Time-Resolved Spectroscopic Studies of the Photochemistry of Riboflavin, Aromatic N-Oxides and the Absolute Reactivity of Hydroxyl Radical

Home , Orellanine, Tirapazamine

TIME-RESOLVED SPECTROSCOPIC STUDIES OF THE PHOTOCHEMISTRY OF RIBOFLAVIN, AROMATIC N-OXIDES AND THE ABSOLUTE REACTIVITY OF HYDROXYL RADICAL

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Xiaofeng Shi, M.S.

*****

The Ohio State University 2005

Dissertation Committee: Approved by Professor Matthew S. Platz, Advisor

Professor Christopher M. Hadad ______Professor Dennis E. T. Bong Advisor Graduate Program in Chemistry Professor Thomas R. Lemberger

ABSTRACT

Both nanosecond and ultrafast laser flash photolysis with UV-visible and infrared detection were used to observe the transient species generated photochemically from a number of photosensitizers. The reactions of these transient species were monitored spectroscopically with the aid of theoretical computation.

In the study of photochemical reactions of riboflavin and nucleosides, it was found that triplet riboflavin can be quenched by a silylated guanosine derivative with a rate constant of 1.0 × 108 M-1 s-1. TRIR spectroscopy demonstrated that a hydroflavin radical is formed by an electron transfer-proton transfer mechanism. This sequential electron transfer-proton transfer between triplet riboflavin and guanosine derivative provides the direct observation of the photoinduced oxidative damage of riboflavin to the

DNA nucleobase.

The triplet states of isoquinoline N-oxide and benzocinnoline N-oxide react sluggishly with electron, proton and hydrogen atom donors. These triplets will react with hydroquinone by hydrogen atom transfer (proton coupled electron transfer). Triplet 4- nitroquinoline N-oxide reacts readily with electron donors to from the radical anions as previously reported. The radical anion is protonated on the oxygen atom of the N-oxide

ii group to from a neutral radical. The three N-oxides of this study are not expected to serve as photochemical sources of hydroxyl radical.

Singlet states of tirapazamine and desoxytirapazamine were identified by picosecond time-resolved absorption spectroscopy. The lifetimes of the S1 states and fluorescence quantum yields of aromatic N-oxides were found to be controlled by reversible cyclization to an oxaziridine. The S1 states of TPZ and dTPZ are reduced to radical anions by KSCN, KI and NaN3.

Using LFP-based methodology, we have determined the rate coefficients for the reaction of hydroxyl radical with a number of monocyclic and polycyclic aromatic hydrocarbons in acetonitrile. We observed the reactivities of hydroxyl radical in acetonitrile. For simple aromatic hydrocarbons, the predominant reaction pathway in acetonitrile is the addition of the hydroxyl radical to the aromatic ring, rather than hydrogen-atom abstraction from the phenyl or benzylic C-H positions. Structure- reactivity analysis, based upon frontier molecular orbital and state correlation models indicate that charge-transfer interactions between hydroxyl radical and a given arene play an important role in the stabilization of the transition state for the reaction

α-Alkoxy and hydroxy radicals were generated through thermal and photochemical reactions. Both the product analysis of the thermal reaction and the direct observation of the transient species involved suggest that α-hydroxy radical can very efficiently react with TPZ to form the TPZ-H radical, probably through a direct hydrogen atom exchange between TPZ and ketyl radicals. α-Alkoxy radicals can not proceed through this mechanism.

iii

Dedicated to My Family

ACKNOWLEDGMENTS

First of all, I would like to thank my advisor, Professor Matthew Platz, for his guidance, support and encouragement. He has always been inspiring and thoughtful in science, patient and progressive as a mentor, as well as caring and considerate as a friend.

He has been truly a role-model for me, in many different aspects.

I want to thank Professor Christopher Hadad for teaching me computational chemistry techniques. He has always been available to answer my questions inside and outside science. I am also grateful to the professors and staff members in this department who helped me in various ways.

Many of the research projects in this dissertation are collaborations with current and formal members in the Platz and Hadad group, from which I learned team work and effective communication. I worked with Drs. Christopher Martin and Meng-Lin Tsao in the riboflavin project, Dr. James Poole and Matthew DeMatteo in the hydroxyl radical project, Drs. Jin Liu and James Poole in the tirapazamine project. The Ohio Laboratory of Kinetic Spectrometry in Bowling Green State University and The Center for Chemical and Biophysical Dynamics provided ultrafast laser spectroscopic instruments for this work. I would like to thank Professor Michael A. J. Rodgers and Dr. Alex Gusev in

BGSU, Professor Terry Gustafson and Dr. Gotard Burdzinski of CCBD at OSU, and

v Professor Burda Clemens in Case Western Reserve University for their help in the ultrafast spectroscopic measurements. I also thank Ohio Supercomputer Center for providing the resources of my computational work.

Without the help and hands-on teaching of the members in the Platz group from the very first day, this work would not have been possible. I thank all of them, especially

Dr. Meng-Lin Tsao, for teaching every thing about instrumentation and sharing experimental skills with me. I have experienced friendship and family-like warmth from the people in the Platz group. This has been great five years and will always be in the best of my memory.

Finally, I want to thank my family: my wife and my parents, for their utmost love, concern, understanding and support. They are the source of my power for every step.

VITA

September 13, 1975...... Born – Zhangjiagang, Jiangsu Province, China

1997...... B.S., Chemistry, Nanjing University, China

2000...... M.S., Chemistry, Nanjing University, China

2000- 2005...... Graduate Teaching and Research Associate The Ohio State University

PUBLICATIONS

Research Publications

1. DeMatteo, Matthew P.; Poole, James S.; Shi, Xiaofeng; Sachdeva, Rakesh; Hatcher, Patrick G.; Hadad, Christopher M.; Platz, Matthew S., “On the Electrophilicity of Hydroxyl Radical: A Laser Flash Photolysis and Computational Study”, J. Am. Chem. Soc. 2005, 127, 7094-7109

2. Poole, James S.; Shi, Xiaofeng,; Hadad, Christopher M.; Platz, Matthew S., “Reaction of Hydroxyl Radical with Aromatic Hydrocarbons in Non-Aqueous Solutions – a Laser Flash Photolysis Study in Acetonitrile”, J.Phy. Chem. A 2005 109, 2547-2554

3. Shi, Xiaofeng, Poole, James S.; Emenike, Ejeoma; Burdzinski, Gotard; Platz, Matthew S., “Time Resolved Spectroscopy of the Excited Singlet States of Tirapazamine and Desoxytirapazamine”, J. Phy. Chem. A, 2005 109, 1491-1496

4. Shi, Xiaofeng, Platz, Matthew S., “Time Resolved Spectroscopy of Some Aromatic N-Oxide Triplets, Radical Anions and Related Radicals”, J. Phys. Chem. B 2004, 108, 4385-4391

vii 5. Martin, Christopher B.; Shi, Xiaofeng; Tsao, Meng-Lin; Karweik, Dale; Brooke, James; Hadad, Christopher M.; Platz, Matthew S., “The Photochemistry of Riboflavin Tetraacetate and Nucleosides. A Study Using Density Functional Theory, Laser Flash Photolysis, Fluorescence, UV-Vis, and Time Resolved Infrared Spectroscopy”, J. Phy. Chem. B 2002, 106, 10263-10271

6. Wan, Shuang-Yi; Shi, Xiao-Feng; Xia, Jiang; Sun, Wei-Yin, “Studies on the Hydrolysis of p-Nitrophenyl Acetate Catalyzed by Zinc(II) Complexes with an S2N2 Binding Set”, Main Group Metal Chemistry 2001, 24, 107-110

7. Shi, Xiaofeng; Sun, Weiyin; Zhang, Li; Li, Chongde, “NH-S Hydrogen Bonding in Zinc Enzyme Model Complex with S2N2 Binding Set Studied by Normal Coordinate Analysis of Vibrational Spectra”, Spectrochimica Acta A 2000, 56A(3), 603-613

8. Sun, Wei-Yin; Shi, Xiao-Feng; Zhang, Li; Hu, Jun; Wei, Jin-Hua, “Aromatic C-H-- -S Interaction in the Arenethiolate Complexes of Cadmium(II) with S2N2 Donor Set Evidenced from 113Cd NMR Spectroscopy”, J. Inorg. Biochem. 1999, 76, 259-263

FIELDS OF STUDY

Major Field: Chemistry

viii

TABLE OF CONTENTS

Page

Abstract...... ii

Dedication...... iv

Acknowledgments ...... v

Vita ...... vii

List of Tables...... xii

List of Figures ...... xiii

Chapters

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

1.1 Photosensitization…………………………………………………………….1 1.2 Reaction Oxygen Species and Hydroxyl Radical…………………………….7 1.3 Riboflavin…………………………………………………………………….9 1.4 Tirapazamine………………………………………………………………..11 1.5 Organization of This Dissertaton…………………………………………...13

2. General Experimental and Computational Methods and Protocols……………..17

2.1 Experimental Details………………………………………………………..17 2.1.1 Nanosecond LFP System with UV-vis Detection…………………..17 2.1.2 Nano-second Time Resolved Infrared Spectroscopy……………….18 2.1.3 Ultrafast Transient UV-Vis Spectroscopy………………………….19 2.1.4 Fluorescence Measurement…………………………………………20 2.2 Computational Details………………………………………………………22

ix 3. Photochemical Reaction of Riboflavin Tetraacetate and Nucleosides………..…23

3.1 Introduction……………………………………………………..…………..23 3.2 Instrumental…………………………………………………………………26 3.3 Results and Discussion……………………………………………………...29 3.4 Conclusion…………………………………………………………………..42

4 Transient UV-vis and Time Resolved Infrared Studies of Some Triplet Aromatic N-oxides …………………………………………………………………………44

4.1 Introduction…………………………………………………………………44 4.2 Experimental………………………………………………………………..49 4.3 Results and Discussion……………………………………………………...50 4.3.1 Isoquinoline N-oxide………………………………………………..50 4.3.2 Benzo[c]cinnoline N-oxide…………………………………………54 4.3.2 4-nitroquinoline N-oxide……………………………………………56 4.4 Conclusion…………………………………………………………………..70

5. Time-resolved Spectroscopy of the Singlet Excited States of Tirapazamine and Desoxytirapazamines and their reduction……………………………………….71

5.1 Introduciton…………………………………………………………………71 5.2 Experimental………………………………………………………………..72 5.3 Results……………………………………………………………….……...75 5.4 Discussion…………………………………………………………………..92 5.5 Conclusions…………………………………………………………………97

6. Reactions of Hydroxyl Radicals with Aromatic Hydrocarbons in Non-Aqueous Solutions ………………………………………………………………………..98

6.1 Introduction…………………………………………………………………98 6.2 Experimental………………………………………………………………100 6.3 Results and Discussion…………………………………………………….101 6.3.1 Laser Flash Photolysis…………………………………………….101 6.3.2 The Rate Constant Differences in Water and Acetonitrile………..117 6.3.3..Measurements of Rate Constants of Hydroxyl Radicals with Hydrocarbons……………………………………………………...120 6.4 Conclusions………………………………………………………………..131

x 7. The Reactions between Radicals and Tirapazamine: Studies of the Mechanism of Action of Tirapazamine…………………………………………………...... 133

7.1 Introduction…..……………………………………………………………133 7.2 Experimental…...………………………………………………………….142 7.3 Results……………………………………………………………………..143 7.3.1 Reaction of TPZ with Benzoketyl and/or Acetoketyl Radical…….143 7.3.2 Thermal Reactions of α-Hydroxy(alkoxy) Radical with TPZ…….168 7.4 Discussion…...…………………………………………………………….171 7.4.1 Thermal Generation of Alkyl Radicals and Their Reactivities……171 7.4.2 Reactions of α-Hydroxy Radicals with TPZ and dTPZ…………...177 7.4 Conclusion…………………………………………………………………178

List of References………………………………………………………………….…...179

LIST OF TABLES

Table Page

5.1 Rates of reaction for electron transfer between desoxytirapazamine and various substrates determined from Stern-Volmer analysis of fluorescence quenching…………………………………………………… 87

5.2 Summary of Photophysical and Computational Data…………………….. 93

6.1 Kinetic data for reaction of hydroxyl radical with aromatic hydrocarbons in acetonitrile (ACN) and water (aq)……………………………………... 116

6.2 Comparison of hydroxyl radical reacting with naphthalene and benzene in different solvents………………………………………………………. 119

6.3 Comparison of the ratio of reaction rate constants of hydroxyl with benzene in water and acetonitrile…………………………………………. 119

6.4 Summary of kinetic data for reaction of hydroxyl radical with aromatic hydrocarbons……………………………………………………………… 122

7.1 Product distribution of TPZ refluxed with benzoyl peroxide in different solvents…………………………………………………………………… 170

7.2 Reaction rate constants of benzophenone ketyl and/or acetone ketyl radical with TPZ………………………………………………………….. 177

xii

LIST OF FIGURES

Figure Page

1.1 Jablonski Diagram of general mechanisms of photosensitizer, adapted from Reference 1……………………………….………………………… 3

1.2 Mechanism of action of Porfimer and 8-MOP…………………………… 6

3.1 Stern-Volmer fluorescence quenching of RBTA by G' in CH2Cl2……...... 24

3.2 Differential UV-vis absorption spectra of 0.12 mM RBTA in CH2Cl2 with 214 mM G', indole, and p-cresol. A scaled (1/10) absolute absorption spectrum of RBTA in CH2Cl2 is shown as a reference……….. 30

3.3 (a) TRIR differential spectra of RBTA in CD3CN upon 355 nm LFP. (top), (b) TRIR differential spectra of RBTA with indole in CD3CN upon 355 nm LFP. (middle), (c) TRIR differential spectra of 2.5 mM RBTA with 10 mM G' in deoxygenated methylene chloride upon 355 nm LFP (bottom)…………………………………………………………………... 33

3.4 Kinetic traces recorded at 1488 cm-1 upon 355 nm LFP of RBTA with (top) and without (bottom) G’……………………………………………. 34

3.5 Transient UV-Vis absorption spectra produced upon LFP (355 nm) of RBTA (top) and RBTA + 10 mM G’ (bottom) was recorded 1 µs after the laser pulse…………………………………………………………….. 38

3.6 Decay of the signal at 388 nm after 355 nm laser flash photolysis of RBTA (top) and of RBTA with 10 mM G’ in methylene chloride. (bottom)…………………………………………………………………... 41

3.7 Plot of kobs vs. [G'] (kobs is the decay constant of 388 nm signal in deoxygenated methylene chloride of RBTA + G’). The slope of the line 3 8 -1 -1 is k G' = 1.0 x 10 M s ………………………………………………….. 42

xiii Figure Page

4.1 The transient UV-vis spectrum of triplet 11 upon 355 nm LFP in methanol………………………………………………………………….. 51

4.2 Calculated reaction enthalpies of different reaction pathways of 311*…… 53

4.3 Transient spectrum produced by LFP (355 nm) of 0.1 mM 12. The spectrum was recorded 5 ns after laser pulse over a 30 ns time window… 54

4.4 Calculated reaction enthalpies of different reaction pathways of 312*…… 55

4.5 The transient spectra of 313* (0.35 mM) produced upon LFP (355 nm). The spectrum was recorded immediately after laser pulse over a 45 ns time window. Solvent: 1, CH2Cl2; 2, benzene; 3, water………………...... 57

4.6 The Calculated UV-Vis spectra of 313* and 13•¯ in the gas phase………. 57

4.7 The transient spectra produced upon LFP (355 nm) of 0.3 mM of 13 and 8 mM DABCO in acetonitrile. 1: immediately after laser pulse; 2: 500 ns after laser pulse; 3: 3 µs after laser pulse; 4: 10 µs after laser pulse………………………………………………………………………. 58

4.8 The transient spectra produced upon LFP (355 nm) of 0.3 mM of 13 and 8 mM DABCO. 1: immediately after the laser pulse, in acetonitrile; 2: 5 µs after the laser pulse, in 80% acetonitrile, 20% water (vol:vol)………... 59

4.9 The calculated UV-Vis spectra of different isomers of 13H• in acetonitrile………………………………………………………………... 59

4.10 The transient IR spectra produced upon LFP (355 nm) of 3 mM 13 in -1 -1 acetonitrile (1100-1300 cm ) and CH2Cl2 (1300-1400 cm ). Window: 1, 0-1 µs; 2, 3-4 µs…………………………………………………………. 62

4.11 The calculated IR spectra of 1: 13 and 2: 313* in acetonitrile……………. 62

4.12 The transient IR spectra produced upon LFP (355 nm) of 3 mM 13 and 8 mM DABCO in acetonitrile. Window: 1, 0-1 µs; 2, 3-4 µs……………… 63

4.13 The transient IR spectra produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile. Window: 1, 0-2 µs; 2, 6-8 µs……………………………………………………………………... 63

xiv Figure Page

4.14 The calculated IR spectra of 13H• (1) and 13•¯(2) and 13 (3) in acetonitrile………………………………………………………………... 64

4.15 The decay of the 1240 cm-1 transient produced upon LFP (355 nm) of 3 mM 13 and 8 mM DABCO in acetonitrile. Double exponential decay: τ = 0.63 and 3.5 µs, respectively…………………………………………… 64

4.16 The decay of the 1240 cm-1 transient produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile. Double exponential decay: τ = 0.9 and 14 µs, respectively………………………. 65

4.17 The calculated IR spectra of 13H• protonated at 1: O11, 2: O14 and 3: O13 in acetonitrile……………………………………………………………... 67

4.18 The transient IR spectra produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile-d3 upon 355 nm LFP. Window: 1, 0-2 µs; 2, 6-8 µs……………………………………………... 67

4.19 The decay of the 1410 cm-1 absorbing transient produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile-d3. Single exponential decay: τ = 9 µs……………………………………….. 68

5.1 Transient absorption spectrum produced upon excitation (400 nm) of TPZ in acetonitrile under an argon atmosphere, 1: 8.29 ps; 2: 53.23 ps; 3: 218.23 ps after laser pulse………………………………………………… 77

5.2 The kinetic traces of the transient absorptions of 1TPZ*…………………. 77

5.3 Transient spectrum produced upon excitation (355 nm) of 0.9 mM TPZ in water containing 4 M KSCN, 1: 10 ns after laser pulse; 2: 5 µs after laser pulse; 3: calculated UV of TPZ radical anion (B3LYP/6-31G*, gas phase)……………………………………………………………………... 78

5.4 The kinetics monitored at 545 nm and 385 nm of the transient species whose spectra are shown in Figure 5.3…………………………………… 78

5.5 Transient spectrum produced upon excitation (355 nm) of 0.9 mM TPZ in water containing 4 M KSCN and ~ 3 M NH4Cl……………………...... 79

5.6 The kinetics monitored at 545 nm and 385 nm of the transient species whose spectra are shown in Figure 5.5…………………………………… 79

xv Figure Page

5.7 Comparison of the calculated UV-vis absorptions for the possible transients whose spectra are given in Figure 5.6…………………………. 82

5.8 The fluorescence spectra of TPZ (λex= 432 nm, A=0.302) and dTPZ (λex= 422 nm, A=0.301), dTPZ’(λex= 420 nm, A=0.310), QXNO (λex= 400 nm, A=0.305) and PNNO (λex= 422 nm, A=0.308) in acetonitrile at ambient temperature……………………………………………………………….. 83

5.9 Transient absorption spectrum produced upon LFP (400 nm) of dTPZ in acetonitrile, 1: 1.29 ps; 2: 1033 ps after laser pulse………………………. 87

5.10 Transient absorption spectrum produced upon LFP (355 nm) of 0.9 mM dTPZ and saturated KSCN in actonitrile, 1: right after laser pulse; 2: 1 µs after laser pulse; 3: calculated radical anion of dTPZ (B3LYP/6-31G*, gas phase)………………………………………………………………… 88

5.11 Kinetic traces recorded at 440 nm and 470 nm of the species whose spectra are given in Figure 5.10………………………………………….. 88

5.12 Transient absorption spectrum produced upon LFP (355 nm) of 0.9 mM dTPZ and saturated KSCN, in the presence water and NH4Cl in actonitrile…………………………………………………………………. 89

5.13 Kinetic traces recorded at 440 nm and 330 nm of the transient species whose spectra are given in Figure 5.12…………………………………… 89

5.14 Transient absorption kinetic traces after photoexcitation of dTPZ` (a), QXNO (b) and PNNO (c) in acetonitrile at 400 nm…………………….... 91

6.1 Transient absorbance spectra obtained following laser flash photolysis of an acetonitrile solution of 0.65 mM 21 (PSH) with 355 nm light in the presence of 15 mM trans-stilbene.………………………………………... 104

6.2 Growth curves measured at 392 nm following photolysis of 0.65 mM 21 in the presence (blue) and absence (red) of 15mM trans-stilbene. The inset shows the transient absorption spectrum after photolysis (355 nm) of 21 at longer time scales in the absence of trans-stilbene………………. 104

6.3 Kinetics curves measured at 392 nm following photolysis of 0.65 mM 21 in the presence (blue) and absence (black) of 15 mM trans-stilbene. The red curve is a difference curve, showing the rapid formation and decay of the intermediate generated by reaction of hydroxyl radical with trans- stilbene……………………………………………………………………. 105

xvi Figure Page

6.4 The transient spectrum of PSH in Freon-113 upon 355 nm LFP, 10 ns after laser pulse…………………………………………………………… 108

6.5 Comparison of kinetic traces monitored at 402/410 nm, of the transient species whose spectra are given in Figure 6.4, in the presence and absence of saturated trans-stilbene……………………………………...... 108

6.6 Observed pseudo-first-order rate constant for the growth of transient absorbance at 392 nm following laser flash photolysis of 21 in acetonitrile as a function of the concentration of trans-stilbene. (Slope= 6.1 ± 0.2 × 109 M-1s-1)……………………………………………………. 112

6.7 Observed pseudo-first-order rate constant for the growth of transient absorbance at 392 nm following laser flash photolysis of 21 in CCl4 as a function of the concentration of trans-stilbene. (Slope = 8.9±1.0 ×109 M- 1s-1)……………………………………………………………………….. 112

6.8 Observed pseudo-first-order rate constant (kobs) for the growth of transient absorbance at 392 nm following LFP of 21 in acetonitrile with 12 mM trans-stilbene as a function of the concentration of benzene. (Slope =1.1 ± 0.1 × 108 M-1s-1)…………………………………………… 115

6.9 Stern-Volmer plots for benzene (open squares), benzene-d6 (open diamonds, offset by -0.5 for clarity) and naphthalene (open circles) as a function of the substrate/trans-stilbene relative concentration…………… 115

6.10 Stern-Volmer data obtained from competitive experiments between trans-stilbene and toluene ( ), m-xylene ( ), and p-xylene ( ) at 298 K. The dotted lines are lines of best fit, and the relative rate coefficients obtained from these experiments are shown in Table 6.4………………… 121

6.11 Correlation of the rate coefficient for the reaction of hydroxyl radical with arenes in acetonitrile with a probability-based structure reactivity parameter. The parent arenes used to define the relative rate coefficients are identified in the text…………………………………………………... 127

6.12 Correlation of the rate coefficient for the reaction of hydroxyl radical with arenes in acetonitrile with the charge transfer structure/reactivity parameter (IPArene – EAHO•). The dashed line and equation correspond to the line of best fit for the data obtained, excluding the data for biphenyl, bibenzyl and diphenylmethane. If those data are included, the line of best fit is y = –0.877x + 14.75, R2 = 0.835……………………………….. 130

xvii Figure Page

7.1 Transient UV-vis absorption of BTA radical (triangles), solid triangles are points corrected for the absorption of the parent dTPZ (solid line), adapted from Reference 48……………………………………………….. 135

7.2 The transient spectrum produced by LFP of 7 mM benzophenone in 2- propanol at 355 nm……………………………………………………….. 145

7.3 The transient spectrum produced by LFP of 7 mM benzophenone in 2- propanol, in the presence of 0.2 mM TPZ at 355 nm…………………….. 145

7.4 The kinetic curves monitored at 470 and 410 nm of transient species whose spectra are given in Figure 7.3……………………………………. 146

7.5 The plot of the observed rate constant of the growth of the transient absorbance at 410 nm in Figure 7.3 versus [TPZ]……………………….. 146

7.6 The transient spectrum produced upon 355 nm LFP of 7 mM benzophenone in dioxane upon 355 nm LFP…………………………….. 149

7.7 The transient spectra produced upon 355 nm LFP of 7 mM benzophenone in dioxane, in the presence of 0.11 mM TPZ…………….. 149

7.8 Kinetic curves recorded at 545, 475 and 405 nm of the transient species whose spectra are given in Figure 7.7……………………………………. 150

7.9 The plot of the observed rate constant of the growth and decay of the transient absorptions at 475 and 545 nm in Figure 7.7 versus [TPZ]…….. 150

7.10 The transient spectra of produced by 355 nm LFP of 7 mM benzophenone in ethanol, in the presence of 0.07 mM TPZ…………… 151

7.11 The plot of the observed rate constant of transient absorption measured at 475 and 550 nm in Figure 7.10 versus [TPZ]…………………………….. 151

7.12 The transient spectra of produced upon 355 nm LFP of 9 mM benzophenone in benzene, in the presence of 0.17 M benzhydrol……….. 152

7.13 The transient spectra produced upon 355 nm LFP of 9 mM benzophenone in benzene, in the presence of 0.17 M benzhydrol and 0.1 mM TPZ…………………………………………………………………... 152

7.14 The transient spectra produced upon 308 nm LFP of 7.5% acetone in 2- propanol, in the presence of 0.1 mM TPZ………………………………... 154

xviii Figure Page

7.15 The plot of the observed rate constant of the growth at 475 and 405 nm in Figure 7.14 versus [TPZ]…………………………………………………. 154

7.16 The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in 2-propanol, in the presence of 0.2 mM TPZ. Top: 320 nm to 530 nm; bottom: 480 nm to 720 nm………………………………………………... 157

7.17 The transient spectra produced by 355 nm LFPof 17% t-butyl peroxide in 2-propanol, in the presence of 0.2 mM TPZ, at longer time scales………. 158

7.18 Plots of the observed rate constants of the growth of transient absorption at 475 and 405 nm in Figure 7.20 versus [TPZ]………………………….. 159

7.19 Plots of the observed rate constants of the growth of the transient absorption at 480 and 350 nm in Figure 7.20 versus [TPZ]……………… 159

7.20 The transient UV-vis absorptions of TPZ radical anion (solid circle) and TPZ-H radical (circle), obtained by pulse radiolysis adapted from Reference 48……………………………………………………………… 160

7.21 The UV-vis absorptions of TPZ-H1, TPZ-H4 and BTA radicals in the gas phase predicted by TD-DFT calculation…………………………………. 160

7.22 The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in dioxane, in the presence of 0.15 mM TPZ…………………………….. 163

7.23 The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in 2-propanol, in the presence of 0.36 mM dTPZ………………………… 166

7.24 Kinetic traces recorded at 350 and 410 nm of the transient species whose spectra are given in Figure 7.25…………………………………………... 166

7.25 Kinetic traces recorded at 350 and 410 nm of the transient species whose spectra are given in Figure 7.25, at longer time scales…………………… 167

7.26 Comparison of UV-vis absorptions of ground states of TPZ, dTPZ, dTPZ’ and ddTPZ in acetonitrile…………………………………………. 167

xix

LIST OF SCHEMES

Scheme Page

1.1 Tirapazamine and its possible mechanism of action……………………. 12

3.1 Molecular structure of riboflavin, lumuflavin and lumichrome………... 24

3.2 Derivation of riboflavin and guanosine………………………………… 26

3.3 Proposed reaction pathway of triplet RBTA with G’…………………... 35

4.1 Some aromatic N-oxides with biological activities…………………….. 48

5.1 Photoinduced reduction of TPZ………………………………………… 80

5.2 Photoinduced reduction of dTPZ……………………………………….. 85

6.1 Photolytic decomposition of N-hydroxypyridin-2-thione………………. 103

6.2 Potential Intermediates from the Reaction of Hydroxyl Radical with trans- stilbene…………………………………………………………… 111

7.1 Proposed different mechanisms of action of tirapazamine……………... 134

7.2 The generation of BTA radical by oxidation of dTPZ and its subsequent reactions with 2-disoxyribose……………………………… 135

7.3 Fixing and repair of carbon-based DNA radical………………………... 138

7.4 Different favored reactions of TPZ and dTPZ with DNA radicals…….. 138

7.5 Denny’s mechanism of TPZ mediated DNA damage………………….. 139

7.6 Comparison of different models of the reaction between a DNA radical and TPZ………………………………………………………………… 141

xx Scheme Page

7.7 Generation and fate of TPZ-H radical by reduction and hydrogen atom exchange………………………………………………………………… 156

7.8 Generation of and possible fate of dTPZ-H radical by hydrogen atom exchange and reduction………………………………………………… 164

7.9 The deoxygenation of TPZ refluxed with azo-t-butane and dioxane…... 168

7.10 Calculated thermodynamics of possible reactions between t-butyl radical and acetonitrile………………………………………………….. 171

7.11 Calculated reaction barriers and enthalpy changes for the addition of t- butyl radical to pyridine N-oxide and TPZ……………………………... 173

7.12 Possible radical species present in refluxing dioxane…………………... 174

7.13 Calculated reaction enthalpies of TPZ with different radical species….. 176

xxi

CHAPTER 1

INTRODUCTION

1.1 Photosensitization

Light is a form of energy which can be described as a beam of photon particles.

The distinctive energies of photons depend inversely on their wavelengths or directly on their frequencies. Light in the UV-vis (200 –800 nm) region has very high energy (~ 36

–143 kcal/mol) which is sufficient to break chemical bonds and facilitate a variety of chemical reactions. Hence, photochemical reactions are extremely important among biological processes, for example, photosynthesis. In photochemical reactions, where light acts as an energetic reagent, photons are absorbed by another chemical reagent and this chemical is excited to a higher electronic state. From these excited states which possess the energy provided by photon, inter or intra-molecular reactions that are unfavorable under normal mild conditions can take place.

This property of photochemical reactions can be very useful for medicinal purposes. A drug functions via specific chemical reactions toward specific substrates in the body. Many of the reactions may not proceed under in vivo conditions, i.e., at the temperature of 37 °C, and at low concentrations due to limited intake dose. If the agent is extremely reactive, the selectivity toward targeted tissues or molecules will usually be lost. Light has the exceptional advantages of being able to be finely tuned both

1 temporally and spatially. The intensity, wavelength, irradiation time, and spatial alignment of light can be precisely controlled by current laser and fiber optic technologies. This precise and convenient “aim and fire” approach, together with the unique chemical reactions between the photo-activated drug with the desired biological substrates, provides great selectivity and control in medical practice. Clinically there is an important category of Photodynamic Therapy (PDT) which utilizes the strategy described above.

The drug which absorbs light is generally called the photosensitizer. The application of photosensitizers is not limited to medical or biological purposes. In this

Chapter, however, we only review photosensitizers in cancer treatment and pathogen inactivation of blood products. In cancer treatment, in order to have high selectivity toward tumor tissues and low toxicity to healthy tissues, the in situ generation of reactive species from the drug (or prodrug) is a very important strategy. The activation usually makes use of some special physiological conditions or endogenous enzymes. A photosensitizer, however, is exogenously activated by light.

The general mechanisms of photochemistry are depicted in a Jablonski diagram as shown in Figure 1.1. Upon absorption of a photon, preferably in the visible region, the sensitizer is excited to a higher electronic state (S2 or higher) or higher vibrational state of

S1. In solution phase, the originally populated excited state relaxes to the first excited state (S1) in picoseconds. Chemistry generally proceeds from this S1 state of the sensitizer. In many cases, however, the S1 state can undergo nonradiative inter-system crossing (ISC) to the T1 state. The relatively long lived T1 state is able to take part in chemical reactions based on electron or hydrogen atom transfer with the generation of

2 radical intermediates. This type of photodynamic reaction involves the direct chemical reaction of the excited state, or subsequent species, with the target, and is called a Type I process. Energy transfer from the triplet state of the photosensitizer to a suitable acceptor, most frequently oxygen, results in generation of highly reactive singlet oxygen involved in photodynamic reactions known as Type II.1

Bio- molecules Free Chemical reactions S1 radicals Type I

1 Σg

1 ∆g

Chemical reactions Type II S 0 3 Σg Photosensitizer oxygen

Figure 1.1. Jablonski Diagram of general mechanisms of photosensitizer, adapted from Reference 1.

There are at least three kinds of damage that can be induced by excitation of the photosensitizer. The oxidative species generated by the photosensitizer can directly kill tumor cells by damaging membranes, nuclei, mitochondria, and other organelles, through

3 necrosis or apoptosis.2, 3 They can also indirectly kill cells by damaging the vasculature which cut the oxygen and nutrient supply to the tumor cells. Finally the inflammatory and immune responses trigged by photosensitizer can also contribute to the anti-tumor action. 4

Generally an ideal photosensitizer should have the following properties. It must be non-toxic and non-mutagenic, and its breakdown products and metabolites should also be non-toxic and non-mutagenic. It must absorb UVA or visible light, and longer wavelength light is preferable. Other properties such as suitable metabolic lifetime, water solubility, eaes to transport across cellular membrane, and inexpensive supply, are also desirable.

In early years photosensitizers commonly used in cancer treatment are based on natural hematoporphyrins or synthetic porphyrins. Many other compounds, such as organic dyes (eosin, rose Bengal, methylene blue), aromatic hydrocarbons (naphthalene, anthracenes, biphenyls, quinones), polypyrrolic and metallopolypyrrolic compounds, and transition-metal complexes, have been tested as potential photosensitizers for application in PDT.5, 6 A new generation of photosensitizers have emerged which consist of the photosensitizer moiety linked to biomolecules such as monoclonal antibodies, polypeptide chains, proteins, etc., which allow their selective delivery.7 This strategy offers molecular recognition and specific binding to the tumor.

Currently there are two classes of compounds that are approved for clinical use in the U.S.: porphyrins5, 8 and furocoumrins.9 Both agents are administered to the affected tissues and are then irradiated. The short-lived intermediates do not diffuse into surrounding healthy tissue.

Porfimer (Photofrin) has been used for the treatment of esophageal and endobronchial cancers. This mixture of oligomeric porphyrins is activated by 630 nm laser light, generating singlet-state porfimer, followed by nonradiative intersystem crossing to convert to the more stable triplet state. The triplet state is sufficiently long- lived to undergo intermolecular reactions, most importantly with oxygen. The singlet oxygen generated in this reaction is believed to be the active agent, oxidizing intracellular targets nonspecifically and generating superoxide and hydroxyl radicals, which can also damage cells. Although porfimer is injected intravenously and is distributed throughout the body, selective generation of singlet oxygen in a tumor is achieved by irradiation with a fiber optic diffuser inserted through an endoscope.10-14(Figure 1.2)

8-Methoxypsoralen (8-MOP) utilizes a different strategy to treat cutaneous T cell lymphoma, a disease in which cancerous T cells proliferate in the blood. Instead of bringing light to the tumor via fiber optics, use of this agent brings the tumor cells to the light. 8-MOP is administered orally, and the blood is withdrawn from the patient.

Leukocytes (including T cells) are then irradiated with UVA light outside the body. The treated leukocytes are then recombined with plasma and red blood cells and reinjected into patients. The chemically reactive species in 8-MOP treatment is itself, a DNA intercalator. 8-MOP forms [2+2] cycloadducts with adjacent bases upon irradiation.

Reaction occurs most commonly with thymine C5-C6 double bonds, forming products similar to the thymine cyclobutane dimer. T cells containing cross-linked DNA do not survive treatment.9, 15, 16

These two mechanisms of action for Porfimer and 8-MOP are representative of two most common reactions, Type I and Type II in photodynamic therapies.

R Me hν Porfimer 1Porfimer* Me 1 3 Me O Porfimer* Porfimer* 3 * 1 * Me Porfimer + O2 Porfimer* + O2 1 * O O2 + Substrates Oxizided substrates

Me n=2-6 NaO2C

H OON O HN DNA hν H N O H DNA H O O O O N O O H OMe DNA H OMe H O O O OMe 8-MOP

DNA intercalation

Figure 1.2 Mechanism of action of Porfimer and 8-MOP.

1.2 Reactive Oxygen Species and Hydroxyl Radical

Reactive Oxygen Species (ROS) are oxidative species generated from water or

1 molecular oxygen. These species include molecules like singlet oxygen ( O2) and

• hydrogen peroxide (H2O2), radicals like the hydroxyl radical ( OH), and ions such as

• superoxide anion (O2‒ , which is also a radical). These reactive intermediates are involved in a range of biological and environmental processes. ROS are formed by several different mechanisms. Some of these species are the unavoidable byproduct of cellular respiration. Some molecules are synthesized by dedicated enzymes, and others are generated by ionization from high frequency radiation such as that used in radiotherapy. Finally as mentioned earlier, some ROS are generated by the photosensitizer.

Although ROS have their own biological functions such as cell signaling in low concentration, strong oxidants like the various ROS can damage other molecules and the cell structures, presenting high cytotoxicities. Therefore, ROS-mediated tissue injury is a common pathway for a number of diseases and aging. The other edge of this sword, however, is to kill unhealthy cells or pathogens if a certain selectivity mechanism can be achieved. Type II photosensitizers completely rely on the generation of singlet oxygen to function. Porphyrins described above are examples of a singlet oxygen generator when locally sensitized by light. Methylene blue is another case of a photosensitized singlet oxygen generator, and it has been used to decontaminate fresh frozen plasma.17

Mitomycin C, a quinone antitumor antibiotic, is found to generate hydrogen peroxide, hydroxyl radical and superoxide in the presence of oxygen, which is believed to be partially responsible for its cytotoxicity under aerobic conditions.18-21 Recently, another

7 anti-tumor agent, tirapazamine, is proposed to generate hydroxyl radical in hypoxic

22 cells. Riboflavin (vitamin B2) is being developed as a pathogen eradicator for the sterilization of blood products.23 Riboflavin is also well known for its ability to generate singlet oxygen, hydrogen peroxide and superoxide when exposed to visible light.24, 25

Hydroxyl radical is an important and the most reactive member of the ROS class.

It is a well known active species induced by ionizing radiation,26 and damage by hydroxyl radicals can lead to mutation as well as cell death. In addition, hydroxyl radical is also important in atmospheric, combustion, biological and geomedia transformations of organic substrates. Of particular interest to environmental chemists is the observation that hydroxyl radical, and other reactive intermediates, may be generated photochemically from dissolved humic materials in water,27 and these species may assist in the decomposition of persistent environmental pollutants, such as polycyclic aromatic hydrocarbons (PAHs).

Hydroxyl radical, especially in its reactions with DNA,28-32 is often categorized as being electrophilic in character; however, other kinetic correlations33 with non-aromatic organic substrates suggest that hydroxyl radical may not always be electrophilic in its addition reactions. Therefore, besides the generation of hydroxyl radicals, the fundamental reactive patterns of hydroxyl radical towards a variety of substrates attract our attention.

1.3 Riboflavin

Riboflavin is a photosensitizer which has recently received enormous attention from industry as a pathogen inactivator for the sterilization of blood products. Riboflavin itself is a vitamin (B2), and exhibits a variety of photochemical reactions in solution when irradiated with visible light. It is converted to lumichrome upon photolysis in neutral aqueous solutions, and into lumiflavin in alkaline conditions.34, 35 Exposure of naturally occurring riboflavin in food and the human body to visible light promotes chemical reactions of riboflavin to produce metabolic break down products such as lumichrome.36-

Tens of millions of blood components of blood derived from apparently healthy donors are transfused to millions of Americans annually. Although these blood products are extensively tested in the U.S. (as in other parts of the developed world) for the presence of pathogens prior to administration, there still exists a small, but finite risk of transmission of infectious agents in a transfusion. This risk primarily comes from the

“window” period, the period of time between the infection of a donor and the development of detectable levels of antibodies. Sadly, in many underdeveloped countries, blood products are still not screened prior to administration due to the high cost of the procedure. Therefore, an inexpensive, simple and safe protocol to inactivate pathogens in blood will be tremendously beneficial for both the developed and developing countries.

Navigant Technologies (formerly Gambro BCT) has shown that riboflavin can effectively inactivate intracellular and extracellular HIV and other pathogenic viruses and bacteria in the presence of plasma proteins, platelets, and red blood cells with excellent recoveries of the properties of the transfusable blood products.23 They suggest that

9 nucleic acids are one target of riboflavin inactivation.39 It has also been shown that riboflavin sensitizes nicks and crosslinks in DNA.40

Riboflavin possesses some natural virtues which make it a good candidate as a photosensitizer. Riboflavin is an essential component of the diet and is abundant in nature. Riboflavin is “Generally Regarded As Safe” by the FDA. Its photochemical breakdown products (lumichrome and lumiflavin) are metabolites of riboflavin and thus are formed naturally. Therefore, riboflavin as a photosensitizer is much more likely to be more efficacious, safer, and less expensive over other synthetic sensitizers.

Riboflavin is known to form adducts with proteins, most likely between the flavin and tryptophan residues.41, 42 However, in photosensitized pathogen eradication, the nucleic acid should be the primary target because all potential pathogens present in the blood supply (viruses, bacteria, and parasites) contain either genomic RNA or DNA, single or double stranded and either enveloped or noneveloped. The most important components in blood products, plasma proteins, platelets, and red cells do not contain nucleic acids. Therefore, a dominant nucleic acid-based chemistry of the sensitizer will have a natural selectivity toward pathogens while keeping the integrity of useful blood components. However, it is also well known that long-lived oxidants such as hydrogen peroxide and superoxide ion and short lived singlet oxygen are produced when riboflavin in water or growth medium are exposed to visible light. Thus, whether riboflavin acts as a Type II photosensitizer, which is based on the generation of ROS such as singlet oxygen or hydrogen peroxide, or as a Type I sensitizer, in which the excited riboflavin or its following photoproduct attacks the nucleic acid, or a combination of both, remains unknown. For Type I mechanisms, whether the excited state of riboflavin abstracts

10 hydrogen atoms from a ribose, thereby damaging the sugar moiety, or undergoes redox reaction with the nucleic bases, is also unclear.

1.4 Tirapazamine

Hypoxia, defined as a low concentration of oxygen in tissues, is a very common occurrence in solid tumor tissues. The origin of hypoxia is the imperfect blood vessel network around tumor. Hypoxic cells are resistant to most anti-tumor drugs and radiotherapy for a variety of reasons.43 Most notably among that is the lack of oxygen in tumor cells does not allow the generation of reaction oxygen species induced by radiation.

However, hypoxia also provides a physiological and chemical distinction between healthy and tumor cells, which tumor-selective drugs can exploit.

Brown and Lee discovered the hypoxic cytotoxicity of tirapazamine (TPZ) almost

20 yeas ago, and this is the first compound to be developed specifically as a hypoxic cytotoxin.22 TPZ was a significant advance over the previously known drugs (quinoline- containing aklylating agents and nitroaromatic compounds) because its differential toxicity towards hypoxic cells were larger (100 to 200-fold for hypoxic cells in culture).44

TPZ has been demonstrated to effectively enhance the activity of the chemotherapeutic drug cisplatin and to be complementary to radiotherapy. TPZ has undergone extensive study and is now in Phase III clinical trial with cisplatin for avalanche non-small-cells lung cancer and in Phase II trial with cisplatin based chemoradiotherapy of advanced head and neck cancer.45

O - e (enzyme) O O N O N N H+ N N N N N + OH N NH2 N NH2 N NH2 O N NH2 O OH O O2

Scheme 1.1 Tirapazamine and its possible mechanism of action.

The mechanism of action of tirapazamine has been under active investigation for several years. It is believed that in vivo DNA cleavage by TPZ is due to a radical species generated by one-electron enzymatic reduction (such as cytochrome P450 and cytochrome P450 reductase).22 Brown and co-workers have shown that addition of the radical scavenger dimethyl sulfoxide (DMSO) to hypoxic cells cultures significantly reduce the cytotoxicity of TPZ.22 Based on the analysis of DNA cleavage patterns and the inhibition of radical scavengers on DNA cleavage, Daniels and Gates proposed in

1996 that the active species which causes the cytotoxicity was the hydroxyl radical

(Scheme 1.1).46 However, ESR studies designed to trap the hydroxyl radical were inconclusive.47 Recently, Denny et al proposed another species, the benzotriazinyl radical generated from the trapping of the hydroxyl radical by the adjacent amino group, as the active radical species.48, 49

The clarification of the mechanism of action of TPZ and development of new antitumor agents of this family requires a thorough identification and understanding of the active species which kills the tumor cells. Because of the likely involvement of hydroxyl radical in this bioreductively activated drug, it is of interest to transplant this

12 unique chemistry of aromatic N-oxides to the much studied photochemical generation of

ROS of photosensitizers. The marriage of two types of anti-tumor therapeutic strategies,

PDT and hypoxia-selective drug, may be able to create a new route toward cancer treatment. Since TPZ is a deeply colored compound and has a large absorbance around

475 nm in aqueous solution, it can be easily excited by visible light and studied as a typical photosensitizer. Using Laser Flash Photolysis methods, the short-lived reactive intermediates may be directly detected and identified spectroscopically with the aid of theoretical prediction.

Many aromatic N-oxides possess some unique biological properties, such as carcinogenesis, mutagenesis, and anti-tumor properties.50 Thus, we focused our attention on this entire family of compounds, and hoped to gather some knowledge of the intrinsic structure-activity relationship governing hydroxyl radical release, electron-transfer reactions and hypoxic selectivity.

1.5 Methodologies to study short-lived reactive intermediates

The chemistry of reactive intermediates is a focal point of mechanistic organic chemistry. It also provides a molecular view of biological chemistry and greatly influences modern synthetic chemistry. In the past thirty years, there has been a shift from product-driven studies, in which the primary information was the analysis of reaction products followed by the deduction of probable mechanisms and intermediates, to the direct detection of these short lived species.

Laser Flash Photolysis is a powerful tool to directly observe these intermediates and understand their structure and reactivity. When exposed to very short pulses (fs, 10-

15 s to ns, 10-9 s) of laser radiation at a certain wavelength, the photosensitizer is excited to a higher energy state. This can be followed by a number of inter or intra-molecular chemical reactions. Similarly, a carefully designed photoliabile precursor molecule is used to generate reactive intermediates such as carbenes, nitrenes, and radicals. In both cases, a snapshot of the spectra of the samples, mainly UV-Vis and infrared, is taken.

These spectra are then contrasted with those obtained in the absence of laser pulse. The differential spectra, therefore, contain information of the bleaching of the starting material, the decay and growth of the species formed after the laser pulse. The assignment of the unprecedented spectral features is now greatly facilitated by theoretical calculations, as will be discussed shortly. The temporal evolution of these spectral features can be resolved to provide kinetic traces, and subsequently, the rate constants of the chemical processes involved.

In principle, almost all observable chemical and physical properties of molecules can be predicted by theoretical calculations based on the wave functions of the electronic structure (ab initio) and numerical electron density (Density Functional Theory, DFT), as long as the quantum mechanical operator of that specific property is known. Among these properties, predictions of geometries, enthalpies, and spectra are the most important and useful to physical organic chemists. Enthalpies of isomeric structures in both the gas phase and solvent continuum reveal the distribution of these isomers in equilibrium and quantitatively access the thermodynamics of chemical reactions. The transition state calculations predict the transition energy barrier to overcome in order for a chemical reaction to proceed, and therefore estimate the reaction rate constants. The two most

14 frequently studied spectra are transient UV-vis, which is quite sensitive, and infrared, which provides more structural information.

The past two decades have witnessed revolutionary advances in both spectroscopic techniques and in computational hardware and software. Laser and electronics technology have pushed the time resolution of chemical/physical processes down to femtoseconds, where even molecular vibrations can be monitored. Nano-second resolution laser coupled with UV-Vis detection has been a popular tool within the physical organic community for 25 years. The development of both powerful personal computers and gigantic computer clusters, as well as commercialized molecular modeling and quantum mechanical software packages (such as GAUSSIAN) can provide computational recourses with exponentially increasing capabilities. The combination and interdependence of experiment and theory has now influenced the entire course of physical organic studies from the optimization of experimental conditions and observables, to the interpretation of ambiguous spectral data, and finally to the rational design of new chemical systems.

In this dissertation, we adopt both spectroscopic and computational methods to attempt to understand the generation and reactivities of reactive species generated from some biologically interesting photosensitizers.

1.6 Organization of this Dissertation

This dissertation is organized primarily according to different chemical systems with the theme of photosensitization and the generation and reactivity of hydroxyl radical.

In Chapter 1, photosensitizer, ROS, riboflavin, tirapazamine and methodologies in

15 physical organic chemistry are introduced. Chapter 2 describes the common LFP, transient UV-Vis and time-resolved protocols, as well as computational methods used throughout this dissertation. Chapter 3 is the study of the triplet state of riboflavin and a guanosine derivative. Chapter 4 covers the reactions of the triplet states of aromatic N- oxides. Chapter 5 deals with the photochemistry of tirapazamine and desoxytirapazamine.

Chapter 6 studies the generation and reactivity of hydroxyl radical in nonaqueous solutions. Chapter 7 focuses on the reaction of alkyl radical with tirapazamine and its mechanism of action.

CHAPTER 2

GENERAL EXPERIMETNAL AND CMPUTATIONAL METHODS AND

PROTOCAL

2.1 Experimental Details

2.1.1 Nanosecond LFP system with UV-Vis detection

The nanosecond Laser Flash system with UV-Vis detection prior to autumn 2004 used an excimer laser (Lambda Physik LPX105EMC, 308 nm, 15 ns) or Nd:YAG laser

(Spectra Physics LAB-150-10, ~5 ns, 355 or 266 nm) was used as the excitation light source. The measurement beam was supplied by a PTI 150 W xenon arc lamp with a

LPS 210 power supply, LPS 221 stand alone igniter, A-500 compact arc lamp housing, and MCP-2010 pulser, which allowed for controlled pulsing of the arc lamp with pulsed

0.5-2.0 ms in duration and up to 160 amps in amplitude.

The pulse light from a Xeon Lamp was focused onto a single ARC SP-308 monochromator/spectrograph, with a 1-015-300 grating. This model features dual ports, one with a slit and a photomultiplier for kinetic measurements and the other with a flat field and a Roper ICCD-Max 512T digital ICCD camera for spectroscopic measurements with up to 2 ns temporal resolution. The single monochromator/spectrograph negates the need for separate optimization of kinetic or spectral measurement system alignment, thus facilitating the usage of both types of measurements. The ICCD controller is directly

17 interfaced to the computer using the Roper WinView software and ST-133A controller.

Kinetic data acquisition uses a Tektronix TDS 680C 5Gs/s 1GHz oscilloscope directly interfaced via a National PCI-GPIB to a computer running a custom LabView control and acquisition program. Laser, arc lamp, shutter and other timing and control signals are routed through a National Instruments PCI-6602 DAQ interface. The measurement beam is supplied by a PTI 150 W xenon arc lamp with a LPS 210 power supply, LPS 221 stand alone igniter, A-500 compact arc lamp housing, and MCP-2010 pulser, which allows for controlled pulsing of the arc lamp with pulsed 0.5-2.0 ms in duration and up to 160 amps in amplitude.

For experiments carried out after autumn 2004, the arc lamp and pulsing system was slightly modified. A 150 W xenon arc lamp (Applied Photophysics) was used in the pulsed mode 0.5 ms in duration with a 1 Hz repetition rate.

2.1.2 Nano-second Time Resolved Infrared Spectroscopy

TRIR experiments were conducted with a JASCO TRIR-1000 dispersive-type IR spectrometer with 16 cm-1. The sample was excited by 355 or 266 nm laser pulses of a

Nd:YAG laser (97 Hz repetition rate, 0.5-0.7 mJ/pulse power), which is crossed with the broadband output of a MoSi2 IR source (JASCO). The intensity change of the IR light induced by photoexcitation is monitored as a function of time by an MCT photovoltaic IR detector (Kolmar Technologies, KMPV11-1-J1), with a 50 ns rise time amplified with a low noise NF Electronic Instruments 5307 differential amplifier, and digitized with a

Tektronix TDS784D oscilloscope. The TRIR spectrum is analyzed by the IGOR PRO program (Wavemetrics Inc.) in the form of a difference spectrum:

∆At = − log (1 + ∆It / I) where It is the intensity change induced by photoreaction at time t, and I is the IR intensity for the sample without photoexcitation. Thus, depletion of reactant and formation of transient intermediates or products lead to negative and positive signals, respectively.

2.1.3 Ultrafast Transient UV-Vis Spectroscopy

Picosecond time-resolved measurements were performed at the Ohio Laboratory for Kinetic Spectrometry at Bowling Green State University and the Center for Chemical and Biophysical Dynamics of The Ohio State University.

The instrumentation setup used for the ultrafast transient absorption spectrometric experiments at BGSU has used an output of a Ti:sapphire laser (Spectra-Physics,

Hurricane) (fwhm = 150 fs), which was coupled into a second-harmonic generator (CSK

Super Tripler) to obtain the 400 nm excitation wavelength. The energy of the probe pulse was less than 1 J/cm2 at the sample. The pump beam used was approximately 5 J/pulse with a spot size of 1-2 mm diameter. After the sample cell both beams were coupled into

200 m fiber optic cables and input to a CCD spectrograph (Ocean Optics, S2000-UV-vis) to obtain time-resolved spectral information (425-800 nm). Five thousand excitation pulses were averaged to obtain the transient spectrum at a particular delay time. From the accumulated spectral data kinetic traces at different wavelengths were assembled. The sample flow cell had an optical path of 2 mm and was connected to a solution reservoir and pump system. Sample solutions were prepared with an absorbance of 0.2-1.0 at the

19 excitation wavelength and studied under an argon atmosphere. All measurements were conducted at ambient temperature, 22 ± 2 oC.

The Ohio State University Center for Chemical and Biophysical Dynamics consists of a short pulse oscillator (Coherent/Positive Light, Mira) generating ~30 fs pulses at ~800 nm that seeds a high-energy regenerative amplifier (Coherent/Positive

Light, Legend HE USP). The regenerative amplifier produces ~2.5 mJ, ~40 fs pulses at 1 kHz. The remaining fundamental is used for harmonic and white light generation. The pump beam is chopped with a frequency of 333 Hz. The probe beam is derived from the small portion of the fundamental output (800 nm). It is passed through an optical delay line consisting of a retroreflector mounted on a computer-controlled motorized translation stage. The probe beam is then used to generate a white light continuum in a 1 mm sapphire plate. This is followed by interference filtering. Transmission signals were detected by Si photodiodes and measured with a digital lock-in amplifier (SRS 830). The samples were circulated in a flow cell; the optical path length was 1 mm. To avoid polarization effects, the angle between polarizations of the pump beam and the probe

λ beam was set to the magic angle by a /2 plate.

2.1.4 Fluorescence Measurement

Fluorescence lifetime measurement performed in The Ohio State University used time correlated single photon counting (TCSPC) method. The pulse train of a synchronously-pumped cavity-dumped dye laser (Coherent 700 Series; Rhodamine 6-G or DCM) was directed to a beam splitter. A portion of the pulse train was routed to a fast photodiode that registered a "start" signal on a Time-to-Amplitude Converter (TAC;

Tennelec TC-864) after passing through a Constant Fraction Discriminator (CFD;

Tennelec TC-455). This established the time position of each pulse with great precision.

The "start" pulses initiated the charging of a capacitor in the TAC. The remainder of the pulse train was frequency doubled in an appropriate nonlinear crystal (Inrad, RDA) to generate UV excitation within the diazirine absorbance band. Fluorescence was detected at 90° after having passed through a polarization analyzer (oriented at 54.7° relative to the polarization of the laser excitation), a depolarizer, and a subtractive double monochromator (American Holographic DB-10S). The detector was a Microchannel

Plate Photomultiplier Tube (MCP-PMT; Hammamatsu R-2809U-07). The signals arising from single photons were amplified (Minicircuit ZHL-42) and passed through a second

CFD. The CFD output pulses then served as the "stop" signals for the capacitor in the

TAC. The TAC capacitance was transferred to a Multi-Channel Analyzer (MCA;

Tennelec PCA-II), where the data were binned and presented as a histogram. This histogram served as a representation of the fluorescence decay.51

The fluorescence lifetime measured in Bowling Green State University used a nanosecond flashlamp operating under an atmosphere of H2 gas (0.50-0.55 bar, 0.7 nm fwhm, 40 kHz repetition rate), whose output was filtered through a monochromator prior to sample excitation. The emission was gathered at 90° and passed through a second monochromator (±2 nm). The luminescence was measured with a Peltier-cooled (-30 °C),

R955 red-sensitive photomultiplier tube (PMT). Excitation spectra were corrected with a photodiode mounted inside the fluorimeter that continuously measures the Xe lamp output. TCSPC data were analyzed by iterative convolution of the luminescence decay

21 profile with the instrument response function using software provided by Edinburgh

Instruments.52

UV-vis absorption spectra were obtained on a HP 8452 diode array spectrophotometer. Steady state fluorescence spectra measurements and florescence quenching measurements are carried using a SPEX Fluorolog 1680 double spectrometer

(JY Horiba Inc., Edison, NJ).

2.2 Computational Details

All calculations were performed using the Gaussian 98 suite of programs53 at the

Ohio Supercomputer Center (OSC). Unless otherwise noted, geometries were optimized at the unrestricted B3LYP/6-31G* level of theory with single point energies obtained at the B3LYP/6-31+G**//B3LYP/6-31G* level of theory. Vibrational frequency analyses at the B3LYP/6-31G* level were utilized to verify the energetic minima from the stationary points. The zero-point vibrational energy correction, scaled by a factor of 0.9804, and infrared spectra, scaled by a factor of 0.9613, were also obtained from the frequency analysis. Solution structures and energies were computed using the polarizable continuum model (PCM)54 provided by Gaussian 98. The electronic spectra were computed using the time-dependent DFT theory of Gaussian 98 at the B3LYP/6-31G* level, and 10 transitions were included.

CHAPTER 3

PHOTOCHEMICAL REACTIONS OF RIBOFLAVIN TETRAACETATE AND

NUCLEOSIDES

Reproduced in part with permission from J. Phys. Chem. B 2002, 106, 10263

3.1 Introduction

Riboflavin (RB, 1), also known as vitamin B2, is a yellow-orange compound. The molecular structure of riboflavin consists of a chromophoric flavin core, which is representative of the electronic properties of the whole molecule, and a hydrophilic ribityl chain (Scheme 3.1). Riboflavin undergoes numerous reactions when exposed to solar radiation. In neutral aqueous solution, riboflavin forms lumichrome (LC, 2) upon photolysis, and in alkaline solution, lumiflavin (LF, 3) is produced.34, 35 Riboflavin is also known to be able to generate reactive oxygen species when irradiated. 41, 42

HO OH

OH CH3 H

H3C N N O H3C N N O H3C N N O N N N H3C N H H3C N H H3C N H O O O RB, 1 LF, 2 LC, 3

Scheme 3.1. Molecular structure of riboflavin, lumuflavin and lumichrome.

Riboflavin undergoes rich photochemistry with biologically relevant molecules such as amino acids and nucleic acids. It is believed that in human serum albumin, linkages of flavin and electron-rich amino acid residues, such as tryptophan, can be formed.42 This kind of flavin-tryptophan is also believed to occur in the lens of bovine and human eyes, and is associated with aging of the eye and the formation of cataracts. 41

Photolysis of riboflavin and nucleic acids was reported to lead to the formation of flavin adducts to AT rich regions of DNA40 and to the generation of single-strand breaks.55-57 The single-strand breaks may be due to the action of superoxide anion or hydrogen peroxide,56, 58 which are also formed upon photolysis of riboflavin. Riboflavin and light sensitize the killing of mammalian cells.59 Analysis of the nucleic acids of these cells reveals guanine oxidation and strand breaks. 60, 61 Interest in this photochemistry has recently surged due to reports that riboflavin can sensitize the inactivation of viruses in the presence of transfusable blood products.23

However, as riboflavin seems to be able to selectively deactivate viruses in blood products with acceptable recovery of plasma protein, platelet, and red cell function,39 the

24 reaction of riboflavin with DNA or nucleotides of pathogens should be equally efficient, if not more, compared with the reaction with the tryptophan moiety of proteins. This has prompted the present study of flavin photochemistry using absorption and fluorescence spectroscopy, laser flash photolysis with UV-vis and infrared spectroscopic detection, combined with density functional theory (DFT) calculations performed earlier in this research group.62 We selected guanosine as the photochemical target for riboflavin, since guanosine has the lowest redox potential among the four nucleosides and is well known to be primary target for oxidative reactions towards DNA. Guanosine was silylated to form 2’,3’,5’-tri(dimethyl-t-butylsilyl)guanosine (G’, 5), which is more soluble than guanosine in organic solvents such as methylene chloride (Scheme 3.2), feasible for time- resolved infrared experiments. For the same reason, riboflavin was acetylated to riboflavin tetraacetate (RBTA, 4), and previous studies have demonstrated that this modification will not change the electronic properties of this molecule.62

OH OCOCH3

HO H3COCO OH OCOCH3

OH OCOCH3 Ac2O, ∆ H3C N N O H3C N N O AcOH N N H3C N H H3C N H O O 1 4

O O N NH N NH HO N N NH2 t RO N Si(CH3)2 Bu-Cl, ∆ N NH2 O O DMF, imidazole OH OH OR OR t R= Si(CH3)2 Bu 5

Scheme 3.2. Derivation of riboflavin and guanosine.

3.2 Experimental

TRIR experiments were carried out on the instrumentation described in details in

Chapter 2. Briefly, a reservoir of the deoxygenated sample solution (20 mL 2.5 mM

RBTA in CH2Cl2, CH3CN, or CD3CN with or without the desired quantity of G' or indole) is continually circulated between two calcium fluoride salt plates with a 0.5 mm path length. The sample was excited by 355 nm laser pulses of a Nd:YAG laser (97 Hz repetition rate, 0.5-0.7 mJ/pulse power), 26

Fluorescence lifetimes were determined using time correlated single photon counting (TCSPC), in The Ohio State University, as described in Chapter 2.

Fluorescence quenching experiments were carried out using a SPEX Fluorolog 1680 double spectrometer (JY Horiba Inc., Edison, NJ).

The LFP experiments with UV-vis detection were performed with the instrumentation described in Chapter 2. Samples analyzed by LFP were prepared by combining various amounts of a quencher solution of known concentration with a stock solution containing RBTA. The mixture was then diluted to achieve the desired volume and degassed by bubbling argon through the solution for a minimum of 10 min. All kinetic data represent the average of triplicate measurements.

Binding of G', indole, and p-cresol to RBTA in CH2Cl2 was detected as a shift of the UV-vis spectrum of RBTA. A solution of the compound (1.8 mL, 0.238 M) suspected to bind to RBTA was added to a solution of RBTA (0.2 mL, 1.2 mM), and the

UV-Vis spectrum was recorded. This spectum was compared to the spectrum of RBTA without a binding agent (0.2 mL RBTA solution and 1.8 mL solvent) and the difference spectrum showing the shift in minima and maxima was interpreted as evidence for binding.

Compound 2',3',4',5'-tetraacetylriboflavin (riboflavin tetraacetate, RBTA) was synthesized according to the method of McCormick.63 Silylated guanosine (G') was synthesized as described by Sheu and Foote.64

Preparation of 2’,3’,5’-Tris((tert-butyldimethylsilyl))guanosine (G’). To a solution of guanosine (dried under vacuum with phosphorus pentoxide over night before use, 2.8 g, 10 mmol) and imidazole (6.8 g, 100 mmol) in 50 mL of anhydrous

27 dimethylformamide (DMF) was added 7.5 g (50 mmol) of t-butyldimethylsilyl chloride.

The reaction mixture was stirred at room temperature under argon atmosphere for 40 h.

The resulting mixture was poured into EtOAc-H2O mixture. The organic layer was washed by dilute sodium bicarbonate solution and water for 3 times and then dried over anhydrous sodium sulfate and evaporated to dryness. Column chromatographic purification using 1:l methylene chloride and acetonitrile gave 5.7 g (80%) of 2’,3’,5’- tris((ter~-butyldimethylsilyl)oxy)guanosine, which was re-crystallized from 95% EtOH

+ to give white needle-like crystals. MS (ESI): 648.340319; Calcd for C28H56O5Si3Na :

l 648.33908. H NMR (δ, ppm, CDCl3): 3.82 (dd, J = 11.6Hz, 2.3Hz, lH,C(5’)-H’), 4.02

(dd, J=11.6 Hz, 3.3Hz, lH,C(5’)-H”), 4.11 (ddd, J =4.3 Hz, 3.3 Hz, 2.3 Hz, lH, C(4’)-H),

4.32 (dd, J = 4.3 Hz, 4.0 Hz, lH, C(3’)-H), 4.44 (dd, J = 4.0 Hz, 3.9 Hz, C(2’)-H), 5.83

(d, J = 3.9 Hz, lH, C(1’)-H), 6.63 (s, br, 2H, C(2)-NH2), 7.97 (s, 1H, C(8)-H), 12.3 (s, br, lH, N(1)-H), 0.99, 0.94, 0.89 (s, 27H, 3 × tert-butyl), 0.17, 0.16, 0.13, 0.11, 0.06, 0.03 (s,

18H, 6 × Me).

Preparation of 2’,3’,4’,5’-tetraacetylriboflavin (RBTA). Riboflavin (2.4g, 6.4 mmol) was dissolved in a solution of 100 mL of acetic acid and 100 mL of acetic anhydride to which 0.5 mL of perchloric acid was added dropwise. The solution was heated at 40-45 °C for 30 minutes, cooled in an ice bath, and slowly diluted with 200 mL of cold water. The mixture was extracted twice with 200 mL of CH2Cl2 and the collected organic fractions were washed with two 50 mL portions of 5% aqueous sodium bicarbonate. The organic layer then was dried over anhydrous MgSO4, filtered, and solvent was removed under reduced pressure to yield an orange solid, which had no detectable acetic order. The solid was placed under vacuum (< 5mmHg) overnight to

1 yield 2’,3’,4’,5’-tetraacetylriboflavin (3.0g, 86%). H NMR (δ, ppm, CDCl3): 1.53 (s,

3H), 1.69 (s, 3H), 2.01 (s, 3H), 2.38 (3, 3H), 2.50 (s, 3H), 4.19 (dd, J=12.3, 5.4 Hz, 1H),

4.37 (dd, J=12.3, 2.7 Hz, 1H), 4.85 (m, 1H), 5.12 (bm, 1H), 5.36 (m, 2H), 5.60 (m, 1H),

7.49 (s, 1H), 8.01 (s, 1H), 8.32 (s, 1H).

3.3 Results and Discussion

Fluorescence Quenching

We find that G' quenches the fluorescence of RBTA in methylene chloride.

-1 Stern-Volmer analysis of the data yields a kQ value of 5.2 M (Figure 3.1). Using time correlated single photon counting, we find that the fluorescence lifetimes of RBTA are 6 and 12 ns in CH2Cl2 and acetonitrile, respectively. This leads to a value of kQ = 0.87 ×

109 M-1s-1,

Knowles65 has previously reported that 5'-guanosine monophosphate (GMP) and

5'-adenosine monophosphate (AMP) quench the fluorescence of riboflavin in buffered aqueous solutions. His Stern-Volmer analysis of the data led to kQ values of 65.5 and

82.6 M-1, respectively. The fluorescence lifetime of riboflavin in water is 5 ns,66 thus the quenching rate constants with GMP and AMP can be estimated to be 13.1 × 109 and 16.5

9 -1 -1 × 10 M s , respectively. Therefore, the kQ we observed for G’ is 15-fold smaller than the rate constant for quenching of excited singlet riboflavin with GMP in aqueous buffer reported by Knowles.65

Figure 3.1 Stern-Volmer fluorescence quenching of RBTA by G' in CH2Cl2.

Figure 3.2. Differential UV-vis absorption spectra of 0.12 mM RBTA in CH2Cl2 with 214 mM G', indole, and p-cresol. A scaled (1/10) absolute absorption spectrum of RBTA in CH2Cl2 is shown as a reference.

Dardare found that riboflavin associates with GMP in aqueous buffer.67 This association is manifest as a GMP dependent shift in the absorption spectrum of riboflavin. As shown in Figure 3.2, indole and p-cresol (200 mM) will induce similar shifts in the spectrum of RBTA in methylene chloride, but a comparable concentration of

G' does not. It seems clear that RBTA and indole (and p-cresol) associate in methylene chloride, but there is no evidence of association of RBTA and G' in this solvent. Thus, the greater fluorescence quenching rate constants observed by Knowles likely reflects the pre-association of riboflavin and GMP in aqueous buffer solution.

Time-Resolved Infrared Spectroscopy

Martin et al have previously reported that laser flash photolysis (LFP) of RBTA at

3 * 355 nm in deoxygenated CD3CN produced the transient spectrum of RBTA with τ = 2

µs. 3RBTA* has intense C=O vibrations at 1652 cm-1 and C=N vibrations at 1484 and

1436 cm-1 (Figure 3.3a), in satisfactory agreement with density functional theory

62 calculations. In the same paper, they also reported that LFP of RBTA in CD3CN containing 20 mM indole produces the TRIR spectrum of neutral radical RBTH, produced by protonation of the flavin radical anion.62 In that study, DFT calculations were used to evaluate the energies of every neutral radical obtained by protonation of each oxygen and nitrogen of the lumiflavin radical anion. The thermodynamically most stable radical (by 8 kcal/mol) corresponds to protonation of N5, as shown for RBTH

(Scheme 3.3). Furthermore, this is the only low energy, neutral flavin radical predicted with a carbonyl vibration close (1650 cm-1) to the experimental value of 1660 cm-1

(Figure 3.3b).

Upon LFP of RBTA in the presence of an electron donor such as sodium iodide,

·- the transient spectrum of the RBTA radical anion (RBTA ) was detected (νC=O = 1636 cm-1, τ = 20 µs). The most important resonance structure of RBTA·- (as deduced by DFT calculations of LF·-)62 is depicted in Scheme 4.3.

LFP (355 nm) of 2.5 mM RBTA in deoxygenated CH2Cl2 containing 10 mM G' produces the transient IR spectrum of the RBTH radical shown in Figure 3.3c. This concentration of G' is not sufficient to quench the excited singlet state of RBTA to a significant extent, based on the fluorescence quenching information. The contour of the transition IR spectra in Figure 3.3c at 8-9 µs is very similar to the spectra in Figure 3.3b, which were assigned to the RBTH radical. The lifetime of the transient at has a long time of tens of microseconds, similar to the carrier of spectra in Figure 3.3b. A few small differences from Figure 3.3b and Figure 3.3c, such as the shoulder peak at 1620 cm-1, are believed to be caused by the presence of neutral G’ radical, as predicted by calculation by

Martin et al. The excellent agreement of the spectral features in Figures 3.3b and 3.3c reveals that RBTH is undoubtedly formed from 3RBTA* by the sequential electron transfer-proton transfer mechanism in Scheme 3.3.

3RBTA* decays exponentially (Figure 3.4) and the observed rate constant of disappearance of G', measured at 1488 cm-1, increases in the presence of G'. From Figure

3.4, one can very crudely estimate that the rate constant for electron transfer (kG', Scheme

3.3) is on the order of 108 M-1 s-1. Because G' quenches 1RBTA*, the yield of 3RBTA* is low and the signal-to-noise of the decay trace becomes less favorable in the presence of

G'.

Figure 3.3 (a) TRIR differential spectra of RBTA in CD3CN upon 355 nm LFP. (top), (b) TRIR differential spectra of RBTA with indole in CD3CN upon 355 nm LFP. (middle), (c) TRIR differential spectra of 2.5 mM RBTA with 10 mM G' in deoxygenated methylene chloride upon 355 nm LFP (bottom).

Figure 3.4. Kinetic traces recorded at 1488 cm-1 upon 355 nm LFP of RBTA with (top) and without (bottom) G’.

OH R 3 N N N O N R'O N N + O N NH2 N H O R = ribityl tetraacetate R'O OR' 3 t *RBTA R' = Si(CH3)2 Bu kG' G'

H-atom electron abstraction transfer

R R N N O N N O

N N N H N H H O RBTH O RBTA

+ + O OH N NH N N R'O N G' N NH2 O N N NH2 R'O OR'

R O N N O + N N N N H N N NH2 H O

RBTH G'

Scheme 3.3. Proposed reaction pathway of triplet RBTA with G’.

However, the formation of RBTH⋅ radical can proceed through two pathways. As shown in Scheme 3.3, a direct hydrogen abstraction, mostly likely from the ribose ring, will lead to the formation of RBTH⋅ radical. The other pathway has two steps: electron transfer between triplet RBTA and G’ to form the radical anion pairs, followed by a proton transfer from G’ radical cation to RBTA radical anion. These two reaction routes produce the same RBTH⋅ radical.

In related experiments, Tsao et al found that adenosine triacetate (ATA) also

6 -1 -1 68 accelerates the decay of triplet RBTA (kATA ≈10 M s ) in acetonitrile. The lower reactivity of ATA relative to G' reflects the relative ease of oxidation of these reagents.69,

70 Ribose tetraacetate (RTA) does not accelerate the decay of 3RBTA*, even at concentrations as large as 100 mM.68 Thus, even though hydrogen-atom abstraction reactions of 3RBTA with sugars are predicted to be exothermic, these reactions are relatively unimportant mechanisms of quenching 3RBTA* with purine nucleosides in solution.

NH2 N O N O

O N N O O O O O

O O O O O O O O

ATA RTA

The similarity of the TRIR spectrum observed upon the LFP of RBTA in the presence of G', indole, and p-cresol leads us to assign the main features of the transient

36 spectrum to the flavin nucleus. The inability of RT to quench the 3RBTA* and the

3 * reduced kQ of ATA lead us to identify the reaction between RBTA and G’ as a sequential electron transfer-proton transfer, just as with indole and p-cresol.

Laser Flash Photolysis with UV-Vis Detection

Flavins have been previously studied by LFP methods with UV-vis detection in aqueous and organic solvents.71 LFP (355 nm) of RBTA in methylene chloride was performed to confirm the electron transfer reaction of 3RBTA* and G’. As expected,

3RBTA* has transient absorption maxima at 300 and 390 nm as well as a broad absorbance above 500 nm (Figure 3.5).71 The presence of G' accelerates the disappearance of the transient absorption (Figures 3.6), although the transient absorption no longer decays to baseline. A plot of the observed rate constant of disappearance of

3 * 8 -1 -1 RBTA versus [G'] is linear (Figure 3.7). The slope of this plot, kG', is 1.0 × 10 M s .

This rate constant is about the same as that estimated from the TRIR kinetic experiment.

Therefore, the transient spectrum observed after 3RBTA* has decayed is assigned to the longer-lived neutral flavin radical RBTH. Triplet RBTA and the radical RBTH have similar UV-vis spectra; hence, 3RBTA* does not decay to baseline in the presence of G'.

Figure 3:5. Transient UV-Vis absorption spectra produced upon LFP (355 nm) of RBTA (top) and RBTA + 10 mM G’ (bottom) was recorded 1 µs after the laser pulse.

The quenching of RBTA triplet has also been studied. The decay rate was monitored at 388 nm as a function of indole concentration in methanol, acetonitrile, and

68 9 -1 -1 methylene chloride. The data show that kindole = 3.2 × 10 M s in acetonitrile, 5.0 ×

109 M-1 s-1 in methanol, and 4.5 × 109 M-1 s-1 in methylene chloride. Triplet RBTA, therefore, reacts much more rapidly with indole than with G' in dichloromethane. This may reflect an intrinsically higher reactivity of indole. It also reflects the fact that indole associates with RBTA, but there is no evidence that G' undergoes complexation at the concentrations employed in this study.

RBTA in methylene chloride was also pulsed with 355 nm light in the presence of

300 mM G'. Under these conditions, greater than 70% of the fluorescence of singlet

RBTA has been quenched. The transient spectrum is that of the neutral hydroflavin radical, RBTH. The intensity of the RBTH spectrum obtained in the presence of 300 mM

G' is only 20% of that obtained in the presence of 10 mM G'. It is possible that the yield of RBTH from 1RBTA* is only 20% that of radical formation from 3RBTA. It seems more likely that G' quenching of 1RBTA* does not produce RBTH on the ns time scale and that the small yield of radical observed is entirely due to the low yield of 3RBTA in the presence of 300 mM G'.

Previous reports by Kuzmin et al72 and Steenken and coworker73 present two very similar situations. Kuzmin et al found that in aqueous solution, a radical cation of pyrene is able to oxidize 2’-deoxyguanosine-5’-monophosphate (dGMP), but not dAMP, dCMP or dTMP. A deprotonation product from the dGMP radical cation, the neutral dGMP(-H) radial was formed in a few microseconds. This kind of G(-H) radical was also detected by Steenken and Candeias when oxidizing guansine in high pH solutions by the SO4•‒ radical generated by pulse radiolysis. Both of their studies indicate a characteristic strong

UV absorption of G(-H) radical at 310 nm, as well as other weaker bands at 350 nm and

470 nm. In Figure 3.5, we can observe a 310 nm peak upon the addition of G’, shifted from the 300 nm triplet absorption. The 350 and 470 nm absorption bands are not observed for the obvious reason that they are buried to the bleaching of the ground state absorption of RBTA. The 388 nm absorption is probably due to the RBTH radical, which happens to overlap with that of RBTA triplet.

As mentioned in the introduction, riboflavin and light sensitize single-strand breaks in DNA. These may be due to processes involving reactive oxygen species or to direct reaction of triplet flavin with the sugar moieties of DNA. The minimum energy geometry of triplet lumiflavin and the most stable neutral radical (LFH·) formed upon

39 partial reduction of the flavin were computed. This radical is calculated to be 8 kcal/mol more stable than any other isomeric neutral radical derived from lumiflavin.62

In early studies of Martin et al,62 the thermodynamic data indicate that the reaction between triplet LF and the hydrogen of a series of simple alcohols and ribose models to form the hydroxyalkyl radical and LFH· are all exothermic with the most favorable being the reaction with 2-propanol. Their calculations demonstrate that the reaction of triplet flavin with the sugar backbone of the nucleic acids residues is thermodynamically feasible. However, experimentally, the hydrogen-atom abstraction from the sugar ring by triplet RBTA was not observed. Instead, only electron transfer followed by a proton transfer was observed for silylated guanosine.

CH3 CH3 N N O H3C N N O H3C + RH NH + R H3C NH N H3C N O H O 3LF* LFH

Figure 3.6 Decay of the signal at 388 nm after 355 nm laser flash photolysis of RBTA (top) and of RBTA with 10 mM G’ in methylene chloride. (bottom)

Figure 3.7. Plot of kobs vs. [G'] (kobs is the decay constant of 388 nm signal in 3 deoxygenated methylene chloride of RBTA + G’). The slope of the line is k G' = 1.0 x 108 M-1s-1.

3.4 Conclusions

3RBTA* can be quenched by a silylated guanosine derivative (G') and the absolute rate constants is 1.0 × 108 M-1 s-1 in methylene chloride at ambient temperature as determined by both transient UV-vis and Time-Resolved Infrared spectroscopy. 3RBTA* reacts with ribose tetraacetate, indole, adenine triacetate, but not ribose tetraacetate.

TRIR spectroscopy demonstrated that 3RBTA* reacts with G' to form a hydroflavin radical RBTH by an electron transfer-proton transfer mechanism. LFP-UV results are consistent with the TRIR data. Both indole and G' quench the fluorescence of 1RBTA* in methylene chloride (τ= 12 ns). In the case of G', singlet chemistry leads to only modest

42 yields of the RBTH radical. Just as the sequential electron transfer-proton transfer between 3RBTA* and p-cresol or indole reveals the high reactivities between riboflavin triplet state and amino residues such as tryptophan and tyrosine, a similar reaction pattern between 3RBTA* and guanosine derivative provides the direct observation of the photoinduced oxidative damage of riboflavin to the DNA nucleobase.

CHAPTER 4

TRANSIENT UV-VIS AND TIME RESOLVED INFRARED STUDIES OF SOME

TRIPLET AROMATIC N-OXIDES

Reproduced in part with permission from J. Phys. Chem. A 2004, 108, 4385.

4.1 Introduction

Tirapazamine is an aromatic-N-oxide used in the treatment of hypoxic tumors.22

It is generally accepted that this drug is enzymatically activated to form a radical anion which is protonated to form a neutral radical.22, 74 Daniels et al75 have postulated that the neutral radical fragments to form hydroxyl radical which subsequently damages cellular

DNA. In this way, the oxygen atom of an aromatic N-oxide is effectively converted to hydroxyl radical (Scheme 4.1).

This special property of TPZ prompted us to review the chemistry of other members of the family of aromatic N-oxides. Although heteroaromatic N-oxides do not have a significant occurrence in nature, many do possess bioactivities. For example, orellanine is a bipyridine dioxide and lethal toxin from the mushroom Cortinensis

Oorellanus. Phenazine di-N-oxide such as iodinin is an antibiotics (Scheme 4.2).76 In addition, N-oxidation is a general process during the metabolization of nitrogen heterocyles, and several N-oxides have been obtained as drug metablites.76

O O2 O O N O N N N N N N N + OH N NH2 N NH2 N NH2 O N NH2 OH OH e-, H+ (enzyme)

Scheme 4.1 Generation of hydroxyl radical, a mechanism of tirapazamine proposed by Daniels and Gates75.

Besides tirapazamine, several other analogue compounds have also been tested and found to have anti-tumor properties, such as 1-cyano-2-amino-6-chlorine-quinoxaline di-N-oxide77 and fused pyrazaine N-oxide (RB90740)78 (Scheme 4.2). These two compounds are bioreductive drugs and are believed to be able to generate nitrogen based radicals, either as the active species or the intermediates of the active species. 4-

Nitroquinoline N-oxide is a well-known mutagen and carcinogen.79 Therefore, many compounds in this family seem to specifically target DNA in cells. There are two common mechanisms of DNA damage by agents with different chemoselectivity: some aim at the base moieties by oxidation, alkylation or cyclization of the purine or pyramidine rings, often accompanied with sequence-specificity. Some reagents attack the backbone of DNA, mostly the sugar moiety, and are usually nonselective. Aromatic

N-oxides generally have relatively high redox potentials and hence are oxidative compounds.80 As proposed by Daniels et al75, tirapazamine is also able to generate hydroxyl radical. Therefore, the mechanism of action exhibited by these aromatic N- oxides interests us greatly.

We adopt a photochemical approach to study these aromatic N-oxides for two reasons:

(1) The activation step of these compounds in vivo is likely to be an enzymatic reduction. As catalysts, enzymes either raise the energies of reactants or lower the energies of the transition states to facilitate reactions that are not facile under in vivo conditions. By photosensitization, we are able to achieve the same goal by energizing these agents with photons so that they are able to overcome reaction barriers that are insurmountable under normal conditions. Under these circumstances, we can consider these agents as photosensitizers.

(2) Secondly, laser photolysis is able to simultaneously generate very high concentrations (up to hundreds of µM) reactive species with very short lifetimes (from ps to ms). This will allow us to spectroscopically detect the reactive species involved directly and to monitor their kinetics.

Reported photochemical reactions of aromatic N-oxides fall into two groups: rearrangement and N-deoxygneation.76 There is now general agreement that the first process usually involves the first excited singlet state and the latter one probably the triplet state. Analogy with the nitrones suggests that an oxaziridine is also the intermediate for the rearrangement of these molecules, and indeed many, but not all, of the observed reactions can be discussed in terms of a series of concerted (sigmatropic and

46 electrocyclic) rearrangments. Although there are a large number of reports of photochemical reactions of this family of compounds, there has been a lack of the basic understanding of the basic photophysics of aromatic N-oxides, and the mechanism of these photoreactions largely is based on hypothesis. Therefore, it is necessary to investigate the photophysics of these compounds and study their energy and electron transfer reactions.

Poole et al have reported that the excited singlet state of tirapazamine will undergo electron transfer reactions.81 The measured singlet lifetime of TPZ is 98 ps in water and 110 ps in acetonitrile. These processes are too rapid to study by nanosecond spectroscopy in our laboratory. Electron-transfer reactions of triplet tirapazamine (T1) are not easily studied because of the low quantum yield of intersystem crossing in the S1 state. That motivated this study of the photochemistry of simpler aromatic N-oxides to learn if their triplet states will undergo electron transfer reactions that will ultimately release hydroxyl radical.

The triplet states of isoquinoline N-oxide (11), benzo[c]cinnoline N-oxide (12) and

4-nitroquinoline N-oxide (13) are formed efficiently upon direct photolysis and the latter compound is biologically active. Thus we chose to study their photoinduced reduction and to test their abilities to release hydroxyl radical.

HO O OH N HO O N O OH N N OH OH O Orellanine Iodinin

O CH 3 N N CN N N N

Cl N NH2 N O O 1-cyano-2-amino-6-chlorine quinoxaline di-N-oxide RB90740

N O O O N N N

NO2 isoquinoline N-oxide Benzo[c]cinnoline N-oxide 4-nitroquinoline N-oxide (11) (12) (13)

Scheme 4.2. Some aromatic N-oxides with biological activities.

4.2 Experimental

Isoquinoline (11) and 4-nitroquinoline N-oxide (13) were purchased from Aldrich.

Benzo[c]cinnoline N-oxide (12) was purchased from Acros. 4-Hydroxyaminoquinoline

N-oxide (16) was purchased from TCI. All of these chemicals were used as received.

Solvents (acetonitrile, dichloromethane and benzene) were distilled over calcium hydride before use.

4-Nitroquinoline (14) was prepared as described in the literature.82 0.34 g 4- nitroquinoline N-oxide (13) and 0.28 mL P(COCH3)3 was dissolved in 200 mL CH2Cl2 in a 500 mL Pyrex flask. The reactant mixture was stirred and irradiated by mercury Hg lamps in a Ray-O-Net photoreactor for 30 minutes under argon. The solution was washed with water to remove phosphate, dried over MgSO4, and evaporated under reduced pressure to give yellow solid (0.28 g, 85%). The product is further purified by running through silica column with EtOAc and hexane to give yellow crystals.

TRIR experiments were performed and analyzed as described in Chapter 2. Each sample solution was prepared with an optical density of 0.6-1.0 with a 0.5 mm path length. A total volume of 20-30 mL of sample was circulated between two barium fluoride or calcium fluoride salt plates. The sample was excited with a Nd:YAG laser of

355 nm wavelength at 97 Hz repetition rate with 0.5~0.7 mJ/pulse.

Laser flash photolysis experiments were also performed on the instrument as described in Chapter 2. Samples were excited with a Spectra Physics LAB-150-10 (~5 ns) water-cooled laser at the third harmonic frequency (355 nm) with a power of ~0.045

J/pulse.

HPLC analyses were performed on a Beckman Coulter liquid chromatography equipped with System Gold 168 diode array detector and controlled using an IBM computer and the 32 Karat 5.0 software package. Separations were achieved on a reverse phase C18 250 mm × 4.6 mm (5µ) column preceded by a guard column. A mixture of

HPLC grade water and methanol (2:3, vol:vol) was used as eluent at a flow rate of 0.75 mL/min. The identity of the products was established by comparison of their retention times and absorption spectra with those of authentic samples.

The concentration of the samples analyzed by HPLC was the same as that used in the Laser Flash Photolysis studies. The solutions contained in pyrex cuvettes were degassed by argon and irradiated with 350-365 nm light (Ray-O-Net reactor) for 8 hours while stirring.

Details of the computational methods are described in Chapter 2.

4.3 Results and Discussion

4.3.1 Isoquinoline N-Oxide

Laser flash photolysis (LFP, 355 nm) of isoquinoline N-oxide (11) produces the previously reported transient spectrum of 311* in benzene, dichloromethane and water at ambient temperature. The transient electronic spectrum exhibits broad absorption from

350 to 420 nm (Figure 4.1), in excellent agreement with the literature.83

0.12 4.2 mM iso-quinoline N-oxide in methanol upon 355 nm LFP, 50 ns after laser pulse 0.10

0.08

0.06

0.04

0.02 Transient absorbance

0.00

300 350 400 450 500 wavelength, nm

Figure 4.1. The transient UV-vis spectrum of triplet 11 upon 355 nm LFP in methanol.

311* has a lifetime of about 4 µs and reacts rapidly with oxygen, as expected.

However, the triplet reacts too slowly to measure with typical electron (KSCN, NaI,

NaN3, 1,3,5-trimethylbenzene, DABCO, indole and silylated guanosine), proton (benzoic acid) and hydrogen-atom donors (tris(trimetylsilyl)silane) to allow measurement of those bimolecular quenching rate constants. The one exception to this pattern is hydroquinone which is reported to undergo a hydrogen atom-transfer reaction (proton coupled-electron transfer).84 A plot of the triplet decay rate constant with the concentration of hydroquinone was linear. The slope (7.0×106 M-1s-1) is the bimolecular rate constant of

311* with hydroquinone.

The reactivity pattern of 311* was analyzed with the aid of DFT calculations. It was found that only a simple hydrogen-atom transfer is a thermodynamically favorable process in both the gas phase and acetonitrile. Simple electron and proton transfer are both endothermic processes (Figure 4.2).

The photochemistry of 11 was also studied by time resolved infrared spectroscopy. Unfortunately, we did not detect IR active transients formed by LFP of 11 in dichloromethane. We suspect that the quantum yield of triplet formation is low due to ring-closure reactions in the singlet excited state, 111*.

Isoquinoline N-oxide:

1) Electron transfer with hydroquinone

OH OH O O N N + +

T1 OH OH

Gas phase: ∆H = 118.5 kcal/mol Acetonitrile: ∆H = 32.4 kcal/mol

2) Hydrogen atom transfer

OH O O OH N N + +

OH T1 OH

Gas phase: ∆H = -1.0 kcal/mol Acetonitrile: ∆H = -1.6 kcal/mol

3) Proton transfer reaction

OH O

O OH N N + +

OH OH T1

Gas phase: ∆H = 78.8 kcal/mol Acetonitrile: ∆H = 16.0 kcal/mol

Figure 4.2. Calculated reaction enthalpies of different reaction pathways of 311*.

4.3.2 Benzo[c]cinnoline N-Oxide

LFP (355 nm) of benzocinnoline-N-oxide (12) in deoxygenated solvent at ambient temperature produces the transient spectrum of Figure 4.3.

0.05

0.00

-0.05 Transient absorbance -0.10

360 400 440 480 520 560 nm

Figure 4.3. Transient spectrum produced by LFP (355 nm) of 0.1 mM 12. The spectrum was recorded 5 ns after laser pulse over a 30 ns time window.

The lifetime of the carrier of the transient spectrum is about 7 µs in deoxygenated solvent at ambient temperature. The transient spectrum is attributed to 312* because the carrier of the spectrum reacts rapidly with oxygen. The assignment of the transient spectrum to 312* is also based on the similarity of its reactivity pattern relative to that of

31*. The triplet state of 12 reacts very sluggishly with electron, proton and hydrogen donors. As with 311*, the triplet state of 12 reacts only with hydroquinone among the quenchers studied. DFT calculations of the reaction enthalpies of 312* with hydroquinone were also performed and predict that hydroquinone can not undergo

54 electron transfer reaction with 312* while hydrogen atom transfer is energetically allowed

(Figure 4.4).

1) Electron transfer with hydroquinone

OH OH NO NO N N

+ +

T1 OH OH

Gas phase: ∆H = 172.8 kcal/mol Acetonitrile: ∆H = 25.3 kcal/mol

2) Hydrogen atom transfer

OH O NO NO N N

+ +

OH OH T1

Gas phase: ∆H = -6.2 kcal/mol Acetonitrile: ∆H = -4.3 kcal/mol

Figure 4.4 Calculated reaction enthalpies of different reaction pathways of 312*.

4.3.3 4-Nitroquinoline N-Oxide

Kasama et al.85 and Seki and co-workers86 have reported that photolysis of 4- nitroquinoline N-oxide 13 leads efficiently to its triplet state (λmax = 410, 590 nm). They report further that 313* readily accepts electrons from donors such as guanosine and tryptophan in buffer solution to form a radical anion (13•¯) which can be protonated to

85, 86 form a neutral hydro radical 3H• (λmax = 460 nm).

O O O N N N hv - H+ e 13H

NO2 NO2 NO2

3 13 13* 13 -

The nitro group has the desirable properties of enhancing the triplet state yield and the ease of its reduction. Both factors make it a promising photochemical source of hydroxyl radical by the Daniels-Gates-Greenburg mechanism.46

Laser Flash Photolysis Studies

LFP (355 nm) of 3 in deoxygenated solvents produces the transient spectra of

Figure 4.5. The transient spectra are in good agreement with previous reports and are attributed to 313*. The transient spectrum of 313* is also in good agreement with Time

Dependent Density Functional Theoretical (TD-DFT) calculations (Figure 4.6).

0.2 1

3 0.1

2 0.0

Transient Absorbance Transient -0.1

-0.2 400 450 500 550 600 nm

Figure 4.5. The transient spectra of 313* (0.35 mM) produced upon LFP (355 nm). The spectrum was recorded immediately after laser pulse over a 45 ns time window. Solvent: 1, CH2Cl2; 2, benzene; 3, water.

triplet 0.10 radical anion

0.08

0.06

0.04

0.02

400 440 480 520 560 600 nm

Figure 4.6. The Calculated UV-Vis spectra of 313* and 13•¯ in the gas phase.

0.6

1 0.4

2 0.2 3

Transient absorbance Transient 0.0 4

400 450 500 550 600 nm

Figure 4.7. The transient spectra produced upon LFP (355 nm) of 0.3 mM of 13 and 8 mM DABCO in acetonitrile. 1: immediately after laser pulse; 2: 500 ns after laser pulse; 3: 3 µs after laser pulse; 4: 10 µs after laser pulse.

The TD-DFT predicted transient spectrum of radical anion 13•¯ is also shown in

Figure 4.6 and its main band is predicted to be near that of 313*.

LFP of 13 at 355 nm in acetonitrile containing 8 mM diazabicyclooctane

(DABCO) produces the transient spectrum of Figure 4.7. The large band at 508 nm is attributed to radical anion 13•¯, which is consistent with the predictions of TD-DFT calculations. The shoulder in the transient spectrum of Figure 4.7 at 480 nm is assigned to the DABCO radical cation which is known to absorb in this region.87

0.8

0.6

1 0.4 2

0.2 Transient Absorbance 0.0

400 450 500 550 600 nm

Figure 4.8. The transient spectra produced upon LFP (355 nm) of 0.3 mM of 13 and 8 mM DABCO. 1: immediately after the laser pulse, in acetonitrile; 2: 5 µs after the laser pulse, in 80% acetonitrile, 20% water (vol:vol).

3H• isomers protonated at O 0.15 11 O14 O13

0.10 Absorbance 0.05

400 450 500 550 600 nm

Figure 4.9. The calculated UV-Vis spectra of different isomers of 13H• in acetonitrile.

The transient spectrum obtained by LFP of 13 in the presence of 8 mM DABCO in acetonitrile with 20% water (V/V) is given in Figure 4.8. The spectrum recorded immediately after the laser pulse is that of the radical ions. The spectrum recorded 5 µs after the laser pulse is that of a different species with a broad absorption around 450 nm.

The carrier of this absorption is attributed to the neutral hydroradical 13H• formed by protonation of radical anion 13•¯ by water, in excellent agreement with the pulse radiolysis studies.85, 86

What is the structure of the hydroradical? A natural population analysis of radical anion 13•¯ gives the charge densities listed below.

O11 N Charge densities: O11= -0.41 O13= -0.33 O = -0.33 N 14 O13 O14

As the oxygen atom of the N-O group bears the most negative charge one might expect that O11 of 13•¯ will be the kinetic site of protonation. We have calculated all the relative energies of the hydroradicals produced by protonation of O11, O13 and O14 in the gas phase and in acetonitrile. In acetonitrile, where most LFP and TRIR experiments were performed, the hydroradical protonated at O14 has the lowest energy by 2.0 kcal/mol. Given the normal error of those calculations, we can not conclusively determine the structure of 13H• solely by comparing the computed relative energies.

However, the predicted UV-Vis spectrum of the hydroradicals (Figure 4.9) in acetonitrile suggests the transient spectrum observed (strong absorption over 400 nm) is the

60 hydroradical protonated at the N-oxide oxygen (O11). This will be later confirmed by

Time-Resolved Infrared Spectroscopy.

O11H O O N N N

N N N OO OO14H HO13 O

Gas Phase: 0.0 kcal/mol - 6.9 kcal/mol - 4.5 kcal/mol Acetonitrile: -0.8 kcal/mol -2.0 kcal/mol 0.0 kcal/mol 13H

Time Resolved Infrared (TRIR) Spectroscopy

Laser flash photolysis (LFP, 355 mm) of 13 in deoxygenated acetonitrile produces the TRIR spectrum of Figure 4.10. Two transient bands are observed at 1220 and 1270 cm-1 with lifetimes of 1.3 µs. Vibrations of the precursor are bleached at 1304 and 1350 cm-1 (also at 1520 and 1566 cm-1, separate spectra) which fully recover with a time constant of 1.5 µs (1304 cm-1) and 1.3 µs (1520 cm-1). This leads us to assign the phototransients absorbing at 1220 and 1270 cm-1 to the triplet state of 13 (313*). These assignments are consistent with DFT calculations (Figure 4.11) which predict that 313* will have prominent vibrations at 1230 and 1280 cm-1 after scaling by a factor of 0.9613.

1 6

3 -

2 x10

2 0

-2

1100 1150 1200 1250 1300 1350 1400 1/cm

Figure 4.10. The transient IR spectra produced upon LFP (355 nm) of 3 mM 13 in -1 -1 acetonitrile (1100-1300 cm ) and CH2Cl2 (1300-1400 cm ). Window: 1, 0-1 µs; 2, 3-4 µs.

40 2

-2 0 1

-4 0

1100 1150 1200 1250 1300 1350 1400 1/cm

Figure 4.11. The calculated IR spectra of 1: 13 and 2: 313* in acetonitrile.

5 1240 4 1 3 1216

2 3 - 2 1148 x10 1

-1 1100 1150 1200 1250 1300 1/cm

Figure 4.12. The transient IR spectra produced upon LFP (355 nm) of 3 mM 13 and 8 mM DABCO in acetonitrile. Window: 1, 0-1 µs; 2, 3-4 µs.

4 1248 3 1 1220

3 - 2 1148 2 x10 1132 1

-1 1100 1150 1200 1250 1300 1/cm

Figure 4.13. The transient IR spectra produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile. Window: 1, 0-2 µs; 2, 6-8 µs.

100

60 1 40 2

20 Transient Aborbance 0 3 -20 1100 1150 1200 1250 1300 1/cm

Figure 4.14. The calculated IR spectra of 13H• (1) and 13•¯(2) and 13 (3) in acetonitrile.

1.6 1.4 1.2 1.0

3 - 0.8

x10 0.6 0.4

0.2 0.0

0 2 4 6 8 µs

Figure 4.15. The decay of the 1240 cm-1 transient produced upon LFP (355 nm) of 3 mM 13 and 8 mM DABCO in acetonitrile. Double exponential decay: τ = 0.63 and 3.5 µs, respectively.

600

-6 400 x10

200

0 10 20 30 40 µs

Figure 4.16. The decay of the 1240 cm-1 transient produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile. Double exponential decay: τ = 0.9 and 14 µs, respectively.

LFP of 13 and 8 mM DABCO in deoxygenated acetonitrile produces the transient spectrum of Figure 4.12. The spectrum exhibits a biexponential decay (τ = 0.63 and 3.5

µs, Figure 4.16) which leads us to assign the transient spectra to a mixture of 313* and its long lived radical anion 13•¯.

LFP of a mixture of 13, 8 mm DABCO and saturated ammonium chloride in deoxygenated acetonitrile produces the transient spectra of Figure 4.13. The transients exhibit biexponential decay (τ = 0.9 and 14 µs) which leads us to assign the spectra to a mixture of radical anion 13•¯ and hydroradical 13H•. DFT calculations predict that 13•¯ and 13H• will have similar vibrational spectra in this region (Figure 4.14). The two species of interest have distinctly different lifetimes, however (Figure 4.15 and 4.16).

DFT calculations were performed to predict the IR spectra of the three possible neutral hydroradicals that can be formed upon protonation of the three different oxygen atoms of the radical anion of 13•¯. The region between 1100 and 1300 cm-1 is not decisive in distinguishing amongst the possible isomers. All the possible hydroradicals

13H• are predicted by theory to have vibrational bands due to the stretching of the ring double bonds in the region between 1400 and 1470 cm-1. However, only the hydroradical

-1 obtained by protonation of O11 has a strong absorption at 1420 cm caused by the N-O-H in-place bending (Figure 4.17), while neither of the other two hydroradical isomers

obtained by protonation of O13 and O14 nor the radical anion has a prominent IR band in this region. As shown in Figure 4.18, LFP of a mixture of 13, 8 mM DABCO and

-1 saturated ammonium chloride in acetonitrile-d3 leads to transient absorption at 1410 cm and 1450 cm-1 with τ = 9 µs (Figure 4.19). This is in good agreement with theory which predicts absorptions at 1420 (O-H bending) and 1455 cm-1 (ring double bond stretching)

(Figure 4.17). The other hydroradicals (protonated at O14 or O13) isomers are predicted to have distinct absorptions between 850 and 950 cm-1 but no absorptions were observed in this region.

20 1 15

10 2 3 5

-5

-10 1390 1400 1410 1420 1430 1440 1450 1460 1470 1/cm

Figure 4.17. The calculated IR spectra of 13H• protonated at 1: O11, 2: O14 and 3: O13 in acetonitrile.

600x10-6 1410

400 1

1450 200 2

0 Transient Absorbance Transient -200

1400 1420 1440 1460 1/cm

Figure 4.18. The transient IR spectra produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile-d3 upon 355 nm LFP. Window: 1, 0-2 µs; 2, 6-8 µs.

120

100

6 - 60 x10 40

-20

0 10 20 30 40 µs

Figure 4.19. The decay of the 1410 cm-1 absorbing transient produced upon LFP (355 nm) of 3 mM 13, 8 mM DABCO and saturated NH4Cl in acetonitrile-d3. Single exponential decay: τ = 9 µs.

Since both TRIR and LFP transient spectroscopy suggest that the protonation proceeds at the O11 position of the N-oxide, it seems likely that this hydroradical will fragment to from hydroxyl radical and 4-nitroquinoline, 14. An HPLC analysis indicates the major product formed on photolysis of 4-nitroquinoline N-oxide in acetonitrile is 4- nitroquinoline. The mechanism by which this product is formed is not known. DFT calculations predict that direct fragmentation of triplet 4-nitroquinoline N-oxide to 4- nitroquinoline and oxene is endothermic by 29.5 kcal/mol in acetonitrile. However, the presence of DABCO and ammonium chloride quenches the formation of 4- nitroquinoline, instead of enhancing the yield of 14 through the dehydroxylation mechanism. Thus, electron transfer conditions do not lead to a hydroradical which fragments to form hydroxyl radical and 4-nitroquinoline.

DFT calculations explain why the fragmentation of the hydroxyl radical formed by the protonation of O11 is not favorable in acetonitrile (∆Hr from 13H• to 14 is computed to be 11.3 kcal/mol). This also explains the relatively long lifetime of this hydroradical.

OH N N + OH

NO2 NO2 13H 14

DFT calculations indicates that protonation of O14 will also form a radical, 13H’, with a lifetime sufficiently long to allow its detection (∆Hr from 13H’• to 15 is predicted to be 34.3 kcal/mol). However, we did not observe the radical (13H’) in LFP and TRIR experiments.

O O N N reduction N OH +

N N NHOH OOH O

13H' 15 16

We conclude that the fragmentation proposed by Seki et al86 is not thermodynamically reasonable in solution. In support of this view we note that 4- hydroxyaminoquinoline N-oxide, 16, the product of reduction of 15 is not observed by

69 the HPLC analysis of the mixture of photolysis products of 4-nitroquinoline N-oxide in the presence of DABCO and ammonium chloride.

4.4 Conclusions

The triplet states of isoquinoline N-oxide and benzocinnoline N-oxide react sluggishly with electron (KSCN, NaI, NaN3, 1,3,5-trimethylbenzene, DABCO, indole and silylated guanosine), proton (benzoic acid) and hydrogen-atom donors

(tris(trimethylsilyl)silane). These triplets will react with hydroquinone by hydrogen-atom transfer (proton coupled electron transfer). Triplet 4-nitroquinoline N-oxide reacts readily with electron donors to from the radical anions as previously reported. The radical anion is protonated on the oxygen atom of the N-oxide group to from a neutral radical. The triplet state, radical anion and neutral hydroradical derived from 4- nitroquinoline N-oxide were detected by TRIR spectroscopy and the spectra were interpreted with Density Functional Theory calculations. The three N-oxides of this study are not expected to serve as photochemical sources of hydroxyl radical.

CHAPTER 5

TIRAPAZAMINE AND DESOXYTIRAPAZMINE

Reproduced in part with permission from J. Phys. Chem. A 2005, 109, 1491

5.1 Introduction

We have long been interested in initiating reactions of tirapazamine and its analogues photochemically to develop a source of hydroxyl radical using visible light.

Access to the Ohio Laboratory of Kinetic Spectrometry in Bowling Green State

University and the establishment of Center of Chemical and Biophysical Dynamics

(CCBD) in The Ohio State University allowed us to use ultrafast spectroscopy to examine the photophysics and photochemistry of these aromatic N-oxides. Here we report our studies of the excited singlet states of tirapazamine (TPZ), 4- desoxytirapazamine (dTPZ) and its isomer 1-desoxytirapazamne (dTPZ’), along with quinoxaline di-N-oxide (QXNO) and phenazine di-N-oxides (PNNO).

5.2 Experimental

Picosecond time resolved measurements were performed at the Ohio Libratory for

Kinetic Spectrometry at Bowling Green State University and the Center for Chemical and

Biophysical Dynamics of The Ohio State University. The experimental details are described in Chapter 2.

Nanosecond time resolved measurements were performed at The Ohio State

University using the instrument described in Chapter 2.

Fluorescence quantum yield (ΦF) and fluorescence quenching experiments were carried out using a Spex Fluorolog 1680 double spectrometer (JY Horiba Inc., Edison NJ,

USA). Measurements of the fluorescence quantum yield were performed according to

88 the protocol in JobinYvon Ltd. Website. Acridine orange (ΦF= 0.46, λex=475 nm, integrated from 500 to 750 nm, in ethanol) was used as the standards for TPZ and PNNO and harmane ((ΦF= 0.81, λex=390 nm, integrated from 405 to 650 nm, in 0.1 M H2SO4) were used as the standard for dTPZ, dTPZ’ and QXNO. A series of fluorescence intensities were plotted versus the corresponding absorptions (controlled under 0.1) at the excitation wavelengths to obtain the slopes (S) of the samples and standards. The

2 quantum yields of samples were calculated by: ΦX=ΦST (SX/SST)(ηX/ηST) , where η is the refractive index of the solution . The lifetime of dTPZ was measured at Bowling Green

52 State University as described in Chapter 2.

TPZ and dTPZ were prepared according to the procedure of Fuchs et al.89

3-Amino-1,2,4-benzotriazine 1-Oxide (dTPZ). 2-Nitroaniline (2.48 g, 18 mmol) and cyanamide (1.51 g, 36 mmol) were heated with stirring at 100 °C until a deep red melt formed. The reaction was cooled to room temperature, and cold concentrated

72 hydrochloric acid (6.5 mL, 78 mmol) was added dropwise. The reaction was then warmed to 100 °C and stirred for 20 min, then cooled again to room temperature. To the cooled reaction mixture NaOH was added 6.5 mL of a 16 M aqueous solution in 0.5 mL portions over approximately 15 min. The resulting cloudy yellow-orange mixture was warmed to 100 °C and stirred for 2 h. Addition of water (~25 mL) produced a yellow precipitate. The suspension was allowed to cool for 10 min, collected by vacuum filtration, and then washed with water followed by ethyl acetate to remove unreacted 2- nitroaniline (until the wash changes from orange to light yellow). The bright yellow solid collected from the filtration (0.9 g, 35% yield) was recrystallized from 2-propanol to

1 yield dTPZ. H NMR (δ, ppm, CDCl3): 8.31 (dd, J = 8.7, 0.9 Hz, 1H), 7.75 (ddd, J = 7.7,

6.9, 1.4 Hz, 1H), 7.61 (dd, J = 7.6, 1.0 Hz, 1H), 7.37 (ddd, J = 8.5, 7.0, 1.3 Hz, 1H), 5.13

(bs, 2H).

3-Amino-1,2,4-benzotriazine 1,4-Dioxide (TPZ). TPZ (500 mg, 3.0 mmol) was suspended in glacial acetic acid (20 mL) and heated to 60 °C. To this suspension was added 50% hydrogen peroxide (10 mL), and the resulting mixture was stirred in the dark for 15 h. Between two and four additional aliquots of acetic acid (2 mL) and 50% hydrogen peroxide (2 mL) were added over the next 24 h until starting material was largely consumed (as judged by HPLC). The reaction was lyophilized to dryness (behind a blast shield), and the resulting burnt orange solid purified by flash-column chromatography on silica gel (eluted with 75:25 ethyl acetate:hexane, followed by 100% ethyl acetate, followed by 25:75 2-propanol:ethyl acetate) to yield TPZ (40-60%). The resulting red solid can be recrystallized from 2-propanol or ethanol to yield red plates. 1H

NMR (δ, ppm, CDCl3): 8.45 (d, J = 9.6 Hz, 1H), 8.10 (t, J = 7.9 Hz, 2H), 7.82 (d, J = 8.4

Hz, 1H), 7.68 (t, J = 7.8 Hz, 1H), 6.84 (bs, 2H).

3-Amino-1,2,4-benzotriazine 4-N-oxide (dTPZ’). 30 mg of TPZ was dissolved in dioxane in the presence of 150 µL of azo-t-butyl and refluxed overnight. The reaction was monitored by HPLC until all the staring material was consumed. The reaction solution was then evaporated under reduced pressure and eluted with ethyl acetate/hexane through a silica column. dTPZ’ was the third main fraction (the first two fractions were dTPZ and 3-amino-1,2,4-benzotriazine, respectively) to yield 15 mg final product (55%

1 yield). H NMR (δ, ppm, CDCl3) 8.38 (dd, J = 5.9, 1.2 Hz, 1H), 8.35 (dd, J = 8.3, 1.2 Hz,

1H), 7.88 (ddd, J = 7.9, 7.8, 1.2 Hz, 1H), 7.65 (ddd, J = 7.8, 7.7, 1.2 Hz, 1H), 6.5 (bs, H2;

UV-vis max 208 ( = 9200), 256 ( = 27000), 326 ( = 3400), and 462 ( = 3600) nm.

Quinoxaline di-N-oxide and phenazine di-N-oxide were prepared by the oxidation of parent quinoxaline and phenazine by acetic acid and hydrogen peroxide according to the literature. 90

Quinoxialine di-N-oxide (QXNO). 1.54 g quinoxaline was dissolved in 40 mL acetic acid. 10 mL hydrogen peroxide was added and the mixture was stirred and heated in oil bath at 55 °C for 24 hours. Another portion of 5 mL H2O2 was added and the mixture was stirred for another 24 hours. 33 g NaOH was dissolved in water and cooled down to room temperature. The alkaline solution by portions was added to the reaction mixture to neutralize acetic acid in ice bath. 300 mL chloroform in four portions was used to extract the reaction mixture to give orange organic solution. This solution was then dried under reduced pressure to give 1.6 g (90%) yellow-orange solid. The product

1 was recrystallized in EtOH. HNMR (δ, ppm, CDCl3): 8.60 (m, 2H), 8.21 (2, H), 7.88 (m,

2H)

Phenazine di-N-oxide (PNNO). 1.3 g quinoxaline was dissolved in 40 mL acetic acid. 10 mL hydrogen peroxide was added and the mixture was stirred and heated in oil bath at 55 °C for 26 hours. 33 g NaOH was dissolved in water and cooled down to room temperature. The alkaline solution by portions was added to the reaction mixture to neutralize acetic acid in ice bath. 300 mL chloroform in four portions was used to extract the reaction mixture to give wine-red organic solution. This solution was then dried under reduced pressure to give 0.85 g (60%) red solid. The product was recrystallized in

1 EtOH to obtain cotton-like red crystals. HNMR (δ, ppm, CDCl3): 8.75 (m, 4H), 7.84 (m,

4H).

DFT and TD-DFT calculation details were performed as described in Chapter 2.

5.3 Results

Tirapazamine

The absorption spectra, and emission spectra and fluorescence lifetime of TPZ

81 have been studied as a function of solvent. The interaction of ground and excited (S1) state TPZ with amino acids and nucleosides have also been reported.81

Laser flash photolysis (LFP, 400 nm) of TPZ in acetonitrile produces the transient absorption spectrum of Figure 5.1. Bleaching of TPZ is evident below 500 nm. The negative peak observed between 600 and 700 nm is attributed to TPZ fluorescence which has been previously reported in the 530-650 nm region.81 The quantum yield of TPZ fluorescence was measured to be 0.002 in acetonitrile. The positive peak at 544 nm

75 decays with the same time constant (130 ps) as the fluorescence decay and the recovery of the ground state absorption below 500 nm. (Figure 5.2) Thus, the peak at 544 nm is attributed to the S1 state of tirapazamine. This assignment is consistent with our previous report that the fluorescence lifetime of TPZ in acetonitrile is 110 ps.81

The addition of the neutral electron donor DABCO increases the pseudo first order decay of the transient absorption of TPZ and its fluorescence, as expected.

However, the transient spectrum of the TPZ radical anion is not observed. Presumably, reverse electron transfer between the TPZ radical anion and DABCO radical cation is extremely rapid leading to undetectably short lifetimes and low concentrations of these species.

-3 10x10

544nm 1 5 2

0 3 Transient Absorbance Transient

-5 625 nm

500 550 600 650 700 750 800 Wavelength, nm

Figure 5.1. Transient absorption spectrum produced upon excitation (400 nm) of TPZ in acetonitrile under an argon atmosphere, 1: 8.29 ps; 2: 53.23 ps; 3: 218.23 ps after laser pulse.

-3 Kinetic traces of tirapazamine in acetonitrile upon 400 nm LFP 6x10 544 nm 625 nm 492 nm

4 6.9x109

Transient Absorbance 7.0x10 -2

-4

-9 0.0 0.2 0.4 0.6 0.8 1.0 1.2x10 Time, s

Figure 5.2 The kinetic traces of the transient absorptions of 1TPZ*..

-3 All columns: 3 80x10 1

60 540 40 3 2 20

Transient Absorbance -20

-40

350 400 450 500 550 600 Wavelength, nm

Figure 5.3. Transient spectrum produced upon excitation (355 nm) of 0.9 mM TPZ in water containing 4 M KSCN, 1: 10 ns after laser pulse; 2: 5 µs after laser pulse; 3: calculated UV of TPZ radical anion (B3LYP/6-31G*, gas phase).

80 5 -1 545 nm, k=1.3x10 s 5 -1 60 385 nm, k=1.7x10 s

-3 40 x10 20

0 10 20 30 40 50 µs

Figure 5.4. The kinetics monitored at 545 nm and 385 nm of the transient species whose spectra are shown in Figure 5.3.

1.4 mM TPZ + 2 M NaI + NH4Cl in w ater upon 355 nm LFP 0.25 right after laser 385 nm 3 us after 0.20 10 us after calculated UV absorption of TPZ radical anion calculated UV absorpiton of TPZ hydroradical 0.15 540 nm

0.10

0.05 Transient Absorbance

0.00

400 450 500 550 600 Wavelength, nm T

Figure 5.5. Transient spectrum produced upon excitation (355 nm) of 0.9 mM TPZ in water containing 4 M KSCN and ~ 3 M NH4Cl.

0.15 5 -1 540 nm decay, k=8.8x10 s 5 -1 385 nm decay, k=6.8x10 s 0.10

0.05

0.00

0 10 20 30 40 50 µs

Figure 5.6. The kinetics monitored at 545 nm and 385 nm of the transient species whose spectra are shown in Figure 5.5.

O O N N N 400/355 nm N acetonitrile/water N NH2 N NH2 O O TPZ 1TPZ*

O O N I N I N + N + SCN N NH2 SCN N NH2 O O 1TPZ* TPZ

O O1 N N N N +NH4 + NH3 N NH 2 N NH2 O O4H TPZ TPZ-H

O O1H N N N N

N NH2 N NH2 O4H O TPZ-H TPZ-H'

Scheme 5.1 Photoinduced reduction of TPZ.

Nanosecond time resolved LFP (355 nm) of 1.9 mM TPZ in water containing either 4-5

M KSCN or 3-5 M KI produces the transient spectrum shown in Figure 5.3. The bleaching observed at 430 and 500 nm is due to the ground state absorption of TPZ in this region of the spectrum. The transient absorptions at 385 and 540 nm decay with the same time constant (7 µs) arguing that a common intermediate is responsible for both spectral features. The carrier of the transient absorption is assigned to the TPZ radical anion and the spectrum itself is consistent with the spectrum of this species obtained by

74 pulse radiolysis. The pulse radiolysis studies demonstrated that the pKa of TPZ-H, the conjugate acid of the radical anion is 5.6 ± 0.2.74 The transient spectrum of TPZ radical anion is also consistent with TD-DFT calculations, which predict vertical absorptions of the following wavelengths: 592 nm (f=0.0376), 493 nm (f=0.0276), 411 nm (f=0.0329).

The lifetime of the radical anion is greatly shortened in the presence of oxygen. (Figure

5.4) This result is consistent with the mechanism of action of TPZ and in particular its selectivity towards hypoxic cells.

We continued to add NH4Cl to the samples to qualitatively observe how the increased acidity will influence the fate of the radical anion of TPZ. As in Figure 5.5 and

Figure 5.6, the decay of 540 and 385 nm accelerate and lifetime of TPZ radial anion is reduced to about 1.2 µs. The 380 nm signal does not decay to the baseline. The spectra in Figure 5.5 shows there is a persistent species with a flat absorption below 400 nm.

Denny et al48 observed a similar absorption of TPZ radical anion reduced by pulse radiolysis under low pH conditions, and assigned the carrier of this absorption to the protonated hydroradical at the O4 position. Our TD-DFT solution also indicates that the

O4-protonated radical has absorption at this region (Figure 5.7).

-3 100x10 Calculated UV-Vis absorptions of 4-hydroradical 1-hydroradical 80 tria zinyl radica l

observed species 60

40 Absorbance

350 400 450 500 550 600 650 Wave length, n m

Figure 5.7 Comparison of the calculated UV-vis absorptions for the possible transients whose spectra are given in Figure 5.6

A systematic study of the dependence of the lifetime of the radical anion on pH could not be established, because the extremely high concentration of electrolytes (3-5 molars of KSCN and KI) does not allow us to accurately control a normal pH value of buffer solution.

Desoxytirapazamine

The visible absorption band of dTPZ has a maximum at 413 nm in water (log10ε =

3.68), 404 nm in acetonitrile (log10ε = 3.63), and 402 nm in dichloromethane (log10ε =

3.70). The solvatochromic behavior of this band is much less pronounced than that of the visible band of tirapazamine, and follows the opposite trend: the absorbance maximum is lower in energy in polar, hydrogen-bonding solvents.

Desoxytirapazamine fluoresces more intensely than does TPZ under comparable experimental conditions (Figure 5.8). We measured the fluorescence quantum yield of dTPZ as 0.12 in acetonitrile.

1.4E+08 dTPZ 1.2E+08 TPZ*5 dTPZ'*5 1.0E+08 QXNO*10 PNNO 8.0E+07

6.0E+07

4.0E+07 Fluorescence Counts 2.0E+07

0.0E+00 450 500 550 600 650 700 Wavelength (nm)

Figure 5.8. The fluorescence spectra of TPZ (λex= 432 nm, A=0.302) and dTPZ (λex= 422 nm, A=0.301), dTPZ’(λex= 420 nm, A=0.310), QXNO (λex= 400 nm, A=0.305) and PNNO (λex= 422 nm, A=0.308) in acetonitrile at ambient temperature.

The rates of electron transfer from various donors to the singlet excited state of dTPZ were determined by fluorescence quenching experiments, and the data obtained are summarized in Table 5.1. The measured quenching rate coefficients have magnitudes approaching the diffusion limit but are in general slightly smaller than the corresponding values obtained with TPZ.81 There is only marginal evidence to suggest that there is any significant curvature in the Stern-Volmer behavior, and so it cannot be determined whether or not dTPZ forms hydrogen bonded complexes with some quenchers in the 83 same manner as TPZ.81 The larger rate constants obtained with TPZ may reflect the greater tendency of this drug to from complexes with quenchers.

Adenosine monophosphate is an exception to the rule. It can be seen that unlike

TPZ, dTPZ will undergo photochemical electron transfer with adenosine monophosphate.

This result may be due to the fact that the photochemical energy available from excitation of dTPZ (assuming that the S0,0 → S1,0 electronic transition for dTPZ in water is approximately 450 nm or 2.8 eV), is greater than that available from excitation of TPZ

(S0,0 → S1,0 approximately 510-520 nm or 2.4 eV).

Picosecond time resolved LFP (400 nm) of dTPZ in acetonitrile produces the transient spectrum of Figure 5.9. The carrier of the transient absorption at 640 nm is assigned to the S1 state of dTPZ. The lifetime of the transient absorption is much greater than 200 ps, the limit of the spectrometer. This is consistent with the fluorescence lifetime measurements. The fluorescence lifetime of dTPZ, determined by the TCSPC technique, was determined to be 5.4 ns, which is about 50 fold longer than that of TPZ.81

Nanosecond time resolved LFP (355 nm) of dTPZ in acetonitrile containing either 0.15

M KSCN or 0.08 M NaN3 produces new transient absorptions with a lifetime of about 3 microsecond lifetimes monitored at 440 nm. (Figure 5.10 and 5.11). The 475 nm absorption in Figure 5.11, which has a longer lifetime, is attributed to the konwn

⋅ (SCN)2 ‒ radical. The carrier of the transient spectrum is attributed to the radical anion of dTPZ based on TD-DFT calculations and by analogy to the TPZ studies and the oxygen dependence of the transient species. The transient absorption is rapidly quenched by oxygen.

O O N N N 400/355 nm N acetonitrile/water N NH2 N NH2 O O dTPZ 1dTPZ*

O O N N I N I + N + N NH2 SCN SCN N NH2

1dTPZ* dTPZ

O OH N N N N +NH4 + NH3

N NH2 N NH2 dTPZ dTPZ-H

Scheme 5.2 Photoinduced reduction of dTPZ.

The introduction of acid, NH4Cl, generated transient spectra in Figure 5.12, as well as kinetic traces in Figure 5.13. (Small amount of water is added to help dissolve the salt.) Similar to TPZ, the radical anion of dTPZ is also able to be protonated, as supported by calculations from Liu’s thermodynamics calculation.91 TD-DFT calculations indicate that the hydroradical of dTPZ, dTPZ-H, absorbs around 340 nm, and less strongly at 470 nm, which is not very different from the radical anion of dTPZ.

Therefore, the spectra after a few microseconds in Figure 5.12 could be that of the dTPZ-

H. Kinetic traces in Figure 5.13 show lengthened lifetime compared with in the absence of acid. And decay at 340 nm does not decay to the base line in tens of microseconds.

Another possibility of the longer-lived species followed by the hydroradical is the extremely fast formation of the persistent 3-amino-1,2,4-benzotriazine, the parent heteroaromatic for TPZ and dTPZ. Due to the instrumental limitation, we were unable to measure species with lifetimes over 50 µs.

a Quencher, Q Solvent [Q] Linear fit kq b 2c kQτf R (M) (M-1) (M-1s-1) AMPd W 0.0-0.10 22.8±0.4 0.995 (4.22±0.07)x109 9 NaN3 W 0.0-0.3 16.4±0.3 0.994 (3.03±0.05)x10 KSCN W 0.0-0.17 20.5±0.6 0.990 (3.80±0.10)x109 Tyrosinee W 0.0-0.15 21.6±0.5 0.993 (4.00±0.09)x109 GMPf W 0.0-0.09 21.8±0.5 0.992 (4.04±0.09)x109 Tryptophane W 0.0-0.06 33.4±1.0 0.987 (6.2±0.2)x109 aW = aqueous phosphate buffer, 0.025M, pH 6.9. bErrors quoted are standard deviations on the line of best fit. cGoodness of fit parameter. dAdenosine monophosphate disodium salt. eAmino acids used as methyl ester hydrochlorides. fGuanosine monophosphate, mono sodium salt.

Table 5.1: Rates of reaction for electron transfer between desoxytirapazamine and various substrates determined from Stern-Volmer analysis of fluorescence quenching.

640 nm -3 1 20x10

10 2

0 Transient Absorbance -10 510 nm

500 550 600 650 700 750 Wavelength, nm

Figure 5.9. Transient absorption spectrum produced upon LFP (400 nm) of dTPZ in acetonitrile, 1: 1.29 ps; 2: 1033 ps after laser pulse

445 nm 0.10 3 475 nm 1 0.05

0.00 2 -0.05 Transient Absorbance all columns: 3 -0.10

350 400 450 500 550 Wavelength, nm

Figure 5.10. Transient absorption spectrum produced upon LFP (355 nm) of 0.9 mM dTPZ and saturated KSCN in actonitrile, 1: right after laser pulse; 2: 1 µs after laser pulse; 3: calculated radical anion of dTPZ (B3LYP/6-31G*, gas phase).

0.20 5 -1 440 nm, k=3.2x10 s 5 -1 475 nm, k=1.3x10 s 0.15

0.10

0.05 Transient Absorbance Transient

0.00

0 10 20 30 40 50 µs

Figure 5.11 Kinetic traces recorded at 440 nm and 470 nm of the species whose spectra are given in Figure 5.10.

0.9 mM desTPZ + KSCN + NH4Cl + w ater in acetontrile upon 355 nm LFP right after laser 0.15 3 us after Calculated UV absorption of desTPZ radical anion Calculated UV absorption of desTPZ hydroradical 0.10

0.05

0.00 Transient Absorbance

-0.05

350 400 450 500 550 Wavelength, nm

Figure 5.12. Transient absorption spectrum produced upon LFP (355 nm) of 0.9 mM dTPZ and saturated KSCN, in the presence water and NH4Cl in actonitrile.

0.12 -1 330 nm, with NH4Cl and trace water, k=0.9X105 s -1 0.10 440 nm, with NH4Cl and trace water, k=1.0X105 s

0.08

0.06

0.04

Transient Absorbance Transient 0.02

0.00

0 10 20 30 40 50 µs

Figure 5.13. Kinetic traces recorded at 440 nm and 330 nm of the transient species whose spectra are given in Figure 5.12.

89 dTPZ’

A second, minor metabolite of tirapazamine is dTPZ’, an isomer of dTPZ.89

O N N N N

N NH2 N NH2 O dTPZ dTPZ'

Laser Flash Photolysis of dTPZ’ produces fluorescence (Figure 5.5 and Figure 5.14a) at

450 nm with a lifetime of 1.6±0.3 ps. The fluorescence quantum yield of dTPZ’ was measured to be 0.0024 in acetonitrile, which is similar to that of TPZ and is much smaller than that of its isomer dTPZ.

Related experiments were performed with quinoxaline di-N-oxide (QXNO) and phenazine-di-N-oxide (PNNO), Figures 5.14(b) and 5.14(c), respectively.

O O N N

N N O O

QXNO PNNO

The excited S1 States of QXNO and PNNO have fluorescence quantum yields of less than 0.001 and 0.011 in acetonitrile, respectively and lifetimes of 3.1 ps and 400 ps respectively.

(a) 0.0

-0.2

-0.4

-0.6

O.D. -0.8 ∆ -1.0 probe = 450 nm -1.2

-3 -1.4x10

-2000 0 2000 4000 6000 Time delay /ps

(b)

-3 4x10 probe at 610 nm

2 O.D.

∆ 1

-1

0 5 10 15 20 Time delay /ps

(c)

1.0 Probe at 550 nm

0.5

0.0 O.D.

∆ -0.5

-1.0

-3 -1.5x10

0 200 400 600 800 1000 Time delay /ps

Figure 5.14 Transient absorption kinetic traces after photoexcitation of dTPZ` (a), QXNO (b) and PNNO (c) in acetonitrile at 400 nm.

5.4 Discussion

Photolysis of TPZ and dTPZ produces the analogous excited singlet states which can be detected by picosecond transient absorption spectroscopy. The excited singlet states can be reduced by anionic electron donors to form observable radical anions, crucial species in the cascade of reactions involved in the biological activity of TPZ.

A key finding is that the excited singlet state of dTPZ is about 50 times longer lived than that of TPZ. Previous workers have often noted interesting differences between the photochemistry of a C=N-O unit in aromatic N-oxides and the analogous

N=N-O units in comparable aromatic system.50, 92, 93

93 a Calculated ∆H of τS1 E(S1) Fluorescence cyclication (ps) kcal/mol quantum Compound Ground State Cyclization Product b (kcal/mol) yield O O O Tirapazamine N N N N (TPZ) N N 31.7 (68.0) 110-130 54 0.0020 N NH N NH N NH 2 2 O 2 O O 4-Desoxy- O O tirapazamine N N N O 62.9 5400 65 0.12 N N (dTPZ) N N NH2 N NH N 2 NH 2 N N 1-Desoxy- N N tirapazamine 34.2 1.6 64 0.0024 N NH N NH (dTPZ’) 2 O 2 O O O Quinoxaline N di-N-oxide N 30.1 3.1 69 <0.001 (QXNO) N (0.00088) N O O O O Phenazine N di-N-oxide N 42.6 400 57 0.011 (PNNO) N N O O a Estimated by the UV-Vis and fluorescence spectra. b Measured in acetonitrile

Table 5.2 Summary of computational and photophysical data.

The photochemistry of aromatic N-oxides is complex and it is unlikely that a single mechanism can explain all known observations.50, 92, 93 Photolysis of nitrones produces oxaziridines which can be isolated as stable compounds.93

O hv O NCH2 NCH2 R R

Photolysis of aromatic N-oxides does not produce isolable bicyclic heterocyclic species.

The evidence suggests that oxaziridines are formed as reactive intermediates by cyclization of the excited singlet state. 93

O O O O N R hv N N N R R R

A complex mixture of stable products are then formed depending on the specific reaction conditions and aromatic N-oxides employed.

To better understand our data, DFT calculations were performed on the N-oxides and di-N-oxides of this study (Table 5.2). The cyclization of TPZ to form an oxaziridine is predicted by DFT to be endothermic by 31.7 kcal/mol. The singlet (S1) energy of TPZ is 54-55 kcal/mol.

O O N N N N ∆ H = 31.7 kcal/mol N NH 2 N NH2 O O TPZ

The cyclization of TPZ to form an oxadiaziridine is a much more endothermic process, 68.0 kcal/mol.

O O N N N N ∆ H = 68.0 kcal/mol

N NH2 N NH2 O O TPZ

Thus, the S1 state of TPZ should readily cyclize to from an oxaziridine and this will result in a low fluorescence quantum yield and a short fluorescence lifetime.

We attempted the analogous calculation with dTPZ. However, the oxadiaziridine is no longer predicted to be a stationary point. Attempts to find a minimum for the oxadiaziridine always led to ring opening.

O O N N N O N N N

N NH2 N NH2 N NH2

dTPZ ∆ H = 62.9 kcal/mol

Cyclization of dTPZ’ produces an oxaziridine which exists in a well-defined minimum.

N N N N ∆ H = 34.2 kcal/mol N NH 2 N NH2 O O dTPZ'

The energetic cost of cyclization of dTPZ’ (34.2 kcal/mol) is well below the singlet energy (64 kcal/mol) of the N-oxide. Thus the fluorescence lifetime and quantum yield are lower than its isomer dTPZ.

The endothermicity of cyclization of quinoxaline-di-N-oxide and phenazine-di-N- oxide are 30.1 and 42.3 kcal/mol, well below their singlet energies of 69 and 57 kcal/mol respectively. The short singlet lifetimes are consistent with these results.

Oxadiaziridine itself is not a known compound but has been studied previously by computational methods.94 We have used DFT calculations to study the following isodesmic reaction in which both the oxaziridine and oxadiaziridine isomers are stable minima:

O O HC NH NNH ∆ H = 40.0 kcal/mol

NH2 CH3

DFT calculations predict that oxadiaziridines are 40 kcal/mol less stable than isomeric oxaziridines. Thus, we conclude that the cyclization of the excited state of

96 dTPZ (S1 = 63-64 kcal/mol) to an oxadiaziridine is an endothermic by at least 7 kcal/mol and as a result is a slow and unimportant process. The excited state of dTPZ does not enjoy a rapid deactivation process and thus has a greater fluorescence quantum yield and longer fluorescence lifetime than TPZ.

5.5 Conclusions

Laser flash photolysis of tirapazamine (TPZ) and desoxytirapazamine (dTPZ) produces their excited singlet states. The S1 states have been observed by picosecond time-resolved absorption spectroscopy. The lifetime of TPZ is much shorter than that of dTPZ and the quantum yield of TPZ fluorescence is much smaller than that of dTPZ.

DFT calculations indicate that the S1 lifetime of TPZ is controlled by reversible cyclization to an oxaziridine. The corresponding process in dTPZ forms a higher energy oxadiaziridine. This latter process is endothermic and does not influence the photophysics of the singlet state of dTPZ. The S1 states of TPZ and dTPZ are reduced to radical anions by KSCN, KI and NaN3. Studies of the S1 states of 4-desoxytirapazamine

(dTPZ’), which can cyclize to form an oxaziridine are consistent with this rule.

CHAPETER SIX

REACTIONS OF HYDROXYL RADICAL WITH AROMATIC HYDROCARBONS IN

NON-AQUEOUS SOLUTIONS

Reproduced in part with permission from

J. Phys. Chem. A 2005, 109, 2547 and J. Am. Chem. Soc. 2005, 127, 7094.

6.1 Introduction

As reviewed in Chapter 1, reactive oxygen species (ROS) interest both biological and environmental chemists. Among them, hydroxyl radical (HO•) is critical over a range of important solution-phase reactions from pollutant photochemistry in natural waters95 to the chemistry of aging and carcinogenesis.26

An important group of environmental pollutants are polycyclic aromatic hydrocarbons (PAHs).96-98 These compounds are common components of crude oils and are also highly publicized products of fossil fuel combustion. Hydroxyl radical is believed to assist in the decomposition of these persistent environmental pollutants. One of the characteristics of these compounds is their low solubility in water, making them difficult to study in aqueous solutions on the laboratory scale. However, the high

98 reactivity of hydroxyl radical with most polar organic compounds (including solvents) makes it difficult to perform nonaqueous experiments.

One of the most important components in the study of these complex reactions is one’s ability to measure the rates of fundamental reactions involving these species. Many groups have studied the kinetics of reaction of hydroxyl radical with organic substrates in the aqueous phase, largely by a combination of laser flash photolysis (LFP)99 and pulse radiolysis (PR)100 techniques. Much of the data has been compiled and is readily available.101

The relatively low reactivity of hydroxyl radical with acetonitrile101 allows us to select this solvent as a nonaqueous media, which will not compete strongly with the PAH substrates we are interested in. Acetonitrle also has the advantage of dissolving readily in both the polar hydroxyl precursor and the poly aromatic hydrocarbons. We utilize the

355 nm YAG laser in our laboratory to photolyze the precursor and thereby generate the hydroxyl radical, and a transient UV-Vis detection system to observe the hydroxyl-probe adduct. The formation rate constant and the yield of the adduct is a function of the presence of other hydrocarbon substrates, from which we are able to deduct the absolute rate constants of hydroxyl radical towards a variety of aromatic hydrocarbons. Structure- reactivity relationships and some interesting observation will be discussed based on these kinetic data.

6.2 Experimental

HPLC grade acetonitrile N-hydroxypyridin-2-thione (PSH, 21), and trans-stilbene

(25) were obtained from Aldrich and used as received. All other materials (95 - 99%+ purity) were obtained from Aldrich. Crystalline solids were used as received, and liquids were passed through a short column of basic alumina prior to use. Solutions (2 mL) consisting of PSH (5×10-4 M), trans-stilbene (held constant between 0.010-0.015 M) and the aromatic hydrocarbons (serial concentrations between 0.00 and 1.10 M, depending on the reactivity of the competitor) were deoxygenated using a stream of acetonitrile- saturated argon. The concentration of PSH, with an overall optical density for PSH of about 1.8 - 2.0 at 355 nm, was chosen to provide optimal signal intensity and appropriate building rate constant). The instrumental details for the transient UV-Vis spectroscopic and nanosecond laser flash photolysis kinetic measurements are described in Chapter 2.

Quantitative gas chromatographic (GC) studies were performed on a Hewlett-

Packard 6890 system fitted with an automatic injector and a flame ionization detector.

The column used for the GC-FID was an Rtx-5 column with 5% diphenyl/95% dimethyl polysiloxane stationary phase (length: 30 M, diameter: 0.25 mm, film thickness: 0.25

µM) manufactured by Restek. The GC conditions were the following: initial temperature of 60 °C, hold for 2 min; increase of 25 °C/min to 100 °C, hold for 8 min; increase of 25

C/min to 250 °C, hold for 10 min; along with an inlet temperature of 280 °C and a detector temperature of 300 °C. GC-MS measurements were performed with a Hewlett-

Packard 6890 GC system with a 5973 mass selective detector. The column used for GC-

MS was an HP-5MS column with parameters identical to those of the Rtx-5. Typically, six solutions with three different ratios of benzene to naphthalene (4.7:1, 14.1:1, and

100

42.3:1) were photolyzed in acetonitrile in the presence of 10 mM PSH at 365 nm for 12 to 15 hours in a Ray-O-Net reactor. The solutions were degassed with argon before photolysis. The resulting solution was derivatized with an equivalent volume of BSTFA

(bis(trimethylsilyl)trifluoroacetamide) from Supelco. The solutions were then analyzed by GC-FID, and the chromatographs were integrated using the area normalization technique to determine the relative ratios of phenol and naphthol derivatives. A series of phenol/1-naphthol solutions with known concentrations were derivatized and analyzed by

GC-FID to obtain the response of the ratio of signal intensity vs concentration of authentic phenol and naphthol. The retention times of phenol (6.66 min) and 1-naphthol

(16.13 min) derivatives were obtained directly by standards on GC-FID, while that of 2- naphthol (16.27 min) was determined based on the GC-MS chromatogram.

6.3 Results and Discussion

6.3.1 Laser Flash Photolysis

Generation of hydroxyl radical

Hydroxyl radical was generated by photolysis of N-hydroxypyridine-2-thione (21) at 355 nm as shown in Scheme 6.1. The photolysis of this compound has been investigated previously by Aveline et al,102, 103, who determined that in non-aqueous solution, PSH is photolyzed relatively cleanly to yield hydroxyl radical and the pyrithiyl radical (22) (Scheme 6.1). Unfortunately, hydroxyl radical does not have a diagnostic

UV-vis absorption at convenient wavelengths (300 – 700 nm) in solution-phase laser flash photolysis (LFP) experiments. Thus, we have utilized trans-stilbene as a probe molecule which generates a useful and convenient UV-vis signature of the adduct of

101 hydroxyl radical with trans-stilbene. The kinetics of formation of the adduct are equal to the kinetics of hydroxyl radical decay and can be used to obtain the absolute reaction rate constants of the “invisible” radical of interest following probe methodology pioneered by

Scaiano for other radicals.99

The intermediate 22 has a broad transient absorbance in the visible region with a maximum at 490 nm (Figure 6.1). We independently generated the same radical by photolyzing the dimer (23) in acetonitrile. The transient spectrum obtained from the dimer was identical to that produced from 21 and confirmed the assignment of the 490 absorption to the pyrithiyl radical. Laser Flash Photolysis (LFP) of PSH (21) in the presence of trans-stilbene generates an absorbance signal at 392 nm in acetonitrile

(Figure 6.1). Analysis of this signal indicates that it is the sum of two chemical processes, on top of the slight bleaching of the precursor: one process involves a reactive intermediate which is rapidly formed on the nanosecond time scale (Figure 6.2). The other process involves an intermediate which is formed and is persistent on the microsecond time scale (Figure 6.3). The second process was observed in the absence of trans-stilbene, and originally by Aveline et al.102, 103 who attributed the signal to secondary reactions of 22 with the precursor 21.

102

S S 355 nm NOH N + HO

21 22

S S S S

N + NOH N N + HO

trans-stilbene OH rapid OH CH CN 3 competing substrates "Reporter" species absorbing at 392 nm slow e.g. benzene

H O 2 OH + H

CH2CN

Scheme 6.1. Photolytic decomposition of N-hydroxypyridin-2-thione.

103

0.15

0.10 S

0.05

0.00

-0 .05 Transient Absorbance (AU) Absorbance Transient

-0 .10 Immediately after laser pulse +100 ns +500 ns +1500 ns -0 .15 350 400 450 500 550 wavelength (nm)

Figure 6.1. Transient absorbance spectra obtained following laser flash photolysis of an acetonitrile solution of 0.65 mM 21 (PSH) with 355 nm light in the presence of 15 mM trans-stilbene.

0.15 20

O.D. 0 ∆ -3 0.10 -20x10

0.0 1.0 µs 0.05 O.D. ∆

0.00

-0.05 20 40 60 80 100 120 140 160 ns

Figure 6.2. Growth curves measured at 392 nm following photolysis of 0.65 mM 21 in the presence (blue) and absence (red) of 15mM trans-stilbene. The inset shows the transient absorption spectrum after photolysis (355 nm) of 21 at longer time scales in the absence of trans-stilbene. 104

0.20

0.15

0.10 394nm

0.05 Absorbance

0.00 PSH alone PSH in the presence of stilbene difference curve -0.05

0.0 0.5 1.0 1.5 2.0µs time

Figure 6.3. Kinetics curves measured at 392 nm following photolysis of 0.65 mM 21 in the presence (blue) and absence (black) of 15 mM trans-stilbene. The red curve is a difference curve, showing the rapid formation and decay of the intermediate generated by reaction of hydroxyl radical with trans-stilbene.

Verification of the stilbene-OH adduct

The photochemical process accompanied by the generation of hydroxyl radical from PSH seems to be complicated. In addition, hydroxyl radical is likely to abstract one hydrogen atom from acetonitrile to form a cyanomethyl radical. Calculations indicate that this process will be exothermic by 22 kcal/mol in the gas phase. It was therefore necessary to ensure that the rapid-growth transient signal was not due to the following processes:

The absorbance of trans-stilbene is poor at 355 nm, and thus it is unlikely that direct irradiation followed by intersystem crossing will yield triplet stilbene under the

105 experimental conditions. Triplet energy transfer from a thione triplet state is also unlikely, since the quantum yield for intersystem crossing in PSH is low (Φ approx.

0.05)102, 103, relative to N-O bond cleavage.

Photoionization of the precursor is unlikely to be a problem, since this process is

“instantaneous” within laser pulse and is much faster than the growth in a timescale of tens of nanoseconds we observed.

We attempted to determine whether the rapid-growth transient absorbance at 392 nm was due to addition of PS• to trans-stilbene. If we assume that the rapid-growth transient is due to addition of PS• to trans-stilbene, it follows that the rate of decay of the signal observed at 493 nm (the absorption maximum of PS•) should increase in the presence of trans-stilbene in solution. The decay of the absorbance signal at 493 nm in

6 -1 6 -1 the presence (kobs = 2.2 ×10 s ) and absence (kobs = 2.6×10 s ) of 15 mM trans-stilbene remains effectively the same. Thus, the decay at 493 nm cannot be due to addition reaction of PS• with trans-stilbene and neither can the growth of absorption at 392 nm.

In a related experiment, we attempted to generate PS• via photolysis of its dimeric disulfide at 355 nm. Once again, a signal was observed at 493 nm. However, the decay kinetics of PS• generated in this fashion proceeds more slowly when PS• is generated from the disulfide precursor than when the radical is produced from PSH. In the latter case, the decay of PS• is accelerated by the reaction of PS• with PSH. Therefore, the addition of PS• to trans-stilbene is not responsible for the growth of probe signal at 392 nm.

It is conceivable that PS• might react with acetonitrile solvent to produce the cyanomethyl radical. Hydroxyl radical can also abstract a hydrogen atom from

106 acetonitrile to form the cyanomethyl radical. We also measured the observed rate constant of formation of the transient species absorbing at 392 nm in CCl4 (Figure S6). A plot of kobs versus [trans-stilbene] is linear (Figure S7).The reactive intermediate formed by photolysis of PSH in CCl4 reacts with trans-stilbene with an absolute second-order rate constant of (8.9 ± 1.0)×109 M-1s-1, a value similar to that obtained in acetonitrile. We think that it is unlikely that two different solvent-derived radicals will react with trans- stilbene with large and similar rate constants. The most economical interpretation of our data is that hydroxyl radical is reacting with trans-stilbene to form the species with transient absorption at 392 nm.

The question of addition of cyanomethyl radical to trans-stilbene may be addressed by photolysis of PSH in the presence of trans-stilbene in a non-hydrogen- atom-donating solvent. Thus, PSH was photolyzed in the presence of trans-stilbene in

Freon 113. In this solvent, the absorbance of PSH is somewhat shifted, but the transient absorption spectrum retains many of the same features (Figure 6.4). In addition, the signal observed at 410 nm in this experimental system consists of a rapid growth feature, and a slower growth feature (Figure 6.5). We observe the same qualitative behavior in

Freon 113 as we do in acetonitrile, and it therefore appears that the cyanomethyl radical plays no part in the rapid short-term behavior of the signal at 392 nm in the experimental system.

107

0.2 PSH in freon-113 (A355=2.2) upon 355 nm LFP 10 ns after laser pulse

0.1

0.0

-0.1 Transient Absorbance

-0.2 350 400 450 500 550 wavelength, nm

Figure 6.4. The transient spectrum of PSH in Freon-113 upon 355 nm LFP, 10 ns after laser pulse.

0.20 PSH in freon-113 (A355=2.2) upon 355 nm LFP 0.15 402 nm kinetics 410 nm kinetics, with satuated stilbene 0.10

0.05

0.00

-0.05

-0.10

0.0 0.5 1.0 1.5 2.0 µs

Figure 6.5. Comparison of kinetic traces monitored at 402/410 nm, of the transient species whose spectra are given in Figure 6.4, in the presence and absence of saturated trans-stilbene

108

Stern-Volmer measurements

Those results ruled out the possibility of the contribution of PS radical and canynomethyl radical to the transient absorbance at 392 nm. It is clear that the 392 nm absorption is the superposition of two processes, with the least possible interferences of other species mentioned previously. The first process relates to the formation of the hydroxyl-trans-stilbene adduct, and the second being the PS radical (22) adding to PSH to form the dimer. The rate of the second process, can be expressed as k[PSH]. The concentration of PSH has been optimized to be about 500 to 600 µM in order to obtain the maximal transient signal intensity. Therefore, the observed rate will typically be of the order of 106 s-1, corresponding a lifetime of about 1µs under these conditions. The concentration of the probe, trans-stilbene, and the substrates that will later be added are controlled such that rate constant of formation measured be about 5×106 s-1 and 5×107 s-1

. Then, to a reasonable degree of approximation, the slow process will be considered

“persistent” in the ns regime, and the growth of the fast process can be fit to a single exponential manner.

The observed pseudo-first-order rate constant for the fast process, kobs, then, can be expressed as:

kobs = 1/τ + k [trans-stilbene]

where τ is the lifetime of the short-lived reactive intermediate (which reacts with trans-stilbene) in the absence of trans-stilbene and k is the second-order rate constant for reaction of that reactive intermediate with trans-stilbene. Thus, plots of kobs versus trans- stilbene concentration should be linear. And indeed, the observed kobs is linearly

109 dependent on the concentration of trans-stilbene (Figure 6.6). This plot of kobs vs. [trans- stilbene] generates a slope of (6.1 ± 0.2) × 109 M-1s-1 at ambient temperature.

A similar LFP experiment was also performed with the hydroxyl-radical precursor, 21, in CCl4 as a function of trans-stilbene concentration (Figure 6.7). In CCl4, a species is generated which reacts with trans-stilbene with a rate constant of 9 ×109 M-

1s-1, and with a lifetime of ~1 µs in the absence of trans-stilbene. Therefore, the species which reacts with trans-stilbene has roughly the same reactivity and lifetime in CCl4 and

CH3CN. This experimental further rules out the possible participation of cyanomethyl radical in the formation of the 390 nm transient.

Figure 6.6 provides the absolute rate constant for reaction of hydroxyl radical with trans-stilbene (via all modes of reaction) of (6.1 ± 0.2) × 109 M-1s-1 in acetonitrile at

298 K. The intercept of this curve provides an estimate of the lifetime of hydroxyl radical in the presence of all other species in this system, including acetonitrile, in the absence of trans-stilbene. The derived lifetime of hydroxyl radical in acetonitrile was about 50 ns. This provides an upper limit for the rate constant for reaction of hydroxyl radical with pure acetonitrile of 1.0 × 106 M-1s-1, assuming that neat acetonitrile has a concentration of 19.2 M at 298 K. This rate constant is approximately a factor of 20 smaller than the value measured for the same reaction in aqueous solution.104 This observation is consistent with subsequent observations of the kinetics of the reaction of hydroxyl radical with aromatic hydrocarbons, both in qualitative behavior and in magnitude, which are discussed in section 6.3.2.

There are a number of potential sites on trans-stilbene where hydroxyl radical may attack. Poole has performed DFT calculations105 to determine which of the potential

110 intermediates (Scheme 6.2) was responsible for the observed transient absorbance. The

B3LYP/6-31+G**//B3LYP/6-31G* level calculations were performed to examine the thermodynamics stabilities of the radical adducts arising from hydroxyl radical addition to different positions of the benzene ring, as well as the alkene carbon. The absorbance spectrum for each was predicted by calculating the gas-phase vertical transitions for each potential intermediate using time-dependent density functional theory. Poole’s results indicate that the ortho- and para- substituted hydroxyl radical adducts are the thermodynamic favored products, and products with UV-vis absorptions around 390 nm.

HO OH +

HO OH

Scheme 6.2. Potential Intermediates from the Reaction of Hydroxyl Radical with trans- stilbene.

111

6 150x10

100 -1 s -1 k, M 50

-3 0 5 10 15 20 25 30 35x10 [stilbene], M

Figure 6.6. Observed pseudo-first-order rate constant for the growth of transient absorbance at 392 nm following laser flash photolysis of 21 in acetonitrile as a function of the concentration of trans-stilbene. (Slope= 6.1 ± 0.2 × 109 M-1s-1)

Figure 6.7. Observed pseudo-first-order rate constant for the growth of transient absorbance at 392 nm following laser flash photolysis of 21 in CCl4 as a function of the concentration of trans-stilbene. (Slope = 8.9±1.0 ×109 M-1s-1)

112

We continued to use this LFP method to measure the kinetics of hydroxyl radical with benzene and naphthalene, molecules we may consider to be models of larger PAHs.

The rate coefficient for reaction of hydroxyl radical with benzene was measured in two ways: by measuring the dependence of the fast pseudo-first-order growth rate constant with the concentration of benzene (Figure 6.6), and by measuring the dependence of the maximum transient absorbance intensity with the concentration of benzene.

For the first method, kobs can be expressed as:

kobs = 1/τ + k [trans-stilbene] + k’[benzene]

where k’ is the absolute second-order rate constant for reaction of hydroxyl radical with benzene. Thus, the slope of Figure 6.8 yields the rate constant of interest k’ for hydroxyl radical reaction with benzene.

The second method is a Stern-Volmer analysis obtained by monitoring the transient absorption at 392 nm. This appraoch enables us to determine the relative rate coefficients for competitive reactions of hydroxyl radical with trans-stilbene and a given aromatic substrate according to the equation below.

o Astilbene-HO = 1 + kArene τ Astilbene-HO

where A˚stilbene-HO is the optical yield of the stilbene-HO adduct in acetonitrile,

Astilbene-HO is the optical yield of the stilbene-HO adduct in the presence of stilbene and arene, kArene is the absolute second-order rate constant for the reaction of arene with HO radical, and τ is the lifetime of HO radical in the presence of constant trans-stilbene

9 concentration but in the absence of arene. We have shown that kstilbene = 6.1 ± 0.3 x 10

M-1s-1 in acetonitrile at ambient temperature. At the concentrations of stilbene of 0.010 –

113

0.015 M used in this work, kstilbene[stilbene] >> kT (where kT represents the sum of all rate constants for all processes which consume hydroxyl radical in the absence of stilbene) and

o Astilbene-HO kArene [Arene] = 1 + k [stilbene] Astilbene-HO stilbene

The concentration of stilbene is kept constant with a particular arene. It is important to note that kArene and kstilbene are the total of all rate constants for reaction of

HO radical with the arene and stilbene, respectively, at all sites and by both addition and hydrogen-atom abstraction mechanisms.

The maximum transient absorbances (A) at 392 nm, which were present 50 – 150

o ns after the laser pulse, were taken to be the values of A 392 (trans-stilbene only) and A392

(trans-stilbene plus arene) in the Stern-Volmer analysis (Figure 6.8). In Figure 6.8, the background signal (zero trans-stilbene) was not subtracted from the signal obtained in the presence of trans-stilbene and arene.

114

6 75x10

70 -1

, s 65 obs k

0.00 0.05 0.10 0.15 0.20 0.25 [benzene], M

Figure 6.8. Observed pseudo-first-order rate constant (kobs) for the growth of transient absorbance at 392 nm following LFP of 21 in acetonitrile with 12 mM trans-stilbene as a function of the concentration of benzene. (Slope =1.1 ± 0.1 × 108 M-1s-1)

7.0 0.30

0.25 [Benzene] = 0.0 M 6.0 0.20

(A.U) 0.15 39 5

0.10 5.0 Baseline Absorbance 0.05 4.0 0.00 395nm -0.05 -6

A 0.0 0.5 1.0 1.5x10

/ tim e (s) 3.0 0,395nm

A 2.0

1.0

0.0 020406080100

[Substrate]/[trans-Stilbene]

Figure 6.9. Stern-Volmer plots for benzene (open squares), benzene-d6 (open diamonds, offset by -0.5 for clarity) and naphthalene (open circles) as a function of the substrate/trans-stilbene relative concentration.

115

The rate coefficients for reaction of hydroxyl radical with benzene-d6 and naphthalene were determined only by Stern-Volmer kinetic analyses (Figure 6.9.). The

Stern-Volmer treatment also yields the ratio of the sum of all rate constants for hydroxyl reaction with the substrate (abstraction of each type of hydrogen and addition to each carbon) relative to the sum of all rate constants for reaction with trans-stilbene

(abstraction of each type of hydrogen and addition to each carbon). The experimental results are shown in Table 6.1.

For benzene, the results obtained by both methods are consistent with one another within experimental uncertainty, and the results obtained for benzene and benzene-d6 yield a kinetic isotope effect (KIE) of 1.0 ± 0.3, indicating that the reaction of hydroxyl radical with this compound is dominated by radical addition, and but not by hydrogen abstraction from the benzene ring.

a 2b kSubstr, ACN kSubstrate, aq Substrate Method kSubstr/kStilbene R -1 -1 c -1 -1 d (M s ) (M s ) benzene Direct (8.3 ± 1.2) × 107 7.8×109 e S-V 0.0209 ± 0.0012 0.98 (1.1 ± 0.1) × 108 7.8×109 e 8 9 f benzene-d6 S-V 0.0198 (1.2 ± 0.1) × 10 4.7×10 naphthalene S-V 0.290 ± 0.007 1.00 (1.8 ± 0.1) × 109 9.4×109 g a Direct = direct measurement of growth kinetics, S-V = Stern-Volmer competitive analysis. b Goodness-of-fit parameter. c Measured by LFP in acetonitrile (ACN), this study. d Measured in aqueous solution by LFP and/or pulse radiolysis. e Reference 106. f Reference 107. g Reference 108.

Table 6.1: Kinetic data for reaction of hydroxyl radical with aromatic hydrocarbons in acetonitrile (ACN) and water (aq).

116

6.3.2 The rate constant differences in water and in acetonitrile

Perhaps the most remarkable feature of the data in Table 6.1 are the absolute second-order rate constants for reaction of hydroxyl radical with these compounds in acetonitrile in comparison with those obtained in aqueous solution.108 Initially, we expected that hydroxyl radical would exhibit lower reactivity in water, since it may form hydrogen bonds with water and form a solvation shell109-an option not readily available for a polar, but non-hydroxylic, solvent such as acetonitrile. The data obtained show the exact opposite trend-the rate coefficient for reaction of hydroxyl radical with naphthalene is some 5.8 times slower in acetonitrile than in water, and the rate coefficient for reaction of hydroxyl radical with benzene is some 71 times slower in acetonitrile than in water.

To test the veracity of this unexpected result, we sought to independently measure this ratio by analysis of the mixture of products formed upon photolysis (365 nm) of N- hydroxypyridine-2-thione (21) in acetonitrile containing benzene and naphthalene.

Twenty-two measurements were made by gas chromatography using three different ratios of the aromatic reagents at concentrations comparable to those used in the LFP experiments. By GC product studies, the ratio of knaphthalene to kbenzene obtained at ambient temperature was varied from 7 to 24 with a mean of 14.4 (N=22). The large standard deviation is due to the low absolute yields of phenol and of the 1- and 2-naphthol adducts. This result is in excellent agreement with the same ratio of rate constants (13.8) as that determined by the LFP approach. The comparison of the knaphthalene to kbenzene is listed in Table 6.2.

We also photolyzed N-hydroxypyridine-2-thione (21) in neat benzene containing naphthalene. Under these conditions, the absolute yields of phenol and the mixture of

117 naphthols were increased by a factor of at least ten. This led to a ratio of rate constants of

6.3 in this nonpolar organic solvent mixture, and the ratio was obtained with greater reproducibility than in acetonitrile. In contrast, the ratio of knaphthalene to kbenzene in water is

1.2.108 The GC study confirms that this ratio changes dramatically in organic solvents.

Computational studies by DeMatteo and Hadad110 at the B3LYP and CBS-QB3 levels confirmed the rate enhancement of hydroxyl radical addition to benzene by calculation of the transition states in the presence of explicit solvent molecules as well as a continuum dielectric field. They discovered that an extremely stable complex between the solvent and hydroxyl radical is not the origin of the observed differences in reactivity.

The origin of the rate enhancement lies entirely in the structures of the transition states and not in the pre-reaction complexes. The calculations reveal that the hydroxyl radical moiety becomes more anionic in the transition state and, therefore, looks more like hydroxide anion. In the transition states, solvation of the incipient hydroxide anion is more effective with water than with acetonitrile and provides the strong energetic advantage for a polar solvent capable of hydrogen bonding. At the same time, the aromatic unit looks more like the radical cation in the transition state.

Their computational results demonstrate that the ratio of the rate constants of one solvent molecule complexed hydroxyl radical adding to benzene ring in water and acetonitrile is 53:1, at B3LYP/6-311+G*//B3LYP/6-31+G** level and 406:1, at CBS-

QB3 level. When two molecules of solvent are complexed with included, the ratio was calculated to be 222:1 at B3LYP/6-311+G*//B3LYP/6-31+G** level.110 As summarized in Table 6.3, these results exhibits a good agreement with the ratios of these rate constants we obtained from three experimental methods.

118

Method knaphthalene/kbenezne In watera 1.2 In benzene 6.3 In acetonitrile Stern-Volmer 13.8 Product analysis 14.4 a Reference 101

Table 6.2. Comparison of hydroxyl radical reacting with naphthalene and benzene in different solvents.

Method kwater/kacetonitrile Experimental Direct 94:1 Stern-Volmer 71:1 Computational a B3LYP/6-311+G*//B3LYP/6-31+G** 53:1 a CBS-QB3 406:1 b B3LYP/6-311+G*//B3LYP/6-31+G** 221:1 a One molecule of solvent is complexed with hydroxyl radical b Two molecules of solvent are complexed with hydroxyl radical

Table 6.3. Comparison of the ratio of reaction rate constants of hydroxyl with benzene in water and acetonitrile.

119

6.3.3 Measurement of rate constants of hydroxyl radical with hydrocarbons

The same Stern-Volmer analysis performed with a set of aromatic hydrocarbons.

Figure 6.10 was obtained from competitive experiments between trans-stilbene and toluene, m-xylene and p-xylene at 298 K. The relative rate coefficients for reaction of hydroxyl radical with trans-stilbene and aromatic substrates determined by Stern-Volmer analysis of kinetic data obtained by this method are summarized in Table 6.4. Where data are available, we have included the values obtained by LFP and PR studies in aqueous solutions.101

120

Figure 6.10. Stern-Volmer data obtained from competitive experiments between trans- stilbene and toluene ( ), m-xylene ( ), and p-xylene ( ) at 298 K. The dotted lines are lines of best fit, and the relative rate coefficients obtained from these experiments are shown in Table 6.4.

121

a 2 b c d d Substrate kArene/kStilb R kArene,ACN kArene,aq IP (M-1s-1) (M-1s-1) (eV) benzene 0.0209 ± 0.0012 0.98 (1.1 ± 0.2) × 108 7.8 × 109 f 9.243 8 9 f benzene-d6 0.0198 (1.1 ± 0.2) × 10 4.7 × 10 toluene 0.0686 ± 0.99 (3.7 ± 0.2) × 108 5.1 × 109 8.828 0.0034 8 toluene-d3 0.0620 (3.4 ± 0.2) × 10 8 toluene-d5 0.0571 (3.1 ± 0.2) × 10 8 toluene-d8 0.0587 (3.2 ± 0.2) × 10 o-xylene 0.107 (5.8 ± 0.3) × 108 6.7 × 109 8.56 m-xylene 0.130 ± 0.007 0.99 (7.0 ± 0.4) × 108 7.5 × 109 8.55 p-xylene 0.107 ± 0.008 0.97 (5.8 ± 0.3) × 108 7.0 × 109 8.44 mesitylene 0.241 ± 0.008 1.00 (1.3 ± 0.1) × 109 6.4 × 109 8.40 durene 0.232 ± 0.008 1.00 (1.3 ± 0.1) × 109 8.06 naphthalene 0.290 ± 0.007 1.00 (1.6 ± 0.1) × 109 9.4 × 109 8.144 1-methyl 0.411 ± 0.022 0.98 (2.2 ± 0.1) × 109 7.96 naphthalene 2-methyl 0.462 ± 0.021 0.99 (2.5 ± 0.1) × 109 7.91 naphthalene 1,3-dimethyl 0.701 ± 0.027 0.91 (3.8 ± 0.2) × 109 7.86 naphthalene 1,4-dimethyl 0.845 ± 0.065 0.97 (4.6 ± 0.3) × 109 7.80 naphthalene biphenyl 0.111 ± 0.007 0.98 (6.0 ± 0.3) × 108 9.5 × 109 8.16 diphenylmethane 0.128 ± 0.011 0.97 (7.0 ± 0.4) × 108 8.9 bibenzyl 0.111 ± 0.007 0.98 (6.0 ± 0.3) × 108 8.9 trans-stilbene (5.4 ± 0.4) × 109 7.656

a Determined by Stern-Volmer analysis (298 K). b R2, linearity of Stern-Volmer Plot. c Rate constants in acetonitrile as determined in this work. d Reference101 Table 6.4. Summary of kinetic data for reaction of hydroxyl radical with aromatic hydrocarbons.

122

Solvent Effects

As we have discussed previously, the rate coefficients for reactions of hydroxyl radical with arenes in acetonitrile are significantly smaller than those observed in water, an unexpected result. High-level ab initio and density functional theory calculations indicates the transition state for the ionic hydroxyl radical/benzene system is stabilized to an greater extent than the starting individual species in water than in acetonitrile.

For the data available, the ratio of rate coefficients for HO• reaction with arenes in water relative to acetonitrile (Table 6.3) varies from a factor of 65 for benzene down to a factor of 4 for mesitylene. It is expected that as the measured rate coefficient in acetonitrile approaches the diffusion-controlled limit, this factor will become smaller and approach a limiting value of approximately 0.37. This value is calculated from the measured viscosities of these solvents at 293 K (0.37 cP for acetonitrile and 1 cP for water) and assumes no change in the reactants' hydrodynamic radii moving from water to acetonitrile. The qualitative behavior of the rate-coefficient ratios is consistent with such an expectation.

Kinetic Isotope Effects (KIE)

The absence of a significant KIE for benzene and benzene-d6 is consistent with the fact that the dominant reaction of hydroxyl radical with benzene is addition to the ring rather than hydrogen-atom abstraction at the phenyl C-H position.111 In fact, the KIE for hydroxyl-radical reaction with benzene in water was measured by Dorfman and co-

107 workers. These authors obtained the same rate constant for benzene-h6 and benzene-

9 -1 -1 d6, specifically 4.7 × 10 M s , and therefore no KIE effect was observed in water. (We

123 should note that, since then, there have been multiple measurements of the reaction of hydroxyl radical with benzene, and the consensus rate constant for benzene-h6 in water is

9 -1 -1 101, 106 slightly faster at 7.8 × 10 M s . ) Our acetonitrile rate constants for benzene-h6 and benzene-d6, as well as the aqueous results of Dorfman, concur in that there is no significant KIE for HO• radical reactions with benzene. These results are consistent with a dominant radical-addition reaction with the aromatic ring.

The measured KIE for the toluene isotopomers allows us to determine the importance of hydroxyl radical addition to the aromatic ring relative to hydrogen-atom abstraction from a benzylic position, a more thermodynamically favored process than abstraction from a phenyl C-H position. The hydrogen-atom abstraction reaction would be expected to exhibit a significant primary KIE. However, comparison of the relative rate coefficients of toluene and toluene-d3 yields a KIE of 1.1. Such a result tends to indicate that abstraction at the benzylic position represents a minor reaction pathway, consistent with results obtained previously in the gas phase.112 It appears that the major influence of the methyl group in the increased reactivity of toluene relative to benzene

(an approximately 3-fold increase) is the activation of the aromatic ring. This is confirmed by comparison of toluene with toluene-d5 and toluene-d8. The KIE measured for addition to the ring (toluene-d5) is small, approximately 1.2, as would be expected if

H-atom abstraction occurs but is only a minor contributor to the overall reaction rate, and the KIE is little changed by deuteration at the benzylic position (toluene-d8).

It should be noted as well that a number of rate constants for hydroxyl radical reactions have been measured in the gas phase.101 The relative ratio of rate constants between benzene and p-xylene is ~12 in the gas phase and only ~5 in acetonitrile between

124 the two substrates. The participation of multiple reaction pathways (addition vs abstraction) and loose complexes on the entrance channel (as the reactants approach each other) may be responsible for this difference.

Structure-Reactivity Relationships

The KIE data for toluene indicate that the relative importance of hydrogen-atom abstraction from the benzylic position is low compared to addition to the aromatic ring.

However, as more methyl groups are added to the system under study, we assume that abstraction from a benzylic site should become increasingly probable. Of course, the addition of methyl groups to an aromatic system will also activate the aromatic ring, particularly toward an electrophilic radical such as the hydroxyl radical. We have considered the data obtained from two approaches: one is a purely probabilistic analysis; the other based on a frontier molecular orbital model.

The probabilistic approach is a simple examination of the rate coefficient ratio for the system under study relative to its "parent" compound. In the case of benzene and naphthalene derivatives, the "parent" compounds are benzene and naphthalene, respectively; for the remaining compounds in Table 6.4, biphenyl is chosen as the

"parent" compound. The relative rate coefficients are then plotted against the ratio of possible hydrogen-atom abstraction sites and hydroxyl-radical addition sites (i.e., the number of benzylic hydrogen atoms/the number of ring carbon atoms). The plot obtained is shown in Figure 6.11 and appears to show some degree of correlation. However, such an approach does not explain why mesitylene is as reactive as durene or indeed why methyl- and dimethylnaphthalene isomers have significantly different reactivity.

125

Moreover, if the probabilistic argument holds true, we would expect that increasing the probability of hydrogen-atom abstraction (presumably the slower of the possible reactions, based on the KIE data from toluene isotopomers) should decrease the overall rate of reaction. Therefore, it seems likely that the major influence of the methyl groups is ring activation.

126

15.00

12.00

9.00 Parent Arene k / 6.00 Arene k

3.00

0.00 0.00 0.50 1.00 1.50 2.00 2.50

No. of Abstraction Sites/No. of Addition Sites

Figure 6.11. Correlation of the rate coefficient for the reaction of hydroxyl radical with arenes in acetonitrile with a probability-based structure reactivity parameter. The parent arenes used to define the relative rate coefficients are identified in the text.

127

Computational studies of the reaction of hydroxyl radical with benzene110 indicate that charge-transfer interactions are important in the transition state. In particular, the transition state resembles the hydroxide anion coordinated to the radical cation of benzene. This is in excellent agreement with our calculations of the benzene-chlorine atom complex.113 The frontier molecular orbital model, as developed by Fukui,114 has been successfully applied to the reactions of alkyl radicals with alkenes by Fischer,

Radom, and others.115 This model indicates that the stabilization due to delocalization of an electrophilic radical-arene system as it evolves from reactant to transition state will depend on the energies of the arene highest-occupied molecular orbital and the radical

SOMO. Of course, the experimental parameters are the ionization potential (IP) and

115 electron affinity (EA) of the arene and HO• radical, respectively. Fischer et al. have used a state correlation approach that uses these same parameters to model the energies of the charge-transfer states: the lower the energy of the charge-transfer states, the greater the configurational mixing and stabilization of the ground state. The term "ground state" refers to the lowest electronic energy level of the transition state of reaction. Favorable interactions of the charge-transfer states are those characterized by large values of the

parameters (IParene - EAHO•) or (IPHO• - EAarene). In this particular case, we will use the first of these parameters as our structure/reactivity parameter.

The IPs for the arenes used in this study have been determined experimentally in the gas phase and are given in Table 6.3. Similarly, the electron affinity of hydroxyl radical has been measured in the gas phase and has a value of 1.828 eV. A plot of

log(kareneMs) against (IParene - EAHO•) is shown in Figure 6.12.

128

The correlation of the data is quite good, particularly when one considers that the rate coefficients were determined in acetonitrile solution, whereas the structure-reactivity parameter we used was derived solely from gas-phase data. However, it is worth noting that, for most of these compounds, radical addition to the aromatic ring was the predominant reaction pathway; no other modes of reaction of hydroxyl radical competed with this process to any significant extent. The data obtained for trans-stilbene does lie on the correlation curve despite the fact that an attractive, alternative mode of reaction is available to the hydroxyl radical, namely, addition to the alkene double bond. It is difficult to determine whether hydroxyl addition to the alkene carbons competes to any significant extent with the addition to the aromatic ring, but it is worth noting that the phenyl rings may represent a possible steric barrier to reaction at the double bond of trans-stilbene. In addition, the transient absorbance observed at 392 nm from reaction of hydroxyl radical with trans-stilbene, the basis of our LFP method, appears to be due to the addition of hydroxyl radical to the aromatic ring ortho and or para to the double bond, based on our earlier calculations. It is not clear whether a similarly good correlation would be obtainable for multiply functionalized compounds, and these systems will be studied further in the future.

129

10.0

y = -1.033x + 15.734, R2 = 0.960

9.5 .Ms)

9.0 Arene

k diphenylmethane ( biphenyl bibenzyl 10

log 8.5

8.0 5.5 6.0 6.5 7.0 7.5 8.0 (IP -EA ) (eV) Arene HO•

Figure 6.12. Correlation of the rate coefficient for the reaction of hydroxyl radical with arenes in acetonitrile with the charge transfer structure/reactivity parameter (IPArene –

EAHO•). The dashed line and equation correspond to the line of best fit for the data obtained, excluding the data for biphenyl, bibenzyl and diphenylmethane. If those data are included, the line of best fit is y = –0.877x + 14.75, R2 = 0.835.

130

6.4 Conclusions

Using our LFP-based methodology, we have determined the rate coefficients for the reaction of hydroxyl radical with a number of monocyclic and polycyclic aromatic hydrocarbons in acetonitrile. From an analysis of the observed kinetic isotope effects for the reaction of hydroxyl radical with isotopomers of benzene and toluene, we conclude that at least for simple aromatic hydrocarbons, the predominant reaction pathway in acetonitrile is the addition of the hydroxyl radical to the aromatic ring, rather than hydrogen-atom abstraction from the phenyl, or somewhat surprisingly, benzylic C-H positions. Structure-reactivity analysis confirms that addition reactions play a dominant role in more complex systems in acetonitrile: the observed structure-reactivity behavior of these compounds cannot be described in terms of a simple probabilistic model.

Structure-reactivity analysis, based upon frontier molecular orbital and state correlation models indicate that charge-transfer interactions between hydroxyl radical and a given arene play an important role in the stabilization of the transition state for the reaction, an observation in concordance with our earlier computational study of the reaction of hydroxyl radical with benzene. A good correlation is observed for the rate

coefficient of reaction with the charge transfer parameter (IPArene – EAHO•).

Observation of the different reaction rates of hydroxyl radical toward aromatics in water and acetonitrile is consistent with the commonly held view of hydroxyl radical as being "electrophilic" when reacting with electron-rich aromatic rings, such as DNA bases, for which the radical cation can be well stabilized by the organic substrate.

However, such an electrophilic role of hydroxyl radical will be heavily dependent on the

131 ability of the organic substrate to stabilize the incipient radical cation. DNA bases, due to their electron-rich nature, are likely to favor such radical cations.

132

CHAPTER 7

THE REACTION BETWEEN RADICALS AND TIRAPAZAMINE: STUDIES OF

THE MECHANISM OF ACTION OF TIRAPAZAMINE

7.1 Introduction

Active radical species

Different theories of the formation and oxygenation of DNA carbon-centered radicals by tirapazamine have been proposed by Gates et al.46, 75 and by Denny et al.48, 49.

Based on the non-selective pattern of DNA cleavage and the inhibition of this cleavage by radical scavengers, Daniels and Gates46 proposed in 1996 that the active species which is responsible for the cytotoxicity is the hydroxyl radical (Scheme 7.1) However,

Patterson’s ESR attempt to spin trap and detect the hydroxyl radical was inconclusive47, casting doubts on this proposal.

In 2003, Denny and Anderson proposed that the active species is the benzotriazinyl radical (BTA, Scheme 7.1).48 Benzotriazinyl radical is the reaction product produced from the hydroradical (TPZ-H•) formed by elimination of a water molecule (Figure 7.1). Another pathway, a disproportionation reaction that generates

133

TPZ, dTPZ and water, is operative only when the concentration of hydroradical, e.g., when generated from pulse radiolysis, is very large.

- e aq

TPZ + dTPZ + H2O Radiolysis

k2, disproportionation

O e- (enzyme) O O O N + N H N ? N N N N N + OH N NH2 N NH2 N NH2 N NH2 O O OH

TPZ O O2 TPZ TPZ-H dTPZ

k1, -H2O hv Donor KSCN, NaI, etc. Photolysis O * N O N N N N NH BTA N NH2 O

1TPZ*

Scheme 7.1. Proposed different mechanisms of action of tirapazamine.

The Denny group produced the benzotriazinyl radical by pulse radiolysis of 4-

•- desoxytirapazamine (Scheme 7.1) and oxidation by SO4 , followed by disproportination

(Scheme 7.2). The benzotriazinyl radical (BTA) was found to have a one-electron reduction potential of 1.31 V and reacts with dGMP and 2-desoxyribose with rate constants of (1.4±0.2)×108 M-1s-1 and (3.7±0.5)×106 M-1s-1, respectively. The BTA

134 radical was also found to absorb between 330 to 430 nm with a lifetime of several milliseconds (Figure 7.1).

O O O HO HO N SO N R N R N 4 N O N + + O N NH H N NH2 HO N NH2 HO dTPZ BTA dTPZ

(R = OH, guanine)

Scheme 7.2. The generation of BTA radical by oxidation of dTPZ and its subsequent reactions with 2-disoxyribose.

Figure 7.1 Transient UV-vis absorption of BTA radical (triangles), solid triangles are points corrected for the absorption of the parent dTPZ (solid line), adapted from Reference 48.

135

Tirapazamine is a surrogate of oxygen

The hypoxia-activated mechanism of TPZ offers a paradox: on the one hand, significant levels of the drug-derived DNA-cleaving radical are produced only under low- oxygen concentration conditions, while on the other hand, it is well-known that O2 enhances radical-mediated DNA strand cleavage in vivo. This effect of oxygen on radical-mediated DNA strand cleavage in biological systems reflects the ability of O2 to compete effectively with endogenous thiols for reaction with DNA radicals. Reaction of

O2 with desoxyribose radicals in DNA (sometimes referred to as “fixing” the radical lesion) generally leads to DNA strand cleavage, whereas the reaction of a DNA radical with a thiol generally represents a chemical repair process.116 In 1998 and 1999, Gates,

Greenberg and coworkers showed that, in addition to initiating the formation of DNA radicals, TPZ can react with these radicals and convert them into base-labile lesions.

TPZ traps the C1-nucleotide radical, generated independently in single stranded DNA with a rate constant of 2×108 M-1s-1 and with double stranded DNA with a rate constant of 4×106 M-1s-1.75, 116

O O NH NH O RO N O hν RO N Alkaline N O N O labile tBu O + N NH lesion R'O 2 O R'O O

The essence of this “fixing” process is actually the oxidation of the DNA sugar radical (most likely at C1 or C4) by the drug to from a more facile lesion, a DNA cation, oxy-radical, or carbonyl moiety (Scheme 7.3).

136

Gates et al further proposed that a reaction between TPZ and the DNA sugar radical will promote a net transfer of oxygen atom from the 4-nitrogen atom to the DNA radical, although it is not clear whether or not an intermediate adduct of TPZ and DNA is involved in this transformation or if it is a concerted process. Liu, Hadad and Platz used

Density Functional Theory to show that this is process is very exothermic. They also provided a computational based explanation as to why the two desoxy compounds of

TPZ are unable to kill hypoxic tumor cells. The desoxy TPZ derivatives are inactive because they will not efficiently transfer the N-oxide oxygen to the DNA sugar radical to make the damage permanent and lethal (Scheme 7.4). The most favorable reaction of dTPZ and dTPZ’ with a sugar radical leads to a stable nitroxyl radical according to the

DFT calculation.

137

O N N

N NH2 O TPZ

O O O O B N B B O N O RSH O + OH and/or H H O N NH O O

BTA C1 radical

O2/[O]

O O B B O O DNA strand O or cleavage O O

Scheme 7.3. Fixing and repair of carbon-based DNA radical.

O O O B B H B O O O DNA breakdown O O O O O O N NH2 N NH2 N N N N N NH2 O O N TPZ N O dTPZ O O B B O O N NH N NH O 2 O 2 N N N N O O dTPZ Stable nitroxide radical

Scheme 7.4. Different favored reactions of TPZ and dTPZ with DNA radicals.

138

Denny et al. reported that a desoxyribose radical reacts with TPZ with a rate constant of 2.5×109 M-1s-1.48 From their kinetic studies, they proposed the formation of a

TPZ-desoxyribose adduct. In this reaction the carbon radical adds to the 4-nitrogen atom to form the N4-adduct, followed by rearrangement to form hydroradical (TPZ-H) and an aldehyde. The hydroradical (TPZ-H) can further eliminate a water molecule in a concerted process to produce benzotrianzinyl radical (BTA), which in turn can abstract a hydrogen atom from another DNA sugar to form a new DNA carbon-centered radical. A short chain reaction mechanism was proposed for this mechanism (Scheme 7.5).

O O N N N R N + H HO N NH2 N NH2 O RCH O TPZ O H N4-adduct O R N CH2 R N HO - H O N NH2 O O dTPZ N N - H2O N N

N NH N NH2 OH BTA TPZ-H

Scheme 7.5. Denny’s mechanism of TPZ mediated DNA damage.

139

There are two potential problems with the Denny mechanism. First, the addition of the ribose radical to the N4 atom is difficult to imagine because the ribose radical is bulky and the N4 atom is surrounded by an aromatic ring, an amino group and an oxygen atom. Secondly, and more importantly, the model compound of DNA in the Denny mechanism is a desoxyribose, which has α-hydroxy group at the likely reactive site. The existence of the α-hydroxy group allows the reaction mechanism which forms an aldehyde. In native DNA, the α-hydroxy groups are substituted with phosphate or nucleotides, thus this mechanism can not proceed (Scheme 7.6).

The controversy over the in vivo mechanism of action of TPZ is centered on two key problems: (1) what is the active species causing initial DNA damage and (2) what is the mechanism by which TPZ oxidizes DNA carbon-centered radicals. We designed a series of thermal and photochemical reactions to generate different alkyl radicals. We will directly observe the transient UV-vis spectra of reactive intermediates and measure their kinetics, and perform product analysis, to better uncover the true reaction mechanism.

140

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

DNA desoxyribose

R1 R3 R1 OR3 R1 OH R2 R 2 R2 (R2 = H, R)

O O O N N N N N N

N NH N NH2 N NH2 2 RCH O O O O RCH RCH H H OH O

Daniels-Gates-Greenberg's Denny's Daniels-Gates-Greenberg's O adduct O adduct 4 N4 adduct 4 homolytic cleavage model Rearrangement model

Scheme 7.6. Comparison of different models of the reaction between a DNA radical and TPZ.

141

7.2 Experimental

The instrument used for Laser Flash Photolysis experiments is described in detail in Chapter 2.

Azo-tert-butane, benzoyl peroxide, benzhydrol, t-butyl peroxide, pyridine N- oxide, and isoquinoline N-oxide were purchased from Aldrich. Solvents such as 2- propanol, acetonitrile, dioxane, benzene and acetone were purchased from Aldrich. The synthesis of TPZ, dTPZ, dTPZ’ has been described in detail in Chapter 5.

Thermal reactions were protected by argon saturated with solvent. The reaction mixture were refluxed at the temperature of the boiling point of the solvent at atmospheric pressure. Typically, 1 mM of the aromatic N-oxide and 8-10 equivalents of azo-t-butane or benzoyl peroxide were employed. The thermal reactions were monitored by HPLC of the reaction products.

HPLC analyses were performed on a Beckman Coulter liquid chromatograph equipped with a System Gold 168 diode array detector and controlled using an IBM computer and the 32 Karat 5.0 software package. Separations were achieved on a reverse phase C18, 250 mm × 4.6 mm (5µ), column preceded by a guard column. A mixture of

HPLC grade water and methanol (1:1, vol:vol) was used as eluent at a flow rate of 0.50 mL/min. The identity of the products was established by comparison of their retention times and absorption spectra with those of authentic samples.

142

7.3 Results

7.3.1 Reaction of TPZ with benzophenone ketyl and acetone ketyl radical mixture

Triplet benzophenone sensitization

Triplet benzophenone abstracts the tertiary-hydrogen atom of 2-propanol which initiates the first documented organic photoreaction. We utilized this property of benzophenone to photochemically generate benzophenone ketyl and acetone ketyl radicals in 2-propanol.

OH 3 O O* OH OH hv H + N-oxides Ph Ph ISC Ph Ph Ph Ph

Laser flash photolysis of 7 mM benzophenone in 2-propanol generates the transient UV-vis spectrum shown in Figure 7.2. The 545 nm and 320 nm bands are well known absorptions for benzophenone ketyl radical. The 530 nm triplet benzophenone absorption was not observed in the transient spectra 50 ns after the laser pulse, suggesting an extremely fast reaction of the hydrogen atom abstraction from 2-propanol. At ambient temperature this radical has a lifetime of several microseconds in the absence of oxygen.

In the presence of 0.2 mM TPZ, the transient spectrum produced by LFP of benzophenone exhibits a bleaching around 470 nm (Figure 7.3), which is the wavelength of maximum absorption of ground state TPZ. Kinetic studies (Figure 7.4) also demonstrate that the observed first order decay rate constant (kobs) recorded at 470 nm is very similar to the observed growth rate constant measured at 410 nm. Therefore, the benzophenone ketyl/acetone ketyl radical mixture probably reacts with TPZ to generate a

143 new species that has weak absorption at 410 nm. The species absorbing at 410 nm shows no decay over hundreds of microseconds.

144

0.12 7 mM benzophenone in iso-propanol upon 355 nm LFP, deoxygenated 0.10 ~ 50 ns 545 nm ~ 250 ns after laser pulse 0.08

0.06

0.04

0.02 Transient Absorbance Transient 0.00

-0.02 400 450 500 550 600 Wavelength/nm

Figure 7.2. The transient spectrum produced by LFP of 7 mM benzophenone in 2- propanol at 355 nm.

0.8 320 nm 8.2 mM benzophenone and 0.2 mM TPZ in 2-propanol upon 355 nm LFP 0.6 1 us after 15 us after 0.4 30 us after laser pulse 545 nm

. O.D. 0.2 ∆

0.0 470 nm

-0.2

300 350 400 450 500 Wavelength, nm

Figure 7.3. The transient spectrum produced by LFP of 7 mM benzophenone in 2- propanol, in the presence of 0.2 mM TPZ at 355 nm.

145

0.2

0.1

0.0 C omprarison of kinetics: O.D. 5 -1 ∆ 470 nm, kobs = 1.8x10 s -0.1 5 -1 410 nm, kobs = 1.1x10 s

-0.2

10 20 30 40 50 µs

Figure 7.4. The kinetic curves monitored at 470 and 410 nm of transient species whose spectra are given in Figure 7.3.

3 600x10

kobs of the growth at 410 nm vs. [TPZ] 500 8 -1 -1 slope= 5.2+/- 0.2 x10 M s 400 -1

, s 300 obs k 200

100

-3 0.0 0.2 0.4 0.6 0.8 1.0x10 [TPZ], M

Figure 7.5. The plot of the observed rate constant of the growth of the transient absorbance at 410 nm in Figure 7.3 versus [TPZ].

146

A plot (Figure 7.5) of kobs of the growth of the transient absorbance at 410 nm vs.

[TPZ] gives a slope of 5.2×108 s-1M-1, the combination of the bimolecular rate constant of reactions of benzophenone ketyl and acetone ketyl radical with TPZ.

We performed similar photoreactions using dioxane as a solvent. In this solvent, we expect a radical pair of benzophenone ketyl/α-dioxanyl radicals will form as a result of hydrogen atom abstraction from triplet benzophenone. From Figure 7.6, a shift of absorption from 530 nm to 545 nm is observed on a 500 ns scale. This suggests that the rate of triplet benzophenone abstraction of a hydrogen atom from dioxane is slower than that from 2-propanol.

3 * O OH

O + + O H O O

520 - 530 nm 545 nm

OH O O N ? N N N + N NH N NH2 O H 2 O O

Upon addition of TPZ, as in Figure 7.7, a bleaching at 470 nm was observed again.

Therefore, the mixture of benzophenone ketyl/α-dioxane radicals also reacts with TPZ.

The observed rate constants of decay at 475 nm, the growth at 405 nm, and the decay at

545 nm of benzophenone ketyl radical all have very similar rate constants (Figure 7.8)

These rate constants were plotted versus the concentration of TPZ in Figure 7.9. The

147 slopes give bimolecular rate constants 4.0×108 M-1s-1 and 4.5×108 M-1s-1 at 475 nm and

405 nm, respectively.

Similar experiments were also carried out using ethanol as solvent. A bleaching was observed at 475 nm due to the reaction of benzophenone ketyl/α-ethanol radical mixture with TPZ (Figure 7.10). A plot (Figure 7.11) of the kobs measured at 550 nm and

475 nm give the biomolecular constants of 3.2×108 M-1s-1 and 6.4×108 M-1s-1, respectively.

3 O * OH

+ OH + OH

520 - 530 nm 545 nm

We attempted to photolyze benzophenone in benzene in the presence of a large concentration of benzhydrol. In this experiment, we hoped to generate two identical benzophenone ketyl radicals as a result of triplet benzophenone abstracting a hydrogen atom from benzhydrol. Then we would be able to determine the rate constants of just benzophenone ketyl radical with TPZ. The transient UV-vis spectra in Figure 7.12 and

Figure 7.16 show the generation of benzophenone ketyl radical and its reaction with TPZ.

However, due to the limited solubility of TPZ in benzene, a plot of these observed constants versus [TPZ] could not be obtained.

3 Ph O O* Ph OH 355nm H 2 Ph Ph ISC Ph Ph Ph Ph

148

530 0.3

0.2 O.D. ∆ 0.1 545

0.0

350 400 450 500 550 Wavelength, nm

8.7 mM benzophenone in dioxane upon 355 nm LFP right after 100 ns 500 ns 3000 ns after laser pulse

Figure 7.6. The transient spectrum produced upon 355 nm LFP of 7 mM benzophenone in dioxane upon 355 nm LFP.

0.3

505 0.2 545

475 0.1

O.D. 405 ∆

0.0

-0.1

350 400 450 500 550 Wavlength, nm

8.7 mM benzophenone in dioxane upon 355 nm LFP, in the presence of 0.11 mM TPZ right after 100 ns 1000 ns 5000 ns 20 us after laser pulse

Figure 7.7. The transient spectra produced upon 355 nm LFP of 7 mM benzophenone in dioxane, in the presence of 0.11 mM TPZ.

149

7 -1 545 nm, kdecay=2.1x10 s 0.2 7 -1 475 nm, kdecay=2.1x10 s 7 -1 405 nm, kdecay=2.5x10 s 0.1

0.0 Transient Absorbance Transient -0.1

0 10 20 30 40 50 µs

Figure 7.8. Kinetic curves recorded at 545, 475 and 405 nm of the transient species whose spectra are given in Figure 7.7.

3 250x10 545 nm, slope = 4.0+/-0.2x108 s-1 M-1 8 -1 -1 200 475 nm, slope = 4.5+/-0.4x10 s M 405 nm

-1 150 , s obs k 100

0 100 200 300 400x10-6 [TPZ], M

Figure 7.9. The plot of the observed rate constant of the growth and decay of the transient absorptions at 475 and 545 nm in Figure 7.7 versus [TPZ].

150

0.5 Benzophenone in ethanol, in the presence of 70 uM TPZ, upon 355 nm LFP 0.4 right after 100 ns 0.3 5 µs 45 µs after laser 0.2 O.D. O.D. ∆ 0.1

0.0

-0.1 350 400 450 500 550nm wavelength

Figure 7.10. The transient spectra of produced by 355 nm LFP of 7 mM benzophenone in ethanol, in the presence of 0.07 mM TPZ.

8 -1 -1 3 550 nm: 3.2+/-0.4x10 s M 200x10 8 -1 -1 475 nm: 6.4+/-0.4x10 s M

150 -1 , s

obs 100 k

-6 0 50 100 150 200 250 300 350x10 [TPZ], M

Figure 7.11. The plot of the observed rate constant of transient absorption measured at 475 and 550 nm in Figure 7.10 versus [TPZ].

151

9 mM benzophenone in benzene in the presence of 0.17 M benzhydrol upon 355 nm LFP right after laser 0.6 100 ns 1000 ns 5000 ns after laser pulse 0.4 O.D. ∆

0.2

0.0 350 400 450 500 550nm wavelength

Figure 7.12. The transient spectra of produced upon 355 nm LFP of 9 mM benzophenone in benzene, in the presence of 0.17 M benzhydrol.

0.6 9 mM benzophenone + 0.17 M benzhydrol in benzene upon 355 nm LFP, in the presence of ~100 uM TPZ right after 1 us 0.4 10 us 40 us after laser pulse O.D.

∆ 0.2

0.0

350 400 450 500 550nm wavelength

Figure 7.13. The transient spectra produced upon 355 nm LFP of 9 mM benzophenone in benzene, in the presence of 0.17 M benzhydrol and 0.1 mM TPZ.

152

Sensitization by acetone

If we use acetone as a photosensitizer and excite it at 308 nm, we will generate triplet acetone. Triplet acetone can abstract the tertiary-hydrogen from 2-propanol to form a pair of identical acetone ketyl radicals. The rate constant of the abstraction process is measured to be about 1.0×106 M-1s-1.120 Under these conditions, the quantum yield of the Norrish Type I cleavage of the triplet acetone can be considered to be a minor process.121 This radical can react further with TPZ.

OH O 3 308 nm O* OH H 2 H CCHISC 3 3 H3C CH3

We flashed a solution of 7.5% acetone in 2-propanol in the presence of 0.1 mM

TPZ. A bleaching at 475 nm was observed, as well as the growth of a new band at 350 nm and a less intense band at 405 nm were formed (Figure 7.14). In this case, there is no interference from the benzophenone ketyl radical, which absorbs strongly at 545 nm and

320 nm.

A plot of the decay rate constant measured at 475 nm and the growth rate constants at 405 nm versus [TPZ] yields the two bimolecular rate constants of 1.1×109 M-

1s-1 and 1.0×109 M-1s-1, respectively (Figure 7.15).

153

0.15 7.5% acetone in 2-propanol upon 308 nm LFP in the presense of 100µM TPZ 0.10 right after 350 1µs 7µs 0.05 405 30 µs after laser pulse

0.00 O.D. ∆ -0.05

-0.10

475 -0.15 350 400 450 500 550nm wavelength, nm

Figure 7.14. The transient spectra produced upon 308 nm LFP of 7.5% acetone in 2- propanol, in the presence of 0.1 mM TPZ.

6 1.0x10 9 -1 decay at 475 nm, k=1.1+/-0.1x10 s 9 -1 0.8 growth at 405 nm, k=1.0+/-0.1x10 s -1

s 0.6 -1 M

obs, 0.4 k

0.2

0.0

-6 0 200 400 600x10 [TPZ], M

Figure 7.15. The plot of the observed rate constant of the growth at 475 and 405 nm in Figure 7.14 versus [TPZ].

154

Generation of acetone ketyl radical from t-butyl peroxide

t-Butyl peroxide was excited at 308 nm to generate the t-butoxy radical. This radical can abstract the tertiary-hydrogen atom from 2-propanol.

OH OH OH O 308 nm O O 2+H

The transient spectra in Figure 7.16 exhibit very similar features to those shown in the spectra of Figure 7.14. The absorptions at 350 nm and 405 nm have same kinetics, implying they are related to the same chemical process. The starting point for 350 nm has a smaller transient absorption than that at 405 nm because TPZ has greater absorption at 350 nm than at 405 nm. Therefore, the signal at 350 nm is a combination of the bleaching of TPZ and the growth of transient absorption of a new species in this region.

The bands at 350 and 405 nm are assigned to the TPZ-H radical (Scheme 7.7). This radical was generated by photoinduced reduction of TPZ in the presence of electron and proton donor in Chapter 5 and has known absorptions at around 450 nm (Figure 5.5).

The observed rate constants at 350, 405 and 475/480 nm measured in two different experiments, ~1.0×109 M-1s-1, are very similar, as shown in Figure 7.22 and Figure 7.23.

At longer time scales, as shown in Figure 7.21, the absorption band at 350 nm eventually decays, while the 405 nm band remains as intense as before. The bleaching at

475 nm actually recovers in part. This process with a lifetime of hundreds of microseconds is attributed to the disproportionation of two TPZ-H radicals to generate one molecule of TPZ and one molecule of dTPZ (Scheme 7.7). The observed constant

155 absorption at 405 nm is due to dTPZ, which has absorption bands at 405 nm, which overlaps with the absorption of the TPZ-H radical. This observation is consistent with

Denny48 and Wardman’s74 pulse radiolysis studies, in which they generated TPZ-H and

8-Me-TPZ-H radical by reduction of the N-oxides in low pH solution (Figure 7.20).

Liu’s DFT calculations predict that the TPZ-H radical fragments in the gas phase with a τ of about 500 ps and 40 µs in aqueous solutions. Therefore it is likely that the transient that absorbs at 350 nm is the TPZ-H radical.

R O N N

N NH2 O TPZ

hv, +e-, +H+

R O OH R O R O R O N N N N N N N + N + H2O N NH N NH 2 N NH2 2 N NH2 O OH O

TPZ TPZ-H TPZ dTPZ

R = H, this work O OH = Me, Denny's work N N N

N NH2 N NH2 OH O

TPZ-H4 TPZ-H1

Scheme 7.7. Generation and fate of TPZ-H radical by reduction and hydrogen atom exchange.

156

0.1

0.0

-0.1 17% t-butyl peroxide upon 308 nm LFP in the presence of 0.2 mM TPZ, 20 ns window right after -0.2 1µs Transient Absorbance 5µs 15µs after laser pulse -0.3 350 400 450 500nm wavelength

100 17% t-butyl peroxide in 2-propanol upon 308 nm LFP in the presence of 0.2 mM TPZ 50 100 µs after laser pulse

Transient Absorbance -50

-3 -100x10 500 550 600 650 700nm wavelength

Figure 7.16. The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in 2-propanol, in the presence of 0.2 mM TPZ. Top: 320 nm to 530 nm; bottom: 480 nm to 720 nm.

157

0.1

0.0

-0.1 17% t-butyl peroxide upon 308 nm LFP in the presence of 0.2 mM TPZ, -0.2 15 µs after laser pules, 20 ns window Transient Absorbance 100 µs after laser pulse, 20 ns window 1ms after laser pulse, 20 ns window -0.3 350 400 450 500nm wavelength

Figure 7.17. The transient spectra produced by 355 nm LFPof 17% t-butyl peroxide in 2-propanol, in the presence of 0.2 mM TPZ, at longer time scales.

158

9 -1 405 nm growth, 1.0+/- 0.1x10 s 3 8 -1 600x10 475 nm decay, 9.4+/- 0.5x10 s -1 s

-1 400 , M obs k 200

-6 0 200 400 600x10 [TPZ], M

Figure 7.18. Plots of the observed rate constants of the growth of transient absorption at 475 and 405 nm in Figure 7.20 versus [TPZ].

3 8 -1 -1 350x10 growth at 350 nm, slope = 8.9 +/- 0.7x10 M s 9 -1 -1 decay at 480 nm, slope = 1.1 +/- 0.1x10 M s 300

250 )

-1 200 (s

obs 150 k 100

-6 0 50 100 150 200 250 300x10 [TPZ], M

Figure 7.19. Plots of the observed rate constants of the growth of the transient absorption at 480 and 350 nm in Figure 7.20 versus [TPZ].

159

Figure 7.20. The transient UV-vis absorptions of TPZ radical anion (solid circle) and TPZ-H radical (circle), obtained by pulse radiolysis adapted from Reference 48.

-3 70x10 Calculated UV-Vis absorptions of TPZ-H4 radical 60 TPZ-H1 radical 50 BTA radical

30 Absorbance 20

0 300 400 500 600 700 800 Wavelength, nm

Figure 7.21. The UV-vis absorptions of TPZ-H1, TPZ-H4 and BTA radicals in the gas phase predicted by TD-DFT calculation.

160

Another possibility is that hydrogen atom exchange reaction proceeds at the O1 position to generate TPZ-H1. Protonation of the TPZ radical anion at O1 position will also generate the TPZ-H1 radical. Liu et al’s DFT calculation predict that the enthalpy of

TPZ-H1 is 6.3 kcal/mol higher than TPZ-H4 isomer. Our TD-DFT calculation for the

UV-vis absorptions of TPZ-H1 and TPZ-H4 shows that although both isomers have absorptions at around 350 nm, theTPZ-H1 radical has its most intense absorption about

540 nm (Figure 7.21). Since we did not observe such absorptions at this region (as in

Figure 7.14), we consider TPZ-H4 as the main product of the reaction of acetone ketyl radical with TPZ. This assignment is consistent with Liu’s energetic prediction.

There exists another possibility, however, that the TPZ-H4 radical will undergo extremely fast hemolytic N-O cleavage to form dTPZ and hydroxyl radical. Before this hydroxyl radical dissociate from the dTPZ molecule, it will be captured by the adjacent amino group to generate BTA radical and a molecule of water. This theory disagrees with Denny et al’s observation of slow (with a rate constant of 102 to 103 s-1) unimolecular reaction of TPZ-H radical to eliminate a water molecule. The UV-vis absorption of BTA radical is also calculated and shown in Figure 7.21. TD-DFT predicts

BTA radical absorbs at 360 to 370 nm and around 620 nm, although Denny’s radiolysis results show that BTA radical absorbs at 430 nm. We did not observe transient absorptions between 600 and 700 nm (Figure 7.16, bottom).

161

Generation of an ether radical from t-butyl peroxide and dioxane

We attempted to generate an α-alkoxy carbon radical by photolysis of t-butyl peroxide in dioxane. In the presence of TPZ, however, no transient absorption or bleaching of the UV-vis spectra is observed, even after a few hundreds of microseconds.

This observation suggests that the α-dioxanyl radical does not react with TPZ at a fast rate (Figure 7.22). This radical can not transfer a hydrogen atom to TPZ. The only reaction of this radical is to add to an oxygen or nitrogen atom of TPZ, or to abstract a hydrogen atom from the amino group. As a result the dioxanyl radical is less reactive with TPZ than an α-hydroxy substituted radical.

It should be noted that the biomolecular rate constant of the hydrogen atom abstraction from dioxane by t-butoxy radical was previously found to be 1.5×106 M-1s-1 by Malatesta and Scaiano.117 A similar rate constant for the abstraction from 2-propanol was found to be 1.8×106 M-1s-1 by the same group. It can be estimated from the molarity of pure dioxane (~12 M) that the abstraction rate of hydrogen atom is 1.8×107 s-1, corresponding to a lifetime of tert-butoxyl radical in 2-propanol of about 50 ns.

Therefore, the absence of transient absorptions of the t-butyl peroxide and dioxane with

TPZ is due to the lack of reactivity of the dioxanyl radical with TPZ.

162

0.3 308 nm LFP of 17% t-butyl peroxide in dioxane in the presence of 0.15 mM TPZ 90 µs after laser pulse 0.2 in 2-propanol, 15µs after lase pulse

0.1

0.0

-0.1 Transient Absorbance

-0.2

300 350 400 450 500nm wavelength

Figure 7.22. The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in dioxane, in the presence of 0.15 mM TPZ.

163 dTPZ

The reactivity of 4-desoxytirapazamine (dTPZ) with acetone ketyl radical generated by reaction of t-butyl butoxyl radical with 2-propanol was also studied (Figure

7.23). It can be seen that there is a bleaching at 410 nm and a growth of transient absorbance at 345 nm. The kinetic traces at the two wavelengths give approximately same rate constants, suggesting they are related processes (Figure 7.24). We assign the absorption at 345 nm to dTPZ-H radical. This radical was generated by a different method, by reduction of dTPZ in acid solution in Chapter 5. Both the TD-DFT calculations and the transient spectra by the reduction method shows that dTPZ-H radical has an absorption band around 340 nm (Figure 5.12).

O N N

N NH2

hv, +e-, +H+

O OH OH O N N N N N N ? N N ++H2O N NH2 N NH2 N NH2 N NH2

dTPZ dTPZ-H dTPZ ddTPZ

Scheme 7.8. Generation of and possible fate of dTPZ-H radical by hydrogen atom exchange and reduction.

164

Due to the limited solubility of dTPZ in 2-propanol, we were unable to carry out an accurate linear plot of the change of kobs at 345 nm and 410 nm by varying the concentration of dTPZ. However, a three point linear fitting gives a rate constant of about 3-6 ×108 s-1M-1. This rate constant is about half of that of TPZ. This lower rate constant may contribute to the inactivity of dTPZ towards tumor cells.

At longer time scales, a partial recovery of the absorption at 410 nm is observed.

This may be due to a disproportionation reaction similar to TPZ-H radical, that two dTPZ-H radicals disproportionate to generate dTPZ, 3-amino-benzotriazine 1-N-oxide

(ddTPZ) and a water molecule. This observation may also imply that the N-O homolytic cleavage in dTPZ-H radical is not an efficient process.

165

345 0.05

0.00

-0.05 410

-0.10 17% t-butyl peroxide in 2-propanol upon 308 nm LFP

Transient Absorbance in the presence of 0.36 mM dTPZ 50 µs after laser pulse -0.15

300 350 400 450 500 nm

Figure 7.23. The transient spectra produced by 355 nm LFP of 17% t-butyl peroxide in 2-propanol, in the presence of 0.36 mM dTPZ.

5 -1 -3 350 nm growth, 3.7x10 s 5 -1 60x10 410 nm decay, 3.4x10 s 40

-20 Transient Absorbance -40

5 10 15 20 µs

Figure 7.24. Kinetic traces recorded at 350 and 410 nm of the transient species whose spectra are given in Figure 7.25.

166

0.10

0.08

0.06

0.04 350 nm 0.02 410 nm

0.00 Transient Absorbance Transient -0.02

50 100 150 200 -6 x10

Figure 7.25. Kinetic traces recorded at 350 and 410 nm of the transient species whose spectra are given in Figure 7.25, at longer time scales.

Molar Extinction Coefficients of TPZ, dTPZ, dTPZ' and ddTPZ in acetonitrile 25000 TPZ 20000 dTPZ dTPZ' 15000 ddTPZ 10000 (1/M)(1/cm)

5000

0 200 300 400 500 600 Wavelength, nm

Figure 7.26. Comparison of UV-vis absorptions of ground states of TPZ, dTPZ, dTPZ’ and ddTPZ in acetonitrile

167

7.3.2 Thermal reactions of α-hydroxy(alkoxy) radicals with TPZ

Azo-tert-butane/dioxane as radical precursor

We refluxed azo-tert-butane in dioxane in the presence of TPZ. HPLC studies of the reaction products show the main product (≥60%) is dTPZ’ (1-desoxytirapazamine), together with smaller percentages of dTPZ’(4-desoxytirapazamine) and ddTPZ (3-amino-

1,2,4-benzotriazine), as shown in Scheme 7.9. This is actually a very effective synthetic method to make dTPZ’, and gives a much higher yield than those methods reported by

Gates et al.89

O O N NN N N N N N N N + + dioxane, argon, ∆ N NH N NH2 N NH2 N NH2 2 O O TPZ ddTPZ dTPZ dTPZ'

R-N=N-R 1.0 : 1.6 : 6.6

89 Enzyamatic reduction 1.0 : 6.5 : 1.1

Scheme 7.9. The deoxygenation of TPZ refluxed with azo-t-butane and dioxane.

Acetonitrile and benzene were also used as solvent and refluxed with azo-tert- butane and TPZ. However, no deoxygenation of TPZ was observed for these two solvents. Other aromatic N-oxides were also refluxed with azo-tert-butane in dioxane.

Both dTPZ and dTPZ’ can be deoxygenated to generate the parent heteroaromatic by this method, but at a slower rate. Neither pyridine N-oxide nor isoquinoline N-oxide can be deoxygenated by refluxing with azo-t-butane in dioxane.

168

However, even in the absence of azo-tert-butane, refluxing just dioxane itself can also deoxygenate TPZ but with a slower rate. The rate of deoxygenation depends on the

“purity” of the dioxane, the way it is distilled and how recently it has been purified.

There is a trend that purer dioxane has a slower rate of deoxygenating TPZ. It was also found that butylated hydroxyl toluene can partially prohibit the deoxygenation of azo-t- butyl in dioxane.

Benzoyl Peroxide in isopropanol and isopropyl ether

We used an alternative method to generate α-hydroxy and alkoxy radicals by refluxing benzoyl peroxide in 2-propanol or diisopropyl either. The products were analyzed by HPLC.

O O ∆ 2 O + 2CO2 O

OH OH TPZ ?

O TPZ O ?

The results are listed in Table 7.1.

169

Benzoyl Reflux Product distribution Solvent Peroxide time TPZ dTPZ’ dTPZ ddTPZ 3 mM 2-propanol < 0.3 hrs ~ 0 % ~ 0 % ~ 50 % ~ 50 % 0 2-propanol 12 hrs ~ 80 % < 10 % < 10 % < 10 % 3 mM diisopropyl ether 4 hrs ~ 0 % ~ 80% < 10 % < 10 % 0 diisopropyl ether 24 hrs ~ 60 % ~ 30 % < 10 % < 10 %

Table 7.1. Product distribution of TPZ refluxed with benzoyl peroxide in different solvents.

We found that refluxing benzoyl peroxide/2-propanol deoxygenates TPZ with a very fast rate (~10 minutes). The products are mainly dTPZ, where the O4 atom was lost, and ddTPZ, where both the O1 and O4 atoms were lost. The fact that there was no dTPZ’ in the product mixture suggests that the reaction of acetone ketyl radical with TPZ heavily favors the 4-position of TPZ.

Another interesting discovery is that in diisopropyl ether, the deoxygenation reaction proceeds, but at a much slower rate (completed in 4 hours). However, the main product is dTPZ’, a different product than that formed in the benzoyl peroxide/2-propanol refluxing system. This indicates that the two radicals react by different mechanism. The acetone ketyl radical reacts by hydrogen atom transfer and the diisopropyl either derived radical must react by addition or hydrogen atom abstraction.

In the absence of benzoyl peroxide, 2-propanol cannot deoxygenate TPZ, although refluxing diisopropyl ether was able to generate a small amount of dTPZ’ at a slow rate, probably due to the presence of impurities.

170

7.4 Discussion

7.4.1 Thermal generation of alkyl radicals and their reactivities

To explain these observations, we first need to identify what reactive species are formed after the thermal decomposition of azo-tert-butane. We performed some computational studies to investigate the fate of the t-butyl radical generated from the thermal degradation of the azo compound in different solvents.

∆H in the gas phase N -5.2 kcal/mol

N CH3CN + 3.0 kcal/mol

H -0.6 kcal/mol CH CN + 2

Scheme 7.10. Calculated thermodynamics of possible reactions between t-butyl radical and acetonitrile.

As shown in Scheme 7.10, t-butyl radical can abstract hydrogen atoms from acetonitrile in a thermodynamically neutral process. However, the addition of the t-butyl radical to the nitrogen atom of the cyano group is 4.6 kcal/mol more exothermic than hydrogen abstraction in the gas phase. It was reported before that the ratio of the addition and abstraction of adamantyl radical to the carbon atom of acetonitrile was 1:4.119 The fate of the addition product, (CH3)3C-N=C-CH3 is unknown, but this species is likely to be less reactive than the parent t-butyl radical. Although the distribution of the addition

171 and abstraction products in this system remains unclear, the alkyl radical that

• • deoxygenates TPZ is likely to be a CH2CN or C(CH3)3 radical. However, the experimental results imply that neither radical reacts with TPZ efficiently at the oxygen atom position.

When azo-t-butane is refluxed in benzene, it is unlikely that the bulky butyl radical will add to the phenyl ring, neither can the butyl radical be stabilized by the aromatic ring in the manner of a benzene-halogen atom complex (our DFT calculation shows no energetic minima can be found when approaching butyl radical to benzene).

It has been proposed that alkyl radicals will add to the oxygen position of aromatic N-oxides to form an adduct (although other positions are also possible), which quickly undergoes homolytic cleavage of the N-O bond to give the heterocyle and alkoxy radical. Liu and Platz performed systematic computational studies of this type of

• reaction, using methyl and CH3OCH2 radicals as models for the ribose radical. Liu calculated a series of the ∆H’s and ∆H≠’s of radical addition reactions with a variety of aromatic N-oxides from pyridine N-oxide to TPZ. The results show that the initial addition step has ∆H≠’s of 8-13 kcal/mol. We calculated the addition of t-butyl radical to

• pyridine N-oxide and TPZ. The results exhibit trends similar to those of the CH3OCH2 radical (Scheme 7.3).

172

O O O N N O + N N +

9.0 kcal/mol -6.3 kcal/mol -27.2 kcal/mol

O O O N N N O N N N + + N NH N NH2 2 N NH2 O O

-16.6 kcal/mol -31.9 kcal/mol

Scheme 7.11. Calculated reaction barriers and enthalpy changes for the addition of t- butyl radical to pyridine N-oxide and TPZ.

Theory predicts that the addition of alkyl radicals to the oxygen atom of TPZ has a moderate transition state barrier. The reactions are thermodynamically favored.

However, the addition of simple alkyl radicals to TPZ actually do not proceed efficiently, based on our experimental observations. It also seems that dioxane solvent does possess a special property that is able to facilitate the deoxygenation process initiated by azo compounds but can also deoxygenate TPZ in the absence of the initiator. We consider that dioxane, as an ether, can be oxidized to form peroxide upon storage unless perfectly protected from atmospheric oxygen. It is likely that the following species will be present in refluxing dioxane (Scheme 7.12):

173

O OH H + + -1.3 kcal/mol O O

O O OO 2 OH

O O

OO OO OH OOH OH

O O O O

Scheme 7.12. Possible radical species present in refluxing dioxane.

Among these species, an α-hydroxy radical species that may play an important role in the deoxygenation process.

It is worth noting that other ethers, such as THF and diethyl ether do not deoxygenate TPZ effectively upon reflux. The deoxygenation of azo-t-butyl in THF proceeds but is much slower that in dioxane.

Denny et al generated a deoxyribose radical from hydroxyl radical generated by pulse radiolysis and investigated the reaction of the ribose radical with TPZ with both optical and conductivity detection of reactive intermediates. They claim that they observed two phases of the reaction, the first being the formation of the adduct and the second being the breakdown of this adduct. When discussing the structure of the adduct of desoxyribose radical adding to TPZ, they proposed the N4-ribose structure as shown in

Scheme 7.5 and 7.6. The structure seems unlikely because the radical and charge in this structure do not have extra stabilization through effective resonance, and the N4 position

174 is sterically shielded. Liu’s DFT calculation has shown that the addition of the methoxymethyl radical to the N4 position is endothermic by 1.3 kcal/mol.

Another possibility, first suggested by Daniels and Gates is that the DNA centered

≠ radical adds to the O4 atom position. Thermodynamically, the ∆H of this process is calculated to be as high as 10-15 kcal/mol, while it has a favorable ∆H of -26 kcal/mol

(Scheme 7.5). Liu’s computational studies suggest that this kind of N-alkoxy radical species will be short-lived because homolytic cleavage of the N-O bond has a very low transition state barrier in the gas phase, although in aqueous solutions this transition barrier is calculated to be higher. Gould et al reported very short (ps) lifetimes of this kind of radical.118 Therefore, Denny et al’s claim that the adduct has a lifetime of tens of microseconds may be problematic.

A simple reaction between the α-hydroxyl radical and TPZ can also be a direct hydrogen atom exchange. A “jump” of the hydrogen atom from the hydroxyl group of the radical to the oxygen of TPZ probably does not have a high transition state barrier, because hydrogen atom is both very small and light. This kind of direct hydrogen exchange also avoids the formation of the intermediate adduct, which may not be very facile because of steric effect. However, both the concerted rearrangement on the O4- adduct and a direct hydrogen atom transfer end up with the same products, the TPZ-H radical and acetone. The enthalpy change was calculated to be –20.3 kcal/mol for both processes. The direct oxidation of the acetone ketyl radical by TPZ to form TPZ radical

+ anion and C(OH)Me2 cation is very endothermic in the gas phase, although this number will be a lot less positive in aqueous solution (Scheme 7.13) .

175

∆H in the gas O phase N OH N + 107.3 kcal/mol N NH2 O O O N OH N N O + N + -20.3 kcal/mol N NH2 O N NH2 OH O O N N O N N + -21. 5 cal/mol OH N NH2 N NH2 O OH -23. 6 cal/mol

O N N + -2.0 kcal/mol O N NH2 N N OH + O O N NH2 N N O N O N + -31.9 kcal/mol

N NH2 N NH2 O -16.6 kcal/mol

Scheme 7.13. Calculated reaction enthalpies of TPZ with different radical species.

176

7.4.2 Reactions of α-hydroxy radicals with TPZ and dTPZ

The rate constants for the photochemically generated α-hydroxyl radicals reacting with TPZ were summarized in Table 7.2.

Reactive radicals 550/545 nm 475 nm 410/405 nm 350 nm

OH + OH 3.2±0.4×108 6.4±0.4×108 NA

OH 8 8 + O H 4.0±0.2×10 4.5±0.4×10 NA O

OH 8 + OH 5.2±0.2×10 NA

OH 10.1±1.0×108 10.0±1.0×108 8.9±1.0×108 (from t-butoxy radical)

9.4±0.5×108 10.0±1.0×108

OH 11.0±1.0×108 10.0±1.0×108 (from acetone sensitization)

Table 7.2. Reaction rate constants of benzophenone ketyl and/or acetone ketyl radical with TPZ.

We previously determined that α-dioxanyl radical does not react with TPZ, therefore, the rate constant of benzophenone ketyl radical with TPZ is about 4.5×108 M-1 s-1. The acetone ketyl radical has a rate constant with TPZ of about 1×109 s-1M-1. This difference probably reflects the different accessibility of benzophenone ketyl and acetone ketyl radical to TPZ.

177

By combining our thermal and photochemical studies, it is very clear that α- hydroxy radical is able to efficiently react with TPZ through direct hydrogen atom transfer. No addition of either hydroxy or alkoxy radical to TPZ was observed.

Therefore, a nucleoside carbon radical will not be efficiently oxidized by TPZ through an addition mechanism. However the hydrogen atom transfer mechanism is not available for nucleosides and natural DNA radicals.

7.5 Conclusions

α-Alkoxy and hydroxy radicals were generated through thermal and photochemical reactions. Both the product analysis of the thermal reaction and the direct observation of the transient species involved suggest that α-hydroxy radical can very efficiently react with TPZ to form the TPZ-H radical, probably through a direct hydrogen atom exchange between TPZ and ketyl radicals. α-alkoxy radicals can not proceed through this mechanism.

178

References

(1) Szacilowski, K.; Macyk, W.; Drzewiecka-Matuszek, A.; Brindell, M.; Stochel, G. Chem. Rev. 2005, 105, 2647-2694.

(2) Ochsner, M. J. Photochem. Photobiol. B. 1997, 39, 1-18.

(3) Oleinick, N. L.; Evans, H. H. Radiat. Res. 1998, 150, S146-56.

(4) Korbelik, M. J. Clin. Laser Med. Surg. 1996, 14, 329.

(5) Ali, H.; van Lier, J. E. Chem. Rev. 1999, 99, 2379.

(6) Bonnett, R. In McCleverty, J. A., Meyer, T. J., Eds.; Comprehensive Coordination Chemistry II; Elsevier: Amsterdam, 2003; .

(7) Lang, K.; Mosinger, J.; Wagnerova, D. M. Coord. Chem. Rev. 2004, 248, 321.

(8) Jori, G. J. Photochem. Photobiol. B. 1996, 36, 87-93.

(9) Edelson, R. L. Yale J. Biol. Med. 1989, 62, 565-577.

(10) Wolkenberg, S. E.; Boger, D. L. Chem. Rev. 2002, 102, 2477-2495.

(11) Sharman, W. M.; Allen, C. M.; van Lier, J. E. Methods Enzymol. 2000, 319, 376-400.

(12) Kochevar, I. E.; Redmond, R. W. Methods Enzymol. 2000, 319, 20-28.

(13) Ahmad, N.; Mukhtar, H. Methods Enzymol. 2000, 319, 342-358.

(14) Rosenthal, I.; Ben-Hur, E. Int. J. Radiat. Biol. 1995, 67, 85-91.

(15) Kanne, D.; Straub, K.; Rapoport, H.; Hearst, J. E. Biochemistry 1982, 21, 861-871.

(16) Peckler, S.; Graves, B.; Kanne, D.; Rapoport, H.; Hearst, J. E.; Kim, S. H. J. Mol. Biol. 1982, 162, 157-172.

(17) Muller-Breitkreutz, K.; Mohr, H.; Briviba, K.; Sies, H. J. Photochem. Photobiol. B. 1995, 30, 63-70.

(18) Lown, J. W.; Chen, H. H.; Plambeck, J. A. Chem. Biol. Interact. 1981, 35, 55-70.

(19) Pan, S. S.; Andrews, P. A.; Glover, C. J.; Bachur, N. R. J. Biol. Chem. 1984, 259, 959-966.

179

(20) Dusre, L.; Rajagopalan, S.; Eliot, H. M.; Covey, J. M.; Sinha, B. K. Cancer Res. 1990, 50, 648-652.

(21) Doroshow, J. H. Proc. Natl. Acad. Sci. U. S. A. 1986, 83, 4514-4518.

(22) Brown, J. M. Br. J. Cancer 1993, 67, 1163-1170.

(23) Goodrich, R. P. Vox Sang. 2000, 78 Suppl 2, 211-215.

(24) Dalle Carbonare, M.; Pathak, M. A. J. Photochem. Photobiol. B. 1992, 14, 105-124.

(25) Kale, H.; Harikumar, P.; Kulkarni, S. B.; Nair, P. M.; Netrawali, M. S. Mutat. Res. 1992, 298, 17-23.

(26) Schwarz, H. A. J. Chem. Edu. 1981, 58, 101.

(27) Vaughan, P. P.; Blough, N. V. Environ. Sci. Technol. 1998, 32, 2947.

(28) Cadet, J.; Berger, M.; Douki, T.; Ravanat, J. L. Rev. Physiol. Biochem. Pharmacol. 1997, 131, 1-87.

(29) Cadet, J.; Bellon, S.; Berger, M.; Bourdat, A. G.; Douki, T.; Duarte, V.; Frelon, S.; Gasparutto, D.; Muller, E.; Ravanat, J. L.; Sauvaigo, S. Biol. Chem. 2002, 383, 933- 943.

(30) Ames, B. N.; Shigenaga, M. K.; Hagen, T. M. Proc. Natl. Acad. Sci. U. S. A. 1993, 90, 7915-7922.

(31) Burrows, C. J.; Muller, J. G. Chem. Rev. 1998, 98, 1109-1152.

(32) Marnett, L. J. Carcinogenesis 2000, 21, 361-370.

(33) Villamena, F. A.; Hadad, C. M.; Zweier, J. L. J. Am. Chem. Soc. 2004, 126, 1816- 1829.

(34) Smith, E. C.; Metzler, D. E. J. Am. Chem. Soc. 1963, 85, 3285.

(35) Ahmad, I.; Rapson, D. C.; Heelis, P. F.; Phillips, G. D. J. Org. Chem. 1980, 45, 731.

(36) McCormick, D. B. Nutr. Rev. 1972, 30, 75-79.

(37) Rivlin, R. S. N. Engl. J. Med. 1970, 283, 463-472.

(38) Zempleni, J.; Galloway, J. R.; McCormick, D. B. Am. J. Clin. Nutr. 1996, 63, 54-66.

(39) Goodrich, R. P. In In Cambridge Healthtech Institute's Sixth Annual Blood Product Safety Conference; 2000; .

(40) Ennever, J. F.; Carr, H. S.; Speck, W. T. Pediatr. Res. 1983, 17, 192-194. 180

(41) Ono, S.; Oikawa, K.; Hirano, H.; Obara, Y. Int. J. Vitam. Nutr. Res. 1986, 56, 259- 262.

(42) Tapia, G.; Silva, E. Radiat. Environ. Biophys. 1991, 30, 131-138.

(43) Denny, W. A. Cancer Invest. 2004, 22, 604-619.

(44) Zeman, E. M.; Brown, J. M.; Lemmon, M. J.; Hirst, V. K.; Lee, W. W. Int. J. Radiat. Oncol. Biol. Phys. 1986, 12, 1239-1242.

(45) Brown, J. M.; Wilson, W. R. Nat. Rev. Cancer. 2004, 4, 437-447.

(46) Daniels, J. S.; Gates, K. S. J. Am. Chem. Soc. 1996, 118, 3380-3385.

(47) Patterson, L. H.; Taiwo, F. A. Biochem. Pharmacol. 2000, 60, 1933-1935.

(48) Anderson, R. F.; Shinde, S. S.; Hay, M. P.; Gamage, S. A.; Denny, W. A. J. Am. Chem. Soc. 2003, 125, 748-756.

(49) Shinde, S. S.; Anderson, R. F.; Hay, M. P.; Gamage, S. A.; Denny, W. A. J. Am. Chem. Soc. 2004, 126, 7865-7874.

(50) Albini, A.; Alpegiani, M. Chem. Rev. 1984, 84, 43.

(51) Buterbaugh, J. S.; Toscano, J. P.; Weaver, W. L.; Gord, J. R.; Hadad, C. M.; Gustafson, T. L.; Platz, M. S. J. Am. Chem. Soc. 1997, 119, 3580.

(52) Tyson, D. S.; Castellano, F. N. J. Phys. Chem. A 1999, 103, 10955.

(53) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; 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.; Baboul, A. G.; Stefanov, J. V.; 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.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head- Gordon, M.; Replogle, E. S.; Pople, J. A.; Gaussian, Inc., Pittsburgh, PA, 1998. 1998, SGI64-G98, revision A.9.

(54) Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

(55) Speck, W. T.; Rosenkranz, S.; Rosenkranz, H. S. Biochim. Biophys. Acta 1976, 435, 39-44.

(56) Korycka-Dahl, M.; Richardson, T. Biochim. Biophys. Acta 1980, 610, 229-234.

181

(57) Joshi, P. C. Indian J. Biochem. Biophys. 1989, 26, 186-189.

(58) Joshi, P. C. Toxicol. Lett. 1985, 26, 211-217.

(59) Edwards, A. M.; Silva, E.; Jofre, B.; Becker, M. I.; De Ioannes, A. E. J. Photochem. Photobiol. B. 1994, 24, 179-186.

(60) Yamamoto, F.; Nishimura, S.; Kasai, H. Biochem. Biophys. Res. Commun. 1992, 187, 809-813.

(61) Hoffmann, M. E.; Meneghini, R. Photochem. Photobiol. 1979, 29, 299-303.

(62) Martin, C. B.; Tsao, M. L.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 2002, 124, 7226-7234.

(63) McCormick, D. B. Heterocycl. Chem. 1970, 7, 447.

(64) Sheu, C.; Foote, C. S. J. Am. Chem. Soc. 1995, 117, 6439.

(65) Knowles, A. Photochem. Photobiol. 1971, 13, 225-236.

(66) Wahl, P.; Auchet, J. C.; Visser, A. J.; Muller, F. FEBS Lett. 1974, 44, 67-70.

(67) Dardare, N.; Platz, M. S. Photochem. Photobiol. 2002, 75, 561-564.

(68) Tsao, M.-L., personal communication.

(69) Steenken, S. Chem. Rev. 1989, 89, 503.

(70) Steenken, S.; Jovanovic, S. V. J. Am. Chem. Soc. 1997, 119, 617.

(71) Knowles, A.; Roe, E. M. Photochem. Photobiol. 1968, 7, 421-436.

(72) Kuzmin, V. A.; Dourandin, A.; Shafrovich, V.; Geacintov, N. E. Phys. Chem. Chem. Phys. 2000, 2, 1531.

(73) Candeias, L. P.; Steenken, S. J. Am. Chem. Soc. 1989, 111, 1094.

(74) Laderoute, K.; Wardman, P.; Rauth, A. M. Biochem. Pharmacol. 1988, 37, 1487- 1495.

(75) Daniels, J. S.; Gates, K. S.; Tronche, C.; Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1254-1257.

(76) Albini, A.; Pietra, S. In Heterocyclic N-oxides; CRC Press, Inc.: 1991; .

(77) Monge, A.; Palop, J. A.; Lopez de Cerain, A.; Senador, V.; Martinez-Crespo, F. J.; Sainz, Y.; Narro, S.; Garcia, E.; de Miguel, C.; Gonzalez, M. J. Med. Chem. 1995, 38, 1786-1792. 182

(78) Naylor, M. A.; Stephens, M. A.; Nolan, J.; Sutton, B.; Tocher, J. H.; Fielden, E. M.; Adams, G. E.; Stratford, I. J. Anticancer Drug Des. 1993, 8, 439-461.

(79) Takahashi, K.; Kawazoe, Y.; Tada, M.; Tada, M.; Ito, N.; Okada, M. Gann 1978, 69, 499-505.

(80) Wardman, P. In Chemistry of Nitroarene and Aromatic N-oxide Radicals; Alfassi, Z. B., Ed.; N-Centered Radicals; John Wiley & Sons Ltd: 1998; .

(81) Poole, J. S.; Hadad, C. M.; Platz, M. S.; Fredin, Z. P.; Pickard, L.; Guerrero, E. L.; Kessler, M.; Chowdhury, G.; Kotandeniya, D.; Gates, K. S. Photochem. Photobiol. 2002, 75, 339-345.

(82) Kaneko, C.; Yamamoto, A.; Gomi, M. Heterocycles 1979, 12, 227.

(83) Ono, I.; Hata, N. Bull. Chem. Soc. Jpn. 1973, 46, 3658.

(84) Binstead, R. A.; McGuire, M. E.; Dovletoglou, A.; Seok, W. K.; Roecker, L. E.; Meyer, T. J. J. Am. Chem. Soc. 1992, 114, 173.

(85) Kasama, K.; Takematsu, A.; Yamamoto, S.; Arai, S. J. Phys. Chem. 1884, 44, 4918.

(86) Seki, H.; Takematsu, A.; Arai, S. J. Phys. Chem. 1987, 91, 176.

(87) Shida, T. In Electronic Absorption Spectra of Radical Ions; Elsevier: Amsterdam, 1988; .

(88) http://www.jobinyvon.com/usadivisions/Fluorescence/applications/quantumyieldstra d.pdf2004).

(89) Fuchs, T.; Chowdhury, G.; Barnes, C. L.; Gates, K. S. J. Org. Chem. 2001, 66, 107- 114.

(90) Kobayashi, Y.; Kumadaki, I.; Sato, H.; Sekine, Y.; Hara, T. Chem. Pharm. Bull. 1974, 22, 2097.

(91) Liu, J., personal communication.

(92) Spence, G. G.; Taylor, E. C.; Buchardt, O. Chem. Rev. 1969, 69, 231.

(93) Albini, A.; Fasoni, E.; Amer, A. In Handbook of Photochemistry; CRC Press: Boca Raton, FL, .

(94) Cimiraglia, R.; Persico, M.; Tomasi, J. J. Phys. Chem. 1977, 81, 1876.

(95) Blough, N. V.; Zepp, R. G. In Reactive Oxygen Species in Natural Waters; Foote, C. S., Valentine, J. S., Greenberg, A. and Liebman, J. F., Eds.; Active Oxygen in

183

Chemistry; Blackie Academic and Professional: New York, 1995; Vol. 2, pp 280- 333.

(96) Schuetzle, D.; Siegl, W. O.; Jensen, T. E.; Dearth, M. A.; Kaiser, E. W.; Gorse, R.; Kreucher, W.; Kulik, E. Environ. Health Persp. 1994, 102, 3.

(97) Richter, H.; Howard, J. B. Prog. Energy Combust. Sci. 2000, 26, 565.

(98) Salloum, M.; Chefetz, B.; Hatcher, P. G. Environ. Sci. Technol. 2002, 36, 1953.

(99) Scaiano, J. C. In Nanosecond laser flash photolysis: a tool for physical organic chemistry; Moss, R. A., Platz, M. S. and Jones, M., Jr., Eds.; Reactive Intermediate Chemistry; John Wiley & Sons, Inc.: Hoboken, NJ, 2004; pp 797-845.

(100) Baxendale, J. H.; Rodgers, M. A. J. Chem. Soc. Rev. 1978, 7, 235.

(101) (a) http://www.rcdc.nd.edu/compilations/Hydroxyl/OH.htm; (b) http://kinetics.nist.gov; (c) http://kinetics.nist.gov/solution.

(102) Aveline, B. M.; Kochevar, I. E.; Redmond, R. W. J. Am. Chem. Soc. 1996, 118, 10124.

(103) Aveline, B. M.; Kochevar, I. E.; Redmond, R. W. J. Am. Chem. Soc. 1996, 118, 10113.

(104) Neta, P.; Schuler, R. H. J. Phys. Chem. 1975, 79, 1.

(105) Poole, J. S., personal communication.

(106) Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B. J. Phys. Chem. Ref. Data 1988, 17, 513.

(107) Dorfman, L.; Taub, I. A.; Buehler, R. E. J. Chem. Phys. 1962, 36, 3051.

(108) Roder, M.; Wojnarovits, L.; Foldiak, G. Radiat. Phys. Chem. 1990, 36, 175.

(109) Engdahl, A.; Karlström, G.; Nelander, B. J. Chem. Phys. 2003, 118, 7797.

(110) DeMatteo, M. P., personal communication.

(111) Barckholtz, C.; Barckholtz, T. A.; Hadad, C. M. J. Phys. Chem. A 2001, 105, 140.

(112) Atkinson, R. Atmos. Environ. , Part A 1990, 24A, 1.

(113) Tsao, M. -.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 2003, 125, 8390.

(114) Fukui, K. Fortschr. Chem. Forsch. 1970, 15, 1.

(115) Fischer, H.; Radom, L. Angew Chem. Int. Ed. 2001, 40, 1340. 184

(116) Hwang, J. T.; Greenberg, M. M.; Fuchs, T.; Gates, K. S. Biochemistry 1999, 38, 14248-14255.

(117) Malatesta, V.; Scaiano, J. C. J. Org. Chem. 1982, 47, 1455.

(118) Lorance, E. D.; Kramer, W. H.; Gould, I. R. J. Am. Chem. Soc. 2002, 124, 15225.

(119) Engel, P. S.; Lee, W. -K.; Marschke, G. E.; Shine, H. J. J. Org. Chem. 1987, 52, 2813.

(120) (a) Charney, D. R.; Dalton, J. C.; Hautala, R. R.; Snyder, J. J.; Turro, N. J. J. Am. Chem. Soc. 1974, 96, 1407; (b) Porter, G.; Dogra, S. K.; Loutfy, R. O.; Sugamori, S. E.; Yip, R. W. J. Chem. Soc., Faraday Trans. 1 1973, 1462.

(121) Haas, Y., Photochem. Photobiol. Sci. 2004, 3, 6

185