Chem. 262, 10914 – 10917.

63. Tullius, T.D., Kurtz, Jr., D.M., Conradson, S.D., Hodgson, K.O. (1979) The molybdenum site of xanthine oxidase. Structural evidence from x-ray absorption spectroscopy. J. Am. Chem. Soc. 101, 2776 - 2779.

64. Huber, R., Hof, P., Duarte, R.O., Moura, J.J.G., Moura, I., Liu, M.Y., LeGall, J., Hille, R., Archer, M., and Romão, M.J. (1996) A structure-based catalytic mechanism for the xanthine oxidase family of molybdenum enzymes. Proc. Nat. Acad. Sci. USA 93, 8846 - 8851.

65. Page, C.C., Moser, C.C., Chen, X. and Dutton, P.L. (1999) Natural engineering principles of electron tunnelling in biological oxidation-reduction. Nature 402, 47 – 52.

66. McWhirter, R.B. and Hille, R. (1991) The reductive half-reaction of xanthine oxidase. Identification of spectral intermediates in the hydroxylation of 2- hydroxy-6-methylpurine. J. Biol. Chem. 266, 23724 – 23731.

67. Ryan, M.G., Ratnam, K., and Hille, R. (1995) The Molybdenum Centers of Xanthine Oxidase and Xanthine Dehydrogenase. J. Biol. Chem. 270, 19209 – 19212.

68. Edmondson, D., Ballou, D., Van Heuvelen, A., Palmer, G., and Massey, V. (1973) Kinetic Studies on the Substrate Reduction of Xanthine Oxidase. J. Biol. Chem. 248, 6135 – 6144.

69. Hunt, J. and Massey, V. (1994) Studies of the reductive half-reaction of milk xanthine dehydrogenase. J. Biol. Chem. 269, 18904 – 18914.

70. Fersht, A. (1998) “Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein Folding”, W. H. Freeman.

71. Hille, R. and Massey, V. (1981) Studies on the oxidative half-reaction of xanthine oxidase. J. Biol. Chem. 256, 9090 – 9095.

72. Leimkühler, S., Stockert, A.L., Igarashi, K., Nishino, T., and Hille, R. (2004) The Role of Active Site Glutamate Residues in Catalysis of Rhodobacter capsulatus Xanthine Dehydrogenase. J. Biol. Chem. 279, 40437 – 40444.

73. Truglio, J.J., Theis, K., Leimkühler, S., Rappa, R., Rajagopalan, K.V., and Kisker, C. (2002) Crystal Structures of the Active and Alloxanthine-Inhibited Forms of Xanthine Dehydrogenase from Rhodobacter capsulatus. Structure 10, 115 - 125.

159 74. Choi, E.Y., Stockert, A.L., Leimkühler, S. and Hille, R. (2004) Studies on the mechanism of action of xanthine oxidase. J. Inorg. Biochem. 98, 841 - 848.

75. Ilich, P. and Hille, R. (1997) Tautomerization of the substrate heterocycle in the course of the reaction of xanthine oxidase. Inorganica Chimica Acta 263, 87 – 93.

76. Kim, J.H., Ryan, M.G., Knaut, H., and Hille, R. (1996) The Reductive Half- reaction of Xanthine Oxidase. J. Biol. Chem. 271, 6771 – 6780.

77. Ellis, P.J., Conrads, T., Hille, R., and Kuhn, P. (2001) Crystal Structure of the 100 kDa Arsenite Oxidase from Alcaligenes faecalis in Two Crystal Forms at 1.64 Å and 2.03 Å. Structure 9, 125 – 132.

78. Kisker, C., Schindelin, H., and Rees, D.C. (1997) MOLYBDENUM- COFACTOR–CONTAINING ENZYMES: Structure and Mechanism. Annu. Rev. Biochem. 66, 233 – 267.

79. Romão, M.J., Archer, M., Moura, I., Moura, J.J.G., LeGall, J., Engh, R., Schneider, M., Hof, P., and Huber, R. (1995) Crystal Structure of the Xanthine Oxidase-Related Aldehyde Oxido-Reductase from D. gigas. Science 270, 1170 – 1176.

80. McNaughton, R.L., Helton, M.E., Rubie, N.D., Kirk, M.L. (2000) The Oxo-Gate Hypothesis and DMSO Reductase: Implications for a Psuedo- Bonding Interaction Involved in Enzymatic Electron Transfer. Inorg. Chem. 39, 4386 - 4387.

81. Burgmayer, S.J.N., Kim, M., Petit, R., Rothkopf, A., Kim, A., BelHamdounia, S., Hou, Y., Somogyi, A., Habel-Rodriguez, D., Williams, A., Kirk, M.L. (2007) Synthesis, characterization, and spectroscopy of model molybdopterin complexes. Journal of Inorganic Biochemistry 101, 1601 – 1616.

82. Nishino, T. and Okamoto, K. (2000) The role of the [2Fe–2S] cluster centers in xanthine oxidoreductase. J. Inorg. Biochem. 82, 43 – 49.

83. Hille, R. and Massey, V. (1991) The kinetic behavior of xanthine oxidase containing chemically modified flavins. J. Biol. Chem. 266, 17401 – 17408.

84. Leimkühler, S., Hodson, R., George, G.N., and Rajagopalan, K.V. (2003) Recombinant Rhodobacter capsulatus Xanthine Dehydrogenase, a Useful Model System for the Characterization of Protein Variants Leading to Xanthinuria I in Humans. J. Biol. Chem. 278, 20802 – 20811.

85. Olson, J.S., Ballou, D.P., Palmer, G., and Massey, V. (1974) The Reaction of Xanthine Oxidase with Molecular Oxygen. J. Biol. Chem. 249, 4350 – 4362.

160 86. Edmondson, D., Massey, V., Palmer, G., Beacham, III, L.M., and Elion, G.B. (1972) The Resolution of Active and Inactive Xanthine Oxidase by Affinity Chromatography. J. Biol. Chem. 247, 1597 – 1604.

87. Harris, C.M. and Massey, V. (1997) Kinetic Isotope Effects and Electron Transfer in the Reduction of Xanthine Oxidoreductase with 4- Hydroxypyrimidine. A COMPARISON BETWEEN OXIDASE AND DEHYDROGENASE FORMS. J. Biol. Chem. 272, 22514 – 22525.

88. Morell, D.B. (1952) The nature of catalytic activities of milk xanthine oxidase. Biochem. J. 51, 657 - 666.

89. Hart, L.I., McGartoll, M.A., Chapman, H.R., AND Bray, R.C. (1970) The composition of milk xanthine oxidase. Biochem. J. 116, 851 – 864.

90. Bordas, J., Bray, R.C., Garner, C.D., Gutteridge, S. and Hasnain, S.S. (1980) X- ray absorption spectroscopy of xanthine oxidase. The molybdenum centres of the functional and the desulpho forms. Biochem. J. 191, 499 – 508.

91. Wahl, R.C. and Rajagopalan, K.V. (1982) Evidence for the inorganic nature of the cyanolyzable sulfur of molybdenum hydroxylases. J. Biol. Chem. 257, 1354 – 1359.

92. Massey, V. and Edmondson, D. (1970) On the Mechanism of Inactivation of Xanthine Oxidase by Cyanide. J. Biol. Chem. 245, 6595 – 6598.

93. Pauff, J.M., Hemann, C.F., Jünemann, N., Leimkühler, S., and Hille, R. (2007) The Role of Arginine 310 in Catalysis and Substrate Specificity in Xanthine Dehydrogenase from Rhodobacter capsulatus. J. Biol. Chem. 282, 12785 – 12790.

94. Pauff, J.M., Zhang, J., Bell, C.E., and Hille, R. (2008) Substrate Orientation in Xanthine Oxidase: Crystal Structure of Enzyme in Reaction with 2-hydroxy-6- methylpurine. J. Biol. Chem. 283, 4818 – 4824.

95. Rayment, I. (2002) Small-Scale Batch Crystallization of Proteins Revisited: An Underutilized Way to Grow Large Protein Crystals. Structure 10, 147 - 151.

96. Collaborative Computational Project, Number 4 (1994) ``The CCP4 Suite: Programs for Protein Crystallography''. Acta Cryst. D50, 760-763.

97. Potterton, E., Briggs, P., Turkenburg, M., and Dodson, E.J. (2003) A graphical user interface to the CCP4 program suite. Acta. Cryst. D59, 1131-1137.

98. Murshudov, G., Vagin, A., and Dodson, E. (1996) The Refinement of Protein structures. Proceedings of Daresbury Study Weekend

161 99. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997) Refinement of Macromolecular Structures by the Maximum-Likelihood Method. Acta Cryst. D53, 240-255.

100. Pannu, N.J., Murshudov, G.N., Dodson, E.J., and Read, R.J. (1998) Incorporation of Prior Phase Information Strengthens Maximum-Likelihood Structure Refinement. Acta Cryst. D54, 1285-1294.

101. Murshudov, G.N., Lebedev, A., Vagin, A.A., Wilson, K.S., and Dodson, E.J. (1999) Efficient anisotropic refinement of macromolecular structures using FFT. Acta Cryst. D55, 247-255.

102. Winn, M., Isupov, M., and Murshudov, G.N. (2001) Use of TLS parameters to model anisotropic displacements in macromolecular refinement. Acta Cryst. D57, 122-133.

103. Schuettelkopf, A.W. and van Aalten, D.M.F. (2004) PRODRG: a tool for high- throughput crystallography of protein-ligand complexes. Acta Crystallographica D60, 1355--1363.

104. Emsley P, Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallographica Section D-Biological Crystallography 60, 2126-2132 Part 12 Sp. Iss.

105. DeLano, W.L. "The PyMOL Molecular Graphics System." DeLano Scientific LLC, San Carlos, CA, USA. http://www.pymol.org

106. Yamaguchi, Y., Matsumura, T., Ichida, K., Okamoto, K., and Nishino, T. (2007) Human Xanthine Oxidase Changes its Substrate Specificity to Aldehyde Oxidase Type upon Mutation of Amino Acid Residues in the Active Site: Roles of Active Site Residues in Binding and Activation of Purine Substrate. J Biochem (Tokyo) 141, 513-524.

107. Bray, R.C. and George, G.N. (1985) Electron-paramagnetic-resonance studies using pre-steady-state kinetics and substitution with stable isotopes on the mechanism of action of molybdoenzymes. Biochem. Soc. Trans. 13, 560-567.

108. Laskowski, R. A., MacArthur, M. W., Moss, D. S. and Thornton, J. M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures J. Appl. Crystallogr. 26, 283–291.

109. Davis, M.D., Olson, J.S., and Palmer, G. (1982) Charge Transfer Complexes between Pteridine Substrates and the Active Center Molybdenum of Xanthine Oxidase. J. Biol. Chem. 257, 14730 – 14737.

110. Davis, M.D., Olson, J.S., and Palmer, G. (1984) The Reaction of Xanthine Oxidase with Lumazine; Characterization of the Reductive Half-Reaction. J. Biol. Chem. 259, 3526 – 3533.

162 111. Hemann, C., Ilich, P., Stockert, A.L., Choi, E.Y., and Hille, R. (2005) Resonance Raman Studies of Xanthine Oxidase: the Reduced Enzyme-Product Complex with Violapterin. J. Phys. Chem. B 109, 3023-3031.

112. Oertling, W.A. and Hille, R. (1990) Resonance-enhanced Raman scattering from the molybdenum center of xanthine oxidase. J. Biol. Chem. 265, 17446 – 17450.

113. Godber, B.L.J., Doel, J.J., Sapkota, G.P, Blake, D.R., Stevens, C.R., Eisenthal, R., and Harrison, R. (2000) Reduction of Nitrite to Nitric Oxide Catalyzed by Xanthine Oxidoreductase J. Biol. Chem. 275, 7757 - 7763.

114. Lovell, S.C., Davis, I.W., Arendall, W.B., de Bakker, P.I.W., Word, J.M., Prisant, M.G., Richardson, J.S., and Richardson, D.C. (2003) Structure validation by C-alpha geometry: phi, psi, and C-beta deviation. Proteins: Structure, Function, and Genetics 50, 437-450.

115. La Mar, G.N., Asokan, A., Espiritu, B., Yeh, D.C., Auclair, K., and Ortiz de Montellano, P.R. (2001) Solution 1H NMR of the Active Site of Substrate- bound, Cyanide-inhibited Human Heme Oxygenase. COMPARISON TO THE CRYSTAL STRUCTURE OF THE WATER-LIGATED FORM. J. Biol. Chem. 276, 15676 - 15687.

116. Asai, R., Nishino, T., Matsumura, T., Okamoto, K., Igarashi, K., Pai, E.F., and Nishino, T. (2007) Two mutations convert mammalian xanthine oxidoreductase to highly superoxide-productive xanthine oxidase J.Biochem.(Tokyo) 141, 525-534.

117. Lee, C., Liu, X., and Zweier, J.L. (2000) Regulation of Xanthine Oxidase by Nitric Oxide and Peroxynitrite J. Biol. Chem. 275, 9369–9376.

118. Trujillo, M., Alvarez, M.N., Peluffo, G., Freeman, B.A., and Radi, R. (1998) Xanthine Oxidase-mediated Decomposition of S-Nitrosothiols J. Biol. Chem. 273, 7828 – 7834

119. Hille, R. and Nishino, T. (1995) Flavoprotein structure and mechanism. 4. Xanthine oxidase and xanthine dehydrogenase. FASEB J. 9, 995–1003.

120. Nishino, T., Okamoto, K., Kawaguchi, Y., Hori, H., Matsumura, T., Eger, B.T., Pai, E.F., and Nishino, T. (2005) Mechanism of the Conversion of Xanthine Dehydrogenase to Xanthine Oxidase: IDENTIFICATION OF THE TWO CYSTEINE DISULFIDE BONDS AND CRYSTAL STRUCTURE OF A NON-CONVERTIBLE RAT LIVER XANTHINE DEHYDROGENASE MUTANT. J. Biol. Chem. 280, 24888-24894.

163 121. de Jong, J.W., Schoemaker, R.G., de Jonge, R., Bernocchi, P., Keijzer, E., Harrison, R., Sharma, H.S., and Ceconi, C. (2000) Enhanced expression and activity of xanthine oxidoreductase in the failing heart. J Mol Cell Cardiol. 32, 2083–2089.

122. Poss, W.B., Huecksteadt, T.P., Panus, P.C., Freeman, B.A, and Hoidal, J.R. (1996) Regulation of xanthine dehydrogenase and xanthine oxidase activity by hypoxia. Am. J. Physiol. Lung Cell. Mol.Physiol. 270, 941 - 946.

123. Kayyali, U.S., Donaldson, C., Huang, H., Abdelnour, R., and Hassoun, P.M. (2001) Phosphorylation of Xanthine Dehydrogenase/Oxidase in Hypoxia. J. Biol. Chem. 276, 14359 – 14365

124. Remy, C.N., Richert, D.A., Doisy, R.J., Wells, I.C., and Westerfeld, W.W. (1955) Purification and Characterization of Chicken Liver Xanthine Dehydrogenase J. Biol. Chem. 217, 293 – 306

125. Watt, W.B. (1972) Xanthine Dehydrogenase and Pteridine Metabolism in Colias Butterflies J. Biol. Chem. 247, 1445 – 1451.

126. Ferrari, R., Cargnoni, A., Bernocchi, P., Pasini, E., Curello, S., Ceconi, C., and Ruigrok, T.J.C. (1996) Metabolic Adaptation During a Sequence of No-Flow and Low-Flow Ischemia: A Possible Trigger for Hibernation. Circulation 94, 2587 - 2596

127. Esumi, K., Nishida, M., Shaw, D., Smith, T. W., and Marsh, J.D. (1991) NADH measurements in adult rat myocytes during simulated ischemia. Am. J. Physiol. Heart Circ. Physiol. 260, H1743 - H1752

128. Moroni, F. (2008) Poly(ADP-ribose)polymerase 1 (PARP-1) and postischemic brain damage. Curr. Opin. Pharmacol. 8, 96 – 103.

129. Hunt, J. and Massey, V. (1992) Purification and Properties of Milk Xanthine Dehydrogenase. J. Biol. Chem. 267, 21479 – 21485.

130. Falasca, G.F. (2006) Metabolic diseases: gout. Clinics in Dermatology 24, 498– 508.

131. Yellon, D.M. and Hausenloy, D.J. (2007) Myocardial Reperfusion Injury N. Engl. J. Med. 357, 1121-35.

132. Li, C. and Jackson, R.M. (2002) Reactive species mechanisms of cellular hypoxia-reoxygenation injury. Am. J. Physiol. Cell. Physiol. 282, 227-241.

133. Terkeltaub, R.A. (2003) Gout. N. Engl. J. Med. 349, 1647-1655.

134. Massey, V., Komai, H., Palmer, G., and Elion, G. B. (1970) On the Mechanism of Inactivation of Xanthine Oxidase by Allopurinol and Other Pyrazolo[3,4- d]pyrimidines. J. Biol. Chem. 245, 2837–2844. 164

135. Khosravan, R., Grabowski, B.A., Wu, J.T., Joseph-Ridge, N., and Vernillet, L. (2006) Pharmacokinetics, Pharmacodynamics and Safety of Febuxostat, a Non-Purine Selective Inhibitor of Xanthine Oxidase, in a Dose Escalation Study in Healthy Subjects. Clin. Pharmacokinet. 45, 821-841.

136. Becker, M.A., Schumacher Jr., H.R., Wortmann, R.L., MacDonald, P.A., Eustace, D., Palo, W.A., Streit, J., and Joseph-Ridge, N. (2005) Febuxostat Compared with Allopurinol in Patients with Hyperuricemia and Gout. N. Engl. J. Med. 353, 2450-2461.

137. Schumacher Jr, H.R. (2005) Febuxostat: a non-purine, selective inhibitor of xanthine oxidase for the management of hyperuricaemia in patients with gout. Expert. Opin. Investig. Drugs 14, 893- 903.

138. Sánchez-Lozada, L.G., Tapia, E., Soto, V., Avila-Casado, C., Franco, M., Zhao, L., and Johnson, R.J. (2008) Treatment with the xanthine oxidase inhibitor febuxostat lowers uric acid and alleviates systemic and glomerular hypertension in experimental hyperuricaemia. Nephrol. Dial. Transplant. 23, 1179 - 1185.

139. Mo, S.F., Zhou, F., Lv, Y.Z., Hu, Q.H., Zhang, D.M., and Kong, L.D. (2007) Hypouricemic Action of Selected Flavonoids in Mice: Structure–Activity Relationships. Biol. Pharm. Bull. 30, 1551-1556.

140. Wu, J.W., Tsai, T.H. (2007) Effect of Silibinin on the Pharmacokinetics of Pyrazinamide and Pyrazinoic Acid in Rats. Drug Metabolism and Disposition 35, 1603-1610.

141. Yu, Z., Fong, W.P., and Cheng, C.H.K. (2006) The Dual Actions of Morin (3,5,7,2',4'-Pentahydroxyflavone) as a Hypouricemic Agent: Uricosuric Effect and Xanthine Oxidase Inhibitory Activity. J. Pharm. Exp. Therap. 316, 169- 175.

142. Cos, P., Ying, L., Calomme, M., Hu, J.P., Cimanga, K., Poel, B.V., Pieters, L., Vlietinck, A.J., Berghe, D.V. (1998) Structure-Activity Relationship and Classification of Flavonoids as Inhibitors of Xanthine Oxidase and Superoxide Scavengers. J. Nat. Prod. 61, 71-76.

143. Knekt, P., Kumpulainen, J., Järvinen, R., Rissanen, H., Heliövaara, M., Reunanen, A., Hakulinen, T., and Aromaa, A. (2002) Flavonoid intake and risk of chronic diseases. Am. J. Clin. Nutr. 76, 560-568.

144. Nijveldt, R.J., van Nood, E., van Hoorn, D.E.C., Boelens, P.G., van Norren, K., van Leeuwen, P.A.M. (2001) Flavonoids: a review of probable mechanisms of action and potential applications. Am. J. Clin. Nutr. 74, 418-425.

145. Chun, O.K., Chung, S.J., and Song, W.O. (2007) Estimated Dietary Flavonoid Intake and Major Food Sources of U.S. Adults. J. Nutrition 137, 1244-1252. 165

146. Lin, J.K., and Shih, C.A. (1994) Inhibitory effect of curcumin on xanthine dehydrogenase/oxidase induced by phorbol-12-myristate-13-acetate in NJH3T3 cells. Carcinogenesis 15, 1717-1721.

147. Lin, H.C., Tsai, S.H., Chen, C.S., Chang, Y.C., Lee, C.M., Lai, Z.Y., and Lin, C.M. (2008) Structure–activity relationship of coumarin derivatives on xanthine oxidase-inhibiting and free radical-scavenging activities. Biochemical Pharmacology 75, 1416-1425.

148. Prince, M., Campbell, C.T., Robertson, T.A., Wells, A.J., and Kleiner, H.E. (2006) Naturally occurring coumarins inhibit 7,12-dimethylbenz[a]anthracene DNA adduct formation in mouse mammary gland. Carcinogenesis 27, 1204 - 1213.

149. Chen, S., Cho, M., Karlsberg, K., Zhou, D., and Yuan, Y.C. (2004) Biochemical and Biological Characterization of a Novel Anti-aromatase Coumarin Derivative. J. Biol. Chem. 279, 48071 - 48078.

150. Agnes Y.Y. Lee, Frederick R. Rickles, Jim A. Julian, Michael Gent, Ross I. Baker, Chris Bowden, Ajay K. Kakkar, Martin Prins, and Mark N. Levine (2005) Randomized Comparison of Low Molecular Weight Heparin and Coumarin Derivatives on the Survival of Patients With Cancer and Venous Thromboembolism. J. Clin. Oncol. 23, 2123 - 2129.

151. Schalekamp, T., Klungel, J.H., Souverein, P.C., and de Boer, A. (2008) Increased Bleeding Risk With Concurrent Use of Selective Serotonin Reuptake Inhibitors and Coumarins. Arch. Intern. Med. 168, 180 - 185.

152. Allan, R.K., Mok, D., Ward, B.K., and Ratajczak, T. (2006) Modulation of Chaperone Function and Cochaperone Interaction by Novobiocin in the C- terminal Domain of Hsp90: EVIDENCE THAT COUMARIN ANTIBIOTICS DISRUPT Hsp90 DIMERIZATION. J. Biol. Chem., 281, 7161 - 7171.

153. Devi, R.R. and Arumughan, C. (2007) Antiradical efficacy of phytochemical extracts from defatted rice bran. Food and Chemical Toxicology 45, 2014- 2021.

154. Ou, S.Y., Jackson, G.M., Jiao, X., Chen, J., Wu, J.Z., Huang, X.S. (2007) Protection against Oxidative Stress in Diabetic Rats by Wheat Bran Feruloyl Oligosaccharides. J. Agric. Food Chem. 55, 3191-3195.

155. Kim, J.H. and Hille, R. (1993) Reductive half-reaction of xanthine oxidase with xanthine. Observation of a spectral intermediate attributable to the molybdenum center in the reaction of enzyme with xanthine. J. Biol. Chem. 268, 44-51.

166 156. Bonagura, C.A. Bhaskar, B., Sundaramoorthy, M., and Poulos, T.L. (1999) Conversion of an Engineered Potassium-binding Site into a Calcium-selective Site in Cytochrome c Peroxidase. J. Biol. Chem. 274, 37827–37833.

157. Guo, M., Bhaskar, B., Li, H., Barrows, T.P., and Poulos, T.L. (2004) Crystal structure and characterization of a cytochrome c peroxidase–cytochrome c site-specific cross-link. Proc. Natl. Acad. Sci. U.S.A. 101, 5940–5945.

158. Williams, J.W. and Bray, R.C. (1981) Kinetic and e.p.r. studies on the inhibition of xanthine oxidase by alloxanthine (1 H-pyrazolo [3, 4-d] pyrimidine-4,6- diol). Biochem. J. 195, 753 – 760.

159. Kelley, E.E., Trostchansky, A., Rubbo, H., Freeman, B.A., Radi, R., and Tarpey, M.M. (2004) Binding of Xanthine Oxidase to Glycosaminoglycans Limits Inhibition by Oxypurinol. J. Biol. Chem. 279, 37231 – 37234.

167