COMPUTATIONAL INVESTIGATIONS OF CYTOCHROME P450 AROMATASE CATALYSIS AND BIOLOGICAL EVALUATION OF ISOFLAVONE AROMATASE INHIBITORS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

By

John C Hackett, B.Sc. Pharm., M. S.

* * * * *

The Ohio State University 2004

Dissertation Committee: Approved by Professor Robert W. Brueggemeier, Adviser

Professor Paul Blower ______Adviser Professor Pui-Kai Li Graduate Program in Pharmacy

Professor Karl Werbovetz

ABSTRACT

Dioxygen-containing P450 catalytic intermediates including the reduced dioxygen and hydroperoxo species beyond the dioxygen-ligated enzyme have eluded direct structural observation. In this study, structures, vibrations, and relative energetics of low and intermediate spin states of these species are characterized with unrestricted density functional theory using gradient-corrected and hybrid exchange correlation potentials. All functionals tested produce quality geometries compared to available experimental data, and the dioxygen vibrational frequencies are in reasonable agreement with data from resonance Raman and infrared spectroscopic results. The red shifts which have been observed in the UV/Visible spectrum of wild-type and D251N- cytochrome P450cam from radiolytic reduction of these enzymes are reproduced well using RI-J time- dependent density functional theory calculations. Interestingly, the computed spectral patterns for the reduced dioxygen and hydroperoxo model systems are quite similar.

B3LYP computations of 14N and 1H hyperfine coupling constants revealed that it is a competent level of theory to reproduce the experimental ENDOR values of the hydroperoxo species. It has been postulated that a species provisionally assigned as the reduced dioxygen intermediate has the same 14N hyperfine coupling constant as the hydroperoxo species due to the spin density distortion effects of protein active-site hydrogen-bond donors. Interestingly, even in the presence of hydrogen-bond donors ii which should mimic a gross excess of hydrogen-bonding potential, the spin density on iron never recovers to give rise to the experimentally observed hyperfine coupling constant. This observation raises some controversy about the true identity of the species assigned as the reduced dioxygen intermediate.

B3LYP density functional theory calculations are used to unravel the mysterious third step of aromatase catalysis. The feasibility of mechanisms in which the reduced ferrous dioxygen intermediate mediates androgen aromatization are explored and determined to be unlikely. Mechanisms for the aromatization/deformylation sequence which are initiated by 1β-hydrogen atom abstraction by P450 Compound I are considered. 1β-

Hydrogen atom abstraction from substrates in the presence of the 2,3-enol encounters strikingly low barriers (5.3-7.8 kcal/mol), whereas barriers for this same process rise to

17.0-27.1 kcal/mol in the keto tautomer. Transition states for 1β-hydrogen atom

abstraction from enolized substrates in the presence of the 19-gem-diol decayed directly

to the experimentally observed products. If the C19 aldehyde remains unhydrated,

aromatization occurs with concomitant decarbonylation, and therefore does not support

dehydration of the C19 aldehyde prior to the final catalytic step. On the doublet surface,

the transition state connects to a potentially labile 1(10) dehydrogenated product, which

may undergo rapid aromatization, as well as formic acid. As the reaction vectors

indicated, ab initio molecular dynamics on the Born-Oppenheimer potential energy

hypersurface confirmed that the 1β-hydrogen atom abstraction and deformylation or

decarbonylation occur in a non-synchronous, coordinated manner. These calculations

iii support a dehydrogenase behavior of aromatase in the final catalytic step, which can be

summarized by 1β-hydrogen atom abstraction followed by gem-diol deprotonation.

Aromatase, a cytochrome P450 hemoprotein that is responsible for estrogen biosynthesis

by conversion of androgens into estrogens, has been an attractive target in the treatment

of hormone-dependent breast cancer. As a result, a number of synthetic steroidal or

nonsteroidal aromatase inhibitors have been successfully developed. In addition, there are

several classes of natural products that exert potent activities in aromatase inhibition, with the flavonoids being most prominent. Previous studies have exploited flavone and flavanone scaffolds for the development of new aromatase inhibitors. In this dissertation, we describe design, synthesis, and biological evaluation of a novel series of 2-(4’- pyridylmethyl)thioisoflavones as the first example of synthetic isoflavone-based

aromatase inhibitors.

The biological evaluation of a series of 2-azole and 2-thioazole isoflavones as potential

aromatase inhibitors are described. Differences in inhibitory activity of triazole and

imidazole inhibitors are rationalized with density functional theory to expose a key

difference in the electronic structure of these molecules. In addition, difference binding

spectra of inhibitors to immunoaffinity-purified aromatase produces classical Type II

spectra consistent with coordination of the nitrogen lone pair electrons to the P450 heme.

iv

For my loving wife and son, Rajini and Dev

v

ACKNOWLEDGMENTS

I wish to thank my adviser, Dean Robert W. Brueggemeier for support over the course of

my Ph.D. study and provision of unconditional freedom to pursue all of my research

ideas, no matter how far the required techniques seemed to deviate from the traditions of

the laboratory. If not for this, this thesis could not have materialized.

I thank all faculty in the Division and in the Department of Chemistry for excellent and

enthusiastic teaching. In particular, I thank Professor Christopher M. Hadad. His

outstanding knowledge and enthusiasm for computational chemistry sparked my intense interest for this area of research. Without his guidance, the marriage of computational chemistry and the traditional enzymological approaches undertaken in our laboratory could not have flourished. Thus, the insight gained in this research into the most fundamental processes occuring within cytochrome P450 enzymes would not have been possible.

I wish to express my gratitude to all my current and past labmates for their friendship over the years; Jon Baker, Jennifer Whetstone, Trevor Petrel, Jeannette Richards,

Surachai Joomprabutra, and Danielle Pellegata, Danyetta Davis, and Edgar Diaz-Cruz. I

want to acknowledge Mike Ivers, Kevin Schaefer, and Jason Holton, if not for their v i support and long-lasting friendship, success in the early years of my graduate study

would have been far more difficult. I am especially grateful to Dr. Young-Woo Kim who

is an amazingly talented scientist, extraordinary collaborator, and friend. I am grateful to

Bin Su, for useful discussions about all aspects of my research and his friendship. I am

especially indebted to Serena Landini, whose unwavering friendship for myself and

family has made our life away from home enjoyable.

Special mention must go to my wife, Rajini, for tolerating a change from a stable lifestyle

in Florida and many sacrifices to endure this adventure. I cannot express the gratitude I

have for her, as her partnership and love have provided the motivation for my intellectual

and personal evolution.

This research was supported by USAMRC Pre-doctoral Fellowship (DAMD17-02-1-

0529). Computations were carried out at the Ohio Supercomputer Center (OSC) and supported by OSC grant PAS0091.

v ii

VITA

July 11, 1976……………………………...Born – Orlando, Florida

1999……………………………………… B.S. in Pharmacy, University of Florida

1999-2002..……………………………….Graduate Teaching Associate

Division of Medicinal Chemistry and

Pharmacognosy, The Ohio State University,

Columbus, Ohio

2002-2004………………………………...Graduate Research Associate

Division of Medicinal Chemistry and

Pharmacognosy, The Ohio State University,

Columbus, Ohio

PUBLICATIONS

Research Publication

1. Kim, Y.-W.; Hackett, J. C.; Brueggemeier, R. W. Synthesis and aromatase

inhibitory activity of novel pyridine-containing isoflavones. J. Med. Chem. 2004,

47, 4032-4040.

viii

FIELDS OF STUDY

Major Field: Pharmacy

Medicinal Chemistry

ix

TABLE OF CONTENTS

Page

Abstract ...... ii

Dedication ...... v

Acknowledgments ...... vi

Vita...... vii

List of Tables ...... xiv

List of Figures...... xviii

Chapters:

1. Introduction: Aromatase Catalysis and Inhibition...... 1

1.1 Physiology and Biosynthesis of Estrogens ...... 1

1.2 The Catalytic Mechanism of Aromatase...... 7

1.3 Inhibitors of Aromatase...... 25

1.4 Natural Product Inhibitors of Aromatase...... 30

1.4.1 Flavonoids ...... 30

1.4.2 Other Natural Products ...... 34

1.5 References...... 39 x 2. Density Functional Theory Investigation of Model Dioxygen Intermediates

in the P450 Catalytic Cycle: Structure and Spectroscopic Properties...... 49

2.1 Introduction...... 49

2.2 Computational Methodology ...... 54

2.3 Results and Discussion...... 57

2.3.1 Ordering of States ...... 57

2.3.2 Geometries and Harmonic Frequencies of P450 Models ...... 61

2.3.3 Excited States...... 72

2.3.4 Spin Distribution and Hyperfine Coupling Constants...... 77

2.4 Conclusions...... 90

2.5 References...... 93

3. The Final Catalytic Step of Cytochrome P450 Aromatase:

A Density Functional Theory Study ...... 106

3.1 Introduction...... 106

3.2 Computational Methodology ...... 110

3.2.1. Model Systems and Nomenclature...... 110

3.2.2. Theoretical Methods ...... 113

3.3 Results and Discussion...... 115

3.3.1. Deformylation by the Ferrous Peroxo Intermediate ...... 115 xi 3.3.2 The Model Steroid-Compound I VdW Complexes...... 120

3.3.3 The 1β-Hydrogen Atom Abstraction/Aromatization

Sequence...... 124

3.3.4 Ab Initio Molecular Dynamics of 2,4ETS and 2,4OTS...... 146

3.4 Conclusions...... 155

3.5 References...... 156

4. Pyridine-Containing Isoflavone Inhibitors of Aromatase ...... 166

4.1 Introduction...... 166

4.2 Inhibitor Design...... 167

4.3 Results and Discussion...... 172

4.4 Conclusions...... 194

4.5 Experimental Section...... 195

4.6 References...... 197

5. Azole Isoflavone Inhibitors of Aromatase...... 201

5.1 Introduction...... 201

5.2 Results and Discussion...... 203

5.3 Conclusions...... 215

5.4 Experimental Section...... 216

5.5 References...... 217 xii 6. Conclusions...... 218

6.1 Computational Studies of P450 Aromatase Catalysis...... 218

6.2 Isoflavone Inhibitors of Aromatase...... 224

6.3 References...... 225

Appendices...... 226

Appendix A: Supporting Information for Chapter 2 ...... 226

Appendix B: Supporting Information for Chapter 3...... 232

Bibliography ...... 247

xiii

LIST OF TABLES

Table Page

2.1. Triplet-Singlet (3A-1A) and (4A-2A) Quartet-Doublet Energetic Preferences

(kcal/mol) for Truncated Models of Cytochrome P450 as well as Electron Affinities (eV)

of the Dioxygen Model for Different Density Functionals in the Gas Phase and in the

presence of a Low Dielectric Solvent (ε = 5.621)...... 60

1- 2.2. Key Geometric Parameters (Å) for [Fe(P)(O2)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 63

2- 2.3. Key Geometric Parameters (Å) for [Fe(P)(O2)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 64

1- 2.4 Key Geometric Parameters (Å) for [Fe(P)(O2H)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 65

-1 1- 2.5. Vibrational Frequencies (cm ) for [Fe(P)(O2)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 66

-1 2- 2.6. Vibrational Frequencies (cm ) for [Fe(P)(O2)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 67

-1 1- 2.7. Vibrational Frequencies (cm ) for [Fe(P)(O2H)(SCysNH2)] for Different Density

Functionals in the Gas Phase...... 68

1- 2.8. Prominent RI-J TD-DFT computed UV/Vis Transitions in [Fe(P)(O2)(SCysNH2)]

...... 73 xi v 2 2.9. Prominent RI-J TD-DFT computed UV/Vis Transitions in [Fe(P)(O2)(SCysNH2)]

...... 74

1- 2.10 Prominent RI-J TD-DFT computed UV/Vis Transitions [Fe(P)(O2H)(SCysNH2)]

...... 75

2.11 DFT/(TZV,6-311+G*)//DFT (TZV/6-31G*) Isotropic Hyperfine Coupling

Constants of Heme Nitrogen Atoms for all Models and the Distal Proton of

1- [Fe(P)(O2H)(SCysNH2)] in the Gas Phase and with ε =5.621...... 81

2.12. DFT/(TZV,6-311+G*)//DFT (TZV/6-31G*) Natural Atomic and Group Spin

2- 1- Densities of the [Fe(P)(O2)(SCysNH2)] and [Fe(P)(O2H)(SCysNH2)] Models of

Reduced Dioxygen and Hydroperoxo P450 in the Gas Phase and ε=5.621...... 82

2.13. B3LYP/(TZV,6-311+G*)//B3LYP (TZV/6-31G*) Isotropic Hyperfine Coupling

Constants of Heme Nitrogen Atoms and Natural Atomic and Group Spin Densities of the

2- Hydrated [Fe(P)(O2)(SCysNH2)] and in the Gas Phase and ε=5.621...... 83

3.1. Bond lengths (in Å) and angles (in degrees) for the 2A(4A) reactant clusters and 1β-

hydrogen atom abstraction transition states...... 123

3.2 4A-2A Energetic splitting for reactant complexes...... 124

3.3. 2,4ERC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well

as spin densities...... 125

3.4. 2,4ORC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities...... 126

3.5. 2,4KRC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities...... 127

xv 3.6. Energies relative to the Reactant Complex of Stationary points on the 2A (4A)1β-H atom abstraction/aromatization PES...... 130

3.7. 2,4ETS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well

as spin densities...... 142

3.8. 2,4OTS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well

as spin densities...... 143

3.9. 2,4KTS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well

as spin densities...... 144

4.1. Aromatase inhibitory activities of isoflavones 5a-k and 6a-g...... 177

4.2. Enzyme kinetic parameters for selected isoflavones and reference compounds...... 185

4.3. Aromatase inhibitory activities of isoflavones 7a-e ...... 188

4.4. Aromatase inhibitory activities of isoflavones 8a-c...... 189

4.5. Aromatase inhibitory activities of isoflavones 9a-c ...... 190

4.6. Aromatase inhibitory activities of isoflavones 10a-c...... 191

5.1. IC50 values for aromatase inhibition by azole isoflavones and

reference compounds...... 206

5.2. Enzyme kinetic parameters for imidazole isoflavones and reference compounds...... 215

B.1 Relevant Geometric Parameters for Select Time Steps of the 2ETS Molecular

Dynamics Trajectory...... 234

B.2 Relevant Geometric Parameters for Select Time Steps of the 2ETS Molecular

Dynamics Trajectory ...... 235

B.3 Relevant Geometric Parameters for Select Time Steps of the 2OTS Molecular

Dynamics Trajectory ...... 240 xvi B.4 Relevant Geometric Parameters for Select Time Steps of the 2OTS Molecular

Dynamics Trajectory ...... 241

B.5 B3LYP/(TZVP,SV(P)) TD-DFT calculations of MD steps 19 and 31 initiated with

2ETS...... 245

xvii

LIST OF FIGURES

Figure Page

1.1 Structures of endogenous estrogens...... 2

1.2 Principal pathways of steroid hormone biosynthesis in the human ovary...... 5

1.3 Prosthetic of cysteinato-heme enzymes including cytochrome P450 aromatase: an

iron-protoporphyrin IX linked with a proximal cysteine ligand...... 8

1.4 Schematic representation of different intermediates generated during the catalytic

cycle of cytochrome P450 enzymes...... 9

1.5 Proposed intermediates in the catalytic turnover of cytochrome P450 aromatase. ...11

1.6 Postulated mechanism for the base-catalyzed aromatization of 2β-hydroxy-19-oxo-

androst-4-ene-3,17-dione...... 13

1.7 Postulated aromatization mechanism of androgen epoxides ...... 15

1.8 Summary of the 18O labeling studies of Akhtar and coworkers ...... 18

1.9 Deformylation of 19-oxo androstenedione by the aromatase peroxo-iron

intermediate as proposed by Akhtar and coworkers...... 19

1.10 Mechanism for the final catalytic step of cytochrome P450 aromatase proposed by

Covey and coworkers...... 23

1.11 Aminoglutethimide (AG) and imidazole inhibitors of aromatase including

antimycotic agents...... 27

1.12 Triazole inhibitors of aromatase ...... 28 xviii 1.13 Flavonoid compounds investigated for aromatase inhibitory activity...... 31

1.14 Isoflavonoid compounds investigated for aromatase inhibitory activity...... 33

1.15 Mammalian lignans investigated for aromatase inhibitory activity...... 35

1.16 TAN-931 and its N,N-dimethyl amide analog isolated from cultures of Penicillium

funiculosum No. 8974...... 36

1.17 Sesquiterpene lactones isolated from different Asteraceae species from northwestern

Argentina...... 37

1.18 Structure of standishnal from Thuja standishii and the diacetyl analog...... 38

2.1 Oxygen-containing catalytic intermediates of the cytochrome P450 catalytic

cycle ...... 50

1-,2- 1- 2.2 Structure of [Fe(P)(O2)(SCysNH2)] (top) and [Fe(P)(O2H)(SCysNH2)] (bottom)

model for cytochrome P450...... 53

2- 2.3 Contour plots of the spin density distribution for the [Fe(P)(O2)(SCysNH2)] (top)

1- and [Fe(P)(O2H)(SCysNH2)] (bottom) models as obtained with B3LYP in the

gas-phase (left) and in the presence of the low dielectric solvent

continuum (ε = 5.621) (right)...... 84

2- 2.4 Structures of hydrated [Fe(P)(O2)(SCysNH2)] ...... 87

3.1 Intermediates in the catalytic turnover of cytochrome P450 aromatase...... 107

3.2 Deformylation of 19-oxo androstenedione as initially proposed by Akhtar...... 109

3.3 Model systems and the androstane-based numbering system used in this study...... 112

3.4 Structure of the hydrated peroxo hemiacetal adduct of the (a) enol-19-oxo steroid and (b) protonated peroxo hemiacetal adduct...... 118

3.5 Results of the B3LYP/(TZV,3-21G*) C1-H1β PES scans...... 119 xix 3.6 Fully optimized structures of (a) 2ERC, (b) 2KRC, and (c) 2ORC

at the B3LYP/(TZV,3-21G*) level of theory...... 122

3.7 Relative energy diagram of stationary points along the 1β- hydrogen atom

abstraction/deformylation reaction for the enol-19,19-diol (E) and ketone-19,19-diol (K) species at the B3LYP/(Wachter’s + f, 6-311+G**)//B3LYP/(TZV,3-21G*) level of theory...... 128

3.8 Relative energy diagram of stationary points along the 1β- hydrogen atom

abstraction/deformylation reaction for the enol-19-oxo (O) species at the

B3LYP/(Wachter’s + f, 6-311+G**)//B3LYP/(TZV,3-21G*) level of theory...... 129

3.9 B3LYP/(TZV,3-21G*) fully optimized transition-state structures and A-ring carbon

B3LYP/(Wachter’s+f, 6-311+G**) NPA spin densities of

(a) 2ETS, (b )2OTS, and (c) 2KTS...... 132

3.10 B3LYP/6-311+G**//B3LYP/6-31G* NPA spin densities of heavy atoms in the model

steroid substrate A ring...... 133

3.11 Representative normal modes for the respective imaginary vibrational frequency of

the (a) 2ETS, (b) 2OTS, and (c) 2KTS transition state structures...... 134

3.12 Fully optimized structures of (a) 2,4EPC, (b) 2OPC, and (c) 2KPC at the

B3LYP/(TZV,3-21G*) level of theory...... 140

3.13 Fully optimized structures of (a) 4OHR and (b) 4KHR at the B3LYP/(TZV,3-21G*)

level of theory...... 141

3.14 Ab initio MD trajectory energy profiles initiated with (a) 2,4ETS and (b) 2,4OTS at the

B3LYP/(TZV,3-21G*) level of theory...... 148

xx 3.15 Gas phase (●) and ε = 5.621 (○) Mulliken atomic charges (a) and β spin densities (b) localized on the steroid substrate at critical points on the ab initio MD trajectory for propagation of 2ETS transition state structure from Figure 3.10...... 149

3.16 Highest-occupied (158) and lowest unoccupied (159) β spin orbitals of the 2ETS MD steps 19 and 31...... 152

3.17 1β-Hydrogen atom abstraction-electron transfer mechanism for aromatization of the steroid A-ring in the final catalytic step of cytochrome P450 aromatase...... 154

4.1 Chemical structures and aromatase inhibitory activities of two isoflavones (genistein, biochanin A) and one flavone (chrysin)...... 167

4.2 Synthesis of isoflavones 5a-k ...... 170

4.3 Synthesis of isoflavones 6a-g...... 171

4.4 Aromatase inhibitory activities of triazole and thioazole isoflavones 5e ( ), 5f (■), 5g

(▲), 5h (▼), 5i (♦), 5j (●), and 5k (□)...... 175

4.5 Aromatase inhibitory activities of triazole and thioazole isoflavones 6d (■), 6e (▲), 6f

(▼), and 6g (♦)...... 176

4.6 Lineweaver-Burk plot of aromatase inhibition by compound 5e...... 178

4.7 Lineweaver-Burk plot of aromatase inhibition by compound 5h...... 179

4.8 Lineweaver-Burk plot of aromatase inhibition by compound 5i...... 180

4.9 Lineweaver-Burk plot of aromatase inhibition by compound 5j...... 181

4.10 Lineweaver-Burk plot of aromatase inhibition by compound 6f...... 182

4.11 Lineweaver-Burk plot of aromatase inhibition by compound 6g...... 183

4.12 Lineweaver-Burk plot of aromatase inhibition by compound AG...... 184

4.13 Modification of aryl group at the 7-position of isoflavones...... 186 xxi 4.14 Aromatase inhibitory activities of isoflavones 7a (■), 7b (▲), 7c (▼), 7d (♦), and 7e

(●)...... 187

4.15 Aromatase inhibitory activities of isoflavones 8a (■), 8b (▲), and 8c...... 189

4.16 Aromatase inhibitory activities of isoflavones 9a (■), 9b (▲), and 9c...... 190

4.17 Aromatase inhibitory activities of isoflavones 10a (■), 10b (▲), and 10c...... 191

5.1 Chemical structures and aromatase inhibitory acitivities of Biochanin A (isoflavone),

Chrysin (flavone), and the 7-benzyloxy-2-(4-pyridylmethylthio)isoflavone aromatase inhibitor reported by our laboratory...... 202

5.2 Synthesis of azole isoflavones 12a-d and 13a-d...... 204

5.3 Aromatase inhibitory activities of imidazole isoflavones 12a (○), 12b (▲), 12c (●), and 12d (□)...... 205

5.4 Aromatase inhibitory activities of triazole and thioazole isoflavones 13a (○), 13b (▲),

13c (●), and 13d (□)...... 205

5.5 B3LYP/6-31G(d) optimized geometries of 12a (left) and 13a (right) displaying the surfaces of the highest-occupied molecular orbital (HOMO)...... 208

5.6 Type II difference spectra of immunoaffinity-purified human placental aromatase induced by 50 µM compound 12a (blue) and compound 13a (red)...... 209

5.7 Lineweaver-Burk plot of aromatase inhibition by compound 12a...... 211

5.8 Lineweaver-Burk plot of aromatase inhibition by compound 12b...... 212

5.9 Lineweaver-Burk plot of aromatase inhibition by compound 12c...... 213

5.10 Lineweaver-Burk plot of aromatase inhibition by compound 12d...... 214

xxii A.1 O2(πg)-Fe(3dyz)-S(2py) lowest unoccupied Kohn-Sham molecular orbital of the

1- [Fe(P)(O2)(SCysNH2)] model intermediate (top) and singly-occupied Kohn-Sham

2- molecular orbital of [Fe(P)(O2)(SCysNH2)] (bottom)...... 227

A.2 O2(πg)-Fe(3dyz)-S(2py) singly-occupied Kohn-Sham molecular orbital of the

1- [Fe(P)(O2H)(SCysNH2)] model intermediate...... 228

A.3 Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2)(SCysNH2)] ...... 229

A.4 Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2H)(SCysNH2)] ...... 230

A.5 Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2H)(SCysNH2)] ...... 231

B.1 Singly occupied molecular orbital of 2ERC ...... 233

B.2 Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory...... 236

B.3 Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory...... 237

B.4 Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory...... 238

B.5 Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory...... 239

B.6 Structures of Select Time Steps of the 2OTS Molecular Dynamics Trajectory ...... 242

B.7 Structures of Select Time Steps of the 2OTS Molecular Dynamics Trajectory ...... 243

B.8 Structures of Select Time Steps of the 2OTS Molecular Dynamics Trajectory ...... 244

B.9 Gas phase (■) and ε = 5.621 (▲) NPA atomic charges and spin densities localized on the substrate at critical points on the 2ETS molecular dynamics trajectory ...... 246

xxiii

CHAPTER 1

INTRODUCTION

AROMATASE: CATALYSIS AND INHIBITION

1.1. PHYSIOLOGY AND BIOSYNTHESIS OF ESTROGENS

Estrogen biosynthesis appears to be ubiquitous throughout the vertebrates including

mammals, birds, reptiles, teleost and elasmobranch fish, and Agnatha species (hagfish

and lampreys). In birds, fish and most mammals, estrogen biosynthesis generally occurs

in the gonads and in the brain. Estrogen biosynthesis in the brain has been implicated in

sex-related behavior such as mating responses. In humans, estrogen biosynthesis is more

extensive among the tissues. [1-3]

Estrogens are endogenous hormones that produce numerous physiological actions. In human females, these include developmental effects, neuroendocrine actions involved in

the control of ovulation, cyclical preparation of the reproductive tract for fertilization and

implantation, and major actions on mineral, carbohydrate, protein, and lipid metabolism.

The most potent naturally-occuring estrogens in humans are 17β-estradiol (Figure 1.1, 1 1a), followed by estrone (1b) and estratriol (1c). Nonsteroidal compounds with estrogenic or antiestrogenic activity including flavones, isoflavones, and coumestane derivatives occur in a variety of plant and fungal species. A number of synthetic agents including pesticides, plasticizers, and a variety of other industrial chemicals also have estrogenic activity. While the affinity of these environmental estrogens or

“xenoestrogens” for estrogen receptors may be low, their quantity and persistence in the environment as well as pharmacokinetic parameters, which favor accumulation in adipose tissue, raise concern about their potential toxicity. [4]

R1

R2

HO

1a: 17b-Estradiol (R1 = OH, R2 = H) 1b: Estrone (R1 = O, R2 = H) 1c: Estriol (R1 = OH, R2 = OH)

Figure 1.1: Structures of endogenous estrogens.

The ovaries are the principal source of circulating estrogen in the premenopausal woman.

The major secretory product is 17β-estradiol, synthesized by the granulosa cells from

androgenic precursors provided by thecal cells (See Figure 1.2 for biosynthesis in the

human ovary). In men and postmenopausal women, the principal source of estrogen is the

adipose stroma where estrogen is synthesized from dihydroepiandrosterone secreted by

the adrenal cortex. Estrogens are largely responsible for the pubertal changes and account

for the secondary sexual characteristics of females. They directly cause growth and 2 development of the vagina, uterus, and fallopian tubes. They also act in concert with other hormones to cause enlargement of the breasts through promotion of ductal growth, stromal development, and deposition of fat. [6] Ductal elongation and branching occurs with positive response to ovarian estrogen and pituitary growth hormone. Growth hormone may mediate its effects primarily through local production of insulin-like growth factor I (IGF-I). [7]

Estrogens, as well as other steroid hormones biosynthesized, utilize cholesterol (Figure

1.2, 1d) as a starting material. Cholesterol can be obtained from circulating lipoproteins, by de novo synthesis from acetyl-CoA subunits, and from cholesterol esters stored within lipid droplets. In the ovary, cholesterol is largely derived from uptake of plasma low- density lipoprotein LDL. LH stimulates the expression of LDL receptors on the membranes of thecal cells to faciliate uptake. Cholesterol liberated from LDL inside cells is transported by the steroidogenic acute regulatory protein, (StAR) into the mitochondria where it is a substrate for the cholesterol side-chain cleavage enzyme CYP11A1, which transforms it to pregnenolone (Figure 1.2, 1e). From pregnenolone, the major biosynthetic pathway in the corpus luteum is the ∆4-pathway, which converts pregnenolone to progesterone (Figure 1.2, 1f) via 3β-hydroxysteroid dehydrogenase/isomerase and some progesterone is further metabolized to 17- hydroxyprogesterone (Figure 1.2, 1g) by CYP17. In the thecal cells of the developing follicle, the ∆5-pathway predominates, in which CYP17 catalyzes the formation of 17- hydroxypregenolone directly followed fission of the steroid C17-20 bond by the same enzyme resulting in dehydroepiandrosterone (DHEA). DHEA (Figure 1.2, 1h) is a 3 substrate for 3β-HSD in the thecal cells where it is converted to androstenedione (Figure

1.2, 1i). Androstenedione diffuses through the basement membrane separating theca and the granulosa and enters granulosa cells. In the granulosa, androstenedione encounters aromatase, whose biosynthesis in under control of FSH. It is here where aromatase synthesizes the estrogens estrone directly and estradiol by the additional reversible action of 17β-hydroxysteroid dehydrogenase. [4]

4 HO Cholesterol (1d)

CYP11A1

O

3β-hydroxysteroid CYP17 HO dehydrogenase (3β-HSD) Pregnenolone (1e) O O OH

HO O 17-Hydroxypregnenolone Progesterone (1f)

CYP17 CYP17

O O OH

HO HO Dehydroepiandrosterone 17-Hydroxyprogesterone (1g) (1h) (DHEA) 3β-HSD 17β-hydroxysteroid dehydrogenase (17β-HSD) O OH

O O Androstenedione (1i Testosterone

CYP19 CYP19 O OH 17β-HSD

HO HO Estrone 17β-Estradiol

Figure 1.2: Principal pathways of steroid hormone biosynthesis in the human ovary. [4]

5 The CYP19 gene encodes the enzyme enzyme aromatase and is part of the cytochrome

P450 superfamily. The protein product of this gene is a 55 kDa protein 503 amino acids

in length. The enzyme in human is expressed in numerous tissues such as placenta,

adipose, ovarian granulosa cells, testicular Sertoli and Leydig cells, and the brain.

Estrogens have an important mitogenic role in hormonally dependent pathologic

processes, such as breast and endometrial cancer. Inhibition of aromatase, which

catalyzes the terminal step for the biosynthesis of estrogens, could be a useful strategy for treatment. [8]

It is well known that estrogens are important in the growth of breast cancers in both pre- and postmenopausal women. However, breast cancer estrogen sensitivity has been

demonstrated to increase with patient age. Two-thirds of breast cancers in

postmenopausal women have tumors that are positive for estrogen and progesterone

receptors, compared to premenopausal women where the presence of these receptors

appears in less than half of tumors. The endogenous ligands for these receptors, estrogens

and progestins, stimulate cell proliferation directly by increasing the rate of early response genes and indirectly through the stimulation of growth factors. [6] Such dependence on sex steroids for tumor growth makes hormonal therapeutics an inviting strategy for treatment. [9]

Based on the concept that estrogen is the proximate regulator of cell proliferation, two general strategies have been developed for the treatment of hormone-dependent breast cancer. The first strategy is to block estrogen receptor action by disrupting its interaction 6 with and activation by estradiol. Antiestrogens, such as tamoxifen, bind to the estrogen receptor and interfere with the transcription of estrogen-induced genes. The efficacy of tamoxifen has been established in the treatment of postmenopausal, hormone-responsive breast cancer. [10] Tamoxifen increases long-term survival, reduces recurrences, and has few side effects. In addition to tamoxifen’s antagonistic effects, it also behaves as a weak or partial agonist. While exhibiting antagonistic activity in the breast, the partial agonist effects in other parts of the body have lead to the formation of secondary tumors of the liver and uterus. [11] A second pharmacological approach is to block estradiol synthesis catalyzed by the cytochrome P450 enzyme aromatase (CYP19). Aromatase has been a particularly attractive target for inhibition in the treatment of hormone-dependent breast cancer since the aromatization of androgen substrates is the terminal and rate-limiting step in estrogen biosynthesis. [12-14]

1.2 THE CATALYTIC MECHANISM OF AROMATASE

Aromatase is a cytochrome P450 enzyme responsible for the biosynthesis of estrogens from androgen precursors. Both androstenedione and testosterone may act as substrates for the enzyme and convert them to estrone and estradiol, respectively. In common with all P450 enzymes, aromatase contains a single iron protoporphyrin IX with a cysteine thiolate anion acting as the proximal iron ligand. (Figure 1.3) It is here where with electrons donated from NADPH-cytochrome P450 reductase activate molecular oxygen to a species capable of performing each step toward the aromatization process.

7 CH3

H3C N N Fe N N H3C CH3

S-Cys HOOC COOH

Figure 1.3: Prosthetic of cysteinato-heme enzymes including cytochrome P450 aromatase: an iron-protoporphyrin IX linked with a proximal cysteine ligand.

8 HHS = 1/2 ROH O RH H O Fe H O 2 2 RH S S = 5/2 H R a Fe O S e- Fe i S b

RH 1- Fe S = 2 c S O RH O2 Fe h S 1- O O RH

H2O d Fe S = 0 H g S O 2- O RH H e O e- Fe f 1- O RH S O RH Fe H O S Fe H+ S H+

Fe = Iron Protoporphyrin IX

Figure 1.4: Schematic representation of different intermediates generated during the catalytic cycle of cytochrome P450 enzymes. The charge and ground state spin quantum number for select intermediates are indicated.

9 The activation of molecular oxygen by P450s occurs in stages, starting with the enzyme

in its resting state. (Figure 1.4, a) At this point, the enzyme is a low-spin ferric iron and is

it believed the sixth distal ligand is water. [15,16] Substrate binding perturbs the water

structure at the P450 active site and shifts the spin state equilibrium from predominately

low-spin to high spin. (Figure 1.4, b) In the case of CYP101, the camphor binding shifts

the redox potential for –303 mV to –173 mV, making reduction by the accessory redox

protein putidaredoxin more feasible. [17] After the redox protein provides the reducing

equivalent to the substrate bound P450, Fe(III) moves to Fe(II) while remaining in the

high-spin state. It is here where molecular oxygen can bind avidly to P450. An additional

electron is injected into dioxygen-P450 by a redox partner, generating the peroxo

intermediate. The details from this point on, including dioxygen bond cleavage, the

structure of the activated species, and the mechanism by which the activated oxygen still

pose a challenge for our understanding of P450 catalysis. Subsequent protonation of this species likely results from proton transfer from an organized conduit of water molecules adjacent to the P450 heme thiolate complex. The available evidence strongly suggests that the activated species consists of a single oxygen atom bound to iron and believed to be a porphyrin cation radical. (Figure 1.4, h)

Biotransformation of androgens by aromatase proceeds in three of the oxidative steps

previously described, each consuming a single mole of reduced nicotinamide adenine

dinucleotide phosphate (NADPH) and molecular oxygen. The intermediates of the

biotransformation are displayed in Figure 1.5.

10 R R HO

O O

R = O Androstenedione R = OH Testosterone

R R OH O +/- H2O HO

O O

R

HO

R = O Estrone R = OH 17β-Estradiol

Figure 1.5: Proposed intermediates in the catalytic turnover of cytochrome P450 aromatase.

11 The first oxidation occurs at the androgen 19-methyl group by the classical hydrogen

atom abstraction-hydroxy radical rebound mechanism originally proposed by Groves and

coworkers. [18] The first hydroxylation occurs with retention of configuration. [19,20]

3 Experiments using [19- H3]androst-4-ene-3,17-dione show that the small tritium kinetic

isotope effect observed with aromatization is exclusively associated with the first hydroxylation step. [21,22] In addition, using the labeled suicide substrate analog [19-

3 H3]androst-3-ene-3,6,17-trione also revealed a marked tritium kinetic isotope effect with the first 19-hydroxylation. [23] The second hydroxylation step stereoselectively removes

the 19-pro-R hydrogen, without an apparent kinetic isotope effect to yield the 19-gem

diol. [24,25] Whether the 19-gem-diol remains or dehydrates to the 19-aldehyde for the

final catalytic step remains an area of active dispute.

The third step oxidatively cleaves the C10-C19 bond resulting aromatization of the

steroid A-ring and release of the 19-methyl group as formic acid. A number of

mechanisms have been proposed for the final catalytic step of aromatase, although a

single unifying mechanism as yet to be agreed upon. It has been known for many years

that the 1β- and 2β- hydrogens are stereoselectively lost to the aqueous medium. [26-28].

Many of the catalytic mechanisms proposed to date are reviewed here with their

underlying experimental details

In 1974, Hosoda and Fishman [29] proposed that the final catalytic step was initiated by

2β-hydroxylation of androstenedione by aromatase. They synthesized 2β-hydroxy-19- oxo-androst-4-ene-3,17-dione as a potential intermediate in estrogen biosynthesis. In 12 addition, 2α-hydroxy-19-oxo-androst-4-ene-3,17-dione, 2α,19-dihydroxy-androst-4-ene-

3,17-dione, and 2β,19-dihydroxy-androst-4-ene-3,17-dione were prepared. In the

presence of water or neutral pH, the two epimeric 2,19-diols as well as the 2α-hydroxy-

19-aldehyde were not aromatized, but the 2β-hydroxy-19-aldehyde was rapidly and

essentially completely converted to estrone. The proposed mechanism is initiated by

anchimerically-assisted elimination of the 2β-hydroxyl group followed by collapse of the

dienone intermediate. (Figure 1.6)

O OH OH HO H O H HO OH H O HO HO HO H O

Figure 1.6: Postulated mechanism for the base-catalyzed aromatization of 2β-hydroxy-

19-oxo-androst-4-ene-3,17-dione. [29]

Morand and coworkers [30] considered the possibility that androgen epoxides could be catalytic intermediates in the biosynthesis of estrogens. They synthesized three epoxides,

17β-hydroxy-4β,5β-oxido-androstan-3-one, 17β,19-dihydroxy-4β,5β-oxido-androstan-3-

one, and 3-acetyl-3,19-dihydroxy-5β,6β-oxido-androstan-17-one. Only when 17β,19-

dihydroxy-4β,5β-oxido-androstan-3-one was incubated with human placental

microsomes followed by extraction of the medium revealed disappearance of the

precursor and formation of 17β-estradiol. The other potential precursors were 13 metabolized to unidentified products. (Figure 1.7) A mechanism was proposed which required hydration of the 19-aldehyde to the gem-diol and accounted for stereoselective removal of the 1β- and 2β- hydrogen atoms. Proton-mediated epoxide ring opening results in the formation of a C5 carbonium ion, which drives elimination of the C10 substituent with concomitant C5-C10 unsaturation. The 1β- hydrogen atom is then removed followed by a concerted shift of the C5-C10 double bond and elimination of the

4β-hydroxyl group. The product is the labile 1(10), 3-dienone, which can be rapidly aromatized by removal of the 2β-hydrogen atom during enolization.

14 OH OH HO HO

O O OH OH

H

O O

OH2

HO

Figure 1.7: Postulated aromatization mechanism of androgen epoxides. [30,31]

15 The acid- and base- catalyzed reactions of the 4β,5β- and 4α,5α- epoxyandrostane-

3,17,19-trione were also examined to see if epoxide ring opening in these compounds

would result in elimination of the C10 substituent and subsequent formation of estrone.

[31] It was found that relatively strong concentrated and aqueous mineral acids (i.e.

H2SO4 and HClO4) or methanolic potassium hydroxide treatment were necessary to

promote reactions of the epoxides at a reasonable rate. Elimination of the C10 substiuent

as originally envisaged did not appear to be the major mode of reaction for these

compounds.

Akhtar and coworkers [32] studied the mechanism of the aromatization of

androstenedione in terms of the involvement of oxygen atoms using a number of labeled

precursors. (Figure 1.8) The most notable of these labeled precursors were 19-hydroxy-

androst-4-ene-3,17-dione and androst-4-ene-3,17,19-trione in which the C19 was labeled

with 2H and 18O. In order to follow the fate of formate in the labeled atoms at C19 during

the aromatization reaction, formic acid was benzylated and analyzed by mass

2 18 spectrometry. [19- H] androst-4-ene-3,17,19-trione was aromatized under an O2 atmosphere with human placental microsomes and analysis of the isolated formate revealed that 90% of the formate retained a single atom of 18O. The non-deuterated

substrate androst-4-ene-3,17,19-trione showed only 70% incorporation of a single 18O

16 18 atom. When this experiment was repeated under an atmospshere of O2, the O

aldehyde oxygen was retained in the formate to an extent of 82%. Aromatization of 19S-

2 18 19-hydroxy [19- H1, O]androst-4-ene-3,17-dione with placental microsomes and analysis of the formate product showed that 90% of the 18O was transferred to the 16 formate product, implying that the oxygen incorporating during the first hydroxylation

2 step is retained. Reactions of 19S-19-hydroxy [19- H1]androst-4-ene-3,17-dione under

18 18 an O2 atmosphere were found to contain 0.9 atom of O, which could have been

incorporated in either the second or third hydroxylation steps. The overall conversion of

androst-4-ene-3,17-dione was studied to 3-hydroxy-estr-1,3,5(10)-triene-17-one (estrone)

was studied on an atmosphere of 18O and the mass spectrum of the biosynthetic formate

18 revealed peak at m/z 140 corresponding to [ O2] formate, which was five-fold more intense than the [18O] formate at m/z 138, supporting that two of the three moles of

molecular oxygen are incorporated in the biosynthetic formate. The possible involvement

of 10β-hydroxy-estr-4-ene-3,17-dione in estrogen biosynthesis was also considered by

Akhtar and coworkers. The 3H-labeled species was incubated with placental microsomes and no significant formation of estrogens could be detected. Later investigations by Caspi

2 [33] using [16,16,19- H3] androst-4-ene-3,17-dione and 10β-hydroxy-estr-4-ene-3,17-

dione confirmed these findings.

17 O 2H 18O 2H

O O 18 16 O2 O2

2 2 18 O H 18 O H 90 % 1 O 82 % 1 O + + atom atom OH HO OH HO

H H HO 2H H18O 2H

O O 18 16 O2 O2

2 2 18 O H 18 O H 90 % 1 O 90 % 1 O + + atom atom OH HO OH HO

H H H 18 18 2 O2 O H Major product + 18OH O HO

Figure 1.8: Summary of the 18O labeling studies of Akhtar and coworkers. [32]

18 Taking the results of the 18O labeling studies and assuming that the 19-gem diol dehydration to the 19-aldehyde occurs, and these species are obligatory intermediates in the biosynthesis of estrogens, Akhtar proposed mechanistic possibilities for the final catalytic step of aromatase. The first assumes that third oxidative step introduces either a

1β- or 2β- hydroxy group which reacts with the 19-carbonyl producing four and five- membered cyclic hemiacetals. The hemiacetals are then proposed to undergo a non- enzymatic electrocyclic rearrangement to produce formate and estrogen. The second mechanistic alternative rationalizes the isotopic labeling data proposes that NADPH and oxygen participate in the formation of an iron peroxo species which reacts with the carbonyl group of the aldehyde to produce a peroxo acetal intermediate, which may rearrange with concomitant 1β-hydrogen atom removal to give the aromatized product.

(Figure 1.9)

O CysS CysS CysS Fe Fe Fe OH O H OO OH OO H OH + H

O O HO

Figure 1.9: Deformylation of 19-oxo androstenedione by the aromatase peroxo-iron intermediate as proposed by Akhtar and coworkers. [32]

19 Cole and Robinson attempted to model the peroxide reaction for aromatase by synthesizing a number of chemical models for the peroxohemiacetal intermediate. [33]

The methoxyhydroperoxide analog of androst-4-ene-3,17-dione failed to afford estrone under a variety of conditions. A possible lack of reactivity for this species was the hypothesis that enolization was a prerequisite for 1β- hydrogen atom removal. After a number of failed attempts to produce an enol ether analog of the methoxyhydroperoxide species, attention was turned to the synthesis of the 19-oxo derivative. Treatment of this compound with 30% hydrogen peroxide resulted in a rather slow, but smooth aromatization affording the doubly protected estrogen derivative. Furthermore, production of an equivalent amount of formic acid occurred.

Graham-Lorence and coworkers [34] rationalized the role of the reduced ferrous dioxygen species in the final catalytic step with a homology model constructed largely from the X-ray crystal structure of CYP102 from Bacillus megaterium. Based on the active site model, they suggested aspartate 309 was ideally positioned to perform the 2β- proton abstraction, while either lysine 473 or histidine 475 donates a proton to 3-keto of androstenedione, accomplishing the prerequisite enolization. Generation of the reactive peroxo species was facilitated by the presence of a “threonine switch.” Previous site- directed mutagenesis studies mutating this amino acid or aspartate 309 to apolar amino acids results in an inactive protein indicating they may be involved in the mechanism which shuttles protons to the distal oxygen atom after the second electron reduction promoting O-O bond scission. This proposal requires dehydration of the 19-gem diol and the resulting aldehyde hydrogen bonds to the side chain hydroxyl of threonine 310. A 20 hydrogen bond of this nature could interrupt normal proton delivery to the peroxo

intermediate extending its lifetime. In addition, the aldehyde carbonyl is polarized

making more susceptible to nucleophilic attack.

Caspi and others [35] revisited the obligatory intermediacy of the 2β-hydroxy-androst-4-

ene-3,17-dione intermediate in estrogen biosynthesis to address the observation that

oxygen atoms from the third oxidative step are incorporated into the biosynthetic

formate. To determine whether this intermediate would give the same product isotope

distribution shown by Akhtar with experiments using 19-oxo precursors, [2β-18O,19-3H]-

2β-hydroxy-androst-4-ene-3,17-19-trione was prepared and incubated with human placental microsomes. Formic acid was recovered and converted to the benzyl formate and analyzed by GC-MS. No peaks were present at m/z = 138 indicating an absence of

18O in the released formate.

Covey and coworkers [40] proposed a mechanism based on observations made during

studies on the metabolism of 19-methyl androgens. Incubation of (1R)- and (1S)-1-

hydroxyethyl-estr-4-ene-3,17-dione with placental microsomes both produced 10-acetyl-

estr-4-ene-3,17-dione as the major product and no conversion to estrogens was detected.

18 18 When an O2 atmosphere was used in incubations with the 1S enantiomer, O was not

incorporated into the 10-acetyl product. In contrast, when this experiment was conducted

with the 1R enantiomer, 18O was incorporated into the 10-acetyl product. These results

supported that the enzyme catalyzed a stereoselective dehydration of the gem-diol

produced after the second oxidative step where the pro-R hydroxyl group was 21 stereoselectively removed. In agreement with a previous proposal by Osawa and coworkers, that the gem-diol is the actual enzyme-bound species that undergoes the third

monoxygenation and can provide an explanation why 10-acetyl-estr-4-ene-3,17-dione is

not metabolized to estrone. The acetyl group is less readily hydrated to the gem-diol than

androst-4-ene-3,17,19-trione, and therefore may bind to aromatase but be unable to form

the gem diol required for conversion to estrone. To account for the stereoselective loss of

the 1β- and 2β- hydrogens from the medium, this mechanism was initiated with 1β-

hydrogen atom abstraction. The resulting C1 radical is then proposed to fragment to a

gem-diol radical and the 1(10),4-dienone, followed by hydroxyradical rebound to C19.

Orthoformate sandwiched between the heme group and the the 1(10),4 dieneone seems

ideally positioned for proton transfer and dehydration reactions to produce formic acid

and estrone.

22 CysS CysS Fe Fe O OH HO HO H HO H HO H

O O

CysS OH CysS Fe Fe HO H OH HO HO HO H H

O O

CysS Fe O HO H

HO

Figure 1.10: Mechanism for the final catalytic step of cytochrome P450 aromatase proposed by Covey and coworkers. [40]

23 Despite all of the mechanisms proposed here, the final step of the reaction sequence has

not been resolved and is still the subject of considerable debate. Most of these proposals

relied on results that were obtained with human placental microsomes. In the course of

aromatase purification, a P450 non-glycoprotein was eluted in the void volume. Although

this protein could not directly affect aromatase activity, it could metabolize androgens.

[41] In addition, Bednarski and Nelson have raised the controversy that 19-hydroxy- and

19-oxo-androstenediones are not free, obligatory intermediates in the aromatization of

androstenedione by human placental aromatase, but rather are products of their own

autonomous cytochrome P-450-dependent, microsomal enzymatic activities. [42] These

observations emphasize that caution should be exercised when interpreting the result of mechanistic studies conducted with microsomes.

The short lifetimes and high reactivity of catalytic intermediates in the later parts of the cytochrome P450 cycle and in the final deformylation step of aromatase make elucidation of detailed descriptions for these processes by current experimental methods difficult. An alternative method to study short-lived, highly reactive intermediates and the ultrafast catalytic steps of enzymes is to use computational quantum chemistry. One particular area of this discipline that has been exceptional for treating relatively large systems within a reasonable timeframe is density functional theory. In particular, the hybrid density functional B3LYP has been a rising star for the accurate treatment of systems containing first and second row elements as well as transition metals. The application of density functional theory to study the structures, mechanisms, and electron transfer processes occurring in biological systems has recently been extensively reviewed. 24 [43-46] In particular, these methods have been particularly successful contributors to our

current understanding of dioxygen activation in cytochrome P450 enzymes and the

mechanisms of P450-mediated oxidation. [47] In Chapter 2, the application of density

functional theory using various functionals is applied to study the structure and

spectroscopic properties of dioxygen intermediates in the P450 catalytic cycle and the

theoretical results are compared with available experimental data. Chapter 3 describes the

results of a comprehensive theoretical study of the final catalytic step of aromatase.

1.3 INHIBITORS OF AROMATASE

Aromatase inhibitors, in general, belong to one of three groups 1) Competitive, reversible inhibitors; 2) Irreversible inhibitors of the affinity-label type; and 3) Mechanism-based irreversible inhibitors. Affinity label and mechanism-based aromatase inhibitors are generally steroidal and their biological activities and structure-activity relationships have been extensively reviewed elsewhere. [48-51] Only compounds with this classification that have experienced significant clinical use will be reviewed here. This review will cover nonsteroidal, reversible competitive and natural product inhibitors of aromatase.

Nonsteroidal aromatase inhibitors are competitive and reversible in nature. They produce a Type II binding spectra (λmin = 385-395 nm and λmax = 425-435 nm) as a result of

coordination of the nitrogen heterocycle to the iron atom of the iron protoporphyrin IX at

the active site of aromatase. [52]

25 Aminoglutethimide (AG; Figure 1.11, 1j) is the prototype nonsteroidal inhibitor of

aromatase. AG was originally used as an antiepileptic agent, but this therapeutic use was

discontinued due to serious side effects. AG inhibits CYP11A1 and other steroidogenic

P450 enzymes resulting in decreases in testosterone, dihydrotestosterone,

androstenedione, progesterone, and 17-hydroxyprogesterone. [53] Our group has

determined that (±) AG inhibits aromatase in human placental microsomes with an IC50 of 2.8 µM and an apparent Ki of 1.4 µM. [54] Graves and coworkers showed that the

dextrorotary enantiomer of AG is 38 times more potent than the levorotary isomer in inhibiting aromatization in human placental microsomes and has a 36-fold better spectral binding constant. [55] AG has been used clinically with some success to treat patients with metastatic breast cancer. Doses of AG must be administered with oral hydrocortisone since the compound effectively blocks endogenous secretion of this hormone. [56]

26 H N N O NO Cl Cl N N Cl

O Cl

Cl NH2 Aminoglutethimide Miconazole Clotrimazole 1j 1k 1l N

N O O O H N Cl Cl O N

CH Ketoconazole 3 1m

N N N N

NC CN CN

Fadrozole CGS18320B 1n 1o

Figure 1.11: Aminoglutethimide (AG) and imidazole inhibitors of aromatase including antimycotic agents.

27 The realization that the nitrogen heteroatom of aminoglutethimide could coordinate with

the heme group of aromatase prompted many groups to consider other nitrogen

heterocycles. An obvious starting point where the imidazole antimycotic drugs (i. e.

miconazole, clotrimazole , and ketoconazole). Miconazole (1k) is a potent aromatase

inhibitor with an IC50 of 0.6 µM and an apparent Ki of 55 nM in human placental microsomes. Clotrimazole (1l) and ketoconazole (1m) are less potent than miconazole with IC50 s of 1.8 and 60 µM, respectively. [57] The discovery of CGS16949A (1n)

(fadrozole) and CGS18320B (1o) has inspired both academic and industrial laboratories to pursue avenues toward aromatase inhibitors with improved selectivity and pharmacological profiles in humans. [58] Fadrozole and CGS18320B still have inhibitory activity with respect to aldosterone, progesterone, and corticosterone biosynthesis. The apparent Ki of racemic fadrozole in human placental microsomes is 1.6 nM. (S)-

Fadrozole is equipotent to the racemate and the (R)- isomer is less potent with an

apparent Ki of 39 nM.

Cl N N N N N N N N N NC CN N N N NC CN Vorozole Anastrozole Letrozole 1p 1q 1r

Figure 1.12: Triazole inhibitors of aromatase.

28 Competitive nonsteroidal inhibitors can also be constructed with a triazole ring. The

triazole derivative R76713 (vorozole; Figure 1.12, 1p) in its racemic form inhibits

aromatization in human placental microsomes with an apparent Ki of 1.3 nM. (+)-

Vorozole is more potent than its enantiomer, (-)-vorozole each having apparent Ki values of 0.7 nM and 18 nM, respectively. The racemate nor the enantiomerically pure compounds had any effect on P450-dependent cholesterol biosynthesis, cholesterol side chain cleavage, 7α-hydroxylation, or 21-hydroxylase activity. [45] Anastrozole (1q) is a triazole aromatase inhibitor lacking an asymmetric center with an IC50 for aromatization

in human placental microsomes of 12 nM. [59] In vitro, anastrozole has little impact on

the activities of other steroidogenic P450s and has been shown to decrease peripheral

aromatase inhibitory activities in monkeys by 50-60 %. [60] An additional triazole

compound potently inhibiting aromatase in vitro and in vivo is CGS20267 (letrozole; 1r).

[61] Letrozole has an IC50 of 11.5 nM in human placental microsomes and inhibits

estradiol production in vitro in LH stimulated hamster ovarian tissue with an IC50 of 20 nM. In adult female rats, a 14 day treatment with 1mg/kg p.o. q.d. completely interrupts ovarian cyclicity and suppresses uterine weight to that seen 14 days after ovariectomy. In adult female rats bearing estrogen-dependent dimethylbenzanthracene-induced mammary tumors, 0.1 mg/kg p.o q.d. for 42 days caused almost complete tumor regression.

Letrozole does not affect adrenal steroidogenesis even at doses several orders of magnitude required to inhibit estrogen biosynthesis.

29 1.4 NATURAL PRODUCT INHIBITORS OF AROMATASE

1.4.1 FLAVONOIDS

Kellis and Vickery found that several naturally occurring and synthetic flavones were

capable of inhibiting the aromatization of androstenedione and testosterone to estrogens

catalyzed by human placental and ovarian microsomes. [62] Activity was measured by

the release of tritium from [1,2-3H] androstenedione. The most potent compound identified was 7,8-benzoflavone (2a; IC50 = 0.07 µM) followed by chrysin (2b; IC50 =

0.50 µM). Dose response studies with apigenin (2c), flavone (2d), flavanone (2e), and quercetin (2f) showed they were less efficacious inhibitors with IC50 values of 1.2, 8, 8,

and 12 µM, respectively. 7,8-benzoflavone and chrysin were competitive with the substrate androstendione in kinetic studies and demonstrated Ki values of 21 nM and 0.26

µM, respectively. When 7,8-benzoflavone was titrated into a solubilized preparation of

aromatase prequillibrated with androstenedione a reverse Type I spectrum is observed

(λmin = 388-390 nm and λmax = 420 nm). Analyses of the spectral binding study indicated

that this compound binds at a single site on the aromatase enzyme and the requirement

for prequillibration with the natural substrate indicates the interaction with aromatase and

not other P450 enzymes.

30 O OH O 5 6 2' 7 3' O HO O 8 10 6' 4' 9 5' 7,8-Benzoflavone Chrysin 2a 2b OH O O 5 6 3 C A HO O 7 O 8 B OH Apigenin Flavone 2c 2d O OH O OH

OH O HO O

OH Flavanone Quercetin 2e 2f

OH O O-Rutinose

OH O HO O

OH 2-phenyl-4H-naptho-[1,2b]furan Rutin 2g 2h OH OH

OH HO O

OH Catechin 2i

Figure 1.13: Flavonoid compounds investigated for aromatase inhibitory activity.

31 Additional studies were conducted to investigate the inhibition of aromatase by several

derivatives of 7,8-benzoflavone. [63] First, elimination of the carbonyl group by conversion of the pyran-4-one to a furan with the synthesis of 2-phenyl-4H-naptho-

[1,2b]furan (2g; IC50 >100 µM) caused almost complete loss of activity. 5-hydroxy- (IC50

= 10 µM) and 6-hydroxy-7,8-benzoflavone (IC50 = 12 µM) decreased activity several

hundred fold. The 7-hydroxy (IC50 = 0.19 µM) and 8-hydroxy (IC50 = 0.35 µM) analogues were only slightly less potent, while the 10-hydroxy compound (IC50 = 0.07

µM) had equal activity to the parent 7,8-benzoflavone. Interestingly, introduction of a 9-

hydroxy group (IC50 = 0.02 µM) increased aromatase inhibitory activity several fold.

Kinetic analysis revealed the 9-hydroxy analog was also competitive with

ansdrostenedione (Ki = 5 nM, Km = 10 nM). This compound also induced a reverse type I

shift of the aromatase Soret band.

Ibrahim and Abul-Hajj [64] found several synthetic flavones were able to inhibit the aromatization of androstendione to estrone by human placental microsomes. In these studies, flavones and flavanones demonstrated higher aromatase inhibitory activity than isoflavones and isoflavanones. The most promising compounds identified among those tested were flavone (IC50 = 10 µM), flavanone (IC50 = 8 µM), 4’-hydroxyflavone (IC50 =

10 µM), 7-hydroxyflavone (IC50 = 0.5 µM), and 7,4’-dihydroxyflavone (IC50 = 2.0 µM).

Kinetic analysis of 7-hydroxyflavone demonstrated it was competitive with the

androstendione substrate (Ki = 0.25 µM, Km = 58 nM). Within the series examined, introduction of a 4’-hydroxyl group in the B ring resulted in a negligible change in

32 inhibitory activity. Hydroxylation at the 3-position in ring C resulted in a significant

decrease in activity compared to flavone (3-hydroxyflavone, IC50 = 140 µM). Reduction

of the carbonyl functionality of almost abolished inhibitory activity, with flavan-4-ol

having an IC50 of 120 µM.

OH OH O

HOO HO O Daidzein Equol 2j 2k

OH OMe OH O OH O

HO O HO O Genistein Biochanin A 2l 2m

Figure 1.14: Isoflavonoid compounds investigated for aromatase inhibitory activity

Eleven flavonoid compounds were compared with aminoglutethimide for their ability to

inhibit aromatase enzyme activity in human preadipocyte cell culture. [65] 7,8-

benzoflavone was the most potent aromatase inhibitor in this bioassay system with an

IC50 of 0.5 µM. Chrysin and aminoglutethimide followed with IC50 values of 4.6 and 7.4

µM, respectively. The least potent compounds were flavone (IC50 = 68 µM) and the isoflavone biochanin A (IC50 = 113 µM,). Catechin (2i), daidzein (2j), equol (2k), genistein (2l), 5,6-benzoflavone, quercetin (2f), and rutin (2h) had no effects on aromatase inhibitory activity in human preadipocytes. Kinetic analyses using the cell

33 culture system again implicated 7,8-benzoflavone as the most active (Ki = 0.2 µM, Km =

62 nM). As in the dose response studies, AG followed with Ki values of 2.4 µM. Flavone

(2d) and Biochanin A (2m) demonstrated Ki values of 22 and 49 µM, respectively.

1.4.2 OTHER NATURAL PRODUCTS

Adlercreutz and coworkers have found the main mammalian lignan enterolactone (2n;

Figure 1.15) and some of its diphenolic derivatives are moderate or weak inhibitors of

human aromatase in placental microsomes and the choriocarcinoma cell line JEG-3. [66]

Dose-response studies with enterolactone on androstenedione aromatization

demonstrated an IC50 value of 14 µM. 4,4’-Dihydroxy enterolactone and nordihydroguaiaretic acid (2q) were more active in this assay system demonstrating IC50

values of 6 µM and 11µM, repectively. Dose response studies using the JEG-3 cell line

produced similar results. Kinetic analysis of enterolactone showed that it was competitive

with androstenedione (Ki = 6 µM, Km = 20 nM). In addition, enterolactone induces a

reverse type I spectrum (λmax = 417 nm, λmin = 392 nm) with a partially purified

aromatase preparation.

34 HO R3 H

O HO R4 H O

OH R3' OH R4' Nordihydroguaiaretic acid (2q) R3 = R3' =OH, R4=R4'=H, Enterolactone (2n) R3 = R4 = R4' = OH,R3' = H, 3'-Demethoxymatairesinol (2o) R3 = R3' =H, R4=R4'=OH, Didemethoxymatairesinol (2p)

Figure 1.15: Mammalian lignans investigated for aromatase inhibitory activity. [66,67]

Seven lignans and six flavonoids were evaluated for their ability to inhibit the aromatase enzyme in human preadipocyte cell culture. Enterolactone (2n; IC50 = 74 µM, Ki = 14.4

µM) and the hydroxylated analogs 3’-demethoxymatairesinol (2o; IC50 = 84 µM, Ki = 5.0

µM) and didemethoxymatairesinol (2p; IC50 = 60 µM, Ki = 7.3 µM) were weak inhibitors.

The Km of androstenedione in these studies averaged 30 nM. Unlike the lignans

considered in this study, coumesterol displayed moderate activity (IC50 = 17 µM, Ki = 1.3

µM). [67]

Hida and coworkers discovered TAN-931 (2r; Figure 1.16) in a culture filtrate of

Penicillium funiculosum No. 8974. This compound had an IC50 value for aromatase

inhibition of 17 µM. The 5-methoxy, N,N-dimethyl amide analog (2s) had a comparable

IC50 value of 18 µM, but in contrast the parent TAN-931, this analog produced a

significant decrease in plasma E2 and uterine weight after a 100 mg/kg dose. [68]

35 OH O CHO OH O CHO

CH3 N COOH CH OH OH OH OMe 3 O TAN-931 2s 2r

Figure 1.16: TAN-931 and its N,N-dimethyl amide analog isolated from cultures of

Penicillium funiculosum No. 8974. [68]

A number of sesquiterpene lactones have been isolated and characterized from different

Asteraceae species of northwestern Argentina. (Figure 1.17) Eleven of these compounds were evaluated for aromatase inhibitory activity with human placental microsomes. [69]

10-Epi-8-deoxycumambrin B (2t; IC50 = 7 µM, Ki = 4 µM), dehydroleucodin (2u; IC50 =

15 µM, Ki = 21 µM), and ludartin (2v; IC50 = 55 µM, Ki = 23 µM) were moderate to weak

inhibitors of androstenedione aromatization. Of the three, only dehydroleucodin and

ludartin induced reverse type I difference spectra with a partially purified aromatase

preparation. None of the compounds inhibited the cholesterol side-chain cleavage enzyme at up to 200 µM concentration. Cytotoxicity of these sesquiterpene lactones has been attributed to the α-methylene-γ-lactone functional group. Sodium borohydride reduction to the 11,13 dihydro analog (2w) abrogated this compounds cytotoxic activity against JEG-3, HeLa, and COS-7 cells. This compound also displayed improved inhibitory activity (IC50 = 2 µM, Ki = 1.5 µM) and was a more potent at inducing a reverse type I spectrum. [70]

36 OH O H

H H O O O O 10-Epi-8-deoxycumambrin B Dehydroleucodin 2t 2u OH H

O H H O O O O

Ludartin 11β,13-Dihydro-10-epi-8-deoxycumambrin B 2v 2w

Figure 1.17: Sesquiterpene lactones isolated from different Asteraceae species from northwestern Argentina [69,70]

Minami and coworkers elucidated the structures of several diterpenoids from the bark of

Thuja standishii. 1 µM screening of these analogs identified two active analogs, standishnal (2x) and the diacetyl analog (2y). These diterpenes inhibited aromatization of androstendione 50.2 and 38.6 percent, respectively compared to the steroidal suicide inhibitor formestane which inhibited activity 63.7 percent. [71]

37 CHO

RO OR

R=H, Standishnal (2x) R=OAc (2y)

Figure 1.18: Structure of standishnal from Thuja standishii and the diacetyl analog. [71]

Since the identification of AG as an inhibitor of aromatase, nonsteroidal competitive inhibitors have evolved from being relatively non selective with respect to their effects on adrenal steroidogenesis to highly selective aromatase inhibitors. This evolution has been the result of identification of an optimal heme-coordinating heterocycle and core scaffold

for its attachment. In parallel with the discovery of these highly efficacious aromatase inhibitors, a number of natural product compounds with aromatase inhibitory activity have been discovered. Many of these compounds demonstrate weak to moderate potency for aromatase inhibition, despite their apparent lack of a heme coordinating functional group as efficiacious as a nitrogen-containing heterocycle. With these properties, it is likely that natural products may offer a series of novel scaffolds for the construction of aromatase inhibitors. The most potent representative of the natural product inhibitors are the flavonoids. Based on data from homology modeling and site-directed mutagenesis of the aromatase protein, a binding orientation was predicted in which the A and C rings of

the flavone mimic the C and D rings of the steroid substrate, respectively. [72] This

orientation places the flavone 4-keto functionality in the same space likely occupied by

38 the steroid 19-angular methyl group and explains the difference spectra obtained with these compounds. In general, isoflavones are less efficacious aromatase inhibitors than the flavones. Deductions based on the modeling results obtained by other research groups assuming that isoflavones bind in a manner reminiscent of flavones have lead us to envision that introduction of a nitrogen heterocycle at the 2-position may result in novel aromatase inhibitors. Chapter 4 and Chapter 5 describe the biological evaluation of a series of novel pyridine-containing and azole isoflavones, respectively and their structure activity relationships for the inhibition of aromatase.

1.5 REFERENCES

1. Callard, G. V.; Petro, Z.; Ryan, K. J.; Phylogenetic distribution of aromatase and

other androgen-converting enzymes in the central nervous system. Endocrinology

1978, 103, 2283-2290.

2. Callard, G. V.; Petro, Z.; Ryan, K. J. Aromatization and 5α-reduction in the brain

and non-neural tissues of a cyclostome (Petromyzan marinus) Gen Comp

Endocrinol 1980, 42, 155-159.

3. Callard, G. V. Aromatization is cyclic-AMP dependent in cultured reptilian brain

cells. Brain Res 1981, 204, 451-454.

4. Williams, C. L.; Stancel, G. M. Estrogens and progestins. In Goodman and

Gilman’s The Pharmacological Basis of Therapeutics, Ninth Edition. Hardman, J.

G.; Limbird, L. E.; Molinoff, P. B.; Ruddon, R. W.; Goodman Gilman, A. Eds.

McGraw-Hill: New York, 1996; 1411-1438. 39 5. Carr, B. R. Disorders of the ovary and female reproductive tract. In Williams

Textbook of Endocrinology, Ninth edition. Wilson, J. D. Ed. W. B. Saunders:

Philadelphia, 1998; 751-779.

6. Dickson, R. B.; Lippman, L. E. Growth factors in breast cancer. Endocrine Rev.

1995, 16, 559-589.

7. Ruan, W.; Newman, C. B.; Kleinberg, D. L. Intact and aminoterminally shortened

forms of insulin-like growth factor induce mammary gland differentiation and

development. Proc. Natl. Acad. Sci. USA 1992, 89, 10872-10876.

8. Simpson. E. R.; Mahendroo, M. S.; Means, G. D.; Kilgore, M. W.; Hinshelwood,

M. M.; Graham-Lorence, S.; Amarneh, B.; Ito, Y.; Fisher, C. R.; Michael, M. D.;

Mendelson, C. R.; Bulun, S. E. Aromatase cytochrome P450, the enzyme

responsible for estrogen biosynthesis. Endocrine Rev. 1994, 15, 342-354.

9. McGuire, W. L. An update on estrogen and progesterone receptors in prognosis

for primary and advanced breast cancer, in: S. Iacobelli, R. J. B. King, H. R.

Lidner, M. E. Lippman. (Eds), Hormones and Cancer, vol. 15, Raven press, New

York, 1980, pp 337-344.

10. Early Breast Cancer Trialist’s Group. Systemic treatment of early breast cancer by

hormonal, cytotoxic, or immune therapy. Lancet 1992, 339, 1-15.

11. Jordan, V.C. Tamoxifen: toxicities and drug resistance during the treatment and

prevention of breast cancer. Ann. Rev. Pharmacol. Toxicol. 1995, 35, 195-211.

12. Brodie, A.; Lu, Q.; Long, B. Aromatase and its inhibitors. J. Steroid Biochem.

Mol. Biol. 1999, 69, 205-210.

40 13. Santen, R. J.; Harvey, H. A. Use of aromatase inhibitors in breast carcinoma.

Endocr. Relat. Cancer 1999, 6, 75-92.

14. Brueggemeier, R. W. Aromatase, aromatase inhibitors, and breast cancer. Am. J.

Ther. 2001, 8, 333-344.

15. Ortiz de Montellano, P. R.; Oxygen activation and reactivity. In Cytochrome

P450: Structure, Mechanism, and Biochemistry, Second Edition; Ortiz de

Montellano, P. R., Ed. Plenum Press: New York, 1995; 245-303.

16. Raag, R.; Poulos, T. L.; X-ray crystallographic structural studies of cytochrome

P450cam: Factors controlling substrate metabolism. Front. Biotransform. 1992, 7,

1-43.

17. Sligar, S. G. Coupling of spin, substrate, and redox equillibria in cytochrome

P450. Biochemistry, 1976, 15, 5399-5406.

18. Groves, J. T.; Watanabe, Y. J. Reactive iron porphyrin derivatives related to the

catalytic cycles of cytochrome P-450 and peroxidase. Studies of the mechanism of

oxygen activation. J. Am. Chem. Soc. 1988, 110, 8443-8452

19. Caspi, E.; Arunachalam, T.; Nelson, P. A. Biosynthesis of estrogens: the steric

mode of the initial C-19 hydroxylation of androgens by human placental

aromatase. J. Am. Chem. Soc. 1983, 105, 6987-6989.

20. Caspi, E.; Arunachalam, T.; Nelson, P. A. Biosynthesis of estrogens:

aromatization of (19R)-,(19S)-,and(19RS)-[19-3H,2H,1H]-3.beta.-hydroxyandrost-

5-en-17-ones by human placental aromatase J. Am. Chem. Soc. 1986, 108, 1847-

1852.

41 21. Miyairi, S. Fishman, J. Novel method of evaluating biological 19-hydroxylation

and aromatization of androgens. Biochem. Biophys. Res. Commun. 1983, 117,

392-398.

22. Miyairi, S. Fishman, J. Radiometric analysis of oxidative reactions in

aromatization by placental microsomes. Presence of differential isotope effects. J.

Biol. Chem. 1985, 260, 320-325

23. Numazawa, M.; Midzuhashi, K.; Nagaoka, M. Metabolic aspects of the 1 beta-

proton and the 19-methyl group of androst-4-ene-3,6,17-trione during

aromatization by placental microsomes and inactivation of aromatase. Biochem.

Pharmacol. 1994, 47, 717.

24. Osawa, Y.; Shibata, K.; Rohrer, D. Weeks, C., Duax, W. L. Reassignment of the

absolute configuration of 19-substituted 19-hydroxysteroids and stereomechanism

of estrogen biosynthesis. J. Am. Chem. Soc. 1975, 97, 4400-4402.

25. Arigoni, D.; Battaglia, R.; Akhtar, M.; Smith, T. J. Chem. Soc., Chem. Commun.

1975, 185.

26. Brodie, H. J.; Kripalani, K. J.; Possanza, G. Studies on the mechanism of estrogen

biosynthesis. VI. The stereochemistry of hydrogen elimination at C-2 during

aromatization. J. Am. Chem. Soc. 1969, 91, 1241-1242.

27. Fishman, J.; Guzik, H. Stereochemistry of estrogen biosynthesis. J. Am. Chem.

Soc. 1969, 91, 2805-2806.

28. Fishman, J.; Guzik, H.; Dixon, D. Sterochemistry of estrogen biosynthesis.

Biochemistry 1969, 8, 4304-4309.

42 29. Hosoda, H.; Fishman, J. Unusually facile aromatization of 2β-hydroxy-19-oxo-4-

androstene-3,17-dione to estrone. Implications for estrogen biosynthesis. J. Am.

Chem. Soc. 1974, 96, 7325-7329.

30. Morand, P.; Williamson, D. G.; Layne, D. S.; Lompa-Krzymien, L.; Salvador, J.

Conversion of an androgen epoxide into 17β-estradiol by human placental

microsomes. Biochemistry, 1975, 14, 635-638.

31. Mastalerz, H.; Morand, P. Acid- and base-catalyzed reactions of 4β,5β- and

4α,5α- epoxyandrostane-3,17,19-trione. J. Chem. Soc., Perkin Trans. I. 1982,

2611.

32. Akhtar, M.; Calder, M. R.; Corina, D. L.; Wright, J. N. Mechanistic studies on C-

19 demethylation in estrogen biosynthesis. Biochem. J. 1982, 201, 569-580.

33. Cole, P. A.; Robinson, C. H. Mechanism and inhibition of cytochrome P-450

aromatase. J. Med. Chem.1990, 33, 2933-2942.

34. Graham-Lorence, S.; Amarneh, B.; White, R. E.; Peterson, J. A.; Simpson, E. R.

A three-dimensional model of aromatase cytochrome P450. Prot. Sci. 1995, 4,

1065-1080.

35. Caspi, E.; Dharmaratne, H. R. W.; Shackleton, C. Biosynthesis of estrogens by

microsomal placental aromatase; isolation and metabolism of 10β-hydroxyestr-4-

ene-3,17-dione. J. Chem. Soc., Chem. Commun. 1989, 1699.

36. Chen, S.; Zhou, D. Functional domains of cytochrome P450 inferred from

comparative analysis of amino acid sequences and substantiated by site-directed

mutagenesis experiments. J. Biol. Chem. 1992, 267, 22587-22594.

43 37. Zhou, D.; Korzekwa, K. R.; Poulos, T.; Chen, S. A site-directed mutagenesis

study of human placental aromatase. J. Biol. Chem. 1992, 267, 762-768.

38. Amarneh, B.; Corbin, C. J.; Peterson, J. A.; Simpson, E. R.; Graham-Lorence, S.

Functional domains of human aromatase cytochrome P450 characterized by

molecular modeling and site-directed mutagenesis. Mol. Endocrinol. 1993, 7,

1617-1624.

39. Caspi, E.; Wicha, J.; Arunachalam, T.; Nelson, P.; Spiteller, G. Estrogen

biosynthesis. Concerning the obligatory intermediacy of 2β-hydroxy-10β-

formylandrost-4-ene-3,17-dione. J. Am. Chem. Soc. 1984, 106, 7282-7283.

40. Covey, D. F.; Carrell, H. L.; Beusen, D. D. Metabolism of 19-methyl substituted

steroids and a proposal for the third aromatase monooxygenation. Steroids 1987,

50, 363-374.

41. Tan, L.; Muto, N. Purification and reconstitution properties of human placental

aromatase. A cytochrome P-450-type monooxygenase Eur. J. Biochem. 1986,

156, 243-250.

42. Bednarski P. J.; Nelson S. D. Dissociation of 19-hydroxy- 19-oxo-, and

aromatizing-activities in human placental microsomes through the use of suicide

substrates to aromatase. J. Steroid Biochem. 1989, 32, 309-316.

43. Siegbahn, P. E. M.; Blomberg, M. R. A. Density functional theory of biologically

relevant metal centers. Annu. Rev. Phys. Chem. 1999, 50, 221-249.

44. Siegbahn, P. E. M.; Blomberg, M. R. A. Transition-metal systems in biochemistry

studied by high-accuracy quantum chemical methods. Chem. Rev. 2000, 100, 421-

437. 44 45. Himo, F.; Siegbahn, P. E. M. Quantum chemical studies of radical-containing

enzymes. Chem. Rev. 2003, 103, 2421-2456.

46. Noodleman, L.; Lovell, T.; Han, W.-G.; Li, J.; Himo, F. Quantum chemical

studies of intermediates and reaction pathways in selected enzymes and catalytic

synthetic systems. Chem. Rev. 2004, 104, 459-508.

47. Meunier, B.; de Visser, S. P.; Shaik, S. Mechanism of oxidation reactions

catalyzed by cytochrome P450 enzymes. Chem. Rev. 2004, 104, 3947-3980.

48. Banting, L.; Nicholls, P. J.; Shaw, M. A.; Smith, H. J. Recent developments in

aromatase inhibition as a potential treatment for oestrogen-dependent breast

cancer. Prog. Med. Chem. 1989, 26, 253-298.

49. Banting, L. Inhibition of aromatase. Prog. Med. Chem. 1996, 33, 147-184.

50. Covey, D. F. Aromatase inhibitors: specific inhibitors of estrogen biosynthesis. In

Sterol biosynthesis inhibitors. Pharmaceutical and agrochemical aspects. Berg,

D.; Plempel, M. Eds. Ellis Horwood Ltd., Chichester (England), 1988: 534-571.

51. O’Reilly, J. M.; Brueggemeier, R. W. Development of steroidal and nonsteroidal

inhibitors of aromatase for the treatment f hormone-dependent breast cancer.

Curr. Med. Chem. 1996, 3, 11-22.

52. Schenkman J. B.; Sligar S. G.; Cinti D. L. Substrate interaction with cytochrome

P-450. Pharmacol. Ther. 1981, 12, 43-71.

53. Santen, R. J.; Samojlik, E.; Wells, S. A. Resistance of the ovary to blockade of

aromatization with aminoglutethimide. J. Clin. Endocrinol. Metab. 1980, 51, 473-

477.

45 54. Kim, Y.-W.; Hackett, J. C.; Brueggememeier, R. W. Synthesis and aromatase

inhibitory activity of novel pyridine-containing isoflavones. J. Med. Chem. 2004,

47, 4032-4040.

55. Graves, P. E.; Salhanick, H. A. Stereoselective inhibition of aromatase by

enantiomers of aminoglutethimide. Endocrinology 1979, 105, 52-57.

56. Santen, R. J.; Worgul, T. J.; Lipton, A.; Harvey, H.; Boucher, A.; Samojlik, E.;

Wells, S. A. Aminoglutethimide as a treatment of postmenopausal women with

advanced breast carcinoma. Ann. Intern. Med. 1982, 96, 94-101.

57. Mason, J. I.; Murry, B. A.; Olcott, M.; Sheets, J. J. Imidazole antimycotics:

inhibitors of steroid aromatase. Biochem. Pharmacol. 1985, 34, 1087-1092.

58. Bossche, H. V. J. Steroid Biochem. Mol. Biol. 1992, 43, 1003.

59. Miller, W. R.; Jackson, J.; The therapeutic potential of aromatase inhibitors.

Expert Opin. Investig. Drugs 2003, 12, 1-15.

60. Plourde, P. V. Dyroff, M.; Dukes, M. Arimidex: a potent and selective fourth-

generation aromatase inhibitor. Breast Cancer Res. Treat. 1994, 30, 103.

61. Bhatnagar, A. S.; Hausler, A.; Schieweck, K.; Lang, M.; Bowman, R. Highly

selective inhibition of estrogen biosynthesis by CGS20267, a new non-steroidal

aromatase inhibitor. J. Steroid Biochem. Mol. Biol. 1990, 37, 1021-1027.

62. Kellis, J. T.; Vickery, L. E. Inhibition of human estrogen synthetase (aromatase)

by flavones. Science 1984, 225, 1032-1034

63. Kellis, J. T., Jr.; Nesnow, S.; Vickery, L. E. Inhibition of aromatase cytochrome

P-450 (estrogen sythetase) by derivatives of α-napthoflavone. Biochem.

Pharmacol. 1986, 35, 2887-2891. 46 64. Ibrahim, A.- R.; Abul-Hajj, Y. Aromatase inhibition by flavonoids. J. Steroid

Biochem. Mol. Biol. 1990, 37, 257-260.

65. Campbell, D. R.; Kurzer, M. S. Flavonoid inhibition of aromatase enzyme activity

in human preadipocytes. J. Steroid Biochem. Mol. Biol. 1993, 46, 381-388.

66. Adlercreutz, H.; Bannwart, C.; Wahala, K.; Makela, T; Brunow, G.; Hase, T.;

Arosemena, P. J.; Kellis, J. T. Jr.; Vickery, L. E. Inhibition of human aromatase

by mammalian lignans and isoflavonoid phytoestrogens. J Steroid Biochem. Mol.

Biol. 1993, 44, 147-153.

67. Wang, C.; Makela, T.; Hase, T.; Adlercreutz, H.; Kurzer, M. S. Lignans and

flavonoids inhibit aromatase in human preadipocytes. J. Steroid. Biochem. Mol.

Biol. 1994, 50, 205-212.

68. Hida, T.; Ishii, T.; Kanamaru, T.; Muroi, M. TAN-931, A novel nonsteroidal

aromatase inhibitor produced by Penicillium funiculosum No. 8974. J. Antibiot.

1991, 44, 600-612.

69. Blanco, J. G.; Gill, R. R.; Alavarez, C. I.; Patrito, L. C.; Genti-Raimondi, S.;

Flury, A. A novel activity for a group of sequiterpene lactones: inhibition of

aromatase. FEBS Lett. 1997, 409, 396-400.

70. Blanco, J. G.; Gil, R. R.; Bocco, J. L.; Meragelman, T. L.; Genti-Raimondi, S.;

Flury, A. Aromatase inhibition by an 11,13-dihydroderivative of a sesquiterpene

lactone. J. Pharmacol. Exp. Ther. 2001, 297, 1099-1105.

71. Minami, T.; Iwamoto, M.; Ohtsu, H.; Ohishi, H.; Tanaka, R.; Yoshitake, A.

Aromatase inhibitory activities of standishnal and the diterpenoids from the bark

of Thuja standishii. Planta Med. 2002, 68, 742-745. 47 72. Kao, Y.-C.; Zhou, C.; Sherman, M.; Laughton, C. A. Molecular basis for the

inhibition of human aromatase (estrogen synthetase) by flavone and isoflavone

phytoestrogens: A site-directed mutagenesis study. Environ. Health Perspect.

1998, 106, 85-92.

48

CHAPTER 2

DENSITY FUNCTIONAL THEORY INVESTIGATION OF MODEL DIOXYGEN

INTERMEDIATES IN THE P450 CATALYTIC CYCLE: STRUCTURE AND

SPECTROSCOPIC PROPERTIES

2.1 INTRODUCTION

The cytochromes P450 represent a ubiquitous class of cysteinato-heme enzymes, which play key roles in the biotransformation of many organic substances. These enzymes are able to catalyze a myriad of difficult chemical transformations including oxidation of

C-H bonds, heteroatom oxidation and dealkylation, alkene epoxidation, aldehyde deformylation, and reductive dehalogenation [15] In addition to metabolism of the many xenobiotics which confront living organisms, these catalytic abilities have earned P450

enzymes prominence in the transformation of numerous physiological molecules. [73,74]

The proposed catalytic cycle of cytochrome P450 contains multiple reactive oxygen

intermediates, each capable, in theory, of initiating substrate oxidation (Figure 2.1). The

high-valent iron oxene, or Compound I (Figure 2.1; d), has gained widespread

acceptance as the primary oxidant. The mechanisms of many oxidations can be

rationalized by the proton-coupled, electron-transfer mechanism attributed to Compound 49

I. [18] However, a number of peculiar reactions exist which are not interpretable by this

mechanism, [75,76] such as the deformylation of aldehydes, suggesting the possibility of

a second oxidant in the P450 catalytic cycle. [77-83]

1- 2- O O O O e- O +H+ O H Fe Fe Fe S S S Cys Cys Cys abc Fe(P) = Iron Protoporphyrin IX = Fe +H+ O Fe -H O S 2 Cys d

Figure 2.1: Oxygen-containing catalytic intermediates of the cytochrome P450 catalytic cycle.

Experimental characterization of dioxygen bound P450 enzymes (Figure 2.1; a) and

synthetic models of P450 reactivity have yielded considerably more information about

this intermediate in the catalytic cycle than any other oxygen-containing intermediate.

18 [87] O2 isotope labeling with resonance Raman and infrared spectroscopic

characterization have allowed assignment of the respective active oxygen frequencies in

synthetic P450 models as well as in wild-type (WT) and mutant enzymes. [85-88] In

addition, the UV/Visible absorption spectrum of these species have also been recorded

and characterized. [85,89,90] However, the stability of the dioxygen adducts of P450

protein and synthetic models makes these intermediates the only oxygen-containing

species which lend themselves to direct structural observation. Some of the first structural

50 information about the Fe coordination sphere including molecular oxygen came from

EXAFS studies of wild-type cytochrome P450cam (CYP101-WT). [91] Years later, synthetic porphyrin complexes with thiolate and oxygen ligands were synthesized and studied by X-ray crystallography. [92] Recently, a high-resolution cryogenic X-ray structure of CYP101-WT with O2 bound was obtained, and has provided significant insight into the conformational changes, which occur during catalysis and the role of an ordered network of water molecules in oxygen activation. [93]

P450 catalytic intermediates en route to the spectroscopically elusive Compound I have been studied by various means, although direct structural observation is inaccessible to date. In contrast to the diamagnetic O2-P450 and its synthetic analogs, the rate-limiting electron injection makes these states paramagnetic, limiting the repertoire of methods available for experimental characterization. Red-shifted UV/Visible spectra of CYP101-

D251N proceeding through the catalytic cycle and CYP101-WT at ~ 77 K are attributed to accumulation of the reduced dioxygen (b) and hydroperoxo (c) species, respectively.

(Figure 2.1) [92,93] Cryogenic radiolytic reduction of heme enzymes [94,95] has opened the door to EPR and 1H and 14N electron-nuclear double resonance (ENDOR) studies of reduced dioxygen and hydroperoxo CYP101–WT, D251N, and T252A proteins. [96] The results of these experiments have provided corroborating evidence for the assignments made to the UV/Visible spectroscopic results.

One avenue to study the transient dioxygen intermediates in the P450 catalytic cycle is to use quantum chemistry. A useful aspect of quantum chemical methods is that short-lived 51 species are treated with equal ease and accuracy as long-lived ones. Unfortunately, high- level ab initio quantum chemical calculations which include electron correlation are reserved for systems restricted to a few heavy atoms, and therefore can only be applied to very small problems of biochemical interest. Many of the studies performed to date on transition metals in biological systems, including models of cytochrome P450, have been accomplished using density functional theory (DFT). [44,45] DFT offers the inclusion of electron correlation with minimal computational cost and may potentially be applied to hundreds of atoms depending on the computing resources available. Given the history of

DFT performance in application to large systems of biological interest, application of this theory is a reasonable first step toward the full characterization of these intermediates, which have eluded direct structural observation.

To this end, we chose a model complex composed of a pristine iron porphyrin, cysteinamide axial ligand, and the appropriate dioxygen-containing distal ligand for geometry optimization using DFT. (Figure 2.2) Since a disproportionate amount of experimental data exists for the dioxygen bound intermediates from studies with model complexes and native enzymes, these data served as standards for comparison and validation of the theoretical approach. The intermediates of the P450 catalysis are marked due to the change in preferred multiplicity as the catalytic cycle proceeds, and it is therefore important to assess the theoretical method in its ability to reproduce the proper ground state multiplicity. To achieve this, the spin state energetics between the low and intermediate spin states of each species was calculated with both hybrid and gradient- corrected density functional theory methods. 52

1-,2- 1- Figure 2.2: Structure of [Fe(P)(O2)(SCysNH2)] (top) and [Fe(P)(O2H)(SCysNH2)]

(bottom) model for cytochrome P450.

53

The properties of the individual low-spin species are examined more closely. Herein, changes in geometric parameters, vibrational frequencies, and properties related to the distribution of unpaired spin are discussed as they depend on the choice of functional and the particular intermediate at hand. In addition, the electronic spectrum in the Soret region of wavelengths was calculated using time-dependent density functional theory for each low-spin intermediate. The results of these computations have allowed us to speculate assignments of the predominant transitions observed in the electronic spectrum of the native enzyme.

2.2 COMPUTATIONAL METHODOLOGY

The cytochrome P450 active site contains and iron protoporphyrin IX, for which the apoprotein provides a deprotonated cysteine axial ligand. In this study, we employed the

-1,-2 truncated hexacoordinate [Fe(P)(O2)(SCysNH2)] (P=Porphyrin) and distally

1- protonated analog, [Fe(P)(O2H)(SCysNH2)] , of the cytochrome P450 active site.

(Figure 2.2) The corresponding atomic coordinates were extracted from the cryogenic crystal structure of CYP101 with oxygen bound (PDB ID: 1DZ8, Chain A). [93] The protein backbone was reduced to a cysteinamide axial ligand which includes an unusual

NH-S hydrogen bond contributed by leucine 358 observed in the crystal structure and may modulate reactivity of intermediates in the catalytic cycle. [97-99] Unrestricted calculations employing the B3LYP hybrid functional [100-103] and Becke’s 1988 gradient-corrected exchange functional [104] with the PW91 [105] and P86 [106] gradient-corrected correlation functionals (BPW91 and BP86) were performed with the 54

Gaussian 98 suite of programs. [107] Calculations employing the BP86 density

functional with the resolution of identity approximation [108,109] for treatment of the

Coulomb matrix elements (RI-J) were performed with the Turbomole 5.6 suite of

programs. [110] In Gaussian 98 geometry optimizations, the TZV (17s10p6d)/ [6s3p3d] basis set [111,112] was used for iron and the 6-31G(d) basis set was used for all other atoms. In RI-J BP86 geometry optimizations, the TZVP (17s11p6d), [6s4p3d] basis set was used for iron and the SV(P) basis set ( (7s4p1d) / [3s2p1d] for N,C,O; (15s4p1d) /

[5s2p1d] for S; and (8s) / [2s] for H) was applied to the remaining atoms. [116,117] The default auxiliary basis sets [108,109] corresponding to those basis sets used for construction of the unrestricted Kohn-Sham orbitals were used for coulomb fitting in

Turbomole. Spherical harmonic basis functions (i.e. 5d, 7f, 9g) were used in all computations. Relativistic effects common to many transition metal systems have been ignored since their contribution in iron systems was shown to be insignificant. [113] The effect of the low dielectric solvent, chlorobenzene (CB, ε = 5.621) on the electronic structure of these models was evaluated with the polarizable continuum model (PCM) of

Tomasi using Gaussian. [114-116] We believe the dielectric constant of this solvent is a reasonable approximation of the environment experienced by the heme center at the protein active site. Population analysis was carried out with the natural population analysis (NPA) method. [117] The presence of a minimum on the potential energy surface was verified by harmonic vibrational frequency analysis whenever possible.

Vibrational frequency analyses of solvated structures were avoided because of the large computational expense required for numerical evaluation of the force constants, and as a result, zero-point vibrational energy corrections were not applied. 55

To study the interactions of unpaired electrons with various magnetic nuclei, N, we calculated the isotropic hyperfine coupling constants Aiso(N) using the aforementioned geometries with the 6-311+G* basis set for the non-Fe atoms and TZV for Fe. The hyperfine interaction tensor A can be separated into its isotropic and anisotropic components. Neglecting spin-orbit effects, the isotropic hyperfine splittings Aiso(N) are

α-β equal to the Fermi contact term AFC and they are related to the spin densities ρ (RN) at the corresponding nuclei by

N 4π −1 α −β = gβg β S ρ ()r Aiso 3 N N z N (1)

where β is the Bohr magneton, βN is the nuclear magneton, and g is the free-electron g- value. The g-value of the nucleus is given by gN = µN/IN (µN is the nuclear magnetic moment of nucleus N in units of βN, and IN is the total nuclear spin for that nucleus). The

α-β spin density ρ (RN) at the position of nucleus N can be expressed as

α −β ψ (r )δ (r )ψ (r ) (2) ∑Pµν µ kN kN ν kN µ,ν where α −β is the spin density matrix. [118] Pµν

Time dependent density functional theory (TD-DFT) allowed the computation of spin- conserved vertical excitation energies, oscillator strengths and excited state compositions in terms of transitions between occupied and virtual orbitals. RI-J TD-DFT calculations of RI-J BP86 optimized geometries were carried out with the ESCF module [119-121] of the Turbomole 5.6 suite of programs. Excited states with vertical excitation energies

56 corresponding to 300 nm and longer were calculated and compared to the available experimental data.

2.3 RESULTS AND DISCUSSION

2.3.1 ORDERING OF STATES.

Of fundamental importance in studies involving transition metals is the ability of the chosen theoretical method to reproduce the correct multiplicity for the species of interest.

Recent Hartee-Fock and density functional theory studies provide ample support for the idea that the sign and magnitude of the energy splittings between states is very sensitive to the choice of functional. Recent theoretical work by Scherlis and Estrin with iron ions and a truncated model of the hemoglobin site in the gas phase indicated that GGA density functionals underestimated the energy gaps, favoring the stabilization of the lowest spin states. [113] Better agreement with experiment was obtained by using hybrid density functionals. [113] In a follow-up study, GGA treatments of aquo- and nitrosylated active sites of P450 predicted the correct, low-spin doublet ground states. [122] We have calculated the energetic preferences of low (S=0, ½) and intermediate (S=1, 3/2) spin states of the three dioxygen P450 intermediates and assessed the contribution of a low dielectric solvent (CB, ε = 5.621) as a model of the environment the heme complex experiences in the active site. These data are listed in Table 2.1.

57

The GGA functionals, BPW91 and BP86, generally predict low spin (S=0) configurations for the dioxygen model complex utilized in this study. The results presented here are in agreement with previously observed trends in which the low spin state is favored with a small triplet-singlet energetic spacing. In contrast, B3LYP calculations favor an intermediate-spin triplet ground state. These results seem to be in disagreement with experimental evidence, which has established that the electronic structure of the molecular oxygen bound active site of CYP101-WT corresponds to a singlet configuration. [123,124] In all cases, but most notably with BPW91, the energetic spacing between spin states is expanded upon immersion in a dielectric constant representative of the active site.

The current discrepancy with the B3LYP result and the experimental observation may be explained with the proposal that thermal equilibrium arises between two different spin states. [125] This effect has been noted to occur in some porphyrins [126] and other complexes [127,128] containing iron. In these calculations, we have ignored thermodynamic contributions, so the B3LYP results in Table 2.1 are predictive of a triplet ground state at 0 K. Thermal equilibrium hypothesizes that as the temperature increases, the singlet state could be observed due to fluctuations of the iron atom out of the heme plane or perturbations in the geometries of the axial and distal ligands. The crossing of spin potential energy surfaces is not unknown to DFT investigations of these systems. This concept was demonstrated at the B3LYP level of theory in pentacoordinated models of hemoglobin [113] by distorting the Fe atom from the heme

58 plane and by stretching the imidazole ligands of a bis(histidine) heme model complex.

[129]

The species provisionally assigned to the reduced dioxygen and hydroperoxo CYP101 intermediates are paramagnetic and have a low-spin (S=1/2) configuration. [99] All of the density functionals tested reproduced the correct ground state multiplicity with the respective model complexes, although some variation exists in the size of the quartet- doublet energetic spacing. Gas-phase GGA treatment of the reduced dioxygen species yields an energetic preference for the doublet state approximately twice that computed by

B3LYP; however, this disparity disappears when a low polarity solvent is included. The contribution of solvent in the case of the hydroperoxo is more density functional method dependent. In BPW91 calculations, the energetic spacing between spin states changed minimally when solvent was included. However, the BP86 spin state splitting contracts greater than five-fold. We were unable to obtain the quartet-doublet spacing with the

B3LYP functional due to lack of convergence of the self-consistent field equations for the quartet state.

The electron affinities were calculated from the relative energies of the low-spin state of the dioxygen and reduced dioxygen model complexes in the gas phase and for ε=5.621, each with their respective gas-phase optimized geometries (Table 2.1). Inclusion of a low-dielectric solvent results in a dramatic shift in the electron affinities favoring reduction of dioxygen P450. This observation illustrates the influence of the surrounding protein on the redox properties of the active site heme and its ligands. 59

Gas Phase ε = 5.621

3 1 1- ( A- A) [Fe(P)(O2)(SCysNH2)] BPW91 -0.37 6.3 BP86 0.24 0.73 B3LYP -12.6 -14.4

4 2 2- ( A- A) [Fe(P)(O2)(SCysNH2)] BPW91 11.3 16.6 BP86 11.5 17.5 B3LYP 6.70 16.6

4 2 1- ( A- A) [Fe(P)(O2H)(SCysNH2)] BPW91 21.8 22.63 BP86 22.8 4.8 B3LYP NCb NCb

Electron Affinities (2A-1A)

BPW91 1.47 -2.12 BP86 1.38 -2.05 B3LYP 1.07 -1.70

Table 2.1: Triplet-Singlet (3A-1A) and (4A-2A) Quartet-Doublet Energetic Preferences

(kcal/mol) for Truncated Models of Cytochrome P450 as well as Electron Affinities (eV) of the Dioxygen Model for Different Density Functionals in the Gas Phase and in the presence of a Low Dielectric Solvent (ε = 5.621). aA negative value for the high spin-low spin splitting indicates that the high spin is favored. bNC=Not calculated due to lack of geometric convergence of the 4A stat

60

2.3.2 GEOMETRIES AND HARMONIC FREQUENCIES OF P450 MODELS

Tables 2.2-2.4 list the key geometric parameters and Tables 2.5-2.7 list the O-O stretching (νO-O), Fe-O stretching (νFe-O), and Fe-O-O bending (δFe-OO) frequencies for the model dioxygen P450 system. A significant amount of experimental data has been collected for the dioxygen P450 and synthetic models of P450 reactivity. All three density functionals produced mean Fe-N distances or porphyrin “core-size”, Fe-O, and

Fe-S distances which agree well with the available experimental data. The experimental

O-O distance is more uncertain than the former bond lengths, since a range of values is available from experiment. This bond length was determined using a synthetic model of dioxygen P450, [Fe(TPpivP)(SC6HF4)(O2)] where TPpivP = meso-tetra(α,α,α,α,o- pivalaamidophenylporphyrin). The authors identified a disordered O2 molecule in all of the complexes, identifying bond lengths of 1.14, 1.15, 1.17, 1.20, and 1.28 Å, which appeared to depend on local environment and the presence of nonbonded interactions with other parts of the complex. With this error in mind, the O-O bond distances calculated at this level of theory also seem reasonable. [92] νO-O is known for CYP101-

WT, CYP101-D251N, and synthetic models from resonance Raman and infrared spectroscopic investigations. [85-88] The gradient-corrected density functionals predict harmonic vibrational frequencies that are ~8% higher than experiment. The νO-O frequency for the dioxygen model calculated by the B3LYP functional, which incorporates 20% HF exchange, predicts a stretching frequency which is ~15% too high.

BPW91 and BP86 calculations identified two closely coupled νO-O frequencies in our

61

P450 model. Resonance Raman spectroscopy has identified multiple components of the

O2 stretching band in WT, D251N, and D251N mutant enzyme complexed with putidaredoxin (Pd). The spectral profile of the D251N/Pd complex revealed two major components within the νO-O band, while three components were identified in the isolated mutant and wild-type bands. This spectral pattern was attributed to perturbations at the enzyme active site resulting in conformational substates of the protein. [88] Each pair of calculated νO-O bands results from coupling of this stretch to underlying vibrational modes of the porphyrin and cysteinamide axial ligand. The present theoretical results support that the appearance of multicomponent bands may arise from Fermi resonance of

νO-O as easily as from conformational substates of the protein. νFe-O stretching and νFe-OO bending modes have also been identified at these levels of theory. A similar relationship between the calculated νFe-OO mode and the resonance Raman frequencies exists, in which the predictions of GGA functionals exceed the experimental value by ~9%, while the B3LYP result is ~15% excessive. When considering the νFe-OO bending mode, GGA functionals predict a lower frequency bending mode by ~ 4%, whereas the value determined with B3LYP is ~ 10 % excessive.

62

BPW91 B3LYP BP86 RI-J BP86 Experiment

Mean Fe-N 2.017 2.023 2.016 2.020 1.994,a1.993,a 2.00b Fe-O 1.806 1.786 1.803 1.812 1.837,a1.850,a1.78,b1.8c Fe-S 2.357 2.382 2.353 2.344 2.367,a2.365,a2.37,b2.3c O-O 1.292 1.277 1.295 1.279 See texta

CysNH2-S 2.315 2.364 2.303 2.252

1- Table 2.2: Key Geometric Parameters (Å) for [Fe(P)(O2)(SCysNH2)] for Different

Density Functionals in the Gas Phase. aRef. [85] (X-Ray diffraction of

b c [Fe(TPpivP)(SC6HF4)(O2)]) Ref. [85] (EXAFS of O2-CYP101) Ref. [93] (X-Ray diffraction of O2-CYP101)

63

BPW91 B3LYP BP86 RI-J BP86

Mean Fe-N 2.007 2.019 2.006 2.012 Fe-O 1.917 1.930 1.908 1.931 Fe-S 2.356 2.596 2.529 2.479 O-O 1.309 1.315 1.312 1.294

CysNH2-S 2.131 2.180 2.133 2.118

2- Table 2.3: Key Geometric Parameters (Å) for [Fe(P)(O2)(SCysNH2)] for Different

Density Functionals in the Gas Phase

64

BPW91 B3LYP BP86 RI-J BP86

Mean Fe-N 2.018 2.026 2.017 2.020 Fe-O 1.847 1.831 1.843 1.861 Fe-S 2.352 2.424 2.350 2.334 O-O 1.447 1.447 1.450 1.427

CysNH2-S 2.245 2.272 2.237 2.196

1- Table 2.4: Key Geometric Parameters (Å) for [Fe(P)(O2H)(SCysNH2)] for Different

Density Functionals in the Gas Phase.

65

BPW91 B3LYP BP86 Experiment

a b νO-O 1231 1302 1228 1139, 1140 1235 1258 (1131,1138)c1 (1128,1136)c2 (1129,1137)c3

c1 c2,c3 νFe-OO 592 615 596 540, 536

c1 c2,c3 δFe-O-O 385 445 385 401, 399 432 435

-1 1- Table 2.5. Vibrational Frequencies (cm ) for the [Fe(P)(O2)(SCysNH2)] Model of

Cytochrome P450 for Different Density Functionals in the Gas Phase. aRef. [92] (Infrared spectrum of [(O2)(2,3,5,6-Fluorophenyl)thiolato]iron(II)Tetraphenylpivaloylporphyrin, b c Ref. [87] (Resonance Raman of O2-CYP101) Ref. [88] (Resonance Raman of: 1.

CYP101-WT, 2. CYP101-D251N, 3. CYP101-D251N/Pd complex)

66

BPW91 B3LYP BP86

νO-O 1177 1224 1171 1182 1176 1183

νFe-OO 474 500 480

δFe-O-O 318 334 321

-1 2- Table 2.6. Vibrational Frequencies (cm ) for the [Fe(P)(O2)(SCysNH2)] Model of

Cytochrome P450 for Different Density Functionals in the Gas Phase.

67

BPW91 B3LYP BP86

νO-O 810 902 806

νFe-OO 536 582 541

δFe-O-O 375 375 375

-1 1- Table 2.7. Vibrational Frequencies (cm ) for the [Fe(P)(O2H)(SCysNH2)] Model of

Cytochrome P450 for Different Density Functionals in the Gas Phase.

The dioxygen model P450 undergoes a number of subtle structural changes upon addition of an extra electron. Electron addition followed by geometry optimization is intended to model the structural changes that occur when a second electron is transferred from the appropriate redox protein to the dioxygen-bound heme active site. Reduction of the dioxygen model induces a minor change in the porphyrin core size. GGA treatments of the core size predict a contraction from the dioxygen model of ~0.01 Å, although the

B3LYP computed contraction is four-fold greater. More notable differences between the oxidation states of the cytochrome P450 models are lengthening of the Fe-O, Fe-S and O-

O bonds. Fe-O lengthening of ~ 0.1 Å is observed in models studied using GGA functionals, a more substantial expansion of the Fe-O bond length of 0.14 Å is observed in the B3LYP calculations. There does not appear to be a relationship between exchange and correlation approximations made in the tested density functionals when comparing the magnitude of the Fe-S lengthening. The largest stretch of 0.214 Å is noted with the

B3LYP functional, though smaller but significant changes in this distance of 0.176 and

0.135 Å are realized with BP86 and RI-J BP86, respectively. Interestingly, calculations 68 performed on both oxidation states with the BPW91 functional reveal a Fe-S bond that is shortened by 0.001 Å upon reduction. In a pioneering study of Harris and Loew who explored cytochrome P450 model systems by DFT methods, similar contractions in the core size and lengthening of the aforementioned bond distances of similar magnitudes were observed in a model which employed a mercaptide axial ligand (-SCH3) to mimic the cysteine residue present in the native enzyme. [130] In contrast to the results presented here, the calculations of Harris and Loew, which also applied the BPW91 density functional, observed lengthening of the Fe-S bond in model calculations ignoring the H-bond interactions considered here. One facet of dioxygen P450 reduction that could not be addressed with the models of previous calculations is how changes in the oxidation states affect the hydrogen-bonding pattern around the sulfur axial ligand. The present calculations reveal that electron injection results in a decrease in the length of the hydrogen bond between the amide hydrogen and the sulfur atom. Oxidation states studied with the BPW91 and B3LYP reveal a 0.184 Å decrease in the hydrogen bond distance.

Treatment with BP86 and RI-J BP86 yield a decrease in this distance by 0.170 and 0.134

Å, respectively. Changes in the NH-S hydrogen bond status may be rationalized in terms of the molecular orbital involved in accepting the electron from the redox protein. The lowest unoccupied MO and singly occupied MO of the dioxygen and reduced dioxygen species are primarily of O2(πg)-Fe(3dyz)-S(2py) antibonding character (See Appendix A).

To rationalize the change in hydrogen bond distance upon reduction, it is likely that as the incoming electron populates the O2(πg)-Fe(3dyz)-S(2py) antibonding orbital, the protein backbone’s hydrogen bond stabilizes the sulfur 2p orbitals to accommodate the developing excess negative charge on the sulfur atom. 69

The νO-O, νFe-O, and δFe-OO vibrational frequencies were also calculated for the reduced oxidation state of the P450 model, although experimental data are not available for comparison because to our knowledge, this species has not been accessible to resonance

Raman or infrared spectroscopic techniques. Like the dioxygen precursor, two closely coupled νO-O modes were identified in models studied with BPW91 and BP86, and a single mode was identified with the B3LYP functional. Injection of an additional electron results in a red shift in three vibrational bands. BPW91 and BP86 predict shifts in the

–1 -1 range of 52-57 cm . A more pronounced 78 cm shift is associated with the B3LYP

-1 functional. The νFe-O and δFe-OO bands are red-shifted by 110 cm at all levels of theory.

The reduced dioxygen catalytic intermediate of the P450 cycle rapidly accepts protons from water molecules within the active site, resulting in the departure of a water molecule and the formation of the high-valent P450 Compound I. The initial protonation of the distal oxygen yields a hydroperoxo intermediate, which has been postulated to be an alternative oxidizing intermediate in P450 catalysis. [131-135] In addition to the dioxygen and reduced dioxygen models of P450 catalytic intermediates, the distally protonated analog was analyzed using the same repertoire of density functional methods.

Distal protonation of the model complexes results in expansion of the heme core to comparable values with the dioxygen complex. Fe-O bond contraction occurs as a result of protonation of the distal oxygen atom, compared to the reduced dioxygen precursor.

With the exception of BPW91, distal protonation also has significant effects on the length of the Fe-S bond. Contraction of this parameter after distal protonation of the reduced dioxygen model calculated with B3LYP and BP86 are similar (0.172 and 0.179 Å, 70 respectively). A smaller contraction of 0.145 Å is calculated at the RI-J BP86 level of theory.

The distal protonation event appears to be communicated to the cysteinamide ligand as changes in the NH-S hydrogen bond distance are observed at all theoretical levels examined here. The NH-S hydrogen bond length expands upon addition of the proton to the distal oxygen. This presumably occurs as a result of further stabilization of the

O2(πg)-Fe(3dyz)-S(2py) antibonding orbital by the distal proton (See Appendix A). The

B3LYP computed value for this parameter is larger than that obtained with GGA functionals. The trend for the values of the NH-S hydrogen bond lengths is the same as the reduced dioxygen model.

The harmonic vibrational frequencies for the hydroperoxo model P450 intermediate were calculated with the described theoretical approaches, and notable differences were observed. A single νO-O, νFe-O, and δFe-OO bending mode was identified in each theoretical

-1 treatment. On average, the νO-O is red shifted by 411 and 346 cm compared to those frequencies identified in the dioxygen and reduced dixoygen models, respectively. The greatest νO-O mode was calculated with the B3LYP method. The mean νFe-O frequency across these theoretical methods for the hydroperoxo species is 69 cm-1 blue shifted compared to its reduced dioxygen precursor, and 48 cm-1 red shifted compared to the dioxygen model, consistent with shortening of the Fe-O bond on protonation of the

71 reduced species. All of the density functionals tested provided the same result for the δFe-

-1 OO at 375 cm .

2.3.3 EXCITED STATES

Time-dependent density functional theory (TD-DFT) was applied to calculate the electronic absorption spectra of each model dioxygen P450 intermediate. Calculation of electronic excitation energies which correspond to absorption in the UV/Visible range of these “large” porphyrin systems require the calculation of >100 states affording significant computational expense. To this end, we employed the RI approximation for the coulomb matrix elements in Turbomole. This technique allows one to approach such a daunting computational task within a reasonable timeframe. We limited our treatment to the singlet transitions in each model. The excitations in these species are extensively multiconfigurational. The most heavily weighted excitations between occupied and virtual orbitals, their energies, and oscillator strengths, which potentially contribute to the experimentally observed Soret bands are listed in Tables 2.8-2.10. As will be soon realized, many of the important excitations arise from π→π* transitions involving the porphyrin macrocycle. Porphyrin orbitals are described using the irreducible representation labels that would be assigned to molecular orbitals arising from Fe(P) in

D4h symmetry. Contour surfaces of the orbitals dominating these excitations and a more detailed description of the nature of important excitations are provided in Appendix A.

72

Expt./Soret Nature of Excitationa λ[nm] Oscillator b Band (∆E[eV]) Strength

417/B’ (3eg→LUMO) 395(3.14) 0.1270 25.6(120→134) +(O2(πg)- +19.9(128→137) Fe(3dyz)- S(2py)→2b1u) +17.8(127→136) +(1a1u→5eg)

23.4(127→135) (1a1u→5eg) 378(3.28) 0.2701

+17.4(123→135) +(2b2u→5eg)

+15.9(129→137) +(5a2u→2b1u)

352/B 24.6(121→135) (4a2u+π*→5eg) 347(3.58) 0.0890

+18.1(122→135) +(4a2u+π*→5eg)

+11.4(123→136) +(5a2u→2b1u)

Table 2.8: Prominent RI-J TD-DFT computed UV/Vis Transitions in Model P450

Intermediates. aUsing the velocity representation from Turbomole. bCoefficients of each transition represent the percent contribution to the excitation .

73

Expt./Soret Nature of Excitationa λ[nm] Oscillator b Band (∆E[eV]) Strength

B’ 9.9α(127→135) (1a1u→5eg) 400(3.10) 0.2507

+9.9β(127→135) +(1a1u→5eg)

B 15.9β(121→135) (3eg→5eg) 338(3.67) 0.0639

+13.5α(122→135) +(3eg→5eg)

+13.0α(121→135) +(3eg→5eg)

+12.4α(134→142) +(SOMO→3b2u)

+9.9β(133→142) +(HOMO→3b2u)

Table 2.9: Prominent RI-J TD-DFT computed UV/Vis Transitions in Model P450

Intermediates. aUsing the velocity representation from Turbomole. bCoefficients of each transition represent the percent contribution to the excitation

74

Expt./Soret Nature of Excitationa λ[nm] Oscillator b Band (∆E[eV]) Strength

441/B’ (π*→5eg) 397(3.12) 0.1058 32.9α(124→135) (π*→2b1u) +11.1β(128→137)

13.3α(123→135) (2b2u→5eg) 381(3.26) 0.1026

347/B 27.3α(120→135) (3eg→5eg) 348(3.56) 0.0443

Table 2.10: Prominent RI-J TD-DFT computed UV/Vis Transitions in Model P450

Intermediates. aUsing the velocity representation from Turbomole. bCoefficients of each transition represent the percent contribution to the excitation

1- The computed spectrum of [Fe(P)(O2)(SCysNH2)] contains two clusters which we have assigned to the B and B’ bands of the Soret. Theory is predictive of an excitation at 347 nm, which is in close agreement with the experimental B band absorption of dioxygen

CYP101 at 352 nm. The predominant contributors to this excitation are 121→135

(4a2u+amide π*→5eg), 122→135 (4a2u+amide π*→ 5eg), and 123→136 (2b2u→ 5eg) transitions. In the spectrum of dioxygen CYP101, the B’ component of the split Soret occurs 65 nm red-shifted from the position of the B component at 352 nm. We have identified two excitations in our model system with significant oscillator strength at 378 and 395 nm whose excitations could contribute to the low-energy component of the observed split Soret. The main contributors to these excitations are transitions between orbitals that are predominately localized on the π system of the porphyrin.

75

During enzymatic turnover, CYP101-D251N displays a significantly red-shifted split

Soret UV/Visible spectrum compared to other catalytic intermediates of CYP101-WT.

[89] The formation of this new spectrally characterized intermediate is required for catalytic turnover of the enzyme and a direct correlation exists between the decay of this intermediate and the appearance of the product, 5-hydroxycamphor. Based on evidence from EPR and ENDOR spectroscopic analysis of intermediates obtained with thermal annealing at cryogenic temperatures, the predominant CYP101-D251N species is the reduced dioxygen intermediate. It is hypothesized that neutralization of the aspartate residue by site-directed mutagenesis forestalls proton transfer to the distal oxygen atom permitting its observation. The calculated excitations are again predictive of the experimental split Soret. The excitation with the largest oscillator strength, which we have considered to correspond to the B component is computed to occur at 338 nm. The

B’ component with significant oscillator strength occurs at 400 nm. The absolute positions of the Soret bands of CYP101-D251N were not reported; only that the position of the bands were red-shifted compared to the unreduced species. Our calculations support this assignment. The 127→135(1a1u→5eg) transition identified in the unreduced model at 378 nm is red-shifted 22 nm in the reduced species, consistent with experiment.

Previous semi-empirical computations utilizing the INDO/S/ROHF/CI procedure predicted a 30 nm red-shift, although in those studies, the major shifted excitation was

predominately of 2a2u→4eg character. [130]

Upon distal protonation resulting in the formation of the hydroperoxo catalytic intermediate, which can be observed in CYP101-WT at 77 K, the positions of the Soret 76 components appear at 371 and 441 nm. We have computed a pair of bands with marked oscillator strength at 348 and 397 nm. In experiments using dioxygen-ligated CYP101-

WT , the positions of the B and B’ bands shift 19 and 24 nm after reduction and distal protonation. In our theoretical results, the position of the B band in the computed spectrum of the hydroperoxo model is essentially the same as that observed in the dioxygen model spectrum (347 vs. 348 nm). On the other hand, if we consider the positions of the B’ bands in these two models, this band is red-shifted 19 nm compared to the 378 nm excitation of the dioxygen model. The predominant contributors to the theoretical B’ band are interesting, as the origin of the occupied orbitals are localized to the cysteinamide ligand and are of predominately antibonding character. The resulting singly occupied orbitals are localized on the porphyrin π system. This observation of excitation from occupied orbitals localized to the truncated “protein” of our model system may partially explain shortcomings to accurately calculate the absolute positions and relative shifts in the split Soret of these catalytic intermediates. In fact, the protein may offer more than a well-controlled polarized environment for the heme chromophore, but may also be a source of electrons in the excited states.

2.3.4 SPIN DISTRIBUTION AND HYPERFINE COUPLING CONSTANTS

An important question regarding the cytochrome P450 catalytic cycle is how the density of the electron is delocalized in the active site after reduction of the dioxygen intermediate by the appropriate redox protein. Delocalization of charge and spin density is necessary, otherwise the hydrophobic environment inside the apoprotein could not 77 sustain the presence of a densely charged, highly energetic species. Controversy over experimental conclusions has appeared regarding the delocalization of spin in heme systems due to an apparent disparity between ESR and NMR data. [136-138] ESR spectral interpretation is indicative of a single electron confined to the iron atom, while

NMR indicates additional unpaired electron density on the porphyrin macrocycle. The extent of the delocalization for many heme systems is not accurately known, although evidence suggests that this phenomenon depends heavily on the electronic state and the nature of the ligands. This apparent contradiction was recently explained by DFT calculations performed on a heme a model as well as on bis-imidazole-ligated iron porphyrin without substituents. [129,139,140] The calculations using the B3LYP method, demonstrated that the integrated spin density at the iron atom was nearly one, in agreement with the ESR measurements. However, significant areas with opposite β spin

(i.e. spin polarization) are found along the Fe-N bond axes, thus evoking a need for additional α-spin density to be present in the porphyrin ring, ring substituents, and the axial ligands to keep the net amount of unpaired spin to be exactly one. [139,140] To address this question in the low-spin reduced dioxygen and hydroperoxo P450 intermediates, we examined the spin distribution properties of our corresponding models with the previously described levels of theory, but without the use of the RI-J approximation.

The isotropic hyperfine coupling constants (hfccs) at nucleus N are related to the spin density at N by expression (1). The hfccs of the distal proton and the heme nitrogens were determined experimentally for species provisionally assigned as hydroperoxo- and 78 reduced dioxygen CYP101 and mutants using 1H- and 14N-ENDOR spectroscopy. [96]

We have calculated the isotropic hfccs as a guide to the level of theory, which may accurately describe the electron spin distribution in these models. The quality of the calculated hfcc values depends greatly on the choice of basis set, sometimes requiring the use of large, flexible, uncontracted basis sets making accurate calculations of somewhat small molecules accessible. [141] Toward this goal, we employed the valence triple-ξ 6-

311+G* basis set for all non-Fe atoms and TZV for Fe employing the previously optimized geometries.

However, consideration of the environment around the iron porphyrin may be important.

A convenient approach toward treatment of solvation of the heme prosthetic group and surrounding ligands buried inside the cytochrome P450 apoprotein is to use the polarizable continuum model of Tomasi and coworkers. We have chosen chlorobenzene,

ε=5.621 to approximate the dielectric continuum where our model system resides in the native protein. This is not a gross approximation of the appropriate dielectric continuum, as results of molecular dynamics of cytochrome c in spherical water droplets has estimated the value to be around ε = 4. [142] One must be aware of the limitations of implicit solvent continuum models within the context of our model system. First, unlike the implicit continuum of solvent used here, the environment of the protein is non- uniform. Also, the unpaired electron is confined to the cavity constructed by the solvation model, and thus the potential for the unpaired spin to accumulate on isolated solvent molecules is ignored. One example is how hydrogen bond interactions may affect spin density. [143] Table 2.11 lists the computed expectation values for the total spin operator, 79 heme nitrogen hfccs, and distal hydrogen hfccs in the gas phase and in the presence of the

PCM salvation model. In none of the cases does the computed value deviate significantly from the expected a 0.75 nominal value for a doublet. Hence, we are not concerned with the intrusion of higher-lying electronic states.

80

BPW91 BP86 B3LYP Experiment

2- [Fe(P)(O2)(S-Cys-NH2)] 0.7604(0.7742) 0.7591(0.7714) 0.7667(0.7677)

b Aiso(N) 1.05 (0.15) 1.11 (0.28) 1.36 (1.36) 5.8-5.9 -0.54 (-1.54) -0.58 (-1.49) -0.46 (-0.51) -0.61 (-1.46) -0.64 (-1.43) -0.51 (-0.55) -1.18 (0.13) 1.20 (0.22) 1.53 (1.51)

1- [Fe(P)(O2H)(S-Cys-NH2)] 0.7741(0.7694) 0.7722 (0.7679) 0.7719 (0.7717)

b Aiso(N) -3.95 (-4.23) -3.54 (-3.83) -4.86 (-5.22) 5.8-5.9 -3.54 (-3.88) -3.06 (-3.47) -5.74 (-6.01) -3.44 (-3.66) -3.21 (-3.40) -5.63 (-5.93) -3.68 (-4.01) -3.28 (-3.66) -5.97 (-6.27)

c b Aiso(H) -2.95 (-2.09) -2.36 (-1.60) 8.11 (6.66) 8.2, 11.2, 10.9,b 11.5b

Table 2.11:. DFT/(TZV,6-311+G*)//DFT (TZV/6-31G*) Isotropic Hyperfine Coupling

Constants of Heme Nitrogen Atoms for all Models and the Distal Proton of

1- a [Fe(P)(O2H)(SCysNH2)] in the Gas Phase and with ε =5.621. The values for ε

=5.621are listed in parentheses. bRef. [96] (sign of hfcc not reported), cRef. [148,149].

81

BPW91 BP86 B3LYP

2- [Fe(P)(O2)(SCysNH2)] Porphine 0.075 (-0.084) 0.078 (-0.071) -0.049 (-0.048) Fe 0.021 (0.274) 0.033 (0.271) 0.062 (0.104) Proximal O 0.426 (0.406) 0.419 (0.402) 0.451 (0.459) Distal O 0.492 (0.417) 0.483 (0.410) 0.541 (0.489) Sulfur -0.014 (-0.013) -0.013 (-0.012) -0.004 (-0.004)

NH2 (Cysteinamide) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000)

1- [Fe(P)(O2H)(SCysNH2)] Porphine -0.100 (-0.082) -0.092 (-0.076) -0.058 (-0.052) Fe 0.690 (0.738) 0.720 (0.729) 0.933 (0.958) Proximal O 0.278 (0.239) 0.277 (0.239) 0.120 (0.098) Distal O 0.070 (0.052) 0.069 (0.052) -0.002 (-0.003) Sulfur 0.059 (0.051) 0.061 (0.052) 0.003 (-0.003)

NH2 (Cysteinamide) 0.003 (0.003) 0.003 (0.003) 0.002 (0.002)

Table 2.12: DFT/(TZV,6-311+G*)//DFT (TZV/6-31G*) Natural Atomic and Group Spin

2- 1- Densities of the [Fe(P)(O2)(SCysNH2)] and [Fe(P)(O2H)(SCysNH2)] Models of

Reduced Dioxygen and Hydroperoxo P450 in the Gas Phase and ε=5.621. The values for

ε =5.621 are listed in parentheses.

82

[Fe(P)(O2)(SCysNH2) [Fe(P)(O2)(SCysNH2) [Fe(P)(O2)(SCysNH2 2- 2- 2- + ] (H2O) ] 2(H2O) ) (H3O /H2O ) 0.7682 (0.7502) 0.7710 (0.7736) 0.8622

Aiso(N) 1.64 (1.65) 1.30 (1.19) -1.75 -0.57 (-0.64) -0.52 (-0.68) -2.42 -0.36 (-0.42) -0.46 (-0.52) -2.18 1.07 (1.06) 0.73 (0.67) -0.92

Spin Densities

Porphine -0.055 (-0.053) -0.057 (-0.056) -0.117 Fe 0.100 (0.129) 0.114 (0.154) 0.481 Proximal O 0.465 (0.479) 0.450 (0.458) 0.397 Distal O 0.500 (0.453) 0.495 (0.447) 0.310 Sulfur -0.005 (-0.005) -0.005 (-0.005) -0.016

NH2 0.000 (0.000) 0.000 (0.000) 0.001 (Cysteinamid e)

Table 2.13:. B3LYP/(TZV,6-311+G*)//B3LYP (TZV/6-31G*) Isotropic Hyperfine Coupling

Constants of Heme Nitrogen Atoms and Natural Atomic and Group Spin Densities of the

2- Hydrated [Fe(P)(O2)(SCysNH2)] and in the Gas Phase and ε=5.621. The values for ε

=5.621are listed in parentheses

83

2- Figure 2.3: Contour plots of the spin density distribution for the [Fe(P)(O2)(SCysNH2)]

1- (top) and [Fe(P)(O2H)(SCysNH2)] (bottom) models as obtained with B3LYP in the gas phase (left) and in the presence of the low dielectric solvent continuum (ε = 5.621)

(right). Blue areas represent excess α-density, and green areas excess β density. An isocontour value of 0.001 e/B3 has been used.

84

14N-ENDOR measurement of the hyperfine coupling to a 14N-pyrrole ligand provides a good index of the spin density on the heme iron atom. Studies of this type on species provisionally assigned to the reduced dioxygen and hydroperoxo intermediates of

CYP101 gave rise to the same mean hfcc of 5.8-5.9 MHz. [96] From this result, the authors postulated that protonation of the distal oxygen atom had little or no impact on the spin density at Fe. At all DFT levels, the computed 14N hfccs for the reduced dioxygen model are significantly low compared to the experimentally determined values, corresponding to a low assignment of net spin density on Fe. The calculated hfcc result is corroborated by the small spin densities localized on Fe using the NPA partitioning scheme (Table 2.12). In the reduced dioxygen model, inspection of the NPA values and the spin density difference density contour plot (Figure 2.3) consistently indicates that the

α unpaired electron is primarily confined to the dioxygen axial ligand. A small amount of

β spin accumulates on the porphyrin macrocycle as a result of spin polarization. Despite its potential role in stabilizing the singly occupied molecular orbital, no appreciable spin density accumulates on the cysteinamide axial ligand for the reduced dioxygen model.

In contrast to the reduced dioxygen model, the isotropic hfccs (Table 2.11) calculated for the hydroperoxo analog are in better agreement with available experimental data.

Hyperfine couplings calculated at the GGA level are consistently low in this case.

However, the B3LYP results are in excellent agreement with experiment. The mean hfcc for the heme nitrogens computed as 5.6 and 5.8 MHz in the gas phase and solvent, respectively. Inspection of the NPA partitioning of the B3LYP total density reveals the presence of almost a full electron localized on the heme Fe (Table 2.12). GGA treatments 85 of this model concentrate less spin density on Fe, instead shifting it to the axial and distal ligands. Excess α spin centered on Fe and the proximal oxygen at the B3LYP level of theory results in polarization of the heme orbitals, allowing the appearance of β spin on the porphyrin macrocycle. Spin density contour plots show this accumulation of β spin is essentially fully localized on the heme nitrogens and along the Fe-N bond axis. (Figure

2.3)

Within our model system, why does B3LYP perform so well for the hydroperoxo model and not for the reduced dioxygen model? In addition to limitations, which arise as a result of using an approximate exchange-correlation potential and finite Gaussian-type basis set, a number of inadequacies arise as a consequence of the choice of truncated model. In

1 the H-ENDOR studies of reduced O2-CYP101-D251N, an exchangeable proton with a hfcc of 14.4 MHz was identified and the possibility that this proton participates in a hydrogen-bond with the dioxygen ligand was suggested. [96] To this point, we have not considered the impact that hydrogen bond donors to the dioxygen ligand may have on spin density of the reduced dioxygen model complex. It is clear from this study, that in the absence of such a hydrogen-bond donor, almost an entire electron is localized on the dioxygen ligand. It is conceivable that a hydrogen bond from an acidic proton may distort the spin density from that of an isolated dioxygen ligand such that it more closely resembles the spin density pattern calculated for the hydroperoxo analog.

86

2- Figure 2.4. Structures of hydrated [Fe(P)(O2)(SCysNH2)] . Water-oxygen ligand hydrogen bonds, the O-O, Fe-O, Fe-S, Mean Fe-N and NH-S hydrogen bond lengths (in

Å) are highlighted.

87

In an effort to explore the consequences of introducing hydrogen-bond donors to the distal face of the heme on the spin density distribution and hence the hfccs, three additional systems were studied. The initial geometries of these species were generated by introducing one and two molecules of water in the vicinity of the distal oxygen atom and both distal and proximal oxygen atoms of the optimized reduced dioxygen model. In light of the superior performance provided by the B3LYP functional to calculate the hfccs in the hydroperoxo model, these systems were fully optimized at the B3LYP/(6-

31G*,TZV) level of theory and properties were subsequently computed with a single- point computation for nonmetal atoms with the triple-ξ basis set as previously described.

The geometries of these species are shown in Figure 2.4 with critical bond distances labeled. Introduction of the water molecules has a neglibible effect on the overall geometry of the reduced dioxygen complex. Interestingly, the presence of neutral water molecules as hydrogen-bond donors also has little impact on the spin densities or the 14N hfccs compared to the unhydrated clusters (Table 2.13).

Considering the proton source model [144] of Sligar et al. that involves T252, D251, and two solvent water molecules in the active site of CYP101 where the ultimate proton donor to the distal oxygen atom is the protonated carboxylic acid of D251, a model was constructed to mimic the presence of an acidic proton in hydrogen-bonding proximity to the distal oxygen. Starting from the optimized geometry of the doubly-hydrated model, an additional proton was added to the distal water molecule and subject to a single-point computation. The presence of a naked hydronium ion in the gas phase adjacent to the distal proton models an extreme excess of acidic protons in the active site. If indeed an 88 acidic proton, such as the one from D251, distorts the spin density of the reduced dioxygen intermediate to a distribution like that of the hydroperoxo species, such an effect should occur in this exaggerated model system.

Introduction of the naked hydronium does induce a minor shift in the spin distribution of the reduced dioxygen model. The most notable changes in spin density occur on the dioxygen ligand and on the Fe atom. A decrease in spin density of approximately 0.2 e and an increase of 0.3 e occur on the dioxygen ligand and Fe, respectively. Although the presence of a strong hydrogen-bond donor shifts spin density toward the Fe atom, the net unpaired spin of 0.481 e is still significantly less than one. As follows, the computed 14N hfccs are also much lower than the experimental value determined for the radiolytically reduced oxy D251N-CYP101, for which the reduced dioxygen intermediate is thought to predominate. The lack of distortion of spin density by the excess acidic proton model raises questions about the identity of the catalytic intermediate present in the active site of

D251N-CYP101 at ~77 K. Recent theoretical studies of proton transfer in models of

CYP101-WT indicate that proton transfer to the distal oxygen is concerted with electron injection and occurs without thermodynamic barrier. [145] D251N-CYP101 is still a catalytically competent enzyme, indicating proton transfer is still active, although less efficient. [85] One possibility is that the ultimate proton donor is the terminal member of a hydrogen-bonded network of water. This mechanism of proton transfer is proposed to occur in P450eryF, with a barrier as low as 3.8 kcal/mol. [146,147] It is possible the species observed in radiolytically reduced oxy D251N-CYP101 by ENDOR spectroscopy

89 is not the reduced dioxygen intermediate, but rather the hydroperoxo intermediate interacting with an alternative hydrogen-bonding network.

We also calculated the isotropic hfcc for the distal proton of the hydroperoxo- analog.

The gas phase B3LYP result of 8.1 MHz is in excellent agreement with the experimentally determined value for the exchangeable proton in hydroperoxo ferric myoglobin, [148] hemoglobin, [149] and heme oxygenase. [149] Inclusion of the dielectric field decreases the calculated result by ~ 20 % to a still appreciable 6.7 MHz.

Unfortunately these values are lower than the isotropic hfcc for this proton in hydroperoxo CYP101-WT and mutants which range from 10.9-11.2 MHz. However, these values are in reasonable agreement considering the lack of hydrogen polarization functions in the basis set.

2.4 CONCLUSIONS

In this study, we have illustrated the effective use of density functional theory as tool to model the properties of dioxygen P450 catalytic intermediates. The results provided here are complementary to the efforts of molecular spectroscopy to identify and characterize transient intermediates in the enzymatic cycle, which elude direct structural observation.

In general, GGA functional treatments predict accurate ground state multiplicities for each of the models studied. The B3LYP result for the ground state multiplicity is in disagreement with the experimental singlet configuration; however, spin multiplicities of 90 these systems can be extremely sensitive to thermal equilibration effects and geometric perturbations which could result from constraints imparted on the system by the protein.

Electron affinity calculations from low-spin states predict a dramatic shift in the electron affinity upon immersion in a low dielectric solvent. This observation illustrates the influence of the hydrophobic core of the enzyme on the redox properties of the P450 prosthetic group.

The DFT treatments used here produce quality geometries for dioxygen model system compared with available structural data from X-Ray diffraction of CYP101 as well as synthetic analogs of the P450 site. The model we describe here is extended from those of other theoretical studies to include the unusual NH-S hydrogen bond. Different oxidation states of the model system show significant differences in the value of this parameter, indicating it may play a role in facilitating the reduction of O2-bound P450. The next logical step to be taken in our laboratory to understand the role of the protein environment on this facet of dioxygen activation would be to expand the model system and embed it as part of a hybrid QM/MM scheme. [150,151] Considering the lack of scaling factors for the level of theory, the calculated frequencies are in reasonable agreement with infrared and resonance Raman studies. Identification of multiple νO-O modes, some of which were coupled to underlying vibrations of the porphyrin macrocycle and cysteinamide ligand affirms the potential for Fermi resonance in vibrational studies of this enzyme.

91

RI-J TD-DFT can approximately reproduce the Soret shifts, which occur in the

UV/Visible spectrum of after dioxygen activation. In the case of the model systems described in this study, both model reduced dioxygen and hydroperoxo catalytic intermediates experience shifts of approximately 20 nm in the B’ band of the split Soret.

These results are in reasonable agreement with experimental observations of CYP101-

D251N and CYP101-WT enzymes for which spectra arising from each of these intermediates may have been observed. Interestingly, these results indicate the reduced dioxygen and hydroperoxo catalytic intermediates may have similar spectra when they are absolutely characterized. The role of UV/Visible spectroscopy to distinguish between these two species is still questionable. Notably, some of the significant, singlet electronic transitions in the hydroperoxo model arose from orbitals localized to “protein” backbone in the present model. This observation indicates that more extensive models of the P450 active site may be required to identify a distinguishing feature.

Computation of isotropic hyperfine coupling constants of the model catalytic intermediates described herein unveiled a competent level of theory to reproduce experimentally derived parameters as well as establish how theory can distinguish between different members of the catalytic cycle. The B3LYP functional proved to be very accurate in reproducing the 14N isotropic hyperfine coupling constants of the hydroperoxo species. Interestingly, the computed values were very low for the reduced dioxygen intermediate with this level of theory, which has been assigned the same 14N isotropic hyperfine coupling constants by experiment. It has been postulated that hydrogen-bond donors in the active site of D251N-CYP101 distort the spin density from 92 the dioxygen ligand such that the net spin density on Fe integrates to nearly one, giving rise to the same hyperfine-coupling constant. To test this hypothesis, we introduced neutral water molecules as well as a hydronium ion/neutral water pair to the distal face of the heme to model such an environment. Despite including hydrogen-bond donors as intense as the gas-phase hydronium ion, a significant amount of spin density failed to accumulate on the iron atom to reproduce the experimental hyperfine coupling constant.

These results raise serious questions about the actual identity of the species observed after radiolytic reduction of oxygenated D251N-CYP101 at 77 K, and it is possible that the observed species is not the reduced dioxygen intermediate. In addition to the results presented here, theoretical studies have indicated that even in the absence of an acidic terminal proton donor, the thermodynamic barrier for proton transfer is quite small. We can speculate that the catalytic intermediate observed in the presence of the D251N mutation is also the hydroperoxo species experiencing an alternative hydrogen-bond milieu, rather than the reduced dioxygen catalytic intermediate.

2.5 REFERENCES

73. Kagawa, N; Waterman, M. R. Regulation of Steroidogenic P450s. In

Cytochrome P450: Structure, Mechanism, and Biochemistry, Second Edition;

Ortiz de Montellano, P. R., Eds. Plenum Press: New York, 1995; 419-442.

74. Capdevila, J. H.; Zeldin, D.; Makita, K.; Karara, A.; Falck, J. R. Cytochrome

P450 and the Metabolism of Arachidonic Acid and Oxygenated Eicosanoids.

In Cytochrome P450: Structure, Mechanism, and Biochemistry, Second 93

Edition; Ortiz de Montellano, P. R., Eds. Plenum Press: New York, 1995;

443-471.

75. Groves, J. T.; Watanabe, Y. J. Reactive iron porphyrin derivatives related to

the catalytic cycles of cytochrome P-450 and peroxidase. Studies of the

mechanism of oxygen activation. J. Am. Chem. Soc. 1988, 110, 8443.

76. Guengerich, F. P. Common and Uncommon Cytochrome P450 Reactions

Related to Metabolism and Chemical Toxicity. Chem. Res. Toxicol. 2001, 14,

611-650.

77. Vaz, A. D. N.; Kessell, K. J.; Coon, M. J. Aromatization of a bicyclic steroid

analog, 3-oxodecalin-4-ene-10-carboxaldehyde, by liver microsomal

cytochrome P450 2B4. Biochemistry 1994, 33, 13651-13661.

78. Raner, G. M.; Chiang, E. W.; Vaz, A. D. N.; Coon, M. J. Mechanism-Based

Inactivation of Cytochrome P450 2B4 by Aldehydes: Relationship to

Aldehyde Deformylation via a Peroxyhemiacetal Intermediate. Biochemistry

1997, 36, 4895-4902.

79. Roberts, E. S.; Vaz, A. D. N.; Coon. M. J. Catalysis by cytochrome P-450 of

an oxidative reaction in xenobiotic aldehyde metabolism: deformylation with

olefin formation. Proc. Natl. Acad. Sci. USA 1991, 88, 8963-8966.

80. Vaz, A. D. N.; Pernecky, S. J.; Raner, G. M.; Coon, M. J. Peroxo-iron and

oxenoid-iron species as alternative oxygenating agents in cytochrome P450-

catalyzed reactions: switching by threonine-302 to alanine mutagenesis of

cytochrome P450 2B4. Proc. Natl. Acad. Sci. USA 1996, 93, 4644-4648.

94

81. Peng, H. M.; Raner, G. M.; Vaz, A. D. N.; Coon, M. J. Oxidative cleavage of

esters and amides to carbonyl products by cytochrome P450 Arch. Biochem.

Biophys. 1995, 318, 333-339.

82. Raner G. M., Hatchell J. A.; Dixon M. U.; Joy T. L.; Haddy A. E.; Johnston E.

R. Regioselective peroxo-dependent heme alkylation in P450(BM3)-F87G by

aromatic aldehydes: effects of alkylation on cataysis. Biochemistry 2002, 41,

9601-9610.

83. Kuo C. L.; Raner G. M.; Vaz A. D.; Coon M. J. Discrete species of activated

oxygen yield different cytochrome P450 heme adducts from aldehydes.

Biochemistry 1999, 38, 10511-10518.

84. Makris, T. M.; Davydov, R.; Denisov, I. G.; Hoffman, B. F.; Sligar, S. G.

Mechanistic enzymology of oxygen activation by the cytochromes P450.

Drug Metab. Rev. 2002, 34, 691-708.

85. Schappacher, M.; Ricard, L.; Weiss, R. Models for the coordination site of

iron in cytochrome P-450. Synthesis and spectroscopic properties of a

dioxygen adduct of (2,3,5,6-tetrafluorophenylthiolato)iron(II)

tetraphenylpivalcylporphyrin J. Am. Chem. Soc. 1981, 103, 7646-7648.

86. Dawson, J. H.; Sono, M. Cytochrome P-450 and chloroperoxidase: thiolate-

ligated heme enzymes. Spectroscopic determination of their active-site

structures and mechanistic implications of thiolate ligation. Chem. Rev. 1987,

87, 1255-1276.

95

87. Bangcharoenpaurpong, O.; Rizos, A. K.; Champion, P. M. Resonance Raman

detection of bound dioxygen in cytochrome P-450cam. J. Biol. Chem. 1986,

261, 8089-8092.

88. Sjodin, T.; Christian, J. F.; Macdonald, I. D. G.; Davydov, R.; Unno, M.;

Sligar, S. G.; Hoffman, B. M.; Champion, P. M. Resonance Raman and EPR

Investigations of the D251N Oxycytochrome P450cam/Putidaredoxin

Complex. Biochemistry 2001, 40, 6852-6859.

89. Benson, D. E.; Suslick, K. S.; Sligar, S. G. Reduced Oxy Intermediate

Observed in D251N Cytochrome P450cam. Biochemistry 1997, 36, 5104-

5107.

90. Denisov, I. G.; Makris, T. M.; Sligar, S. G. Cryotrapped reaction

intermediates of cytochrome p450 studied by radiolytic reduction with

phosphorus-32. J. Biol. Chem. 2001, 276, 11648-11652.

91. Dawson, J. H.; Kau, L. -S.; Penner-Hahn, J. E.; Sono, M.; Eble, K. S.; Bruce,

G. S.; Hager, L. P.; Hodgson, K.O. Oxygenated cytochrome P-450-CAM and

chloroperoxidase: direct evidence for sulfur donor ligation trans to dioxygen

and structural characterization using EXAFS spectroscopy J. Am. Chem. Soc.

1986, 108, 8114-8116.

92. Schappacher, M.; Ricard, L.; Fischer, J.; Weiss, R.; Bill, E.; Montiel-

Montoya, R.; Winkler, H.; Trautwein, A. X. Synthesis, structure and

spectroscopic properties of two models for the active site of the oxygenated

state of cytochrome P450. Eur. J. Biochem. 1987, 168, 419-429

96

93. Schlicting, I.; Berendzen, J.; Chu, K.; Stock, A. M.; Maves, S. A.; Benson, D.

E.; Sweet, R. M.; Ringe, D.; Petsko, G. A.; Sligar, S. G. The catalytic pathway

of cytochrome p450cam at atomic resolution. Science 2000, 287, 1615-1622.

94. Denisov, I. G.; Makris, T. M.; Sligar, S. G. Formation and decay of

hydroperoxo-ferric heme complex in horseradish peroxidase studied by

cryoradiolysis. J. Biol. Chem. 2002, 277, 42706-42710.

95. Denisov, I. G.; Ikeda-Saito, M.; Yoshida, T.; Sligar, S. G. Cryogenic

absorption spectra of hydroperoxo-ferric heme oxygenase, the active

intermediate of enzymatic heme oxygenation. FEBS Lett. 2002, 532, 203-206.

96. Davydov, R.; Makris, T. M.; Kofman, V.; Werst, D. E.; Sligar, S. G.;

Hoffman, B. M. Hydroxylation of Camphor by Reduced Oxy-Cytochrome

P450cam: Mechanistic Implications of EPR and ENDOR Studies of Catalytic

Intermediates in Native and Mutant Enzymes. J. Am. Chem. Soc. 2001, 123,

1403-1415.

97. Poulos, T. L.; Finzel, B. C.; Howard, A. J. Crystal structure of substrate-free

Pseudomonas putida cytochrome P-450. Biochemistry 1986, 25, 5314-5322.

98. Fisher, M. T.; Sligar, S. G. Control of heme protein redox potential and

reduction rate: linear free energy relation between potential and ferric spin

state equilibrium. J. Am. Chem. Soc. 1985, 107, 5018-5019.

99. Ueyama, N.; Nishikawa, N.; Yamada, Y.; Okamura, T.; Nakamura, A.

Cytochrome P-450 Model (Porphinato)(thiolato)iron(III) Complexes with

Single and Double NH···S Hydrogen Bonds at the Thiolate Site J. Am. Chem.

Soc. 1996, 118, 12826-12827. 97

100. Stevens, P. J.; Devlin, F. J.; Chablowski, C. F.; Frisch, M. J. Ab Initio

Calculation of Vibrational Absorption and Circular Dichroism Spectra Using

Density Functional Force Fields. J. Phys. Chem. 1994, 98, 11623-11627.

101. Becke, A. D. Density-functional thermochemistry. III. The role of exact

exchange. J. Chem. Phys. 1993, 98, 5648-5652

102. Becke, A. D. Density-functional thermochemistry. I. The effect of the

exchange-only gradient correction. J. Chem. Phys., 1992, 96, 2155-2160.

103. Becke, A. D. Density-functional thermochemistry. II. The effect of the

Perdew–Wang generalized-gradient correlation correction J. Chem. Phys.

1992, 97, 9173-9177.

104. Becke, A. D. Density-functional exchange-energy approximation with correct

asymptotic behavior. Phys. Rev. A. 1988, 38, 3098-3100.

105. Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.;

Singh, D. J.; Fiolhais, C. Atoms, molecules, solids, and surfaces: Applications

of the generalized gradient approximation for exchange and correlation. Phys.

Rev. B 1992, 46, 6671-6687.

106. Perdew, J. P. Density-functional approximation for the correlation energy of

the inhomogeneous electron gas. Phys. Rev. B. 1986, 33, 8822-8824.

107. Gaussian 98 (Revision A.11.3), Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.;

Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V.G.;

Montgomery, J. A.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J.

M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.;

Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; 98

Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma,

K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.;

Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,

P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-

Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe,

M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.;

Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian, Inc., Pittsburgh

PA, 1998.

108. Eichkorn, K.; Treutler, H.; Öhm, H.; Häser, M.; Ahlrichs, R. Auxiliary basis

sets to approximate Coulomb potentials. Chem. Phys. Lett. 1995, 242, 652-

660.

109. Eichkorn, K.; Weigend, F.; Treutler, O. Ahlrichs, R. Auxiliary basis sets for

main row atoms and transition metals and their use to approximate Coulomb

potentials. Theor. Chem. Acc. 1997, 97, 119-124.

110. Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C. Electronic structure

calculations on workstation computers: the program system TURBOMOLE

Chem. Phys. Lett. 1989, 162, 165-169

111. Schaefer, A.; Horn, H.; Ahlrichs, R. Fully optimized contracted Gaussian

basis sets for atoms Li to Kr. J. Chem. Phys. 1992, 97, 2571-2577.

112. Schaefer, A.; Huber, C.; Ahlrichs, R. Fully optimized contracted Gaussian

basis sets of triple zeta valence quality for atoms Li to Kr J. Chem. Phys.

1994, 100, 5829-5835.

99

113. Scherlis, D.A.; Estrin, D.A. Structure and spin-state energetics of an iron

porphyrin model: An assessment of theoretical methods. Int. J. Quantum

Chem. 2002, 87, 158-166

114. Miertus, S.; Scrocco, E.; Tomasi, J. Electrostatic interaction of a solute with a

continuum. A direct utilization of ab initio molecular potentials for the

prevision of solvent effects. Chem. Phys. 1981, 55, 117-129.

115. Miertus, S.; Tomasi, J. Approximate evaluations of the electrostatic free

energy and internal energy changes in solution processes. Chem. Phys. 1982,

65, 239-245.

116. Cossi, M.; Barone, V.; Cammi, R.; Tomasi, J. Ab initio study of solvated

molecules: a new implementation of the polarizable continuum model. Chem.

Phys. Lett., 1996, 255, 327-335.

117. Reed, A. E.; Weinstock, R. B.; Weinhold, F. Natural population analysis. J.

Chem. Phys. 1985, 83, 735-746.

118. Weil, J. A.; Bolton, J. R.; Wertz, J. E. Electron paramagnetic resonance:

elementary theory and practical applications. John Wiley & Sons: New York,

1994.

119. Weiss, H.; Ahlrichs, R.; Häser, M. A direct algorithm for self-consistent-field

linear response theory and application to C60: Excitation energies, oscillator

strengths, and frequency-dependent polarizabilities. J. Chem. Phys. 1993, 99,

1262-1270.

100

120. Bauernschmitt, R.; Ahlrichs, R. Treatment of electronic excitations within the

adiabatic approximation of time dependent density functional theory Chem.

Phys. Lett. 1996, 256, 454-464.

121. Bauernschmitt, R.; Häser, M.; Treutler, O.; Ahlrichs, R. Calculation of

excitation energies within time dependent density functional theory using

auxiliary basis sets. Chem. Phys. Lett. 1997, 264, 573-578.

122. Scherlis, D. A.; Martí, M. A.; Ordejón, P. Estrin, D. A. Environment effects

on chemical reactivity of heme proteins. Int. J. Quantum. Chem. 2002, 90,

1505-1514.

123. Davydov, R.; Kappl, R.; Hutterman, J.; Peterson, J. A. EPR-spectroscopy of

reduced oxyferrous-P450cam. FEBS 1991, 295, 113-115.

124. Kobayashi, K.; Iwamoto, T.; Honda, K. Spectral intermediate in the reaction

of ferrous cytochrome P450cam with superoxide anion. Biochem. Biophys.

Res. Commun. 1994, 201, 1348-1355.

125. Li, X. Y.; Zgierski, M. Z. Iron motion in a five-coordinated heme model.

Chem. Phys. Lett. 1992, 188, 16-20.

126. Scheidt, W. R.; Geiger, D. K. Molecular stereochemistry and crystal structure

of a 'spin-equilibrium' (S = 1/2, S = 5/2) hemichrome salt: bis-(3-

chloropyridine)octaethylporphinatoiron(III) perchlorate. J. Chem. Soc., Chem.

Commun. 1974, 1154-1155.

127. Létard, J. F.; Guionneau, P.; Rabardel, L.; Howard, J. A. K.; Goeta, A. E.;

Chasseau, D.; Kahn, O. Structural, Magnetic, and Photomagnetic Studies of a

Mononuclear Iron(II) Derivative Exhibiting an Exceptionally Abrupt Spin 101

Transition. Light-Induced Thermal Hysteresis Phenomenon Inorg. Chem.

1998, 37, 4432-4441.

128. Blakesley, D. W.; Payne, S. C.; Hagen, K. S. Spin-State Variation in Solid

State and Solution of Mononuclear Iron(II) 1,4,7-Trimethyl-1,4,7-

triazacyclonane Complexes. Inorg. Chem. 2000, 39, 1979-1989.

129. Smith, D. M. A.; Dupuis, M.; Vorpagel, E. R.; Straatsma, T. P.

Characterization of Electronic Structure and Properties of a Bis(histidine)

Heme Model Complex. J. Am. Chem. Soc. 2003, 125, 2711-2717.

130. Harris, D.; Loew, G.; Waskell, L. Structure and Spectra of Ferrous Dioxygen

and Reduced Ferrous Dioxygen Model Cytochrome P450. J. Am. Chem. Soc.

1998, 120, 4308-4318.

131. Newcomb, M.; Aebisher, D.; Shen, R.; Chandrasena, R. E. P.; Hollenberg, P.

F.; Coon, M. J. Kinetic Isotope Effects Implicate Two Electrophilic Oxidants

in Cytochrome P450-Catalyzed Hydroxylations J. Am. Chem. Soc. 2003; 125,

6064-6065.

132. Chandrasena, R. E. P.; Vatsis, K. P.; Coon, M. J.; Hollenberg, P. F.;

Newcomb, M. Hydroxylation by the Hydroperoxy-Iron Species in

Cytochrome P450 Enzymes J. Am. Chem. Soc. 2004, 126, 115-126.

133. Toy, P. H.; Newcomb, M.; Coon, M. J.; Vaz, A. D. N. Two Distinct

Electrophilic Oxidants Effect Hydroxyl- cation in Cytochrome P-450-

Catalyzed Reactions. J. Am. Chem. Soc., 1998, 120, 9718-9719.

102

134. Newcomb, M.; Toy, P. H. Hypersensitive Radical Probes and the Mechanisms

of Cytochrome P450-Catalyzed Hydroxylation Reactions Acc. Chem. Res.

2000, 33, 449-455.

135. Volz, T. J.; Rock, D. A.; Jones, J. P. Evidence for Two Different Active

Oxygen Species in Cytochrome P450 BM3 Mediated Sulfoxidation and N-

Dealkylation Reactions. J. Am. Chem. Soc. 2002, 124, 9724-9725.

136. Horrocks, W. D.; Greenberg, E. S. Evaluation of dipolar nuclear magnetic

resonance shifts in low-spin hemin systems: ferricytochrome c and

metmyoglobin cyanide. Biochim. Biophys. Acta. 1973, 322, 38-44.

137. Goff, H. M. Iron(III) porphyrin-imidazole complexes. Analysis of carbon-13

nuclear magnetic resonance isotropic shifts and unpaired spin delocalization J.

Am. Chem. Soc. 1981, 103, 3714-3722.

138. Tan, H.; Simonis, U.; Shokhirev, N. V.; Walker, F. A. Substituent Effects on

the Choice of the Orbital Preferred for Electron Spin Delocalization in Two

Paramagnetic Low-Spin Iron(III) Porphyrins: Mapping the Spin Density

Distribution at the Pyrrole Positions by 1H COSY and NOESY Techniques. J.

Am. Chem. Soc. 1994, 116, 5784-5790.

139. Johansson, M. P.; Blomberg, M. R. A.; Sundholm, D.; Wikström, M. Change

in electron and spin density upon electron transfer to haem. Biochim. Biophys.

Acta. 2002, 1553, 183-187.

140. Johansson, M. P.; Sundholm, D.; Gerfen, G.; Wikström, M. The Spin

Distribution in Low-Spin Iron Porphyrins. J. Am. Chem. Soc. 2002, 124,

11771- 11780. 103

141. Munzarová, M.; Kaupp, M. A Critical Validation of Density Functional and

Coupled-Cluster Approaches for the Calculation of EPR Hyperfine Coupling

Constants in Transition Metal Complexes. J. Phys. Chem. A 1999, 103, 9966-

9983.

142. Simonson, T.; Perahia, D. Internal and interfacial dielectric properties of

cytochrome c from molecular dynamics in aqueous solution. Proc. Natl. Acad.

Sci. USA 1995, 92, 1082-1086.

143. O’Malley, P. J. Density Functional Calculations Modelling the Spin Density

Distribution, Hyperfine Couplings, and Hydrogen Bonding Environment of

the Ascorbate (Vitamin C) Free Radical. J. Phys. Chem. B 2001, 105, 11290-

11293.

144. Vidakovic, M.; Sligar, S. G.; Li, H.; Poulos, T. L. Understanding the Role of

the Essential Asp251 in Cytochrome P450cam Using Site-Directed

Mutagenesis, Crystallography, and Kinetic Solvent Isotope Effect

Biochemistry 1998, 37, 9211-9219.

145. Kamachi, T.; Yoshizawa, K. A Theoretical Study on the Mechanism of

Camphor Hydroxylation by Compound I of Cytochrome P450 J. Am. Chem.

Soc. 2003, 125, 4652-4661.

146. Guallar, V.; Harris, D. L.; Batista, V. S.; Miller, W. H. Proton-Transfer

Dynamics in the Activation of Cytochrome P450eryF J. Am. Chem. Soc. 2002,

124, 1430-1437.

104

147. Harris, D. L. Oxidation and electronic state dependence of proton transfer in

the enzymatic cycle of cytochrome P450eryF. J. Inorg. Biochem. 2002, 91,

568-585.

148. Davydov, R.; Macdonald, I. D. G.; Makris, T. M.; Sligar, S. G.; Hoffman, B.

M. EPR and ENDOR of Catalytic Intermediates in Cryoreduced Native and

Mutant Oxy-Cytochromes P450cam: Mutation-Induced Changes in the Proton

Delivery System. J. Am. Chem. Soc. 1999, 121, 10654-10655.

149. Davydov, R. M.; Yoshida, T.; Ikeda-Saito, M.; Hoffman, B. M. Hydroperoxy-

Heme Oxygenase Generated by Cryoreduction Catalyzes the Formation of -

meso-Hydroxyheme as Detected by EPR and ENDOR J. Am. Chem. Soc.

1999, 121, 10656-10657.

150. Ryde, U. Combined quantum and molecular mechanics calculations on

metalloproteins. Curr. Opin. Chem. Biol. 2003, 7,136-142.

151. Morokuma, K. New challenges in quantum chemistry: Quests for accurate

calculations for large molecular systems. Philos. Transact. Ser. A Math. Phys.

Eng. Sci. 2002, 360, 1149-1164.

105

CHAPTER 3

THE FINAL CATALYTIC STEP OF CYTOCHROME P450 AROMATASE: A

DENSITY FUNCTIONAL THEORY STUDY

3.1 INTRODUCTION

Aromatase (CYP19) is the cytochrome P450 enzyme responsible for the conversion of androgens, including androstenedione and testosterone, to the estrogen products, estrone and estradiol. [15,152] This enzyme plays a key role in the regulation of these sex steroids in reproductive and adipose tissue and may modulate sexual differentiation in the brain. Aromatase has been a particularly attractive target for inhibition in the treatment of hormone-dependent breast cancer since the aromatization of androgen substrates is the terminal and rate-limiting step in estrogen biosynthesis [153] Inhibitors of this enzyme have proven very efficacious in attenuating this disease. [154-156] From a mechanistic standpoint, aromatase is very interesting because it is one of few enzymes capable of constructing an aromatic ring.

106

Biotransformation of androgens by aromatase proceeds in three oxidative steps, each

consuming a single mole of molecular oxygen and NADPH (Figure 3.1). The first oxidation occurs at the 19-methyl group (Figure 3.1, step a) by the classical enzyme- mediated hydrogen atom abstraction-hydroxyl radical rebound mechanism originally

proposed by Groves and coworkers. [18] The first hydroxylation proceeds with retention

3 of configuration. [19,20] Experiments using [19- H3]androst-4-ene-3,7-dione show that

the small tritium kinetic isotope effect observed with aromatization is exclusively

associated with the first hydroxylation step. [21,22] Studies using the labeled suicide

3 14 substrate analog [19- H3, 4- C]androst-4-ene-3,6,17-trione also revealed a marked

tritium isotope effect with the first 19-hydroxylation. [23] The second hydroxylation

reaction (Figure 3.1, step b) stereoselectively removes the 19-pro-R hydrogen without an

apparent kinetic isotope effect to yield the 19-gem diol, which may dehydrate to the 19-

aldehyde (Figure 3.1, step c). [24,25]

R H HO OH H Hβ19 H OH Hα C D 1 a b A B O O O c +/- H2O R=O Androstenedione R=OH Testosterone H O

H O + O OH HO d

R=O Estrone R=OH 17β-Estradiol

Figure 3.1: Intermediates in the catalytic turnover of cytochrome P450 aromatase. 107

The third step oxidatively cleaves the C10-C19 bond resulting in aromatization of the

steroid A-ring and release of formic acid. (Figure 3.1, step d) A number of mechanisms

have been proposed and shown to be experimentally unlikely. These include 2β-

hydroxylation, [29,39,157-159] 4,5-epoxidation, [30,31] Baeyer-Villiger oxidation of

C19, [32] and 10β-hydroxyestr-4-ene-3,17-dione formation. [235,160,161] It has been

known for many years that the 1β and 2β hydrogens are lost stereospecifically to the

aqueous medium. [26-28] A number of model reaction studies have suggested that

formation of the 2,3-enol is a prerequisite for aromatization. [162-165] Differences in the stereoselectivity of the 2-hydrogen atom removal between androstenedione and testosterone indicate that the enzyme mediates the enolization process. [66] A homology model of the aromatase enzyme has been constructed using the X-ray structures of soluble bacterial P450s. The active site model suggests the presence of an aspartic acid residue near the 2β hydrogen and lysine or histidine residues near the 3-ketone of androstenedione. The orientation of these residues is supportive of an enzyme acid-base catalyzed enolization process to selectively remove the 2β hydrogen. [34] One

mechanism for the oxidative deformylation step that has received significant favor

involves nucleophilic attack of the 19-aldehyde by the reduced ferrous dioxygen intermediate (Figure 3.2). The resulting peroxo hemiacetal is suggested to decay via a process by which the proximal oxygen atom removes the 1β-hydrogen, resulting in

aromatization of the steroid A ring and formic acid release. [179] Studies monitoring the

18 incorporation of enzymatically-consumed O2 into the formate product have been

performed using crude microsomal preparations containing aromatase. [32] However, 108

these results are open to interpretation since evidence exists that isolated 19-hydroxy and

19-oxo androstenedione metabolites may be products of P450 enzymes other than aromatase. [42]

Cys S FeII O Cys O S FeII O O O O H H

HO HO

Cys O O S FeII OH H

HO

Figure 3.2: Deformylation of 19-oxo androstenedione as initially proposed by Akhtar.

[32]

In this chapter, a model system of the cytochrome P450 active site and truncated steroid substrates are studied with density functional theory to unravel the mysterious third step of aromatase catalysis. First, the feasibility of the mechanism in which the reduced ferrous dioxygen intermediate mediates androgen aromatization is explored. Secondly, we consider processes and locate transition states in which the widely accepted P450

109

oxidant, Compound I (see Figure 3.3), abstracts the 1β-hydrogen atom initiating the

aromatization and deformylation cascade.

3.2 COMPUTATIONAL METHODOLOGY

3.2.1 MODEL SYSTEMS AND NOMENCLATURE

The reduced ferrous dioxygen intermediate and Compound I of Cytochrome P450 were

2- 0 modeled as [(Porphyrin)(Fe-O2)(SH)] and [(Porphyrin)(FeO)(SH)] complexes,

respectively (Figure 3.3). QM/MM investigations by Shaik and coworkers including the

full protoporphyrin IX and steric and electric field contributions of the apoprotein reveal

that this truncated system may be a reliable mimic for computationally intensive

mechanistic studies. [167] The aromatase enzyme aromatizes the A ring of steroid

substrates androstenedione and testosterone. These substrates are considerably large for

full quantum mechanical treatment, especially in combination with the P450 model

systems. In this study, we are exclusively interested in the chemical transformation which

occurs in the A ring of the steroid system. Atoms in the C and D rings of the substrate are

sufficiently distant and likely have minimal influence on the computed relative energetics

of stationary points on the aromatization potential energy surface (PES). For this reason,

the trans-decalin structures displayed in Figure 3.3 are employed, which share common

structural features with the A and B rings of the steroid substrates. The hydration status of

the doubly oxidized substrate which participates in the third catalytic step and at what

point in the enzymatic transformation the prerequisite tautomerization occurs is 110

unknown. These facts suggested the use of three model steroid systems. The first two represent the enol and keto tautomers of the hydrated form of the steroid and therefore contain the 19,19-gem diol functional group (Scheme 1, step c). The third model is in the enol form likely necessary for aromatization, although it contains the 19-aldehyde to explore the possibility of this hydration status to produce the experimentally observed products.

Herein, three steroid model systems and two spin multiplicity PESs are investigated and therefore a myriad of species are discussed. To simplify the discussion, a simple nomenclature will be laid out to aid the reader in navigating through the chapter (Figure

3.3). Each species will be identified unless noted otherwise with a three-letter code preceded by a numerical superscript indicating its spin multiplicity. The first letter will indicate the substrate (E = enol-19,19-gem diol, K = keto-19,19-gem diol, and O = enol-

19-oxo) and the second pair of letters will identify this species’ place on the PES. RS =

Reactant Supermolecule (reactants separated by infinite distance), RC =van der Waals complex of reactants, TS = Transition state, PC = van der Waals complex of products,

PS = Product Supermolecule, and HR = Hydroxyradical intermediate. For example,

2ETS is the transition state of the enol-19,19-gem diol model steroid on the doublet potential energy surface.

111

O O FeII S Fe = H 2- [(Porphyrin)(Fe-O2)(SH)] Model Peroxo Intermediate O N N FeIV Fe S N N H [(Porphyrin)(FeO)(SH)]0 Model Compound I

HO OH HO OH H H

HO O Enol-19,19-Diol (E) Ketone-19,19-Diol (K) O 19 H 1 9 2 8

HO HO 3 5 7 Enol-19-Oxo (O) 4 6

Figure 3.3: Model systems and the androstane-based numbering system used in this study.

112

3.2.2 THEORETICAL METHODS

The systems were studied with the unrestricted hybrid density functional theory method

UB3LYP, as defined in the Gaussian 03 suite in which the VWN functional III is used instead of functional V. [100] Geometry optimizations of transition states and minima were computed with the Gaussian 03 suites of programs [170] and utilized the TZV basis

set [111,112] for iron and the 3-21G* basis set [171] for the remaining atoms. The 3-

21G* basis set is comparatively small relative to those recently used to study P450 model

systems, but in this study proves to produce geometric parameters comparable to similar

species previously described in the literature. In addition, this basis set is a reasonable

choice to bring ab initio molecular dynamics (AIMD) within reach of the computational

resources available. Unless otherwise noted, properties (i.e barrier heights, spin densities)

of the optimized species were derived from single-point energy computations using the

Wachter’s + f basis set [174-176] for iron and the 6-311+G** basis set [177,178] for the

remaining atoms. Spherical harmonic basis functions (i.e. 5d, 7f) were used in all

calculations. Minima were verified to have all real vibrational frequencies, and transition

states had a single imaginary vibrational frequency. These frequency analyses also

provided the necessary parameters to compute zero-point vibrational energy corrections.

The effect of a low dielectric continuum (ε = 5.621) on the electronic structure of these

systems and their relative energetics were evaluated with the polarizable continuum

model (PCM) of Tomasi and coworkers. [114-116] We believe the dielectric constant of

this solvent is a reasonable approximation of the environment experienced by the heme

113

center at the protein active site. In many of the computations including this solvation model, unbound hydrogen atoms, or those transferred as the complex proceeds to products, were treated explicitly for their contribution to the solute cavity. This approximation has been applied in a number of elegant studies to investigate the consequences of a polarized environment on the structure and reactivity of P450

Compound I. [179-186]

Some of the reaction mechanisms described in this study are concerted, in which a complex series of transformations occur in a barrierless fashion on the path from 1β- hydrogen atom abstraction to eventual product formation. Unfortunately, intrinsic reaction coordinate (IRC) calculations on complexes described here with such a large number of active internal coordinates are computationally intractable. [187,188] As an alternative, AIMD simulations were carried out on the Born-Oppenheimer potential energy hypersurface in the Turbomole 5.6 suite of programs. [110,189] To initialize each

AIMD run, 1β-hydrogen atom abstraction transition state geometries were used for the initial atomic positions and random velocities were applied at 310.15 K. A time step, ∆t, of 40 a.u. (0.97 fs) was chosen. This time step is approximately one order of magnitude smaller than the shortest vibrational period of the system (the O-H stretch, 9.5 fs). Some simulations were repeated with a 20 a.u. time step, which traced out nearly identical trajectories, validating the choice of a longer, and computationally convenient, time step.

The wave function was fully converged and the gradient computed at each time step at

114

the B3LYP/(TZV,3-21G*) level of theory. The Leapfrog Verlet algorithm [190] propagated the nuclear coordinates along the dynamical trajectory for 200 time steps:

r(t + ∆t) = r(t) + ∆tv(t + ∆t / 2)

v(t + ∆t / 2) = v(t − ∆t / 2) + ∆t[ f (t) / m]

Important geometries which correspond to inflection points on the molecular dynamics energy profiles were characterized with single-point energy calculations using the TZVP basis set for iron and the SV(P) basis set [111,112] for the remaining atoms in the gas phase with Turbomole. Single-point energies of the geometries in the presence of a low- dielectric continuum (ε = 5.621) with the PCM model were computed in Gaussian 03.

Population analyses was carried out with the natural population analysis (NPA) [117] and

Mulliken methods. Time-dependent density functional theory (TD-DFT) calculations at the B3LYP/(TZVP,SV(P)) level of theory were carried out with the ESCF module [119-

121] of the Turbomole 5.6 suite of programs.

3.3 RESULTS AND DISCUSSION

3.3.1 DEFORMYLATION BY THE FERROUS PEROXO INTERMEDIATE

A mechanism that has remained consistent with all available experimental aromatase data was proposed by Akhtar and involves attack of the reduced ferrous dioxygen intermediate

115

on C19 of the 19-oxo enolized steroid. [32] The resulting peroxohemiacetal intermediate is believed to fragment into the aromatized product, formic acid, and a hydroxo ferric enzyme intermediate, which may be rapidly protonated leaving the aqua-bound P450.

This mechanism is outlined in Figure 3.2. To study whether the operation of this mechanism in the final catalytic step of aromatase is feasible, we constructed and optimized models of the ferrous peroxo hemiacetal intermediate with the enolized substrate and performed rigid PES scans of the C1-H1β distance coordinate to estimate the barrier of hydrogen-atom abstraction by the proximal oxygen of the ferrous peroxo unit.

Three models of the ferrous peroxo hemiacetal intermediate were constructed and optimized at the B3LYP/(TZV,3-21G*) level of theory. Structures and key geometric parameters are displayed in Figure 3.4. Only the doublet state was considered in these studies since previous density functional theoretical work in our laboratory with extended models of the P450 active site have indicated this to be the ground state and well- separated from the higher-lying quartet state. (Table 2.1) Two models are dianionic, one with a water molecule hydrogen-bonded to the aldehyde oxygen. In the third model, the aldehyde oxygen is protonated with a net system charge of -1. Formation of the unhydrated and hydrated dianionic models from their isolated components was 31.7 and

67.8 kcal/mol exothermic, respectively. [191] The rigid PES scans of the C1-H1β coordinate were conducted in 0.1 Å increments, the energies of which are plotted relative to that of the respective ferrous peroxo hemiacetal reactant. (Figure 3.5) Results of the 116

PES scans are supportive of very high barriers for hydrogen-atom abstraction by the proximal oxygen (Op). For the protonated species, the PES briefly plateaus at approximately 80 kcal/mol when the C1-H1β coordinate is 2.6 Å, placing the hydrogen atom 1.3 Å from Op. Further scanning does not provide evidence of an additional potential energy well beyond this point. Complete geometry optimization of the species with C1-H1β equal to 2.6 Å returns to the reactant complex. PES scans of the same coordinate in the unhydrated and singly hydrated analog support lower barriers (~ 60 kcal/mol) for 1β-H abstraction by the proximal oxygen; however, these values are still too immense to be accessible by enzymatic catalysis. The approximate barrier for hydrogen-atom abstraction from the dianionic species is 63 kcal/mol and occurs at a C1-

H1β distance of 2.2 Å where the H1β-Op distance is 1.2 Å. SCF convergence difficulties prevented scanning beyond a C1-H1β value of 2.1 Å in the singly hydrated intermediate.

The potential energy curve prior to this value followed that of the unhydrated species identically, indicating that hydrogen bonding to the aldehyde oxygen has no impact on the 1β-H abstraction barrier. In the unhydrated complex, a shallow potential energy well is encountered at 2.6 Å and H1β-Op of 1.12 Å. Geometry optimization from this well does not result in fragmentation to the experimentally observed products or return to reactants; alternatively, it optimizes to a stable, proximal-oxygen protonated species.

117

(a)

Fe-S = 2.370 (2.388) Fe-O = 1.839 (1.831) O-O = 1.518 (1.511) C1-H1β = 1.099 (1.098) H1β-Op = 2.030 (2.068) C19-OAld = 1.364 (1.341) OAld-H2O = 1.457

(b)

Fe-S = 2.310 Fe-O = 1.845 O-O = 1.528 C1-H1β = 1.099 H1β-Op = 2.145 C19-OAld = 1.446

Figure 3.4: Structure of the hydrated peroxo hemiacetal adduct of the (a) enol-19-oxo steroid and (b) protonated peroxo hemiacetal adduct. OAld denotes the aldehyde oxygen and distances are given in Å. Geometric parameters for the unhydrated species are in parentheses.

118

90 80 70 60 50 40

E (kcal/mol) 30

∆ 20 10 0 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5 2.7 2.9 3.1 C1-H1β (Å)

Figure 3.5: Results of the B3LYP/(TZV,3-21G*) C1-H1β PES scans. Protonated (■), hydrated (○) and unhydrated (▼) peroxo hemiacetal adducts are plotted with regard to the C1-H1β bond distance.

In light of the apparently high energetic barrier for hydrogen-atom abstraction in the peroxo hemiacetal species and lack of fragmentation upon transfer of the 1β-H to the proximal oxygen atom, we have considered the possibility that Compound I may be the active oxidant in the final step of aromatase catalysis. Given that 2β hydrogen loss to the aqueous medium can be accounted for by a 2,3-enolization step prerequisite for aromatization, we have considered several mechanisms of aromatization which are initiated by 1β-H abstraction by Compound I. Due to the lack of knowledge regarding the

2,3 tautomeric and C19 hydration status during this catalytic step, these possibilities have been incorporated into the model systems. The results of these studies are discussed next.

119

3.3.2 THE MODEL STEROID-COMPOUND I VDW COMPLEXES.

The electronic structure of P450 Compound I has been a flourishing debate between theoreticians over the last decade, although more recently a working consensus has evolved. [179-186, 192-195] There is now widespread agreement that Compound I has two closely lying spin states. Each state has two electrons in a triplet configuration localized to the FeO moiety, while a third electron resides in an orbital of mixed porphyrin π and sulfur p character from the distal thiolate ligand. The contributing porphyrin π orbital shares similar properties to the highest-lying a2u orbital of iron porphyrin in D4h symmetry. Antiferromagnetic and ferromagnetic coupling of the electron in this mixed character orbital to the FeO triplet pair gives rise to nearly degenerate doublet and quartet states, respectively. The optimized geometries of 2ERC,

2KRC, and 2ORC are representative of these geometries on both the doublet and quartet spin surfaces and are displayed in Figure 3.6. Critical geometric parameters for these species and the 4A-2A energetic spacings are listed in Tables 3.1 and 3.2, respectively.

The geometries of these species were identified by distorting the reaction vector in the corresponding 1β-hydrogen atom transition state toward reactants as discussed below and subjecting the system to full optimization. Geometric parameters for the

[(Porphyrin)(FeO)(SH)] component of the reactant complexes are in agreement within the estimated experimental uncertainty with the crystal structure of oxyferrous cytochrome P450cam. [93] The van der Waals complexes identified here have doublet

120

and quartet states separated by small energetic spacings (< ~ 1.5 kcal/mol) whose values are listed in Table 3.2.

RC spin densities illustrate these intermediates generally share the same electronic structure properties as species studied previously. Group and atomic charges and spin densities illustrating this point are listed in Table 3.3. Fe and O atomic spin densities indicate the presence of two parallel, unpaired electrons centered on these atoms. In the cases of 2,4KRC and 2,4ORC, a third antiferromagnetically or ferromagnetically coupled electron to the triplet pair is shared between the porphyrin macrocycle and the thiolate ligand. No appreciable spin density accumulates on the substrate in these cases. In contrast, 2,4ERC has spin density parallel to that localized to the porphyrin and the thiolate ligand appearing on the model steroid. Analysis of the singly-occupied Kohn-

Sham orbital of 2ERC (See Appendix B) indicates the accumulation of spin density on the substrate arises from mixing of an antibonding π* orbital localized on the diene system with the a2u porphyrin/sulfur p orbital. The presence of a low dielectric continuum enhances the porphyrin’s radical character of these intermediates and diminishes the thiolate contribution. Inclusion of the solvation model has effects on the spin density localized to the substrate of 2,4ERC. β Spin density is abolished from the substrate portion of 2ERC in the presence of solvation, significantly enhancing the porphyrin’s radical character. 4ERC retains the substrate-centered spin density, although it is also slightly decreased in favor of the porphyrin radical.

121

(a) (b)

(c)

Figure 3.6: Fully optimized structures of (a) 2ERC, (b) 2KRC, and (c) 2ORC at the

B3LYP/(TZV,3-21G*) level of theory.

122

Geometric Parameter 2ERC(4ERC) 2ETS(4ETS) 2KRC(4KRC) 2KTS(4KTS) Fe-O 1.664 (1.676) 1.715 (1.732) 1.656 (1.658) 1.810 (1.781) Fe-S 2.447 (2.408) 2.324 (2.298) 2.455 (2.459) 2.288 (2.268) Fe-N 2.007 (2.007) 2.010 (2.010) 2.004 (2.003) 2.003 (2.001) H1β-O(Fe) 2.413 (2.487) 1.601 (1.484) 2.537 (2.524) 1.129 (1.226) C1-H1β 1.093 (1.091) 1.158 (1.186) 1.096 (1.096) 1.447 (1.359) ∠(Fe)O-H1β-C1 127.2 (115.2) 170.6 (173.0) 140.8 (141.5) 146.4 (154.4) O-Hpro-S-O(Fe) 1.616 (1.610) 1.437 (1.484) 1.657 (1.671) 1.533 (1.596) O-Hpro-S 1.020 (1.023) 1.068 (1.052) 1.015 (1.014) 1.044 (1.028) O-Hpro-R 0.996 (0.997) 0.997 (0.997) 0.997 (0.997) 0.997 (0.997) Geometric Parameter 2ORC(4ORC) 2OTS(4OTS) Fe-O 1.652 (1.658) 1.689 (1.730) Fe-S 2.479 (2.477) 2.359 (2.312) Fe-N 2.004 (2.004) 2.006 (2.005) H1β-O(Fe) 2.190 (2.145) 1.526 (1.414) C1-H1β 1.103 (1.103) 1.174 (1.222) ∠(Fe)O-H1β-C1 165.1 (166.2) 170.8 (171.1) HAldehyde-O(Fe) 2.140 (2.176) 1.984 (1.988) CAldehyde-H 1.104 (1.104) 1.104 (1.102)

Table 3.1. Bond lengths (in Å) and angles (in degrees) for the 2A(4A) reactant clusters and 1β-hydrogen atom abstraction transition states. The quartet surface value is listed in parentheses.

123

Theory ERC KRC ORC

B3LYP/(TZV, 3-21G*) -1.64 0.33 0.04 B3LYP/(TZV, 3-21G*)+ ZPE -1.33 0.42 0.10 B3LYP/(Wachter’s + f, 6-311+G**)b -0.08 0.39 0.22

PCM (ε = 5.621)-B3LYP/(Wachter’s + f, 6-311+G**)b 1.11 0.62 0.12

Table 3.2. 4A-2A Energetic splitting for reactant complexes. A negative value for the high spin-low spin splitting indicates that the high spin is favored. bSingle-point energy computation using the B3LYP/(TZV,3-21G*) fully-optimized geometry.

3.3.3 THE 1β-HYDROGEN ATOM ABSTRACTION/AROMATIZATION

SEQUENCE

Transition states for 1β-hydrogen atom abstraction from the substrates on the doublet and quartet spin surfaces were computed and representative 2ETS, 2KTS, and 2OTS transition-state structures are displayed in Figure 3.9. A complete list of critical geometric parameters is provided in Table 3.1. Computed energy profiles for the enol and keto species are shown in Figure 3.7 and for the 19-oxo species in the Figure 3.8 The relative energetics compared to the RC on the unique PESs are listed in Table 3.6.

124

2ERC 4ERC Charges Fe 0.95 (0.95) 0.93 (0.93) O -0.57 (-0.58) -0.59 (-0.60) Porphyrin -0.40 (-0.09) -0.45 (-0.27) SH -0.15 (-0.24) -0.16 (-0.27) Substrate 0.17 (0.04) 0.28 (0.21) Spin Densities Fe 1.20 (1.23) 1.13 (1.15) O 0.81 (0.78) 0.84 (0.83) Porphyrin -0.35 (-0.64) 0.24 (0.42) SH -0.45 (-0.36) 0.41 (0.31) Substrate -0.22 (0.00) 0.37 (0.29)

Table 3.3. 2,4ERC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

125

2ORC 4ORC Charges Fe 0.94 (0.95) 0.94 (0.95) O -0.56 (-0.59) -0.56 (-0.58) Porphyrin -0.28 (-0.18) -0.27 (-0.17) SH -0.07 (-0.15) -0.09 (-0.17) Substrate -0.03 (-0.03) -0.02 (-0.03)

Spin Densities Fe 1.16 (1.19) 1.09 (1.11) O 0.86 (0.83) 0.91 (0.88) Porphyrin -0.46 (-0.55) 0.45 (0.55) SH -0.56 (-0.47) 0.54 (0.45) Substrate 0.00 (0.00) 0.00 (0.00)

Table 3.4. 2,4ORC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

126

2KRC 4KRC Charges Fe 0.93 (0.94) 0.92 (0.94) O -0.48 (-0.53) -0.48 (-0.52) Porphyrin -0.36 (-0.23) -0.34 (-0.21) SH -0.10 (-0.19) -0.12 (-0.21) Substrate 0.02 (0.01) 0.02 (0.01) Spin Densities Fe 1.09 (1.14) 1.02 (1.06) O 0.90 (0.86) 0.97 (0.93) Porphyrin -0.42 (-0.53) 0.43 (0.54) SH -0.57 (-0.46) 0.54 (0.44) Substrate -0.01 (0.00) 0.03 (0.03)

Table 3.5. 2,4KRC B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

127

+40

2KTS +20 4KTS

2KRS 4ETS 4KRS 4ERS RC 2ETS 4KHR 0 2ERS ) l o m

/ -20 l ca k E (

-40

4KPS

2KPC 2KPS -60

4EPS -80

4EPC 2EPS

2EPC -100

Figure 3.7: Relative energy diagram of stationary points along the 1β- hydrogen atom abstraction/deformylation reaction for the enol-19,19-diol (E) and ketone-19,19-diol (K) species at the B3LYP/(Wachter’s + f, 6-311+G**)//B3LYP/(TZV,3-21G*) level of theory.

128

+40

+20

4OTS RC 2OTS 0 2ORS 4ORS ) l o

m -20 / l 4 ca OHR k E (

-40

-60

-80 2 2OPC OPS 4OPS

-100

Figure 3.8: Relative energy diagram of stationary points along the 1β- hydrogen atom abstraction/deformylation reaction for the enol-19-oxo (O) species at the

B3LYP/(Wachter’s + f, 6-311+G**)//B3LYP/(TZV,3-21G*) level of theory.

129

Theory RS TS HR PC PS

E

1 20.1 (20.8) 4.3 (5.9) [5.5 (6.8)] -80.2(-62.4) -50.1 (-33.9) 2 19.1 (21.6) 2.0 (3.4) -80.1 (64.2) -53.1 (-38.3) 3 5.6 (5.9) 7.5 (7.8) [ 6.9 (7.2)] -94.8(-88.3) -88.8 (-76.4) 4 -5.6 (-6.7) 5.9 (5.3) [ 5.4 (4.8)] -95.2(-94.6) -101.6 (-89.6) K 1 24.6 (24.3) 27.1 (21.4) (10.1) -41.9 -15.0 (-0.8) 2 23.1 (22.8) 22.3 (17.0) (7.5) -43.9 -20.2 (-7.1) 3 9.5 (9.3) 26.6 (23.1) (3.3) -60.1 -57.0 (-45.1) 4 -0.81 (-1.4) 25.1(22.9) (3.9) -65.1 -69.3 (-56.9) O 1 8.7 (8.7) 3.9 (3.0) (-14.8) -70.7 -52.3 (-52.4) 2 8.1 (8.1) 1.7 (0.52) (-16.2) -71.6 -55.9 (-56.0) 3 -0.58 (-0.62) 6.5 (7.0) (-22.9) -87.5 -85.2 (-85.4) 4 -5.2 (-5.2) 5.3 (6.8) (-19.1) -84.3 -88.2 (-88.3)

Table 3.6. Energies relative to the Reactant Complex of Stationary points on the 2A

(4A)1β-H atom abstraction/aromatization PES. Levels of theory (1= B3LYP/(TZV,3-

21G*); 2 = B3LYP/(TZV,3-21G*) + ZPE; 3 = B3LYP/(Wachter’s+f, 6-311+G**); 4 =

PCM-B3LYP/(Wachter’s+f, 6-311+G**)). Energies (kcal/mol) are relative to the reactant complex (RC) in each case. See Figure 3.3 and the text for definitions of the E, K, and O structures. The quartet energetics are listed in parentheses.

130

Hydrogen-atom abstraction from the enol-19,19-diol species (i.e. 2ETS and 4ETS) occurs with an exceptionally low energetic barrier (Table 3.6), and the barriers are 7.5 and 7.8 kcal/mol on the doublet and quartet surfaces, respectively. Inclusion of a dielectric continuum with ε = 5.621 lowers these barriers to 5.9 and 5.3 kcal/ mol, now with a slight preference for the quartet state. Consideration of zero-point vibrational energy contributions lowers these barriers even further by 2.3 and 2.5 kcal/mol on the doublet and quartet surfaces, respectively. Inspection of 2,4ETS geometries reveal that this reaction proceeds with an “early” transition state consistent with its energetic similarity to the reactant complex. This is illustrated by the minimal lengthening of the C1-H1β bond in the transition state, compared to the corresponding value in the reactant complexes (see

Table 3.1).

131

(a) 2A(4A) C1 -0.04 (0.11) C2 -0.27 (0.22) C3 -0.07 (0.11) C4 -0.05 (0.02) C5 -0.11 (0.12) C10 -0.01 (0.00)

(b) 2A(4A) C1 -0.09 (0.19) C2 -0.10 (0.05) C3 -0.08 (0.14) C4 -0.01 (0.02) C5 -0.07 (0.10) C10 0.00 (-0.01)

(c) 2A(4A) C1 0.09 (-0.50) C2 0.00 (-0.10) C3 0.00 (0.00) C4 0.00 (0.00) C5 0.00 (0.00) C10 0.00 (-0.01)

Figure 3.9: B3LYP/(TZV,3-21G*) fully optimized transition-state structures and A-ring

carbon B3LYP/(Wachter’s+f, 6-311+G**) NPA spin densities of (a) 2ETS, (b )2OTS,

and (c) 2KTS. C1-H1β and H1β-O(Fe) distances are given in Å. The quartet surface value is in parentheses.

132

(a) (b)

(c)

Figure 3.10: B3LYP/6-311+G**//B3LYP/6-31G* NPA spin densities of heavy atoms in the model steroid substrate A ring. (a) enol-19,19-diol, (b) enol-19-oxo, (c) keto-19,19- diol.

133

(a) 2ETS = 201i cm-1 4ETS = 162i cm-1

(b) 2OTS = 729i cm-1 4OTS = 556i cm-1

(c) 2KTS = 629i cm-1 4KTS = 1599i cm-1

Figure 3.11: Representative normal modes for the respective imaginary vibrational frequency of the (a) 2ETS, (b) 2OTS, and (c) 2KTS transition state structures.

Displacement vectors are depicted with arrows and the values for the doublet and quartet states are given.

134

Two possible sources for these strikingly low energy barriers in these model systems were considered. In 2,4ETS , the pro-S 19-OH forms an apparent hydrogen bond with the ferryl oxygen. As a result, the pro-S O-H distance (1.068 Å) is slightly longer than the pro-R OH (0.997 Å) in 2ETS. To investigate the role of this hydrogen bond in stabilizing the hydrogen-atom abstraction transition states, this bond was frozen to the value observed for the pro-R OH and single-point energies were computed. These frozen O-H barrier heights are listed in Table 3.6 in brackets, and demonstrate little impact on the abstraction barrier height. It appears that stabilization of the transition states of 2,4ETS species lies principally with the substrate itself. Spin densities of A-ring carbon atoms in the transition states (Figure 3.9) and the derived substrate C1 radicals (Figure 3.10) are delocalized over the entire ring, with the most prominent spin density appearing on the

C2 and C5 atoms.

Connection of the 2,4ETS transition states to the 2,4EPC (Figure 3.12; a) complexes indicates that this is a very exothermic process, as would be expected with a significant driving force of aromatization. The transition states on both spin surfaces are directly connected to a product complex consisting of the experimentally observed aromatized product, formic acid, and the resting aqua complex of the model enzyme. The identification of these pairs of stationary points indicate that after 1β-hydrogen atom abstraction, concomitant transfer of the pro-S hydrogen of the C19 gem diol occurs in a barrierless (concerted) manner to the hydroxy iron intermediate. The imaginary vibrational frequency in Figure 3.11 is consistent with hydrogen-atom transfer from C1 to

135

Compound I; however, the normal mode for the imaginary vibrational frequency lacks any significant motion of the pro-S hydroxyl group of the gem diol. Therefore, each hydrogen transfer does not occur in a synchronized manner, rather in two nonsynchronous distinct steps. A more detailed discussion of the electronic details of the second hydrogen transfer will be discussed later in this chapter. Noting this observation, we also tried to identify a similar transition state in which the pro-R hydrogen would be poised for transfer to give the experimentally observed products. Several trials failed to produce the desired stationary point, although all of our partial optimizations toward this goal decayed to the same trio of products.

The 19-oxo metabolites of aromatase substrates have been isolated as potential catalytic intermediates in the biosynthesis of estrogens. However, no experimental evidence exists supporting the presence of either the 19,19-gem diol or the dehydrated 19-oxo species at the enzyme active during catalysis site. As part of our initial hypothesis, we envisioned the possibility that after 1β-hydrogen atom abstraction from the dehydrated intermediate, the steroid A-ring might aromatize concomitantly liberating the formyl radical. This species was then expected to rebound to the ferric hydroxy radical intermediate leading to the experimentally observed products. To investigate this possibility, we identified the

19-oxo 1β-hydrogen atom abstraction transition states (2,4OTS) on both spin surfaces

(Figure 3.9). The barrier heights are similar to those computed with the hydrated analog and are 6.5 and 7.0 kcal/mol on the doublet and quartet surfaces, respectively. These barriers are lowered to 5.3 and 6.8 kcal/mol in the presence of ε= 5.621 dielectric field. 136

Inclusion of zero-point vibrational energy corrections lowers these barriers by 2.2 and 2.5 kcal/mol. C1-H1β-O(Fe) geometric parameters are similar to those for 2,4ETS, consistent with an early transition state. The spin densities of the A-ring in the transition states and

C1 radical underscore the conclusion that the low barriers are a direct result of the substrate’s ability to delocalize the developing radical character. Unlike the hydrated analog, only 2OTS decays directly to a product complex containing the aromatic entity.

Distortion of the 2OTS structure along the normal mode for the imaginary vibrational frequency (Figure 3.11; b), which is consistent with hydrogen transfer between C1 and the ferryl oxygen, and optimizing the resulting species toward products, proved that the

2OTS transition state is directly connected to 2OPC (Figure 3.12; b), The 2OPC structure is a vdW complex of the aromatized product, carbon monoxide and the ferric porphyrin aqua complex. As hypothesized, the formyl group is liberated from the bicyclic substrate en route to aromatization, but instead of rebound to formic acid, a ferric hydroxy species abstracts hydrogen from the formyl species, resulting in decarbonylation. The normal mode for the imaginary vibrational frequency has no significant motion of the 19-oxo hydrogen atom, indicating decarbonylation likely occurs in a second distinct step without an observable barrier. Carbon monoxide production is not observed experimentally, so indeed, these calculations are not supportive of a dehydration step following the second

C19 hydroxylation. For the quartet surface, 4OTS encounters an intermediate 4OHR

(Figure 3.13; a) on the PES indicative of a subsequent barrier for formyl rebound or decarbonylation. Analysis of the spin density distribution of 4OHR shows a low atomic charge localized to the substrate species of 0.02 and 0.03 e in the gas phase and with ε = 137

5.621, respectively, and a substrate spin density of 1.00 e in both environments. These parameters indicate the intermediate on the quartet surface is a complex of the substrate radical and Fe bound hydroxy radical intermediate. Elongating the C10-C19 bond distance of this species followed by partial optimization also resulted in the decarbonylation by the ferric hydroxyl radical or hydroxide bound intermediate. Our efforts to identify a transition state for formyl radical rebound on both spin surfaces repeatedly failed, always decaying via the decarbonylation route to generate carbon monoxide.

Hydrogen-atom abstraction from the keto form of the steroid substrate encounters a significantly higher energetic barrier than the two enolized substrates. The computed barriers for 1β-hydrogen atom abstraction are 26.6 and 23.1 kcal/mol on the doublet and quartet surfaces, respectively. Inclusion of an ε = 5.621 dielectric field lowers these barrier heights to 25.1 and 22.9 kcal/mol, respectively. Inclusion of the zero-point vibrational energy correction reduces these values by 4.8 and 4.4 kcal/mol. The energetics for the keto analog are comparable to those previously computed for ethane

[196,197] and camphor [145] with an analogous model system for the P450 enzyme in which the cysteinate ligand is modeled with a methylthiolate anion. This observation is to be expected, as inspection of the spin densities in 2,4KTS (Figure 3.9; c) and the isolated

C1 radical (3.11;c) indicate that the developing radical cannot delocalize the impending spin density over the steroid’s A ring.

138

2KTS is connected directly to 2KPC (Figure 3.12; c) which consists of the aqua-ligated model P450, formic acid, and the analogous dehydrogenated 3-oxo-1(10),4-diene product. Although considerably exothermic, the resulting product complex lies 28.7 kcal/mol higher in energy than the doublet surface of the enol tautomer. In contrast, the quartet state encounters 4KHR (Figure 3.13; c), which indeed can be described as a complex of the substrate radical and hydroxy radical intermediate, as spin density analysis reveals the presence of a full electron (spin density of 1.00 e) localized to the steroid substrate. 4KHR is only 3-4 kcal/mol below 4KTS indicating the presence of a subsequent barrier for secondary hydrogen-atom or proton transfer from the model substrate to the hydroxy radical bound to iron.

Formation of the 1(10) olefin from the keto substrate with energetics similar to those for other cytochrome P450 enzymes which are known to occur is very interesting, and the mechanism described here may have implications in the final step of lanosterol-14α- demethylase (CYP51). This enzyme is responsible for the removal of the 14α-methyl group from lanosterol in the biosynthesis of cholesterol. [198,199] Like aromatase, the first two hydroxylations appear to be unexceptional hydrogen atom abstraction-hydroxy radical rebound steps yielding the 14α-gem-diol, although the mechanism of the final

15α-hydrogen atom abstraction with concomitant deformylation has also been controversial. [200-202] In analogy to the mechanism described here, it is likely that the

Compound I intermediate might abstract the 15α-hydrogen atom yielding the hydroxy radical intermediate. In a subsequent step, a hydrogen atom from the gem-diol could be 139

lost to the hydroxyradical intermediate yielding the experimentally observed mixture of formic acid, the dehydrogenated steroid, and the resting state of the enzyme.

(a) (b)

(c)

Figure 3.12: Fully optimized structures of (a) 2,4EPC, (b) 2OPC, and (c) 2KPC at the

B3LYP/(TZV,3-21G*) level of theory. Selected distances are given in Å. Quartet surface values are given in parentheses.

140

(a) (b)

Figure 3.13: Fully optimized structures of (a) 4OHR and (b) 4KHR at the

B3LYP/(TZV,3-21G*) level of theory. Selected distances are given in Å.

141

2ETS 4ETS Charges Fe 0.96 (0.97) 0.94 (0.96) O -0.73 (-0.74) -0.73 (-0.74) Porphyrin -0.57 (-0.57) -0.59 (-0.57) SH -0.24 (-0.33) -0.19 (-0.29) 1β-H 0.34 (0.34) 0.35 (0.36) Substrate 0.25 (0.33) 0.22 (0.29) Spin Densities Fe 1.40 (1.39) 1.34 (1.38) O 0.46 (0.50) 0.67 (0.65) Porphyrin -0.12 (-0.09) 0.00 (0.00) SH -0.01 (0.01) 0.21 ( 0.12) 1β-H -0.02 (-0.03) 0.02 (0.02) Substrate -0.70 (-0.78) 0.76 (0.82)

Table 3.7: 2,4ETS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

142

2OTS 4OTS Charges Fe 0.94 (0.96) 0.91 (0.94) O -0.61 (-0.63) -0.63 (-0.66) Porphyrin -0.51 (-0.50) -0.56 (-0.53) SH -0.24 (-0.33) -0.17 (-0.26) 1β-H 0.32 (0.33) 0.34 (0.35) Substrate 0.09 (0.17) 0.09 (0.16) Spin Densities Fe 1.36 (1.36) 0.91 (0.94) O 0.39 (0.40) 0.80 (0.76) Porphyrin -0.18 (0.17) 0.07 (0.07) SH -0.11 (-0.05) 0.29 (0.19) 1β-H -0.01(-0.02) 0.00(0.01) Substrate -0.43 (-0.52) 0.59 (0.65)

Table 3.8: 2,4OTS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

143

2KTS 4KTS Charges Fe 0.90 (0.92) 0.95 (0.96) O -0.80 (-0.80) -0.76 (-0.76) Porphyrin -0.40 (-0.39) -0.46 (-0.41) SH -0.18 (-0.25) -0.06 (-0.11) 1β-H 0.37 (0.37) 0.37 (0.37) Substrate 0.11 (0.15) 0.32 (0.33)

Spin Densities Fe 0.90 (0.92) 0.95 (0.96) O -0.80 (-0.80) -0.76 (-0.76) Porphyrin -0.40 (-0.39) -0.46 (-0.41) SH -0.18 (-0.25) -0.06 (-0.11) 1β-H 0.37 (0.37) 0.37 (0.37) Substrate 0.11 (0.15) 0.32 (0.33)

Table 3.9: 2,4KTS B3LYP/(Wachter’s + f, 6-311+G**) NPA group and atomic charges as well as spin densities. Values computed in the presence of the PCM model with

ε=5.621 are in parentheses.

144

To date, aromatase has resisted structural elucidation by X-ray crystallography, so no template exists in which to construct an expanded QM system or QM/MM system to study the electronic and steric contributions of the protein to aromatization energetics.

Full QM/MM studies of the hydroxylation of camphor by cytochrome P450cam have been completed, and provide significant insight into the role of the surrounding protein.

Guallar et al. studied the 5-exo hydrogen-atom abstraction from camphor by the quartet state of P450cam. [203] Their results revealed a 11.7 kcal/mol barrier for hydrogen-atom abstraction (8.2 kcal/mol with zero-point vibrational energy corrections included). When the QM subsystem was extracted from the QM/MM model and the barrier recomputed in the gas phase, it rose to 20 kcal/mol. This study went on to illustrate the individual steric and electronic contributions of the surrounding protein to the lowering of the barrier.

Much of the energetic discrepancy was attributed to electrostatic interactions of the peripheral heme propionate substituents with positively charged amino acid side chains in the pocket, but this is still an area of debate. [167,203]

If the surrounding aromatase apoprotein could be considered here for the derived transition states, we may find the final aromatization step with enolized substrates to have a negligible barrier or be apparently barrierless and the keto substrates dehydrogenated with a minimal barrier. Another possibility that would allow steroid substrates in the keto form to contribute to the pool of aromatized product lies in utilizing the kinetic energy liberated during O-O bond cleavage. Using a model of the P450cam active site, which included amino acid side chains proposed to be involved in proton delivery to the distal oxygen of the peroxo intermediate by Sligar, [144] Kamachi and Yoshizawa computed

145

the O-O bond cleavage to be 50 kcal/mol exothermic. [145] A more recent QM/MM study by Guallar and Friesner of these steps in P450cam found that 80 kcal/mol was released in the bond-breaking step alone. [204] As proposed by Kamachi and Yoshizawa, the intermediates involved in dioxygen activation and substrate hydroxylation might decay faster than the system equilibrates; therefore, the kinetic energy released during O-

O scission may only be partially transferred to the protein. The remaining kinetic energy may be used by Compound I to overcome the high potential energy barrier for hydrogen- atom abstraction. [145]

3.3.4 AB INITIO MOLECULAR DYNAMICS OF 2,4ETS AND 2,4OTS

As noted above, concomitant formation of formic acid occurs directly after C1-H1β hydrogen-atom abstraction, and we were interested in examining the mechanistic structures along the reaction path. Therefore, to study the potential energy hypersurface between the hydrogen-atom abstraction transition states 2,4ETS and 2,4OTS to the respective product complexes, Born-Oppenheimer molecular dynamics was used. A similar strategy was used by Yoshizawa and coworkers to effectively follow the geometric and energetic changes which occur along the ethane to ethanol reaction pathway mediated by a similar model of P450 Compound I. [205] In this study, we limited our simulations to approximately 200 time steps of ~ 1 fs. This time scale was more than sufficient for the transition states to decay to products. Moreover, the formation of products occurred in ~ 50 fs in each simulation. Single trajectory calculations were performed for each system and spin surface. Single trajectories only

146

provide a sampling of the dynamical behavior of this very important enzymatic reaction.

These calculations are limited to providing information about the correct reaction pathway and how the electronic structure of the system changes as the reaction proceeds.

The computation of accurate thermodynamic parameters requires statistical treatments of data obtained from multiple trajectory runs, and are not considered here. However, we were very interested in understanding the electronic structure in the transformation sequence.

Energy profiles for 2,4ETS and 2,4OTS are plotted versus time in Figure 3.14. These plots include time points relevant to product formation only. All simulations decay to the respective products identified in the stationary point evaluations discussed above. The low-spin state of the aqua-bound P450 model produced in all combinations of products is energetically preferable in all simulations in agreement with the experimental low-spin resting state of cytochrome P450 enzymes. [206] Indeed, the doublet state is preferred energetically through the course of both MD trajectories, with the exception of a near degeneracy to the quartet trajectory at some time points.

147

(a) 2,4ETS

15

5 9

-5 15 19 -15 17 -25 26 37 12 -35 31 E (kcal/mol)

∆ -45 23 29 Doublet (40 au) 34 -55 Quartet (40 au) Doublet (20 au) -65 46 -75

400 560 720 880 1040 1200 1360 1520 1680 1840 Time (au) (b) 2,4OTS

15 10 5 9 0 -5 -10 19 30 -15 14 25 -20 E (kcal/mol) E

∆ -25 12 -30 16 27 52

-35 33 Doublet (40 au) 22 -40 Quartet (40 au) -45 Doublet (20 au) 48 -50 400 560 720 880 1040 1200 1360 1520 1680 1840 2000 Time (au)

Figure 3.14: Ab initio MD trajectory energy profiles initiated with (a) 2,4ETS and (b)

2,4OTS at the B3LYP/(TZV,3-21G*) level of theory. Only time points relevant to product formation are shown. Structures at time points subjected to B3LYP/(TZVP,SV(P)) single-point energy computations are labeled.

148

(a)

0.75 26 31 25

24 34 37 0.50 41 46

Atomic Charge Atomic 9 23 0.25 12 15 17 19

0.00 320 640 960 1280 1600 Time (au)

19 0.75 23 24

0.50 25

26

Spin Density 0.25 β

31 0.00

720 800 880 960 1040 1120 1200 (b) Time (au)

Figure 3.15: Gas phase (●) and ε = 5.621 (○) Mulliken atomic charges (a) and β spin densities (b) localized on the steroid substrate at critical points on the ab initio MD trajectory for propagation of 2ETS transition state structure from Figure 3.9.

149

Dynamics initiated with the 2,4ETS structures corroborate the connection of transition states to the experimentally observed products. The first 26 steps (~24 fs) of simulation are dominated by a decrease in the H1β-O-Fe angle from the transition state value of

141° to approximately 109°. As the simulation proceeds past step 26, the H1β-O-Fe angle further decreases to ~76° to allow maximal overlap of the oxygen lone pair with the hydrogen-bound pro-S hydroxyl group. At step 34 (~ 32 fs), the hydrogen transfer to the hydroxy-bound Fe is complete, forming the aqua-bound complex. From this point, the

H1β-O-Fe angle begins to increase to 109°. Pro-S hydrogen transfer is followed by C10-

C19 bond fission, releasing formic acid with simultaneous aromatization of the A-ring.

The hydrogen transfer from the gem-diol group is very interesting since it is demonstrative of an alternative reaction channel that can be taken by the hydroxy bound radical intermediate in P450 enzymes other than rebound to give the hydroxylated product. In order to characterize this as a hydrogen atom or proton transfer mechanism, we studied select time points along the preferred doublet trajectory with

B3LYP/(TZVP,SV(P)) single-point energy computations. The structures and critical geometric properties of these species are provided in the Appendix B. Due to program limitations of Turbomole, Mulliken atomic charges are used. While Mulliken analyses have often been criticized, the qualitative trends identified with these analyses were verified by NPA analysis in Gaussian by recomputation of the electronic wave function

(see Appendix B). Figure 3.15 shows the net Mulliken atomic charge of the substrate at these select time points. From the beginning of the simulation through step 19 (~ 17 fs), the net atomic charge remains small consistent with a bound substrate radical. The radical

150

nature of step 19 is confirmed with a substrate spin density of 0.7 e. As the simulation proceeds past step 23 (~21 fs), a pronounced accumulation of positive charge occurs on the steroid substrate and reaches a maximum at step 31 (~ 29 fs), which immediately precedes the hydrogen transfer. With the increase in positive charge on the substrate over this time frame, there is a simultaneous loss of spin density on the substrate. Analysis of the singly-occupied molecular orbital of the selected time steps prior to charge accumulation show that the SOMO is localized exclusively on the substrate. Time steps beyond the charge accumulation are characterized by a SOMO of mixed sulfur/porphyrin

π character. (Figure 3.16)

Taken together, the ab initio molecular dynamics results confirm deformylation to formic acid occurs in two distinct steps. The first is a classical hydrogen-atom abstraction, resulting in formation of the steroid substrate radical. An electron transfer follows from the substrate to the hydroxy radical intermediate. Therefore, the following deformylation step can be most accurately described as a proton transfer from the substrate cation to the hydroxide-bound P450 intermediate. These steps are outlined in Figure 3.17.

151

(a) Step 19 158β -0.186 eV (b) Step 19 159β -0.114 eV

(c) Step 31 158β -0.158 eV (d) Step 31 159β -0.132 eV

Figure 3.16: Highest-occupied (158) and lowest unoccupied (159) β spin orbitals of the

2 3 ETS MD steps 19 and 31. A contour value of 0.03 e/B is used.

152

We have considered that the apparent electron transfer arises as a result of a substrate cation and substrate radical state crossing as recently described by Kumar et al. [207] to delineate whether this phenomenon occurs in the processes studied here, time-dependent

DFT (TD-DFT) computations at the B3LYP/(TZVP,SV(P)) level have been employed to study the low-lying excited states of the 2ETS MD steps 19 and 31, which are akin to a substrate radical and cation, respectively. The step 19 and step 31 structures have low- lying excited states (0.38 eV and 0.18 eV) whose major contributing configuration is characterized by a transition from the highest occupied (158) and lowest unoccupied

(159) β spin orbitals (Figure 3.16). In the step 19 structure, the lowest excitation is a transition between an antibonding π* orbital of the substrate diene system and the mixed porphyrin π/sulfur p orbitals, and vice versa for the step 31 structure. As an approximation to the vertical excited state wave functions, the occupations of the spin orbitals were switched and the populations recomputed. The substrate Mulliken charges and spin densities for the approximate excited state of the step 19 were +0.8 and –0.1 e, respectively, consistent with a cationic excited state. In contrast, the values for the approximate excited state of the step 31 structure were –0.3 and –0.9 e, respectively, consistent with a substrate radical. These results suggest that in the early course of the

MD simulation, a substrate radical state is preferred energetically and has a low-lying cationic excited state. As the simulation proceeds, the MD propagates through the state crossing, giving energetic preference to the substrate cation surface and initiating the proton transfer.

153

1β−H atom H Cys Cys e- transfer Cys H S FeIV O H O abstraction S FeIV O H O S FeIII O H O OH H OH OH H H H

HO HO HO

O Cys H S FeIII O OH H H

HO

Figure 3.17: 1β-Hydrogen atom abstraction-electron transfer mechanism for aromatization of the steroid A-ring in the final catalytic step of cytochrome P450 aromatase.

Simulations initiated with the 2,4OTS geometries confirm that the decarbonylation channel proceeds toward the aromatized product. Following the 1β-hydrogen atom abstraction, the H1β-O-Fe angle equilibrates to nearly 109° and remains near this value until decarbonylation occurs at step 48 (~45 fs). Prior to decarbonylation, the simulation is dominated by rotation about the Fe-O bond. The 147° rotation orients the hydrogen atom away from the formyl group and repositions the oxygen lone pair for hydrogen transfer from the formyl group. Examination of the relative energetics of specific time points in the vicinity of the actual decarbonylation corroborates the previous PES calculation that the process is barrierless on the doublet surface. B3LYP/(TZVP,SV(P)) single-point energy computations of select geometries along the doublet trajectory confirm a similar reaction process to the hydrated analog. The charge on the steroid substrate remains low throughout the first 25 steps of MD simulation consistent with initial hydrogen-atom abstraction. As the simulation proceeds past this point, the positive charge on the substrate accumulates to a maximum of +0.60 e immediately preceding the 154

decarbonylation step to generate carbon monoxide. Therefore, an electron transfer followed by formyl deprotonation is operative in this process. Although the decarbonylation mechanism is not functional in the aromatase enzyme, the information gleaned here in combination with future work could contribute to a predictive paradigm for aldehyde metabolism by cytochrome P450.

3.4 CONCLUSIONS

We have presented a B3LYP density functional theory study of historically accepted and also novel mechanisms for the aromatization/deformylation sequence catalyzed by aromatase. Preliminary studies of peroxo hemiacetal adduct model systems indicate that

1β-hydrogen atom removal by the proximal oxygen of the iron peroxo species requires a high-energy barrier and does not initiate fragmentation to the experimentally observed products.

Alternative mechanistic processes initiated by 1β-hydrogen atom abstraction by the widely-accepted ultimate P450 oxidant, Compound I, were considered. Through these studies, a novel mechanism for the final catalytic step was delineated. A strikingly low barrier height (< 7 kcal/mol) for the 1β-hydrogen-atom abstraction was computed for steroid models containing the 2,3-enol moiety and computations using the keto tautomers confirmed the low barrier results from the enolized species ability to delocalize the developing radical. Noting that the deformylation and dehydrogenation of the keto species occurs with similar barriers computed for other known P450 substrates such as

155

camphor, we can speculate the mechanism described here may be operative in other P450 enzymes where demethylation and dehydrogenation occur with identical oxygen and cofactor requirements.

These studies do not support the dehydration of the 19-gem-diol prior to the final catalytic step. Results with model steroids containing the 19-aldehyde are subject to decarbonylation, to yield carbon monoxide, a process that is not consistent with the experimentally observed formic acid product. The characteristics of the 2,4ETS reaction vectors and their trajectory for decay to products without subsequent barrier suggest that the second hydrogen removal from the gem-diol occurs in a non-synchronous, concerted manner. Ab initio molecular dynamics confirmed that both deformylation and decarbonylation occur in two distinct steps. However, removal of the second hydrogen can be more adequately described as a deprotonation of the substrate cation by a ferric hydroxy P450 intermediate formed from a preceeding electron transfer event as depicted in Figure 3.17.

3.5 REFERENCES

152. Ryan, K. J. Conversion of androstenedione to estrone by placental microsomes.

Biochim. Biophys. Acta 1958, 27, 658-659.

153. Brodie, A. M. H. ISI Atlas of Science: Pharmacology 1987, 266.

154. Baum, M.; Budzar, A. U.; Cuzik, J.; Forbes, J.; Houghton, J. H.; Klijn, J. G.;

Sahmoud, T. Anastrozole alone or in combination with tamoxifen versus

tamoxifen alone for adjuvant treatment of postmenopausal women with early 156

breast cancer: first results of the ATAC randomised trial. Lancet 2002, 359,

2131-2139.

155.Bonneterre, J.; Buzdar, A.; Nabholtz, J. M.; Robertson J. F.; Thurlimann, B.; von

Euler, M.; Sahmoud, T.; Webster, A.; Steinberg, M. Anastrozole is superior to

tamoxifen as first-line therapy in hormone receptor positive advanced breast

carcinoma. Cancer 2001, 92, 2247-2258.

156. Brueggemeier R.W. Aromatase inhibitors in breast cancer therapy. Expert Rev.

Anticancer Ther. 2002, 2, 181-191.

157. Goto, J.; Fishman, J. Participation of a nonenzymatic transformation in the

biosynthesis of estrogens from androgens. Science 1977, 195, 80-81.

158. Fishman, J.; Raju, M. S. Mechanism of estrogen biosynthesis. Stereochemistry

of C-1 hydrogen elimination in the aromatization of 2 beta-hydroxy-19-

oxoandrostenedione. J. Biol. Chem. 1981, 256, 4472-4477.

159. Hahn, E. F.; Fishman, J. Immunological probe of estrogen biosynthesis.

Evidence for the 2 beta-hydroxylative pathway in aromatization of androgens.

J. Biol. Chem. 1984, 259, 1689-1694.

160. Covey, D. F.; Hood, W. R.; Beusen, D. D.; Carrell, H. L. Hydroperoxides as

inactivators of aromatase: 10 beta-hydroperoxy-4-estrene-3,17-dione, crystal

structure and inactivation characteristics. Biochemistry, 1984, 23, 5398-5406

161. Townsley, J. D.; Possanza, G. J.; Brodie, H. J. Fed. Proc., Fed. Am. Soc. Exp.

Biol. 1966, 25, 282 (Abstract 520).

157

162. Cole, P. A.; Robinson, C. H. Synthesis and study of a methoxy hydroperoxide

androstenedione derivative; analog of a potential aromatase intermediate. J.

Chem. Soc., Chem. Commun. 1986, 1651-1653.

163. Cole, P. A.; Robinson, C. H. Synthesis of and reactivity studies with 19-

peroxide-androstenedione derivatives: analogs of a proposed aromatase

intermediate. J. Chem. Soc., Perkin Trans. 1. 1990, 2119-2125.

164. Cole, P. A.; Robinson, C. H. Peroxide model reaction for placental aromatase.

J. Am. Chem. Soc. 1988, 110, 1284-1285.

165. Cole, P. A.; Bean, J. M.; Robinson, C. H. Conversion of a 3-desoxysteroid to 3-

desoxyestrogen by human placental aromatase. Proc. Natl. Acad. Sci. U. S. A.

1990, 87, 2999-3003.

166. Cole, P. A.; Robinson, C. H. Conversion of 19-oxo[2 beta-2H]androgens into

oestrogens by human placental aromatase. An unexpected stereochemical

outcome. Biochem. J. 1990, 268, 553-561

167. Schöneboom, J. C.; Lin, H.; Reuter, N.; Thiel, W.; Cohen, S.; Ogliaro, F.;

Shaik, S. The elusive oxidant species of cytochrome P450 enzymes:

characterization by combined quantum mechanical/molecular mechanical

(QM/MM) calculations. J. Am. Chem. Soc. 2002, 124, 8142-8151.

168. Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-

energy formula into a functional of the electron density Phys. Rev. B 1988, 37,

785-789.

158

169. Vosko, S. H.; Wilk, L.; Nusair, M. Accurate spin-dependent electron liquid

correlation energies for local spin density calculations: a critical analysis. Can.

J. Phys. 1980, 58, 1200-1211.

170. Gaussian 03, Revision B.04, Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.;

Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.;

Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi,

J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G.

A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.;

Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.;

Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts,

R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.;

Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.;

Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M.

C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J.

B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov,

B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D.

J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe,

M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople,

J. A. Gaussian, Inc.: Pittsburgh, PA, 2003.

171. Binkley, J. S.; Pople, J. A.; Hehre, W. J. Self-consistent molecular orbital

methods. 21. Small split-valence basis sets for first-row elements. J. Am. Chem.

Soc. 1980, 102, 939-947.

159

172. Gordon, M. S.; Binkley, J. S.; Pietro, W. J.; Hehre, W. J. Self-consistent

molecular-orbital methods. 22. Small split-valence basis sets for second-row

elements. J. Am. Chem. Soc. 1983, 104, 2797-2803.

173. Pietro, W. J.; Francl, M. M.; Hehre, W. J.; Defrees, D. J. Pople, J. A.; Binkley,

J. S. Self-consistent molecular orbital methods. 24. Supplemented small split-

valence basis sets for second-row elements. J. Am. Chem. Soc. 1982, 104,

5039-5048.

174. Wachters, A. J. H. Gaussian Basis Set for Molecular Wavefunctions Containing

Third-Row Atoms. J. Chem. Phys. 1970, 52, 1033-1036.

175. Bauschlicher, C. W.; Langhoff, S. R.; Barnes, L. A. Theoretical studies of the

first- and second-row transition-metal methyls and their positive ions. J. Chem.

Phys. 1989, 91, 2399-2411.

176. Basis sets were obtained from the Extensible Computational Chemistry

Environment Basis Set Database, Version 02/25/04, as developed and

distributed by the Molecular Science Computing Facility, Environmental and

Molecular Sciences Laboratory which is part of the Pacific Northwest

Laboratory, P.O. Box 999, Richland, Washington 99352, USA, and funded by

the U.S. Department of Energy. The Pacific Northwest Laboratory is a multi-

program laboratory operated by Battelle Memorial Institute for the U.S.

Department of Energy under contract DE-AC06-76RLO 1830. Contact David

Feller or Karen Schuchardt for further information.

160

177. Krishnan, R.; Binkley, J. S.; Seeger, R. Pople, J. A. Self-consistent molecular

orbital methods. XX. A basis set for correlated wave functions. J. Chem. Phys.

1980, 72, 650-654.

178. Clark, T.; Chandrasekhar, J.; Schleyer, P. v. R. J. Comput. Chem. 1983, 4, 294.

179. Ogliaro, F.; Cohen, S.; de Visser, S.; Shaik, S. Medium Polarization and

Hydrogen Bonding Effects on Compound I of Cytochrome P450: What Kind of

a Radical Is It Really? J. Am. Chem. Soc. 2000, 122, 12892-12893.

180. Ogliaro, F.; de Visser, S.; Cohen, S.; Sharma, P. K.; Shaik, S. Searching for the

Second Oxidant in the Catalytic Cycle of Cytochrome P450: A Theoretical

Investigation of the Iron(III)-Hydroperoxo Species and Its Epoxidation

Pathways J. Am. Chem. Soc. 2002, 124, 2806-2817.

181. de Visser, S. P.; Ogliaro, F.; Sharma, P. K.; Shaik, S. What Factors Affect the

Regioselectivity of Oxidation by Cytochrome P450? A DFT Study of Allylic

Hydroxylation and Double Bond Epoxidation in a Model Reaction. J. Am.

Chem. Soc. 2002, 124, 11809-11826.

182. de Visser, S. P.; Shaik, S. A Proton-Shuttle Mechanism Mediated by the

Porphyrin in Benzene Hydroxylation by Cytochrome P450 Enzymes J. Am.

Chem. Soc. 2003, 125, 7413-7424.

183. Sharma, P. K.; de Visser, S. P.; Shaik, S. Can a Single Oxidant with Two Spin

States Masquerade as Two Different Oxidants? A Study of the Sulfoxidation

Mechanism by Cytochrome P450. J. Am. Chem. Soc. 2003, 125, 8698-8699.

161

184. de Visser, S. P.; Shaik, S.; Sharma, P. K.; Kumar, D.; Thiel, W. Active Species

of Horseradish Peroxidase (HRP) and Cytochrome P450: Two Electronic

Chameleons. J. Am. Chem. Soc. 2003, 125, 15779-1578

185. Ogliaro, F.; de Visser, S. P.; Cohen, S.; Kaneti, J.; Shaik, S. The experimentally

elusive oxidant of cytochrome P450: a theoretical "trapping" defining more

closely the "real" species. ChemBioChem. 2001, 11, 848-851.

186. Shaik, S.; Cohen, S.; de Visser, S. P.; Sharma, P. K.; Kumar, D.; Kozuch, S.;

Ogliaro, F.; Danovich, D. The “Rebound Controversy: An Overview and

Theoretical Modeling of the Rebound Step in Mono-Oxygenations by

Cytochrome P450. Eur. J. Inorg. Chem. 2004, 207-226.

187. Gonzalez, C.; Schlegel, H. B. An improved algorithm for reaction path

following. J. Chem. Phys. 1989, 90, 2154-2161.

188. Gonzalez, C.; Schlegel, H. B. Reaction path following in mass-weighted

internal coordinates. J. Phys. Chem. 1990, 94, 5523-5527.

189. A description of the ab initio molecular dynamics code implemented in the

FROG module of Turbomole 5.6 can be found in: Elliott, S. D.; Ahlrichs, R.;

Hampe, O.; Kappes; M. M. Auto-ionised products from the reaction of sodium

clusters with dioxygen: Theory and experiment. Phys. Chem. Chem. Phys.

2000, 2, 3415-3424.

190. Verlet, L. Computer "Experiments" on Classical Fluids. I. Thermodynamical

Properties of Lennard-Jones Molecules. Phys. Rev. 1967, 159, 98-103.

162

191. We were unable to perform a rigid PES scan of the distal oxygen-C19

coordinate for the initial attack of C19 by the model intermediate due to

problems with self-consistent field equation convergence.

192. Du, P.; Loew, G. H. Role of axial ligand in the electronic structure of model

compound I complexes. Int. J. Quantum Chem. 1992, 44, 251-261.

193. Antony, J.; Grodzicki, M.; Trautwein, A. X. Local Density Functional Study of

Oxoiron(IV) Porphyrin Complexes and Their One-Electron Oxidized

Derivatives. Axial Ligand Effects. J. Phys. Chem. A. 1997, 101, 2692-2701.

194. Green, M. T. Evidence for Sulfur-Based Radicals in Thiolate Compound I

Intermediates. J. Am. Chem. Soc. 1999, 121, 7939-7940.

195. Ogliaro F.; de Visser S. P.; Groves J. T.; Shaik S. Chameleon States: High-

Valent Metal-Oxo Species of Cytochrome P450 and Its Ruthenium Analogue.

Angew. Chem. Int. Ed. Engl. 2001, 40, 3503.

196. Yoshizawa, K.; Shiota, Y.; Kagawa, Y. Energetics for the Oxygen Rebound

Mechanism of Alkane Hydroxylation by the Iron-Oxo Species of Cytochrome

P450. Bull. Chem. Soc. Jpn. 2000, 73, 2669-2673.

197. Yoshizawa, K.; Kagawa, Y; Shiota, Y. Kinetic Isotope Effects in a C-H Bond

Dissociation by the Iron-Oxo Species of Cytochrome P450. J. Phys. Chem. B

2000, 104, 12365-12370.

163

198. Alexander, K.; Akhtar, M.; Boar, R. B.; McGhie, J. F.; Barton, D. H. R.

Pathway for the removal of C-32 in cholesterol biosynthesis. J. Chem. Soc.,

Chem. Commun. 1972 383-385.

199. Mitropoulos, K. A.; Gibbons, G. F.; Reeves, E. A. Lanosterol 14alpha-

demethylase. Similarity of the enzyme system from yeast and rat liver. Steroids

1976, 27, 821-829.

200. Alexander, K. T. W.; Akhtar, M.; Boar, R. B.; McGhie, J. F.; Barton, D. H.

R.Removal of the 32-carbon atom as formic acid in cholesterol biosynthesis. J.

Chem. Soc., Chem. Commun. 1971, 1479-1481.

201. Akhtar, M.; Freeman, C. W.; Wilton, D. C.; Boar, R. B.; Copsey, D. B.The

pathway for the removal of the 15α-methylgroup of lanosterol. The role of

lanost-8-ene-3β,32-diol in cholesterol biosynthesis. Bioorg. Chem. 1977, 6,

473.

202. Akhtar, M.; Alexander, K.; Boar, R. B.; McGhie, J. F.; Barton, D. H. R.

Chemical and enzymic studies on the characterization of intermediates during

the removal of the 14alpha-methyl group in cholesterol biosynthesis. The use of

32-functionalized lanostane derivatives. Biochem. J. 1978, 169, 449-463.

203. Guallar, V.; Baik, M.-H.; Lippard, S. J.; Friesner, R. A. Peripheral heme

substituents control the hydrogen-atom abstraction chemistry in cytochromes

P450. Proc. Natl. Acad. Sci. USA. 2003, 100, 6998-7002.

164

204. Guallar, V.; Friesner, R. A. Cytochrome P450CAM Enzymatic Catalysis Cycle:

A Quantum Mechanics/Molecular Mechanics Study. J. Am. Chem. Soc. 2004,

126, 8501-8508.

205. Yoshizawa, K.; Kamachi, T.; Shiota, Y. A Theoretical Study of the Dynamic

Behavior of Alkane Hydroxylation by a Compound I Model of Cytochrome

P450. J. Am. Chem. Soc. 2001, 123, 9806-9816.

206. Dawson, J. H.; Sono, M. Cytochrome P-450 and chloroperoxidase: thiolate-

ligated heme enzymes. Spectroscopic determination of their active-site

structures and mechanistic implications of thiolate ligation. Chem. Rev. 1987,

87, 1255-1276.

207. Kumar, D.; de Visser, S. P.; Shaik, S. Oxygen Economy of Cytochrome P450:

What Is the Origin of the Mixed Functionality as a Dehydrogenase-Oxidase

Enzyme Compared with Its Normal Function? J. Am. Chem. Soc. 2004, 126,

5072-5073.

165

CHAPTER 4

PYRIDINE –CONTAINING ISOFLAVONE INHIBITORS OF AROMATASE

4.1 INTRODUCTION

Flavonoids are a diverse group of plant-derived chemicals that are produced by various higher plants, which can therefore be found in numerous food sources such as fruits, vegetables, legumes, and whole grains. [208] Compounds in this class have shown a wide variety of biological activities such as antiviral, antiinflammatory, antibacterial, antifungal, and anticancer activities. [209-213] In particular, due to their structural and functional similarities to endogenous estrogens, flavonoids have attracted considerable interest as alternative estrogens, termed phytoestrogens, and extensively studied for their potential role in many estrogen-dependent diseases including breast cancer. In fact, numerous flavonoids have shown interesting pharmacological activities in breast cancer biology, including binding affinities for estrogen receptors, [220] antiproliferative activities, [218,219] and inhibitory activities against aromatase enzyme. [62]

Inspired by the versatility of genistein, our research group has been investigating exploitation of the isoflavone nucleus as a potential privileged structure [219] for the development of new therapeutic agents for hormone-dependent breast cancer. [12-14]

As part of this effort, we became interested in development of new isoflavone derivatives

166

targeting aromatase (CYP19), a cytochrome P450 hemoprotein that is responsible for

estrogen biosynthesis by conversion of androgens into estrogens. [8] Aromatase has been

a particularly attractive target for inhibition in the treatment of hormone-dependent breast

cancer. [12-14]

OH O OH OCH OH O OH O 3 C A HO O B HO O HO O

Genistein Biochanin A Chrysin

(apparent Ki = 123 ± 8 µM) (apparent Ki = 12 ± 5 µM) (apparent Ki = 2.6 ± 0.1 µM)

Figure 4.1: Chemical structures and aromatase inhibitory activities of two isoflavones

(genistein, biochanin A) and one flavone (chrysin). [72]

4.2 INHIBITOR DESIGN

Most nonsteroidal aromatase inhibitors developed to date are reversible inhibitors that

compete with the natural substrates, androstenedione and testosterone. Strong evidence

for the binding of flavones to the active site of aromatase was obtained by difference

spectral absorption studies by Kellis and Vickery. [222] They reported α-napthoflavone could displace androstenedione from the aromatase active site and induce a spectrum consistent with the low-spin state of iron. Ibrahim and Abul-Hajj later reported reduction of the flavone 4-keto group was detrimental to aromatase inhibition by these compounds. [64] Based on data obtained from site directed mutagenesis studies and 167

ligand docking into a homology model of the aromatase protein, a binding orientation

was predicted in which the A and C rings of the flavone (Figure 4.1) mimic the C and D

rings of the steroid substrate, respectively. Therefore, the 2-phenyl substituent is oriented

in a region similar to that occupied by the A ring of the steroid. This analysis places the

flavone 4-keto functionality in the same position as the steroid 19-angular methyl group

with respect to the heme iron. [72] Deductions based on the modeling results obtained by

other research groups, assuming the isoflavones bind to the aromatase active site in an

orientation reminiscent of the flavones, we hypothesize the translation of the phenyl group from the 2 to 3 position of the flavone core may diminish the ability of the 4-keto functionality to adequately coordinate to the heme iron and introduce unfavorable steric interactions within the enzyme active site.

Competitive nonsteroidal aromatase inhibitors usually possess a heteroatom that interferes with hydroxylation of steroids by coordinating with the heme iron of aromatase. Although several heteroatoms, such as sulfur, oxygen, and nitrogen, are known to show abilities to bind to heme iron, the majority of compounds in this class

possess a nitrogen-containing heterocyclic moiety such as imidazole, triazole, pyrimidine,

or pyridine. [223] We have envisioned that introduction of the appropriate heme-

coordinating functional group to the 2-position of the isoflavone could result in a ligand

whose binding mode is now biased to mimic the predicted binding mode of the flavones.

This orientation places the 3-phenyl functionality in the region occupied by the steroid A-ring, with the 4-keto functionality now pointing away from the heme.

168

On the basis of this rationale, we designed a library of 2,4’,7-trisubstituted isoflavones as

shown in Figure 4.2. In this study, we have focused on the biological evaluation of

isoflavones containing a (pyridylmethyl)thio moiety at the 2-position, which would be

easily prepared via the phase transfer catalysis procedure (Figure 4.2) developed in our laboratory. [224-226] We were also interested in the preparation of isoflavones containing a non-nitrogenated moiety at the same position such as benzylthio or allylthio group for comparison. Previous studies have indicated that the presence of hydroxyl group at 4’-position seems to be unfavorable for aromatase inhibitory activity. For example, GEN is 10-fold less potent than its 4’-methoxy analog, biochanin A (BCA,

Figure 4.1). [72] Therefore, we were interested in nonpolar alternatives to the 4’-OH such as methoxy, methyl, and hydrogen. In fact, it was proposed that isoflavones without the

4’-hydroxyl group might more efficiently inhibit aromatase. [227] With regard to the 7- position, a 7-hydroxyl group appears to be the most promising, as its presence is almost ubiquitous among biologically active natural isoflavones. However, 7-methoxy analogs might also be interesting to investigate the consequences of masking the hydroxyl group at the position.

169

R R O 1 O 1 MeOH or BnOH

Ph3P, DIAD o HO OH THF, 0 C, 0.5 h R2 OH

3a: R1 = H 4a: R1 = H, R2 = OMe 3b: R1 = Me 4b: R1 = Me, R2 = OMe 3c: R1 = OMe 4c: R1 = OMe, R2 = OMe 4d: R1 = H, R2 = OBn 4e: R1 = OMe, R2 = OBn

R O 1 n-Bu4N·HSO4, 50% aq. NaOH

CS2, R3X, H2O, THF rt, overnight R2 OSR3

5a: R1 = OMe, R2 = OBn, R3 = allyl 5b: R1 = OMe, R2 = OBn, R3 = Bn 5c: R1 = H, R2 = OMe, R3 = allyl 5d: R1 = H, R2 = OMe, R3 = Bn 5e: R1 = H, R2 = OMe, R3 = (4-pyridyl)methanethio 5f: R1 = H, R2 = OMe, R3 = (3-pyridyl)methanethio 5g: R1 = H, R2 = OMe, R3 = (2-pyridyl)methanethio 5h: R1 = Me, R2 = OMe, R3 = (4-pyridyl)methanethio 5i: R1 = OMe, R2 = OMe, R3 = (4-pyridyl)methanethio 5j: R1 = H, R2 = OBn, R3 = (4-pyridyl)methanethio 5k: R1 = OMe, R2 = OBn, R3 = (4-pyridyl)methanethio

Figure 4.2: Synthesis of isoflavones 5a-k .[224-226]

170

R R O 1 O 1 BBr3, CH2Cl2, rt, overnight

or BF3·OEt2, Me2S, CH2Cl2 R2 OSR3 (for 4g from 9l) HO OSR3

5a: R1 = OMe, R2 = OBn, R3 = allyl 6a: R1 = OH, R3 = allyl 5b: R1 = OMe, R2 = OBn, R3 = Bn 6b: R1 = OH, R3 = Bn 5d: R1 = H, R2 = OMe, R3 = Bn 6c: R1 = H, R3 = Bn 5e: R1 = H, R2 = OMe, R3 = (4-pyridyl)methanethio 6d: R1 = H, R3 = (4-pyridyl)methanethio 5f: R1 = H, R2 = OMe, R3 = (3-pyridyl)methanethio 6e: R1 = H, R3 = (3-pyridyl)methanethio 5k: R1 = OMe, R2 = OBn, R3 = (4-pyridyl)methanethio 6f: R1 = OH, R3 = (4-pyridyl)methanethio 6g: R1 = OMe, R3 = (4-pyridyl)methanethio

Figure 4.3: Synthesis of isoflavones 6a-g. [224-226]

The aromatase assay was performed according to the modified method of the procedure

previously reported by our laboratory, in which human placental microsomes were used

as the aromatase source. [224] IC50 values of the compounds were determined from dose- response curves (Figure 4.4 for compounds 5e-k and Figure 4.5 for compounds 6e-g) and

are listed in Table 4.1. The IC50 values of (±)-aminoglutethimide (AG) and BCA were

also determined in our assay system for comparison. In order to examine the mode of

aromatase inhibition and more accurately distinguish relative potencies, kinetic studies for selected compounds (5e, 5h-j, 6f, and 6g) were also performed, and their apparent Ki values are listed in Table 4.2 with apparent Km values and Ki/Km ratios. Lineweaver-Burk

plots are shown in Figures 4.6-4.15.

171

4.3 RESULTS AND DISCUSSION

In screening assay studies, analogs lacking a pyridyl group (5a-d and 6a-c) showed no

significant inhibition in enzyme assays containing up to 100 µM of the compound.

Therefore, dose-response studies of these compounds were not pursued. On the other

hand, most of the pyridine-containing analogs showed inhibitory activities, suggesting

that the functionality might play a role in inhibiting aromatase. It is obvious that the

position of the nitrogen atom in the pyridyl moiety is important for inhibitory activity. 4’-

Pyridyl analog 5e is ~ 6-fold more potent that the 3’-pyridyl analog 5f, whereas the 2’-

pyridyl analog 5g has no observable inhibitory activity in our assay system. This trend in activity is echoed in the 7-hydroxy analogs of this series; the 4’-pyridyl analog 6d is also

~ 6-fold more potent than 3’-pyridyl analog 6e. These results suggest that the position of the pyridyl nitrogen influences inhibitory activity. One possible explanation is that the pyridyl nitrogen may coordinate the active site heme iron. A method to probe this potential interaction is to obtain difference spectra with the active compounds and the enzyme preparation. Interestingly, we are unable to obtain spectra supportive of Type II binding with human placental microsomes or with CYP19 baculovirus-transfected insect cell microsomes (B-D Biosciences, Woburn, MA).

The 7-hydroxy analogs 6d and 6e exhibit greater inhibitory activity than the corresponding 7-methoxy analogs 5e and 5f, respectively. This result suggests the presence of a hydrogen bond donor at the C- 7 position may elicit a favorable interaction with the enzyme. With regard to the 4’ substituent, the presence of a hydrogen atom, as in

5e, at this position is ~2-fold more favorable in achieving aromatase enzyme activity 172

compared to the 4’-methyl and 4’–methoxy substituted analogs 5h and 5i. The IC50 values of 5h and 5i are quite similar, and it is difficult to cast judgment about the superiority of one compound over another based on dose-response data alone. Toward this goal, we conducted enzyme kinetic studies (Table 4.2) of compounds 5e, 5h, and 5i to verify the IC50 results and to more accurately discern the relative activities of these analogs. These compounds demonstrated typical competitive-type inhibition with androstenedione in Lineweaver-Burk plots (Figures 4.6-4.8). As computed from the relative potencies (Ki/Km), 4’-methyl analog 5h is 37% more active than the 4’methoxy

analog, 5i. Interestingly, this structure-activity relationship does not translate to the 7-

hydroxy analogs. In this case, the 4’-methoxy analog is more potent than the 4’-hydrogen

analog 6d. In addition, the potency of 4’,7-dihydroxy analog 6f was comparable to the 7-

hydroxy-4’-methoxy analog 6g, but ~ 2-fold greater than that of the 7-hydroxy-4’-

hydrogen analog 6d. This result suggests the presence of a proton acceptor in the 4’

position in the hydroxy analogs may be beneficial for aromatase activity.

To our surprise, 7-benzyloxy analog 5k, originally prepared as a precursor of 6g,

displayed promising inhibitory activity. This observation was unexpected because the

compound was assumed to be too bulky to fit within the enzyme active site. This led us to

synthesize the 7-benzyloxy-4’-hydrogen analog 5j with the hypothesis that 7-benzyloxy

analogs would display similar structure- activity relationships at the 4’ position as

observed in the 7-methoxy series. Indeed, the elimination of the 4’-methoxy resulting in

5j, proved effective in producing a compound with enhanced aromatase inhibitory 173 activity. In addition, 5j exhibits ~ 8-fold improvement compared to the corresponding 7- methoxy analog 5e. Finally, we obtained IC50 values in our assay system for the isoflavone lead compound, BCA, and the extensively studied aromatase inhibitor AG.

The most potent compound in this series in terms of data obtained from dose-response studies, 5j, demonstrated 162- and 13- fold improvements in aromatase inhibition over these compounds, respectively.

174

120

100

80

60

40

% Control Activity Control % 20

0 1 2 3 4 5

log [nM]

Figure 4.4: Aromatase inhibitory activities of isoflavones 5e ( ), 5f (■), 5g (▲), 5h

(▼), 5i (♦), 5j (●), and 5k (□). Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

175

120

100

80

60

40

% Control Activity Control % 20

0 1 2 3 4 5

log [nM]

Figure 4.5: Aromatase inhibitory activities of isoflavones 6d (■), 6e (▲), 6f (▼), and 6g

(♦). Error bars represent standard error (n = 3), and the data were analyzed by the

nonlinear regression analysis method.

176

R O 1

R2 O SR3

a compound R1 R2 R3 IC50 (µM) Log IC50 (nM) (±S.E.) 5a OMe OBn allyl >100  5b OMe OBn benzyl >100  5c H OMe allyl >100  5d H OMe benzyl >100  5e H OMe (4’-pyridyl)methyl 1.6 3.21 ± 0.11 5f H OMe (3’-pyridyl)methyl 9.2 3.96 ± 0.16 5g H OMe (2’-pyridyl)methyl >100  5h Me OMe (4’-pyridyl)methyl 3.0 3.48 ± 0.05 5i OMe OMe (4’-pyridyl)methyl 3.1 3.49 ± 0.02 5j H OBn (4’-pyridyl)methyl 0.21 2.33 ± 0.03 5k OMe OBn (4’-pyridyl)methyl 0.53 2.72 ± 0.11 6a OH OH allyl N.D.b  6b OH OH benzyl N.D.b  6c H OH benzyl N.D.b  6d H OH (4’-pyridyl)methyl 0.61 2.79 ± 0.11 6e H OH (3’-pyridyl)methyl 3.6 3.56 ± 0.06 6f OH OH (4’-pyridyl)methyl 0.28 2.44 ± 0.07 6g OMe OH (4’-pyridyl)methyl 0.22 2.34 ± 0.04 AG 2.8 3.45 ± 0.05 BCA 34 4.53 ± 0.06

a Table 4.1. Aromatase inhibitory activities of isoflavones 5a-k and 6a-g. IC50 values were calculated by a nonlinear regression analysis (GraphPad Prism). Each dose- response curve contained ten concentrations, each in triplicate. b Not determined.

177

0.00100

0.00075

0.00050 formed/min

nmoles product 0.00025

0.00000 0 100 200 300 400 500

nM Androstenedione 25000

20000

15000

10000 formed/min]

1/[nmol product 5000

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 4.6. Lineweaver-Burk plot of aromatase inhibition by compound 5e. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 2000 nM (▲), 5000 nM

(▼), and 10000 nM (♦) inhibitor. Each point represents the average of three

determinations ± standard error.

178

0.00100

0.00075

0.00050 formed/min

nmoles product 0.00025

0.00000 0 100 200 300 400 500

nM Androstenedione

7500

5000

formed/min] 2500 1/[nmol product

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 4.7. Lineweaver-Burk plot of aromatase inhibition by compound 5h. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 500 nM (▼),

and 2000 nM (♦) inhibitor. Each point represents the average of three

determinations ± standard error.

179

0.00100

0.00075

0.00050 formed/min

nmoles product 0.00025

0.00000 0 100 200 300 400 500

[nM Androstenedione]

6000

5000

4000

3000

formed/min] 2000

1/[nmoles product 1000

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 4.8: Lineweaver-Burk plot of aromatase inhibition by compound 5i. Various concentrations of androstenedione (50 to 500 nM) were incubated with microsomal enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 500 nM (▼), and 2000 nM (♦) inhibitor. Each point represents the average of three

determinations ± standard error.

180

0.00100

0.00075

0.00050 formed/min

nmoles product 0.00025

0.00000 0 100 200 300 400 500

[nM Androstenedione]

8000

6000

4000 formed/min]

1/[nmol product 2000

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 4.9: Lineweaver-Burk plot of aromatase inhibition by compound 5j. Various concentrations of androstenedione (50 to 500 nM) were incubated with microsomal enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 500 nM (▼), and 2000 nM (♦) inhibitor. Each point represents the average of three determinations ±

standard error.

181

0.004

0.003

0.002 formed/min

nmoles product 0.001

0.000 0 100 200 300

nM Androstenedione 15000

12500

10000

7500

formed/min] 5000 1/[nmol product 2500

0 0.000 0.025 0.050 0.075 0.100

1/[nM Androstenedione ]

Figure 4.10: Lineweaver-Burk plot of aromatase inhibition by compound 6f. Various

concentrations of androstenedione (10 to 300 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 500 nM (▼), and 2000 nM (♦) inhibitor. Each point represents the average of three determinations ±

standard error.

182

0.0050

0.0025 formed/min nmoles product

0.0000 0 100 200 300

nM Androstenedione

12500

10000

7500

5000 formed/min]

1/[nmoles product 2500

0 0.000 0.025 0.050 0.075 0.100

1/[nM Androstenedione]

Figure 4.11: Lineweaver-Burk plot of aromatase inhibition by compound 6g. Various concentrations of androstenedione (10 to 300 nM) were incubated with microsomal enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 500 nM (▼), and 2000 nM (♦) inhibitor. Each point represents the average of three determinations ± standard error.

183

0.0015

0.0010

formed/min 0.0005 nmoles product

0.0000 0 100 200 300 400 500

[nM Androstenedione]

7500

5000

formed/min] 2500 1/[nmol product

0 0.000 0.005 0.010 0.015 0.020 -1 [nM Androstenedione]

Figure 4.12: Lineweaver-Burk plot of aromatase inhibition by compound AG. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 500 nM (▲), 2000 nM (▼), and 5000 nM (♦) inhibitor. Each point represents the average of three

determinations ± standard error.

184

Apparent K (µM) Apparent K (µM) compound i m K /K (±S.E.)a (±S.E.)a i m 5e 0.90 ± 0.04 0.11 ± 0.06 8.18 5h 1.16 ± 0.07 0.12 ± 0.01 9.67 5i 1.69 ± 0.18 0.11 ± 0.01 15.4 5j 0.22 ± 0.02 0.13 ± 0.01 1.69 6f 0.31 ± 0.02 0.11 ± 0.07 2.82 6g 0.26 ± 0.02 0.10 ± 0.02 2.60 AG 1.41 ± 0.10 0.09 ± 0.01 15.7 BCA 12.0 ± 5b  

Table 4.2. Enzyme kinetic parameters for selected isoflavones and reference

a compounds. Apparent Km, apparent Ki, and S.E. values were calculated by weighted

regression analysis. [227] b See reference [72]

185

O O

Me2S, BF3OEt2, CH2Cl2

BnO O S HO O S N N

O

K2CO3, DMF

R X OOS R Y N

7a: R = phenylmethyl 7b: R = beta-napthyl 7c: R = alpha-napthyl 7d: R = 4'-biphenyl 7e: R = cyclohexyl 8a: R = p-nitrophenyl 8b: R = m-nitrophenyl 8c: R = o-nitrophenyl 9a: R = 2'-pyridyl 9b:R = o-methoxy 9c: R = p-methoxy 10a: R = p-bromo 10b: R = p-fluoro 10c: R = p-chloro

Figure 4.13: Modification of aryl group at 7-position of isoflavone.

186

120

100

80

60

40

% Control Activity Control % 20

0 1 2 3 4 5

log [nM]

Figure 4.14: Aromatase inhibitory activities of isoflavones 7a (■), 7b (▲), 7c (▼), 7d

(♦), and 7e (●). Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

187

O

RO O S N

a compound R IC50 (nM) LogIC50(nM)±S.E. 359 2.55 ± 0.04 7a

7b 90 1.96 ± 0.05

85 1.93 ± 0.08 7c

154 2.18 ± 0.06 7d

553 2.74 ± 0.08 7e

a Table 4.3: Aromatase inhibitory activities of isoflavones 7a-e. IC50 values were calculated by a nonlinear regression analysis (GraphPad Prism). Each dose-response curve contained ten concentrations, each in triplicate.

188

120

100

80

60

40

% Control Activity Control % 20

0 1 2 3 4 5

log [nM]

Figure 4.15: Aromatase inhibitory activities of isoflavones 8a (■), 8b (▲), and 8c. Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

O

O O S R N

a compound R IC50 (nM) LogIC50(nM)±S.E.

8a p-NO2 132 2.12 ± 0.08

8b m-NO2 113 2.06 ± 0.05

8c o-NO2 138 2.14 ± 0.03

a Table 4.4. Aromatase inhibitory activities of isoflavones 8a-c. IC50 values were calculated by a nonlinear regression analysis (GraphPad Prism). Each dose-response curve contained ten concentrations, each in triplicate.

189

120

100

80

60

40

% Control Activity Control % 20

0 0 1 2 3 4 5

log [nM]

Figure 4.16: Aromatase inhibitory activities of isoflavones 9a (■), 9b (▲), and 9c. Error

bars represent standard error (n = 3), and the data were analyzed by the nonlinear

regression analysis method.

O

O O S R X N

compound R X IC50 (nM) LogIC50(nM)±S.E. 9a H N 378.2 2.58 ± 0.06 9b o-OMe C 243 2.38 ± 0.04 9c p-OMe C 161 2.21 ± 0.02

a Table 4.5. Aromatase inhibitory activities of isoflavones 9a-c IC50 values were calculated by a nonlinear regression analysis (GraphPad Prism). Each dose-response curve contained ten concentrations, each in triplicate.

190

120

100

80

60

40

% Control Activity Control % 20

0 0 1 2 3 4 5

log [nM]

Figure 4.17: Aromatase inhibitory activities of isoflavones 10a (■), 10b (▲), and 10c.

Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

O

O O S R N

compound R IC50 (nM) LogIC50(nM)±S.E. 10a p-Br 213 2.33 ± 0.05 10b p-F 337 2.53 ± 0.05 10c p-Cl 233 2.37 ± 0.05

a Table 4.6. Aromatase inhibitory activities of isoflavones 10a-c. IC50 values were calculated by a nonlinear regression analysis (GraphPad Prism). Each dose-response curve contained ten concentrations, each in triplicate.

191

Enzyme kinetic studies were also conducted with the most potent compounds of this

series identified in the dose-response studies. Compounds 5j, 6f, and 6g demonstrated competititive-type inhibition with the androstenedione substrate (Figure 3, plots for 6f and 6g not shown). The computed relative inhibitory potency ratios reflected the same trend (6j > 6g > 6f) determined in the dose-response studies. As reflected by the relative potency, 5j is the most potent compound within this series and is more potent than AG.

This result is in agreement within the estimated experimental uncertainty of the result obtained from the dose-response studies. Compound 5j also demonstrates 50-fold enhancement in potency compared to the natural product lead, BCA. The present enzyme

kinetic results emphasize the importance of these structural modifications for optimization of aromatase inhibition by isoflavones.

Realization that introduction of a benzyl group at the 7-position of the 4’-

(pyridylmethylthio)isoflavones led us to explore further the structure-activity relationships in this region of the molecule. Extension of the hydrocarbon chain from benzyloxy to phenylethyloxy by introduction of a single methylene group (7a) decreased activity almost two-fold. Expansion of the aromatic entity to napthyl (7b,c) increased aromatase inhibitory activity greater than two-fold. In terms of IC50 values, it is not

possible to determine whether attachment at the α- or β- carbon of the naphthalene ring

will produce more desirable activity. Introductiion of an additional phenyl ring resulting in the biphenyl analog (7d) enhances aromatase inhibitory activity

approximately 25% compared to 3j. The cyclohexyl analog (7e) demonstrates a greater 192

than two-fold loss of activity, thus illustrating the importance of unsaturation in the 7-

substituent to achieve potent aromatase inhibitory activity. These biological activities of these compounds suggest the 7-functionality may access an aromatic pocket in or near the aromatase active site where π-π van der Waals interactions may potentiate activity.

Introduction of a nitro-group (8a-c) to the 7-benzyloxy functionality leads to an almost

two-fold enhancement in activity. This enhancement appears to independent of the

position of this substituent on the aromatic ring. The regional independence of this

electron-withdrawing substituent underscores the postulate that increased activity arises

from a π-π stacking interaction. Based on the IC50 values alone, it appears that

introduction of a ortho methoxy (9b) group to the 7-benzyloxy substituent does not

impart a significant change in inhibitory activity. However, introduction of this functional

group in the para position (9c) leads to a slight increase in activity. Whether this minor

enhancement in activity arises from the hydrophobic nature of this substituent or from is

hydrogen-bonding potential is unclear. The introduction of a nitrogen into the 2-position

(9a) of the 7-benzyloxy substituent leads to an almost two-fold decrease in activity.

Introduction of chlorine (10c) or bromine (10a) at the para position apparently has no

effect on the aromatase inhibitory activity compared to 5j. Interestingly, the p-fluoro

analog (10b) is approximately 50% less active. This effect may arise as a result from the

hydrogen-bonding potential of this atom. The loss of activity as a result of introduction of

a hydrogen-bond donor indicates the slight enhancement achieved with the para methoxy substituent is a hydrophobic effect.

193

4.4 CONCLUSIONS

To date, the isoflavone ring system has been considered an inappropriate scaffold for

development of aromatase inhibitors. This study has shown that aromatase inhibitory

activity can be achieved in the isoflavone nucleus by introducing functional groups with

the potential to coordinate the heme iron. The structure-activity relationships indicate that the binding modes of 7-protected analogs 5e-k might be different from those of 7- hydroxy analogs 6d-g. Hydrogen-bonding potential appears to be important in the 7- hydroxy analogs 6d-g for their aromatase inhibitory activity, whereas hydrophobic interactions appear to play a role in 7-protected analogs 5e-k. Among the tested isoflavones, compounds 5j, 6f, and 6g show more potent aromatase inhibitory activities than the others. Based on the kinetic studies, it is clear that these compounds compete at the active site of aromatase with the natural substrate, androstenedione. Especially of great interest as a new lead is compound 5j, containing a benzyl group that seems to be a key feature for the inhibitory activity. Further optimization of the 7-substituent of compound 5j realized compounds with greater than two-fold activity improvements. In addition these structure-activity studies highlighted the potential role for π-π stacking interactions within the aromatase active site and the importance of strong electron withdrawing substituents. A number of the 5j analogs display very similar potencies for aromatase inhibition in terms of IC50 values, therefore additional enzyme kinetic studies

are ongoing. While investigations are currently underway to evaluate the exact

nature of the enzyme-ligand interactions, these compounds could be new leads for

the development of more potent inhibitors in this series.

194

4.5 EXPERIMENTAL SECTION

Preparation of human placental microsomes. Human term placentas from nonsmoking

females were acquired from The Ohio State University Department of Pathology/Tissue

Procurement in normal saline. The placenta was washed with ice-cold normal saline and connective and vascular tissue was removed. The entire procedure is carried out at 4° C.

Microsomes were prepared from the remaining tissue using a modified version of the method described by Kellis and Vickery. [222] The remaining tissue is minced with scissors and blended in homogenization buffer (0.25 M sucrose, 25 mM potassium phosphate pH 7.4, 1 mM EDTA). The resulting homogenate is centrifuged at 15,000 × g for 25 minutes. The supernatant is saved and the pellet discarded. The supernatant is ultracentrifuged at 140,000 × g for 30 minutes if 10 mL tubes are used or 45 minutes if

30 mL tubes are used. The supernatant is discarded and the pellet is washed with 3-4 volumes of microsome washing buffer (100 mM potassium phosphate pH 7.4, 1 mM

EDTA, 5 mg/L butylated hydroxytoluene). The microsomes are resuspended in two volumes microsomal washing buffer by vortexing, then pooled. The pooled microsomes are homogenized using a Potter-Elvejhem type tissue homogenizer and drill. The homogenized microsomes are centrifuged at 140,000 × g for 45 minutes. The resulting supernatant is discarded and the microsomes are resuspended in microsome extraction

(storage) buffer (50 mM potassium phosphate pH 7.4, 100 mM potassium chloride, 1mM

EDTA, 1mM dithiothreitol, 5 mg/L butylated hydroxytoluene, 20 % (v/v) glycerol). The resuspended microsomes are aliquoted into cryogenic storage vials, flushed with nitrogen, and stored at −80 °C until required

195

Preparation of unlabeled androstenedione. Androstenedione was recrystallized from hexane/ethyl ether (1:1). The crystals were washed with ice-cold hexane and dried in the

oven. The purity of the resulting material was determined by reversed-phase high

performance liquid chromatography using Beckman Ultrasphere 25 × 0.46 cm ODS

column and an isocratic mobile phase of 40 % methanol/water and flow rate of 0.5

mL/minute. Androstenedione elutes as a single peak with a retention time of 14.3 minutes

under these conditions.

Inhibition study. Inhibition of human placental aromatase was determined by

3 3 monitoring the amount of H2O released as the enzyme converts [1β- H]androst-4-ene-

3,17-dione to estrone. Ten inhibitor concentrations ranging from 100 nM to 50 µM were

evaluated. Aromatase activity assays were carried in 0.1 M potassium phosphate buffer

(pH 7.0) with 5% propylene glycol. All samples contained a NADPH regenerating

system consisting of 2.85 mM glucose-6-phosphate, 1.8 mM NADP+ and 1.5 units of

glucose-6-phosphate dehydrogenase (Sigma, St. Louis, MO). Samples contained 100 nM

androst-4-ene-3,17-dione (400,000−450,000 dpm). Reactions were initiated with the

addition of 50 µg microsomal protein. The total incubation volume was 2.0 mL.

Incubations were allowed to proceed for 15 minutes in a shaking water bath at 37 °C.

Reactions were quenched by the addition of 2.0 mL of chloroform. Samples were then

vortexed and centrifuged for 5 minutes and the aqueous layer was removed. The aqueous

layer was subsequently extracted twice in the same manner with 2.0 mL chloroform.

A 0.5 mL aliquot of the final aqueous layer was combined with 5 mL 3a70B scintillation cocktail (Research Products International Corp., Mt. Prospect, IL) and the amount of

196

radioactivity determined. Each sample was run in triplicate and background values were

determined with microsomal protein inactivated by boiling. Samples containing 50 µM

(±) aminoglutethimide (Sigma, St. Louis, MO) were used a positive control. IC50 sigmoidal dose-response data were analyzed with the Graphpad Prism (Version 3.0) program.

Kinetic study. Enzyme kinetic studies of compounds 3j, 4f, and 4g were conducted to investigate the nature of aromatase inhibition. Michaelis-Menten enzyme kinetic parameters were determined by varying the concentration of androst-4-ene-3,17-dione from 10 to 300 nM or 50 to 500 nM in the prescence of a fixed concentration of inhibitor.

Assay conditions were the same as those described in the IC50 studies except reactions

were initiated by the addition of 15 µg microsomal protein. Analysis of the enzyme kinetic data was performed with the weighted linear regression analysis previously

described by Cleland. [227]

4.6 REFERENCES

208. Dixon, R. A.; Steele, C. Flavonoids and isoflavonoids-gold mine for metabolic

engineering. Trends Plant Sci. 1999, 4, 394-400.

209. Barnes, J.; Anderson, L. A.; Phillipson, J. D. St. John’s wort (Hypericum

perforatum L.): a review of its chemistry, pharmacology and clinical properties.

J. Pharm. Pharmacol. 2001, 53, 583-600.

210. Vrijsen, R.; Everaert, L.; Boeye, A. Antiviral activity of flavones and

potentiation by ascorbate. J. Gen. Virol. 1988, 69, 1749-1751. 197

211. Ito, M.; Ishmoto, S.; Nishida, Y.; Shiramizu, T.; Yumoki, H. Effects of

baicalein, a flavonoid and other anti-inflammatory agents on glyoxalase-I

activity. Agric. Biol. Chem. 1986, 50, 1073-1074.

212. Miski, M.; Ulubelen, A.; Johansson, C.; Mabry, T. J. Antibacterial activity

studies of flavonoids from Salvia palaestina. J. Nat. Prod. 1983, 46, 874-875.

213. Tahara, S.; Katagiri, Y.; Ingham, J. L.; Mizutani, J. Prenylated flavonoids in the

roots of yellow lupin. Phytochemistry 1994, 36, 1261-1271.

214. Markaverich, B. M.; Roberts, R. R.; Alejandro, M. A.; Johnson, G. A.;

Middleditch, B. S.; Clark, J. H. Bioflavonoid interaction with rat uterine type II

binding sites and cell growth inhibition. J. Steroid Biochem. 1988, 30, 71-78.

215. Jordan, V. C.; Koch, R.; Bain, R. R. In Estrogens in the Environment II:

Influences on Development; McLachlan, J. A., Ed.; Elsevier: New York, 1985;

pp 221-234.

216. Shutt, D. A.; Cox, R. I. Steroid and phyto-oestrogen binding to sheep uterine

receptors in vitro. J. Endocrinol. 1972, 52, 299-310.

217. Martin, P. M.; Horowitz, K. B.; Ryan, D. S.; McGuire, W. L. Phytoestrogen

interaction with estrogen receptor in human breast cancer cells. Endocrinology

1978, 103, 1860-1867.

218. So, F. V.; Guthrie, N.; Chambers, A. F.; Carroll, K. K. Inhibition of

proliferation of estrogen receptor-positive MCF-7 human breast cancer cells by

198

flavonoids in the presence and absence of excess estrogen. Cancer Lett. 1997,

112, 127-133.

219. Evans B. E.; Rittle, K. E.; Bock, M. G.; DiPardo, R. M.; Freidinger, R. M.;

Whitter, W. L.; Lundell, G. F.; Veber, D. F.; Anderson, P. S.; Chang, R. S.

Methods for drug discovery: development of potent, selective, orally effective

cholecystokinin antagonists. J. Med. Chem. 1988, 31, 2235-2246.

220. Adlercreutz, H. Phytoestrogens: epidemiology and a possible role in cancer

protection. Environ. Health Perspect. 1995, 103 (Suppl. 7), 130-132.

221. Hodek, P.; Trefil, P.; Stiborová, M. Flavonoids-potent and versatile biologically

active compounds interacting with cytochromes P450. Chem. Biol. Interact.

2002, 139,1-21.

222. Kellis Jr., J. T.; Vickery, L. E. Purification and characterization of human

placental aromatase cytochrome P-450. J. Biol. Chem. 1987, 262, 4413-4420.

223. Recanatini, M.; Cavalli, A.; Valenti, P. Nonsteroidal aromatase inhibitors:

Recent Advances. Med. Res. Rev. 2002, 22, 282-304.

224. Kim, Y.-W., Ph. D. Thesis., The Ohio State University 2004.

225. Kim, Y.-W.; Brueggemeier, R. W. A convenient one-pot synthesis of 2-

(alkylthio)isoflavones from deoxybenzoins using a phase

transfer catalyst. Tetrahedron Lett. 2002, 43, 6113-6115.

199

226. Kim, Y.-W.; Mobley, J. A.; Brueggemeier, R. W. Synthesis and estrogen

receptor binding affinities of 7-hydroxy-3-(4-hydroxyphenyl)-4H-1-benzopyran-

4-ones containing a basic side chain. Bioorg. & Med. Chem. Lett. 2003, 13, 1475-

1478.

227. Cleland, W. W. The statistical analysis of enzyme kinetic data. Method Enzymol.

1979, 63, 103-138.

200

CHAPTER 5

AZOLE ISOFLAVONE INHIBITORS OF AROMATASE

5.1 INTRODUCTION

Our previous study demonstrated proof of the principle that introduction of appropriate nitrogen-containing heterocycle at the 2-position of the isoflavone nucleus is beneficial

for aromatase inhibition. We have now focused on the synthesis of azole isoflavones,

which were easily prepared by the phase transfer catalysis developed in our laboratory.

[224-226] In this series we are interested in exploring the potential contribution of the

imidazole, triazole, thioimidazole, and thiotriazole towards aromatase inhibition. Our

previous study with 2-(4-pyridylmethylthio)isoflavones identified the importance of a

hydrophobic functionality in the 7- position of the isoflavone, and the reduced necessity

of the 4’-substituent to achieve desirable affinity for the enzyme. Herein, we also

describe some diversification of these positions to examine their contribution to the

structure-aromatase inhibition relationships of azole isoflavones.

201 OMe OH O

HO O Biochanin A (apparent Ki = 12 ± 5 µM) OH O

HO O Chrysin (apparent Ki = 2.6 ± 0.1 µM) O

BnOO S N 7-Benzyloxy-2-(4-pyridylmethylthio)isoflavone (apparent Ki = 0.22 ± 0.02 µM)

Figure 5.1. Chemical structures and aromatase inhibitory acitivities of Biochanin A [72]

(isoflavone), Chrysin (flavone), [62] and the 7-benzyloxy-2-(4- pyridylmethylthio)isoflavone aromatase inhibitor reported by our laboratory.

202 5.2 RESULTS AND DISCUSSION

2-(Methylsulfonyl)isoflavones 11a-d were prepared from the corresponding 2-

(methylthio)isoflavones with mCPBA using the method previously described by our laboratory. [224-226] Azole isoflavones 12a-d were prepared from 2-

(methylsulfonyl)isoflavones 11a-d and thioazole isoflavones 13a-d were prepared from

11a in a single step in moderate to good yields by displacing the 2-methylsulfonyl group

with various nucleophiles. (Figure 5.2)

Evaluation of the compounds for aromatase inhibitory activity was performed using the

tritiated water release aromatase assay previously used by our laboratory in which human

placental microsomes are the aromatase source. IC50 values of the compounds were

determined from dose-response curves (Figure 5.3 and 5.4). The IC50 values for (±)-

aminoglutethimide (AG) and BCA were also determined for comparison. Kinetic studies

were undertaken for promising compounds to examine their mode of aromatase

inhibition. The apparent Ki values for each compound, assay-specific apparent Km values, and Ki/Km ratios are listed in Table 5.2. The Lineweaver-Burk plot of imidazole

isoflavones 12a-d are shown in Figures 5.7-5.10

203 R R2 O 2 O

R3H, NaH o DMF, 0 C, 0.5 h R O R R1 O SO2Me 1 3

1a: R1 = OMe, R2 = H, 12a: R1 = OMe, R2 = H, R3 = imidazol-1-yl 1b: R1 = OMe, R2 = OMe 12b: R1 = OMe, R2 = OMe, R3 = imidazol-1-yl 1c: R1 = OBn, R2 = H 12c: R1 = OBn, R2 = H, R3 = imidazol-1-yl 1d: R1 = OBn, R2 = OMe 12d: R1 = OBn, R2 = OMe, R3 = imidazol-1-yl 13a: R1 = OMe, R2 = H, R3 = 2H-1,2,4-triazol-1-yl 13b: R1 = OMe, R2 = H, R3 = 1H-imidazolyl-2-thio 13c: R1 = OMe, R2 = H, R3 = 1-methyl-1H-imidazolyl-2-thio 13d: R1 = OMe, R2 = H, R3 = 2H-1,2,4-triazolyl-3-thio

Figure 5.2: Synthesis of Azole Isoflavones 12a-d and 13a-d. [224-226]

204 120

100

80

60

40

% Control Activity Control % 20

0 1 2 3 4 5 log[nM]

Figure 5.3: Aromatase inhibitory activities of imidazole isoflavones 12a (○), 12b (▲),

12c (●), and 12d (□). Aminoglutethimide (■) is used as a reference. Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

120

100

80

60 % Control Activity Control % 40

2 3 4 5

log [nM]

Figure 5.4: Aromatase inhibitory activities of triazole and thioazole isoflavones 13a (○),

13b (▲), 13c (●), and 13d (□). Error bars represent standard error (n = 3), and the data were analyzed by the nonlinear regression analysis method.

205 a IC50 (µM) Log IC50 (nM) (±S.E.)

12a 0.77 2.89 ± 0.04

12b 2.0 3.29 ± 0.04

12c 0.52 2.71 ± 0.03

12d 4.7 3.67 ± 0.09

13a 18 4.26 ± 0.09 13b >50 13c >50 13d >50

AG 2.8 3.45 ± 0.05

BCA 34 4.53 ± 0.06

Table 5.1. IC50 values for aromatase inhibition by azole isoflavones and reference

compounds. IC50 values were calculated by nonlinear regression analysis in the GraphPad prism program. Data for each dose-response curve were obtained by evaluating

aromatase inhibition at ten concentrations of the compound, each in triplicate.

206 In dose-response studies (Table 5.1) up to 50 µM 2-thioimidazole (13b) and 2-

thiotriazole (13c,d) isoflavone analogs displayed less than 50% inhibition of control

aromatase activity and were judged inactive. In contrast to the thioazole analogs,

compounds with the heterocycle attached directly to the isoflavone scaffold showed

promising inhibitory activity. The consequence of the nature of the azole heterocycle is

exemplified by comparing the IC50 values of 7-methoxy analogs 12a and 13a. The imidazole analog 12a is 23-fold more potent than the corresponding triazole. This difference in potency was surprising due to the number of efficacious triazole aromatase inhibitors reported. [223] Recanatini and coworkers previously postulated the coordinating ability of the nitrogen heterocycle is related to the distribution of the highest-occupied molecular orbital (HOMO) and the resulting electron density over the potentially iron-coordinating heterocycle. This postulate was used to partially rationalize the observed differences in potency between imidazomethyl- and triazomethyl- xanthones. [228]

To test this hypothesis, we fully optimized the geometries of 12a and 13a with the

B3LYP [100] hybrid density functional and 6-31G(d) basis set in the Gaussian 98 [107] suite of programs. The optimized geometries of these molecules share similar conformational preferences and may bind similarly to the aromatase active site. (Figure

5.5) As expected, the HOMO of the imidazole analog 12a is more densely localized on the nitrogen heterocycle than in 13a. (Figure 5.5) Difference spectra of this pair of compounds were obtained with immunoaffinity-purified human placental aromatase

(Hauptman-Woodward Medical Research Institute, Buffalo, NY). Both compounds 207 produce classical Type II spectra consistent with an increase in concentration of low-spin iron as a result of azole ligation. (Figure 5.6) Compound 12a and 13a have absorbtion minima at 394 nm and 391 nm, respectively, while both spectra have maxima at 430 nm.

The smaller change in absorbance induced by compound 13a is consistent with triazole being a weaker ligand for the heme iron. Taken together with the density functional theory (DFT) calculations, these observations explain the superior potency of the imidazole analog relies on the availability of the nitrogen lone pair electrons.

Figure 5.5: B3LYP/6-31G(d) optimized geometries of 12a (left) and 13a (right)

displaying the surfaces of the highest-occupied molecular orbital (HOMO). A contour

value of 0.05 e/B3 was used.

In general, imidazole isoflavone inhibitors presented in this work share similar structure-

activity relationships to those reported for 2-(4-pyridylmethyl)thioisoflavones. Regardless

of the 7-substituent, a decrease in potency was observed when a methoxy group was introduced at the 4’-position. However, a disparity exists when the relative decrease in potency is compared considering the nature of the nature of the 7-substituent. 4’,7-

Dimethoxy analog, 12b, is 2.6- fold less potent than the 7-methoxy analog 12a. The loss 208 of potency is more pronounced when considering the 7-benzyloxy analogs. The presence

of the 4’-methoxy in 12d results in a nine-fold loss of activity when compared to 12c.

4’,7-Dimethoxy 2-(4-pyridylmethylthio)-isoflavone and 7-benzyloxy-4’-methoxy-2-(4-

pyridyl-methylthio)isoflavone are 1.2 and 2.5 less active than their 4’-H analogs. This observation may indicate the nature of the 4’-substituent may be a more important factor in the imidazole isoflavones in terms of aromatase inhibition.

0.06

0.04

0.02 Absorbance 0.00

-0.02 360 380 400 420 440 460 480 500 λ (nm)

Figure 5.6: Type II difference spectra of immunoaffinity-purified human placental aromatase induced by 50 µM compound 12a (_____) and compound 13a ( - - - -).

In enzyme kinetic studies, imidazole isoflavones 12a-d demonstrated typical competitive type inhibition in the Lineweaver-Burk plots (Figure 5.7-5.10) supporting their inhibition of aromatase by competing with the natural substrate for the active site. Ki /Km ratios of imidazole isoflavones were calculated as relative inhibitory potency and the same activity trend is observed as in the dose-response studies. Compound 12c demonstrates 48-fold 209 enhancement in potency compared to the natural product lead, BCA. As reflected by its

relative potency, imidazole analog 12c is 4.4-fold more potent than the widely characterized inhibitor AG, which is in agreement within the expected experimental uncertainty of the result determined in the dose-respone studies (5.4-fold enhancement in potency compared to AG). The present enzyme kinetic study results emphasize the importance of these structural modifications for optimization of aromatase inhibition by

isoflavones.

210 0.100

0.075

0.050 formed/min nmol product 0.025

0.000 0 100 200 300 400 500

nM Androstenedione

7500

5000

formed/min] 2500 1/[nmoles product

0 0.000 0.005 0.010 0.015 0.020

1/ [nM Androstenedione]

Figure 5.7: Lineweaver-Burk plot of aromatase inhibition by compound 12a. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 400 nM (▼), and 2000 nM (♦). Each point represents the average of three determinations ± standard

error.

211 0.0015

0.0010

formed/min 0.0005 nmoles product

0.0000 0 100 200 300 400 500

nM Androstenedione

7500

5000

formed/min] 2500 1/[nmol product

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 5.8: Lineweaver-Burk plot of aromatase inhibition by compound 12b. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 500 nM (▲), 2000 nM (▼), and 5000 nM (♦). Each point represents the average of three determinations ± standard

error.

212 0.0015

0.0010

formed/min 0.0005 nmoles product

0.0000 0 100 200 300 400 500

nM Androstenedione

10000

7500

5000 formed/min] 2500 1/[nmoles product

0 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 5.9: Lineweaver-Burk plot of aromatase inhibition by compound 12c. Various concentrations of androstenedione (50 to 500 nM) were incubated with microsomal enzyme preparations at inhibitor concentrations of 0 nM (■), 100 nM (▲), 400 nM (▼), and 2000 nM (♦). Each point represents the average of three determinations ± standard error.

213 0.0015

0.0010

formed/min 0.0005 nmoles product

0.0000 0 100 200 300 400 500

nM Androstenedione

4500

3500

2500 formed/min] 1500 1/[nmoles product

500 0.000 0.005 0.010 0.015 0.020

1/[nM Androstenedione]

Figure 5.10: Lineweaver-Burk plot of aromatase inhibition by compound 12d. Various

concentrations of androstenedione (50 to 500 nM) were incubated with microsomal

enzyme preparations at inhibitor concentrations of 0 nM (■), 2000 nM (▲), 5000 nM

(▼), and 10000 nM (♦). Each point represents the average of three determinations ± standard error.

214 a a Apparent Ki Apparent Km Ki /Km

12a 0.68 ± 0.04 0.10 ± 0.00 6.8

12b 1.82 ± 0.15 0.10 ± 0.01 18

12c 0.25 ± 0.02 0.07 ± 0.00 3.6

12d 4.40 ± 0.34 0.12 ± 0.01 37

AG 1.41 ± 0.10 0.09 ± 0.01 16

BCA 12 ± 5b

Table 5.2: Enzyme kinetic parameters for imidazole isoflavones and reference compounds. aValues were calculated by weighted regression analysis [227] and expressed

in µM ± standard error. bRef. [72]

5.3 CONCLUSIONS

With the exception of our recent work on 2-(4-pyridylmethylthio)isoflavones, to our

knowledge there has been no effort to construct isoflavone-based aromatase inhibitors.

This study further underlines aromatase inhibitory activity can be achieved with the

isoflavone nucleus by introduction of an appropriate heme-coordinating nitrogen

heterocycle at the 2- position of the isoflavone. Comparable potency to the 2-(4-

pyridylmethylthio)-isoflavones can be obtained by introducing an imidazole moiety at the

2-position. Enzyme kinetic analyses of imidazole isoflavones reveal these compounds

interact competitively with the aromatase active site. As revealed by DFT calculations

and the spectra of ligand binding, the superiority of this heterocycle to achieve aromatase 215 inhibition over others examined may lie in the electronic structure of the nitrogen heterocycle itself. While investigations are currently underway to resolve further aspects of the enzyme-ligand interactions, these compounds are additional leads in our repertoire of potent aromatase inhibitors.

5.4 EXPERIMENTAL SECTION

Microsome preparation, inhibition studies, and kinetic studies were performed as described in Chapter 4.

Determination of P450 Difference Spectra. Immunoaffinity-purified human placental cytochrome P450 aromatase (Hauptman-Woodward Medical Research Institute, Buffalo,

NY) was diluted to 7.0 µg/mL with 0.1 M Tris Buffer (pH 7.4) in a 50 µL quartz UV cell and the absorbance was scanned from 350 nm to 500 nm. 0.1 M Tris Buffer (pH 7.4) was used as a reference. Compounds 12a and 13a were introduced into the sample cell in 1 uL of ethanol for a final concentration of 50 µM and spectrum was rescanned for the aforementioned wavelength range. Multiple scans were taken to ensure spectrum stability for aromatase and aromatase/inhibitor mixtures. Absorbance values of aromatase were subtracted from the aromatase/inhibitor values to generate the difference spectrum.

216 5.5 REFERENCES

228. Recanatini, M.; Bisi, A.; Cavalli, A., Belluti, F.; Gobbi, S.; Rampa, A.; Valenti,

P.; Palzer, M.; Palusczak, A.; Hartmann, R.W. A new class of nonsteroidal

aromatase inhibitors: Design and Synthesis of chromone and xanthone

derivatives and inhibition of the P450 enzymes aromatase and 17α-

hydroxylase/C17,20-lyase. J. Med. Chem. 2001, 44, 672-680.

217

CHAPTER 6

CONCLUSIONS

6.1 COMPUTATIONAL STUDIES OF P450 AROMATASE CATALYSIS

In this study, we have illustrated the effective use of density functional theory as a tool to model the properties of dioxygen P450 catalytic intermediates. The results provided here are complementary to the efforts of molecular spectroscopy to identify and characterize transient intermediates in the enzymatic cycle, which elude direct structural observation.

In general, GGA functional treatments predict accurate ground state multiplicities for each of the models studied. The B3LYP result for the ground state multiplicity is in disagreement with the experimental singlet configuration; however, spin multiplicities of these systems can be extremely sensitive to thermal equilibration effects and geometric perturbations which could result from constraints imparted on the system by the protein.

Electron affinity calculations from low-spin states predict a dramatic shift in the electron affinity upon immersion in a low dielectric solvent. This observation illustrates the influence of the hydrophobic core of the enzyme on the redox properties of the P450 prosthetic group. 218 The DFT treatments used here produce quality geometries for dioxygen model system

compared with available structural data from X-Ray diffraction of CYP101 as well as

synthetic analogs of the P450 site. The model we describe here is extended from those of

other theoretical studies to include the unusual NH-S hydrogen bond. Different oxidation

states of the model system show significant differences in the value of this parameter,

indicating it may play a role in facilitating the reduction of O2-bound P450. The next

logical step to be taken in our laboratory to understand the role of the protein

environment on this facet of dioxygen activation would be to expand the model system

and embed it as part of a hybrid QM/MM scheme. [144,145] Considering the lack of

scaling factors for the level of theory, the calculated frequencies are in reasonable

agreement with infrared and resonance Raman studies. Identification of multiple νO-O

modes, some of which were coupled to underlying vibrations of the porphyrin

macrocycle and cysteinamide ligand affirms the potential for Fermi resonance in

vibrational studies of this enzyme.

TD-DFT can approximately reproduce the Soret shifts, which occur in the UV/Visible spectrum of after dioxygen activation. In the case of the model systems described in this

study, both model reduced dioxygen and hydroperoxo catalytic intermediates experience

shifts of approximately 20 nm in the B’ band of the split Soret. These results are in

reasonable agreement with experimental observations of CYP101-D251N and CYP101-

WT enzymes for which spectra arising from each of these intermediates may have been

observed. Interestingly, these results indicate the reduced dioxygen and hydroperoxo

catalytic intermediates may have similar spectra when they are absolutely characterized. 219 The role of UV/Visible spectroscopy to distinguish between these two species is still questionable. Notably, some of the significant, singlet electronic transitions in the hydroperoxo model arose from orbitals localized to “protein” backbone in the present model. This observation indicates that more extensive models of the P450 active site may be required to identify a distinguishing feature.

Computation of isotropic hyperfine coupling constants of the model catalytic intermediates described herein unveiled a competent level of theory to reproduce experimentally derived parameters as well as establish how theory can distinguish between different members of the catalytic cycle. The B3LYP functional proved to be very accurate in reproducing the 14N isotropic hyperfine coupling constants of the

hydroperoxo species. Interestingly, the computed values were very low for the reduced

dioxygen intermediate with this level of theory, which has been assigned the same 14N

isotropic hyperfine coupling constants by experiment. It has been postulated that

hydrogen-bond donors in the active site of D251N-CYP101 distort the spin density from

the dioxygen ligand such that the net spin density on Fe integrates to nearly one, giving

rise to the same hyperfine-coupling constant. To test this hypothesis, we introduced

neutral water molecules as well as a hydronium ion/neutral water pair to the distal face of

the heme to model such an environment. Despite including hydrogen-bond donors as

intense as the gas-phase hydronium ion, a significant amount of spin density failed to

accumulate on the iron atom to reproduce the experimental hyperfine coupling constant.

These results raise serious questions about the actual identity of the species observed

after radiolytic reduction of oxygenated D251N-CYP101 at 77 K, and it is possible that 220 the observed species is not the reduced dioxygen intermediate. In addition to the results presented here, theoretical studies have indicated that even in the absence of an acidic terminal proton donor, the thermodynamic barrier for proton transfer is quite small. We can speculate that the catalytic intermediate observed in the presence of the D251N mutation is also the hydroperoxo species experiencing an alternative hydrogen-bond milieu, rather than the reduced dioxygen catalytic intermediate.

We have presented a B3LYP density functional theory study of historically accepted and also novel mechanisms for the aromatization/deformylation sequence catalyzed by aromatase. Preliminary studies of peroxo hemiacetal adduct model systems indicate that

1β-hydrogen atom removal by the proximal oxygen of the iron peroxo species requires a high-energy barrier and does not initiate fragmentation to the experimentally observed products.

Alternative mechanistic processes initiated by 1β-hydrogen atom abstraction by the widely accepted ultimate P450 oxidant, Compound I, were considered. Through these studies, a novel mechanism for the final catalytic step was delineated. A strikingly low barrier height (< 7 kcal/mol) for the 1β-hydrogen-atom abstraction was computed for steroid models containing the 2,3-enol moiety and computations using the keto tautomers confirmed the low barrier results from the enolized species ability to delocalize the developing radical. Noting that the deformylation and dehydrogenation of the keto species occurs with similar barriers computed for other known P450 substrates such as camphor, we can speculate the mechanism described here may be operative in other P450 221 enzymes where demethylation and dehydrogenation occur with identical oxygen and

cofactor requirements. Ab initio molecular dynamics confirmed that both deformylation

and decarbonylation occur in two distinct steps. However, removal of the second hydrogen can be more adequately described as a deprotonation of the substrate cation by a ferric hydroxy P450 intermediate formed from a preceeding electron transfer event as depicted in Figure 3.18.

The novel dehydrogenase mechanism operative in the final catalytic step of aromatase identified here may provide insight into other biosynthetic P450-reactions whose mechanisms for C-C bond cleavage lack understanding. These enzymes are the cholesterol side chain cleavage enzyme (CYP11A), lanosterol 14α-demethylase

(CYP51), and progesterone 17α-hydroxylase/17,20-lyase. Interestingly, all three enzymes, like aromatase, require three moles of NADPH and molecular oxygen to carry out their respective C-C bond cleavage reaction, the final step is preceded by two classical P450 hydroxylation mechanisms, and the peroxo intermediate has been proposed as the active P450 oxidant in all three cases. Of the three enzymes, the 14α- demethylation of lanosterol is the most similar to aromatase, where the 14α-methyl gem- diol is formed, followed by 15α-hydrogen atom abstraction to introduce the 14-15 olefin.

[194-196]

CYP11A cleaves the C20-C22 bond of cholesterol in mitochondria. The first two steps are 22R- and 20S- hydroxylation. The efficient conversion is insured by the fact that the

222 dihydroxylated intermediate has an affinity for the enzyme 300 time that of cholesterol.

[229] Based on the aromatase mechanism delineated by theoretical means, it is easy to

envision that the final C-C bond cleavage step could be initiated by hydrogen atom

abstraction from either of the diol hydroxyl groups. The resulting oxy radical could

rearrange one of the aldehyde products and the carbon radical. The remaining radical

could transfer an electron to the P450 hydroxy-radical intermediate, leaving a protonated

aldehyde., which could potentially be deprotonated in a barrierless manner leaving

pregnenolone, 4-methylpentanal, and the aqua-bound resting state of the enzyme.

CYP17A introduces a 17α-hydroxyl group of pregnenolone (or progesterone) and cleaves the C17-20 olefin to give dehydroepiandrosterone (or androstenedione). If pregnenolone and progesterone are free intermediates, the hydration of the 20-one would be required prior to C17-C20 cleavage step prior to the dehydrogenase bond cleavage reaction. As in CYP11A, the reaction could be initiated by hydrogen atom abstraction from the 17α-hydroxyl group. In either order, an electron could be transferred to the

hydroxyradical P450 intermediate resulting in a Fe-bound hydroxide able to deprotonate

the gem-diol, or the 17α-oxy radical could undergo rearrangement to produce the C19

steroid and an orthoacetate radical which could transfer and electron to the

hydroxyradical intermediate and undergo subsequent deprotonation.

223 6.2 ISOFLAVONE INHIBITORS OF AROMATASE

To date, the isoflavone ring system has been considered an inappropriate scaffold for

development of aromatase inhibitors. This study has shown that aromatase inhibitory

activity can be achieved in the isoflavone nucleus by introducing functional groups with

the potential to coordinate the heme iron. The structure-activity relationships indicate that the binding modes of 7-protected analogs might be different from those of 7-hydroxy analogs. Hydrogen-bonding potential appears to be important in the 7-hydroxy analogs for their aromatase inhibitory activity, whereas hydrophobic interactions appear to play a role in 7-protected analogs. Among the tested isoflavones, compounds 5j, 6f, and 6g show more potent aromatase inhibitory activities than the others. Based on the kinetic studies, it is clear that these compounds compete at the active site of aromatase with the natural substrate, androstenedione. Especially of great interest as a new lead is compound

5j, containing a benzyl group that seems to be a key feature for the inhibitory activity.

Further optimization of the 7-substituent of compound 5j realized compounds with greater than two-fold activity improvements. In addition these structure-activity studies highlighted the potential role for π-π stacking interactions within the aromatase active site and the importance of strong electron withdrawing substituents. A number of the 5j

analogs display very similar potencies for aromatase inhibition in terms of IC50 values,

therefore additional enzyme kinetic studies are ongoing.

Further studies underlined how aromatase inhibitory activity could be achieved with the

isoflavone nucleus by introduction of an appropriate heme-coordinating nitrogen heterocycle at the 2- position of the isoflavone. Comparable potency to the 2-(4- 224 pyridylmethylthio)-isoflavones can be obtained by introducing an imidazole moiety at the

2-position. Enzyme kinetic analyses of imidazole isoflavones reveal these compounds interact competitively with the aromatase active site. As revealed by DFT calculations and the spectra of ligand binding, the superiority of this heterocycle to achieve aromatase inhibition over others examined may lie in the electronic structure of the nitrogen heterocycle itself. While investigations are currently underway to resolve further aspects of the enzyme-ligand interactions, these compounds are additional leads in our repertoire of potent aromatase inhibitors.

6.3 REFERENCES

229. Lambeth, J. D.; Kitchen, S. E.; Farooqui, A. A.; Tuckey, R.; Kamin, H.

Cytochrome P450scc-substrate interactions: Studies of binding and catalytic

activity using cholesterols. J. Biol. Chem. 1982, 257, 1876-1884.

230. Zuber. M. X.; Simpson, E. R.; Waterman, M. R. Expression of bovine 17α-

hydroxylase cytochrome P450 cDNA in nosteroiogenic COS-1 cells. Science,

1986, 234, 1258-1261.

231. Barnes, H. J.; Arlotto, M. P.; Waterman, M. R. Expression and enzymatic

activity of recombinant cytochrome P450 17α-hydroxylase in Escherica coli.

Proc. Natl. Acad. Sci. USA. 1991, 88, 5597-5601.

225

APPENDIX A

SUPPORTING INFORMATION FOR CHAPTER 2

226

Figure A.1: O2(πg)-Fe(3dyz)-S(2py) lowest unoccupied Kohn-Sham molecular orbital of

1- the [Fe(P)(O2)(SCysNH2)] model intermediate (top) and singly-occupied Kohn-Sham

2- 3 molecular orbital of [Fe(P)(O2)(SCysNH2)] (bottom). An isocontour value of 0.03 e/B has been used.

227

Figure A.2: O2(πg)-Fe(3dyz)-S(2py) singly-occupied Kohn-Sham molecular orbital of the

1- 3 [Fe(P)(O2H)(SCysNH2)] model intermediate. An isocontour value of 0.03 e/B has been used.

228 120(3eg) 121 (4a2u + amide π*) 122 (4a2u + amide π*)

123(2b2u) 127 (1a1u) 128(O2(πg)-Fe(3dyz)-S(2py))

129(5a2u) 134(LUMO) 135 (5eg)

136 (5eg) 137 (2b1u)

Figure A.3: Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2)(SCysNH2)] .

229 121(3eg) 122(3eg) 127(1a1u)

133(HOMO) 134(SOMO) 135(5eg)

142(3b2u)

Figure A.4: Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2H)(SCysNH2)] .

230 120(3eg) 123(2b2u) 124(amide π*)

128(amide π*) 135(5eg) 137(2b1u)

Figure A.5: Contour plots of prominent molecular orbitals active in the RI TD-

1- BP86/(TZVP,SV(P)) electronic transitions of [Fe(P)(O2H)(SCysNH2)] .

231

APPENDIX B

SUPPORTING INFORMATION FOR CHAPTER 3

232

Figure B.1: Singly occupied molecular orbital of 2ERC.

233 Step 9 Step 12 Step 15 Step 17 Step 19 Step 23 Step 24

Fe-O 1.734 1.732 1.734 1.740 1.749 1.778 1.787

Fe-S 2.359 2.370 2.380 2.383 2.386 2.385 2.384

H1β-O(Fe) 1.388 1.304 1.237 1.181 1.103 0.930 0.939

C1-H1β 1.246 1.293 1.348 1.386 1.455 1.698 1.750

∠O-H1β-C1 168.0 167.3 167.1 169.3 169.1 150.1 141.9

OHproSO(Fe) 1.487 1.359 1.322 1.340 1.388 1.479 1.468

O-Hpro-S 1.002 1.118 1.165 1.151 1.107 1.023 1.034

O-Hpro-R 1.027 0.994 0.963 1.007 1.038 0.964 0.953

∠H1β-O-Fe-N* 10.37 14.46 18.25 18.58 17.79 14.21 13.29

∠H-O-Fe 141.0 145.7 147.7 145.9 142.1 126.7 120.7

Table B.1: Relevant Geometric Parameters for Select Time Steps of the 2ETS Molecular

Dynamics Trajectory. * Value relative to the dihedral angle in the transition state.

234 Step Step Step Step Step Step Step Step

25 26 29 31 34 37 41 46

Fe-O 1.796 1.805 1.840 1.866 1.907 1.951 1.999 2.051

Fe-S 2.382 2.379 2.368 2.358 2.338 2.312 2.272 2.217

H1β-O(Fe) 0.977 1.026 1.080 1.008 0.941 1.061 1.030 1.005

C1-H1β 1.796 1.850 2.095 2.284 2.486 2.543 2.443 2.194

∠O-H1β-C1 133.2 124.8 106.1 98.20 88.87 83.92 88.7 101.2

OHproSO(Fe) 1.442 1.405 1.271 1.189 1.082 0.998 1.039 1.112

O-Hpro-S 1.056 1.084 1.176 1.236 1.342 1.446 1.462 1.581

O-Hpro-R 0.961 0.984 1.036 1.008 0.963 1.026 1.013 1.014

∠H1β-O-Fe-N* 12.45 11.66 8.92 5.92 -1.68 -11.58 -23.34 -36.82

∠H-O-Fe 114.6 108.8 93.80 85.14 75.72 75.58 85.24 105.6

Table B.2: Relevant Geometric Parameters for Select Time Steps of the 2ETS Molecular

Dynamics Trajectory. * Value relative to the dihedral angle in the transition state.

235

Step 9 (320 a.u. , 7.7 fs) Step 12 (440 a.u., 10.6 fs)

Step 15 (560 a.u., 13.5 fs) Step 17 (640 a.u., 15.5 fs)

Figure B.2: Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory.

236

Step 19 (720 a.u., 17.4 fs) Step 23 (880 a.u., 21.3 fs)

Step 24 (920 a.u., 22.2 fs) Step 25 (960 a.u., 23.2 fs)

Figure B.3: Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory.

237

Step 26 (1000 a.u., 24.2 fs) Step 29 (1120 a.u., 27.1 fs)

Step 31 (1200 a.u., 29.0 fs) Step 34 (1320 a.u., 31.9 fs)

Figure B.4: Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory.

238

Step 37 (1440 a.u., 34.8 fs) Step 41 (1600 a.u., 38.7 fs)

Step 46 ( 1800 a.u., 43.5 fs)

Figure B.5: Structures of Select Time Steps of the 2ETS Molecular Dynamics Trajectory.

239 Step 9 Step 12 Step14 Step 16 Step 19 Step 22 Step 25

Fe-O 1.750 1.775 1.799 1.815 1.842 1.864 1.883

Fe-S 2.349 2.340 2.326 2.319 2.296 2.267 2.251

H1β-O(Fe) 1.112 0.935 1.085 1.096 1.068 1.071 0.917

C1-H1β 1.492 1.641 1.634 1.741 1.979 2.183 2.425

∠O-H1β-C1 168.7 165.5 156.0 150.7 126.3 117.0 108.5

Hformyl-O 1.800 1.764 1.783 1.785 1.781 1.727 1.666

Cformyl-Hformyl 1.127 1.086 1.084 1.103 1.124 1.110 1.156

∠H1β-O-Fe-N* 3.10 2.8 5.49 9.76 30.16 37.42 48.24

∠H-O-Fe 118.1 114.5 111.0 110.3 110.45 109.5 109.3

Table B.3. Relevant Geometric Parameters for Select Time Steps of the 2OTS Molecular

Dynamics Trajectory. * Value relative to the dihedral angle in the transition state.

240 Step 27 Step 30 Step 33 Step 48 Step 52

Fe-O 1.894 1.911 1.933 1.987 1.998

Fe-S 2.240 2.224 2.210 2.192 2.199

H1β-O(Fe) 0.967 1.107 1.021 1.027 1.098

C1-H1β 2.553 2.738 2.905 3.547 3.655

∠O-H1β-C1 98.71 85.1 78.33 42.72 37.99

Hformyl-O 1.636 1.628 1.656 1.403 1.063

Cformyl-Hformyl 1.181 1.188 1.165 1.378 1.645

∠H1β-O-Fe-N* 56.85 67.54 78.05 147.1 164.1

∠H-O-Fe 108.7 107.8 107.3 113.2 112.7

Table B.4: Relevant Geometric Parameters for Select Time Steps of the 2OTS

Molecular Dynamics Trajectory. * Value relative to the dihedral angle in the transition

state.

241

Step 9 (320 a.u., 7.7 fs) Step 12 (440 a.u., 10.6 fs)

Step 14 (520 a.u., 12.6 fs) Step 16 (600 a.u., 14.5 fs)

Figure B.6: Structures of Select Time Steps of the 2OTS Molecular Dynamics

Trajectory.

242

Step 19 (720 a.u., 17.4 fs) Step 22 (840 a.u., 20.3 fs)

Step 25 (960 a .u., 23.2 fs) Step 27 (1040 a.u., 25.3 fs)

Figure B.7: Structures of Select Time Steps of the 2OTS Molecular Dynamics

Trajectory.

243

Step 30 (1160 a. u., 28.1 fs) Step 33 (1280 a.u., 31.0 fs

Step 48 (1880 a.u., 45.5 fs) Step 52 (2040 a. u., 49.3 fs)

Figure B.8: Structures of Select Time Steps of the 2OTS Molecular Dynamics

Trajectory.

244 Excitation ∆E (eV) Oscillator Comments Strengtha Step 19 75.8(158β→159β) 1.04 0.013 Corresponds to a transition between an orbital centered on the substrate diene system and the porphyrin π system Step 31 73.2(158β→159β) 0.18 0.003 Corresponds to a transition between a mixed porphyrin π/sulfur p orbital and the substrate diene system

Table B.5: B3LYP/(TZVP,SV(P)) TD-DFT calculations of MD steps 19 and 31 initiated with 2ETS. aVelocity representation in Turbomole.

245 0.8

0.7

0.6

0.5

0.4

0.3 Substrate Atomic Charge Atomic Substrate

0.2

0.1 720 760 800 840 880 920 960 1000 Time (a.u.)

0.8

0.7

0.6

0.5

Spin Density Spin 0.4 β

0.3

0.2 Substrate

0.1

0.0 720 760 800 840 880 920 960 1000 Time (a.u.)

Figure B.9: Gas phase (■) and ε = 5.621 (▲) NPA atomic charges and spin densities localized on the substrate at critical points on the 2ETS molecular dynamics trajectory.

246

BIBLIOGRAPHY

Adlercreutz, H. Phytoestrogens: epidemiology and a possible role in cancer protection. Environ. Health Perspect. 1995, 103 (Suppl. 7),130-132.

Adlercreutz, H.; Bannwart, C.; Wahala, K.; Makela, T; Brunow, G.; Hase, T.; Arosemena, P. J.; Kellis, J. T. Jr.; Vickery, L. E. Inhibition of human aromatase by mammalian lignans and isoflavonoid phytoestrogens. J Steroid Biochem. Mol. Biol. 1993, 44, 147-153.

Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C. Electronic structure calculations on workstation computers: the program system TURBOMOLE. Chem. Phys. Lett. 1989, 162, 165-169.

Akhtar, M.; Alexander, K.; Boar, R. B.; McGhie, J. F.; Barton, D. H. R. Chemical and enzymic studies on the characterization of intermediates during the removal of the 14alpha-methyl group in cholesterol biosynthesis. The use of 32-functionalized lanostane derivatives. Biochem. J. 1978, 169, 449-463.

Akhtar, M.; Calder, M. R.; Corina, D. L.; Wright, J. N. Mechanistic studies on C-19 demethylation in estrogen biosynthesis. Biochem. J. 1982, 201, 569-580.

Akhtar, M.; Freeman, C. W.; Wilton, D. C.; Boar, R. B.; Copsey, D. B. The pathway for the removal of the 15α-methyl group of lanosterol. The role of lanost-8-ene- 3β,32-diol in cholesterol biosynthesis. Bioorg. Chem. 1977, 6, 473.

Alexander, K. T. W.; Akhtar, M.; Boar, R. B.; McGhie, J. F.; Barton, D. H. R. Pathway for the removal of C-32 in cholesterol biosynthesis. J. Chem. Soc., Chem. Commun. 1971, 1479-1481.

Alexander, K.; Akhtar, M.; Boar, R. B.; McGhie, J. F.; Barton, D. H. R. Removal of the 32-carbon atom as formic acid in cholesterol biosynthesis. J. Chem. Soc., Chem. Commun. 1972, 383-385.

Amarneh, B.; Corbin, C. J.; Peterson, J. A.; Simpson, E. R.; Graham-Lorence, S. Functional domains of human aromatase cytochrome P450 characterized by

247 molecular modeling and site-directed mutagenesis. Mol. Endocrinol. 1993, 7, 1617- 1624.

Antony, J.; Grodzicki, M.; Trautwein, A. X. Local Density Functional Study of Oxoiron(IV) Porphyrin Complexes and Their One-Electron Oxidized Derivatives. Axial Ligand Effects. J. Phys. Chem. A. 1997, 101, 2692-2701.

Arigoni, D.; Battaglia, R.; Akhtar, M.; Smith, T. J. Chem. Soc., Chem. Commun. 1975, 185.

Bangcharoenpaurpong, O.; Rizos, A. K.; Champion, P. M. Resonance Raman detection of bound dioxygen in cytochrome P-450cam. J. Biol. Chem. 1986, 261, 8089-8092.

Banting, L. Inhibition of aromatase. Prog. Med. Chem. 1996, 33, 147-184.

Banting, L.; Nicholls, P. J.; Shaw, M. A.; Smith, H. J. Recent developments in aromatase inhibition as a potential treatment for oestrogen-dependent breast cancer. Prog. Med. Chem. 1989, 26, 253-298.

Barnes, H. J.; Arlotto, M. P.; Waterman, M. R. Expression and enzymatic activity of recombinant cytochrome P450 17α-hydroxylase in Escherica coli. Proc. Natl. Acad. Sci. USA. 1991, 88, 5597-5601.

Barnes, J.; Anderson, L. A.; Phillipson, J. D. St. John’s wort (Hypericum perforatum L.): a review of its chemistry, pharmacology and clinical properties. J. Pharm. Pharmacol. 2001, 53, 583-600.

Basis sets were obtained from the Extensible Computational Chemistry Environment Basis Set Database, Version 02/25/04, as developed and distributed by the Molecular Science Computing Facility, Environmental and Molecular Sciences Laboratory which is part of the Pacific Northwest Laboratory, P.O. Box 999, Richland, Washington 99352, USA, and funded by the U.S. Department of Energy. The Pacific Northwest Laboratory is a multi-program laboratory operated by Battelle Memorial Institute for the U.S. Department of Energy under contract DE-AC06-76RLO 1830. Contact David Feller or Karen Schuchardt for further information.

Bauernschmitt, R.; Ahlrichs, R. Treatment of electronic excitations within the adiabatic approximation of time dependent density functional theory. Chem. Phys. Lett. 1996, 256, 454-464.

Bauernschmitt, R.; Häser, M.; Treutler, O.; Ahlrichs, R. Calculation of excitation energies within time dependent density functional theory using auxiliary basis set expansions. Chem. Phys. Lett. 1997, 264, 573-578.

248 Baum, M.; Budzar, A. U.; Cuzik, J.; Forbes, J.; Houghton, J. H.; Klijn, J. G.; Sahmoud, T. Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: first results of the ATAC randomised trial. Lancet 2002, 359, 2131-2139.

Bauschlicher, C. W.; Langhoff, S. R.; Barnes, L. A. Theoretical studies of the first- and second-row transition-metal methyls and their positive ions. J. Chem. Phys. 1989, 91, 2399-2411.

Becke, A. D. Density-functional thermochemistry. I. The effect of the exchange-only gradient correction. J. Chem. Phys., 1992, 96, 2155-2160.

Becke, A. D. Density-functional thermochemistry. II. The effect of the Perdew–Wang generalized-gradient correlation correction J. Chem. Phys. 1992, 97, 9173-9177.

Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648-5652

Becke, A.D. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A. 1988, 38, 3098-3100.

Bednarski P. J.; Nelson S. D. Dissociation of 19-hydroxy- 19-oxo-, and aromatizing- activities in human placental microsomes through the use of suicide substrates to aromatase. J. Steroid Biochem. 1989, 32, 309-316.

Benson, D. E.; Suslick, K. S.; Sligar, S. G. Reduced Oxy Intermediate Observed in D251N Cytochrome P450cam. Biochemistry 1997, 36, 5104-5107.

Bhatnagar, A. S.; Hausler, A.; Schieweck, K.; Lang, M.; Bowman, R. Highly selective inhibition of estrogen biosynthesis by CGS20267, a new non-steroidal aromatase inhibitor. J. Steroid Biochem. Mol. Biol. 1990, 37, 1021-1027.

Binkley, J. S.; Pople, J. A.; Hehre, W. J. Self-consistent molecular orbital methods. 21. Small split-valence basis sets for first-row elements. J. Am. Chem. Soc. 1980, 102, 939-947.

Blakesley, D. W.; Payne, S. C.; Hagen, K. S. Spin-State Variation in Solid State and Solution of Mononuclear Iron(II) 1,4,7-Trimethyl-1,4,7-triazacyclonane Complexes. Inorg. Chem. 2000, 39, 1979-1989.

Blanco, J. G.; Gil, R. R.; Bocco, J. L.; Meragelman, T. L.; Genti-Raimondi, S.; Flury, A. Aromatase inhibition by an 11,13-dihydroderivative of a sesquiterpene lactone. J. Pharmacol. Exp. Ther. 2001, 297, 1099-1105.

249 Blanco, J. G.; Gill, R. R.; Alavarez, C. I.; Patrito, L. C.; Genti-Raimondi, S.; Flury, A. A novel activity for a group of sequiterpene lactones: inhibition of aromatase. FEBS Lett. 1997, 409, 396-400.

Bonneterre, J.; Buzdar, A.; Nabholtz, J. M.; Robertson J. F.; Thurlimann, B.; von Euler, M.; Sahmoud, T.; Webster, A.; Steinberg, M. Anastrozole is superior to tamoxifen as first-line therapy in hormone receptor positive advanced breast carcinoma. Cancer 2001, 92, 2247-2258.

Bossche, H. V. J. Steroid Biochem. Mol. Biol. 1992, 43, 1003.

Brodie, A. M. H. ISI Atlas of Science: Pharmacology 1987, 266.

Brodie, A.; Lu, Q.; Long, B. Aromatase and its inhibitors. J. Steroid Biochem. Mol. Biol. 1999, 69, 205-210.

Brodie, H. J.; Kripalani, K. J.; Possanza, G. Studies on the mechanism of estrogen biosynthesis. VI. The stereochemistry of hydrogen elimination at C-2 during aromatization. J. Am. Chem. Soc. 1969, 91, 1241-1242.

Brueggemeier R.W. Aromatase inhibitors in breast cancer therapy. Expert Rev. Anticancer Ther. 2002, 2, 181-191.

Brueggemeier, R. W. Aromatase, aromatase inhibitors, and breast cancer. Am. J. Ther. 2001, 8, 333-344.

Callard, G. V. Aromatization is cyclic-AMP dependent in cultured reptilian brain cells. Brain Res 1981, 204, 451-454.

Callard, G. V.; Petro, Z.; Ryan, K. J. Aromatization and 5α-reduction in the brain and non-neural tissues of a cyclostome (Petromyzan marinus) Gen Comp Endocrinol 1980, 42, 155-159.

Callard, G. V.; Petro, Z.; Ryan, K. J.; Phylogenetic distribution of aromatase and other androgen-converting enzymes in the central nervous system. Endocrinology 1978, 103, 2283-2290.

Campbell, D. R.; Kurzer, M. S. Flavonoid inhibition of aromatase enzyme activity in human preadipocytes. J. Steroid Biochem. Mol. Biol. 1993, 46, 381-388.

Capdevila, J. H.; Zeldin, D.; Makita, K.; Karara, A.; Falck, J. R. Cytochrome P450 and the Metabolism of Arachidonic Acid and Oxygenated Eicosanoids. In Cytochrome P450: Structure, Mechanism, and Biochemistry, Second Edition; Ortiz de Montellano, P. R., Eds. Plenum Press: New York, 1995; 443-471.

250 Carr, B. R. Disorders of the ovary and female reproductive tract. In Williams Textbook of Endocrinology, Ninth edition. Wilson, J. D. Ed. W. B. Saunders: Philadelphia, 1998; 751-779.

Caspi, E.; Arunachalam, T.; Nelson, P. A. Biosynthesis of estrogens: the steric mode of the initial C-19 hydroxylation of androgens by human placental aromatase. J. Am. Chem. Soc. 1983, 105, 6987-6989.

Caspi, E.; Arunachalam, T.; Nelson, P. A. Biosynthesis of estrogens: aromatization of (19R)-,(19S)-,and(19RS)-[19-3H,2H,1H]-3.beta.-hydroxyandrost-5-en-17-ones by human placental aromatase J. Am. Chem. Soc. 1986, 108, 1847-1852.

Caspi, E.; Dharmaratne, H. R. W.; Shackleton, C. Biosynthesis of estrogens by microsomal placental aromatase; isolation and metabolism of 10β-hydroxyestr-4-ene- 3,17-dione. J. Chem. Soc., Chem. Commun. 1989, 1699.

Caspi, E.; Wicha, J.; Arunachalam, T.; Nelson, P.; Spiteller, G. Estrogen biosynthesis. Concerning the obligatory intermediacy of 2β-hydroxy-10β-formylandrost-4-ene- 3,17-dione. J. Am. Chem. Soc. 1984, 106, 7282-7283.

Chandrasena, R. E. P.; Vatsis, K. P.; Coon, M. J.; Hollenberg, P. F.; Newcomb, M. Hydroxylation by the Hydroperoxy-Iron Species in Cytochrome P450 Enzymes J. Am. Chem. Soc. 2004, 126, 115-126.

Chen, S.; Zhou, D. Functional domains of cytochrome P450 inferred from comparative analysis of amino acid sequences and substantiated by site-directed mutagenesis experiments. J. Biol. Chem. 1992, 267, 22587-22594.

Clark, T.; Chandrasekhar, J.; Schleyer, P. v. R. J. Comput. Chem. 1983, 4, 294.

Cleland, W. W. The statistical analysis of enzyme kinetic data. Method Enzymol. 1979, 63, 103-138.

Cole, P. A.; Bean, J. M.; Robinson, C. H. Conversion of a 3-desoxysteroid to 3- desoxyestrogen by human placental aromatase. Proc. Natl. Acad. Sci. U. S. A. 1990, 87, 2999-3003.

Cole, P. A.; Robinson, C. H. Conversion of 19-oxo[2β-2H]androgens into oestrogens by human placental aromatase. An unexpected stereochemical outcome. Biochem. J. 1990, 268, 553-561.

Cole, P. A.; Robinson, C. H. Synthesis and study of a methoxy hydroperoxide androstenedione derivative; analog of a potential aromatase intermediate. J. Chem. Soc., Chem. Commun. 1986, 1651-1653

251 Cole, P. A.; Robinson, C. H. Synthesis of and reactivity studies with 19-peroxide- androstenedione derivatives: analogs of a proposed aromatase intermediate. J. Chem. Soc., Perkin Trans. 1. 1990, 2119-2125.

Cole, P. A.; Robinson, C. H. Mechanism and inhibition of cytochrome P-450 aromatase. J. Med. Chem.1990, 33, 2933-2942.

Cole, P. A.; Robinson, C. H. Peroxide model reaction for placental aromatase. J. Am. Chem. Soc. 1988, 110, 1284-1285.

Cossi, M.; Barone, V.; Cammi, R.; Tomasi, J. Ab initio study of solvated molecules: a new implementation of the polarizable continuum model. Chem. Phys. Lett., 1996, 255, 327-335.

Covey, D. F. Aromatase inhibitors: specific inhibitors of estrogen biosynthesis. In Sterol biosynthesis inhibitors. Pharmaceutical and agrochemical aspects. Berg, D.; Plempel, M. Eds. Ellis Horwood Ltd., Chichester (England), 1988: 534-571

Covey, D. F.; Carrell, H. L.; Beusen, D. D. Metabolism of 19-methyl substituted steroids and a proposal for the third aromatase monooxygenation. Steroids 1987, 50, 363-374.

Covey, D. F.; Hood, W. R.; Beusen, D. D.; Carrell, H. L. Hydroperoxides as inactivators of aromatase: 10 beta-hydroperoxy-4-estrene-3,17-dione, crystal structure and inactivation characteristics. Biochemistry, 1984, 23, 5398-5406.

Davydov, R. M.; Yoshida, T.; Ikeda-Saito, M.; Hoffman, B. M. Hydroperoxy-Heme Oxygenase Generated by Cryoreduction Catalyzes the Formation of -meso- Hydroxyheme as Detected by EPR and ENDOR J. Am. Chem. Soc. 1999, 121, 10656- 10657.

Davydov, R.; Kappl, R.; Hutterman, J.; Peterson, J. A. EPR-spectroscopy of reduced oxyferrous-P450cam. FEBS 1991, 295, 113-115.

Davydov, R.; Macdonald, I. D. G.; Makris, T. M.; Sligar, S. G.; Hoffman, B. M. EPR and ENDOR of Catalytic Intermediates in Cryoreduced Native and Mutant Oxy- Cytochromes P450cam: Mutation-Induced Changes in the Proton Delivery System. J. Am. Chem. Soc. 1999, 121, 10654-10655.

Davydov, R.; Makris, T. M.; Kofman, V.; Werst, D. E.; Sligar, S. G.; Hoffman, B. M. Hydroxylation of Camphor by Reduced Oxy-Cytochrome P450cam: Mechanistic Implications of EPR and ENDOR Studies of Catalytic Intermediates in Native and Mutant Enzymes. J. Am. Chem. Soc. 2001, 123, 1403-1415.

252 Dawson, J. H.; Kau, L. -S.; Penner-Hahn, J. E.; Sono, M.; Eble, K. S.; Bruce, G. S.; Hager, L. P.; Hodgson, K.O. Oxygenated cytochrome P-450-CAM and chloroperoxidase: direct evidence for sulfur donor ligation trans to dioxygen and structural characterization using EXAFS spectroscopy J. Am. Chem. Soc. 1986, 108, 8114-8116.

Dawson, J. H.; Sono, M. Cytochrome P-450 and chloroperoxidase: thiolate-ligated heme enzymes. Spectroscopic determination of their active-site structures and mechanistic implications of thiolate ligation. Chem. Rev. 1987, 87, 1255-1276. de Visser, S. P.; Ogliaro, F.; Sharma, P. K.; Shaik, S. What Factors Affect the Regioselectivity of Oxidation by Cytochrome P450? A DFT Study of Allylic Hydroxylation and Double Bond Epoxidation in a Model Reaction. J. Am. Chem. Soc. 2002, 124, 11809-11826. de Visser, S. P.; Shaik, S. A Proton-Shuttle Mechanism Mediated by the Porphyrin in Benzene Hydroxylation by Cytochrome P450 Enzymes J. Am. Chem. Soc. 2003, 125, 7413-7424. de Visser, S. P.; Shaik, S.; Sharma, P. K.; Kumar, D.; Thiel, W. Active Species of Horseradish Peroxidase (HRP) and Cytochrome P450: Two Electronic Chameleons. J. Am. Chem. Soc. 2003, 125, 15779-1578.

Denisov, I. G.; Ikeda-Saito, M.; Yoshida, T.; Sligar, S. G. Cryogenic absorption spectra of hydroperoxo-ferric heme oxygenase, the active intermediate of enzymatic heme oxygenation. FEBS Lett. 2002, 532, 203-206.

Denisov, I. G.; Makris, T. M.; Sligar, S. G. Cryotrapped reaction intermediates of cytochrome p450 studied by radiolytic reduction with phosphorus-32. J. Biol. Chem. 2001, 276, 11648-11652.

Denisov, I. G.; Makris, T. M.; Sligar, S. G. Formation and decay of hydroperoxo- ferric heme complex in horseradish peroxidase studied by cryoradiolysis. J. Biol. Chem. 2002, 277, 42706-42710.

Dickson, R. B.; Lippman, L. E. Growth factors in breast cancer. Endocrine Rev. 1995, 16, 559-589.

Dixon, R. A.; Steele, C. Flavonoids and isoflavonoids-gold mine for metabolic engineering. Trends Plant Sci. 1999, 4, 394-400.

Du, P.; Loew, G. H. Role of axial ligand in the electronic structure of model compound I complexes. Int. J. Quantum Chem. 1992, 44, 251-261.

253 Early Breast Cancer Trialist’s Group. Systemic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy. Lancet 1992, 339, 1-15.

Eichkorn, K.; Treutler, H.; Öhm, H.; Häser, M.; Ahlrichs, R. Auxiliary basis sets to approximate Coulomb potentials. Chem. Phys. Lett. 1995, 242, 652-660.

Eichkorn, K.; Weigend, F.; Treutler, O. Ahlrichs, R. Auxiliary basis sets for main row atoms and transition metals and their use to approximate Coulomb potentials. Theor. Chem. Acc. 1997, 97, 119-124.

Elliott, S. D.; Ahlrichs, R.; Hampe, O.; Kappes; M. M. Auto-ionised products from the reaction of sodium clusters with dioxygen: Theory and experiment. Phys. Chem. Chem. Phys. 2000, 2, 3415-3424.

Evans B. E.; Rittle, K. E.; Bock, M. G.; DiPardo, R. M.; Freidinger, R. M.; Whitter, W. L.; Lundell, G. F.; Veber, D. F.; Anderson, P. S.; Chang, R. S. Methods for drug discovery: development of potent, selective, orally effective cholecystokinin antagonists. J. Med. Chem. 1988, 31, 2235-2246.

Fisher, M. T.; Sligar, S. G. Control of heme protein redox potential and reduction rate: linear free energy relation between potential and ferric spin state equilibrium. J. Am. Chem. Soc. 1985, 107, 5018-5019.

Fishman, J.; Guzik, H. Stereochemistry of estrogen biosynthesis. J. Am. Chem. Soc. 1969, 91, 2805-2806.

Fishman, J.; Guzik, H.; Dixon, D. Sterochemistry of estrogen biosynthesis. Biochemistry 1969, 8, 4304-4309.

Fishman, J.; Raju, M. S. Mechanism of estrogen biosynthesis. Stereochemistry of C-1 hydrogen elimination in the aromatization of 2 beta-hydroxy-19-oxoandrostenedione. J. Biol. Chem. 1981, 256, 4472-4477.

Gaussian 03, Revision B.04, Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; 254 Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian, Inc.: Pittsburgh, PA, 2003.

Gaussian 98 (Revision A.11.3), Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V.G.; Montgomery, J. A.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian, Inc., Pittsburgh PA, 1998.

Goff, H. M. Iron(III) porphyrin-imidazole complexes. Analysis of carbon-13 nuclear magnetic resonance isotropic shifts and unpaired spin delocalization J. Am. Chem. Soc. 1981, 103, 3714-3722.

Gonzalez, C.; Schlegel, H. B. An improved algorithm for reaction path following. J. Chem. Phys. 1989, 90, 2154-2161.

Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. Reaction path following in mass- weighted internal coordinates. 1990, 94, 5523-5527.

Gordon, M. S.; Binkley, J. S.; Pietro, W. J.; Hehre, W. J. Self-consistent molecular- orbital methods. 22. Small split-valence basis sets for second-row elements. J. Am. Chem. Soc. 1983, 104, 2797-2803.

Goto, J.; Fishman, J. Participation of a nonenzymatic transformation in the biosynthesis of estrogens from androgens. Science 1977, 195, 80-81.

Graham-Lorence, S.; Amarneh, B.; White, R. E.; Peterson, J. A.; Simpson, E. R. A three-dimensional model of aromatase cytochrome P450. Prot. Sci. 1995, 4, 1065- 1080.

Graves, P. E.; Salhanick, H. A. Stereoselective inhibition of aromatase by enantiomers of aminoglutethimide. Endocrinology 1979, 105, 52-57.

Green, M. T. Evidence for Sulfur-Based Radicals in Thiolate Compound I Intermediates. J. Am. Chem. Soc. 1999, 121, 7939-7940.

255 Groves, J. T.; Watanabe, Y. J. Reactive iron porphyrin derivatives related to the catalytic cycles of cytochrome P-450 and peroxidase. Studies of the mechanism of oxygen activation. J. Am. Chem. Soc. 1988, 110, 8443-8452.

Groves, J. T.; Watanabe, Y. J. Reactive iron porphyrin derivatives related to the catalytic cycles of cytochrome P-450 and peroxidase. Studies of the mechanism of oxygen activation. J. Am. Chem. Soc. 1988, 110, 8443.

Guallar, V.; Baik, M.-H.; Lippard, S. J.; Friesner, R. A. Peripheral heme substituents control the hydrogen-atom abstraction chemistry in cytochromes P450. Proc. Natl. Acad. Sci. USA. 2003, 100, 6998-7002.

Guallar, V.; Friesner, R. A. Cytochrome P450CAM Enzymatic Catalysis Cycle: A Quantum Mechanics/Molecular Mechanics Study. J. Am. Chem. Soc. 2004, 126, 8501-8508.

Guallar, V.; Harris, D. L.; Batista, V. S.; Miller, W. H. Proton-Transfer Dynamics in the Activation of Cytochrome P450eryF J. Am. Chem. Soc. 2002, 124, 1430-1437.

Guengerich, F. P. Common and Uncommon Cytochrome P450 Reactions Related to Metabolism and Chemical Toxicity. Chem. Res. Toxicol. 2001, 14, 611-650.

Hahn, E. F.; Fishman, J. Immunological probe of estrogen biosynthesis. Evidence for the 2 beta-hydroxylative pathway in aromatization of androgens. J. Biol. Chem. 1984, 259, 1689-1694.

Harris, D. L. Oxidation and electronic state dependence of proton transfer in the enzymatic cycle of cytochrome P450eryF. J. Inorg. Biochem. 2002, 91, 568-585.

Harris, D.; Loew, G.; Waskell, L. Structure and Spectra of Ferrous Dioxygen and Reduced Ferrous Dioxygen Model Cytochrome P450. J. Am. Chem. Soc. 1998, 120, 4308-4318.

Hida, T.; Ishii, T.; Kanamaru, T.; Muroi, M. TAN-931, A novel nonsteroidal aromatase inhibitor produced by Penicillium funiculosum No. 8974. J. Antibiot. 1991, 44, 600-612.

Himo, F.; Siegbahn, P. E. M. Quantum chemical studies of radical-containing enzymes. Chem. Rev. 2003, 103, 2421-2456.

Hodek, P.; Trefil, P.; Stiborová, M. Flavonoids-potent and versatile biologically active compounds interacting with cytochromes P450. Chem. Biol. Interact. 2002, 139, 121.

256 Horrocks, W. D.; Greenberg, E. S. Evaluation of dipolar nuclear magnetic resonance shifts in low-spin hemin systems: ferricytochrome c and metmyoglobin cyanide. Biochim. Biophys. Acta. 1973, 322, 38-44.

Hosoda, H.; Fishman, J. Unusually facile aromatization of 2β-hydroxy-19-oxo-4- androstene-3,17-dione to estrone. Implications for estrogen biosynthesis. J. Am. Chem. Soc. 1974, 96, 7325-7329.

Ibrahim, A.- R.; Abul-Hajj, Y. Aromatase inhibition by flavonoids. J. Steroid Biochem. Mol. Biol. 1990, 37, 257-260.

Ito, M.; Ishmoto, S.; Nishida, Y.; Shiramizu, T.; Yumoki, H. Effects of baicalein, a flavonoid and other anti-inflammatory agents on glyoxalase-I activity. Agric. Biol. Chem. 1986, 50, 1073-1074.

Johansson, M. P.; Blomberg, M. R. A.; Sundholm, D.; Wikström, M. Change in electron and spin density upon electron transfer to haem. Biochim. Biophys. Acta. 2002, 1553, 183-187.

Johansson, M. P.; Sundholm, D.; Gerfen, G.; Wikström, M. The Spin Distribution in Low-Spin Iron Porphyrins. J. Am. Chem. Soc. 2002, 124, 11771- 11780.

Jordan, V. C.; Koch, R.; Bain, R. R. In Estrogens in the Environment II: Influences on Development; McLachlan, J. A., Ed.; Elsevier: New York, 1985; pp 221-234.

Jordan, V.C. Tamoxifen: toxicities and drug resistance during the treatment and prevention of breast cancer. Ann. Rev. Pharmacol. Toxicol. 1995, 35, 195-211.

Kagawa, N; Waterman, M. R. Regulation of Steroidogenic P450s. In Cytochrome P450: Structure, Mechanism, and Biochemistry, Second Edition; Ortiz de Montellano, P. R., Eds. Plenum Press: New York, 1995; 419-442.

Kamachi, T.; Yoshizawa, K. A Theoretical Study on the Mechanism of Camphor Hydroxylation by Compound I of Cytochrome P450 J. Am. Chem. Soc. 2003, 125, 4652-4661.

Kao, Y.-C.; Zhou, C.; Sherman, M.; Laughton, C. A. Molecular basis for the inhibition of human aromatase (estrogen synthetase) by flavone and isoflavone phytoestrogens: A site-directed mutagenesis study. Environ. Health Perspect. 1998, 106, 85-92.

Kellis Jr., J. T.; Vickery, L. E. Purification and characterization of human placental aromatase cytochrome P-450. J. Biol. Chem. 1987, 262, 4413-4420.

257 Kellis, J. T., Jr.; Nesnow, S.; Vickery, L. E. Inhibition of aromatase cytochrome P- 450 (estrogen sythetase) by derivatives of α-napthoflavone. Biochem. Pharmacol. 1986, 35, 2887-2891.

Kellis, J. T.; Vickery, L. E. Inhibition of human estrogen synthetase (aromatase) by flavones. Science 1984, 225, 1032-1034.

Kim, Y.-W. Ph. D. Thesis. The Ohio State University 2004.

Kim, Y.-W.; Brueggemeier, R. W. A convenient one-pot synthesis of 2- (alkylthio)isoflavones from deoxybenzoins using a phase transfer catalyst. Tetrahedron Lett. 2002, 43, 6113-6115.

Kim, Y.-W.; Hackett, J. C.; Brueggememeier, R. W. Synthesis and aromatase inhibitory activity of novel pyridine-containing isoflavones. J. Med. Chem. 2004, 47, 4032-4040.

Kim, Y.-W.; Mobley, J. A.; Brueggemeier, R. W. Synthesis and estrogen receptor binding affinities of 7-hydroxy-3-(4-hydroxyphenyl)-4H-1-benzopyran-4-ones containing a basic side chain. Bioorg. & Med. Chem. Lett. 2003, 13, 1475-1478.

Kobayashi, K.; Iwamoto, T.; Honda, K. Spectral intermediate in the reaction of ferrous cytochrome P450cam with superoxide anion. Biochem. Biophys. Res. Commun. 1994, 201, 1348-1355.

Krishnan, R.; Binkley, J. S.; Seeger, R. Pople, J. A. Self-consistent molecular orbital methods. XX. A basis set for correlated wave functions. J. Chem. Phys. 1980, 72, 650-654.

Kumar, D.; de Visser, S. P.; Shaik, S. Oxygen Economy of Cytochrome P450: What Is the Origin of the Mixed Functionality as a Dehydrogenase-Oxidase Enzyme Compared with Its Normal Function? J. Am. Chem. Soc. 2004, 126, 5072-5073.

Kuo C. L.; Raner G. M.; Vaz A. D.; Coon M. J. Discrete species of activated oxygen yield different cytochrome P450 heme adducts from aldehydes. Biochemistry 1999, 38, 10511-10518.

Lambeth, J. D.; Kitchen, S. E.; Farooqui, A. A.; Tuckey, R.; Kamin, H. Cytochrome P450scc-substrate interactions: Studies of binding and catalytic activity using cholesterols. J. Biol. Chem. 1982, 257, 1876-1884.

Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density Phys. Rev. B 1988, 37, 785-789.

258 Li, X. Y.; Zgierski, M. Z. Iron motion in a five-coordinated heme model. Chem. Phys. Lett. 1992, 188, 16-20.

Létard, J. F.; Guionneau, P.; Rabardel, L.; Howard, J. A. K.; Goeta, A. E.; Chasseau, D.; Kahn, O. Structural, Magnetic, and Photomagnetic Studies of a Mononuclear Iron(II) Derivative Exhibiting an Exceptionally Abrupt Spin Transition. Light- Induced Thermal Hysteresis Phenomenon Inorg. Chem. 1998, 37, 4432-4441.

Makris, T. M.; Davydov, R.; Denisov, I. G.; Hoffman, B. F.; Sligar, S. G. Mechanistic enzymology of oxygen activation by the cytochromes P450. Drug Metab. Rev. 2002, 34, 691-708.

Markaverich, B. M.; Roberts, R. R.; Alejandro, M. A.; Johnson, G. A.; Middleditch, B. S.; Clark, J. H. Bioflavonoid interaction with rat uterine type II binding sites and cell growth inhibition. J. Steroid Biochem. 1988, 30, 71-78.

Martin, P. M.; Horowitz, K. B.; Ryan, D. S.; McGuire, W. L. Phytoestrogen interaction with estrogen receptor in human breast cancer cells. Endocrinology 1978, 103, 1860-1867.

Mason, J. I.; Murry, B. A.; Olcott, M.; Sheets, J. J. Imidazole antimycotics: inhibitors of steroid aromatase. Biochem. Pharmacol. 1985, 34, 1087-1092.

Mastalerz, H.; Morand, P. Acid- and base-catalyzed reactions of 4β,5β- and 4α,5α- epoxyandrostane-3,17,19-trione. J. Chem. Soc., Perkin Trans. I. 1982, 2611.

McGuire, W. L. An update on estrogen and progesterone receptors in prognosis for primary and advanced breast cancer, in: S. Iacobelli, R. J. B. King, H. R. Lidner, M. E. Lippman. (Eds), Hormones and Cancer, vol. 15, Raven press, New York, 1980, pp 337-344.

Meunier, B.; de Visser, S. P.; Shaik, S. Mechanism of oxidation reactions catalyzed by cytochrome P450 enzymes. Chem. Rev. 2004, 104, 3947-3980.

Miertus, S.; Scrocco, E.; Tomasi, J. Electrostatic interaction of a solute with a continuum. A direct utilization of ab initio molecular potentials for the prevision of solvent effects. J. Chem. Phys. 1981, 55, 117-129.

Miertus, S.; Tomasi, J. Approximate evaluations of the electrostatic free energy and internal energy changes in solution processes. Chem. Phys. 1982, 65, 239-245.

Miller, W. R.; Jackson, J.; The therapeutic potential of aromatase inhibitors. Expert Opin. Investig. Drugs 2003, 12, 1-15.

259 Minami, T.; Iwamoto, M.; Ohtsu, H.; Ohishi, H.; Tanaka, R.; Yoshitake, A. Aromatase inhibitory activities of standishnal and the diterpenoids from the bark of Thuja standishii. Planta Med. 2002, 68, 742-745.

Miski, M.; Ulubelen, A.; Johansson, C.; Mabry, T. J. Antibacterial activity studies of flavonoids from Salvia palaestina. J. Nat. Prod. 1983, 46, 874-875.

Mitropoulos, K. A.; Gibbons, G. F.; Reeves, E. A. Lanosterol 14alpha-demethylase. Similarity of the enzyme system from yeast and rat liver. Steroids 1976, 27, 821-829.

Miyairi, S. Fishman, J. Novel method of evaluating biological 19-hydroxylation and aromatization of androgens. Biochem. Biophys. Res. Commun. 1983, 117, 392-398.

Miyairi, S. Fishman, J. Radiometric analysis of oxidative reactions in aromatization by placental microsomes. Presence of differential isotope effects. J. Biol. Chem. 1985, 260, 320-325.

Morand, P.; Williamson, D. G.; Layne, D. S.; Lompa-Krzymien, L.; Salvador, J. Conversion of an androgen epoxide into 17β-estradiol by human placental microsomes. Biochemistry, 1975, 14, 635-638.

Morokuma, K. New challenges in quantum chemistry: Quests for accurate calculations for large molecular systems. Philos. Transact. Ser. A Math. Phys. Eng. Sci. 2002, 360, 1149-1164.

Munzarová, M.; Kaupp, M. A Critical Validation of Density Functional and Coupled- Cluster Approaches for the Calculation of EPR Hyperfine Coupling Constants in Transition Metal Complexes. J. Phys. Chem. A 1999, 103, 9966-9983.

Newcomb, M.; Aebisher, D.; Shen, R.; Chandrasena, R. E. P.; Hollenberg, P. F.; Coon, M. J. Kinetic Isotope Effects Implicate Two Electrophilic Oxidants in Cytochrome P450-Catalyzed Hydroxylations J. Am. Chem. Soc. 2003; 125, 6064- 6065.

Newcomb, M.; Toy, P. H. Hypersensitive Radical Probes and the Mechanisms of Cytochrome P450-Catalyzed Hydroxylation Reactions Acc. Chem. Res. 2000, 33, 449-455.

Noodleman, L.; Lovell, T.; Han, W.-G.; Li, J.; Himo, F. Quantum chemical studies of intermediates and reaction pathways in selected enzymes and catalytic synthetic systems. Chem. Rev. 2004, 104, 459-508.

Numazawa, M.; Midzuhashi, K.; Nagaoka, M. Metabolic aspects of the 1 beta-proton and the 19-methyl group of androst-4-ene-3,6,17-trione during aromatization by

260 placental microsomes and inactivation of aromatase. Biochem. Pharmacol. 1994, 47, 717.

O’Malley, P. J. Density Functional Calculations Modelling the Spin Density Distribution, Hyperfine Couplings, and Hydrogen Bonding Environment of the Ascorbate (Vitamin C) Free Radical. J. Phys. Chem. B 2001, 105, 11290-11293.

O’Reilly, J. M.; Brueggemeier, R. W. Development of steroidal and nonsteroidal inhibitors of aromatase for the treatment f hormone-dependent breast cancer. Curr. Med. Chem. 1996, 3, 11-22.

Ogliaro F.; de Visser S. P.; Groves J. T.; Shaik S. Chameleon States: High-Valent Metal-Oxo Species of Cytochrome P450 and Its Ruthenium Analogue. Angew. Chem. Int. Ed. Engl. 2001, 40, 3503.

Ogliaro, F.; Cohen, S.; de Visser, S.; Shaik, S. Medium Polarization and Hydrogen Bonding Effects on Compound I of Cytochrome P450: What Kind of a Radical Is It Really? J. Am. Chem. Soc. 2000, 122, 12892-12893.

Ogliaro, F.; de Visser, S. P.; Cohen, S.; Kaneti, J.; Shaik, S. The experimentally elusive oxidant of cytochrome P450: a theoretical "trapping" defining more closely the "real" species. ChemBioChem 2001, 11, 848-851.

Ogliaro, F.; de Visser, S.; Cohen, S.; Sharma, P. K.; Shaik, S. Searching for the Second Oxidant in the Catalytic Cycle of Cytochrome P450: A Theoretical Investigation of the Iron(III)-Hydroperoxo Species and Its Epoxidation Pathways. J. Am. Chem. Soc. 2002, 124, 2806-2817.

Ortiz de Montellano, P. R.; Oxygen activation and reactivity. In Cytochrome P450: Structure, Mechanism, and Biochemistry, Second Edition; Ortiz de Montellano, P. R., Ed. Plenum Press: New York, 1995; 245-303.

Osawa, Y.; Shibata, K.; Rohrer, D. Weeks, C., Duax, W. L. Reassignment of the absolute configuration of 19-substituted 19-hydroxysteroids and stereomechanism of estrogen biosynthesis. J. Am. Chem. Soc. 1975, 97, 4400-4402.

Peng, H. M.; Raner, G. M.; Vaz, A. D. N.; Coon, M. J. Oxidative cleavage of esters and amides to carbonyl products by cytochrome P450. Arch. Biochem. Biophys. 1995, 318, 333-339.

Perdew, J. P. Density-functional approximation for the correlation energy of the inhomogeneous electron gas. Phys. Rev. B. 1986, 33, 8822-8824.

Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Pederson, M. R.; Singh, D. J.; Fiolhais, C. Atoms, molecules, solids, and surfaces: Applications of the 261 generalized gradient approximation for exchange and correlation. Phys. Rev. B 1992, 46, 6671-6687.

Pietro, W. J.; Francl, M. M.; Hehre, W. J.; Defrees, D. J. Pople, J. A.; Binkley, J. S. Self-consistent molecular orbital methods. 24. Supplemented small split-valence basis sets for second-row elements. J. Am. Chem. Soc. 1982, 104, 5039-5048.

Plourde, P. V. Dyroff, M.; Dukes, M. Arimidex: a potent and selective fourth- generation aromatase inhibitor. Breast Cancer Res. Treat. 1994, 30, 103.

Poulos, T. L.; Finzel, B. C.; Howard, A. J. Crystal structure of substrate-free Pseudomonas putida cytochrome P-450. Biochemistry 1986, 25, 5314-5322.

Raag, R.; Poulos, T. L.; X-ray crystallographic structural studies of cytochrome P450cam: Factors controlling substrate metabolism. Front. Biotransform. 1992, 7, 1- 43.

Raner G. M., Hatchell J. A.; Dixon M. U.; Joy T. L.; Haddy A. E.; Johnston E. R. Regioselective peroxo-dependent heme alkylation in P450(BM3)-F87G by aromatic aldehydes: effects of alkylation on cataysis. Biochemistry 2002, 41, 9601-9610.

Raner, G. M.; Chiang, E. W.; Vaz, A. D. N.; Coon, M. J. Mechanism-Based Inactivation of Cytochrome P450 2B4 by Aldehydes: Relationship to Aldehyde Deformylation via a Peroxyhemiacetal Intermediate. Biochemistry 1997, 36, 4895- 4902.

Recanatini, M.; Bisi, A.; Cavalli, A., Belluti, F.; Gobbi, S.; Rampa, A.; Valenti, P.; Palzer, M.; Palusczak, A.; Hartmann, R.W. A new class of nonsteroidal aromatase inhibitors: Design and Synthesis of chromone and xanthone derivatives and inhibition of the P450 enzymes aromatase and 17a-hydroxylase/C17,20-lyase. J. Med. Chem. 2001, 44, 672-680.

Recanatini, M.; Cavalli, A.; Valenti, P. Nonsteroidal aromatase inhibitors: Recent Advances. Med. Res. Rev. 2002, 22, 282-304.

Reed, A. E.; Weinstock, R. B.; Weinhold, F. Natural population analysis. J. Chem. Phys. 1985, 83, 735-746.

Ren, W.; Qiao, Z.; Wang, H.; Zhu, L.; Zhang, L. Flavonoids: promising anticancer agents. Med. Res. Rev. 2003, 23, 519-534.

Roberts, E. S.; Vaz, A. D. N.; Coon. M. J. Catalysis by cytochrome P-450 of an oxidative reaction in xenobiotic aldehyde metabolism: deformylation with olefin formation. Proc. Natl. Acad. Sci. USA 1991, 88, 8963-8966.

262 Ruan, W.; Newman, C. B.; Kleinberg, D. L. Intact and aminoterminally shortened forms of insulin-like growth factor induce mammary gland differentiation and development. Proc. Natl. Acad. Sci. USA 1992, 89, 10872-10876.

Ryan, K. J. Conversion of androstenedione to estrone by placental microsomes. Biochim. Biophys. Acta 1958, 27, 658-659.

Ryde, U. Combined quantum and molecular mechanics calculations on metalloproteins. Curr. Opin. Chem. Biol. 2003, 7,136-142.

Santen, R. J.; Harvey, H. A. Use of aromatase inhibitors in breast carcinoma. Endocr. Relat. Cancer 1999, 6, 75-92.

Santen, R. J.; Samojlik, E.; Wells, S. A. Resistance of the ovary to blockade of aromatization with aminoglutethimide. J. Clin. Endocrinol. Metab. 1980, 51, 473- 477.

Santen, R. J.; Worgul, T. J.; Lipton, A.; Harvey, H.; Boucher, A.; Samojlik, E.; Wells, S. A. Aminoglutethimide as a treatment of postmenopausal women with advanced breast carcinoma. Ann. Intern. Med. 1982, 96, 94-101.

Schaefer, A.; Horn, H.; Ahlrichs, R. Fully optimized contracted Gaussian basis sets for atoms Li to Kr. J. Chem. Phys. 1992, 97, 2571-2577.

Schaefer, A.; Huber, C.; Ahlrichs, R. Fully optimized contracted Gaussian basis sets of triple zeta valence quality for atoms Li to Kr. J. Chem. Phys. 1994, 100, 5829- 5835.

Schappacher, M.; Ricard, L.; Fischer, J.; Weiss, R.; Bill, E.; Montiel-Montoya, R.; Winkler, H.; Trautwein, A. X. Synthesis, structure and spectroscopic properties of two models for the active site of the oxygenated state of cytochrome P450. Eur. J. Biochem. 1987, 168, 419-429.

Schappacher, M.; Ricard, L.; Weiss, R. Models for the coordination site of iron in cytochrome P-450. Synthesis and spectroscopic properties of a dioxygen adduct of (2,3,5,6-tetrafluorophenylthiolato)iron(II) tetraphenylpivalcylporphyrin J. Am. Chem. Soc. 1981, 103, 7646-7648.

Scheidt, W. R.; Geiger, D. K. Molecular stereochemistry and crystal structure of a 'spin-equilibrium' (S = 1/2, S = 5/2) hemichrome salt: bis-(3- chloropyridine)octaethylporphinatoiron(III) perchlorate. J. Chem. Soc., Chem. Commun. 1974, 1154-1155.

Schenkman J. B.; Sligar S. G.; Cinti D. L. Substrate interaction with cytochrome P- 450. Pharmacol. Ther. 1981, 12, 43-71. 263

Scherlis, D. A.; Martí, M. A.; Ordejón, P. Estrin, D. A. Environment effects on chemical reactivity of heme proteins. Int. J. Quantum. Chem. 2002, 90, 1505-1514.

Scherlis, D.A.; Estrin, D.A. Structure and spin-state energetics of an iron porphyrin model: An assessment of theoretical methods. Int. J. Quantum Chem. 2002, 87, 158- 166.

Schlicting, I.; Berendzen, J.; Chu, K.; Stock, A. M.; Maves, S. A.; Benson, D. E.; Sweet, R. M.; Ringe, D.; Petsko, G. A.; Sligar, S. G. The catalytic pathway of cytochrome p450cam at atomic resolution. Science 2000, 287, 1615-1622.

Schöneboom, J. C.; Lin, H.; Reuter, N.; Thiel, W.; Cohen, S.; Ogliaro, F.; Shaik, S. The elusive oxidant species of cytochrome P450 enzymes: characterization by combined quantum mechanical/molecular mechanical (QM/MM) calculations. J. Am. Chem. Soc. 2002, 124, 8142-8151.

Shaik, S.; Cohen, S.; de Visser, S. P.; Sharma, P. K.; Kumar, D.; Kozuch, S.; Ogliaro, F.; Danovich, D. The “Rebound Controversy: An Overview and Theoretical Modeling of the Rebound Step in Mono-Oxygenations by Cytochrome P450. Eur. J. Inorg. Chem. 2004, 207-226.

Sharma, P. K.; de Visser, S. P.; Shaik, S. Can a Single Oxidant with Two Spin States Masquerade as Two Different Oxidants? A Study of the Sulfoxidation Mechanism by Cytochrome P450. J. Am. Chem. Soc. 2003, 125, 8698-8699.

Shutt, D. A.; Cox, R. I. Steroid and phyto-oestrogen binding to sheep uterine receptors in vitro. J. Endocrinol. 1972, 52, 299-310.

Siegbahn, P. E. M.; Blomberg, M. R. A. Density functional theory of biologically relevant metal centers. Annu. Rev. Phys. Chem. 1999, 50, 221-249.

Siegbahn, P. E. M.; Blomberg, M. R. A. Transition-metal systems in biochemistry studied by high-accuracy quantum chemical methods. Chem. Rev. 2000, 100, 421- 437.

Simonson, T.; Perahia, D. Internal and interfacial dielectric properties of cytochrome c from molecular dynamics in aqueous solution. Proc. Natl. Acad. Sci. USA 1995, 92, 1082-1086.

Simpson. E. R.; Mahendroo, M. S.; Means, G. D.; Kilgore, M. W.; Hinshelwood, M. M.; Graham-Lorence, S.; Amarneh, B.; Ito, Y.; Fisher, C. R.; Michael, M. D.; Mendelson, C. R.; Bulun, S. E. Aromatase cytochrome P450, the enzyme responsible for estrogen biosynthesis. Endocrine Rev. 1994, 15, 342-354.

264 Sjodin, T.; Christian, J. F.; Macdonald, I. D. G.; Davydov, R.; Unno, M.; Sligar, S. G.; Hoffman, B. M.; Champion, P. M. Resonance Raman and EPR Investigations of the D251N Oxycytochrome P450cam/Putidaredoxin Complex. Biochemistry 2001, 40, 6852-6859.

Sligar, S. G. Coupling of spin, substrate, and redox equillibria in cytochrome P450. Biochemistry, 1976, 15, 5399-5406.

Smith, D. M. A.; Dupuis, M.; Vorpagel, E. R.; Straatsma, T. P. Characterization of Electronic Structure and Properties of a Bis(histidine) Heme Model Complex. J. Am. Chem. Soc. 2003, 125, 2711-2717.

So, F. V.; Guthrie, N.; Chambers, A. F.; Carroll, K. K. Inhibition of proliferation of estrogen receptor-positive MCF-7 human breast cancer cells by flavonoids in the presence and absence of excess estrogen. Cancer Lett. 1997, 112, 127-133.

Stevens, P. J.; Devlin, F. J.; Chablowski, C. F.; Frisch, M. J. Ab Initio Calculation of Vibrational Absorption and Circular Dichroism Spectra Using Density Functional Force Fields. J. Phys. Chem. 1994, 98, 11623-11627.

Tahara, S.; Katagiri, Y.; Ingham, J. L.; Mizutani, J. Prenylated flavonoids in the roots of yellow lupin. Phytochemistry 1994, 36, 1261-1271.

Tan, H.; Simonis, U.; Shokhirev, N. V.; Walker, F. A. Substituent Effects on the Choice of the Orbital Preferred for Electron Spin Delocalization in Two Paramagnetic Low-Spin Iron(III) Porphyrins: Mapping the Spin Density Distribution at the Pyrrole Positions by 1H COSY and NOESY Techniques. J. Am. Chem. Soc. 1994, 116, 5784- 5790.

Tan, L.; Muto, N. Purification and reconstitution properties of human placental aromatase. A cytochrome P-450-type monooxygenase Eur. J. Biochem. 1986, 156, 243-250. Townsley, J. D.; Possanza, G. J.; Brodie, H. J. Fed. Proc., Fed. Am. Soc. Exp. Biol. 1966, 25, 282 (Abstract 520).

Toy, P. H.; Newcomb, M.; Coon, M. J.; Vaz, A. D. N. Two Distinct Electrophilic Oxidants Effect Hydroxyl- cation in Cytochrome P-450-Catalyzed Reactions. J. Am. Chem. Soc., 1998, 120, 9718-9719.

Ueyama, N.; Nishikawa, N.; Yamada, Y.; Okamura, T.; Nakamura, A. Cytochrome P-450 Model (Porphinato)(thiolato)iron(III) Complexes with Single and Double NH···S Hydrogen Bonds at the Thiolate Site J. Am. Chem. Soc. 1996, 118, 12826- 12827.

265 Vaz, A. D. N.; Kessell, K. J.; Coon, M. J. Aromatization of a bicyclic steroid analog, 3-oxodecalin-4-ene-10-carboxaldehyde, by liver microsomal cytochrome P450 2B4. Biochemistry 1994, 33, 13651-13661.

Vaz, A. D. N.; Pernecky, S. J.; Raner, G. M.; Coon, M. J. Peroxo-iron and oxenoid- iron species as alternative oxygenating agents in cytochrome P450-catalyzed reactions: switching by threonine-302 to alanine mutagenesis of cytochrome P450 2B4. Proc. Natl. Acad. Sci. USA 1996, 93, 4644-4648.

Verlet, L. Computer "Experiments" on Classical Fluids. I. Thermodynamical Properties of Lennard-Jones Molecules. Phys. Rev. 1967, 159, 98-103.

Vidakovic, M.; Sligar, S. G.; Li, H.; Poulos, T. L. Understanding the Role of the Essential Asp251 in Cytochrome P450cam Using Site-Directed Mutagenesis, Crystallography, and Kinetic Solvent Isotope Effect Biochemistry 1998, 37, 9211- 9219.

Volz, T. J.; Rock, D. A.; Jones, J. P. Evidence for Two Different Active Oxygen Species in Cytochrome P450 BM3 Mediated Sulfoxidation and N-Dealkylation Reactions. J. Am. Chem. Soc. 2002, 124, 9724-9725.

Vosko, S. H.; Wilk, L.; Nusair, M. Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Can. J. Phys. 1980, 58, 1200-1211.

Vrijsen, R.; Everaert, L.; Boeye, A. Antiviral activity of flavones and potentiation by ascorbate. J. Gen. Virol. 1988, 69, 1749-1751.

Wachters, A. J. H. Gaussian Basis Set for Molecular Wavefunctions Containing Third-Row Atoms. J. Chem. Phys. 1970, 52, 1033-1036.

Wang, C.; Makela, T.; Hase, T.; Adlercreutz, H.; Kurzer, M. S. Lignans and flavonoids inhibit aromatase in human preadipocytes. J. Steroid. Biochem. Mol. Biol. 1994, 50, 205-212.

Weil, J. A.; Bolton, J. R.; Wertz, J. E. Electron paramagnetic resonance: elementary theory and practical applications. John Wiley & Sons: New York, 1994.

Weiss, H.; Ahlrichs, R.; Häser, M. A direct algorithm for self-consistent-field linear response theory and application to C60: Excitation energies, oscillator strengths, and frequency-dependent polarizabilities. J. Chem. Phys. 1993, 99, 1262-1270.

Williams, C. L.; Stancel, G. M. Estrogens and progestins. In Goodman and Gilman’s The Pharmacological Basis of Therapeutics, Ninth Edition. Hardman, J. G.; Limbird,

266 L. E.; Molinoff, P. B.; Ruddon, R. W.; Goodman Gilman, A. Eds. McGraw-Hill: New York, 1996; 1411-1438.

Yoshizawa, K.; Kagawa, Y; Shiota, Y. Kinetic Isotope Effects in a C-H Bond Dissociation by the Iron-Oxo Species of Cytochrome P450. J. Phys. Chem. B 2000, 104, 12365-12370.

Yoshizawa, K.; Kamachi, T.; Shiota, Y. A Theoretical Study of the Dynamic Behavior of Alkane Hydroxylation by a Compound I Model of Cytochrome P450. J. Am. Chem. Soc. 2001, 123, 9806-9816.

Yoshizawa, K.; Shiota, Y.; Kagawa, Y. Energetics for the Oxygen Rebound Mechanism of Alkane Hydroxylation by the Iron-Oxo Species of Cytochrome P450. Bull. Chem. Soc. Jpn. 2000, 73, 2669-2673.

Zhou, D.; Korzekwa, K. R.; Poulos, T.; Chen, S. A site-directed mutagenesis study of human placental aromatase. J. Biol. Chem. 1992, 267, 762-768.

Zuber. M. X.; Simpson, E. R.; Waterman, M. R. Expression of bovine 17α- hydroxylase cytochrome P450 cDNA in nosteroiogenic COS-1 cells. Science, 1986, 234, 1258-1261.

267