Towards Targeted Photodynamic Therapy: Synthesis and Characterization of Aziridine Aldehyde-Cyclized Cancer- Targeting Peptides and Bacteriochlorin Photosensitizers

Áron Roxin

A thesis submitted in conformity with the requirements for the degree of Doctorate of Philosophy Graduate Department of Pharmaceutical Sciences University of Toronto

Áron Roxin

Doctorate of Philosophy

Graduate Department of Pharmaceutical Sciences University of Toronto

2014 Abstract

This thesis presents the contributions we made towards achieving targeted photodynamic therapy

(PDT) by synthesizing and characterizing new aziridine aldehyde-cyclized cancer-targeting peptides and bacteriochlorin photosensitizers (PSs). The new peptides are based on the integrin- targeting sequence, arginine-glycine-aspartic acid (RGD), and were cyclized by aziridine aldehyde-driven macrocyclization chemistry. We developed a versatile conjugation strategy, and created two useful RGD macrocycles that specifically bound to integrin receptors in vitro.

Computer modeling and competitive binding studies found that this macrocyclization chemistry modulated the binding affinity of these peptides by tuning the geometry of peptides of different lengths. The new PSs described herein include bacteriochlorophyll derivatives that efficiently produced reactive oxygen species (ROS) upon illumination with red light. We discovered two simple structural modifications that enhanced the photogeneration of ROS, and suggest that future studies make use of these design parameters to create next generation bacteriochlorin PSs with superior photoactivity compared to known derivatives. In addition, we optimized a facile synthetic reaction for creating a new natural product analog with a distinct exocyclic F-ring, and

ii found it was capable of PDT. Unlike some of its natural chlorin counterparts, our bacteriochlorin was not a potent antioxidant. Yet, we suggest that future efforts make use of our simple reaction to expand the library of F-ring containing bacteriochlorins to elucidate key structural modifications that can tune the PDT efficacy and antioxidant activity of this distinct class of bacteriochlorins. Finally, this thesis presents preliminary investigations that attempted to develop new cancer-targeting PDT agents. While the goal of achieving optimal cancer-targeted PDT was not accomplished in this work, it is envisioned that this thesis will provide useful data and insights that will contribute to the development of optimal aziridine aldehyde-cyclized peptide- bacteriochlorin PS conjugates for efficient in vivo cancer-specific PDT.

iii Acknowledgments

I would like to acknowledge several people who have helped me throughout my PhD experience at the University of Toronto. First and foremost, I thank my supervisor and mentor, Dr. Gang Zheng. His constant support, encouragement and guidance gave me a rich and rewarding graduate school experience. Working in Dr. Zheng‘s lab gave me the opportunity to pursue my research interests and allowed me to develop many useful skills. Thanks also to Dr. Zheng‘s senior scientist, Dr. Juan Chen. Dr. Chen taught me a lot about pharmaceutical research and always made time to discuss my experiments. I‘ll always look up to Dr. Zheng and Dr. Chen, and appreciate that they‘ve greatly helped me progress to the next level in my research career. I would also like to thank the members of my advisory committee, Dr. Andrei K. Yudin, Dr. David Jaffray and Dr. Raymond Reilly, for their valuable guidance.

There are many people in Dr. Zheng‘s lab and close group of collaborators who I would like to thank for helping me and for enriching my work experience. Dr. Lili Ding taught me many important microbiology skills. Dr. Andrew Cao greatly contributed to my development as a chemist. Also, the other graduate students and postdocs in Dr. Zheng‘s lab; Dr. Jon Lovell, Ken Ng, Cheng Jin, Dr. Ben Scott, T.D. MacDonald, Ben Luby, Sophie Wang, Arash Farhadi, Elizabeth Huynh, Sarah Cui, Dr. Jiyun Shi and Dr. Neeshma Dave were all great colleagues and I will always treasure their friendship.

I am very thankful for my loving partner, Dr. Katie Marshall. As a fellow scientist, she was always understanding and encouraging throughout my program. I am also grateful to my parents, Dr. George and Dr. Sara Roxin for their constant love and support.

As I continue to pursue my career goals, I will forever be thankful to everyone who has made my experience as a PhD student at the University of Toronto a great one.

iv Table of Contents

Abstract ...... ii

Acknowledgments ...... iv

Table of Contents ...... v

List of Tables ...... xii

List of Figures, Charts and Schemes ...... xiv

List of Abbreviations ...... xxvi

List of Appendices ...... xxx

Chapter 1 ...... 1

Introduction to Cyclic Cancer-Targeting Peptides and Bacteriochlorin Photosensitizers ...... 1

Preamble ...... 1

Section 1. Flexible or fixed: a comparative review of linear and cyclic cancer-targeting peptides ...... 2

Summary ...... 2

Introduction ...... 2

Geometry of linear and cyclic peptides ...... 3

Library screenings and structure-activity studies ...... 7

Receptor subtype specificity of linear and cyclic peptides ...... 9

Stability of linear and cyclic peptides ...... 11

Unconjugated peptides as anti-cancer therapeutics ...... 14

Cancer-targeting peptide conjugates ...... 16

Cell penetrating peptides ...... 19

Limitations of peptide cyclization ...... 20

Unique aspects of peptide cyclization ...... 23

Conclusions ...... 27

Section 2. Towards potent bacteriochlorin photosensitizers: insights into structural modifications with potential to enhance PDT-relevant photophysical properties ...... 28

v Summary ...... 28

Introduction ...... 28

Metal insertion ...... 31

The heavy atom effect ...... 36

The electron-donating meso 5-methoxy group ...... 39

Modifications along the y-axis ...... 43

Cationic substituents along the x-axis and the y-axis ...... 47

Additional fused rings ...... 49

Photosensitizers capable of type 1 PDT ...... 51

Conclusions ...... 55

Section 3. Towards Cancer-Targeted Photodynamic Therapy: Overarching Thesis Goals ..... 56

Aziridine aldehyde-cyclized integrin-targeting peptides ...... 56

Structural features for modulating the photoproperties of bacteriochlorin photosensitizers ...... 56

Expanding the structural library of bacteriochlorins for photodynamic therapy ...... 57

Towards achieving cancer-targeted photodynamic therapy using aziridine aldehyde- cyclized peptides and bacteriochlorin photosensitizers ...... 57

Chapter 2 Conformational Modulation of In Vitro Activity of Cyclic RGD Peptides via Aziridine Aldehyde-Driven Macrocyclization Chemistry ...... 59

Preamble ...... 59

Introduction ...... 59

Results ...... 62

Aziridine ring-opening and fluorophore/ chelator conjugation ...... 62

Computer modeling ...... 63

Binding affinity ...... 66

Confocal microscopy ...... 67

Discussion ...... 68

Conclusions ...... 71

vi Materials and Methods ...... 72

Chemicals ...... 72

General methods ...... 72

Synthesis of unsubstituted aziridine-aldehyde (18) ...... 73

Manual linear peptide synthesis ...... 74

Aziridine-driven cyclization ...... 74

Aziridine ring-opening with cysteamine ...... 74

Amide cyclization ...... 75

Fluorescein conjugation ...... 75

DOTA conjugation ...... 76

Pyropheophorbide a conjugation ...... 76

Computer modeling ...... 76

Binding affinity ...... 77

Confocal microscopy ...... 77

Acknowledgements ...... 78

Supporting Information ...... 79

HPLC-MS characterization of the side chain-protected aziridine aldehyde-cyclized peptides cPRGDA (7), cPRGDAA (8) and cPRDGA (9) bonded to cysteamine by sulfhydryl aziridine ring-opening ...... 79

HPLC-MS characterization of the side chain-protected amide-cPRGDK (21) peptide after amide cyclization ...... 82

HPLC-MS characterization of the side chain-deprotected aziridine aldehyde-cyclized peptides cPRGDA (13), cPRGDAA (14) and cPRDGA (15) bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to fluorescein ...... 83

HPLC-MS characterization of the side chain-deprotected amide-cPRGDK (23) peptide conjugated to fluorescein ...... 86

HPLC-MS characterization of the side chain-protected aziridine aldehyde-cyclized peptide cPRGDA bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to DOTA (24) or pyro (25) ...... 87

Chapter 3 Modulation of reactive oxygen species photogeneration of bacteriopheophorbide a derivatives by exocyclic E-ring-opening and charge modifications ...... 89 vii Preamble ...... 89

Introduction ...... 89

Results ...... 93

Synthesis of bacteriochlorins ...... 93

Spectroscopic properties of bacteriochlorins ...... 96

Molecular modeling of potential PDT activity ...... 100

ROS photogeneration detected with AUR...... 101

ROS photogeneration detected with SOSG ...... 107

Discussion ...... 112

Conclusions ...... 115

Materials and Methods ...... 115

Materials ...... 115

General chemistry information ...... 116

Palladium bacteriopheophorbide a (1a) ...... 117

Palladium 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (1b) ...... 117

31-Oxo-rhodobacteriochlorin 173-(2-trimethylaminoethyl)ester (2a) ...... 118

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin-131-(2-sulfoethyl)amide- 173-(2-trimethylaminoethyl)ester (2b) ...... 119

31-Oxo-rhodobacteriochlorin 173-methyl ester (3a) ...... 119

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin-13'-(2-sulfoethyl)amide- 173-methyl ester (3b) ...... 120

Bacteriopheophorbide a (4a) ...... 121

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) 121

31-Oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide (5a) ...... 122

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131,173-di(2- sulfoethyl)amide (5b) ...... 122

ROS photogeneration detected with AUR and SOSG ...... 123

viii Computational studies ...... 124

Acknowledgments ...... 124

Supporting Information ...... 125

31-Oxo-rhodobacteriochlorin 173-(2-trimethylaminoethyl)ester (2a) ...... 125

31-Oxo-rhodobacteriochlorin 173-methyl ester (3a) ...... 129

Bacteriopheophorbide a (4a) ...... 133

31-Oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide (5a) ...... 137

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131-(2-sulfoethyl)amide- 173-(2-trimethylaminoethyl)ester (2b) ...... 141

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2-sulfoethyl)amide- 173-methyl ester (3b) ...... 145

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) 149

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131,173-di(2- sulfoethyl)amide (5b) ...... 153

Palladium bacteriopheophorbide a (1a) ...... 157

Palladium 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (1b) ...... 158

Computational studies ...... 159

ROS-specificity of AUR and SOSG using bacteriopheophorbide a (4a) ...... 161

In vitro biocompatibility of detergent with A549 cancer cells ...... 162

Solubility of 1a-5a and 1b-5b in PBS containing 2% DMSO and 0.5% cremophor (v/v) 164

Chapter 4 Synthesis and characterization of a new natural product analogue, 132-173- bacteriochlorophyllone a ...... 166

Preamble ...... 166

Introduction ...... 166

Results ...... 169

One-step synthesis of 12 from bacteriopheophorbide a (16) ...... 169

ix Investigation of reaction conditions for the one-step synthesis of 12 from bacteriopheophorbide a (16) ...... 169

Structure elucidation of 12 ...... 172

Investigation of the antioxidant activity of 12 ...... 179

PDT activity of 12 ...... 180

Discussion ...... 181

Conclusions ...... 183

Methods and Materials ...... 183

Chemicals and reagents ...... 183

Equipment ...... 183

One-step synthesis of 12 from bacteriopheophorbide a (16) ...... 184

Investigation of reaction conditions for the one-step synthesis of 12 from bacteriopheophorbide a (16) ...... 184

Investigation of the antioxidant activity of 12 ...... 186

PDT activity of 12 ...... 186

Acknowledgements ...... 187

Supporting Information ...... 188

Chapter 5 Future Directions ...... 197

Preamble ...... 197

Introduction ...... 197

Cancer-targeting aziridine aldehyde-cyclized RGD peptides ...... 198

Promising new bacteriochlorin photosensitizers ...... 200

Future bacteriochlorophyllones ...... 203

Targeted photodynamic therapy with RGD peptides and bacteriochlorin photosensitizers .. 204

Closing comments ...... 206

References ...... 207

Appendices ...... 230

x Cancer-targeting aziridine aldehyde-cyclized RGD-based peptides ...... 230

Promising new bacteriochlorin photosensitizers ...... 234

Towards targeted photodynamic therapy with RGD peptides and bacteriochlorin photosensitizers ...... 236

xi List of Tables

Table 2.2. Calculated heavy atom distances between macrocycles and the extracellular binding regions of the αVβ3 integrin receptor. Reproduced with permission from (179). Copyright © 2012, American Chemical Society...... 66

Table S3.1. DFT-based molecular modeling calculations of the singlet-state molecular orbital energies (eV) of 2a – 5a and 2b – 5b.a Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 159

Table S3.2. TD-DFT-based molecular modeling calculations of the triplet excited-state energies a for optimized S0 geometries (eV) of 2a – 5a and 2b – 5b Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 160

Table 4.2. Proton NMR shift assignments (δ ppm), splitting patterns, and J-values of the bacteriochlorins 16 and 12, and the chlorins 17 and 8a (with significant proton shifts bolded). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co. 177

xii Table 4.3. Carbon NMR shift assignments (DMSO-d6, δ ppm) of the bacteriochlorins, 16 and 12 (with significant carbon shifts bolded). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 178

Table S4.1. Proton NMR shift assignments (CDCl3, δ ppm) and splitting patterns of the chlorins, 17 and 8b (with significant proton shifts bolded). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 194

xiii List of Figures, Charts and Schemes

Figure 1.1.1. Ribbon representations of the NMR structures of the RGD-A and RGD-B peptides (top panel). The superimposed solution structures of 19 structures of RGD-A and RGD-B (middle panel). The NMR structures of the distorted type I β-turn for RGD-A (C2RGD segment) and the type II‘ β-turn for RGD-B (RGDC3 segment) (bottom panel). Reproduced with permission from (19). Copyright © 2001, American Chemical Society...... 6

Figure 1.1.2. A summary of the typical strategies involved in creating a novel cyclic peptide with a high binding affinity to cancer biomarker receptors. Reproduced with permission from (1). Copyright © 2012, Future Science Ltd...... 8

Figure 1.1.3. A representation of the principle that peptides can specifically bind to different receptor subtypes due to particular constrained peptide geometries. Reproduced with permission from (1). Copyright © 2012, Future Science Ltd...... 10

Figure 1.1.4. An illustration showing that cyclic peptides have high proteolytic stability compared to linear peptides and are therefore more useful for cancer biomarker targeting. Reproduced with permission from (1). Copyright © 2012, Future Science Ltd...... 12

Figure 1.1.5. Epifluorescence microscopy showing the in vitro binding of A) the linear KNGRG-OG (20 μM) and B) the cyclic cKNGRE-OG (20 μM) to HT-1080 (CD13+) cells following 30 min. incubations at 37 °C. Green fluorescence signals show binding of the peptide- OG conjugates and the Blue fluorescence signals show the nucleus-staining of DAPI. Images were acquired with identical exposure times and displayed consistent window and level. Scale bar = 100 μm. Reproduced with permission from (68). Copyright © 2010, Elsevier Publishers. 19

Figure 1.1.6. Radioimaging of a mouse bearing a 400-mg B16-F1 melanoma tumor capture with a γ-camera 30 min. after the injection of 99mTcCCMSH (25 mCi). This lateral image shows a high accumulation of the radioactivity in the tumor (b) and lower levels in the kidneys (a), bladder (c) and the tail vein injection site (d). The intensity of the γ-emission is color-coded to represent high radioactivity in white-yellow, medium radioactivity in orange and low radioactivity in dark red. Reproduced with permission from (84). Copyright © 1998, Proceedings of the National Academy of Sciences...... 24

xiv Figure 1.2.1. Simplified conventional labelling system for the trans-bacteriochlorin scaffold that denotes the four characteristic pyrrole rings and the carbon positions amenable to chemical modifications...... 30

Figure 1.2.2. Enhancements of the ΦISC (A, C, and D), ΦT (B) and ΦΔ (E) of bacteriochlorins by insertion of the metals Zn(II), In(III) (A) and Pd(II)...... 33

Figure 1.2.3. Enhancements of the ΦISC of bacteriochlorins by insertion of Zn(II)...... 36

Figure 1.2.4. Enhancements of ΦΔ by the heavy atom effect, showing the benefit of meso-tetra- arylchlorides (A), the advantage of the meta-SO3H moiety (B), and the importance of 2,6- dichlorophenyl-3-phenylsulfonic acid moieties (C)...... 38

Figure 1.2.5. Enhancements of ΦISC (A and B) and ΦT (C) afforded by the electron-donating methoxy group at the meso 5-position...... 41

Figure 1.2.6. Noted exceptions where the electron-donating methoxy group at the meso 5- position hindered the ΦISC...... 43

Figure 1.2.7. Potential strategies for enhancing the ΦISC of bacteriochlorins by modifications along the y-axis, namely by the use of unmodified phenyl moieties at the 3- and 13-positions (A), the avoidance of EWGs at the 2- and 12-positions (B), and the creation of asymmetric bacteriochlorins using the acetyl moiety (C)...... 45

Figure 1.2.8. Noted trend wherein the ΦISC of bacteriochlorins was hindered by cationic moieties positioned at the y-axis (A) and the x-axis (B)...... 48

Figure 1.2.9. Examples where the ΦISC was hindered by fusing aryl groups along the x-axis of bacteriochlorins...... 51

Chart 1.2.1. Bacteriochlorins known to generate type 1 PDT ROS upon photoirradiation...... 54

Figure 2.1. General strategy for synthesizing the fluorescein-labeled aziridine aldehyde-cyclized RGD macrocycles. Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 61

Figure 2.3. 3D models of A) cPRGDA (orange) and B) cPRGDAA (purple) separately showing Pro-Cα- Asp-Cα vectors and C) overlain with amide-cPRGDK (yellow). The cPRGDA peptide (orange) was overlain with cRGDf(N-Me)V (green) in a docked configuration with the αVβ3 protein to illustrate structural similarity at the RGD-motif. Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 64

Table 2.1. Calculated Pro-Cα- Asp-Cα distances of macrocycles using averaged conformer geometries...... 64

Figure 2.4. 3D modeling showing the comparative docking interactions of A) cPRGDA (yellow) and cRGDf(N-Me)V (blue) and B) amide-cPRGDK (orange) and cRGDf(N-Me)V (blue) with the extracellular binding regions of the αVβ3 integrin receptor. Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 65

Table 2.2. Calculated heavy atom distances between macrocycles and the extracellular binding regions of the αVβ3 integrin receptor...... 66

Table 2.3. Cell adhesion competition assay summary of IC50 values of fluorescein-labeled macrocycles...... 67

Figure 2.5. Confocal microscopy fluorescence images merged with DIC showing the differential binding of fluorescein-labeled macrocycles to U87 (αVβ3 +) (A – F) and HT-29 (αVβ3 -) cells (G – J). Reproduced with permission from (169). Copyright © 2012, American Chemical Society. 68

Figure S2.1. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening (7) after prep-HPLC purification. A) Scanning the collected UV channel at 254 nm shows a > 95 % purity of the product (9.6 min) while the peak < 3 min shows the void volume. B) Analysis of the ESI-MS(+) at 9.6 min identifies the molar mass of the desired product (calculated molar mass = 1036.31 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 79

xvi Figure S2.2. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDAA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening (8) after prep-HPLC purification. A) Scanning the collected UV channel at 254 nm shows a > 70 % purity of the product (9.3 min) while the peak < 3.5 min shows the void volume. B) Analysis of the ESI-MS(+) at 9.3 min identifies the molar mass of the desired product (calculated molar mass = 1107.39 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 80

Figure S2.3. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRDGA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening (9) after prep-HPLC purification. A) Scanning the collected UV channel at 254 nm shows a > 95 % purity of the product (9.6 min) while the peak < 3 min shows the void volume. B) Analysis of the ESI-MS(+) at 9.6 min identifies the molar mass of the desired product (calculated molar mass = 1036.31 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 81

Figure S2.4. HPLC-MS spectral data confirming the identity and purity of the side chain- protected amide-cPRGDK peptide (21) directly after amide cyclization and without purification. A) Scanning the collected UV channel at 254 nm shows a < 50 % purity of the product (10.4 min). The peak < 3 min shows the void volume, the peaks > 10.7 min shows oligomer byproducts and the peak at 6 min shows excess HBTU. B) Analysis of the ESI-MS(+) at 10.4 min identifies the molar mass of the desired product (calculated molar mass = 962.16 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 82

Figure S2.5. HPLC-MS spectral data confirming the identity and purity of the side chain- deprotected aziridine aldehyde-cyclized cPRGDA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to fluorescein (13) after prep-HPLC purification. A) Scanning the collected UV channel at 441 nm shows a > 95 % purity of the product (8.3 min). The peak < 3 min shows the void volume and the high baseline between 6 – 15 min is due to low absorbance values. B) Analysis of the UV-Vis spectrum at 8.3 min shows the characteristic spectrum of fluorescein in the desired product. C) Analysis of the ESI-MS(+) at 8.3 min identifies the molar mass of the desired product (calculated molar mass = 1086.18 g/mole)...... 83

Figure S2.6. HPLC-MS spectral data confirming the identity and purity of the side chain- deprotected aziridine aldehyde-cyclized cPRGDAA peptide bonded to cysteamine by sulfhydryl xvii aziridine ring-opening and conjugated to fluorescein (14) after prep-HPLC purification. A) Scanning the collected UV channel at 441 nm shows a > 95 % purity of the product (8.1 min). The peak at ~ 2 min shows the void volume and the high baseline between 6 – 15 min is due to low absorbance values. B) Analysis of the UV-Vis spectrum at 8.1 min shows the characteristic spectrum of fluorescein in the desired product. C) Analysis of the ESI-MS(+) at 8.1 min identifies the molar mass of the desired product (calculated molar mass = 1157.25 g/mole)...... 84

Figure S2.7. HPLC-MS spectral data confirming the identity and purity of the side chain- deprotected aziridine aldehyde-cyclized cPRDGA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to fluorescein (15) after prep-HPLC purification. A) Scanning the collected UV channel at 441 nm shows a > 95 % purity of the product (8.2 min) while the peak at ~ 2 min shows the void volume. B) Analysis of the UV-Vis spectrum at 8.2 min shows the characteristic spectrum of fluorescein in the desired product. C) Analysis of the ESI-MS(+) at 8.2 min identifies the molar mass of the desired product (calculated molar mass = 1086.18 g/mole)...... 85

Figure S2.8. HPLC-MS spectral data confirming the identity and purity of the side chain- deprotected amide-cPRGDK peptide conjugated to fluorescein (23) after prep-HPLC purification. A) Scanning the collected UV channel at 441 nm shows a > 95 % purity of the product (7.8 min). B) Analysis of the UV-Vis spectrum at 7.8 min shows the characteristic spectrum of fluorescein in the desired product. C) Analysis of the ESI-MS(+) at 7.8 min identifies the molar mass of the desired product (calculated molar mass = 911.91 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 86

Figure S2.9. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to DOTA (24) after prep-HPLC purification. A) Scanning the collected UV channel at 254 nm shows a > 90 % purity of the product (9.6 min). The peak < 3.5 min shows the void volume. B) Analysis of the ESI-MS(+) at 9.6 min identifies the molar mass of the desired product (calculated molar mass = 1422.71 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 87

Figure S2.10. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to pyro (25) after prep-HPLC purification. A) Scanning xviii the collected UV channel at 663 nm shows a > 95 % purity of the product (11.6 min). B) Analysis of the UV-Vis spectrum at 11.6 min shows the characteristic spectrum of pyro in the desired product. C) Analysis of the ESI-MS(+) at 11.6 min identifies the molar mass of the desired product (calculated molar mass = 1552.94 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society...... 88

Figure 3.4. Correlation of A) Qx λmax (MeOH) B) B3LYP/6-31G* (SPARTAN ‗-6) calculated HOMO−1 to LUMO energy gap, and C) 1H NMR shift of H23-NH due to the presence (■; 2a- 5a) or absence (□; 2b-5b) of the exocyclic E-ring. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 99

Figure 3.5. The calculated ΔEST energy gap (B3LYP/6-31G*) for compounds with (■; 2a-5a) and without (□; 2b-5b) the exocyclic E-ring. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 101

Figure S3.1. 1D 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2- trimethylaminoethyl)ester (2a) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 125

Figure S3.2. 2D COSY 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2- trimethylaminoethyl)ester (2a) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 126

Figure S3.3. 13C Jmod NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2- trimethylaminoethyl)ester (2a) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 127

Figure S3.4. HPLC-MS characterization of 31-oxo-rhodobacteriochlorin 173-(2- trimethylaminoethyl)ester (2a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass

Figure S3.5. 1D 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-methyl ester (3a) in

Figure S3.7. 13C Jmod NMR spectrum of 31-oxo-rhodobacteriochlorin 173-methyl ester (3a) in

Figure S3.8. HPLC-MS characterization of 31-oxo-rhodobacteriochlorin 173-methyl ester (3a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 132

Figure S3.12. HPLC-MS characterization of bacteriopheophorbide a (4a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 136

Figure S3.13. 1D 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide

xx Figure S3.14. 2D COSY 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2- sulfoethyl)amide (5a) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 138

Figure S3.15. 13C Jmod NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide

Figure S3.16. HPLC-MS characterization of 31-oxo-rhodobacteriochlorin 173-(2- sulfoethyl)amide (5a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 140

Figure S3.17. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 1 3 13 -(2-sulfoethyl)amide-17 -(2-trimethylaminoethyl)ester (2b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 141

Figure S3.18. 2D COSY 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131-(2-sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b) in DMSO- d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society. 142

Figure S3.19. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131-(2-sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b) in DMSO- d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society. 143

Figure S3.20. HPLC-MS characterization of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131-(2-sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 144

Figure S3.21. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 3 13'-(2-sulfoethyl)amide-17 -methyl ester (3b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 145

Figure S3.22. 2D COSY 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 3 rhodobacteriochlorin 13'-(2-sulfoethyl)amide-17 -methyl ester (3b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 146

xxi Figure S3.23. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 3 rhodobacteriochlorin 13'-(2-sulfoethyl)amide-17 -methyl ester (3b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 147

Figure S3.24. HPLC-MS characterization of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide-173-methyl ester (3b) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 148

Figure S3.25. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin

Figure S3.26. 2D COSY 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 150

Figure S3.27. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 151

Figure S3.28. HPLC-MS characterization of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 152

Figure S3.29. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 1 3 13 ,17 -di(2-sulfoethyl)amide (5b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 153

Figure S3.30. 2D COSY 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 1 3 rhodobacteriochlorin 13 ,17 -di(2-sulfoethyl)amide (5b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 154

xxii Figure S3.31. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 1 3 rhodobacteriochlorin 13 ,17 -di(2-sulfoethyl)amide (5b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 155

Figure S3.32. HPLC-MS characterization of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131,173-di(2-sulfoethyl)amide (5b) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 156

Figure S3.33. HPLC-MS characterization of palladium bacteriopheophorbide a (1a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 157

Figure S3.34. HPLC-MS characterization of palladium 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide (1b) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 158

Table S3.1. DFT-based molecular modeling calculations of the singlet-state molecular orbital energies (eV) of 2a – 5a and 2b – 5b...... 159

Table S3.2. TD-DFT-based molecular modeling calculations of the triplet excited-state energies for optimized S0 geometries (eV) of 2a – 5a and 2b – 5b ...... 160

Figure S3.36. Determination that combinations of DMSO (1% - 3%, v/v) and cremophor (0.1% - 1.0%, v/v) in PBS are biocompatible with A549 cancer cells after 1 h incubations at 37 ˚C by MTT cell viability analysis (in quintuplicate, error bars show ± STDERR). Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 164

Figure S3.37. UV-vis spectra of 1a-5a and 1b-5b (~50 μM) in PBS containing 2% DMSO (v/v) and 0.5% cremophor (v/v) showing either dissolution or aggregation indicated by spectral red- shifting of Qy absorption bands beyond 800 nm. Reproduced with permission from (163). Copyright © 2014, American Chemical Society...... 165

xxiii Figure 4.1. Previously described natural porphyrins (1-4), chlorins (5-10), the previously synthesized bacteriochlorin 11, and the novel bacteriochlorin 12, each containing the seven- membered exocyclic F-ring...... 167

Figure 4.2. Synthetic route of 11 (261). (i) pyridine, 105 °C, 72 h (ii) CH(OCH3)3, HCl, MeOH,

Figure 4.3. Synthetic route of 12. (i) 1-propanol; (ii) dil. HCl (aq); (iii) 80% TFA (aq) (v/v); (iv) choline chloride, HBTU, DMAP, DIPEA, DMSO, Ar (g). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 169

Figure 4.5. Structures and proton positions of the bacteriochlorins 16 and 12, and the chlorins 17 and 8a. Reproduced with permission from ...... 173

1 Figure 4.6. 2D H COSY NMR spectrum (Bruker 400 MHz, DMSO-d6, δ ppm) of 12 illustrating partial proton shift assignments due to the coupling of adjacent protons. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 174

1 Figure 4.7. Partial H NMR spectra (Bruker 400 MHz, DMSO-d6, δ ppm), proton assignments and partial structures of a) 16 and b) 12 with curved arrows showing downfield shifts of the 17, 171, 172 and 181 protons and the absence of the 132 proton of 12. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 175

13 Figure 4.8. Partial Jmod C NMR spectra (Bruker 100 MHz, DMSO-d6, δ ppm), carbon assignments and partial structures of a) 16 and b) 12 with curved arrows showing the significant downfield shift of the 173 ketone carbon, a slight upfield shift of the 131 carbonyl carbon and the presence of the quaternary 132 carbon of 12. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 179

Figure 4.9. Investigation of the concentration-dependant antioxidation of AUR by 12 in 0.5 μM

xxiv Figure 4.10. Dose-response curves showing the concentration-dependent viability of U87 cells treated with 12 under dark conditions (solid trace) and after irradiation with a 740 nm LED light box (dashed trace), elucidated by MTT analysis (N = 3, error bars indicate ± STDERR). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co. 181

Figure S4.1. HPLC absorbance chromatogram showing that the described reaction conditions (Table 4.1, entry 11) produced an esterified product of 17, but did not result in the 132-173- cyclization of 17 after 8 days (% of compounds in crude calculated by integrations at 663 nm). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co. 188

Figure S4.2. Biotage absorbance chromatograms showing that a) the consumption of the precursor, 16, due to choline esterification (Table 4.1, entry 12) enhanced the resolution of the product, 12, during purification compared to b) excluding choline chloride (Table 4.1, entry 13) during 132-173-cyclization of 16 at ambient temperature. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 189

Figure S4.4. HRMS (ESI+) spectrum showing the [M + H]+ (593.3 m/z) and the [M + Na]+ adduct (615.3 m/z) of 12. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 191

Figure S4.8. HPLC-MS spectra showing the a) purity, b) the UV-Vis spectrum and c) the ESI+MS spectrum of 12. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co...... 196

xxv List of Abbreviations

α-MSH - α-melanocyte stimulating hormone

A2bu - α,β-diaminobutyric acid

Abu – 2-aminobutyric acid

A2pr – 2,3-diaminopropanoic acid

Ac - acetyl

ACN – acetonitrile

AUR – Amplex UltraRed

Boc – tert-butoxy carbamate

CI – confidence interval

Chg - α-cyclohexylglycine

COSY – correlation spectroscopy

Crem – cremophor

Cyp – cypate dAc – des-acetyl

DCC – N,N'-dicyclohexylcarbodiimide

DCM – dichloromethane

DFT – density functional theory

DIAD – diisopropyl azodicarboxylate

DIBAL – diisobutylaluminum hydride

DIC – differential interference contrast

DIPEA – N,N-diisopropylethylamine

DMAP – 4-(N,N-dimethylamino)pyridine

DMEM – Dulbecco‘s Modified Eagle Medium

DMF – dimethylformamide

DMSO – dimethyl sulfoxide

xxvi DMTU – N,N'-dimethyl thiourea

DOTA – 1,4,7,10-tetraazacyclododecane-1,4,7-tri(t-butyl acetate)-10-acetate

Dpr – 2,3-diaminopropionic acid

EDG – electron donating group

EDTA – ethylenediamine-tetraacetic acid erbb-2 – epidermal growth factor receptor-type 2 gene

ESI – electrospray ionization

ESR – electron spin resonance

EWG – electron withdrawing group

FBS – fetal bovine serum

Fmoc – fluorenylmethyloxycarbonyl

GB/SA – Generalized-Born/Surface-Area

HBTU – O-benzotriazole-N,N,N',N'-tetramethyl-uronium-hexafluoro-phosphate

HIF – hypoxia-inducible factor

HOMO – highest occupied molecular orbital

HPLC – high-performance liquid chromatography

HRMS – high-resolution mass spectrometry

HRP – horseradish peroxidase

Hyp – hydroxylproline

IC – internal conversion

ICP-MS - Inductively coupled plasma mass spectrometry

IFN - interferone

IgG1 – Immunoglobulin G subclass 1

ISC – intersystem crossing

ITC - isothiocyanate

Jmod – J-modulated spin-echo

xxvii KLK2 - kallikreain

LC – liquid chromatography

LED – light-emitting diode

LTSLs – lysolipid-containing temperature sensitive liposomes

LUMO – lowest unoccupied molecular orbital

MCMM – Monte Carlo Multiple Minima

MEM – Modified Eagle Medium

MeOH – methanol

MS – mass spectrometry

MT1-MMP –membrane type 1-matrix metalloproteinase

MTT – 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

NGF – nerve growth factor

NHS – N-hydroxysuccinimide

NIR – near-infrared

NMR – nuclear magnetic resonance

OD – optical density

OG – Oregon Green

Orn - ornithine

OtBu – O-tert-butyl

Pbf – 2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl

PBS – phosphate-buffered saline

PDB – Protein Data Bank

PDT – photodynamic therapy

Pen – 8,8-dimethylcysteine

PMT – photomultiplier tube

Prep – preparatory

xxviii PS – photosensitizer

Pyro – pyropheophorbide a

ROS – reactive oxygen species

SAR – structure-activity relationship

SDex – dexamethasone-derived steroid

SE – standard error

SOSG – Singlet Oxygen Sensor Green

SRIF – somatostatin-14

SST2 – somatostatin receptor subtype 2

Suc – 3-carboxypropanoyl

TCEP - tris-(2-carboxyethyl)phosphine

TDDFT – time-dependent density functional theory

TEAA – triethylammonium acetate

TFA – trifluoroacetic acid

TFE - trifluoroethanol

THF - tetrahydrofuran

Tiron – 1,2-dihydroxybenzene-3,5-disulfonic acid disodium salt hydrate

TIS - triisopropylsilane

TLC – thin-layer chromatography

TNF – tumor necrosis factor

TrkA – tyrosine kinase receptor A

UV-vis – ultraviolet-visible

VEGF – vascular endothelial growth factor

xxix List of Appendices

Figure A1.1. Investigating modified aziridine aldehydes to assess their compatibility with the sulfhydryl-aziridine ring opening-based conjugation strategy described in chapter 2 (169)...... 230

Figure A1.2. The noted limitation of the current cysteamine-based aziridine ring opening reaction (A and B) and the proposed optimization of the desired sulfhydryl-aziridine ring opening using (C) Fmoc-protected cysteamine monomers that eliminates the undesired amino- aziridine ring opening pathway (D and E)...... 231

Figure A1.3. Preliminary exploration of the binding specificity of cPRGDA, cPRGDAA and the negative control cPRDGA into buffered solutions of purified cancer-related integrin variants A)

αVβ5, B) αVβ6 and C) α5β1, using fluorescence polarization spectroscopy (N = 1)...... 232

Figure A1.4. HPLC spectra showing the successful labeling of the amide-cyclized cPRGDK(DOTA) macrocycle using A) ‗cold‘ 63Cu(II) and B) radioactive 64Cu(II)...... 233

Figure A2.1. MTT-based cell viability screening assays for bacteriochlorins (50 μM, 4 h incubations at 37 °C in (2:0.5:97.5) DMSO: cremophor: media (v/v) used to test dark toxicities and PDT efficacies with A) 22Rv1 prostate cancer, B) A549 non-small cell lung cancer, C) HT29 colon cancer and D) U87 glioblastoma cell lines (N=3, error bars represent ± STDERR)...... 234

Table A2.1. The charge and water-solubility of selected bacteriochlorin photosensitizers from Chapter 3...... 234

Figure A2.2. Clonogenic assay results showing the surviving fractions of RIF-1 cells treated with the controls bacteriochlorins (1a or 1b) and 2b (24 h incubations at 37 °C in RPMI-1640 media containing (10:1:0.05) FBS: DMSO: cremophor) under either dark or light (740 nm irradiation) conditions in a normoxic atmosphere (N = 3, error bars represent 95% confidence intervals)...... 235

Figure A2.3. Western blot showing the expression of HIF1-α in RIF-1 cells after exposures to hypoxia (0.2 % O2) ranging from 1 h to 48 h‘s, compared to the control (0 h) cells grown under normoxic conditions...... 235

xxx Figure A3.1. Competitive cell adhesion assay results showing the concentration-dependent displacement of integrin-expressing U87 cells from vitronectin-coated wells using increasing concentrations of Ac-PRGDK(Fluorescein) after 30 minute co-incubations at 37 °C (A). Confocal microscopy images showing the specificity of Ac-PRGDK(Fluorescein) (10 μM, 2 h incubations at 37 °C) to integrin-expressing U87 cells (B and C) compared to the negative control, HT29 cells (D and E)...... 236

Figure A3.2. Evaluations of ROS photogeneration by 4b and 1a in (97.5:2:0.5) RPMI-1640: DMSO: cremophor (v/v/v) using A) AUR and B) SOSG (data from Chapter 3 (163); N = 3, error bars represent ± STDERR). MTT-based cell viability assays showing the PDT efficacy (4 J/cm2 irradiation at 740 nm) and dark toxicity of 4b and 1a (50 μM, 4 h incubations at 37 °C in (97.5:2:0.5) media: DMSO: cremophor (v/v/v)) using C) 22Rv1 prostate cancer, D) A549 non- small cell lung cancer, E) HT29 colon cancer and F) U87 glioblastoma cell lines (data from Figure A2.1; N = 3, error bars represent ± STDERR)...... 237

Scheme A3.1. Synthesis of the integrin-targeting peptide-photosensitizer conjugate, Ac- PRGDK(4b)...... 238

Scheme A3.2. Synthesis of the integrin-targeting peptide-photosensitizer conjugate, Ac- PRGDK(1a)...... 239

Figure A3.3. Evaluation of the A) ROS photogeneration (with SOSG), B) MTT-based (4 h incubations at 37 ºC) dark toxicity and PDT efficacy (4 J/cm2) of the photosensitizers, 1a and 4b, and the integrin targeting peptide-PS conjugates, Ac-PRGDK(1a) and Ac-PRGDK(4b) in DMEM media (with 2% DMSO and 0.5% cremophor, v/v) (N = 3, error bars represent ± STDERR)...... 240

xxxi Chapter 1 Introduction to Cyclic Cancer-Targeting Peptides and Bacteriochlorin Photosensitizers

Preamble

The first section of this Introduction chapter is comprised of the manuscript, ―Flexible or fixed: a comparative review of linear and cyclic cancer-targeting peptides‖ that was adapted from Roxin, Á., and Zheng, G., Future Medicinal Chemistry 2012, 4(12), 1601-1618, (1) Copyright © 2012 Future Science Ltd. with permission of Future Science Ltd. This publication has been reformatted to the required SGS style. This first section of the chapter is meant to present the rationale as to why cyclic cancer-targeting peptides were the focus of study in Chapter 2 of this thesis. Its purpose is to convey the advantages of cyclic cancer-targeting peptides and to impress upon the importance of contributing to the body of knowledge in this growing field biomedical research. I was the primary contributor to this review. Dr. Zheng reviewed and edited the manuscript for submission.

The second section of this chapter is a review of both known and potentially new strategies for tuning the photophysical properties of bacteriochlorins with specific structural modifications. This section of the Introduction chapter is meant to present a summary of methods that could aid the design and development of new second-generation bacteriochlorin photosensitizers. Its purpose is also to introduce the rationale as to why the particular structural modifications of bacteriochlorins were investigated in Chapter 3 of this thesis. I was the primary contributor to this review, while Dr. Zheng reviewed and edited this section of the thesis.

The third section of this chapter is an introduction to the overarching goals of the work presented in Chapters 2, 3 and 4 of this thesis. It is meant to present the rationale as to why the particular studies were conducted and how the results and insights of these chapters can contribute to achieving the goal of cancer-targeted photodynamic therapy through the development of new third-generation cancer-specific bacteriochlorin photosensitizers.

2 Section 1.1. Flexible or fixed: a comparative review of linear and cyclic cancer-targeting peptides

Summary

Peptides can serve as versatile cancer-targeting ligands and have been used for clinically relevant applications such as cancer imaging and therapy. A current and long-standing focus within peptide research is the creation of structurally constrained peptides generated through cyclization. Cyclization is envisioned to enhance the selective binding, uptake, potency and stability of linear precursors. This review compares closely related linear and cyclic peptides in these respects. Peptide cyclization generally improves the selective binding and stability of linear precursors. However, not all cyclization strategies and constrained geometries enhance these properties to the same extent. In some instances, linear analogues actually have better cancer- targeting properties compared to their cyclic counterparts. Though cyclization does not necessarily improve the cancer-targeting properties of linear analogues, cyclic peptides may obtain properties that allow them to be used for additional applications. This review is meant to convey the advantages and limitations of cyclic cancer-targeting peptides.

Introduction

Peptides are useful receptor-binding ligands. Several other classes of targeting ligands sharing this binding property include small molecules, endogenous proteins, antibodies and affibodies. Peptides can be synthesized with relative ease using well-established solid-state Fmoc and Boc synthetic techniques. Compared to proteins and antibodies, this ease of synthesis makes peptides relatively inexpensive to produce and modify for research and commercial purposes. Targeting peptides can be conjugated to small molecule drugs, imaging contrast agents and nanoparticles for a plethora of applications. The geometry of linear peptides can also be modified using a wide variety of traditional and modern chemistries. A highly-studied geometric modification in the field of peptide research is the creation of cyclic peptides.

As new biomarkers have been, and are continuing to become identified, peptides have been chosen for targeting these biomarkers. Efficient peptide ligands are discovered by screening random peptide libraries and also by using computer-aided modeling techniques. Optimized peptide sequences have been cyclized for several key reasons. First, it is envisioned that by gaining a fixed geometry, cyclic peptides could bind more efficiently to their respective

3 receptors. If those receptors exist as several subtypes, it is also hoped that the fixed geometry of cyclic peptides will make these constrained sequences selective to particular receptor subtypes compared to their linear analogues. In addition, cyclic peptides are often chosen over their linear analogues due to their enhanced enzymatic stability. Several studies have identified the properties of novel cyclic peptides. However, no review has yet summarized the properties of closely related linear and cyclic peptides. This review will present examples where linear peptides have been compared to their direct cyclic analogues or optimized cyclic derivatives for cancer imaging or cancer therapy. These comparisons will explore differences between these linear and cyclic peptides in terms of their geometry, binding affinity, biological activity, receptor selectivity and stability. This review will also show that cyclic peptides can be used in ways that linear peptides cannot. Though cyclic peptides often show improvements in some respect(s) compared to their linear precursors, peptide cyclization is not a panacea. Therefore, we will highlight instances where cyclic analogues did not perform as favourably as their linear counterparts. The goal of this comparative review is to aid current peptide research by conveying the advantages, limitations and unique properties of cyclic cancer-targeting peptides.

Geometry of linear and cyclic peptides

Linear peptides that contain 2 to 10 amino acid residues are especially flexible in solution. Once the length of linear peptides extends to between 10 to 20 amino acids residues, random linear peptide sequences can begin to obtain secondary structures including α-helices, turns and β- strands (2). Computer-aided optimization studies can lead to designing linear peptides that contain such secondary structures. For example, D‘Andrea et al. have designed the linear 15- amino acid peptide QK-Ac-KLTWQELYQLKYKGI-NH2 that mimicked the α-helix of the vascular endothelial growth factor protein fragment (VEGF17-25) and was more biologically active than the unstructured isolated sequence of VEGF17-25 (VEGF15, Ac-

KVKFMDVYQRSYCHP- NH2) (3). These secondary structures impose constraints that reduce the free energy of linear peptides and limit their conformations to those that may be more biologically active.

In lieu of in-depth computer modeling, the geometry of short peptide sequences can be easily constrained by cyclization. Several groups have reviewed the geometries that result from peptide cyclization (4-9).The constraints imposed by cyclization will force cyclic peptides to adopt a limited number of molecular conformations in solution. When peptides are to be used as ligands

4 for targeting disease biomarkers, only a limited number of conformations will allow the ligand to bind efficiently. While refining a Monte Carlo simulation method for improving ligand-receptor docking, Deem and Bader noted that if the conformations of cyclic peptides are limited, the time spent in each conformation in solution is proportional to the inverse of the exponential of the free energy of the conformation (10). Cyclic peptides may continue to lose entropy upon interacting with a target (2); the result is a compounded reduction of the free energy of cyclic peptides when they are used as targeting ligands. Generally, if cyclization limits conformations to those required for optimum receptor binding, these cyclic peptides would be more useful compared their linear counterparts which can adopt more conformations that are not useful for receptor binding.

Cyclization has been shown to increase the propensity for β-turn formation in peptides. The formation of this secondary structure is important since β-turns are often found in native proteins (11, 12). Lau et al. generated a cyclic peptide library by the one-bead-one-compound (OBOC) technique and discovered several disulfide-cyclized cNGXGXXc (X = interchangeable residues) peptides that bound to the non-small cell lung cancer biomarker α3β1 in A549 cells (13). Three- dimensional modeling of the best hit within the library, cNGQGEQc, suggested that the D- cysteine to L-glutamine motif adopted a β-turn that was responsible for the improved binding affinity. Beta-turns naturally exist in nine specific conformations which include type I, I‘, II, II‘, IV, VIa1, VIa2, VIb and VIII. Though cyclization can support the formation of these reverse turns, it is difficult to create the exact types of β-turn that will match those of native proteins. This is an ongoing avenue of cyclic peptide research that is being refined. Several groups have summarized the various synthetic methods used to generate cyclic peptides (14-17). It is possible that particular synthetic methods can support the formation of specific geometries.

Though peptide cyclization generally induces structural constraints, the site of cyclization within the sequence can affect the binding affinity of cyclic peptides. Kumar et al. compared 20 sidechain modified linear peptides and 11 cyclic analogues to the original linear sequence Ac- CIYKYY for Src tyrosine kinase inhibition (18). The cyclic derivatives included the original CIYKYY sequence but were cyclized by either head-to-tail, N- or C-termini to sidechain or sidechain-to-sidechain coupling strategies. This study found that the head-to-tail cyclized peptide had an inhibitory potency that was 62.5-fold higher (IC50 = 6.4 µM) compared to the original linear peptide (IC50 = 400 µM). This cyclic peptide was also more potent than the seven other cyclic peptides with the same sequence, which were cyclized by the different aforementioned chemistries (IC50 = 16 - >400 µM). The differences in the binding affinities of the

5 aforementioned peptides appeared statistically significant since the upper limit of the standard errors for all values presented in this report were within ±10% of the calculated means. Two other cyclic peptides with the same sequence had low to sub-micro molar IC50‘s (1.9 µM and 0.28 µM) due to their constrained geometry attained by cyclization through the sidechains of residues 3 (tyrosine) and 4 (lysine) by extended linkers. This study illustrates that peptide cyclization can improve the biological activity of linear analogues, but that the position of cyclization will constrain particular motifs and modulate the biological activity of the active motif.

Peptides containing the same residues, which are cyclized by the same chemistry, but at different positions, can create alternate geometries that elicit different affinities to targeted biomarkers. Assa-Munt et al. created two synthetic derivatives of the native RGD-containing peptide sequence ACDCRGDCFCG (RGD-4C) (19). These were cyclized by oxidizing two different pairs of cysteine residues to create two isoforms labeled RGD-A (C1-C4 and C2-C3) and RGD-B 1 3 2 4 (C -C and C -C ). Competitive binding studies found that the RGD-A isoform bound to αvβ3 (Kd

~ 100 nM) and αvβ5 (Kd ~ 100 nM) integrin receptors while the binding of RGD-B was ~10-fold less efficient for these receptors. These results were presented graphically, using calculated means and standard deviation error ranges, and were deemed statistically significant. Modeling analysis showed that RGD-A contained a hydrophobic pocket and type I β-turn while the RGD-B isoform contained type II‘ β-turn (Figure 1.1.1). It wasn‘t clear which structural components were responsible, but results showed that the geometry of the RGD-A isoform allowed for better integrin receptor binding.

The type of chosen cyclization chemistry and even the orientation of the cyclizing amide bond can affect the resulting biological activity of cyclic peptide analogues. García-Aranda et al. attempted to mimic the β-hairpin fragment of the VEGF81-91 protein by creating cyclic peptides which were constrained by either disulfide or amide cyclization (20). They found that the cyclic 2,10 peptide Ac-M-c(CH2-NH-CO-CH2) GIKPHQGQGI-NH2 had a 2-fold lower IC50 (87.6 ± 5

µM) for displacing the VEGF165 protein from the D1-D3 domains of VEGFR-1 which was significantly lower compared to the linear H-MRIKPHQGQHI-OH peptide (160.0 ± 10.7 µM) that had a similar sequence (error ranges represent the standard errors of the calculated means). At 100 µM concentrations, this cyclic peptide displayed the best inhibition of biotinylated

VEGF165 (53.0 ± 2.2 %) from whole extracellular domain compared to the derivative with a reversed amide components (15.8 ± 3.2 %) and the derivative that was disulfide-cyclized (11.2 ±

6 2.1 %) but contained the same sequences. These results were deemed statistically significant due to the presented standard error ranges. Though biological activity was improved, molecular modeling and 2D NMR studies showed that the whole native β-hairpin structure was not stabilized. However, amide cyclization favoured the formation of a β-turn mimic of the VEGF81-

91 β-hairpin. These constrained structures were likely responsible for improving the binding affinity of the amide-cyclized sequences compared to the original linear sequence.

7 Library screenings and structure-activity studies

Novel cancer-targeting peptide ligands can be identified by screening libraries generated by phage display or combinatorial methods. Several groups have reviewed the topic of peptide library generation (21-24). Some studies generated and screened libraries of linear sequences while others created cyclic analogues of the most promising linear peptides in order to study the influence cyclization had on peptide binding activities (Figure 1.1.2). Bonetto et al. generated a library of peptides using phage display to discover novel ligands that mimicked the functional epitope of the crystallizable fragment of immunoglobulin G (IgG1-Fc) for targeting the Fc-γ receptor 1 (FcγRI) (25). The FcγRI target has several complex biological roles and a potent targeting ligand was proposed to have therapeutic properties against cancer and inflammatory disorders. After phage screening, they identified the conserved sequence TXXCXXΘPXLLGCΦXE (Θ = hydrophobic residues, Φ = acidic residues, X = any interchangeable residue) and found that two of the 27 selected sequences were linear. When these two linear peptides were compared to 15 of the most promising cyclic sequences at 50 µM peptide concentrations, they found that most of the cyclic sequences were able to inhibit the binding of IgG1 to FcγRI better than the linear peptides. The remaining cyclic peptides that had a similar binding activity to FcγRI as the linear peptides were originally found in low (< 3 %) frequencies during round 3 of the phage biopanning studies. Since there were no cyclic peptides that contained the same sequence as either of the two linear peptides, it cannot be concluded that cyclization was solely responsible for improving binding to FcγRI. However, since there were ~ 5 times more cyclic peptides discovered which contained the aforementioned conserved sequence; it is possible that these linear peptides would target FcγRI more efficiently if each were cyclized.

Phage display was used by Hsiao et al. to identify promising ligands with high binding affinities to the αvβ6 integrin receptor on human oral squamous cell carcinoma (OSCC) and, by extension, to major head and neck squamous cell carcinomas (HNSCC) (26). This cyclic peptide phage display library generated 7-amino acid sequences with N- and C-terminal cysteine residues that allowed disulfide cyclization. Biopanning found that the best hit included the sequence CRGDLASLC (cRGD). Among their comparative studies, the cyclic cRGD peptide was compared to the cyclic CRGELASLC (cRGE) peptide. The cRGD peptide significantly reduced the number of HSC-3 cells to 41 % compared to the PBS control sample at 300 µM concentrations (p < 0.001) while the cRGE had a similar activity at 300 µM concentrations as the cRGD peptide had at 100 µM concentrations. In a separate experiment, neither the linear RGDLASL peptide (RGD) nor the linear RGELASL peptide (RGE) could reduce the number of HSC-3 cells compared to the PBS control even at 1 mM concentrations. These comparisons showed that the αvβ6 binding affinity of the RGDLASL sequence was improved by disulfide cyclization. It was further shown that even the RGE-containing disulfide-cyclized peptide could attain αvβ6 binding activity due to cyclization.

The phage display method was also recently used by Duncan et al. to discover novel ligands for targeting the phosphorylation-dependent peptidylprolyl isomerase, Pin 1 (27). Since Pin 1overexpression has been observed in cervical, colon, lung and oral cancers, a potent Pin 1- targeting ligand would have potential for diagnosing and treating these cancers. A library was

9 generated that contained 7-amino acid binding sequences that were flanked by cysteine residues to allowed disulfide cyclization. The most promising sequence within this library contained the residues CRYPEVEIC. They compared the binding of the cyclic and reduced linear analogue of CRYPEVEIC by inhibiting Pin 1 phosphorylation-dependent peptidylprolyl isomerase (PPIase) activity at various concentrations of the substrate peptide cis-Suc-AEPF-pNA (Suc = 3- carboxypropanoyl). They found that the cyclic analogue had an 85-fold higher receptor binding affinity (Ki = 0.52 ± 0.07 µM) compared to the linear peptide (Ki = 44 ± 14 µM). The geometry induced by disulfide cyclization was required for this activity since amide cyclization, afforded by replacing the N- and C-terminal cysteines with lysine and glutamic acid (respectively), did not inhibit Pin 1 PPIase activity. These results suggest that disulfide cyclization was essential for improving the binding of the CRYPEVEIC sequence to Pin 1 and that this sequence can be pursued for cancer therapy.

The binding activity of cyclic cancer-targeting peptides can be optimized using structure-activity studies. Pomilio et al have compiled an extensive review on this topic (28). One such structure- activity study investigated the effect cyclization had on the binding of the modified cyclin recognition motif R-Xaa-L-Yaa-Zaa (Xaa = diamino residues, Yaa and Zaa = hydrophobic residues) to the cyclin binding groove. Andrews et al. compared linear and cyclic peptides for the inhibition of cyclin A- and cyclin E-associated cyclin-dependent kinase-2 (CDK2) activities to discover new ligands that could be useful for inducing apoptosis in tumor cells (29). They synthesized 7 cyclic peptides that contained the Xaa-LFG sequence that were cyclized by tail-to- sidechain coupling by the C-terminal glycine and Xaa (lysine, ornithine (Orn), α,γ- diaminobutyric acid (A2bu) or 2,3-diaminopropanoic acid (A2pr)) and contained 1-5 residues towards the N-terminus beside Xaa. The binding affinities of these peptides were compared to 4 linear peptides using a competitive cyclin A binding assay. Results showed that the cyclic peptide Ac-AA-Abu-R(KLFG) (Abu = 2-aminobutyric acid) had the lowest IC50 (0.63 ± 0.09 µM), which was 94-fold lower compared to a similar linear analogue with the sequence Ac-AA-

Abu-RNLFG-NH2 (IC50 = 59.4 ± 8.26 µM). This suggested that cyclization and the replacement of asparagine with lysine were key factors that improved binding.

Receptor subtype specificity of linear and cyclic peptides

The cyclization of linear peptide sequences can create constrained geometries that can alter the specificity of cyclic peptides to different isoforms or subtypes of targeted receptors (Figure

10 1.1.3). The linear RGD-motif can bind to several different integrin subtypes (generally with

IC50‘s > 100 nM) and therefore binds to integrins with little specificity (30). However, the cyclization of RGD peptides can create derivatives that are specific to particular integrin receptors. Pfaff et al. conducted a structure-activity study where they compared the binding affinity of 16 cyclic RGD peptides to the linear GRGDS and RGDFv peptides with respect to several different integrin receptors (31). Of all of those compared, the cyclic RGDFv had a 3.3- fold better binding affinity (IC50 = 11.3 ± 1.2 µM) to soluble αvβ3 and a 5.8-fold better binding affinity (IC50 = 0.4 ± 0.1 µM) to immobilized αvβ3 compared to the linear RGDFv peptide (IC50‘s = 37 ± 2 µM and 2.3 ± 0.7 µM, respectively). Interestingly, the linear RGDFv showed a 1.3-fold

(IC50 = 0.11 ± 0.07 µM) and 1.5-fold (IC50 = 11 ± 0.2 µM) better binding affinity to the soluble and immobilized αIIbβ3 integrin (respectively) compared to the cyclic RGDFv (IC50‘s = 0.14 ± 0.09 µM and 16 ± 1.4 µM, respectively). This linear RGD also had a 4.3-fold higher binding affinity (IC50 = 32 ± 2 µM) to immobilized α5β1 compared to the cyclic sequence (IC50 = 138 ± 20 µM). Though cyclization did not improve the binding affinity of RGD sequences to all integrin receptors, this study showed that cyclization created a fixed geometry that made the

RGDFv sequence more specific to αvβ3 compared to its linear counterpart.

11 A thorough structure-activity study by Haubner et al. compared the specificity of 27 cyclic RGD- containing pentapeptides to the linear GRGDSPK standard in terms of binding affinity to αvβ3 and αIIbβ3 (32). They found that almost every tested cyclic RGDxV, RGDfX, RGDXv and RGDFx (X = interchangeable L-amino acid and x = interchangeable D-amino acid) derivative bound with a better binding affinity to αvβ3 compared to the linear standard. However, the affinities of these cyclic peptides were similar to the linear reference in terms of αIIbβ3 integrin binding. Interestingly, Bach et al. suggested that if particular cyclic RGD peptides can be tuned to contain type I instead of a type II‘ β-turn, the resulting constrained cyclic RGD peptides can allowed preferential binding to αvβ3 compared to the αIIbβ3 integrin (33). This is not the case for all RGD-containing sequences. Wermuth et al. synthesized 18 derivatives of the cyclic RGDFV sequences which included five parent sequences and their respective inverso, retro and retro- inverso isomers (34). When the binding affinities of each peptide were compared to the linear reference GRGDSPK for the αvβ3 and αIIbβ3 integrins, they found that the cyclic RGDFv (parent), RGDfV (parent) and vfdGR (retro-inversion) peptides had 40- to 250-fold better binding affinities to αvβ3 but did not show any improvements for αIIbβ3 binding. It should be noted though that this report did not state any error ranges for their bind affinity measurements. Structural analyses continued to show that the aforementioned cyclic peptides contained a type II‘ β-turn. In addition, the retro-inverso isomers of RGDFv and RGDfV contained type II β-turns and had a 5- fold and 3.5-fold lower binding affinity (respectively) to αvβ3 compared to the linear standard. Again, this report did not present any error ranges or statistical comparisons. While statistical comparisons were not presented in this report, the findings suggest that the RGDFV-peptides preferentially bound to αvβ3 integrins and that this may have been partially due to the presence of a type II‘ β-turn. It is possible that constraining the RGD-motif to other types of β-turns can further modulate its specificity to other integrin subtypes.

Stability of linear and cyclic peptides

Short linear peptides exist in fast equilibrium of interchanging conformations while in solution. Of these rapidly interchanging conformations, only few will attain an orientation which has receptor selectivity. Though some conformations will preferentially interact with desired receptors, particular conformers can also fit into active sites of proteolytic enzymes. Peptides can be cyclized in order to reduce the overall numbers of interchanging conformers in hopes of limited them to those selective for desired receptors while avoiding degradation by not forming conformers susceptible to interacting with proteolytic enzymes (Figure 1.1.4) (35). Though the

12 theory is intuitive, achieving both goals is not at all trivial. In general though, cyclization often increases the stability of peptides which can prolong their biological activity. This prolonged activity may even be the result of additional resistance to enzymatic degradation by exoproteases that preferentially cleave near the N- or C-termini of peptide sequences (2).

A combination of computer modeling and experimental evaluations can deduce important conclusions about cyclic peptide conformations and their effects on stability compared to linear derivatives. A study by Bogdanowich-Knipp et al. modeled the structures of cyclic and linear

RGD peptides with the sequence cyclo-(1, 6)-Ac-CRGDF-Pen-NH2 (Pen = 8,8- dimethylcysteine) and H2N-RGDF-OH, respectively (36). Their modeling suggested that a salt- bridge at the guanidinium of arginine and the carboxyl of aspartic acid within the disulfide- cyclized RGD could stabilize the labile aspartic acid and reduce the degradation of the cyclic peptide compared to the linear RGD. A second study by Bogdanowich-Knipp et al. tested these models experimentally (37). The cyclo-(1, 6)-Ac-CRGDF-Pen-NH2 peptide was synthesized and the solution stability of this sequence was compared to the linear H2N-RGDF-OH peptide while

13 in solutions ranging from pH 2 to 12. They found that the cyclic RGD peptide was 30-fold more stable at pH 7 compared to linear RGD and was also generally more stable from pH 3 to 7. It should be noted though that no error ranges of the stability measurements were presented in this report. Once the pH ranged from 8 to 12, the disulfide bond likely became reduced and facilitated the higher degradation rate of the cyclic RGD compared to the linear peptide. This experimental evidence supported the suggestion that the disulfide bond constrained the cyclic RGD analogue and could stabilize the labile aspartic acid residue from degradation at biological pH ranges.

Not all cyclization chemistries will provide peptide sequences with the same degree of stability. Somatostatin-14 (SRIF) is an endogenous disulfide-cyclized peptide with the sequence AG(CKNFFWKTFTSC) that plays a role in the regulation of several hormones, notably insulin, glucagon and growth hormone (38). There are comparative studies that investigated the hormone-regulating ability of closely related linear and cyclic SRIF derivatives, particularly by Diel et al. and Vécsei et al. Surprisingly, Diel et al. found that the linear SRIF analogue was better able to protect histamine-containing gastric mucosal mast cells from ethanol-induced erosion in rats (39). Interestingly, Vécsei et al. also found that the linear SRIF derivative was more able to treat electroshock-induced amnesia in rats (40). It is important to note though, that early investigations of SRIF in the 1970‘s and 1980‘s significantly contributed to the advancement of cyclic peptides for medicinal applications. In addition to hormone-regulating investigations, SRIF and its analogues (most notably octreotide, f(CFwKTC)T (41)) were subsequently tested for the treatment of hormone-secreting tumors. A review of the early anti-cancer applications of SRIF was provided by Evers et al. (42) and the more recent applications of octreotide have been reviewed by Anthony and Freda (43). Though the disulfide-cyclized SRIF and its subsequent derivatives often displayed the desired biological activities, Besser et al. found that amide cyclization could create SRIF derivatives that were more stable than the original SRIF peptide (44). They created nine derivatives of the somatostatin receptor-binding octapeptide with the sequence fFYwKVFT-NH2 and cyclized these sequences by amide coupling. The derivatives were all cyclized at the D-phenylalanine and valine residues, but they differed in terms of the orientation of the amide bond and the use of extended linkers. The in vitro somatostatin-14 125 11 ([ I]Y -SRIF) displacement studies with BON-1 cells showed that the IC50‘s of these cyclic peptides were moderate and ranged from 10 µM to 0.1 µM. More importantly, the amide- cyclized peptide J1738 generated in this study was stable for 15 hours in homogenized rat liver

14 while the disulfide-cyclized SRIF standard degraded within 1 hour. This showed that the amide- driven cyclization was responsible for improving enzymatic stability and suggested that the disulfide bridge in the native SRIF is prone to enzymatic reduction.

In some cases, both the biological activity and the stability of peptides can be improved by cyclization. Pakkala et al. studied the activity and stability of linear and cyclic human glandular kallikreain (KLK2)-targeting peptides with the sequences ARRPAPAPG (KLK2a) and GAARFKVWWAAG (KLK2b) (45). The KLK2 protein is a highly prostate-specific serine protease that can be inhibited to mediate the metastasis of prostate cancer. They found that disulfide-cyclized derivatives of the KLK2b sequence obtained enhanced KLK2 inhibition compared to the linear sequence while the cyclic KLK2a peptides were all inactive. They created a series of head-to-tail cyclized KLK2b peptides and found that the inhibitory activity of these peptides was also greater compared to the linear derivative. Next, they tested the stability of the linear, disulfide-cyclized and the head-to-tail cyclized KLK2b derivative in trypsin and human plasma. These stability studies found that 57 % of the head-to-tail cyclized KLK2b ht1 peptide ARFKVWWN remained stable after a 4 hour treatment with trypsin while the disulfide-cyclized GCAARFKVWWACG peptide and the linear ARFKVWWG and Ac-ARFKVWWGG peptides each completely degraded after 30 minutes. They continued to show that the KLK2b ht1 peptide was completely stable in human plasma after 24 hours while the other three peptides consistently degraded within 30 minutes. This study again showed that disulfide cyclization might improve receptor binding affinity but that enzymatic reduction can limit their use in vivo. Therefore, amide coupling may instead be used in select cases to improve the biological activity and stability of linear peptides.

Unconjugated peptides as anti-cancer therapeutics

Linear and cyclic peptides can elicit biological responses that can result in the treatment of cancer and other diseases. Several groups have reviewed the topic of bioactive peptides (2, 46- 51). It was found that the key binding regions of endogenous proteins could be synthesized to create relatively low molecular weight targeting peptides that mimic the structure and activity of native proteins. These targeting peptides can act as antagonists due to competition with endogenous proteins and can prevent downstream cellular signalling in order to reduce the propagation of cancer growth. The synthetic linear P14 peptide (DFPQIMRIKPHQGQHIGE) and the cyclic CBO-P11 (cyclo-VEGI) peptide (DFPQIMRIKPHQGQHIGE) derived from

VEGF79-93 were investigated by Zilberberg et al. to determine if cyclization could increase the binding affinity of this sequence to the kinase receptor domain of VEGF receptors (52). 125 Comparative studies showed that cyclo-VEGI successfully displaced I-VEGF165 from CHO-

VEGFR2, CHO-VEGFR1, and BAE cells in a dose-dependent manner with IC50‘s of 1.3 µM, 0.7 µM, and 12 µM, respectively, while P14 did not. The cyclo-VEGI peptide abolished the

VEGF165-induced migration of BAE cells in a dose-dependent manner while the linear peptide had no effect on the stimulated cell migration. The cyclo-VEGI peptide also displayed antiangiogenic activity on chick choriallantioc membrane (CAM) growth while treatment with P14 resulted in a similar expression of microvessels as treatments with VEGF165. This showed that the cyclic analogue cyclo-VEGI successfully inhibited VEGF activity and that this property was the result of peptide cyclization.

The therapeutic activity of peptide sequences can be improved by cyclization. Haier et al. induced a colon carcinoma tumor model in rats using 1,2-dimethylhydrazin and compared linear and cyclic RGD-containing peptides to determine their relative anti-cancer potencies (53). Tumors were treated with the cyclic RGDfV, a cyclic inactive control peptide (EMD 135981) or the linear RGDS peptide to investigate the therapeutic activity of these RGD derivatives. They found that rats treated with the cyclic RGDfV peptide had an average of 7 tumors. In comparison, treatments with either the cyclic control peptide or the linear RGDS peptide resulted in the presence of an average of 15 and 18 tumors, respectively. This study demonstrated that the cyclic RGDfV peptide can effectively treat tumors in vivo. Though the anti-cancer activity of the cyclic RGDfV peptide was greater than that of the linear RGDS peptide, the linear peptide contained different residues at the C-terminus. Therefore, it is not clear if cyclization was the sole reason why the RGD-motif of the cyclic peptide exhibited an improved activity compared to the linear peptide.

The RGDFV-sequence has been identified as having great potential for treating human cancers that express αVβ3 and αVβ5 integrin receptors. The peptide cilengitide (EMD 121974, Cyclo-

RGDf(N-Me)V, Merck KGaA, Darmstadt, Germany) has IC50‘s of 1 nM and 140 nM for these integrin receptors, respectively, using vitronectin competition assays and was investigated in a Phase I and pharmacokinetic clinical trial for patients with advanced solid tumors (54). This study showed that a twice-weekly administration of cilengitide was safe and had a predictable clearance profile. A case study later showed that combined cilengitide and gemcitabine therapy was able to inhibit rapid tumor growth of a patient with highly vascularized head and neck

16 cancer (55). A recent Phase II clinical trial report by Gilbert et al. suggested that cilengitide alone has modest activity for treating recurring glioblastoma but that it may be used in a combination therapy regime for newly diagnosed glioblastoma (56). These clinical trial studies are promising and show an example of a cyclic peptide with the potential to successfully treat human patients with various types of cancer.

Cancer-targeting peptide conjugates

The conjugation of pharmacokinetic-modifying substituents can improve the properties of cancer-targeting peptides. For example, Haubner et al. created cyclic RGD-carbohydrate conjugates to determine if glycosylation could improve the pharmacokinetic and biodistribution properties of cyclic RGD peptides using either melanoma-bearing or osteosarcoma-bearing mouse models (57). During their investigations, they synthesized the glycosylated cyclic c(RGDyK(SAA)) peptide (GP1) to determine if the sugar substituent (SAA) or cyclization of the RGD-motif could improve the binding affinity of the linear GRGDSPK peptide to various immobilized integrins. They found that the cyclic GP1 peptide had a 30-fold lower IC50 for the

αVβ3 integrin and a 2.5-fold lower IC50 for the αVβ5 integrin compared to the linear standard.

However, the IC50 of the linear peptide was 3-fold lower for αIIbβ3 compared to GP1. A comparison with the cyclic RGDfV peptide showed that this non-glycosylated peptide had a 3- fold lower IC50 for αVβ3, a 4-fold lower IC50 for αVβ5 and a 3-fold lower IC50 for αIIbβ3 compared to GP1. Though the sequences were not identical, this suggested that cyclization improved the receptor binding of the linear derivative for αVβ5 and αVβ3 but that glycosylation does not improve the in vitro integrin receptor binding properties of all cyclic peptide sequences to the same extent. It should be noted though that this report did not present any error ranges for the binding affinity measurements. So it is difficult to conclude that statistical significance of differences in the aforementioned binding affinity values. They continued to radiolabel the GP1 peptide to create GP2 (c(RGD(125I)yK(SAA))) and compared in vivo biodistribution and elimination with the cyclic non-glycosylated P2 peptide (c(RGD(125I)yV)) peptide. These studies showed that peptide cyclization improved the binding affinity of the RGD-motif to αVβ3 and αVβ5 integrin receptors compared to a linear standard and that glycosylation contributed to improved tumor retention and allowed faster elimination of the radiotracer from the liver and kidneys.

The conjugation of sugar moieties or contrast agents can have unexpected effects on the receptor binding or biodistribution of linear and cyclic peptides. Interestingly, Achilefu et al. discovered

17 that the conjugation of the near-IR dye cypate (Cyp) was responsible for creating a high affinity conjugate with the previously inactive linear GRDSKP (Cyp-GRD) peptide (58). During these studies, they also compared the cypate conjugates of the RGD-containing linear GRGDSPK

(Cyp-RGD) and cyclic RGDfV (Cyp-cyclo-RGD) peptides for targeting the αVβ3 integrin receptor of A549 tumors in mouse xenograft models. This investigation showed that the cyclic Cyp-cyclo-RGD conjugate had a 3-fold higher in vivo tumor uptake compared to the linear Cyp- RGD conjugate. Surprisingly, the linear Cyp-GRD had a ~2-fold higher tumor uptake compared to the Cyp-cyclo-RGD and had lower accumulation in healthy tissues. These calculated values appeared to be statistically significant from the data that was presented graphically. However, no details were presented for the statistical methods employed for their comparisons. This comparison showed that the improved tumor uptake of the Cyp-GRD peptide conjugate was due to the rearrangement of glycine from the RGD-motif. This study also suggested that the favorable biodistribution of Cyp-cyclo-RGD compared to Cyp-RGD was due to peptide cyclization.

Peptides can be radiolabelled with radioisotopes to create radiopharmaceuticals capable of imaging and treating cancer and other diseases. This topic has been reviewed by several groups (59-61). Radiolabelling studies have compared linear and cyclic peptides to investigate the effects cyclization has on radiotracer biodistribution. Fani et al. compared the biodistribution of the 99mTc-radiolabeled linear and cyclic RGDfKH sequence using MDA-MB 435 breast cancer mouse xenograft models (62). They found that the tumor uptake of cyclic RGDfKH- 99m Tc(H2O)(CO)3 (3.74 ± 1.51 % ID/g) was 4-fold higher compared to the linear analogue (0.91 ± 0.08 % ID/g) while the accumulation of these peptides were similar in healthy tissues (30 min post injection). Since radiolabelling efficiency and chelation stability (saline containing either 10 mM cysteine or 10 mM histidine) were similar, the improved tumor accumulation of the cyclic RGD peptide was likely due to the constraints imposed by cyclization. It was suggested that the in vivo biodistribution of this cyclic RGD peptide could further be improved using multimeric cyclic RGD peptides that could bind to multiple adjacent αVβ3 integrin receptors. The topic of multimeric cyclic RGD‘s has since been reviewed (63).

Radiolabelling studies have investigated peptide cyclization for targeting the α-melanocyte stimulating hormone (α-MSH). Chen et al. developed the rhenium-cyclized peptide ReCCMSH that contained three cysteines residues (64). The in vivo biodistribution of the 111In-labelled peptides (from 2 hours to 24 hours) showed that the B16/F1 murine melanoma tumor

18 accumulation of 111In-DOTA-ReCCMSH (DOTA = 1,4,7,10-tetraazacyclododecane-1,4,7-tris(t- butyl acetate)-10-acetate) was from 1.2- to 1.5-fold higher compared to the disulfide-cyclized 111In-DOTA-CMSH peptide while the kidney accumulation of 111In-DOTA-ReCCMSH was from 2.6- to 3.9-fold lower compared to 111In-DOTA-CMSH. These results were especially significant since the linear derivative 111In-DOTA-CCMSH had a lower tumor accumulation and higher healthy tissue accumulation compared to the disulfide cyclized 111In-DOTA-CMSH peptide. This report presented extensive statistical comparisons to support their conclusions about the respective uptakes of these compounds in tumors and in healthy tissues. Though both cyclic α-MSH derivatives performed desirably, these results show that not all cyclization methods are able to equally improve the biodistribution of radiotracers in vivo.

Cyclic peptides have been used to create nanoparticles capable of actively targeting cancer biomarkers. Several groups have reviewed this topic (65-67). In particular, Negussie et al. created lysolipid-containing temperature sensitive liposomes (LTSLs) which used the NGR- motif to target CD13/aminopeptidase N that is selectively overexpressed in tumor vasculature (68). They synthesized the linear KNGRG peptide and used on-resin amide cyclization to create the cyclic KNGRE peptide (cKNGRE). These peptides were conjugated to the fluorescent dye Oregon Green (OG) and subsequently to the LTSLs in order to investigate the effect cyclization had on binding to HT-1080 (CD13+) and MCF7 (CD13−) cancer cells. Their results showed that the cKNGRE-OG (EC50 = 61.0 μM) had a 3.6-fold better binding to HT-1080 cells compared to the linear KNGRG-OG peptide (EC50 = 219.9 μM) and this comparison was illustrated using epifluorescence microscopy (Figure 1.1.5). Due to the multivalency effect, the binding affinity of each peptide-OG conjugate increased 10-fold when they we conjugated to LTSLs where the cyclic NRG-OG peptide (EC50 = 6.2 μM) maintained a higher binding affinity to HT-1080 cells compared to the linear analogue (EC50 = 21.5 μM). It should be noted though that this report did not present error ranges or details of the statistical methods that were employed during their data analysis. Yet, these studies suggested that cyclization of the KNGRE-motif was responsible for the possibly improved EC50‘s compared to the linear KNGRG sequence. It is also possible that the glutamic acid residue could have contributed to improving the binding of the cKNGRE derivatives. They continued to show that the cKNGRE-OG-LTSLs could release doxorubicin upon heating as a proof of concept experiment to show its potential as a cancer therapeutic delivery system.

Cell penetrating peptides

Cyclization can create peptides with the ability to penetrate tumors in order to enhance the potency of anti-cancer drugs. The field of cell-penetrating peptides has been reviewed by several groups (69-72). A recent study by Sugahara et al. described the discovery of the cyclic peptide with the sequence CRGDK/RGPD/EC, labeled iRGD, which was shown to have these properties (73). When iRGD was injected along with non-conjugated Nab-paclitaxel (ABX), the tumor uptake of ABX was increased by an order of magnitude within orthotopic BT474 and 22Rv1 xenograft mouse models compared to ABX alone. This resulted in a significant reduction in tumor size and weight. Similar results were found using iRGD in combination with free doxorubicin and with liposomes loaded with doxorubicin using orthotopic 22Rv1 mouse models. The cyclic iRGD was also co-administered with a clinical dose of trastuzumab and resulted in the eradication of tumors in orthotopic BT474 mouse models after 24 days. This report suggested that the mechanism underlying the observed enhancement of anti-cancer drug activity was due to integrin binding, proteolytic cleavage to CRGDK/R, binding to neuropilin-1 due to the C-end Rule motif (R/KXXR/K) (X = interchangeable residue) and cell internalization. Though a linear analogue was not compared to iRGD, this report is the first to show a novel cyclic peptide with

20 the versatile ability to enhance the therapeutic efficacy of several anti-cancer drugs by cell penetration.

Peptide cyclization can potentially improve the cell-penetrating ability of linear sequences. Mandal et al. synthesized 11 cyclic peptides which contained either R, K, E, F or W residues (74). These residues created peptides that contained charged sidechains which were expected to facilitate the nuclear transport of fluorescein labels and to aid the delivery of the anti-cancer drug doxorubicin by enhancing cell penetration. Among their results, the comparative fluorescence- activated cell sorting (FACS) study showed that the fluorescein-labeled cyclic F-[W4R3K] conjugate had an over 2-fold higher uptake in CCRF-CEM human leukemia cells compared to the linear F-KR3W4 peptide. These calculated uptake values were likely significant due to the non-overlapping error ranges that were presented graphically. However, no clear details of the statistical analysis were presented in this report. Since the charges of these two analogues were the same, the possibly improved cell penetrating property of the cyclic peptide may be attributed to the constrained geometry obtained by cyclization. However, peptide cyclization does not guarantee an improvement in cell penetration compared to linear analogues. This will be addressed later in this review.

Limitations of peptide cyclization

Cyclic peptides can potentially obtain desirable constrained geometries that are responsible for increasing their binding affinity, specificity or stability compared to their linear counterparts. However, it is important to note that cyclization doesn‘t necessarily lead to improvements to all or sometimes any of these properties. Linear peptides can contain sequences that can support rigid structures without the need for cyclization. Colombo et al. synthesized the disulfide- cyclized CNGRC and linear GNGRG peptides, which were subsequently conjugated to the tumor necrosis factor (TNF) for anti-CD13 targeted melanoma therapy (75). The anti-tumor therapeutic properties of these conjugates were tested in vivo using B16/F1 murine melanoma tumors. They suggested that the anti-tumor activity of the cyclic CNGRC-TNF conjugate was over 10-fold higher compared to the linear analogue. This suggestion was made since a 10-fold higher dose of the linear GNGRG-TNF (0.1 ng) conjugate was required to reduce tumor volume to the same size as using 0.01 ng of the cyclic CNGRC-TNF conjugate. However, the biological activities of both conjugates were similar after 14 days using 10 ng of either of these agents. Structural characterization of these two analogues surprisingly showed that the uncyclized CNGRC

21 sequence peptide contained a β-turn while the disulfide-cyclized sequence contained a bent configuration. They concluded that the linear NGR-motif itself may induce a rigid conformation to explain why the linear GNGRG sequence still showed desirably anti-tumor activity.

Peptide cyclization can improve biological activity of linear analogues but not in all cases. Mizejewski et al. extensively studied the estrogen-dependent anti-cancer activity of several variations of the linear and cyclic derivatives of human α-fetoprotein (HAFP) and compared their ability to inhibit cancer growth in several cell lines. Among these studies, they developed the linear and cyclic 34-amino acid peptides (P149) taken from a fragment of the third loop of HAFP. They reported that the linear P149 peptide could inhibit MCF-7 breast cancer cell growth while the disulfide-cyclized analogue was not as effective (76). In addition, they found that the linear 149c peptide (EMTPVNPG) fragment of P149 inhibited the growth of MCF-7 tumors to a similar extent (>75 %) as the cyclic analogue (>80 %) (77). Similarly, Mesfin et al. created truncated linear (EMTOVNOG) and cyclic (cyclo-(EMTOVNOGQ)) derivatives of P149 and found that they provided a similar anti-estrotrophic tumor growth inhibition of MCF-7 tumors (78). These comparative studies showed that these cyclic derivatives of HAFP fragments generally did not show improved growth inhibitions of MCF-7 tumors compared to the linear analogues.

Though particular native peptides may exist in cyclic conformations, linear derivatives of these native cyclic peptides may display a superior binding affinity to targeted receptors compared to their cyclic counterparts. As previously described, the endogenous disulfide-cyclized somatostatin-14 (SRIF) and its analogues were investigated for various endocrine and cancer- related applications. In terms of anti-cancer therapy, the most clinically-relevant binding target of

SRIF is the somatostatin receptor subtype 2 (SST2) since this receptor subtype is responsible for regulating several synthetic and secretory processes within human tumors (79). Interestingly,

Baumach et al. showed that a linear derivative of SRIF displayed superior SST2 binding compared to its cyclic counterpart (80). They initially screened a synthetic combinatorial linear hexapeptide library to discover a potent SRIF-mimicking peptide with a high binding affinity to

SST2. After library screening, they discovered the promising N-acetylated and C-amidated Ac- hfirwf-NH2 peptide. Among their studies, they compared the binding affinity of this linear peptide to the amide-cyclized hfirwf peptide and the disulfide-cyclized ChfirwfC peptide. Their competitive binding assay investigated the binding affinity of each peptide to the SST2 by using the 125I-labeled 14-amino acid somatostatin (125I-S-14) as the competitor. This assay showed that

22 the disulfide-cyclized ChfirwfC peptide surprisingly had a 12-fold lower binding affinity (Ki =

1.95 µM) to SST2 compared to the lead linear peptide (Ki = 0.16 µM) and that the amide- cyclized analogue displayed a similar binding affinity (Ki = 0.14 µM) to the same linear peptide.

Their report thoroughly investigated the in vivo biological activity of the linear Ac-hfirwf-NH2, but the lack of improved activity observed with the disulfide-cyclized derivative was not discussed.

In contrast to the previous examples, peptide cyclization has been shown to reduce the degree of cell membrane penetration. Kwon et al. compared the cell permeability of nine linear and nine cyclic peptide derivatives with the sequence E-Dpr-Dpr-Dpr (Dpr = 2,3-diaminopropionic acid) which were conjugated to the dexamethasone-derived steroid SDex (81). These conjugates were incubated with HeLa cells that expressed high levels of an artificial transactivator protein complex comprised of the Gal4 DNA-binding domain, the glucocorticoid receptor ligand binding domain and the VP16 transactivation domain (Gal4 DBD-GR LBD-VP16). This intricate protein complex was created so that the desired binding of SDex to Gal4 DBD-GR LBD-VP16 resulted in luciferase expression. After correcting for IC50 variations of the peptide sequences, the luciferase signals indicated the relative cell penetrating ability of the linear and cyclic peptides. These results were presented as the ―relative permeability factor.‖ Interestingly, they suggested that the linear peptides were 2- to 6-fold more permeable to cells compared to their corresponding cyclic analogues. This was an intriguing result since the sequences compared were the same and the N- and C-termini of the linear analogues were capped to remove charges. It should be noted though that no error ranges or details of statistical comparisons were presented in this report. A similar observation was noted by Fischer et al. while conducting ligand optimization studies of biotinylated penetratin derivatives for HaCaT or A549 cell internalization (82). Among their results, they found that the linear biotinylated 19-amino acid peptide was internalized to a greater extent compared to its cyclic counterpart that contained the same sequence. Since this linear sequence contained cysteine residues near the N- and C-termini, oxidation may have occurred under physiological conditions to prevent it from becoming internalized better than the original linear 17-amino acid penetratin peptide. It was not clear why these linear peptides were more permeable. It is, however, possible that the lower cell penetration of particular cyclic peptides could be due to reductions in enthalpy which may be required for stronger membrane interactions (83).

23 Unique aspects of peptide cyclization

Peptide cyclization allows the use of unique chemistries that can generate conformations not available to linear sequences. Giblin et al. engineered cyclic analogues of the α-melanocyte stimulating hormone (α-MSH, Ac-SYSMEHFRWGKPV-NH2) whereby cyclization involved forming intramolecular rhenium or technetium complexes (84). They synthesized the rhenium- 4,10 7 3,4,10 7 cyclized peptides ReMSH (Re–(C , f )–α-MSH4-13) and ReCCMSH (Re–(C , f )–α-MSH3- 3,4,10 7 13) as well as the disulfide-cyclized CCMSH (C , f )–α-MSH3-13) peptide. The binding 4,10 7 affinity of these peptides were compared to the cyclic APOMSH ((C , f )–α-MSH4-13) derivative of α-MSH. They found that 99mTc and 188Re gave the cyclic CCMSH sequence a greater stability in solutions of phosphate and cysteine compared to their cyclic counterparts with only two cysteine residues. Though ReCCMSH had a better binding affinity (Ki = 2.9 nM) compared to ReMSH (Ki = 66 nM) and CCMSH (Ki = 7.6 nM), the previously reported

APOMSH sequence out-performed all these cyclic compounds (Ki = 0.68 nM). The CCMSH sequence was subsequently radiolabelled with 99mTc (99mTcCCMSH) to display the in vivo tumor-targeting ability of the high-affinity cyclic CCMSH peptide using a γ-camera (Figure 1.1.6). This study showed that rhenium-cyclization of this α-MSH derivative required three cysteine residues to create a geometry that gave a low nanomolar Ki. It should be noted, however, that no error ranges or details of statistical comparisons were presented in this report. Though metal-induced cyclization did not improve the receptor binding property of the cyclic APOMSH sequence, the incorporation of a radiometal during cyclization makes this strategy intriguing for cancer radioimaging and radiotherapy.

Cyclic peptides can have geometries that allow preferentially binding to multiple binding sites of targeted receptors. Colangelo et al. created synthetic cyclic peptides to target loop 1 and loop 4 of the nerve growth factor protein (NGF) by mimicking the geometry of either one, both or the same loop twice to discover new sequences which can stimulate tyrosine kinase receptor A (TrkA) activation (85). This activation would lead to a cascade of signals for the differentiation and survival of NGF responsive neurons (86-88). They found that the bicyclic L1L4, containing the desired mimics of both loops 1 and 4, was able to stimulate the differentiation of the dorsal root ganglia of chick embryos similarly to using the native mNGF protein. The bicyclic L1L4 could also stimulate the differentiation of PC12 rat adrenal medulla cells better than derivatives

25 containing sequences that only mimicked the sequence of one of the targeted loops. The cyclic L1L4 peptide also selectively activated the desired TrkA receptor and not TrkB. This may explain how it improved pain sensitivity and reaction times after chronic sciatic constriction injury (CCI) almost as well as the dose of rat recombinant beta-NGF. They concluded that both loop mimics of the native NGF were required for the desired activity and selectivity. These finding can be useful to cancer research since NGF has been investigated as a potential target for anti-cancer therapeutics (89).

Peptide cyclization can be used to deliberately reduce the biological activity of peptide sequences. Zhong et al. investigated the use of a cyclic 25-amino acid (cycL-25) derivative (83) of the linear 18-amino acid membrane lytic peptide based on the α-helical peptide that was previously developed by Steiner et al. (L-18, LRLALKLALKALKAALKL) (90). The cycL-25 derivative included a membrane type 1-matrix metalloproteinase (MT1-MMP) cleavage site that was expected to cleave upon incubating the peptide with MMP-overexpressing cancer cell lines, thus restoring the activity of the linear membrane lytic sequence. They found that though the linear L-18 peptide was able to reduce the viability of both MDA-MB-435 (high MT1-MMP expression) and MCF-7 (low MT1-MMP expression) better compared to cycL-25, the cyclic peptide was selectively more toxic to the MDA-MB-435 cell line. In addition, cycL-25 displayed lower undesirable hemolytic activity compared to the linear L-18 analogue due its constrained structure. Though the desired anti-cancer toxicity of the cyclic peptide was not high as the linear analogue, cyclization was able to reduce undesirable hemolytic activity to improve the specificity of the membrane lytic sequence. This strategy of selective activation could be expanded to create novel molecular beacons. For example, it might be possible to create cyclic cancer-specific peptide-based molecular Förster resonance energy transfer (FRET) beacons by modifying cyclic cancer-targeting peptides to contain cleavable regions, a fluorophore and a fluorescence quencher. This method could potentially combine the improved binding affinity of cyclic peptides with the selective activation offered by enzymes that are overexpressed by cancers for efficient cancer imaging or therapy. The topic of peptide-based molecular beacons has been reviewed by Liu et al. (91).

Peptides have the potential to bind to selected targets in vivo while not interfering with the immune system to the same extent as their endogenous protein counterparts. Blank et al. synthesized a cyclic chimeric peptide mimetic of the human interferon α2b (IFN-α2b) peptide (122-137GG30-35) and compared the apoptosis-inducing activity of this cyclic peptide to the

26 native protein IFN-α2b (92). The cyclic peptide was previously shown to inhibit 40% of WISH cell growth at a 6 µM concentration while both the linear chimeric peptide and the IFN122–139-α2b peptide only inhibited 30% growth and the IFN29–35-α2b peptide gave 10% inhibition at the same concentration (93). The cyclic peptide appeared to have significantly better binding activity due to the non-overlapping error ranges that were presented graphically. However, there was no mention of statistical methods employed for assessing statistical significance of the presented values. Both the native IFN-α2b protein and the cyclic chimeric peptide induced apoptosis through activation of caspases 8 and 9 and the release of cytochrome C. However, the native IFN-α2b protein induced cell cycle arrest by increasing the distribution of cells in S phase while the cyclic chimeric peptide had no affect compared to control conditions. These results suggest that the cyclic chimera was able to retain its apoptotic activity due to its rigidity while avoiding induction of an immune response due to its small molecular weight relative to the endogenous IFN-α2b protein.

Cyclic peptides can be used to deliberately stimulate an immune response for vaccine research purposes. Witsch et al. created a series of cyclic 12-amino acid peptides and their linear analogues that mimicked human epidermal growth factor receptor 2 (erbb-2) gene protein epitopes that were specific for the L-26 and N-12 antibodies (94). Flow cytometry studies showed that the cyclic L-26-19 and cyclic L-26-24 peptides could displace the L-26 antibody from binding to the surface of BT474 breast cancer cells overexpressing the erbb-2 gene. Their linear analogues, in a dendritic octavalent multiple antigenic peptide (MAP) configuration, did not show this displacement. Similar results were found using the cyclic N-12-1 and cyclic N-12- 2 peptides compared to their linear MAP sequences. In addition to showing that cyclization was responsible for improving receptor binding, this study investigated the use of keyhole limpet hemocyanin (KLH) conjugates of the cyclic L-26-19 and cyclic L-26-24 peptides. The cyclic peptide-KLH conjugates elicited an immune response from mice and this study suggested that these cyclic ligands may be useful for creating vaccines against erbb-2 gene-overexpressing tumors.

Cyclization chemistry can be used to activate linear peptide-prodrug conjugates capable of controlled drug-release. Brady et al. bonded linear peptides to the anti-cancer drug vinblastine to create inactive prodrugs (95). Upon contact with prostate-specific antigen (PSA), four of the tested peptide sequences were cleaved at the glutamine-serine residues, which initiated the formation of an intramolecular bond that created a cyclic diketopiperizine (DKP). This cleavage

27 and subsequent cyclization was responsible for releasing the active drug cargo. The PSA-induced proteolytic cleavage was most efficient using the octapeptide 4-O-(Ac-Hyp-SS-Chg-QSSP)-dAc (Hyp = hydroxylproline, Chg = α-cyclohexylglycine and dAc = des-acetyl) that was conjugated to the 4-position of vinblastine. The topic of cyclization-activated prodrugs has since been reviewed by Gomes et al. (96).

Conclusions

This review has presented studies that have compared closely related linear and cyclic peptide analogues in an attempt to illustrate the advantages, the limitations and the unique properties of cyclic peptides. Cyclic peptides often have higher binding affinities compared to linear counterparts with the same or very similar sequences. Cyclic peptides contain fixed geometries that limit the number of conformations in solution. This property can tune the specificity of particular cyclic peptides compared to linear analogues or even other cyclic analogues, which are cyclized by different chemistries and/or at different sites. Peptide cyclization generally increases enzymatic stability compared to linear peptides, but not all cyclization chemistries can improve this stability to the same degree. There are several aspects of cyclic peptides that are unique from linear peptides and these properties allow cyclic peptides to be used for additional applications. In conclusion, peptide cyclization offers several advantages but cyclization is not guaranteed to induce all intended improvements to linear standards.

28 Section 1.2. Towards potent bacteriochlorin photosensitizers: insights into structural modifications with potential to enhance PDT-relevant photophysical properties

Summary

Bacteriochlorins are a class of porphyrinoids that hold great potential for generating reactive oxygen species (ROS) upon photoirradiation and are among the most promising classes of photosensitizers (PSs) for photodynamic therapy (PDT). Though many useful bacteriochlorins PSs have been reported, it is extremely difficult to design novel bacteriochlorins with optimal photophysical properties pertinent to PDT. Therefore, it would be of great importance to have a simplified set of guidelines to assist the fruitful design of new bacteriochlorin PSs. The goal of this review is to reiterate established methods of enhancing the PDT-relevant photophysical properties of bacteriochlorins, namely the quantum yields of intersystem crossing (ΦISC), excited triplet-states (ΦT), and singlet oxygen photogeneration (ΦΔ), and to introduce new potential strategies for enhancing these photophysical properties. The new suggested strategies herein are based on trends from literature reports where structural modifications can be linked to enhancements of desirable bacteriochlorin photoproperties. It is envisioned that future studies would attempt to validate the suggested strategies and use the refined parameters to create novel bacteriochlorin PSs that exhibit superior photoactivity compared to established derivatives.

Introduction

Photodynamic therapy (PDT) may be defined as the controlled generation of reactive oxygen species (ROS) by illuminated photosensitizers (PSs). At its core, this process requires three essential elements. These include i) a light source, ii) a molecule that absorbs light (the PS), and iii) either molecular oxygen (O2) or an alternate oxidizable substrate. While light and O2 are ubiquitous, useful PSs are very difficult to attain. One would desire several properties for an ideal PS. Previous reports have discussed these ever-improvable properties in great detail (97- 100). While each desirable property offers advantages, the arguably essential attributes of useful PSs include minimal dark toxicity (i.e. inactivity towards biological systems without illumination), high water-solubility, efficient ROS photogeneration, and absorbance of light at wavelengths within the PDT optical ‗window‘ (ex. 600 nm ~ 800 nm) (101, 102).

Bacteriochlorins are a class of porphyrinoids which hold immense potential for PDT. Due to their characteristic absorbance spectra, the most powerful attribute of monomeric

29 bacteriochlorins is their ability to absorb light within the near-infrared (NIR) region at ~700 nm – 800 nm. This particular wavelength range minimizes the absorbance of excitation light by haemoglobin in biological systems below this range and also minimizes the absorbance of light beyond this range by water (98, 103, 104). This ensures that bacteriochlorins can absorb light relatively deeply within tissue when compared to structurally similar classes of porphyrinoids (ex. porphyrins and chlorins) (102, 105-107).

The unmodified bacteriochlorin scaffold contains 14 distinct positions that can be subjected to structural modifications in attempts to tailor PDT-related properties (Figure 1.2.1). Briefly, the 2- , 3- (A-ring), 12-, and 13-positions (C-ring) correspond to the ‗y-axis‘ of bacteriochlorins while the 7-, 8- (B-ring), 17-, and 18-positions (D-ring) correspond to the ‗x-axis.‘ The 5-, 10-, 15-, and 20-positions represent the ‗meso’-positions that connect the four pyrrole rings within the bacteriochlorin macrocycle. Free base bacteriochlorins contain two protonated nitrogens at the 21- and 23-positions at the center of the trans-bacteriochlorin macrocycle, while metalo- bacteriochlorins are synthesized by inserting compatible metal atoms through the aforementioned 21- and 23- nitrogens. With 14 positions at which bacteriochlorins can be modified, the number of potential bacteriochlorins available for study is quite vast. Therefore, one must deeply consider the modifications that should be made to the bacteriochlorin scaffold in order to discover novel bacteriochlorins with optimal photosensitizing properties.

Figure 1.2.1. Simplified conventional labelling system for the trans-bacteriochlorin scaffold that denotes the four characteristic pyrrole rings and the carbon positions amenable to chemical modifications.

Much effort has been devoted by Lindsey, Holten, Tamiaki, Hamblin and associates (in particular) to the synthesis of novel bacteriochlorins that contain optical property-modulating

‗auxochromes‘ to tune the characteristic Qx (~500 nm – 600 nm) and Qy (~700 nm – 800 nm) absorbance bands. In respect to PDT, these studies provide useful insights for tailoring bacteriochlorins with long Qy (and possibly Qx) absorbance maxima for enhancing the depth of light penetration into tissues. The Qy and Qx absorbance band maxima are ultimately governed by the HOMO to LUMO+1 (also the HOMO to LUMO) and the HOMO-1 to LUMO energy gaps of bacteriochlorins, respectively (108-111). However, the particular moieties that have the potential to modify these aforementioned energies of bacteriochlorins as intended are not obvious. Through extensive analysis, it became apparent that the Qy and/or Qx absorbance peaks can be red-shifted by the inclusion of strong electron-withdrawing groups (EWGs) positioned along the y-axis or along the x-axis (112-119). However, several nuances exist when attempting to tailor the optical properties of bacteriochlorins by means of SAR-based strategies. For example, the same moiety can have varying or quite opposite effects on the absorbance spectra of bacteriochlorins depending on the modified position of the macrocycle (120-123).

While outlining simplified strategies of modulating the absorbance characteristics of bacteriochlorins is an achievable goal, it is far more challenging of a task to identify specific structural modifications that can reliably tune their photophysical properties. In terms of PDT, PSs should ideally undergo efficient intersystem crossing (ISC, inverting the spin direction of the

31 excited electron to the same direction as its unexcited pair), and subsequent excited triplet-state conversion (the spin multiplicity that results from ISC of PSs) to maximize the photogeneration 1 of singlet oxygen ( O2) and other transient cytotoxic ROS. While a few general modification strategies are highly likely to enhance the aforementioned photoproperties of bacteriochlorins, it is largely unknown how the plethora of potential structural modifications will affect these photoproperties. The modifications highlighted in this review include the known strategies of metal insertion and the use of the ‗heavy atom effect‘ (124-127), while the remaining suggested strategies have not yet been extensively proposed in past reports. To summarize, these latter suggestions include i) the inclusions of electron-donating groups (EDGs) at the meso 5-position, ii) the inclusion of phenyl moieties along the y-axis, iii) the avoidance of EWGs along the y-axis, iv) the potential advantage of bacteriochlorins with asymmetries across the x-axis, v) the avoidance of cationic moieties, and vi) the avoidance of particular fused aromatic rings. It is very difficult to clearly assign structure-activity relationships (SARs) in terms of bacteriochlorin structure and photophysical properties as they relate to PDT-activity. Therefore, it is not surprising that sparse examples exist in the literature to provide useful insights into these SARs. Yet, the more obvious relationships are presented herein to aid future efforts in their design of novel bacteriochlorins with favourable, or at least less hindered, photoproperties relevant to PDT applications. It should be noted though, that few reports clearly present the error ranges of photophysical measurements and statistical methods used to remark differences. Therefore, the likely error ranges of the cited values will be presented to clarify the potential significant differences of the presented values.

Metal insertion

Many metals are suitable for insertion within the bacteriochlorin macrocycle. Yet, not all metal atoms will afford the bacteriochlorin scaffold with photoproperties favourable for PDT. It is commonly noted that the excited triplet-state lifetimes of PSs should ideally be > 0.5 µsec (or longer) to allow suitable photosensitization through the excited triplet-state (106). However, paramagnetic metals (metals attracted to a magnetic field) generally reduce the excited triplet- state lifetimes of bacteriochlorins to degrees that are far too short to accommodate efficient ROS photogeneration (97, 106, 128, 129). While diamagnetic metals may also reduce the excited triplet-state lifetimes of free base bacteriochlorins, this seemingly deleterious effect is less pronounce compared to paramagnetic metals. In fact, diamagnetic metals can actually enhance

32 ROS photogeneration efficiency by improving the population of the excited triplet-state of bacteriochlorins and their structurally similar porphyrinoid counterparts (112, 128-131).

Numerous studies have investigated the effects of diamagnetic metal insertion into bacteriochlorins and have shown that these metalo-bacteriochlorins have photophysical properties well suited for PDT applications. Not all diamagnetic metals, however, will enhance these desirable photoproperties to the same extent. Therefore, the choice of the appropriate diamagnetic metal should be an important consideration when designing new metalo- bacteriochlorin PSs. For synthetic bacteriochlorins with geminal dimethyl modifications at the 8- and 18-positions, it is clear that zinc (Zn), indium (In) and palladium (Pd) were responsible for enhancing the ΦISC (Figure 1.2.2 A and C) and ΦT (Figure 1.2.2 B) of these metalo- bacteriochlorins (1-3, 5, 6, 8, and 9) compared to their respective free base precursors (4, 7, and

10) (132, 133). Palladium (11 and 14) and Zn (12 and 15) were also shown to enhance the ΦISC

(Figure 1.2.2 D) and ΦΔ (Figure 1.2.2 E) of natural bacteriochlorophyll a (13) and its free-base derivative (15) (134, 135). While these metals are each useful for creating bacteriochlorin PSs that have superior photosensitizing properties compared to their free base precursors, it is important to appreciate which metals are most useful for enhancing these photoproperties.

Figure 1.2.2. Enhancements of the ΦISC (A, C, and D), ΦT (B) and ΦΔ (E) of bacteriochlorins by insertion of the metals Zn(II), In(III) (A) and Pd(II).

For the reasons previously stated, bacteriochlorins show amazing potential for PDT applications. The greatest hindrance to bacteriochlorins, however, is their susceptibility to photo-oxidation at the B- and D-rings (Figure 1.2.1) (97, 104). When this oxidation occurs, the parent bacteriochlorin can form a C=C between the 7- and 8-carbon positions and will become converted to a chlorin. If this type of oxidation continues between the 17- and 18-carbon positions, the chlorin will be further transformed to a porphyrin (136, 137). Due to this inherent disadvantage, several researchers (namely Lindsey, Holten, Tamiaki, Hamblin and associates) have devised strategies of eliminating this particular oxidation pathway. The most brilliant of these strategies includes the use of geminal alkyl groups at the 8- and 18-positions that prevent C=C formation at the B- and D-rings, respectively. Once methods of creating these ‗stable‘ bacteriochlorins were optimized, their efforts focused on the generation of large libraries of bacteriochlorins to elucidate structural modifications that can tune the photophysical properties of these compounds. Among these libraries are bacteriochlorins that were subjected to metal

34 insertion. It should be noted that several of the values cited herein were calculated from transient absorption spectroscopy-based characterizations and that the cited ΦISC/ΦT values were noted to have error ranges of ± 0.09, at most (111). Therefore, differences within 0.18 may likely be significant. As shown in Figure 1.2.2, a free base bacteriochlorin containing para-tolyl groups at the 2- and 12-positions (4) (Figure 1.2.2 A) was metaled to contain Zn(II), In(III)-Cl, or Pd(II)

(132). Subsequent photophysical characterizations revealed that the ΦISC increased by ~50% (3, Zn(II); likely a significant improvement), almost doubled (2, In(III)-Cl), and reached near unity (1, Pd(II)) upon metal insertion compared to the free base precursor (4). This trend was similar to bacteriochlorins containing cyano groups at the 3- and 13-positions (Figure 1.2.2 B). Here, Zn(II) provided a ~50 % enhancement (likely a significant improvement) to the ΦT (6) compared to the free base precursor (7), while Pd(II) again increased the ΦT to near unity (5) (133). These results were nearly mirrored by insertion of Zn(II) (9) and Pd(II) (8) into the free base bacteriochlorin with 2,4,6-trimethylbenzene groups at the 2- and 12-positions and ethyl esters at the 3- and 13- positions (10) (Figure 1.2.2 C) (132).

Synthetic metalo-bacteriochlorins will likely have improved PDT-related photophysical properties where Pd(II) > In(III) > Zn(II), compared to their parent free base precursor. Fortunately, this trend also applies to natural bacteriochlorin analogues. As shown in Figure

1.2.2, Zn(II) and Pd(II) enhanced the ΦISC (Figure 1.2.2 D) and ΦΔ (Figure 1.2.2 E). Specifically, replacement of Mg(II) within the natural bacteriochlorophyll a (13) with Zn(II) (12) enhanced the ΦISC by 10% (possibly not a significant improvement), while Pd(II) (11) again enhanced the

ΦISC to near unity (Figure 1.2.2 D) (134). This pattern of enhancements matched that for the ΦΔ 2 3 of bacteriochlorophyll a derivatives modified with a 13 ketone and a 17 methyl ester (i.e. ΦΔ = 14 (Pd(II)) > 15 (Zn(II)) > 16 (2H)) (Figure 1.2.2 E) (135). Since the PDT-relevant photoproperties of both synthetic bacteriochlorins and natural bacteriochlorin derivatives have been enhanced using Zn(II) and Pd(II), future efforts can be reasonably confident that ROS photogeneration will increase in the order Pd(II) > Zn(II) > free base.

Many reports have suggested that the ΦISC, ΦT, and ΦΔ of porphyrinoids are enhanced by Pd(II) insertion due to the ‗heavy atom effect.‘ Since Zn(II) is not a ‗heavy‘ transition metal, proposed mechanisms by which Zn(II) can enhance PDT-relevant photoproperties are not as conclusive. Yet, several examples exist that support the suggestion that Zn(II) will improve the photosensitizing activity of free base bacteriochlorin precursors. Figure 1.2.3 (A) shows that

Zn(II) (17 and 3) enhanced the ΦISC of the ‗unmodified‘ free base bacteriochlorin (18) (by a

35 modest 8%; may not be a significant improvement) and of a free base with para-tolyl groups at the 2- and 12-positions (4) (by ~50 %; likely a significant improvement) (111, 132). This was similar to bacteriochlorins that contained heptyl groups at the 2- and 12-positions and ethyl esters at the 3- and 13-positions (Figure 1.2.3 B). Specifically, Zn(II) insertion afforded a 33% enhancement (may not be a clearly significant improvement) to the ΦISC (19 compared to 20) and was also responsible for the 43% enhancement (likely a significant improvement) seen for bacteriochlorins with the meso 5-methoxy group (21 compared to 22) (132). This trend was again seen for synthetic bacteriochlorins with ethyl esters at the 3- and 13-positions that contained either ethyl or ethyl ester groups at the 2- and 12-positions that were either modified with a meso 5-methoxy group or were left unmodified at this 5-position (Figure 1.2.3 C). Again, Zn(II) insertion enhanced the ΦISC of 23 by 29% (compared to 24; may not be a significant improvement) and of 25 by 333% (compared to 26; likely a significant improvement) (111). While exceptions to these trends may exist, one can be reasonably confident that Zn(II) insertion within the bacteriochlorin scaffold should improve the ΦISC, ΦT, or ΦΔ of free base precursors, albeit to unpredictable extents.

Figure 1.2.3. Enhancements of the ΦISC of bacteriochlorins by insertion of Zn(II). The heavy atom effect

The label coined the ‗heavy atom effect‘ has be reported by several groups (124, 125) and refers to the phenomenon whereby ‗heavy‘ metals and ‗heavy‘ halogens have been observed to enhance the ΦISC, ΦT, and ΦΔ of PSs. Several groups have made use of the heavy atom effect for synthesizing novel bacteriochlorins with powerful photosensitizing properties. While this has been commonly achieved using high molecular weight metals (ex. Pd, In, etc.), fewer examples exist whereby bacteriochlorins were modified with relatively high molecular weight halogens to

37 exploit the heavy atom effect. It should also be noted that the cited ΦΔ values herein where the 1 result of O2 phosphorescence characterizations that did not include error ranges associated with these calculated values or descriptions of statistical analyses for remarking differences of the reported values.

Pineiro, Periera, Silva, Dabrowski and associates have provided reports that explored the intricacies of the heavy atom effect as it applies to the use of halogens on bacteriochlorin scaffolds for enhancing the ΦΔ. Each of these reports involved the synthesis and photophysical characterization of bacteriochlorins modified with phenyl derivatives at all of the four meso (5-, 10, 15- and 20-) positions. The phenyl groups of these bacteriochlorins were further modified with chloride or fluoride at the ortho-positions and where either left otherwise unmodified or contained a sulfonic acid or a sulfonamide derivative at the meta-position of these phenyl groups. The clearest observation from these studies was that the bacteriochlorin modified with 2,6- dichlorophenyls (27) showed a 25% enhancement to the ΦΔ compared to the counterpart modified with 2,6-difluorophenyls (28) (Figure 1.2.4 A) (138, 139). This was similar to a report by Azenha et al. that showed that chloride at the ortho positions of meso phenyl moieties enhanced the ΦT of porphyrins (140). Amazingly, the meta-sulfonic acid moiety of 29 (that also contained 2,6-dichlorophenyl modifications) was shown to further enhance the ΦΔ by 29% compared to 27 (Figure 1.2.4 A and B) (138, 141). It is not clear how the sulfonic acid group afforded this enhancement. The more obvious possibilities include its electron-withdrawing potential and its position at the meta-position of the meso phenyl rings. While the importance of the position of this sulfonic acid group was not explored, the electron-withdrawing strength was altered by means of converting it to various sulfonamides. While these bacteriochlorins showed modest enhancements to the ΦΔ compared to 27, it was clear that the ethyl sulfonamide (30) and the heptyl sulfonamide (31) derivatives were both likely less capable of generating singlet oxygen upon photoirradiation compared to 29 (Figure 1.2.4 B) (141). Therefore, these studies showed that the ΦΔ of tetra meso-phenyl bacteriochlorins can possibly be improved using two chlorides at ortho-positions of the phenyl rings. In addition, these studies raise the possibility that the sulfonic acid group on tetra meso-phenyl bacteriochlorins may either independently improve the ΦΔ due to its electron-withdrawing strength at the meta-position or that it may somehow act synergistically with the two ortho chlorides by an unknown mechanism.

The noted use of chloride- and sulfonic acid-modified tetra meso-phenyls for possibly enhancing the ΦΔ of bacteriochlorins was further explored to discover potential nuances for these types of modifications. While two ortho-chlorides (27) and the meta-sulfonic acid group (29; likely a significant improvement) enhanced the ΦΔ (compared to 28), the aforementioned researchers attempted to determine if both chlorides were required to elicit the heavy atom effect and

39 whether the meta-sulfonic acid would improve the ΦΔ of both mono ortho chlorophenyl and di- ortho fluorophenyl derivatives (Figure 1.2.4 C) (138, 139, 141, 142). These studies showed that the mono ortho-chlorophenyl derivative (32) displayed a ΦΔ that was on par with its di-ortho chlorophenyl (27) counterpart. Surprisingly, the additional meta-sulfonic acid possibly reduced the ΦΔ of 33 compared to 32 by 29%. Similarly, but to a lesser extent, this meta-sulfonic acid group of 34 may have reduced the ΦΔ compared to 28 by 8%. These aforementioned studies suggested several fascinating potential trends. They suggested that ortho chloride-modified tetra meso phenyl bacteriochlorin derivatives may have displayed the ‗heavy atom effect‘ and that even one ortho chloride on each of the four meso-phenyls was responsible for possibly improving the ΦΔ. Also, examples were provided wherein the meta-sulfonic acid group serendipitously possibly enhanced the ΦΔ of the tetra meso di-ortho chlorophenyl bacteriochlorin while this same moiety at the same position hindered the ΦΔ in all other cases. While an explanation for this latter observation is not clear, future studies can still make use of these particular modifications for PDT applications.

The electron-donating meso 5-methoxy group

Electron-donating groups (EDGs) are substituents within molecules that hold a relative excess of electron density that may become delocalized to nearby regions of the molecule with relatively less electron density. The reverse is true for electron-withdrawing groups (EWGs). If one were to design novel bacteriochlorins with potent PDT activity, it would be of great significance to be able to simply rely on the use of EDGs or EWGs for enhancing PDT-relevant photophysical properties. Indeed patterns exist for the use of EDGs and EWGs at various positions of the bacteriochlorin scaffold for tuning key absorbance properties (i.e. the maxima for the Qx and Qy bands). However, it is far less clear if and how additional EDGs and EWGs may affect the ΦISC,

ΦT, and ΦΔ of bacteriochlorins.

After reviewing several reports by Lindsey, Holten, Tamiaki, Hamblin and associates, a pattern emerged. Examples were noted where the ΦISC or the ΦT of bacteriochlorins were possibly enhanced by the electron-donating methoxy moiety at the meso 5-position of various geminal dimethyl-modified (8- and 18-positions) bacteriochlorin derivatives. It should be noted though that several of the values cited herein were calculated from transient absorption spectroscopy- based characterizations and that the cited ΦISC/ΦT values were noted to have error ranges of ± 0.09, at most (111). Therefore, differences within 0.18 may likely be significant. This pattern

40 was apparent for bacteriochlorins containing ethyl groups at the 2- and 12-position that either had methyl esters or ethyl esters at the 3- and 13-positions (Figure 1.2.5 A). Analysis showed that the meso 5-methoxy group was responsible for enhancing the ΦISC of the methyl ester derivative (35) by 15% compared to its precursor (36) (in toluene; may not be a significant improvement) (143). This was identical to the analogous ethyl ester derivatives (37 compared to 24) (111). While this type of potential improvement was not as pronounced, the meso 5-methoxy group of 38 may have been responsible for the modest 11% improvement of the ΦISC (compared to 39 in DMF) for bacteriochlorin derivatives synthesized to have cationic N-methyl-piperidonyl moieties at the 8- and 18-positions (Figure 1.2.5 B) (143). However, this was likely not a significant improvement.

Figure 1.2.5. Enhancements of ΦISC (A and B) and ΦT (C) afforded by the electron-donating methoxy group at the meso 5-position.

42 This trend was shared by class of bacteriochlorins that contained N-modified 6-membered cycloimide rings fused to the macrocycle core through the 13- and 15-positions (Figure 1.2.5 C). These derivatives either contained an aryl group emanating from the nitrogen of the cycloimide (40 and 41) or had a maleimide functionality adjoined to the para-position of this aryl group by an amide linker (42 and 43) (144). Photophysical characterizations showed that the meso 5- methoxy group was responsible for improving the ΦT of 40 by 55% (compared to 41; likely a significant improvement). Similarly, the electron-donating 5-methoxy group of 42 enhanced the

ΦT of 43 by 59% (likely a significant improvement). It is not clear how the meso 5-methoxy group led to these likely significant enhancements. However, future studies can still consider using this modification to improve the photosensitizing properties of known bacteriochlorin PSs.

While several examples were provided that showed that the electron-donating meso 5-methoxy group led to the discovery of bacteriochlorins with enhanced PDT-relevant photophysical properties, instances were also found where this EDG likely had the reverse affect or had no affect (Figure 1.2.6). For bacteriochlorins with geminal dimethyl-modifications (8- and 18- positions), the meso 5-methoxy substituent of 44 reduced the ΦISC of 18 by 11% (111, 145). This hindrance was even more pronounced for similar analogues that contained 2- and 12-para-tolyl groups (4 and 45). Specifically, the ΦISC of 4 was reduced by 24% (for 45) due to the inclusion of the meso 5-methoxy group (111, 145). However, it should be noted that the possible error ranges (likely ± 0.09, at most (111)) associated with these calculated values cannot show that these particular cited values were significantly different. And again, it is not clear how this EDG reduced the ΦISC for these pairs of bacteriochlorins. What is clear, however, is that the PDT- relevant photophysical properties of several bacteriochlorins were potentially modulated by modifications at the meso 5-position. This is an important insight because it presents a potential site on the bacteriochlorin scaffold that can be subjected to modifications that can possibly tune photophysical properties. Future studies should continue to explore additional modifications at the meso 5-position to refine guidelines that can be used to create novel bacteriochlorin PSs with superior photosensitizing properties.

Figure 1.2.6. Noted exceptions where the electron-donating methoxy group at the meso 5- position hindered the ΦISC.

Modifications along the y-axis

The 2-, 3-, 12-, and 13-positions of the bacteriochlorin scaffold are sites that correspond to the ―y-axis.‖ Over the decades, synthetic methods have been developed to create purely synthetic bacteriochlorins that contain a wide variety of substituents at these positions. Even some natural bacteriochlorins (ex. bacteriochlorophyll a (13) and b (46)) can be easily modified along the y- axis, namely at the 3- and 13- positions. It is well known that modifications along the y-axis can affect the optical properties of bacteriochlorins. It is far less known if and/or how modifications at the 2-, 3-, 12-, and the 13-positions will affect the PDT-relevant photophysical properties of bacteriochlorins. Now that synthetic methods are available for creating virtually endless libraries of new bacteriochlorins, one only needs to consider which moieties are likely to enhance, or at least not hinder, the photophysical properties of bacteriochlorins intended for PDT applications.

While alkyl moieties can be subjected to several types of chemical modifications, the phenyl ring is debatably even more amenable to synthetic manipulation. Therefore, studies should carefully consider which specific phenyl derivatives should be introduced to the bacteriochlorin scaffold, particularly to positions along the y-axis. Yang et al. and Huang et al. created bacteriochlorins with phenyl derivatives at the 3- and 13-positions to tune optical and photophysical properties, and analysis of their results revealed a potentially useful insight (Figure 1.2.7 A). Compared to

44 the ‗unmodified‘ bacteriochlorin analogue (18) with a ΦISC of 0.62, the derivative containing a phenyl ring at the 3- and 13-positons (47) showed a 10% enhancement to the ΦISC (111). On the contrary, the bacteriochlorin with a 3,5-dihydroxyphenyl group at the 3- and 13-positions (48) experienced a 16% reduction to the ΦISC compared to 18 (146). It should be noted though that these cited values were calculated from transient absorption spectroscopy-based characterizations and likely have error ranges of ± 0.09, at most (111). Therefore, these particular values may not be significantly difference. Hence, it is not clear if the inclusion of an unmodified phenyl moiety at both the 3- and 13-positions can enhance the ΦISC or if phenyl derivatives with meta-EDGs may hinder the ΦISC.

46 While clear conclusions could not be drawn in regards to the effect of aryl groups at the y-axis and the PDT-relevant photophysical properties of bacteriochlorins using the aforementioned examples, the use of alkyl EWGs along the y-axis may significantly modulate particular properties. The bacteriochlorin 37 was synthesized by Yang et al. and Reddy et al. to contain an ethyl group at the 2- and 12-positions, ethyl esters at the 3- and 13-positions, and a methoxy group at the 5-position (Figure 1.2.7 B) (111, 143). Photophysical characterizations subsequently calculated a ΦISC (in toluene) of 0.63 for 37. When this result was compared to that of a similar derivative with ethyl ester groups at the 2- and 12-positions, it was found that the electron- withdrawing ester groups of 26 significantly reduced the ΦISC by 62% (compared to 37). Since other derivatives within their series were too dissimilar to reveal the effects of powerful EDGs at positions along the y-axis, clear trends could not be elucidated. However, it is obvious that, in this instance, EWGs at both the 2- and 12-positions of the bacteriochlorin scaffold severely hindered the ΦISC. Therefore, future studies should consider this and design new bacteriochlorins with the possibility in mind that alkyl EWGs at the y-axis may have deleterious consequences to their PDT-relevant photophysical properties.

The effects of alkyl EWGs at the 3- and 13-positions of bacteriochlorins were also investigated.

As previously suggested, the same alkyl EWG at both the 2- and 12-positions reduced the ΦISC. This was similar to the effects of the electron-drawing acetyl group positioned at the 3- and 13- positions along the y-axis of the bacteriochlorin scaffold (Figure 1.2.7 C). These acetyl groups of

50 were responsible for possibly reducing the ΦISC of 18 by 16% (not likely a significant difference) (111). This may add support for the suggestion that identical EWGs at either the 2- and 12-positions or the 3- and 13-positions may hinder the photosensitizing properties of bacteriochlorins. This insight also opens the possibility that alkyl EDGs placed along the y-axis of the bacteriochlorin scaffold may serve to enhance PDT-relevant photophysical properties. It is extremely interesting to note, however, that a single acetyl group at the 3-position of 49 enhanced the ΦISC of 18 by 29% (likely a significant enhancement) (111). Therefore, it is possible that disrupting the overall structural symmetry of bacteriochlorins across the x-axis (along the B- and D-rings) using an EWG at the y-axis can potentially improve the photosensitizing properties of bacteriochlorins. Future studies should take note of this observation and consider synthesizing novel bacteriochlorins with various non-identical groups at a single pyrrole (A- or C-ring) along the y-axis to further investigate strategies of improving PDT-relevant photophysical properties using either EWGs or EDGs or EWG-EDG pairs.

47 Cationic substituents along the x-axis and the y-axis

Introducing anionic (ex. carboxylates, sulfonates, etc.) and cationic (ex. tetra-substituted amines) moieties to the bacteriochlorin scaffold is relatively easy. However, the effect of different charged moieties on the photophysical properties of bacteriochlorins has not yet been thoroughly investigated. Therefore, knowing if and how anionic and/or cationic groups could modulate PDT-relevant photophysical properties would be a great asset for designing novel bacteriochlorins for PDT applications. Since few studies have focused on elucidating the effects of charged moieties on the photophysical properties of bacteriochlorins, this review can provide only examples of potential trends.

Many groups have synthesized broad libraries of bacteriochlorins with modifications at the y- axis and the x-axis. Among these compounds were derivatives with cationic substituents at either the 3- and 13-positions along the y-axis or at the 8- and 18-positions along the x-axis. When compared to the bacteriochlorin 18 with geminal dimethyl groups at the 8- and 18-positions, derivatives that contained additional cationic substituents at both the 3- and 13-positions showed reduced ΦISC‘s. Specifically, the two cationic trimethyl ammonium moieties of 51 reduced the

ΦISC of 18 by 17% (Figure 1.2.8 A) (111, 132, 146). It is possible that the additional amide across the 31-32 and the 131-132 positions of 51 may have contributed to this reduction of the

ΦISC. However, the ΦISC was further reduced by 12% (compared to 18) for a bacteriochlorin that contained similar modifications at the 3- and 13-positions which each contained two trimethyl ammonium groups (52) (146). Again, it must be noted that these cited values were calculated from transient absorption spectroscopy-based characterizations and the error ranges may be as high as ± 0.09 (111). Therefore, the cationic features of 51 and 52 may not have significantly affected the ΦISC of 18.

Figure 1.2.8. Noted trend wherein the ΦISC of bacteriochlorins was hindered by cationic moieties positioned at the y-axis (A) and the x-axis (B).

Another example was found wherein cationic moieties may have lowered the ΦISC of a bacteriochlorin. As previously stated, Reddy et al. have synthesized ‗stable‘ bacteriochlorins with geminal dimethyl groups at the 8- and 18-positions to eliminate the possibility that these bacteriochlorins would undergo photo-oxidation to form chlorin and porphyrin byproducts upon illumination. They also explored the use of cationic N-dimethyl piperidonyl rings at the 8- and 18-positions along the x-axis for making new ‗stable‘ bacteriochlorins. Though both strategies are useful for protecting the bacteriochlorin macrocycle from this type of photoreaction, the charged tetra-substituted ammonium portion of the N-dimethyl piperidonyl groups of 38 possibly had a detrimental effect on the ΦISC (Figure 1.2.8 B). In DMF, the ΦISC of the geminal dimethyl- modified bacteriochlorin 35 was 21% greater compared to its counterpart with cationic N- dimethyl piperidonyl groups (38) at the same positions (111, 143). Again, due to the error associated with these measurements, it is difficult to claim that these cited values were significantly different. Therefore, it is not obvious why (or if) cationic moieties introduced at the y-axis and the x-axis of the bacteriochlorin scaffold could hinder the photosensitizing potential of the noted bacteriochlorins. Previous reports have, however, shown that the photoactivity of porphyrinoids increased with the incorporation of sulfonic acid moieties to the macrocycle scaffold (97, 98). This lends support to the suggestion that future studies should continue to investigate potential trends as they relate to ΦISC and various peripheral anionic/cationic substituents associated with bacteriochlorins.

Additional fused rings

The characteristic Qy absorbance band of bacteriochlorins can be red-shifted further into the NIR region by aromatic rings fused to the B- and D-rings (147). While optical properties can be modulated by fusing additional aryl rings to the bacteriochlorin scaffold, little is known about their effects on the photophysical properties of bacteriochlorins. Samankumara et al., however, have made a fascinating observation. This group discovered methods of synthesizing a new class of oxazolo-bacteriochlorins that contain six-membered oxazolo-rings that replaced the D- and/or B-pyrrole rings of typical bacteriochlorins (148). In addition to these distinguishing modifications, these compounds either contained peripheral aryl rings tethered to the four meso positions (53 and 55) or had phenyl rings fused to the core macrocycle at the B- and D-ring positions (54) or only at the B-ring position (56) (Figure 1.2.9). Upon characterizing the photophysical properties of these unique oxazolo-bacteriochlorins, it was clear that the derivatives with free aryl rings at the meso positions, 53 and 55, displayed 129% and 57%

50 greater ΦISC‘s compared to their respective counterparts, 54 and 56, synthesized by fusing meso aryl rings to the macrocycle core. This report stated that the error associated with their ΦISC characterizations were ±3%. Therefore, the aforementioned cited values can be deemed significant. Interestingly, ring-fusion had the opposite effect on the ΦF. Fusing meso aryl rings to the oxazolo-rings actually significantly improved the ΦF of 54 (ΦF = 0.18) and 56 (ΦF = 0.08) compared to 53 (ΦF = < 0.01) and 55 (ΦF = 0.03), respectively (148). This study showed that, when designing novel bacteriochlorins for PDT applications, one should consider refraining from increasing the overall rigidity of the bacteriochlorin scaffold. If, however, new bacteriochlorins are intended for fluorescence imaging applications, the aforementioned study illustrates how the generally poor ΦF of typical bacteriochlorins can be enhanced.

This aforementioned study clearly presented examples where additional rings that were fused to the central bacteriochlorin scaffold reduced the ΦISC. This remarkable finding lends support for our chosen research direction of investigating the effects exocyclic E-ring-opening had on the photophysical properties of naturally-derived bacteriochlorin photosensitizers. These investigations will be presented in Chapter 3 of this thesis.

Figure 1.2.9. Examples where the ΦISC was hindered by fusing aryl groups along the x-axis of bacteriochlorins.

Photosensitizers capable of type 1 PDT

Photosensitizers can potentially undergo several distinct decay processes once they become excited by photoirradiation (102, 149). Upon absorbing light, PSs first become excited to a singlet excited-state whereby an outermost electron is raised to higher energy level. The excited PS may then emit light through the process of fluorescence, temporarily alter its geometry and generate heat through the process of internal conversion (IC), or undergo photoreactions that

52 break existing bonds or form new bonds. However, to be useful for PDT, PSs must efficiently undergo intersystem crossing (ISC) and enter an excited triplet-state. Here, PS may emit light through a radiative decay process called luminescence. In terms of PDT, however, PSs should efficiently generate ROS.

Reactive oxygen species can be generated by two distinct mechanisms. The first, and typically the most dominant, mechanism has been labelled as ‗type 2‘ PDT. This involves the transfer of 3 energy from the PS in its excited triplet-state to molecular oxygen ( O2) in the triplet ground- 1 state. This triplet-triplet ―annihilation‖ is responsible for the formation of singlet oxygen ( O2), which is the most common ROS that is generated by PSs during PDT (112). While PDT typically 1 involves the photogeneration of O2, a second (and more elusive) ROS-generating pathway exists. Labelled ‗type 1‘ PDT, this mechanism is characterized by electron transfer or hydrogen transfer between the PS in its excited triplet-state and either O2, water or solvents (128, 129). These reactions can yield ROS including the superoxide radical anion, hydrogen peroxide, and/or hydroxyl radical (102, 128). Due to their powerful oxidizing potential, these type 1 PDT ROS 1 can be just as, or even more cytotoxic compared to O2 (106, 128, 150). These type 1 PDT can then continue to oxidize cholesterol (112), polyunsaturated fatty acids (106), susceptible portions of proteins (ex. cysteine, histidine, tyrosine, tryptophan and methionine) (106) and nucleic acids (ex. guanine, thymine, and uracil) (106, 128) in biological systems to cause cytotoxicity. The excited PS can even undergo electron transfer or hydrogen transfer with the aforementioned biological molecules directly through a process sometimes referred to as ‗type 3‘ PDT (more commonly still considered as type 1 PDT) (103, 151).

Type 1 and type 2 PDT each have merits and caveats. If photoirradiation does not lead to rapid 1 photobleaching of the PS, the type 2 PDT process can generate O2 catalytically (152). However, 3 3 since type 2 PDT relies solely on the availability of O2, type 2 PDT is abolished once O2 is consumed in the local irradiated region (150). On the contrary, type 1 PDT can occur under hypoxic or anoxic conditions (106, 130). However, this process may lead to more rapid photobleaching compared to type 2 PDT. Ideally, PSs should efficiently exploit both PDT pathways. Yet, few suggestions have been offered for enhancing the efficiency of the elusive type 1 PDT mechanism. This section of the review will provide examples of bacteriochlorins that are known to produce type 1 PDT ROS upon photoirradiation. In addition, it will present insights into studies that may be done in the future to explore methods of favouring the type 1 PDT pathway.

53 The bacteriochlorin, WST09 (Tookad, 57), is a natural product derivatives of bacteriochlorophyll a (13) that is synthesized by Pd-insertion into bacteriopheophorbide a (58) (Chart 1.2.1) (153). The photophysical properties of 57 have been extensively studied and this compound has even been clinically evaluated for anti-cancer PDT applications (154-157). Photophysical characterizations of 57 have revealed some amazing traits. First, Pd-insertion afforded 57 with a

ΦT that is near unity in organic solvents (158). However, under aqueous conditions, the ΦΔ is dramatically reduced to ~0.5 while the type 1 PDT pathway is slightly enhanced (158). This example suggests that the efficiencies of the aforementioned PDT pathways are solvent- dependent (106). Therefore, even if the photophysical properties of bacteriochlorin PSs are characterized in organic solvents in situ, these properties may change dramatically during in vitro and in vivo PDT (159). Future studies should take note, and make efforts to characterize the photophysical properties of bacteriochlorins in aqueous solutions (in situ) before evaluating these PSs for in vitro or in vivo PDT.

The PS, WST11 (Tookad-soluble, 59), is a derivative of 57 that is synthesized by taurine-driven E-ring opening at the 131 position (Chart 1.2.1) (160, 161). The most interesting characteristic of 59 is that it primarily undergoes type 1 PDT under aqueous conditions and can generate type 1 PDT ROS catalytically in the presence of serum proteins (162). Noting that the dominant PDT mechanism of 57 switch from type 2 PDT to the type 1 PDT pathway (59), we investigated the effects of E-ring opening with taurine and net charge variations on derivatives of the free base 58 (163). These studies were designed to explore which of these modifications could modulate the photogeneration of type 1 and type 2 PDT ROS. These results will be presented and discussed in Chapter 3. Briefly, we found that taurine-driven E-ring opening and increasing negative charge (with carboxylates and sulfonates) generally enhanced the photogeneration of ROS. Though we could not delineate the generation of type 1 and type 2 PDT ROS, our results strongly suggest that all of the 10 bacteriochlorins within this study were capable of undergoing type 1 PDT. Therefore, future efforts should continue to explore methods of the enhancing the type 1 PDT pathway using bacteriochlorins that do not contain an exocyclic E-ring, and continue to investigate the use of anionic moieties.

Chart 1.2.1. Bacteriochlorins known to generate type 1 PDT ROS upon photoirradiation.

The Zn-derivative of the free base 58, Zn-bacteriopheophorbide a (60), was also capable of generating type 1 PDT ROS in aqueous solutions, albeit likely to a lesser extent than the potent type 1 PDT agent 59 (Chart 1.2.1) (164). Indeed, both 57 (Pd) and 60 (Zn) were found to produce hydroperoxides (ROOH) and endoperoxides (ROOR) through the type 1 PDT pathway upon photoirradiation (164). In addition to these aforementioned metalo-bacteriochlorins (57, 59, and 60), the highly-studied free base bacteriochlorin a (61) was also found to generate superoxide radical anions and hydroxyl radicals with quantum efficiencies of 0.05 for each of these processes (Chart 1.2.1) (165). Even purely synthetic bacteriochlorins can undergo type 1 PDT. For example, 5,15-diphenyl-bacteriochlorin (62), generated type 1 PDT ROS under both aerobic and anaerobic atmospheres (Chart 1.2.1) (166).

55 These previous examples illustrate that it is very likely that all monomeric bacteriochlorins will produce type 1 PDT ROS upon photoirradiation in aqueous solutions to some extent. Ultimately, the redox potential of bacteriochlorins and the local O2 concentration/pressure will govern the favoured PDT pathway (104, 167). While it is difficult to conclude how bacteriochlorins should be modified to enhance the type 1 PDT pathway, some studies have offered suggestions. For example, Rajagopalan et al. reported that the azido (N=N=N), azo (N=N), sulfenato (S–O) and oxaza (N–O) groups of PSs are likely sites for excited-state electron transfer (168). Because the type 1 PDT mechanism is very elusive and because this pathway would be valuable for successful PDT under hypoxic conditions, future studies should consider all potential methods of enhancing this pathway. As described, these may include metalation (ex. Zn or Pd), E-ring opening of natural bacteriochlorin derivatives, and the use of the aforementioned functional groups.

Conclusions

Over the past several decades, researchers have made great strides in developing methods for synthesizing natural bacteriochlorin derivatives and purely synthetic bacteriochlorins. Many groups have subsequently characterized the photophysical properties of these compounds.

However, few have attempted to discover key structures that can modulate the ΦISC, ΦISC, ΦΔ,

ΦO*-, ΦHOOH, and ΦHO* of bacteriochlorin PSs. It is well known that insertion of diamagnetic metals and the incorporation of relatively high molecular weight halogens can enhance some of these photophysical properties. This review provided examples to reiterate the reliability of these strategies and described some nuances associated with them. In addition, several less-obvious insights were offered for enhancing the PDT-relevant photophysical properties of bacteriochlorins. While clear conclusions could not be drawn in several instances in this review, the suggestions herein were based on links between specific bacteriochlorin structures and modulations of photophysical properties that affect the photogeneration of ROS. It is envisioned that future efforts will explore these strategies further in attempts to improve the photosensitizing properties of known bacteriochlorins, and to use these design parameters to create novel bacteriochlorins with desirable photoproperties. It is highly unlikely that the aforementioned strategies will directly yield the ideal bacteriochlorin PS. Yet, future studies that explore these strategies will certainly help refine the best methods of designing optimal bacteriochlorins for PDT.

56 Section 1.3. Towards Cancer-Targeted Photodynamic Therapy: Overarching Thesis Goals

Aziridine aldehyde-cyclized integrin-targeting peptides

As introduced in Section 1 of the Introduction chapter, cyclization induces conformational restrictions to peptide sequences. The constrained geometry of cyclic peptides may potentially alter the chemical and biological properties of peptides that contain bioactive sequences. Generally, peptide cyclization may enhance the receptor subtype specificity of cancer biomarker- targeting sequences, and will likely improve chemical and proteolytic stabilities. Yet, the chosen cyclization chemistry and the sites of cyclization are highly-likely to influence the chemical and biological properties of cancer biomarker-targeting sequences. With a new ‗modified-Ugi‘ peptide cyclization method in hand, we chose to evaluate the conformational and biological properties of aziridine aldehyde-cyclized peptides that contained the known integrin-targeting sequence, RGD. Aziridine aldehyde-driven macrocyclization was reported to be a highly- modular and efficient cyclization method. However, new macrocycles were not yet extensively investigated for cancer-targeting applications. Chapter 2 presents the synthesis, structural characterization and biological evaluations of new αVβ3 integrin-targeting cyclic peptides in an effort to determine i) if aziridine aldehyde-driven macrocyclization can be useful for cancer- targeting applications, and ii) if insights can be provided to aid future efforts in the development of superior cancer-targeting RGD-based cancer-specific ligands. If successful, such cancer- targeting agents can be envisioned to specifically deliver anti-cancer photosensitizers for achieving cancer-targeted PDT.

Structural features for modulating the photoproperties of bacteriochlorin photosensitizers

Section 2 of this chapter introduced the advantages of bacteriochlorins for PDT. The focus of the aforementioned section was to present known and potentially new strategies of optimizing the photoproperties of this class of porphyrinoids in order to aid the development of superior bacteriochlorin photosensitizers. This review showed that the clearest structural modification for affording bacteriochlorins with high ΦΔ, ΦT, and ΦISC was the incorporation of high molecular weight diamagnetic metals (i.e. Pd and In) into the core of the bacteriochlorin scaffold. Yet, a fascinating report by Samankumara et al. showed that rings fused to the periphery of the bacteriochlorin scaffold clearly hindered the population of the triplet excited state via ISC (148).

57 This report lends support for our investigation of the affect taurine-driven exocyclic E-ring- opening had on the PDT-relevant photoproperties of bacteriochlorins. Inspired by previous reports that the dominant PDT-mechanism of clinically-evaluated bacteriochlorin photosensitizers (WST09 and WST11) switched due to E-ring opening, we designed and synthesized a library of bacteriochlorins in an effort to determine if the presence of the exocyclic E-ring of bacteriochlorophyll a derivatives could alter PDT-relevant photoproperties. The results of these studies from Chapter 3 were envisioned to provide insights into the future development of new bacteriochlorins with photosensitizing properties that could possibly rival those of leading bacteriochlorin photosensitizers.

Expanding the structural library of bacteriochlorins for photodynamic therapy

It is currently unclear how the plethora of potential peripheral structural modifications available for creating new bacteriochlorins will affect PDT-relevant photophysical properties. In order to develop truly optimized bacteriochlorin for PDT, researchers should have access to all available bacteriochlorin structures. To this end, we presented our modest stride for expanding the structural library of bacteriochlorins. In Chapter 4, we describe a simple and efficient one-pot synthetic method for creating a currently rare class of bacteriochlorins that contain a distinct seven-membered exocyclic F-ring. Such F-ring-containing porphyrinoids (i.e. porphyrins and chlorins) were exclusively identified in marine and pelagic organisms and sediments. Interestingly, some of the chlorin counterparts were found to hold strong antioxidant properties. These reports inspired our efforts to explore the potential biomedical applications of this new compound. In addition to evaluating its use as an antioxidant, we tested its photosensitizing ability since the precursor and its derivatives have been extensively tested for anti-cancer PDT. It was envisioned that the results of this chapter would provide researchers with a simple synthetic method for creating vast libraries of F-ring-containing bacteriochlorins for developing new derivatives that behave as either strong antioxidants or potent photosensitizers.

Towards achieving cancer-targeted photodynamic therapy using aziridine aldehyde-cyclized peptides and bacteriochlorin photosensitizers

Chapters 2, 3 and 4 present the synthesis and characterization of new cyclic cancer-targeting peptides and bacteriochlorin photosensitizers. It is envisioned that future efforts would expand upon the research presented within each individual chapter. In addition, we propose that new

58 aziridine aldehyde-cyclized RGD peptide-bacteriochlorin conjugates should be developed and thoroughly evaluated for cancer-specific PDT. It is currently unknown how this class of cyclic RGD peptides or the new bacteriochlorin photosensitizers presented in this thesis will behave in vivo, separately and as conjugates. While successful cancer-targeting PDT was not achieved in this thesis work using the aforementioned conjugates, we present our latest preliminary results toward this goal in Chapter 5. These studies should provide insights for aiding the development of a new class of cancer-targeting peptide-photosensitizer conjugates for cancer-targeted PDT using the described classes of cyclic peptides and bacteriochlorins.

Chapter 2 Conformational Modulation of In Vitro Activity of Cyclic RGD Peptides via Aziridine Aldehyde-Driven Macrocyclization Chemistry

Preamble

This chapter is comprised of the manuscript, ―Conformational modulation of in vitro activity of cyclic RGD peptides via aziridine aldehyde-driven macrocyclization chemistry‖ that was adapted from Roxin, Á., Chen, J., Scully, C. C. G., Rotstein, B. H., Yudin, A. K., and Zheng, G., Bioconjugate Chemistry 2012, 23(7), 1387-1395, (169) Copyright © 2012 American Chemical Society, with permission from the American Chemical Society. This publication has been reformatted to the required SGS style. I was the primary contributor to this project. I planned and performed all chemical and biological experiments. Briefly, I synthesized, purified, and identified all chemical precursors, peptides and bioconjugates, and the performed all in vitro microscopy and cell adhesion assays. I also interpreted all of the data and wrote the majority of the manuscript. Dr. Chen offered advice for chemical synthesis, helped plan biological experiments, and contributed to structuring and editing the manuscript. Dr. Scully performed the computational experiments, interpreted the computational results, wrote the computational sections of the manuscript, and helped edit the manuscript. Dr. Rotstein offered advice for the synthesis of the unmodified aziridine aldehyde dimer, and helped edit the manuscript. Dr. Yudin helped plan the chemical and computational experiments, helped to interpret the results of the project, and contributed to structuring and editing the manuscript. Dr. Zheng helped plan the chemical and biological experiments, and contributed to structuring and editing the manuscript.

Introduction

Research into peptides targeting biomarkers associated with cancer actively pursues the creation of cyclic peptides that display specific biological activity and that can be functionalized for targeted imaging or therapy. A recent cyclization method developed in our labs employs amphoteric aziridine aldehydes and isocyanides to cyclize linear peptides through a rerouted Ugi reaction (170). Aziridine aldehyde-driven cyclization proceeds rapidly, giving high yields at higher concentrations than is possible for conventional peptide cyclizations. This original 59

60 procedure reported the conjugation of a thiolated coumarin fluorophore after cyclization and has since been refined to include several functional inputs. For instance, in addition to post- cyclization fluorophore labeling, macrocycles have been labeled with a solvatochromic naphthalimide isocyanide during cyclization (171). Most recently, novel thioester isocyanide reagents were developed with the goal of subsequent macrocycle-peptide ligation to access cycle-tail peptide scaffolds (172). We sought to expand these recent conjugation strategies by developing a versatile modification chemistry whereby the conjugation of a wide variety of fluorophores, radiometal chelators and other biological entities could be achieved.

Conventional amide-cyclized peptides often require additional lysine, cysteine or glutamate residues to support side chain-selective conjugations. However, conjugations through these additional residues may hinder binding of the biologically relevant peptide regions (173-175) and often require cumbersome protection/ deprotection protocols (14, 176-178). A method that enables one to constrain a linear epitope of interest into a macrocycle, while offering an exocyclic handle that can be modified at a late stage of synthesis without perturbing the desired conformation of the molecule (179), is highly desirable. Here we show a novel modification strategy wherein the thiol of cysteamine nucleophilically attacks aziridine aldehyde-cyclized macrocycles, resulting in aziridine ring-opening. Fluorescein was then conjugated to the macrocycles through the primary amine of cysteamine using N-hydroxysuccinimide (NHS)- chemistry (Figure 2.1). This strategy removes the necessity of incorporating superfluous residues and allows the conjugation of a variety of biological entities containing carboxylic acid, isothiocyanate (ITC) or NHS functionalities by well-established amide coupling.

Arg-Gly-Asp (RGD)-containing peptides were chosen as a model for our investigations. The

RGD motif is known to bind to the αVβ3 integrin receptor which is overexpressed by several types of cancers and by tumor neovasculature (49, 180-182). To obtain biologically active RGD motifs, computer models of RGD-based aziridine aldehyde-cyclized pentapeptide and hexapeptide macrocycles were constructed in attempts to recapitulate the observed backbone geometry from solution phase and crystallographic studies of potent integrin binding peptides.

Docked conformations of promising RGD macrocycles and the αVβ3 integrin receptor were examined to investigate if these RGD macrocycles could attain binding geometries that were similar to the known high-affinity ligand cRGDf(N-Me)V (Cilengitide) (183-185). We proposed that if our RGD macrocycles could stabilize the RGD motif in a bioactive conformation similar to that of cRGDf(N-Me)V, then these peptides would be suitable for in vitro αVβ3 integrin- binding. Linear RGD-containing peptides were cyclized and labelled with fluorescein to experimentally validate the biological activity of the proposed functional macrocycles (Figure 2.2). Cell adhesion assays compared the binding affinity of the pentapeptide cPRGDA (13) and the hexapeptide cPRGDAA (14) to investigate how the geometry of these peptides correlated to their binding affinity for the αVβ3 integrin receptor of U87 glioblastoma cells. Confocal microscopy was then used to further investigate if these fluorescein-labeled macrocycles could specifically target the αVβ3 integrin receptor of U87 cells for in vitro cancer imaging. Our study

62 provides a guide to experimentally verify the biological activity and specificity of our constrained RGD macrocycles proposed by computational geometry-based modeling that can be extended for screening other biologically active receptor-targeting motifs.

Results

Aziridine ring-opening and fluorophore/ chelator conjugation

The general synthetic strategy included four distinct steps: 1) aziridine aldehyde-driven cyclization of linear peptides, 2) aziridine ring-opening by a hetero-bifunctional linker, 3) dye conjugation and 4) full deprotection of amino acid side chains. The linear peptides with the sequences PR(Pbf)GD(OtBu)A, PR(Pbf)GD(OtBu)AA and PR(Pbf)D(OtBu)GA were successfully cyclized by our recently reported aziridine aldehyde-driven cyclization chemistry. Cysteamine was attached to these macrocycles by zinc chloride-catalyzed aziridine ring-opening. The cysteamine-modified macrocycles (7, 8 and 9) were then conjugated to fluorescein-NHS using DIPEA in DMF in 30 minutes (10, 11 and 12). The side chain protecting groups were then removed using TFA to produce cPRGDA (13), cPRGDAA (14) and cPRDGA (15). We also extended the exocylic conjugation of the cysteamine-modified macrocycles to other reporters. The radiometal chelator DOTA was efficiently conjugated to the free amine of cysteamine on the macrocycle PR(Pbf)GD(OtBu)A (7) using NHS-chemistry (24). The multimodal fluorescent dye, photosensitizer and radiometal chelator pyropheophorbide a was similarly conjugated to the free amine of cysteamine on the macrocycle PR(Pbf)GD(OtBu)A (7) using NHS-chemistry (25).

63 Computer modeling

Macromodel was used to generate conformational ensembles of cPRGDA (13), cPRGDAA (14) and amide-cPRGDK (23). The average conformations of cPRGDA (Figure 2.3A), cPRGDAA (Figure 2.3B) and the amide-cPRGDK were analyzed to investigate suitability for binding to the

αVβ3 integrin receptor. A preliminary study indicated that the pentapeptides, cPRGDA and amide-cPRGDK, preferred a backbone conformation that overlaid well with the RGD motif of cRGDf(N-Me)V (bound to the extracellular αVβ3 integrin protein (PDB: 1L5G) (186))(Figure 2.3D) while the hexapeptide cPRGDAA showed significant backbone deviation in this area (Figure 2.3C). Conformational searches generated between 233 and 368 conformers for each of these macrocycles from which the mean Pro-Cα – Asp-Cα distances within cyclic peptides cPRGDA, cPRGDAA and amide-cPRGDK (Table 2.1) were calculated. Both cPRGDA and cPRGDAA contain a γ-turn-like backbone conformation similar to cRGDf(N-Me)V at the RGD motif (Arg-Cα – Asp-Cα < 7Å). However, the cycle topology is significantly different when the residue preceding the RGD motif is included in the backbone comparison. The Val-Cα-Asp-Cα distance was measured at 6.9 Å in the crystal structure of cRGDf(N-Me)V which is similar to the distance calculated for Pro-Cα - Asp-Cα in cPRGDA (6.1 Å). The Pro-Cα- Asp-Cα distance in cPRGDK is shorter (5.2 Å) and is much longer in cPRGDAA (8.7 Å). This suggests that differing peptide backbone topology could account for the enhanced binding of the RGD-motif of cPRGDA to the αVβ3 integrin binding site compared to the larger cPRGDAA.

Peptide Pro-Cα- Asp-Cα distance (Å) Na

amide-cPRGDK 5.24 ± 0.15 368

cPRGDA 6.11 ± 0.17 264

cPRGDAA 8.37 ± 0.24 233 a. Number of generated conformers.

65 ε Ligand docking to the extracellular αVβ3 integrin protein showed that cPRGDA and the N - acetyl-lysine analogue of amide-cPRGDK possessed similar stabilizing binding interactions to cRGDf(N-Me)V from the crystal structure. The Lys residue of amide-cPRGDK (22) was Nε- acetylated in its docking study to mimic the presence of fluorescein in amide-cPRGDK (23). It was found that cPRGDA and cRGDf(N-Me)V docked to the αVβ3 integrin protein through their respective Arg residues to Asp150 and Asp218 of αVβ3 and through their Asp residues to Mn1 (Figure 2.4A). The Nε-acetyl-lysine of the amide-cPRGDK analogue formed contacts to the same residues and also to Asp126 through the amide proton of the Nε-acetyl-lysine residue (Figure

2.4B). Heavy atom distances between the respective docking regions of the αVβ3 integrin protein to cPRGDA, amide-cPRGDK and cRGDf(N-Me)V were calculated (Table 2.2). These measurements showed that while cPRGDA and cRGDf(N-Me)V used all three nitrogen atoms of their respective Arg residues for optimal binding to Asp150 and Asp218, the binding mode of the Arg of amide-cPRGDK is inferior. This observed binding mode uses two of its three available nitrogen atoms to bind to the dual aspartate motif of the integrin.

66 Table 2.2. Calculated heavy atom distances between macrocycles and the extracellular binding regions of the αVβ3 integrin receptor. Reproduced with permission from (169). Copyright © 2012, American Chemical Society.

Peptide Ligand atom Receptor atom Heavy atom distance (Å)

cPRGDA Asp-OD Mn1 2.1

Arg-NE D218-OD1 3.1

Arg-NZ1 D218-OD2 3.1

Arg-NZ2 D150-OD 3.2

amide-cPRGDK Asp-OD Mn1 2.1

Arg-NZ1 D218-OD1 3.1

Arg-NZ1 D218-OD2 2.9

Arg-NZ2 D150-OD 3.0

Ac-Lys-NZ D126-OD 3.6

cRGDf(N-Me)V Asp-OD Mn1 2.7

Arg-NE D218-OD1 2.9

Arg-NZ1 D218-OD2 2.8

Arg-NZ2 D150-OD 3.3

Binding affinity

With computational insights in hand, the binding affinity of each fluorescein-labeled RGD- macrocycle for the αVβ3 integrin receptor was investigated experimentally using competitive cell adhesion assays. The peptide concentration-dependent displacement of U87 cell adherence from vitronectin-coated wells was used to calculate the respective IC50 values for each fluorescein- labeled macrocycles cPRGDA (13), cPRGDAA (14) and amide-cPRGDK (23). The IC50 value of these RGD-containing macrocycles was on the same order (low μM) as previously reported amide-cyclized RGD sequences analyzed by this competitive cell adhesion method (Table 2.3) (187). Though cRGDf(N-Me)V was not tested in this study, a review by Rüegg et al. reported that Cilengitide displays a similar αVβ3 integrin binding affinity (IC50 ~1 μM) to displace αVβ3

67 integrin-expressing cells from adhering to vitronectin (188). The IC50 of cPRGDA (4.2 μM) was significantly lower compared to cPRGDAA (11.1 μM, p < 0.05) (Table 2.3). These differential

IC50 values were attributed to modulated geometries as a result of peptide length and the presence of the constrained RGD motif in cPRGDA. It was also found that there was not a significant difference in the IC50 of cPRGDA compared to amide-cPRGDK.

a b c Peptide log IC50 (M) SE (log M) IC50 (μM)

amide-cPRGDK -5.114 ± 0.2 7.7

cPRGDA -5.381 ± 0.2 4.2*

cPRGDAA -4.955 ± 0.06 11.11*

cPRDGA n.d.d n.d.d - a. Calculated in Graph-Pad PRISM using a log(inhibitor) vs. response three parameter non-linear regression from an xy scatter plot (bottom constraint at 18) of all normalized values from 3 independent experiments. b. Standard error calculated for log IC50 presented as log of molar concentrations. c. Transformed from log IC50. d. n.d., Not determined. * Significantly different using a two-tailed unpaired t-test (p < 0.05).

Confocal microscopy

Confocal microscopy investigated the specific binding of the fluorescein-labeled cPRGDA (13) and cPRGDAA (14) peptides to the human αVβ3 integrin receptor of U87 glioblastoma cells. These studies showed that cPRGDA, cPRGDAA and the amide-cPRGDK (23) each bound to U87 cells (Figures 2.5A, 2.5C and 2.5D) but none bound to HT-29 cells (Figures 2.5G, 2.5I and 2.5J). The cPRDGA scramble control (15) showed non-specific binding to U87 cells (Figure 2.5B) but not to HT-29 cells (Figure 2.5H). The targeting specificity of cPRGDA and cPRGDAA was investigated by blocking the αVβ3 integrin receptor of U87 cells with a 50-fold molar excess of the linear RGD peptide (20). This blocking reduced specific αVβ3 integrin binding for both

68 RGD-containing macrocycles but the remaining fluorescence signals suggested additional non- specific binding (Figures 2.5E and 2.5F).

Discussion

The aziridine aldehyde-driven cyclization method has been successfully developed to synthesize functional αVβ3 integrin-targeting fluorescein-labeled macrocycles. We were able to show that increasing the peptide length (from five to six residues) and switching the Gly and Asp residues in the middle of the sequence did not affect the aziridine aldehyde-driven cyclization reaction. We validated that a simple zinc chloride salt would catalyze the sulfhydryl-driven aziridine ring- opening of the exocyclic unsubstituted aziridine handle of our macrocycles by the hetero- bifunctional linker cysteamine. The versatility of this synthetic strategy was illustrated by conjugating the common radiometal chelator DOTA and the multimodal fluorescent dye/ radiometal chelator/ photosensitizer pyropheophorbide a using NHS-chemistry. This conjugation could be easily adapted to a wide variety of other fluorescent dyes, radiometal chelators or extended linkers using conventional amide-coupling, the described NHS-chemistry or by analogous ITC-chemistry.

Aziridine aldehyde-driven macrocyclization reactions showed clear advantages over conventional end-to-end amide cyclization. The former reactions were performed at a relatively high concentration (50 mM in TFE) wherein no noticeable polymerization was observed. In

69 contrast, the amide cyclization required high dilution (0.6 mM in DMF) to ensure that amide coupling was predominantly due to intra-molecular bonding of the free N- and C-termini within each monomeric linear PRGDK peptide (16). In addition, the conventional amide cyclization reactions require an additional amino acid to provide a thiol, carboxylic acid or primary amine for subsequent functional group conjugations. These amino acids should be specially protected and selectively deprotected for functional group conjugation (14, 176-178). On the contrary; the aziridine aldehyde-driven macrocyclization offers a free amine for conjugation once cysteamine has become bonded to the macrocycle by aziridine ring-opening. We are thereby able to avoid the challenges of protection/ deprotection often required after amide cyclization. Finally, after the aziridine ring-opening of the aziridine aldehyde-cyclized macrocycles, the space between the binding motif of the macrocycle and the conjugation site may have allowed the desired receptor binding to proceed without imposing significant steric hindrance.

Analysis of the X-ray crystal structure of integrin bound cRGDf(N-Me)V shows significant difference between it and the solution phase structure proposed by Kessler et al. (186). Cyclic pentapeptides with (ldldl) relative stereochemistry often adopt a βII′/γ-turn architecture in solution. In the case of cRGDf(N-Me)V, Gly occupies the i+1 position of the γ turn while d-Phe occupies the i+1 position of the βII′ turn. The X-ray crystal structure of integrin-bound cRGDf(N-Me)V shows that the γ turn is maintained in the bound state, with a 6.4 Å separation between the Arg and Asp α carbons and a 8.7 Å distance separating the Arg and Asp β carbons. cPRGDA docked to αVβ3 and shows that it adopts the same γ turn conformation as the docked cRGDf(N-Me)V which is critical for the polar side chains of Arg and Asp to engage in receptor binding. The Arg - Asp inter-carbon (Cα and Cβ) distances of cPRGDA were similar to those of cRGDf(N-Me)V where the Arg-Asp (Cα-Cα) distance is 6.0 Å and the Arg-Asp (Cβ-Cβ) distance is 8.7 Å. Kessler et al. have shown that the identity of the remaining residues in the peptide are important for structural rigidity but due to the fact that residues not in the RGD motif do not make contact with the integrin surface, ample room exists to splice different residues and functional handles into the macrocycle sequence (31), hence the observed binding affinity for cPRGDA.

Computer modeling data showed that cPRGDA contained a similar backbone conformation at the RGD motif to cRGDf(N-Me)V, while cPRGDAA differs significantly. This suggested that the stabilized RGD motif within cPRGDA would contribute to a better binding affinity for the

αVβ3 integrin receptor compared to cPRGDAA. Competitive cell adhesion assays experimentally

70 showed that the IC50 of cPRGDA for αVβ3 integrin-overexpressing U87 glioblastoma cells was significantly lower compared to cPRGDAA. This may be explained by the backbone shape and macrocycle ring size differences between the two peptides. It is possible that other RGD- containing pentapeptide sequences cyclized by the aziridine aldehyde-driven reaction may also contain this desired turn at the RGD motif. Nevertheless, confocal analysis showed that both the fluorescein-labeled cPRGDA and cPRGDAA macrocycles could successfully bind to U87 cells for in vitro fluorescence cancer imaging while avoiding binding to HT-29 cells that express low levels of the αVβ3 integrin receptor.

The amide-cPRGDK also fulfills the α carbon requirement for a γ-turn-like structure, having a distance of 4.6 Å between the Arg and Asp α carbons. However, the β carbons are much closer than in either PRGDA or cRGDf(N-Me)V where a distance of 7.2 Å separates them. This means that the backbone dihedral angles are significantly different for amide-cPRGDK, which could contribute to a reduction in affinity compared to cRGDf(N-Me)V despite its similar ring size.

Computational docking studies identified the binding mode of cPRGDA and the Nε-acetyl-lysine analogue of amide-cPRGDK with the extracellular αVβ3 integrin receptor. The Lys of amide- cPRGDK was acetylated to mimic the hindrance imposed by fluorescein at the Lys residue in amide-cPRGDK (23). We found that both macrocycles interacted with the Asp150 and Asp218 residues of the protein through their respective Arg residues, much the same as cRGDf(N-Me)V. Interestingly, amide-cPRGDK only used two of its three available Arg nitrogen atoms to bind to Asp150 and Asp218 while cPRGDA and cRGDf(N-Me)V used all three nitrogen atoms of their respective Arg residues for optimal integrin binding. In addition, the Nε-acetyl-lysine residue of amide-cPRGDK interacted with Asp126. Therefore, it was not obvious which binding interaction would have the greatest contribution to the binding affinity of cPRGDA and amide-cPRGDK.

Competitive cell adhesion assays experimentally showed that the IC50 of cPRGDA for U87 cells was not significantly different compared to amide-cPRGDK. This suggests that the aforementioned difference observed in the described docking studies compensated to create geometries that allowed similar in vitro binding to the αVβ3 integrin receptor on U87 cells.

The chemical structure of the cPRGDA, cPRGDAA and cPRDGA macrocycles is inherently different from analogous peptide macrocycles generated through backbone amide coupling. The aziridine aldehyde-driven cyclization and the subsequent aziridine ring-opening reaction result in the formation of an sp3 hybridized carbon atom at the non-amino acid portion of the

71 macrocycles. The described geometric and binding affinity comparisons between cPRGDA and amide-cPRGDK are similar to the investigations published by Geyer et al. in 1994 (189). They developed the cyclic peptide cRGDfψ[CH2-NH]V that contained a reduced amide bond that imposed geometric differences when compared to the parent cRGDfV. Whereas with cRGDfψ[CH2-NH]V, they found that a β-turn forms between the Val and Asp residues (a γ turn lies between Arg and Asp in potent RGD cyclic peptides). They also found that the structural perturbation caused by the introduction of a reduced amide bond into a cyclic RGD peptide caused a significant drop in activity. This suggests that RGD macrocycles containing only amide bonds would stabilize geometries that result in superior αVβ3 integrin binding affinities compared to analogues that do not contain a fully amide-bonded backbone. Interestingly, the binding affinity of cPRGDA to the αVβ3 integrin receptor was not significantly different compared to amide-cPRGDK. Though the amino acid residues of cPRGDA and amide-cPRGDK were not identical, it could be inferred that the sp3 carbon-containing macrocycles generated through aziridine aldehyde-driven cyclization and subsequent aziridine ring-opening may not hinder the binding affinity of cyclic RGD sequences to the same extent as the aforementioned study when compared to analogues that contain only amide bonds.

We have shown that the aziridine aldehyde-driven cyclization modulated the geometry of similar five- and six-membered RGD macrocycles and was responsible for tuning the binding affinity of cPRGDA and cPRGDAA for the αVβ3 integrin receptor on U87 cancer cells. It is possible that the geometry-modulating ability of this macrocyclization reaction can create other RGD-based macrocycles that have varying binding affinities and also varying specificities to different integrin subtypes that are also overexpressed by cancers. This geometry-modulating ability could also be extended to other cancer biomarker-targeting peptide sequences. Using our current modification strategy, these new potential macrocycles could then be conjugated to a variety of functionalities for additional imaging and therapeutic applications.

Conclusions

We have created a versatile strategy for conjugating a common fluorescent dye to RGD- containing αVβ3 integrin receptor-targeting macrocycles which were cyclized by the aziridine aldehyde-driven reaction. This same general method was used to conjugate the commonly used radiometal chelator DOTA and the multifunctional porphyrin pyropheophorbide to a novel RGD- containing macrocycle to display the versatility of our conjugation strategy. Computer modeling

72 studies showed that the aziridine aldehyde-driven cyclization modulated the geometry of similar five and six amino acid-containing RGD macrocycles. This cyclization chemistry was responsible for stabilizing a γ turn at the RGD motif of cPRGDA. Though each fluorescein- labeled RGD-containing peptide specifically targeted the αVβ3 integrin receptor of U87 glioblastoma cells in vitro, it is likely that a stabilized γ turn and the significantly shorter Pro-Cα- Asp-Cα distance were responsible for improving the binding affinity of cPRGDA compared to cPRGDAA. Future studies should continue using pentapeptides to constrain the RGD motif into an active conformation and replace the C-terminal Ala with other natural or synthetic l- or d- amino acid residues to further improve αVβ3 integrin binding or possibly to selectively target different integrin subtypes. Additional studies could use the aziridine aldehyde-driven cyclization chemistry to modulate the geometry of other cancer biomarker-targeting sequences and utilize the described modification strategy for a variety of cancer imaging and therapeutic applications.

Materials and Methods

Chemicals

Organic solvents were purchased from Fisher Scientific and Sigma Aldrich. Amino acids and resins were purchased from Nova Biochem. Cysteamine, zinc chloride, fluorenylmethyloxycarbonyl chloride (Fmoc-Cl), l-serine, triphenyl phosphine (PPh3), diisopropyl azodicarboxylate (DIAD), diisobutylaluminum hydride (DIBAL), tert-butyl isocyanide and crystal violet were purchased from Sigma Aldrich. Fluorescein-NHS was purchased from Thermo Scientific. The chelator 1,4,7,10-tetraazacyclododecane-1,4,7-tris(t- butyl acetate)-10-acetate mono-N-hydroxysuccinimide ester (DOTA-NHS) was purchased from Macrocyclics. Pyropheophorbide-N-hydroxysuccinimide ester (Pyro-NHS) was synthesized following a previously published procedure (190).

General methods

Proton NMR was performed using a Bruker Ultrashield 400 Plus 400 MHz NMR. Mass spectrometry was performed using an Agilent Technologies 6130 Mass Spectrometer. The identity and purity of all peptide derivatives were assessed using a Waters 2695 HPLC with a Waters Delta Pak C18, 5 μm 3.9 x 150 mm column under reverse-phase conditions (0 – 100 % acetonitrile in 0.1 % TFA over 15 minutes at 0.8 mL/ minute) while monitoring with a Waters 2996 photodiode array detector and a Waters Micromass ZQ mass spectrometer. Macrocycles were purified using a Waters Prep LC System with an Agilent 300SB-C3, 5 μm 9.4 x 250 mm

73 column under reverse-phase conditions (0 – 100 % acetonitrile in 0.1 % TFA over 40 minutes at 3.0 mL/ minute) while monitoring with a Waters 2487 Dual λ absorbance detector set to 254 nm, 300 nm, 411 nm or 441 nm. Peptide derivatives were quantified spectrophotometrically by measuring the absorbance of Fmoc ((1:4) piperidine: DMF) or fluorescein (0.1 M NaOH) using a Varian 50 Bio UV-Visible spectrophotometer. Binding affinity assays measured absorbance values using a Molecular Devices Spectra Max M5 plate reader. Confocal experiments were analyzed with an Olympus FV1000 laser confocal scanning microscope.

Synthesis of unsubstituted aziridine-aldehyde (18)

The unsubstituted aziridine-aldehyde (18) was synthesized as previously reported in our lab (170, 191). Briefly, 10 g (95 mmoles) of l-serine was esterified to a serine propyl ester HCl salt (16) using 200 mL 1-propanol and 7.7 mL (106 mmoles) SOCl2. This reaction yielded 16.77 g (91.2 1 1 mmoles) of precursor (96 % yield). The identity was confirmed with H NMR. H NMR (CDCl3, 400 MHz): δ 0.90 (t, J = 7.4 Hz, 3H), 1.62 (m, 2H), 3.51 (t, J = 4.1 Hz, 1H), 3.68 (dd, J = 10.9, 5.5 Hz, 1H), 3.76 (dd, J = 10.7, 4.0 Hz, 1H), 4.05 (t, J = 6.6 Hz, 2H).

The HCl salt (16) was converted to the free base (13.37 g or 90.9 mmoles) using NH4OH. A

Mitsonobu reaction was performed using 23.85 g (90.9 mmoles) of PPh3 and 17.96 ml (90.9 mmoles) of DIAD in anhydrous DCM at -10 °C for 3 hours and at room temperature for 18 hours. The work-up required Et2O precipitation to remove the white PPh3O byproduct and a Kugelrohr apparatus (heated to 86 °C) to distill the aziridine propyl ester (17). This gave a 4.44 g (34.4 mmoles) sample of this precursor (38 % yield). The identity was confirmed with 1H NMR. 1 H NMR (CDCl3, 400 MHz): δ 0.92 (t, J = 7.4 Hz, 3H), 1.22 (m, 3H), 1.65 (m, 2H), 1.82 (d, J = 4.6 Hz, 1H), 1.96 (s, 1H), 2.49 (s, 1H), 4.08 (m, 2H).

The aziridine propyl ester (17) (2.0 g or 15.5 mmoles) was reduced with 38.8 mL (38.8 mmoles) of DIBAL (1.0 M in toluene) at -78 °C for 4.5 hours. The reaction was quenched using MeOH after 2 hour. Gelation was initiated with sat. Na2SO4 (aq) after 1 hour and 45 minutes. The gel was filtered and washed with MeOH to give 532 mg (7.44 mmoles) of the unsubstituted aziridine aldehyde dimer (18) (48 % yield). Mass spectrometry was used to identify the product (ESI-MS:

143.1 (M+1) m/z ). TLC was performed using acetonitrile: water (4:1) and showed an Rf = 0.4

(lit. Rf = 0.32) (170, 191). KMnO4 staining did not reveal any byproducts.

74 Manual linear peptide synthesis

The linear peptides PRGDA (1), PRGDAA (2), PRDGA (3), PRGDK (19) and RGD (20) were synthesized on the solid-phase with 2-chlorotrityl chloride resin (loading ~1.5 mmoles/ g) using standard Fmoc-chemistry. The amino acids Arg, Asp and Lys had side chains protected by Pbf, OtBu and Boc, respectively. Amide coupling of Fmoc-protected amino acids (3 eq.) was achieved after 2 hours using HBTU (3 eq.) in (1:1) DCM: DMF containing 5 % DIPEA (v/v). Resin capping was performed using 1-acetylimidazole (3 eq.) in a solution of (1:1) DCM: DMF containing 5 % (v/v) DIPEA. Peptides were cleaved from the resin using (1:2:97) TFA: TIS: DCM after 2 hours. Peptides were synthesized on a 250 μmole – 550 μmole scale.

Aziridine-driven cyclization

The peptides PR(Pbf)GD(OtBu)A (1), PR(Pbf)GD(OtBu)AA (2) and PR(Pbf)D(OtBu)GA (3) were cleaved from the 2-chlorotrityl chloride resin as described and the N-terminal Fmoc was removed using standard piperidine cleavage. The peptides were precipitated from a minimal volume of DMF using cold Et2O seven times. These peptides (10 µmoles) were cyclized by the aziridine aldehyde-driven reaction (170, 191) using 2.5 equivalents of the unsubstituted aziridine-aldehyde dimer (18) and 3.5 equivalents of tert-butyl isocyanide in TFE. The progress of cyclization reactions were assessed using HPLC-MS monitoring by integrating UV-Vis spectra scanned at 254 nm. After 17 – 19 hours, these cyclization reactions produced the desired macrocycles containing aziridine rings with the sequences PR(Pbf)GD(OtBu)A (4), PR(Pbf)GD(OtBu)AA (5) and PR(Pbf)D(OtBu)GA (6) where yields were estimated at ~47 %, ~38 % and ~43 %, respectively. Mass spectrometry showed the production of 4 (ESI-MS: 960 (M+1) m/z), 5 (ESI-MS: 1031 (M+1) m/z) and 6 (ESI-MS: 960 (M+1) m/z). The reaction mixtures were precipitated using Et2O. The crude mixtures were dried with a speed-vac and were stored at -20 ˚C.

Aziridine ring-opening with cysteamine

The reaction crudes from the previous aziridine aldehyde-driven cyclization were dissolved in

CHCl3. Cysteamine was dissolved in DMSO to make a 3 mg/µL stock solution while zinc chloride was dissolved in (95:5) CHCl3: DMF to make a 0.1 mg/µL stock solution. Cysteamine and zinc chloride was added to the CHCl3 solution containing the peptide crude. The molar ratio of peptide: cysteamine: zinc chloride was 1: 3: 1. The progress of this aziridine ring-opening reaction was assessed using HPLC-MS monitoring as described. Once the starting material was

75 consumed, the reaction mixture was dried using a speed-vac. The crude was dissolved in a minimal volume of DMSO and was purified using a prep-HPLC. This gave the cysteamine- modified macrocycles with the sequences PR(Pbf)GD(OtBu)A (7), PR(Pbf)GD(OtBu)AA (8) and PR(Pbf)D(OtBu)GA (9) with yields of ~23 %, ~33% and ~5 %, respectively. The identity and purity of each product was confirmed using HPLC-MS (Figure S2.1, S2.2 and S2.3). Mass spectrometry showed the isolation of 7 (ESI-MS: 1037 (M+1) m/z), 8 (ESI-MS: 1108 (M+1) m/z) and 9 (ESI-MS: 1037 (M+1) m/z).

Amide cyclization

The linear PR(Pbf)GD(OtBu)K(Boc) peptide (19) was cleaved from the 2-chlorotrityl chloride resin and the Fmoc-protecting group was removed as described. The NH- PR(Pbf)GD(OtBu)K(Boc)-COOH peptide (15 µmoles) was cyclized using HBTU (1 eq.) and HOBt (1 eq.) while dissolved in 25 mL of DMF (1.76 mM) with 2.3 % DIPEA (v/v). Reaction progress was monitored with HPLC-MS as described. The starting material was consumed within 20 hours. The solution was aliquoted to 1.5 mL Eppendorf tubes and dried with a speed- vac. The dried product (21) was precipitated with DMF and Et2O as described and dried with a speed-vac. The identity and purity of the product was confirmed using HPLC-MS (Figure S2.4). Mass spectrometry showed the production 21 (ESI-MS: 963 (M+1) m/z).

Fluorescein conjugation

Fluorescein-NHS (1.5 eq.) was conjugated to the cysteamine-modified macrocycles in DMF containing 4 % DIPEA (v/v) within 30 minutes. This reaction gave the side chain-protected fluorescein labeled macrocycles with the sequences PR(Pbf)GD(OtBu)A (10), PR(Pbf)GD(OtBu)AA (11) and PR(Pbf)D(OtBu)GA (12). ESI-MS(+) showed the production of 10 (ESI-MS: 1395 (M+1) m/z), 11 (ESI-MS: 1466 (M+1) m/z), and 12 (ESI-MS: 1395 (M+1) m/z). The Pbf and OtBu protecting groups were removed using TFA for 2 hours. This side chain- deprotection step gave the final fluorescein-labeled macrocycles cPRGDA (13), cPRGDAA (14) and cPRDGA (15). These solutions were precipitated with Et2O and purified using a prep-HPLC using the described method while detecting products with the 254 nm and 441 nm absorbance channels. HPLC-MS was used to determine the purity (> 95 % pure) and to confirm the identity of each product (Figure S2.5, S2.6 and S2.7). Mass spectrometry showed the isolation of 13 (ESI-MS: 1087 (M+1) m/z), 14 (ESI-MS: 1158 (M+1) m/z), and 15 (ESI-MS: 1087 (M+1) m/z). The peptides were then quantified spectrophotometrically.

76 The Pbf, OtBu and Boc protecting groups were removed from amide-cyclized PR(Pbf)GD(OtBu)K(Boc) peptide (21) using (95:5) TFA: DCM after 2 hours. The production of the side chain-deprotected amide-cyclized PRGDK (22) was confirmed by mass spectrometry

(ESI-MS: found 550 m/z). Fluorescein-NHS (1.3 eq.) was conjugated to the ε-NH2 of Lys residue of the macrocycle (22) (7.5 µmoles) in DMF containing 2.5 % DIPEA (v/v) within 30 minutes.

This solution were precipitated with Et2O and purified using a prep-HPLC using the described method while detecting product with the 254 nm and 441 nm absorbance channels. HPLC-MS was used to determine the purity (> 95 % pure) and to confirm the identity of this fluorescein- labeled cyclic peptide amide-cPRGDK (23) (Figure S2.8). Mass spectrometry showed the isolation of amide-cPRGDK (23) (ESI-MS: 913 (M+1) m/z). The peptide was then quantified spectrophotometrically.

DOTA conjugation

DOTA-NHS (3.3 eq.) was conjugated to the side chain-protected cysteamine-modified PR(Pbf)GD(OtBu)A macrocycle (7) (0.2 µmoles or 1 eq.) in 50 µL of DMF containing 6% DIPEA (v/v) within 1 hours. The crude was purified by prep-HPLC as described while detecting the product using a 254 nm absorbance channel. HPLC-MS was used to determine the purity (> 90 % pure) and to confirm the identity of the product (Figure S2.9). Mass spectrometry showed the production of cPRGDA-DOTA (24) (ESI-MS: 1423 (M+) m/z).

Pyropheophorbide a conjugation

Pyro-NHS (2 eq.) (190) was conjugated to the side chain-protected cysteamine-modified cPR(Pbf)GD(OtBu)A (7) (0.4 µmoles or 1 eq.) in 70 µL of DMF containing 4% DIPEA (v/v) within 2 hours. The crude was purified by prep-HPLC as described while detecting the product using 254 nm and 411 nm absorbance channels. HPLC-MS was used to determine the purity (> 95 % pure) and to confirm the identity of the product (Figure S2.10). Mass spectrometry showed the production of cPRGDA-Pyro (25) (ESI-MS: 1553 (M+) m/z).

Computer modeling

The cPRGDA (13) and cPRGDAA (14) peptides, as well as the amide-cPRGDK (23) peptide, were modeled using the Monte Carlo Multiple Minima (MCMM) algorithms implemented in Macromodel (192) using the OPLS2005 force field (193) and a Generalized-Born/Surface-Area (GB/SA) water model (194). During this simulation, torsion angles of the macrocycles were

77 varied randomly using a Monte Carlo procedure (195) and the conformational states were selected based on the Metropolis algorithm (196) according to the Boltzmann probabilities for each state based on its internal energy. We performed 1,000,000 steps of Monte Carlo sampling using 100 steps per each rotatable bond with retention of unique conformers. A 21 kJ/mol energy window was applied to discard high energy conformers and we used a 0.5 Å maximum atom deviation cutoff to discard redundant conformers. An average structure was calculated for each peptide from their respective conformer families using Macromodel. Docking of cPRGDA, the Nε-lysine-acetylated amide-cPRGDK and the highly-studied cRGDf(N-Me)V peptide with the extracellular αVβ3 integrin protein (PDB entry 1L5G) (186) was performed using Glide 5.7 (Schrodinger Inc.). The average ligand conformations used in the docking studies were obtained from the conformational search detailed above.

Binding affinity

U87 glioblastoma cells were cultured in Corning 75 cm2 flasks in MEM containing 10% FBS. Cells were allowed to reach ~70 % confluence and were treated with trypsin containing EDTA for 15 minutes. The U87 cells (5x104 cells/ well) were co-incubated with various concentrations of the fluorescein-labeled cyclic peptides cPRGDA (13), cPRGDAA (14), cPRDGA (15) and amide-cPRGDK (23) (0 – 20 µM) on human vitronectin-coated 96-well plates (R&D Systems, Minneapolis USA) for 1 h at 37 ºC. After washing, U87 cells were stained with 0.5 % (v/v) crystal violet. The wells were washed and absorbance values were recorded at 627 nm in 0.1 M

HCl using a plate reader (187). IC50 values were calculated using Graph-Pad PRISM by a log(inhibitor) vs. response three parameter non-linear regression from xy scatter plots (bottom constraint at 18) of all normalized values from 3 independent experiments.

Confocal microscopy

The U87 glioblastoma cells served as a positive cell line for the αVβ3 integrin receptor while HT- 29 colon cancer cells served as a negative cell line since these express low levels of the human 2 αVβ3 integrin receptor (197). U87 cells and HT-29 cells were cultured in Corning 75 cm flasks in MEM and DMEM media, respectively, containing 10% FBS. Cells were allowed to reach ~70 % confluence and were treated with trypsin containing EDTA for 15 minutes. Cells (3x104 cells/ well) were transferred to 8-well Lab-Tek 155411 #1.0 borosilicate plates and were incubated with 10 μM solutions of fluorescein-labeled cPRGDA (13) and cPRGDAA (14) (0.5 % v/v DMSO in media) for 2 h at 37 °C. The fluorescein-labeled cPRDGA scramble peptide (15) was

78 used as a negative control while the fluorescein-labeled amide-cPRGDK peptide (23) served as a positive control. The targeting specificity of cPRGDA and cPRGDAA was investigated by blocking the αVβ3 integrin receptor of U87 cells with a 50-fold molar excess of the linear RGD peptide (20). Laser confocal scanning microscopy was then performed with excitation wavelength of 488 nm using 5 % laser power, PMT voltage of 760 V and 120 µM pinhole diameter.

Acknowledgements

Gang Zheng and Andrei K. Yudin thank NSERC and CIHR for financial support of this project.

79 Supporting Information

HPLC-MS characterization of the side chain-protected aziridine aldehyde- cyclized peptides cPRGDA (7), cPRGDAA (8) and cPRDGA (9) bonded to cysteamine by sulfhydryl aziridine ring-opening

Figure S2.2. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDAA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening (8) after prep-HPLC purification. A) Scanning the collected UV channel at 254 nm shows a > 70 % purity of the product (9.3 min) while the peak < 3.5 min shows the void volume. B) Analysis of the ESI-MS(+) at 9.3 min identifies the molar mass of the desired product (calculated molar mass = 1107.39 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society.

82 HPLC-MS characterization of the side chain-protected amide-cPRGDK (21) peptide after amide cyclization

83 HPLC-MS characterization of the side chain-deprotected aziridine aldehyde-cyclized peptides cPRGDA (13), cPRGDAA (14) and cPRDGA (15) bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to fluorescein

Figure S2.6. HPLC-MS spectral data confirming the identity and purity of the side chain- deprotected aziridine aldehyde-cyclized cPRGDAA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to fluorescein (14) after prep-HPLC purification. A) Scanning the collected UV channel at 441 nm shows a > 95 % purity of the product (8.1 min). The peak at ~ 2 min shows the void volume and the high baseline between 6 – 15 min is due to low absorbance values. B) Analysis of the UV-Vis spectrum at 8.1 min shows the characteristic spectrum of fluorescein in the desired product. C) Analysis of the ESI-MS(+) at 8.1 min identifies the molar mass of the desired product (calculated molar mass = 1157.25 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society.

86 HPLC-MS characterization of the side chain-deprotected amide-cPRGDK (23) peptide conjugated to fluorescein

87 HPLC-MS characterization of the side chain-protected aziridine aldehyde- cyclized peptide cPRGDA bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to DOTA (24) or pyro (25)

Figure S2.10. HPLC-MS spectral data confirming the identity and purity of the side chain- protected aziridine aldehyde-cyclized cPRGDA peptide bonded to cysteamine by sulfhydryl aziridine ring-opening and conjugated to pyro (25) after prep-HPLC purification. A) Scanning the collected UV channel at 663 nm shows a > 95 % purity of the product (11.6 min). B) Analysis of the UV-Vis spectrum at 11.6 min shows the characteristic spectrum of pyro in the desired product. C) Analysis of the ESI-MS(+) at 11.6 min identifies the molar mass of the desired product (calculated molar mass = 1552.94 g/mole). Reproduced with permission from (169). Copyright © 2012, American Chemical Society.

Chapter 3 Modulation of reactive oxygen species photogeneration of bacteriopheophorbide a derivatives by exocyclic E-ring- opening and charge modifications

Preamble

This chapter is comprised of the manuscript, ―Modulation of reactive oxygen species photogeneration of bacteriopheophorbide a derivatives by exocyclic E-ring opening and charge modifications‖ that was adapted from Roxin, Á., Chen, J., Paton, A. S., Bender, T. P., and Zheng, G. Journal of Medicinal Chemistry 2014, 57(1), 223-237, (163) Copyright © 2014, American Chemical Society, with permission from the American Chemical Society. I was the primary contributor to this publication. I developed the project‘s hypothesis, designed the chemical library and synthetic routes, synthesized and characterized all chemicals, designed and performed all chemical and biological experiments, proposed the computational experiments, analyzed and interpreted all results, and structured, wrote, and edited the majority of the manuscript for publication. Dr. Chen helped define the appropriate data processing methods for quantifying reactive oxygen species generation, offered advice for chemical synthesis, and helped edit the manuscript for publication. Dr. Paton performed the computational calculations, helped analyze the computational results, contributed to writing the computational sections of the manuscript, and helped edit the manuscript for publication. Dr. Bender offered advice for performing the computational calculations. Dr. Zheng helped design the chemical library, helped plan the chemical experiments, helped interpreting the results, and helped edit the manuscript for publication.

Introduction

Photosensitizers (PSs) are compounds that generate reactive oxygen species (ROS) upon photoirradiation. PSs have been used for anti-cancer and anti-microbial applications due to their ability to damage nearby cells by ROS generated through the process of photodynamic therapy (PDT) (100, 198-201). Among their qualifications, ideal PSs should be inactive until the time of irradiation, should be soluble in aqueous solutions and should efficiently generate ROS (99). In

90 addition, the wavelength of excitation should be within the PDT optical ‗window‘ (650 – 850 nm) to maximize the depth of tissue penetration and to minimize the absorption of excitation light by water and haemoglobin in vivo (198-200, 202). PSs can generate ROS by excited-state electron transfer to solution, molecular O2 and directly to cell substrates (type 1 PDT), and by 1 excited-state energy transfer to molecular O2 (type 2 PDT) to produce singlet oxygen ( O2) (202). Typical PSs undergo type 2 PDT (100); however, PSs that exclusively undergo type 2 PDT will become inactive when O2 is depleted from the local irradiated region (167, 203). To circumvent this dependence on O2, versatile PSs should also undergo type 1 PDT to generate radicals, peroxides and superoxide radical anions. This presents a challenge, since strategies of designing novel PSs with enhanced type 1 and type 2 PDT mechanisms are elusive (204).

Bacteriochlorins are a class of tetrapyrrole macrocycles with two unsaturated pyrrolic rings that typically have a Qy absorption band within the PDT optical window at ~750 nm in the near- infrared (NIR) range. This property allows in vivo tissue penetration to ~8 mm (107) which makes bacteriochlorins a useful class of PSs for PDT. Two clinically evaluated bacteriochlorins include 1a (WST09) and 1b (WST11) (Figure 3.1). Compound 1a is synthesized from palladium insertion into the precursor, bacteriopheophorbide a (4a) (153). Compound 1a is a highly-potent

PS (154, 156, 157, 205) with a triplet excited-state quantum yield (ΦT) of ~0.99 that primarily undergoes type 2 PDT in organic solutions (158). It is the precursor of 1b, which is synthesized by taurine-driven exocyclic E-ring-opening at the 131-carbon position of 1a (160, 161). Compound 1b is also a highly-potent PS (161, 206-209), and recent investigations found that 1b primarily undergoes type 1 PDT in aqueous solutions (162).

It is currently unclear why the dominant PDT mechanism of 1a (type 2) switches to type 1 PDT upon taurine-driven exocyclic E-ring-opening (1b). However, this observation suggested that i) exocyclic E-ring-opening and/or ii) variations in net charge may influence the PDT mechanisms of bacteriochlorins. To this end, we designed and synthesized a series of eight bacteriochlorophyll a derivatives (Figure 3.2) to investigate the influence of these two potential PDT-modulating factors on the photogeneration of ROS. These bacteriochlorins were chosen to investigate i) the effect of taurine-driven exocyclic E-ring-opening by pairing compounds with and without the exocyclic E-ring between the 13- and 15-carbon positions (2a vs. 2b, 3a vs. 3b, 4a vs. 4b, and 5a vs. 5b) and ii) the effect of net charge by conjugating charged moieties at the 173-carbon position of compounds with (2a - 5a) and without the exocyclic E-ring (2b - 5b).

The spectroscopic properties of the bacteriochlorin series were first characterized to elucidate the effect of taurine-driven exocyclic E-ring-opening on the absorbance and fluorescence properties of 2b-5b compared to 2a-5a, respectively. Density functional theory (DFT)-based molecular modeling (B3LYP/6-31G*) (109-111) and NMR spectroscopy were then performed to investigate if taurine-driven exocyclic E-ring-opening modulated molecular orbital energies and aromatic ring-current to alter the absorbance spectra. While the spectroscopic properties of 2a-5a and 2b-5b were explored, the primary goal of this study was to investigate if taurine-driven exocyclic E-ring-opening of bacteriochlorophyll a derivatives and net charge variations would modulate ROS photogeneration. Using the time-dependent extension of DFT (TDDFT) (210- 218) the vertical excited-state energy was calculated to determine whether 2a-5a and 2b-5b showed the potential for type 2 PDT ROS photogeneration (215-217). ROS-activated fluorescence probes were then used to experimentally elucidate the relative photogeneration of ROS in organic-based solution, aqueous-based solution and in cell culture media. These experimental results were then analyzed to delineate the effects of taurine-driven exocyclic E- ring opening and net charge variations on ROS photogeneration. The summary of our observations were then used to express structure-activity relationships, and to propose simple structural modifications for enhancing the ROS photogeneration of free-base bacteriochlorins.

93 Results

Synthesis of bacteriochlorins

The PS, bacteriopheophorbide a (4a), was synthesized according to literature procedures (219, 220). Briefly, bacteriochlorophyll a was extracted from a bacterial culture of Rhodobacter sphaeroides, was demetalated with dilute HCl (aq), and was subsequently hydrolyzed with 3 concentrated TFA (aq) to cleave the phytyl group at the 17 -carbon position to produce 4a. These methods gave 942 mg (1.54 mmol) of 4a, which was used as the precursor for all of the bacteriochlorins in this study (Figure 3.3). The net charge of the exocyclic E-ring-containing bacteriochlorins was adjusted by conjugating charge-modifying moieties at the 173-carbon position of 4a. The cationic choline moiety was conjugated to 4a on the 70 mg scale by esterification according to literature procedures (221) to synthesize 31-oxo-rhodobacteriochlorin 173-(2-trimethylaminoethyl)ester (2a) with a 14% yield (Figure 3.3, step v). The carboxylic acid moiety of 4a was esterified on the 40 mg scale with MeOH to synthesize the neutral 31-oxo- rhodobacteriochlorin 173-methyl ester (3a) with a 34% yield (Figure 3.3, step iv). The anionic taurine moiety was conjugated to 4a on the 25 mg scale by amide conjugation under mildly basic conditions to synthesize 31-oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide (5a) with a 54% yield (Figure 3.3, step iii). These reactions generated the bacteriochlorins 2a, 3a, 4a and 5a, which contained the exocyclic E-ring and varied depending on the charged moiety at the 173- carbon position. Using a combination of 1D 1H NMR, 2D COSY 1H NMR, 13C Jmod NMR and HPLC-MS, the identity and purity of the exocyclic E-ring-containing free-base bacteriochlorins 2a-5a were elucidated (Figure S3.1-S3.16).

Figure 3.3. Synthesis of the bacteriochlorins 1a-3a, 5a, and 1b-5b. Reaction conditions: (i)

Pd(OAc)2, sodium ascorbate, DCM, MeOH, Ar(g). (ii) Taurine, K2HPO4(aq) (pH 8.2), DCC,

DMAP, DMSO, N2(g). (iii) Taurine, K2HPO4(aq) (pH 8.2), HBTU, DMSO, Ar(g), 40 °C. (iv)

MeOH, HBTU, DMAP, DMSO, N2(g). (v) Choline chloride, HBTU, DMAP, DIPEA, DMSO,

Ar(g). (vi) Taurine, K2HPO4(aq) (pH 8.2), DCC, DMAP, DMSO, N2(g). (vii) Taurine, K2HPO4(aq)

(pH 8.2), HBTU, DMSO, Ar(g), 40 °C. (viii) MeOH, HBTU, DMAP, DMSO, Ar(g). (ix) Choline chloride, HBTU, DMAP, DIPEA, ACN, Ar(g). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

95 Taurine-driven exocyclic E-ring-opening at the 131-carbon position of 4a was achieved according to literature procedures (160, 161) with the addition of DCC and DMAP. Taurine was conjugated to 4a under mildly basic conditions on the 80 mg scale to synthesize 31-oxo-15- methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) with a 37% yield (Figure 3.3, step vi). Compound 4b was then used as the precursor for the synthesis of all other free-base bacteriochlorins which lacked the exocyclic E-ring due to taurine conjugation at the 131-carbon position. The cationic choline moiety was conjugated to 4b on the 40 mg scale by esterification at the 173-carbon position to synthesize 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131-(2-sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b) with a 46% yield (Figure 3.3, step ix). The 173-carbon position of 4b was esterified on the 20 mg scale with MeOH to synthesize 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide-173-methyl ester (3b) with a 59% yield (Figure 3.3, step viii). The anionic taurine moiety was conjugated to 4b on the 39 mg scale by amide conjugation at the 173-carbon position under mildly basic conditions to synthesize 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131,173-di(2-sulfoethyl)amide (5b) with a 28% yield (Figure 3.3, step vii). These reactions generated the bacteriochlorins 2b, 3b, 4b and 5b which contained a taurine moiety at the 131-carbon position due to exocyclic E-ring-opening and varied depending on the charged moiety at the 173-carbon position. A combination of 1D 1H NMR, 2D COSY 1H NMR, 13C Jmod NMR and HPLC-MS was used to elucidate the identity and purity of the free-base bacteriochlorins 2b-5b which were synthesized by taurine-driven exocyclic E-ring-opening (Figures S3.17-S3.32).

Compound 4a was used as a precursor for the synthesis of 1a and 1b. These two palladium- containing bacteriochlorins were used as controls when assaying the photogeneration of ROS. Compound 1a was synthesized by palladium insertion into 4a on the 70 mg scale according to literature procedures (153) with a 69% yield (Figure 3.3, step i). Compound 1b was synthesized by taurine-driven exocyclic E-ring-opening at the 131-carbon position of 1a according to literature procedures (160, 161) with the addition of DCC and DMAP (Figure 3.3, step ii). This reaction generated the product, 1b, as well as the chlorin analogue of 1b due to oxidation of an unsaturated pyrrolic ring. This required purification by HPLC and, consequently, gave a very low (~ 1%) yield. A combination of 1D 1H NMR, 2D COSY 1H NMR, 13C Jmod NMR and HPLC-MS (Figure S3.33) was used to elucidate the identity and purity of 1a, whereas HPLC-MS was used to characterize 1b (Figure S3.34).

96 Spectroscopic properties of bacteriochlorins

The UV-vis absorption spectra of 2a-5a and 2b-5b were recorded in MeOH (Table 3.1) to investigate if the Qx-bands (500 – 550 nm) and Qy-bands (725 – 800 nm) of 2a-5a shifted upon taurine-driven exocyclic E-ring-opening (2b-5b). The results showed that the Qy-bands of 2b-5b are slightly red-shifted (λmax = 749, 750 nm) compared to 2a-5a (λmax = 744, 748 nm), while the

Qx-bands of 2b-5b are consistently blue-shifted (λmax = 517 – 518 nm) by 9 – 10 nm compared to

2a-5a (λmax = 527, 528 nm). Molecular modeling at the DFT level was subsequently performed (Tables S3.1 and S3.2), using the B3LYP functional and the 6-31G* basis set as implemented in SPARTAN ‘06, to calculate the HOMO to LUMO and the HOMO−1 to LUMO energy gaps of

2a-5a and 2b-5b (109, 110), to explain the minor Qy-band shifts and the consistent Qx-band shifts, respectively (111) observed upon taurine-driven exocyclic E-ring-opening (Table 3.1). The calculated HOMO to LUMO energy gaps of the derivatives containing the exocyclic E-ring (2a-5a) were slightly greater (0.02387 – 0.08016 eV) than that of their respective derivatives without the exocyclic E-ring (2b-5b), which could not explain the bathochromic Qy-band shifts. However, the low magnitude of the HOMO to LUMO energy gaps (less than 0.1 eV) was consistent to the minor Qy-band shifts (less than 5 nm). On the other hand, taurine-driven exocyclic E-ring-opening consistently increased the energy gaps of HOMO−1 to LUMO between 2a and 2b (0.11 eV), 3a and 3b (0.15 eV), 4a and 4b (0.06 eV), and 5a and 5b (0.14 eV), respectively, which resulted in the consistent hypsochromic shifts of the Qx-bands of 2b-5b compared to 2a-5a.

Qy λmax Qx λmax λem HOMO to HOMO − 1 to a a b c d d Compound (nm) (nm) (nm) Φf LUMO gap (eV) LUMO gap (eV)

2a 745 527 769 0.04 2.05 2.47

2b 749 517 760 0.03 2.10 2.58

3a 748 527 765 0.02 2.05 2.47

3b 749 517 757 0.01 2.13 2.62

4a 744 527 773 0.03 2.05 2.47

4b 749 518 771 0.04 2.08 2.53

5a 747 528 776 0.03 2.05 2.48

5b 750 518 761 0.02 2.13 2.61

a. MeOH. b. MeOH (λex = 525 nm). c. MeOH (λex = 525 nm; λem = 700 – 900 nm), using 1a as a standard (Φf = 0.004) (222). d. B3LYP/6-31G* (SPARTAN ‘06).

Compound Exocyclic E-ring status H21-NH H23-NH

2a Present -0.82 +0.70

2b Not-present -1.32 -1.32

3aa Present -0.95 +0.48

3b Not-present -1.32 -1.31

4a Present -0.96 +0.47

4b Not-present -1.30 -1.30

5a Present -0.84 +0.68

5b Not-present -1.29 -1.28

a. CDCl3

The local electron densities at the C-ring, adjacent to the exocyclic E-ring of 2a-5a and adjacent to the conjugated taurine moiety at the 131-carbon position of 2b-5b, were further investigated by 1H NMR analysis to determine if taurine-driven exocyclic E-ring-opening modulated aromatic ring currents to increase the HOMO−1 to LUMO gap energies of 2b-5b compared to 2a-5a. It was found that the H23-NH protons (δ = +0.70 ppm to +0.47 ppm) were consistently de-shielded compared to the H21-NH protons (δ = -0.82 ppm to -0.96 ppm) in 2a-5a that contained the coplanar electron-withdrawing ketone substituent (115) within the exocyclic E-ring (Table 3.2). Upon taurine-driven exocyclic E-ring-opening, the H21-NH and H23-NH protons became shifted up-field (δ = -1.28 ppm to -1.32 ppm) in the spectra of 2b-5b due to the relatively increased shielding at the adjacent C-ring. This showed that the amide substituent at the 131 carbon position in 2b-5b displayed a lower electron-withdrawing effect on the H23-NH protons compared to the coplanar ketone substituent within the exocyclic E-ring of 2a-5a. These results suggest that the local electron density at the C-ring increased between compounds 2a and 2b, 3a and 3b, 4a and 4b, and between 5a and 5b. Therefore, these local electron density enhancements

99 modulated aromatic ring currents and contributed to the increased HOMO−1 to LUMO energy gaps and the hypsochromic Qx-band shifts of 2b-5b compared to 2a-5a (Figure 3.4).

The fluorescence properties of 2a-5a and 2b-5b were investigated in MeOH to elucidate the effect of taurine-driven exocyclic E-ring-opening on the fluorescence spectra and the fluorescence quantum yields (Φf) (Table 3.1). Analysis of the fluorescence spectra showed that the maximum emission wavelengths of the bacteriochlorins are consistently blue-shifted upon

100 taurine-driven exocyclic E-ring-opening: 9 nm (2a → 2b), 8 nm (3a → 3b), 2 nm (4a → 4b), and 15 nm (5a → 5b) Consequently, their Stokes shifts are slightly decreased: 13 nm (2b vs. 2a), 9 nm (3b vs. 3a), 7 nm (4b vs. 4a), and 18 nm (5b vs. 5a). These fluorescence results suggest that taurine-driven exocyclic E-ring-opening had a minor effect on the singlet excited- state rigidity (223-228). Further investigations of the Φf‘s of 2a-5a (Φf = 0.02 – 0.04) and 2b-5b

(Φf = 0.01 – 0.04) showed that taurine-driven exocyclic E-ring-opening had little effect on Φf. Therefore, potential variations of ROS photogeneration of 2a-5a compared to 2b-5b would not be primarily due to modulations of the quantum yields of internal conversion (ΦIC) (223-228) or

Φf.

Molecular modeling of potential PDT activity

Time-dependent DFT (TDDFT) methods were employed to calculate the vertical excitation energies of 2a-5a and 2b-5b in order to estimate whether the compounds were photodynamically active. According to literature procedures, the use of TDDFT with the B3LYP functional and the 6-31G* basis set provides a useful estimate of the energy difference between the singlet ground- state (S0) and the vertical triplet excited-state (212, 213, 216). We optimized the geometry using symmetry unrestricted DFT methods at B3LYP/6-31G*, then performed single-point energy calculations using TDDFT methods at the same level of theory to determine the vertical excitation energies of 2a-5a and 2b-5b (214-218). The calculations were performed in the gas phase, since it has been shown that for excitation calculations at this level the effects of the solvent are negligible (210, 211). These calculations were used to investigate the potential of these bacteriochlorins to generate type 2 PDT ROS. Type 2 PDT require the vertical singlet- triplet excitation energy gap (ΔEST) of PSs to be greater than 0.98 eV to transfer energy from 3 1 their triplet excited-state to ground-state molecular oxygen ( O2) to produce singlet oxygen ( O2) ST (215-217). These ΔE calculations showed that 2a-5a and 2b-5b were each capable of undergoing type 2 PDT (Table S3.3), and that these ΔEST values of the compounds without the ST exocyclic E-ring (2b-5b: ΔE = 1.5445 – 1.6218 eV) are consistently higher than that of ST compounds with the exocyclic E-ring (2a-5a; ΔE = 1.4926 – 1.5172 eV) (Figure 3.5).

101

ROS photogeneration detected with AUR.

We next experimentally investigated the relative ROS photogeneration of 2a-5a and 2b-5b using the ROS-activated fluorescent probe, Amplex® UltraRed (AUR) (229-233). Compounds 1a and 1b were used as controls. It was found that the known hydrogen peroxide quencher, dimethyl thiourea (DMTU) (234-236), the known superoxide quencher, 1,2-dihydroxybenzene-3,5- disulfonic acid disodium salt hydrate (tiron) (237-239), and the known singlet oxygen quencher, sodium azide (240, 241), could quench the fluorescence of AUR while irradiating compound 4a with a 740 nm LED light box in (70:17.5:12.5) H2O:cremophor:DMF (v/v) (Figure S3.35A), indicating that AUR could detect both type 1 (superoxide and hydrogen peroxide) and type 2 1 ( O2) PDT ROS.

A report by Vakrat-Haglili et al. showed that the known type 2 PDT PS, 1a, generates significant amounts of type 1 PDT ROS in aqueous solutions, but not in organic solutions (158). This suggests that 1a would generate different proportions of type 1 and type 2 PDT ROS depending on the aqueous content of tested solutions. Therefore, our PDT ROS experiments were conducted in a MeOH-based solution, in H2O with high concentrations of detergents, and in cell culture media appropriate for many cancer cell lines. These solvents were primarily chosen to investigate the ROS photogeneration of 2a-5a and 2b-5b in a variety of solvents that differed in

102 aqueous content. These investigations included the use of cell culture media to mimic ROS photogeneration under biocompatible in vitro solution conditions.

AUR was used to investigate ROS photogeneration in (70:30) MeOH:PBS (v/v), (70:17.5:12.5)

H2O:cremophor:DMF (v/v), and (97.5:2:0.5) RPMI-1640 cell culture media:DMSO:cremophor

(v/v). The optical density (OD) of the bacteriochlorins was matched with OD740nm = 0.2, and the compounds were irradiated with a 740 nm LED light box to determine the relative photogeneration of ROS. The photobleaching of PSs was monitored (λAbs = 740 nm) after each light treatment (ΔAbsPS) (242-245) while concomitantly monitoring increasing AUR fluorescence (ΔFlrAUR) to express the relative ROS photogeneration of each PS as the linear 2 slopes (R = 0.859 – 0.9999) of the corrected AUR fluorescence (ΔFlrAUR/ ΔAbsPS) over the light dose range according to literature methods (245-248). Solutions required either high concentrations of MeOH or detergents to fully dissolve the bacteriochlorins in this series. The particular concentrations of DMSO (2%, v/v) and cremophor (0.5%, v/v) in cell culture media were chosen because these concentrations did not induce toxicity to A549 non-small cell lung cancer cells, as indicated by MTT analysis (Figure S3.36). Consequently, the hydrophobic bacteriochlorins, 3a and 4a, were restricted from the cell culture media studies due to aggregation observed by spectral red-shifting of Qy absorption bands beyond 800 nm (Figure S3.37).

Our AUR-based analysis involved pairing derivatives that varied based on exocyclic E-ring status (2a vs. 2b, 3a vs. 3b, 4a vs. 4b, and 5a vs. 5b). The results showed that exocyclic E-ring- opening enhanced the photogeneration of ROS for 2b compared to 2a in all the tested solution conditions. Specifically, there was a 102% increase (p < 0.005) in the MeOH-based solution

(Figure 3.6A), a 47% increase (p < 0.005) in H2O with high detergent concentrations (Figure 3.6B), and a 52% increase (p < 0.005) in cell culture media (Figure 3.6C) for 2b compared to 2a. The photogeneration of ROS was also enhanced for 4b compared to 4a in the two tested solutions including a 33% increase (p < 0.05) in the MeOH-based solution (Figure 3.6A) and a

42% increase (p < 0.005) in H2O with high detergent concentrations (Figure 3.6B). The photogeneration of ROS also increased for 3b compared to 3a by 39% (p < 0.005) in the MeOH- based solution (Figure 3.6A) and for 5b compared to 5a by 18% (p < 0.01) in H2O with high detergent concentrations (Figure 3.6B). In summary, there was a general trend observed wherein taurine-driven exocyclic E-ring-opening enhanced the photogeneration of ROS, especially for compounds 2b (compared to 2a) and 4b (compared to 4a) in solutions containing significant

103 proportions of organic solvent or detergents. A trend was also observed for the palladium- containing compounds, 1a and 1b. Specifically, the photogeneration of ROS increased for 1b compared to 1a by 36% (p < 0.001) in H2O with high detergent concentrations (Figure 3.6B), and by 64% (p < 0.001) in cell culture media (Figure 3.6C), while no difference was observed in the MeOH-based solution (Figure 3.6A).

104

Figure 3.6. Photogeneration of ROS by bacteriochlorins with (1a-5a) and without (1b-5b) the exocyclic E-ring indicated by the linear slopes of corrected AUR fluorescence (ΔFlrAUR/ΔAbsPS) over light dose range (ΔLight dose) after 740 nm irradiation in MeOH-based solution (A, light 2 dose = 0 – 100 mJ; R > 0.986), in H2O with high detergent concentrations (B, light dose = 0 – 1500 mJ; R2 > 0.983) and in cell culture media (C, light dose = 0 – 100 mJ; R2 = 0.859 – 0.945) (N = 3, error bar show ± STDERR; * p < 0.05, ** p < 0.01, *** p < 0.005). Compounds 3a and 4a were excluded from assays in cell culture media (C) due to their observed aggregation (Figure S3.37). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

105 The aforementioned AUR fluorescence results were analyzed to delineate the influence of exocyclic E-ring status and net charge on the photogeneration of ROS in the three aforementioned solutions. Analysis showed that taurine-driven exocyclic E-ring-opening significantly enhanced the photogeneration of ROS (2b-5b > 2a-5a) in the MeOH-based solution by 40% (p < 0.005), in H2O with high detergent concentrations by 25% (p < 0.005), and in cell culture media by 29% (p < 0.005). In addition, the net charge of these bacteriochlorins influenced ROS photogeneration in all the tested solutions, whereby increasing net negative charge significantly enhanced ROS photogeneration in the MeOH-based solution (Figure 3.7A, p

< 0.01), in H2O with high detergent concentrations (Figure 3.7B, p < 0.005), and in cell culture media (Figure 3.7C, p < 0.005). Analysis of the predicted 95% confidence interval (CI) ranges and slopes (fluorescence vs. charge) of the regressions showed that the net negative charge was most influential for enhancing ROS photogeneration in cell culture media (Figure 3.7C, 95% CI range = ± 0.2128; slope = -0.11), followed by H2O with high detergent concentrations (Figure 3.7B, 95% CI range = ± 0.1816; slope = -0.08), and the organic-based solution (Figure 3.7A, 95% CI range = ± 0.2548; slope = -0.08).

106

Figure 3.7. Photogeneration of ROS by bacteriochlorins indicated by the linear slopes of corrected AUR fluorescence (ΔFlrAUR/ΔAbsPS; normalized to 1b) over light dose range (ΔLight dose) after 740 nm irradiation in MeOH-based solution (A, light dose = 0 - 100 mJ; R2 > 2 0.9841), in H2O with high detergent concentrations (B, light dose = 0 – 1500 mJ; R > 0.9816) and in cell culture media (C, light dose = 0 – 100 mJ; R2 = 0.8327 – 0.9303) compared to the net charge of derivatives with (2a-5a) and without (2b-5b) the exocyclic E-ring (N = 3, dashed lines show predicted 95% CI). Compounds 3a and 4a were excluded from assays in cell culture media (C) due to their observed aggregation (Figure S3.37). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

107 ROS photogeneration detected with SOSG

The ROS photogeneration of 1a-5a and 1b-5b was investigated using the ROS-activated fluorescent probe, Singlet Oxygen Sensor Green® (SOSG) (240, 245, 249-252). As DMTU, tiron, and sodium azide could quench the fluorescence of SOSG while irradiating compound 4a with a 740 nm LED light box in (70:17.5:12.5) H2O:cremophor:DMF (v/v) (Figure S3.35B), SOSG could be an indicator to determine the presence of both type 1 (superoxide and hydrogen 1 peroxide) and type 2 ( O2) PDT ROS. This was in agreement with a previous report that showed that SOSG can detect superoxide, hydrogen peroxide and hydroxyl radicals (253). SOSG was subsequently used to investigate the ROS photogeneration of the aforementioned bacteriochlorins (OD740nm = 0.2) in (70:30) MeOH:PBS (v/v), (70:17.5:12.5)

H2O:cremophor:DMF (v/v), and (97.5:2:0.5) RPMI-1640 cell culture media:DMSO:cremophor (v/v) after irradiation with a 740 nm LED light box. The photobleaching of PSs was monitored

(λAbs = 740 nm) after each light treatment (ΔAbsPS) (242-245) while concomitantly monitoring increasing SOSG fluorescence (ΔFlrSOSG) to express relative ROS photogeneration of each PS as 2 the linear slopes (R = 0.92 – 0.999) of the corrected SOSG fluorescence (ΔFlrSOSG/ΔAbsPS) over the light dose range according to literature methods (245-248).

Our SOSG-based analysis involved pairing derivatives that varied based on exocyclic E-ring status (2a vs. 2b, 3a vs. 3b, 4a vs. 4b, and 5a vs. 5b). These investigations showed that taurine- driven exocyclic E-ring-opening consistently enhanced the photogeneration of ROS for 2b compared to 2a in all the tested solution conditions. Specifically, there was an 80% increase (p <

0.01) in the MeOH-based solution (Figure 3.8A), a 36% increase (p < 0.005) in H2O with high detergent concentrations (Figure 3.8B), and a 76% increase (p < 0.005) in cell culture media (Figure 3.8C) for 2b compared to 2a. The photogeneration ROS was also consistently enhanced for 5b compared to 5a in all of the tested solution conditions. This included a 75% increase (p <

0.05) in the MeOH-based solution (Figure 3.8A), a 28% increase (p < 0.01) in H2O with high detergent concentrations (Figure 3.8B), and a 26% increase in cell culture media (Figure 3.8C, p < 0.005) for 5b compared to 5a. The photogeneration of ROS was also enhanced for 3b compared to 3a in the MeOH-based by 81% (Figure 3.8A, p < 0.05), and for 4b compared to 4a by 43% (p < 0.005) in H2O with high detergent concentrations (Figure 3.8B). In summary, there was a general trend observed wherein taurine-driven exocyclic E-ring-opening enhanced the photogeneration of ROS, especially for compounds 2b (compared to 2a) and 5b (compared to 5a) in all of the tested solutions. A trend was also observed for the palladium-containing

108 compounds, 1a and 1b. Specifically, the photogeneration of ROS increased for 1b compared to

1a by 32% (p < 0.001) in H2O with high detergent concentrations (Figure 3.8B), and by 112% (p < 0.05) in cell culture media (Figure 3.8C), while no difference was observed in the MeOH- based solution (Figure 3.8A).

109

Figure 3.8. Photogeneration of ROS by bacteriochlorins with (1a-5a) and without (1b-5b) the exocyclic E-ring indicated by the linear slopes of corrected SOSG fluorescence

(ΔFlrSOSG/ΔAbsPS) over light dose range (ΔLight dose) after 740 nm irradiation in MeOH-based 2 solution (A, light dose = 0 – 250 mJ; R = 0.92 – 0.999), in H2O with high detergent concentrations (B, light dose = 0 – 1000 mJ; R2 > 0.992) and in cell culture media (C, light dose = 0 – 750 mJ; R2 > 0.986) (N = 3, error bar show ± STDERR; * p < 0.05, ** p < 0.01, *** p < 0.005). Compounds 3a and 4a were excluded from assays in cell culture media (C) due to their observed aggregation (Figure S3.37). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

110 The aforementioned SOSG fluorescence results were analyzed to delineate the influence of exocyclic E-ring status and net charge on the photogeneration of ROS in the three aforementioned solutions. Analysis showed that taurine-driven exocyclic E-ring-opening significantly enhanced the photogeneration of ROS (2b-5b > 2a-5a) in the MeOH-based solution by 68% (p < 0.005), in H2O with high detergent concentrations by 25% (p < 0.005), and in cell culture media by 51% (p < 0.005). However, the net charge of these bacteriochlorins did not consistently influence ROS photogeneration in all the tested solutions. Analysis of the predicted 95% CI ranges and slopes (fluorescence vs. charge) of the regressions showed that increasing net negative charge enhanced ROS photogeneration in H2O with high detergent concentrations (Figure 3.9B, p < 0.005; 95% CI range = ± 0.1736; slope = -0.07) and in cell culture media (Figure 3.9C, p < 0.01; 95% CI range = ± 0.5434; slope = -0.15), but was not influential in the organic-based solution (Figure 3.9A, p = 0.26; 95% CI range = ± 0.3792; slope = -0.04).

111

Figure 3.9. Photogeneration of ROS by bacteriochlorins indicated by linear slopes of corrected

SOSG fluorescence (ΔFlrSOSG/ΔAbsPS; normalized to 1b) over light dose range (ΔLight dose) after 740 nm irradiation in MeOH-based solution (A, light dose = 0 – 250 mJ; R2 > 0.9978), in 2 H2O with high detergent concentrations (B, light dose = 0 – 1000 mJ; R > 0.9827) and in cell media (C, light dose = 0 – 750 mJ; R2 > 0.9846) compared to the net charge of derivatives with (2a-5a) and without (2b-5b) the exocyclic E-ring (N = 3, dashed lines show predicted 95% CI). Compounds 3a and 4a were excluded from assays in cell culture media (C) due to their observed aggregation (Figure S3.37). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

112 Discussion

The primary objective of our studies was to determine whether taurine-driven exocyclic E-ring- opening and net charge would influence the photogeneration of ROS by free-base bacteriopheophorbide a derivatives. To this end, eight bacteriochlorins were synthesized that varied in structure due to the presence (2a-5a) or absence (2b-5b) of the exocyclic E-ring, and due to net charge (+1, 0, -1 and -2) variations as a result of moieties conjugated through the 173- carbon position and by the taurine moiety conjugated through the 131-carbon position upon exocyclic E-ring-opening. Compound 4a was synthesized from bacteriochlorophyll a and was used as the precursor for the synthesis of 1a, 2a, 3a, 5a, and 1b-5b. The procedures used to synthesize the bacteriochlorins 1a-5a and 2b-5b were facile and generally high-yielding. The spectroscopic properties, the calculated molecular energies, and the ROS photogeneration of 2a- 5a and 2b-5b were then studied to elucidate structure-activity relationships as a result of taurine- driven exocyclic E-ring-opening and net charge variations.

Analysis of the absorbance spectra of 2a-5a and 2b-5b showed that taurine-driven exocyclic E- ring-opening resulted in consistent hypsochromic shifts of the Qx-bands of 2b-5b compared to 2a-5a. Proton NMR analysis showed a consistent increase in the shielding of the H23-NH protons of 2b-5b compared to 2a-5a. This suggests that taurine-driven exocyclic E-ring-opening resulted in enhancing the local electron density at the C-ring of 2b-5b compared to the C-ring adjacent to the coplanar ketone moiety within the exocyclic E-ring of 2a-5a. We propose that these local electron density enhancements modulated the aromatic ring current of compounds 2b- 5b. DFT-based molecular modeling was then used to calculate the energy gaps between the HOMO-1 and LUMO energy levels, since this energy gap reflects the relative maximum wavelength of the Qx absorption bands of free-base bacteriochlorins (111). These calculations showed that taurine-driven exocyclic E-ring-opening consistently increased the HOMO-1 to LUMO energy gaps of 2b-5b compared to 2a-5a. Therefore, we propose that taurine-driven exocyclic E-ring-opening was responsible for the hypsochromic shifts of the Qx absorption bands of 2b-5b, compared to 2a-5a, as a result of the enhanced local electron densities of the adjacent C-rings of 2b-5b. Though these 1H NMR and DFT studies explain the distinct absorbance spectral shifts of 2b-5b compared to 2a-5a, we cannot directly suggest how variations in the local electron densities at the C-ring of 2b-5b can modulate the photophysical properties of 2a- 5a. However, an analogous investigation by Monteiro et al. concluded that electron withdrawing substituents can reduce the ΦT of meso-tetraphenyl-substituted free-base bacteriochlorins (254).

113 Ding et al. presented a similar report of enhancements of type 1 PDT ROS photogeneration by meso-tetra(hydroxyphenyl) porphyrin due to the encapsulation of the porphyrin within electron- donating micelles (255). The summary of these observations suggest that the inclusion of electron-donating groups and the avoidance of electron-withdrawing groups may potentially enhance the photogeneration of type 1 and/or type 2 PDT ROS of bacteriochlorins.

The ΔEST of 2a-5a and 2b-5b were calculated to theoretically predict if these bacteriochlorins 1 would be capable of generating O2 via type 2 PDT using TDDFT molecular modeling. It is generally accepted that, if the ΔEST is calculated to exceed 0.98 eV (corresponding to 94 kJ/mol or 22.5 kcal/mol), the chromophore has the potential to undergo excited triple-state energy 3 1 transfer with molecular O2 to form O2 (215-217). A report by Rodgers et al. presents an 1 intriguing example where a Si-naphthalocyanine was shown to reversibly generate O2 while possessing a ΔEST energy ~0.5 kcal/mol below the theoretical threshold (256). While such calculations have their limitations, these studies proved useful for our purposes. The results 1 showed that each bacteriochlorin had the potential to generate O2 by transferring their triplet 3 ST excited-state energy to triplet ground-state molecular oxygen ( O2). In addition, since the ΔE energies were greater for 2b-5b compared to the corresponding 2a-5a, we predicted that taurine- driven exocyclic E-ring-opening could potentially modulate the photogeneration of ROS. Analysis of our experimental PDT ROS studies found that taurine-driven exocyclic E-ring- opening (2b-5b vs. 2a-5a) consistently enhanced the photogeneration of ROS in each of the tested solutions using both AUR and SOSG-based analysis. In light of our ROS-specificity studies, and due to the known capture of interrelated type 1 PDT ROS (202) by AUR (229-233) and SOSG (253), we propose that a combination of superoxide, hydrogen peroxide, hydroxyl 1 radicals and O2 were detected during our ROS assays. We observed the most consistent ROS photogeneration enhancements for 2b, 4b, and 5b compared to 2a, 4a and 5a, respectively. The largest enhancements of ROS photogeneration was observed with 2b compared to 2a (AUR and SOSG-based studies) in all of the tested solutions. Our structure-activity relationships corroborate a report by Joshi et al. that showed that exocyclic E-ring-opening consistently enhanced the in vitro PDT activity of three pairs of neutral free-base bacteriochlorins containing ketone functional groups either at the B-ring or at the D-ring (121). In light of our experimental PDT ROS results, it is possible that the in vitro PDT activity enhancements described by Joshi et al. were the result of enhanced ROS photogeneration due to exocyclic E-ring-opening. These

114 findings suggest that exocyclic E-ring-opening may generally enhance the ROS photogeneration of free-base bacteriochlorins to potentially improve in vitro PDT activity.

The palladium-containing controls, 1a and 1b, also showed an interesting trend wherein 1b generated higher levels of ROS compared to 1a in aqueous-based solutions, but not in the 1 MeOH-based solution. It has previously been shown that 1a efficiently produces O2 in organic solution (158), while 1b primarily produces superoxide and hydroxyl radicals in aqueous media 1 (162). The formation of O2 via type 2 PDT is limited by the presence of dissolved and diffusing ground-state molecular O2 (167, 203). The type 1 PDT mechanisms, however, can involve electron transfer to molecular O2 to form superoxide, which can continued to react with H2O to form hydrogen peroxide (202). Therefore, the ROS generation of 1b was higher compared to 1a in aqueous-based solutions in our report likely due to the increased photogeneration of type 1

PDT ROS by 1b through a cascade of reactions involving both O2 and H2O in the aqueous-based solutions (202).

Our analysis also showed that varying the charge of the bacteriochlorins influenced the photogeneration of ROS, whereby increasing net negative charge consistently resulted in an increased photogeneration of ROS in the aqueous-based solutions (AUR and SOSG-based studies). Silva et al. have previously identified the sulfonic acid functional groups of the meso- tetra(2,6-dichloro-3-sulfonatophenyl) bacteriochlorin (TDCPBSO3H) as the key sites of excited- state electron transfer for type 1 PDT (141). This finding suggests that the sulfonic acid functional group of taurine on 5a and 2b-5b, and possibly the carboxylic acid functional groups of compounds 4a and 4b, were involved during excited-state electron transfer to generate type 1 PDT ROS. Therefore, we propose that the anionic functional groups of 4a, 5a, and 2b-5b were influential for enhancing ROS photogeneration compared to the cationic compound, 2a, in aqueous solutions.

We calculated the theoretical potential for type 2 PDT ROS generation of 2a-5a and 2b-5b, and experimentally observed variations in the ROS photogeneration by these PSs. While we observed that taurine-driven exocyclic E-ring-opening and increasing the net negative charge enhanced ROS photogeneration, we cannot conclude the effects these structural modifications had on the photophysical properties of 2a-5a and 2b-5b. However, since taurine-driven exocyclic E-ring- opening had little effect on the Φf and the Stokes shifts of 2a-5a compared to 2b-5b, we can infer that the observed variations of ROS photogeneration were likely the result of modulations of the

115 quantum yields of intersystem crossing (ΦISC), ΦT, and/or singlet oxygen formation (ΦΔ), but not due to major differences in ΦIC (223-228). Extensive studies involving ESR analysis, transient 1 absorption spectroscopy, and O2 luminescence would further elucidate the role taurine-driven exocyclic E-ring-opening and charge variations have on the photophysical properties of bacteriochlorins to truly conclude how our aforementioned structural modifications resulted in enhancements of ROS photogeneration. Such experiments could continue to refine structure- activity relationships that would facilitate the design of novel bacteriochlorin PSs for efficient type 1 and type 2 PDT.

Conclusions

We synthesized a series of eight free-base bacteriochlorophyll a derivatives to develop simple strategies of enhancing the photogeneration of ROS. TDDFT molecular modeling calculations 1 predicted that each derivative would potentially generate O2 upon photoirradiation. Further experimental investigations validated these predictive calculations, and found that taurine-driven exocyclic E-ring-opening and increasing net negative charge generally enhanced ROS photogeneration in aqueous solutions. These structure-activity relationships are meant to aid current PDT research by providing simple strategies of designing novel bacteriochlorins that may efficiently generate type 1 and type 2 PDT ROS. These strategies may be particular useful when the type 1 PDT mechanism is required for anti-cancer or anti-microbial PDT applications under hypoxic or anoxic conditions.

Materials and Methods

Materials

Bacterial cultures of R. sphaeroides were purchased from Frontier Scientific Inc. The reagents taurine, sodium ascorbate, palladium acetate, choline chloride, N,N’-dicyclohexylcarbodiimide (DCC), O-benzotriazole-N,N,N',N'-tetramethyl-uronium-hexafluoro-phosphate (HBTU), 4-(N,N- dimethylamino)pyridine (DMAP) and N,N-diisopropylethylamine (DIPEA) were purchased from

Sigma Aldrich and used without further purification. Deuterated NMR solvents, DMSO-d6 and

CDCl3, were purchased from Cambridge Isotope Laboratories. The ROS-activated fluorescent probes Amplex® UltraRed (AUR) and Singlet Oxygen Sensor Green® (SOSG) were purchased from Invitrogen. The ROS quenchers, N,N'-dimethyl thiourea (DMTU), 1,2-dihydroxybenzene- 3,5-disulfonic acid disodium salt hydrate (tiron), and sodium azide were purchased from Sigma

116 Aldrich and used without further purification. The MTT reagent (3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyltetrazolium bromide) was purchased from Invitrogen-Gibco.

General chemistry information

All synthetic reactions were performed in round-bottom flasks using magnetic stirring while sealed with rubber septa and under either Ar (g) or N2 (g). Rotary evaporation was performed while heating solutions at 40 °C and appropriately adjusting vacuum pressures for various solvents. Normal-phase column chromatography was performed during the synthesis of 4a using Merck Grade 60 silica gel (70-230 mesh size, 60 Å). Compounds 2a, 3a, 5a, and 2b-5b were purified using a Biotage Isolera One flash chromatography system with a 25 g Biotage C-18 cartridge under reverse-phase conditions (gradient of 0–100% ACN in 0.1% TFA over 12 minutes using a flow of 40 mL/ minute) while monitoring eluted products at 357 nm and byproducts at 200 nm. Compound 1b was initially purified with the aforementioned Biotage system under reverse- phase conditions (gradient 0 – 100% ACN in 5 mM K2HPO4 (aq) over 12 minutes using a flow of 40 mL/minute) while monitoring the eluted product at 383 nm, and was further purified using a Waters 2695 HPLC with a 25 cm C-18 column under reverse-phase conditions (gradient of 20 –

100% ACN with 0.1 M TEAA (aq) over 40 minutes using a flow of 0.5 mL/minute) while monitoring the eluted fractions of 1b (λabs = 517 nm and 750 nm) and the chlorin analogue of 1b

(λabs = 628 nm) using a Waters 2996 photodiode array detector (200 – 800 nm). The identity and purity of all compounds (1a-5a and 1b-5b) was assessed using the aforementioned HPLC system with a Waters Delta Pak C18, 5 μm 3.9 x 150 mm column under reverse-phase conditions (gradient of 0 – 100% acetonitrile in 0.1% TFA over 15 minutes using a flow of 0.8 mL/minute) while monitoring eluted samples with the aforementioned photodiode array detector (λabs = 750 nm) and a Waters Micromass ZQ mass spectrometer (200 – 2000 m/z) in ESI-positive mode. HPLC-MS analysis showed that all compounds were ≥ 95% pure. The identity of 1a-5a and 2b- 5b was also assessed by 1D 1H NMR, 2D COSY 1H NMR, and 13C Jmod NMR analysis using a 1 Bruker Ultrashield 400 Plus 400 MHz NMR with either DMSO-d6 or CDCl3. Proton H NMR analysis showed that compounds were ≥ 95% pure, except if stated otherwise (see 3b and 4a). The solubility of all compounds (in PBS with 2.0% DMSO (v/v) and 0.5% cremophor, (v/v)) was assessed using a Varian 50 Bio UV−visible spectrophotometer. The fluorescence spectra of all bacteriochlorins were recorded in MeOH using a Horiba Jobin Yvon Fluoromax-4 spectrofluorometer. The comparative ROS photogeneration assays involved irradiating all

117 compounds using a 740 nm LED light box and recording absorbance values and fluorescence intensities using a Molecular Devices Spectra Max M5 plate reader.

Palladium bacteriopheophorbide a (1a)

Compound 1a was synthesized according to literature procedures (153). The precursor, 4a (70 mg, 114.8 μmol or 1.0 equiv) and Pd(OAc)2 (56 mg, 252.6 μmol or 2.2 equiv) were dissolved in 42 mL of (1:5) DCM:MeOH. Sodium ascorbate (140 mg, 711.8 μmol or 6.2 equiv) was suspended in the solution and the mixture was purged with Ar (g) for 2 minutes. The mixture was then stirred in the dark at room temperature overnight. Chloroform (40 mL) was added and the mixture was washed four times with 30 mL of sat. NaCl (aq). The organic layer was then dried using Na2SO4, was filtered and was concentrated by rotary evaporation. The crude was redissolved in 42 mL of (1:5) DCM:MeOH and the previously described amounts of Pd(OAc)2 and sodium ascorbate were added. The mixture was then stirred at room temperature in the dark overnight. The same extraction was repeated and the same reaction conditions and extraction was performed a total of five times. This yielded 56.7 mg (79.4 μmol) of 1a (69% yield). ESI+MS: + 1 [M+H] = 715 m/z, UV-vis (MeOH, λmax): 756, 533, 383, 329 nm. H NMR (400 MHz, CDCl3): δ (ppm) = 9.21 (s, 1H, 5-H), 8.52 (s, 1H, 10-H), 8.47 (s, 1 H, 20-H), 5.93 (s, 1H, 132-H), 4.38 (m, 1 3 1H, 18-H), 4.36 (m, 1H, 7-H), 4.08 (m, 2H, 17-H + 8-H), 3.87 (s, 3H, 12 -CH3), 3.46 (s, 3H, 15 - 1 2 2 OCH3), 3.38 (s, 3H, 2 -CH3), 3.08 (s, 3H, 3 -COCH3), 2.52 (m, 2H, 17a -CH2), 2.35 (m, 2H, 17b1-H + 17a2-H), 2.20 (m, 2H, 17b2-H + 8a1-H), 2.07 (m, 1H, 8b1-H), 1.77 (d, 3H, J = 7.2 Hz, 1 1 2 13 18 -CH3), 1.67 (d, 3H, J = 7.0 Hz, 7 -CH3), 1.07 (t, 3H, J = 7.3 Hz, 8 -CH3). C NMR (100

MHz, CDCl3): δ (ppm) = 198.21, 187.56, 169.57, 159.13, 158.02, 151.05, 150.03, 144.69, 141.33, 140.71, 140.01, 136.24, 134.87, 129.70, 126.60, 126.41, 109.22, 102.32, 100.94, 98.05, 64.38, 54.34, 52.85, 48.58, 46.77, 34.34, 32.88, 32.13, 30.12, 23.23, 23.19, 20.84, 13.98, 12.11, 10.26.

Palladium 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (1b)

The precursor, 1a (24.7 mg, 34.5 μmol or 1.0 equiv) was dissolved in 3.7 mL of DMSO. DCC (50.3 mg, 241.8 μmol or 7.0 equiv) and DMAP (14.8 mg, 241.8 μmol or 7.0 equiv) was then dissolved in the DMSO solution. A 1.235 mL solution was prepared using 34.5 mg (274.3 µmol or 7.94 equiv) of taurine in 1M K2HPO4 (aq) with a pH of 8.2 according to literature procedures (160, 161). The solutions were combined and the resulting suspension was stirred in the dark at

118 room temperature for 4 days while constantly bubbling N2 (g). The crude was purified using a reverse-phase Biotage system. The collected fraction (eluted at ~ 50% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. The dried crude was dissolved in 200 μL of DMSO and was purified by reverse-phase HPLC. The product was quantified spectrophotometrically, the identity was characterized using ESI+MS and UV-vis spectroscopy, and the purity was found to be >95% using HPLC-MS. This yielded 0.21 mg (250 + nmol) of 1b (0.7% yield). ESI+MS: [M] = 840 m/z, UV-vis (MeOH, λmax): 748, 517, 385, 332 nm.

31-Oxo-rhodobacteriochlorin 173-(2-trimethylaminoethyl)ester (2a)

Compound 2a was synthesized according to literature procedures (221). The precursor, 4a (70 mg, 114.8 µmol or 1.0 equiv) was dissolved in 3.4 mL of DMSO. HBTU (88.9 mg, 229.6 µmol or 2.0 equiv), DMAP (29.9 mg, 229.6 µmol or 2.0 equiv) and DIPEA (99.2 µL, 574 µmol or 5.0 equiv) were then added to the DMSO solution. Choline chloride (63 mg, 574 µmol or 5.0 equiv) was dissolved in 63 μL of H2O and was added to the DMSO solution. The mixture was purged with Ar (g) for 2 minutes and was stirred in the dark at room temperature for 117 h. The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at ~75% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 11 mg (15.7 µmol) of the product 2a (14% yield). ESI+MS: [M+H]+ = 698 m/z. UV-vis 1 (MeOH, λmax): 745, 527, 357 nm. H NMR (400 MHz, DMSO-d6): δ (ppm) = 8.95 (s, 1H, 5-H), 2 8.72 (s, 1H, 10-H), 8.66 (s, 1 H, 20-H), 6.14 (s, 1H, 13 -H), 4.35 (m, 2H, choline CH2), 4.33 (m, 1 1H, 7-H), 4.27 (m, 1H, 18-H), 4.01 (m, 1H, 8-H), 3.81 (m, 1H, 17-H), 3.76 (s, 3H, 12 -CH3), 3 1 2 3.53 (m, 2H, choline CH2), 3.47 (s, 3H, 15 -OCH3), 3.33 (s, 3H, 2 -CH3), 3.13 (s, 3H, 3 - 1 1 COCH3), 2.98 (s, 9H, choline 3CH3) 2.32 (m, 1H, 8 -CH), 2.30 (m, 2H, 17 -H), 2.18 (m, 2H, 2 1 1 1 17 -H), 2.03 (m, 1H, 8 -CH), 2.01 (m, 2H, 17 -H), 1.72 (d, 3H, J = 7.1 Hz, 18 -CH3), 1.66 (d, 1 2 3H, J = 7.1 Hz, 7 -CH3), 1.00 (t, 3H, J = 7.3 Hz, 8 -CH3), 0.70 (s, 1H, 23-NH), -0.82 (s, 1H, 21- 13 NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 199.39, 188.93, 172.36, 172.01, 171.54, 169.67, 163.73, 157.83, 147.81, 139.02, 138.47, 137.35, 133.67, 128.14, 120.02, 108.51, 100.00, 97.44, 97.15, 64.26, 64.11, 58.29, 54.28, 53.29, 53.09, 50.34, 49.37, 48.55, 40.68, 33.69, 30.81, 29.86, 29.17, 23.11, 22.92, 13.40, 11.47, 10.90.

119 31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin-131-(2- sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b)

The precursor, 4b (39.2 mg, 53.2 µmol or 1.0 equiv) was dissolved in 11.2 mL of ACN. HBTU (41.7 mg, 106.4 µmol or 2.0 equiv), DMAP (19.7 mg, 106.4 µmol or 2.0 equiv) and DIPEA (91.9 µL, 127.5 µmol or 10.0 equiv) were then added to the ACN solution. Choline chloride

(29.2 mg, 266 µmol or 5.0 equiv) was dissolved in 29 µL of H2O and was added to the ACN solution. The solution was purged with Ar (g) for 1 minute and was stirred in the dark at room temperature for 20 h. The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at ~60% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 20 mg (24.3 µmol) of the product 2b (46% yield). + 1 ESI+MS: [M] = 822 m/z. UV-vis (MeOH, λmax): 749, 517, 354 nm. H NMR (400 MHz,

DMSO-d6): δ (ppm) = 9.30 (s, 1H, 5-H), 8.93 (s, 1H, 10-H), 8.90 (s, 1H, taurine NH), 8.77 (s, 1 H, 20-H), 5.34 (d, 1H, J = 19.3 Hz, 14a2-H), 5.13 (d, 1H, J = 17.3 Hz, 14b2-H), 4.39 (m, 2H, choline CH2), 4.35(m, 1H, 7-H), 4.33 (m, 1H, 18-H), 4.22 (m, 1H, 17-H), 4.20 (m, 1H, 18-H), 1 3 3.80 (m, 2H, taurine CH2), 3.65 (s, 3H, 12 -CH3), 3.58 (s, 3H, 15 -OCH3), 3.57 (m, 2H, choline 1 2 CH2), 3.24 (s, 3H, 2 -CH3), 3.16 (s, 3H, 3 -COCH3), 3.02 (s, 9H, choline 3CH3), 2.92 (m, 2H, 2 2 1 taurine CH2), 2.66 (m, 1H, 17a -H), 2.41 (m, 1H, 17b -H), 2.33 (m, 1H, 8 -CH), 2.06 (m, 1H, 1 1 1 1 17a -H), 2.04 (m, 1H, 8 -H), 1.80 (d, 3H, J = 6.9 Hz, 7 -CH3), 1.55 (d, 3H, J = 6.8 Hz, 18 -CH3), 1 2 13 1.49 (m, 1H, 17b -H), 0.99 (t, 3H, J = 7.3 Hz, 8 -CH3), -1.32 (s, 2H, 21-NH + 23-NH). C NMR

(100 MHz, DMSO-d6): δ (ppm) = 198.43, 173.00, 172.62, 167.70, 167.63, 167.19, 166.28, 163.42, 134.37, 134.04, 132.22, 131.73, 131.57, 131.32, 128.02, 105.25, 98.24, 97.96, 96.57, 64.04, 58.32, 57.02, 53.30, 53.00, 52.40, 50.76, 47.18, 46.10, 37.15, 33.56, 30.99, 29.68, 29.36, 24.01, 23.58, 13.73, 11.82, 10.90.

31-Oxo-rhodobacteriochlorin 173-methyl ester (3a)

The precursor, 4a (40 mg, 65.6 µmol or 1.0 equiv) was dissolved in 5 mL of DMSO. HBTU (51.4 mg, 131.2 µmol or 2.0 equiv) and DMAP (25.7 mg, µmol or 2.0 equiv) was then dissolved in the DMSO solution. Methanol (10 mL) was added to the DMSO solution and the reaction was stirred in the dark for 3 h while gently bubbling N2 (g). The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at 100% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 14 mg (22.4 + µmol) of the product 3a (34% yield). ESI+MS: [M+H] = 625 m/z. UV-vis (MeOH, λmax): 748, 1 527, 357 nm. H NMR (400 MHz, CDCl3): δ (ppm) = 8.99 (s, 1H, 5-H), 8.50 (s, 1H, 10-H), 8.43

120 (s, 1 H, 20-H), 6.10 (s, 1H, 132-H), 4.31 (m, 1H, 18-H), 4.28 (m, 1H, 7-H), 4.04 (m, 1H, 17-H), 1 3 1 4.03 (m, 1H, 8-H), 3.86 (s, 3H, 12 -CH3), 3.61 (s, 3H, 15 -OCH3), 3.51 (s, 3H, 2 -CH3), 3.46 (s, 2 4 2 1 3H, 3 -COCH3), 3.18 (s, 3H, 17 -OCH3), 2.51 (m, 2H, 17 -CH2), 2.34 (m, 1H, 8 -H), 2.26 (m, 1 1 1 1 1H, 17a -H), 2.08 (m, 1H, 17b -H), 2.06 (m, 1H, 8 -H), 1.80 (d, 3H, J = 7.1 Hz, 18 -CH3), 1.74 1 2 (d, 3H, J = 7.2 Hz, 7 -CH3), 1.12 (t, 3H, J = 7.3 Hz, 8 -CH3), 0.48 (s, 1H, 23-NH), -0.95 (s, 1H, 13 21-NH). C NMR (100 MHz, CDCl3): δ (ppm) = 199.10, 189.04, 173.31, 171.04, 169.53, 163.65, 157.92, 148.09, 139.07, 138.29, 136.83, 136.25, 133.28, 128.66, 121.36, 108.03, 99.62, 97.64, 95.78, 64.35, 54.97, 52.78, 51.67, 50.61, 49.71, 48.87, 33.32, 30.95, 30.16, 29.93, 22.91, 22.85, 13.40, 11.51, 10.75.

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin-13'-(2- sulfoethyl)amide-173-methyl ester (3b)

The precursor, 4b (20 mg, 27.1 µmol or 1.0 equiv) was dissolved in 2 mL of DMSO. HBTU (21 mg, 54.3 µmol or 2.0 equiv), and DMAP (6.5 mg, 54.3 µmol or 2.0 equiv) were dissolved in the DMSO solution. Methanol (2 mL) was added to the DMSO solution, the mixture was purged with Ar (g) for 2 minutes, and was stirred in the dark for 21 h. The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at ~ 70% ACN (v/v)) was concentrated by rotary evaporation and dried using a speed-vac. This yielded 12 mg (16 µmol) of the product 3b (59% yield). Compound 3b was assessed to be ~94% pure by 1H NMR analysis. + 1 ESI+MS: [M+H] = 751 m/z. UV-vis (MeOH, λmax): 749, 517, 356 nm. H NMR (400 MHz,

DMSO-d6): δ (ppm) = 9.30 (s, 1H, 5-H), 8.93 (s, 1H, 10-H), 8.91 (s, 1H, taurine N-H), 8.76 (s, 1H, 20-H), 5.31 (d, 1H, J = 18.7 Hz, 14a2-H), 5.14 (d, 1H, J = 18.5 Hz, 14b2-H), 4.33 (m, 1H,

7-H), 4.31 (m, 1H, 18-H), 4.22 (m, 1H, 17-H), 4.17 (m, 1H, 8-H), 3.79 (m, 2H, taurine CH2), 1 3 4 1 3.65 (s, 3H, 12 -CH3), 3.58 (s, 3H, 15 -OCH3), 3.56 (s, 3H, 17 -OCH3), 3.24 (s, 3H, 2 -CH3), 2 2 2 3.16 (s, 3H, 3 -COCH3), 2.92 (m, 2H, taurine CH2), 2.63 (m, 1H, 17a -H), 2.61 (m, 1H, 17b -H), 2.40 (m, 1H, 17a1-H), 2.32 (m, 1H, 81-CH), 2.06 (m, 1H, 81-H), 2.03 (m, 1H, 17b1-H), 1.81 (d, 1 1 2 3H, J = 7.0 Hz, 18 -CH3), 1.53 (d, 3H, J = 6.8 Hz, 7 -CH3), 0.99 (t, 3H, J = 7.2 Hz, 8 -CH3), - 13 1.31 and -1.32 (each s, 21-NH + 23-NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 198.43, 173.58, 172.99, 167.68, 167.65, 167.33, 166.18, 163.49, 134.36, 134.05, 134.01, 132.13, 131.72, 131.57, 131.36, 128.00, 105.29, 98.28, 97.91, 96.50, 57.04, 53.12, 52.32, 51.77, 50.84, 47.18, 46.10, 37.12, 33.79, 33.55, 30.94, 29.66, 29.40, 24.00, 23.50, 13.73, 11.78, 10.89.

121 Bacteriopheophorbide a (4a)

Compound 4a was synthesized according to literature procedures (219, 220) and was used as the precursor for the synthesis of all other bacteriochlorins in this study. A 450 mL volume of R. sphaeroides was processed to yield 942 mg (1.54 mmol) of 4a. HPLC-MS analysis showed one distinct eluted product with the corresponding [M+H] value of 4a (Figure S3.12), while 1H NMR revealed the presence of a stereoisomer of 4a (~ 10%) with shifts of 71-H, 82-H, 132-H, 181-H at + chiral positions (Figure S3.9). ESI+MS: [M+H] = 611 m/z, UV-vis (MeOH, λmax): 744, 527, 357 1 nm. H NMR (400 MHz, DMSO-d6): δ (ppm) = 8.95 (s, 1H, 5-H), 8.72 (s, 1H, 10-H), 8.63 (s, 1 H, 20-H), 6.17 (s, 1H, 132-H), 4.34 (m, 1H, 7-H), 4.25 (m, 1H, 18-H), 3.99 (m, 1H, 8-H), 3.82 1 3 1 (m, 1H, 17-H), 3.79 (s, 3H, 12 -CH3), 3.45 (s, 3H, 15 -OCH3), 3.32 (s, 3H, 2 -CH3), 3.11 (s, 3H, 2 2 1 2 1 3 -COCH3), 2.35 (m, 3H, 17a -H + 8 -CH2), 2.18 (m, 1H, 17b -H), 2.03 (m, 2H, 17 -CH2), 1.71 1 1 2 (d, 3H, J = 6.9 Hz, 18 -CH3), 1.64 (d, 3H, J = 7.0 Hz, 7 -CH3), 1.02 (t, 3H, J = 6.8 Hz, 8 -CH3), 13 0.65 (s, 1H, 23-NH), -0.85 (s, 1H, 21-NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 199.37, 188.97, 174.43, 171.82, 171.55, 169.71, 163.64, 158.37, 147.85, 138.89, 138.45, 138.40, 137.28, 133.53, 128.13, 119.98, 108.67, 99.87, 97.30, 97.24, 64.23, 54.32, 53.03, 50.52, 49.42, 48.53, 33.66, 31.06, 29.84, 29.52, 23.10, 22.99, 13.39, 11.46, 10.89.

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (4b)

The precursor, 4a (80 mg, 131.2 µmol or 1.0 equiv) was dissolved in 12 mL of DMSO. DCC (190.4 mg, 918.4 µmol or 7.0 equiv) and DMAP (56 mg, 918.4 µmol or 7.0 equiv) were then dissolved in the DMSO solution. A 4 mL solution was prepared using 131.2 mg (1.042 mmol or

7.94 equiv) of taurine in 1M K2HPO4 (aq) with a pH of 8.2 according to literature procedures (160, 161). The solutions were mixed and the resulting suspension was stirred in the dark at room temperature for 10 days with constant gentle N2 (g) bubbling. The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at ~70% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 36.2 mg (49.1 + µmol) of the product 4b (37% yield). ESI+MS: [M+H] = 737 m/z. UV-vis (MeOH, λmax): 749, 1 518, 354 nm. H NMR (400 MHz, DMSO-d6): δ (ppm) = 9.30 (s, 1H, 5-H), 8.93 (s, 1H, 10-H), 8.93 (s, 1H, taurine NH), 8.75 (s, 1 H, 20-H), 5.33 (d, 1H, J = 20.0 Hz, 14a2-H), 5.22 (d, 1H, J = 18.1 Hz, 14b2-H), 4.33 (m, 1H, 7-H), 4.31(m, 1H, 18-H), 4.21 (m, 1H, 17-H), 4.20 (m, 1H, 8-H), 1 3 1 3.76 (m, 2H, taurine CH2), 3.65 (s, 3H, 12 -CH3), 3.58 (s, 3H, 15 -OCH3), 3.24 (s, 3H, 2 -CH3), 2 2 1 3.15 (s, 3H, 3 -COCH3), 2.93 (m, 2H, taurine CH2), 2.54 (m, 1H, 17a -H), 2.39 (m, 2H, 8 -CH +

122 17b2 -H), 2.24 (m, 1H, 17a1-H), 2.04 (m, 1H, 17b1-H), 2.02 (m, 1H, 81-H), 1.81 (d, 3H, J = 7.0 1 1 1 2 Hz, 18 -CH3), 1.54 (d, 3H, J = 6.8 Hz, 7 -CH3), 1.48 (m, 1H, 17 -H), 0.98 (t, 3H, J = 7.0 Hz, 8 - 13 CH3), -1.30 (s, 2H, 21-NH + 23-NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 198.43, 174.85, 173.05, 167.73, 167.67, 167.59, 166.22, 163.52, 134.40, 134.06, 134.01, 132.15, 131.77, 131.49, 131.30, 127.92, 105.38, 98.32, 97.89, 96.45, 57.04, 53.24, 52.30, 50.86, 47.20, 46.07, 37.13, 33.56, 32.66, 31.08, 29.66, 29.46, 24.01, 23.58, 13.74, 11.79, 10.90.

31-Oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide (5a)

The precursor, 4a (25 mg, 16.4 µmol or 1.0 equiv) was dissolved in 3.75 mL of DMSO. HBTU (111 mg, 286.1 µmol or 7.0 equiv) was then dissolved in the DMSO solution. A 1.25 mL solution was prepared using 41.2 mg (326.8 µmol or 8.0 equiv) of taurine in 1M K2HPO4 (aq) with a pH of 8.2 according to literature procedures (160, 161). The solutions were combined and the resulting suspension was purged with Ar (g) for 2 minutes. The mixture was then stirred in the dark at 40 ˚C for 22 h. The crude was then purified using a reverse-phase Biotage system. The collected fraction (eluted at ~75% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 16 mg (22.2 µmol) of the product 5a (54% yield). + 1 ESI+MS: [M+H] = 719 m/z, UV-vis (MeOH, λmax): 747, 528, 357 nm, H NMR (400 MHz,

DMSO-d6): δ (ppm) = 8.95 (s, 1H, 5-H), 8.73 (s, 1H, 10-H), 8.63 (s, 1 H, 20-H), 7.62 (s, 1H, taurine NH), 6.17 (s, 1H, 132-H), 4.31 (m, 1H, 7-H), 4.26 (m, 1H, 18-H), 3.99 (m, 1H, 8-H), 3.79 1 3 1 (s, 3H, 12 -CH3), 3.76 (m, 1H, 17-H),3.47 (s, 3H, 15 -OCH3), 3.33 (s, 3H, 2 -CH3), 3.24 (m, 2H, 2 1 taurine CH2), 3.12 (s, 3H, 3 -COCH3), 2.71 (m, 2H, taurine CH2), 2.44 (m, 2H, 17 -CH2) 2.34 (m, 1H, 81-CH), 2.10 (m, 1H, 17a1-H), 2.02 (m, 1H, 17b1-H), 1.97 (m, 1H, 81-H), 1.72 (d, 3H, J 1 1 2 = 7.3 Hz, 18 -CH3), 1.64 (d, 3H, J = 6.8 Hz, 7 -CH3), 1.01 (t, 3H, J = 7.3 Hz, 8 -CH3), 0.68 (s, 13 1H, 23-NH), -0.84 (s, 1H, 21-NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 199.38, 189.04, 171.766, 171.714, 171.573, 169.77, 163.59, 158.60, 147.87, 138.92, 138.477, 138.422, 137.28, 133.47, 128.12, 119.93, 108.62, 99.82, 97.32, 97.25, 64.17, 54.32, 53.02, 51.01, 50.81, 49.52, 48.52, 35.97, 33.66, 32.82, 30.33, 29.82, 23.10, 22.95, 13.41, 11.47, 10.90.

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131,173-di(2- sulfoethyl)amide (5b)

The precursor, 4b (38.5 mg, 52.2 µmol or 1.0 equiv) was dissolved in 5.775 mL of DMSO. HBTU (138.7 mg, 365.7 µmol or 7.0 equiv) was then dissolved in the DMSO solution. A 1.925 mL solution was prepared using 51.9 mg (414.8 µmol or 7.94 equiv) of taurine in 1M K2HPO4

123

(aq) with a pH of 8.2 according to literature procedures (160, 161). The solutions were combined and the resulting suspension was purged with Ar (g) for 2 minutes. The mixture was stirred in the dark at 40 ˚C for 19 h. The crude was then purified using a reverse-phase Biotage system (with a 25 g C-18 Biotage cartridge using a 20–40% ACN gradient in 0.1% TFA over 12 column volumes and a flow of 40 mL/ minute) while monitoring impurities at 200 nm and collecting the product by monitoring it at 357 nm. The collected fraction (eluted at ~40% ACN (v/v)) was concentrated by rotary evaporation and was dried using a speed-vac. This yielded 9.5 mg (10.7 + µmol) of the product 5b (28% yield). ESI+MS: [M+H] = 844 m/z. UV-vis (MeOH, λmax): 750, 1 518, 353 nm. H NMR (400 MHz, DMSO-d6): δ (ppm) = 9.29 (s, 1H, 5-H), 8.92 (s, 1H, 10-H), 8.90 (s, 1H, taurine NH), 8.73 (s, 1 H, 20-H), 7.82 (s, 1H, taurine NH), 5.31 (d, 1H, J = 18.6 Hz, 14a2-H), 5.10 (d, 1H, J = 18.6 Hz, 14b2-H), 4.31 (m, 1H, 7-H), 4.21 (m, 1H, 18-H), 4.08 (d, 1H, 1 J = 9.5 Hz, 17-H), 3.96 (m, 1H, 8-H), 3.81 (m, 2H, taurine CH2), 3.67 (s, 3H, 12 -CH3), 3.58 (s, 3 1 2 3H, 15 -OCH3), 3.32 (m, 2H, taurine CH2), 3.24 (s, 3H, 2 -CH3), 3.15 (s, 3H, 3 -COCH3), 2.95 2 2 (m, 2H, taurine CH2), 2.56 (m, 2H, taurine CH2), 2.53 (m, 1H, 17a -H), 2.39 (m, 1H, 17b -H), 2.37 (m, 1H, 81-CH), 2.04 (m, 1H, 81-H), 2.02 (m, 2H, 171-H), 1.81 (d, 3H, J = 7.1 Hz, 181- 1 2 CH3), 1.54 (d, 3H, J = 6.9 Hz, 7 -CH3), 0.99 (t, 3H, J = 7.2 Hz, 8 -CH3), -1.27 and -1.28 (each s, 13 2H, 21-NH + 23-NH). C NMR (100 MHz, DMSO-d6): δ (ppm) = 198.44, 172.98, 171.85, 168.05, 167.72, 167.51, 166.19, 163.63, 134.42, 134.07, 133.98, 132.13, 131.68, 131.45, 131.30, 127.87, 105.34, 98.33, 97.87, 96.36, 57.06, 53.28, 52.33, 51.06, 50.87, 47.36, 46.07, 37.02, 35.98, 33.55, 32.95, 30.31, 29.65, 24.00, 23.56, 13.76, 11.79, 10.92.

ROS photogeneration detected with AUR and SOSG

The PSs were dissolved in 100 μL of (70:30) MeOH:PBS, (70:17.5:12.5) H2O:cremophor:DMF or (97.5:2:0.5) RPMI-1640 cell culture media:DMSO:cremophor on sterile black-sided, clear- bottomed Costar 96-well plates (N = 3 for each solution). Compounds 3a and 4a were excluded from assays in cell culture media due to their observed aggregation in 97.5% (v/v) aqueous solution containing 2.0% DMSO (v/v) and 0.5% cremophor (v/v) (Figure S37). Stock solutions of AUR (1 mg/ 340 μL DSMO) and SOSG (1 mg/ 220 μL DMF) were prepared, and 1 μL aliquots of either AUR or SOSG stock solutions were added to each of the wells (25 μM AUR/well or 25 μM SOSG/well) containing the PSs (and with either AUR or SOSG alone as controls). The absorbance of each sample was recorded at 740 nm to ensure the absorbance of all wells were matched (OD740nm = 0.2) after blanking (using AUR or SOSG alone as baseline). The fluorescence of AUR (λex = 550 nm; λem = 581 nm) and SOSG (λex = 485nm; λem = 536 nm) was

124 then recorded to determine the baseline intensities of AUR and SOSG. Plates were then irradiated with a 740 nm LED light box (12.2 mW/cm2 fluence rate). After irradiation, the absorbance of all samples was again recorded at 740 nm to determine the relative amount of PSs

(ΔAbsPS) (242-245) remaining after photobleaching by the generated PDT ROS (136, 257-260). The fluorescence of AUR and SOSG was concomitantly recorded to determine the increase of activated AUR and SOSG due to the trapping of ROS generated by the irradiated PSs. These fluorescence values were then blanked according to the baseline fluorescence of AUR and SOSG to determine the increasing fluorescence of AUR (ΔFlrAUR) and SOSG (ΔFlrSOSG) for each light dose treatment. Since both the photobleaching of the PSs (243, 259, 260) and the increased fluorescence intensities of the ROS-indicators illustrated the relative amounts of ROS that were generated during the irradiation of the PSs, these two metrics were combined to express the relative ROS photogeneration as the slopes of increasing corrected ROS-indicator fluorescence

(ΔFlrSOSG/AUR/ΔAbsPS) over light dose range (ΔLight dose) according to literature methods (244- 247). These slopes were normalized to the slope of the control, 1b, in each of the tested solutions to illustrate the relationship between relative ROS photogeneration and net charge variations.

Computational studies

The DFT methods were implemented in SPARTAN ‘06 (Wavefunction Inc., Irvine, CA). The compounds were first geometry optimized at the DFT level using the B3LYP functional and 6- 31G* basis set (109, 110), with no symmetry restrictions and in the gas phase (210, 211). From these calculations, the energy levels for all compounds were extracted (110). Following this, symmetry unrestricted, gas-phase TDDFT single point energy calculations were performed for the triplet excited-state, using the ground-state optimized geometry. The vertical excitation energies for the compounds were obtained from this calculation (210-212, 215-218).

Acknowledgments

Gang Zheng thanks Natural Sciences and Engineering Research Council of Canada, Canadian Institute of Health Research, Canadian Cancer Society Research Institute, Canadian Foundation for Innovation, Joey and Toby Tanenbaum/Brazilian Ball Chair in Prostate Cancer Research for funding.

125

Supporting Information

31-Oxo-rhodobacteriochlorin 173-(2-trimethylaminoethyl)ester (2a)

126

127

128

Figure S3.4. HPLC-MS characterization of 31-oxo-rhodobacteriochlorin 173-(2- trimethylaminoethyl)ester (2a) showing the A) purity, B) UV-vis spectrum and C) ESI+ mass spectrum. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

129

31-Oxo-rhodobacteriochlorin 173-methyl ester (3a)

Figure S3.5. 1D 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-methyl ester (3a) in

130

131

Figure S3.7. 13C Jmod NMR spectrum of 31-oxo-rhodobacteriochlorin 173-methyl ester (3a) in

132

133

Bacteriopheophorbide a (4a)

134

135

136

137

31-Oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide (5a)

Figure S3.13. 1D 1H NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide

138

139

Figure S3.15. 13C Jmod NMR spectrum of 31-oxo-rhodobacteriochlorin 173-(2-sulfoethyl)amide

140

141

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131-(2- sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b)

Figure S3.17. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 131-(2-sulfoethyl)amide-173-(2-trimethylaminoethyl)ester (2b) in DMSO- d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

142

143

144

145

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide-173-methyl ester (3b)

Figure S3.21. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 3 rhodobacteriochlorin 13'-(2-sulfoethyl)amide-17 -methyl ester (3b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

146

147

Figure S3.23. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 3 rhodobacteriochlorin 13'-(2-sulfoethyl)amide-17 -methyl ester (3b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

148

149

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (4b)

Figure S3.25. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- rhodobacteriochlorin 13'-(2-sulfoethyl)amide (4b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

150

151

152

153

31-Oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 131,173-di(2- sulfoethyl)amide (5b)

Figure S3.29. 1D 1H NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 1 3 rhodobacteriochlorin 13 ,17 -di(2-sulfoethyl)amide (5b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

154

155

Figure S3.31. 13C Jmod NMR spectrum of 31-oxo-15-methoxycarbonylmethyl- 1 3 rhodobacteriochlorin 13 ,17 -di(2-sulfoethyl)amide (5b) in DMSO-d6. Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

156

157 Palladium bacteriopheophorbide a (1a)

158 Palladium 31-oxo-15-methoxycarbonylmethyl-rhodobacteriochlorin 13'-(2- sulfoethyl)amide (1b)

159 Computational studies

Compound HOMO-2 HOMO-1 HOMO LUMO LUMO+1 LUMO+2

2a -8.17953 -7.00678 -6.59002 -4.53714 -3.13176 -3.02793

2b -7.9201 -6.83746 -6.35049 -4.25309 -3.07732 -2.93377

3a -6.6453 -5.26289 -4.84438 -2.79644 -1.37442 -1.03604

3b -5.93055 -5.00416 -4.51613 -2.38803 -1.11175 -0.49105

4a -6.69122 -5.31478 -4.89649 -2.84424 -1.42788 -1.09453

4b -6.48566 -5.11357 -4.65848 -2.58236 -1.22259 -0.69743

5a -6.8348 -5.47594 -5.04836 -2.99653 -1.5721 -1.26011

5b -5.99796 -5.0584 -4.56927 -2.44363 -1.16886 -0.54126 a. B3LYP/6-31G*, SPARTAN ‗06.

160 Table S3.2. TD-DFT-based molecular modeling calculations of the triplet excited-state energies a for optimized S0 geometries (eV) of 2a – 5a and 2b – 5b Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

Compound HOMO-2 HOMO-1 HOMO LUMO LUMO+1 LUMO+2

2a -7.18738 -7.08666 -5.43999 -3.29768 -3.18868 -2.92926

2b -7.02299 -6.86101 -5.20589 -3.17227 -3.07492 -2.76299

3a -5.46517 -5.33101 -3.71311 -1.5664 -1.21791 -0.65004

3b -5.18081 -5.01543 -3.34191 -1.30194 -0.71197 0.11501

4a -5.51611 -5.38268 -3.76135 -1.61776 -1.27425 -0.69489

4b -5.30323 -5.1522 -3.51903 -1.40786 -0.85912 -0.07739

5a -5.67548 -5.54115 -3.91687 -1.76911 -1.44209 -0.84722

5b -5.23525 -5.06861 -3.39405 -1.35865 -0.75386 0.03247 a. B3LYP/6-31G*, SPARTAN ‗06.

Compound E-ring status ΔEST (eV)

2a Present 1.5008

2b Not-present 1.5850

3a Present 1.5172

3b Not-present 1.6213

4a Present 1.4926

4b Not-present 1.5445

5a Present 1.4984

5b Not-present 1.6218 a. B3LYP/6-31G*, SPARTAN ‗06.

ROS-specificity of AUR and SOSG using bacteriopheophorbide a (4a)

The ROS-specificity of AUR and SOSG were investigated to determine if each could specifically detect either type 1 (hydrogen peroxide and superoxide) or type 2 (singlet oxygen)

PDT ROS. The compound 4a (~30 μM) was dissolved in in (70:17.5.12.5) H2O:cremophor:DMF alone, with the type 1 PDT ROS quenchers DMTU (10 mM for hydrogen peroxide) and tiron (10 mM for superoxide), with the type 2 PDT ROS quencher sodium azide (10 mM for singlet oxygen) and with both the aforementioned type 1 and type 2 PDT ROS quenchers. Either AUR or SOSG was added to each of the described samples alone and with each sample containing compound 4a. All samples (N=3) were irradiated on Costar black-sided, clear-bottomed 96-well plates (100 μL total volume) using a 740 nm LED light box (12.17 mW/cm2 fluence rate) to deliver sequential fluences of 0.5 J/cm2, 1.0 J/cm2, 2.0 J/cm2 and 3.0 J/cm2. The fluorescence of the ROS probes AUR (λex 550 nm, λem 581 nm) and SOSG (λex 485 nm, λem 535 nm) were then

162 recorded for all samples and the fluorescence of each sample was subtracted from the baseline of each pair that did not contain compound 4a. The amount of remaining 4a was estimated by recording the absorbance at 740 nm after each subsequent irradiation and normalizing values to the absorbance of 4a at 740 nm without irradiation. The final corrected ROS probe fluorescence was then calculated by compensating for the degree of estimated photobleaching to compare the fluorescence of each ROS probe at each subsequent fluence. These results showed that all of the ROS quenchers could reduce the fluorescence of AUR and SOSG (Figure S3.35A and S3.35B). This suggests that either AUR and SOSG are not completely specific to type 1 and type 2 PDT ROS, respectively, or that the aforementioned type 1 and type 2 PDT ROS quenchers did not specifically quench their respective ROS. Therefore, we could not conclude that AUR and SOSG could specifically detect their respective ROS. We can, however, conclude that the combination of these ROS indicators can indicate the overall ROS photogeneration of the compounds within our series.

Figure S3.35. Investigations of the ROS-specificity of A) AUR and B) SOSG in (70:17.5.12.5)

H2O:cremophor:DMF using 4a (~30 μM), the type 1 PDT ROS quenchers dimethyl thiourea (DMTU, 10 mM) and 1,2-dihydoxybenzene-3,5-disulfonic acid disodium salt hydrate (tiron, 10 mM) and the type 1 PDT ROS quencher sodium azide (10 mM) (N=3, error bars show ± STDERR). Reproduced with permission from (163). Copyright © 2014, American Chemical Society.

In vitro biocompatibility of detergent with A549 cancer cells

The in vitro toxicity of various combinations of DMSO and cremophor were investigated using A549 cancer cells to determine which detergent concentrations should be added to the cell culture media to properly model the ROS photogeneration of the compounds within our series under biocompatible in vitro solution conditions. The A549 cancer cells were cultured in RPMI-

163 1640 cell culture media with 10% FBS (v/v) until they reached ~70% confluence. Cells were then washed with PBS, were treated with trypsin and counted. Cells were seeded (50,000 cells per well) in sterile 96-well plates and were left to grow overnight in media containing 10% FBS (v/v). The media was removed and various combinations of DMSO and cremophor (in PBS) and with cell media containing 10% FBS (v/v) were incubated with cells for 1 h at 37 ˚C (in quintuplicate). The solutions were removed and cells were left to grow overnight in media with 10% FBS (v/v) at 37 ˚C. The cell media was removed and cells were treated with 200 μL of 0.25 mg/mL MTT reagent (in media with 10% FBS (v/v)) for 2 h at 37 ˚C. The MTT solution was removed and cells were treated with 200 μL of (1:1) DMSO:70% isopropanol (v/v) in 0.1 M HCl

(aq). Each plate was shaken on a plate shaker at 550 rpm for ~10 minutes until the purple precipitates completely dissolved. The absorbance of each well was then recorded at 570 nm on a plate reader. After blanking each sample absorbance against the background absorbance of (1:1)

DMSO:70% isopropanol (v/v) in 0.1 M HCl (aq) alone, the cell viability of each of the tested detergent combination was calculated by normalizing their absorbance to the absorbance of cells treated with media with 10% FBS (Figure S3.36). These results showed that even the highest concentrations of DMSO (3%, v/v) and cremophor (1%, v/v) did not cause noticeable toxicity to A549 cancer cells. Therefore, to ensure that the subsequent ROS photogeneration experiments would model biocompatible in vitro cell media conditions, the 2% DMSO (v/v) and 0.5% (v/v) cremophor concentrations were chosen.

164

Solubility of 1a-5a and 1b-5b in PBS containing 2% DMSO and 0.5% cremophor (v/v)

The solubility of each compound was investigated in PBS containing 2% DMSO (v/v) and 0.5% cremophor (v/v) to determine which of these compounds (~50 μM) would remain dissolved during subsequent investigations of ROS photogeneration. Once the UV-vis absorption spectra were recorded, it was found that aggregation caused spectral red-shifting of the Qy absorption band from ~750 nm to >800 nm (Figure S3.37). These investigations showed that compounds 3a and 4a aggregated at concentrations used in subsequent ROS photogeneration experiments in cell culture media containing the same concentrations of detergents. Therefore, compounds 3a and 4a were excluded from these experiments and comparisons.

165

Chapter 4 Synthesis and characterization of a new natural product analogue, 132-173-bacteriochlorophyllone a

Preamble

This chapter is comprised of the manuscript, ―Synthesis and characterization of a new natural product analogue, 132-173-bacteriochlorophyllone a‖ that was adapted from Roxin, Á., MacDonald, T. D., and Zheng, G. Journal of Porphyrins and Phthalocyanines 2013, DOI: 10.1142/S1088424613501058, (221) Copyright © 2013, World Scientific Publishing Co., with permission from World Scientific Publishing Co. I was the primary contributor to this publication. I synthesized, purified, and characterized all chemicals, designed and performed all chemical and biological experiments, analyzed and interpreted all results, and structured, wrote, and edited the majority of the manuscript for publication. Mr. MacDonald helped confirm the structure of 132-173-bacteriochlorophyllone a, and helped edit the manuscript for publication. Dr. Zheng helped structuring the manuscript, and helped edit the manuscript for publication. Dr. Juan Chen offered advice for the chemical experiments and Dr. Matthew Forbes at the AIMS Mass Spectrometry Laboratory in the Department of Chemistry at the University of Toronto for performed HRMS.

Introduction

The scaffolds of tetrapyrrole macrocycles can be modified at several positions to form interesting and unusual structures. The first synthetic bacteriochlorin and chlorins that contained the unusual seven-membered exocyclic F-ring, which emanate from the 132/15 and 17 carbon positions, were reported by Falk et al. (261) and Isenring et al. (262) in 1975. Several natural porphyrins and chlorins displaying this distinct exocyclic F-ring were reported since the early 1980‘s (Figure 4.1). These porphyrins were isolated from sediments and include derivatives that only contain the exocyclic F-ring (1) (263-266), and those with the exocyclic F-ring and E-ring (2 - 4) (265- 271). Several chlorins analogues (5 – 10) (265-267, 269-285) have also been isolated from sediment samples (5 - 8) and marine organisms (6 - 10). Synthetic derivatives of 7 (261, 262) were produced using multi-step synthetic procedures prior to the isolation of a derivative of 166

167

cyclopheophorbide a enol (7; R1 = CH=CH2, R2 = CH3, M = 2H) from natural sources (266, 278). In addition to isolating the chlorins, 8, 9 and 10, Sakata et al. discovered that these compounds act as natural antioxidants (276, 277). The reports of the natural chlorins 6-8, and the elucidation of the antioxidant properties of 8-10, subsequently inspired Ma et al. to discover several synthetic routes for the synthesis of 6, 8, 9 and 10 (286-289).

Figure 4.1. Previously described natural porphyrins (1-4), chlorins (5-10), the previously synthesized bacteriochlorin 11, and the novel bacteriochlorin 12, each containing the seven- membered exocyclic F-ring. Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co.

168 The discovery of these natural exocyclic F-ring-containing porphyrins and chlorins, and the subsequent revelation of antioxidant activity of some of these natural products inspired further investigations of other of tetrapyrroles with this exocyclic structure. To the best of our knowledge, the first bacteriochlorin analogue of these natural porphyrins and chlorins containing the exocyclic F-ring (132-173-cyclobacteriopheophorbide a-enol, 11) was synthesized by Falk et al. (261). This involved the isolation of methyl-bacteriopheophorbide a (13) from Rhodospirillum rubrum, cleavage of the 132-methyl ester (methyl-bacteriopyropheophorbide a, 14), and formation of the 32-ketal-derivative, 15, which was cyclized in a two-step reaction to produce 11 (261) (Figure 4.2). Here we show the one-step cyclization and structural confirmation of a novel natural product analogue containing the exocyclic F-ring, 132-173- bacteriochlorophyllone a (12), from the precursor, bacteriopheophorbide a (16). In addition to studying its antioxidant activity, we investigated the potential utility of 12 as a photosensitizer for photodynamic therapy (PDT) since derivatives of the precursor, 16, have been explored by us (198-200, 290-294) and many others (154-158, 161, 162, 206-209) for PDT and near-infrared (NIR) fluorescence imaging. While the biological activity of 12 was explored, the primary goal of this report is to expand the structural library of tetrapyrrole macrocycles by providing a facile synthetic method to produce a bacteriochlorin analogue of natural porphyrins and chlorins containing the exocyclic F-ring.

Figure 4.2. Synthetic route of 11 (261). (i) pyridine, 105 °C, 72 h (ii) CH(OCH3)3, HCl, MeOH,

169 Results

One-step synthesis of 12 from bacteriopheophorbide a (16)

The product, 12, was synthesized by a facile 132-173-cyclization of 16 (Figure 4.3). Briefly, the precursor (16) was synthesized according to literature procedures (219, 220) by extracting bacteriochlorophyll a (19) from R. sphaeroides, removing the central magnesium, and cleaving the phytyll group at the 173 position. Compound 16 was then cyclized at the 132 and 173 carbon positions giving a good yield (62%) of 12 on the 60 mg scale.

Investigation of reaction conditions for the one-step synthesis of 12 from bacteriopheophorbide a (16)

Several combinations of the reagents, choline chloride, HBTU, DMAP and DIPEA were incubated with the precursor, 16, to gain insight into the 132-173-cyclization reaction mechanism used to produce 12 (Table 4.1). The 132-173-cyclization of 16 did not occur without HBTU (entry 5) and either DMAP or DIPEA (entry 4), suggesting that the mechanism of cyclization involved deprotonation of the 132 position and activation of the 173 carboxylic acid. The yield and purity improved significantly when using both bases together (entry 10; 75% isolated yield, > 95% purity) as compared to using each individually (entry 6; 41% isolated yield, ~ 85% purity and entry 7; 50% isolated yield, > 95% purity). The 132-173-cyclization reaction did not require the use of choline chlorine (entry 8; 62% isolated yield, > 95% purity and entry 9; 44% isolated yield, > 95% yield), but the isolated yields of 12 slightly improved with the use of this reagent.

170 The chlorin, 17, was incubated with all of the aforementioned reagents under the described conditions for 8 days to determine if a substrate that contained a secondary carbon at the 132 position would undergo 132-173-cyclization (Table 4.1, entry 11). Though 17 underwent esterification with choline (~ 29% of crude), an analogous exocyclic F-ring-containing product was not produced (Figure S4.1). Therefore, the described 132-173-cyclization likely requires an acidic tertiary carbon at the 132 position.

Further reaction condition investigations were conducted to determine the effect of increasing temperature, increasing concentration, and the use of choline chloride under these conditions (Table 4.1). Since each of these conditions was expected to generate the product, 12, the product was purified after 1 hour to compare the effects of the aforementioned conditions. While the product, 12, was produced under both of these temperature conditions, the isolated yield of 12 was highest under ambient conditions (entry 12; 46% isolated yield, > 95% purity). As previously described, the use of choline chloride slightly increased the isolated yield of 12 at ambient temperature (entry 12 vs. entry 13; 36% isolated yield, > 95% purity) and at 80 °C (entry 14; 38% isolated yield, > 95% purity vs. entry 15; 31% isolated yield, > 95% purity). The improved isolated yield of 12 was due to the increased consumption of 16 by choline esterification which enhanced the resolution of the eluted product, 12, during purification (Fugure S4.2). Heating (80 °C) the solutions (entry 14-17; 26-38% isolated yield) and increasing the reaction concentrations (entry 18; 26% isolated yield, ~ 85% purity and entry 19; 28% isolated yield, > 95% purity) lead to the reduction of the isolated yields of 12 compared to incubations at ambient temperatures (entry 12). This suggested that the reduction of the isolated yields of 12 during heating (80 °C) was possibly due to a side-reaction that converted the product, 12, to the precursor, 16, or due to base-catalyzed and/or heat-induced degradation.

Substrate Choline-Cl HBTU DMAP DIPEA DMSO T Isolated Purity Entry Time (1 equiv) a (equiv) (equiv) (equiv) (equiv) (μL) (°C) yield (%) (%) b

1 16 5 - - - 972 rt 3 d nr -

2 16 - 2 - - 972 rt 3 d nr -

3 16 - - 2 5 972 rt 3 d nr -

4 16 5 2 - - 972 rt 3 d nr -

5 16 5 - 2 5 972 rt 3 d nr -

6 16 5 2 2 - 972 rt 3 d 41 ~ 85

7 16 5 2 - 5 972 rt 3 d 50 > 95

8 16 - 2 2 5 972 rt 3 d 62 > 95

9 16 - 2 - 5 972 rt 3 d 44 > 95

10 16 5 2 2 5 972 rt 3 d 75 > 95

11 17 5 2 2 5 972 rt 8 d na ~ 71 c

12 16 5 2 2 5 972 rt 1 h 46 > 95

13 16 - 2 2 5 972 rt 1 h 36 > 95

14 16 5 2 2 5 972 80 1 h 38 > 95

15 16 - 2 2 5 972 80 1 h 31 > 95

16 16 5 2 2 5 243 80 1 h 26 ~ 85

17 16 - 2 2 5 243 80 1 h 28 > 95

18 12 - - - 5 486 rt 3 d na ~ 77 d

19 12 - - - 5 486 80 1 h na ~ 49 d

20 12 - - - - 486 80 1 h na > 95 d a. 16 (1 equiv = 16.4 μmoles), 17 (1 equiv = 9.35 μmoles), 12 (1 equiv = 8.2 μmoles); b. Purity of product assessed by HPLC (spectral integrations at 750 nm); c. Purity of substrate in reaction crude assessed by HPLC (spectral integration at 663 nm); d. Purity of substrate in reaction crude assessed by HPLC (spectral integrations at 750 nm); rt, Ambient temperature; nr, No reaction observed by HPLC-MS; na, The product was not isolated.

172 Purified samples of 12 were incubated with and without DIPEA at 80 °C for 1 hour, and with DIPEA at ambient temperature for 3 days (Table 4.1). Incubating 12 with DIPEA at ambient temperature (entry 18; 77% 12 in crude) slowly degraded 12 (Figure S4.3a), heating (80 °C) 12 with DIPEA (entry 19; ~ 49% 12 in crude) degraded 12 and lead to the formation of 16 (Figure S4.3b), while heating (80 °C) alone (entry 20; > 95% 12 in crude) had no effect on 12 (Figure S4.3c). Instability of the product, 12, in the presence of both heat and base therefore leads to a loss in yield.

Structure elucidation of 12

The UV-Vis and fluorescence spectra of 12 were recorded to show the characteristic bacteriochlorin absorbance and emission profiles of 12 (Figure 4.4). Characterization of the ΦF‘s of 12 and 16, relative to the standard 18 (ΦF = 0.004) (222), showed that the ΦF of 12 (ΦF = 0.04) was similar to that of its precursor, 16 (ΦF = 0.03). The HRMS (ESI+) spectrum of 12 showed that the [M + H]+ of 12 (+ 593.2760 m/z) (Figure S4.4) was 17 amu lower than the calculated molecular mass of 16 (calc. C35H38N4O6 = 610.279 amu). Since the calculated molecular mass of

12 (calc. C35H36N4O5 = 592.269 amu) is 18 amu lower than 16, this initially suggested that 12 formed by the dehydration of 16, and possibly by the formation of the exocyclic F-ring. The 1H and 13C spectra of 12 and 16 were subsequently analysed to determine which positions of 16 were involved during the formation of 12 (Figure 4.5).

173

The 2D COSY 1H NMR spectrum of 12 was analysed to determine the shifts of protons 8, 81, 17, 1 2 1 17 , 17 , 18 and 18 (Figure 4.6). The shift assignments of these particular protons were especially important for investigating the presence of the exocyclic F-ring. The shifts of the two distinct 81 protons were assigned by observing their coupling with the adjacent 82 methyl protons. Then, the shift of proton 8 was confirmed by the observed coupling of the two distinct 81 protons. The shifts of proton 17 and one of the two 171 protons were assigned due to their coupling to the 181 methyl protons. These assignments were then used to identify the shifts of the two 172 protons by observing their coupling with protons 181 and 171, and by the coupling between the two individual 172 protons.

174

1H NMR spectra of 12 and 16 revealed the absence of the 132 proton in 12 (Figure 4.7). This implied the presence of an exocyclic F-ring. Significant downfield shifts of the protons at the 171 and 172 positions in 12, relative to those in 16, further indicate the exocyclic F-ring. We propose that these four protons were shifted downfield in 12 due to the formation of the 173 ketone after 132–173-cyclization of 16.

175

The combination of the described 2D COSY 1H NMR results, the partial 1D 1H NMR results, and the complete 1D 1H NMR of 12 (Figure S4.5) and 16 were used to assign the shifts of all protons for these bacteriochlorins (Table 4.2). In addition to comparing the proton shifts of 12 and 16, the proton shifts of the chlorins, pyropheophorbide a (17) and its exocyclic F-ring- containing derivative, 132(S)-173-hydroxychlorophyllone a (8a) (276, 277), were analysed to corroborate the presence of an exocyclic F-ring in 12 (Table 4.2). The chlorin, 17, has a similar ring F as the bacteriochlorin 12, and the comparison of the proton shifts between the two pairs of bacteriochlorins and chlorins illustrate the pattern of proton shifts observed during 132-173- cyclization of 16 and 17 (Figure 4.5). The protons at 17, 171, 172, and 181 were each shifted downfield in the 1H NMR spectrum of the exocyclic F-ring-containing chlorin, 8a, compared to the chlorin 17. The downfield shifts of these protons matched the patterns described for the

176 bacteriochlorin 12 compared to its precursor, 16. In addition, 8a lacks a 132 proton due to the presence of the exocyclic F-ring and the 132 OH group.

We inferred that the chirality of the 132 position of 12 was the S configuration since the chirality of the 132 position of the precursors, 16 (219, 220) and 19 (295), are known to be in the S configuration, and due to the patterns observed for the 1H NMR shifts of protons at the 17 and 171 positions for the bacteriochlorin, 12, compared to its precursor, 16 (Table 4.2), and the chlorins, 8a and 132(R)-173-hydroxychlorophyllone a (8b) (Figure S4.6), compared to their precursor, 17 (Table S4.1). The proton at the 17 position was shifted slightly downfield in the 1H NMR spectra of 12 compared to 16 (Δδ = + 0.25 ppm), as with 8a compared to 17 (Δδ = + 0.65 ppm) (276, 277), whereas the proton at the 17 position of 8b was shifted upfield (Δδ = - 0.43 ppm) compared to 17 (276). In addition, the protons at the 171 position of 12 shifted slightly downfield in the spectra of 12 compared to 16 (Δδ = + 0.64, + 0.47 ppm), as with 8a compared to 17 (Δδ = + 0.54, + 0.07 ppm) (276, 277), whereas these protons of 8b were shifted relatively further downfield (Δδ = + 1.34, + 0.48 ppm) compared to 17 (276).

177 Table 4.2. Proton NMR shift assignments (δ ppm), splitting patterns, and J-values of the bacteriochlorins 16 and 12, and the chlorins 17 and 8a (with significant proton shifts bolded). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co.

Proton 16 a 12 a 17 b 8a c

1 2 -CH3 3.32 (s) 3.31 (s) 3.36 (s) 3.44 (s)

31-CH - - 7.90 (dd, J = 11.5, 6.3 Hz) 8.03 (dd, J = 18.0, 11.4 Hz)

32- 6.22 (dd, J = 17.9, 0.9 Hz), 6.31 (dd, J = 18.0, 1.5 Hz), 3.11 (s) 3.11 (s) CH2/CH3 6.12 (dd, J = 11.5, 1.0 Hz) 6.21 (dd, J = 11.5, 1.5 Hz)

5-CH 8.95 (s) 8.91 (s) 9.23 (s) 9.50 (s)

7-CH 4.34 (m) 4.26 (m) - -

1 7 -CH3 1.64 (d, J = 7.0 Hz) 1.70 (d, J = 7.2 Hz) 3.14 (s) 3.26 (s)

8-CH 3.99 (m) 3.99 (m) - -

1 8 a/b-CH2 2.34 (m), 2.00 (m) 2.30 (m), 1.97 (m) 3.58 (m) 3.68 (q, J = 7.7 Hz)

2 8 -CH3 1.02 (t, J = 6.8 Hz) 0.97 (t, J = 7.3 Hz) 1.64 (t, J = 7.6 Hz) 1.70 (t, J = 7.7 Hz)

10-CH 8.72 (s) 8.68 (s) 9.32 (s) 9.55 (s)

1 12 -CH3 3.79 (s) 3.80 (s) 3.56 (s) 3.71 (s)

132-CH/ 5.23 (d, J = 19.8 Hz), 132a/b-CH 6.17 (s) - 5.09 ( d, J = 19.9 Hz) -

3 15 -OCH3 3.45 (s) 3.44 (s) - -

17-CH 3.82 (m) 4.07 (m) 4.26 (d, J = 9.2 Hz) 4.91 (dt, J = 12.8, 3.6 Hz)

2.90 (dddd, J = 13.2, 4.3, 3.6, 3.0 Hz), 1 17 a/b-CH 2.06 (m) 2.70 (m), 2.53 (m) 2.36 (m), 2.19 (m) 2.26 (ddt, J = 12.8, 2.6, 13.2 Hz)

4.36 (ddd, J = 13.2, 11.2, 3.0 Hz), 172a/b-CH 2.36 (m), 2.18 (m) 3.65 (m), 2.82 (m) 2.66 (m) 2.81 (ddd, J = 11.2, 4.3, 2.6 Hz)

18-CH 4.25 (m) 4.15 (m) 4.44 (q, J = 7.3 Hz) 4.36 (dq, J = 3.6, 7.3 Hz)

1 18 -CH3 1.71 (d, J = 6.9 Hz) 1.98 (d, J = 7.3 Hz) 1.80 (d, J = 7.3 Hz) 2.20 (d, J = 7.3 Hz)

20-CH 8.63 (s) 8.63 (s) 8.50 (s) 8.66 (s)

21-NH - 0.85 (br. s) - 0.62 (br. s) - 1.75 (br. s) - 1.90 (br. s)

a. DMSO-d6; b. CDCl3; c. From ref. (276, 277) (CDCl3).

178 In addition to comparing the 1H NMR spectra of 16, 12, 17 and 8a, the Jmod 13C NMR spectra of 12 (Figure S4.7) and 16 were analysed. After assigning the 13C shifts of 12 and 16 (Table 4.3), two significant spectral differences were observed (Figure 4.8). Specifically, the 173 carbonyl carbon of the carboxyl group of 16 significantly shifted downfield, indicating the formation of a 173 ketone in 12. Also, the tertiary 132 carbon of 16 was converted to the quaternary 132 carbon in 12. The later observation is supported by the absence of the 132 proton and the presence of the quaternary 132 carbon in 12 from the aforementioned 1D 1H NMR and Jmod 13C NMR comparisons of 12 to 16. Therefore, the summary of the described HRMS, 2D COSY 1H NMR, 1D 1H NMR, and Jmod 13C NMR analyses confirmed the presence of the exocyclic F-ring of 12.

Table 4.3. Carbon NMR shift assignments (DMSO-d6, δ ppm) of the bacteriochlorins, 16 and 12 (with significant carbon shifts bolded). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co.

Carbon position Carbon type 16 12

31, 131, 152, 173 Carbonyl 199.4, 189.0, 174.5, 171.8 201.1, 199.4, 185.4, 172.6

132 Exocyclic E-ring 64.2 (tertiary) 83.2 (quaternary)

81, 171, 172 Secondary 31.1, 29.8, 29.5 29.7, 23.5, 19.7

41.0, 33.7, 23.1, 23.0, 13.4, 33.6, 23.0, 22.1, 13.9, 13.4, 21, 32, 71, 82, 121, 153, 181 Methyl 11.5, 10.9 11.5, 10.8

147.9, 138.9, 138.5, 138.4, 146.5, 139.9, 138.9, 138.4, 1, 4, 6, 9, 11, 14, 16, 19 Internal pyrrole 137.3, 133.5, 128.1, 120.0 137.5, 134.1, 127.8, 120.0

2, 3, 12, 13 External sat. pyrrole 171.6, 169.7, 163.6, 158.4 172.2, 167.7, 164.0, 159.7

7, 8, 17, 18 External unsat. pyrrole 54.3, 53.0, 50.5, 49.4 54.1, 53.9, 51.8, 51.5

5, 10, 15, 20 Meso 108.7, 99.9, 97.3, 97.2 106.4, 100.3, 97.9, 96.7

179

Investigation of the antioxidant activity of 12

The antioxidant activity of 12 was explored by investigating if increasing concentrations of 12 (0 μM – 100 μM) could reduce the oxidation of the reactive oxygen species (ROS)-indicator, AUR

(229-233), by H2O2 and hydroxyl radicals in solutions containing either 0.5 μM or 5.0 μM H2O2 with HRP (0.2 U/well) (Figure 4.9). As expected, the 5.0 μM H2O2 solution oxidized AUR to a greater extent compared to the 0.5 μM H2O2 solution. However, the fluorescence of AUR did not

180 decrease in either 0.5 μM or 5.0 μM H2O2 with increasing concentrations of 12. Since a 20-fold

(in 5.0 μM H2O2) and even a 200-fold (0.5 μM H2O2) molar excess of 12 (100 μM) did not reduce the oxidation of AUR by H2O2 and hydroxyl radicals, these results suggest that 12 does not possess strong antioxidant activity.

Figure 4.9. Investigation of the concentration-dependant antioxidation of AUR by 12 in 0.5 μM

PDT activity of 12

The in vitro PDT activity of 12 was explored by investigating the dose-dependent (0 μM – 100 μM) toxicity of 12 (in PBS with 0.5% DMSO, v/v) after 740 nm irradiation compared to dark conditions using U87 glioblastoma cells. MTT analysis was used to measure the viability of U87 cells, and GraphPad PRISM software was subsequently used to calculate the LD50‘s of 12 under

PDT and dark conditions (Figure 4.10). Since the LD50 of 12 under the PDT condition (4.2 ± 0.45 μM) was ~ 40-fold lower compared to the dark condition (169.6 ± 54.1 μM), this experiment illustrated that 12 is capable of in vitro PDT.

181

Figure 4.10. Dose-response curves showing the concentration-dependent viability of U87 cells treated with 12 under dark conditions (solid trace) and after irradiation with a 740 nm LED light box (dashed trace), elucidated by MTT analysis (N = 3, error bars indicate ± STDERR). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co.

Discussion

The combination of HRMS, 2D COSY 1H NMR, 1D 1H NMR, and Jmod 13C NMR results were used to confirm the presence of the exocyclic F-ring of 12. The HRMS spectrum of 12 initially showed that 16 underwent a dehydration reaction while forming 12. Analysis of the 1H NMR spectra of 12 and 16 showed that 12 did not contain a 132 proton. The absence of the 132 proton of 12 was confirmed by observing the quaternary 132 carbon in the Jmod 13C NMR spectrum of 12. The formation of the 173 ketone of 12 was indicated by the significant downfield shifts of the 171 and 172 protons of 12, compared to 16. Comparisons of the proton shifts of the bacteriochlorins, 16 and 12, and their chlorin analogues, 17 and 8a, showed that the 171 and 172 protons of 17 also significantly shifted downfield upon formation of the exocyclic F-ring- containing 8a. Further comparisons of the Jmod 13C NMR spectra of 12 and 16 confirmed that the 173 carboxyl carbon of 16 was converted to the 173 ketone carbon of 12. The summary of these characterizations led to the conclusion that 12 contained an exocyclic F-ring due to the 132- 173-cyclization of 16.

182 Compound 12 was synthesized by a facile method. Sakata et al. proposed that the chlorin analogue of 12, 8a, was formed by the removal of the 132 proton of 14, and the subsequent nucleophilic attack of the 173 carbonyl (276). We propose that DMAP and DIPEA deprotonated the 132 position of 16, and that HBTU activated the carboxylic acid at the 173 position for nucleophilic attack. A minor intermolecular esterification of 16 with choline was observed, but the intramolecular 132-173-cyclization of 16 was the dominant reaction. Though choline chloride was not required for the 132-173-cyclization of 16, the competing esterification reaction consumed 16 and slightly enhanced the isolated yield and purity of 12 by facilitating the resolution and purification of 12 during flash chromatography. Since the 132-173-cyclization reaction did not occur using 17, which contained a secondary carbon at the 132 position, this reaction likely requires the deprotonation of a tertiary carbon at the 132 position. Finally, we found that our 132-173-cyclization of 16 is a reversible reaction, and that increasing the reaction temperature in the presence of base leads to the enhanced conversion of the product, 12, to the precursor, 16. Therefore, we propose that our one-step cyclization method may be extended to other classes of tetrapyrrole macrocycles, containing a carboxylic acid at the 173 position and an acidic tertiary 132 carbon, under ambient temperatures to further synthesize additional tetrapyrrole macrocycles containing the exocyclic F-ring.

Due to the antioxidant properties of the natural exocyclic F-ring-containing chlorins, 8-10 (276, 277), we investigated if the bacteriochlorin analogue, 12, could also act as an antioxidant. The antioxidant activity of 12 was explored by investigating if increasing concentrations of 12 could reduce the oxidation of the ROS-indicator, AUR (229-233), in solutions of H2O2 containing HRP. These studies found that even the highest concentrations of 12 could not reduce the oxidation of AUR by H2O2 and hydroxyl radicals. This suggests that, unlike the previously described natural chlorins 8, 9 and 10, the bacteriochlorin analogue, 12, does not possess strong antioxidant activity. Therefore, the exocyclic F-ring may not be the only structure required to impart antioxidant properties to tetrapyrrole macrocycles. Future studies may derivatize 12 to investigate key structural modifications required to produce strong bacteriochlorin antioxidants.

Metallated and free-base derivatives of 16 have been extensively used for PDT and NIR- fluorescence imaging applications (154-158, 161, 162, 198-200, 206-209, 290-294). These studies prompted our investigation of the PDT activity of 12 to determine if 12 could also generate cytotoxic ROS once activated by NIR-irradiation. The NIR-activated PDT activity of 12 was investigated using U87 cells, and our studies showed that 12 was capable of PDT in vitro.

183 The PDT activity of 12 may be enhanced by metal insertion and by the conjugation of biomarker-targeting moieties through the 31 carbonyl of 12. While the biological activity of 12 was explored, the primary goal of this project was to present a novel bacteriochlorin analogue of natural porphyrins and chlorins containing the distinct exocyclic F-ring. Future studies could pursue the synthesis of additional F-ring-containing tetrapyrrole macrocycles for use as photosensitizers, antioxidants and imaging probes.

Conclusions

We presented a simple reaction that converted the precursor, 16, to 12 by 132-173-cyclization. Analysis of the HRMS, 1D 1H NMR, 2D 1H COSY NMR, and Jmod 13C NMR spectra of 12 clearly illustrated the presence of the exocyclic F-ring. Our report of 12 expands the library of tetrapyrrole macrocycle structures by presenting a new bacteriochlorin natural product analogue. We propose that our one-step synthetic method may be applied to precursors structurally similar to 16 to synthesize other classes of tetrapyrrole macrocycles containing the exocyclic F-ring. Since 12 is fluorescent and capable of PDT, future studies can pursue the synthesis of additional exocyclic F-ring-containing bacteriochlorins for imaging and therapeutic applications.

Methods and Materials

Chemicals and reagents

Bacterial cultures of Rhodobacter sphaeroides were purchased from Frontier Scientific. The reagents choline chloride, O-benzotriazole-N,N,N‘,N‘-tetramethyl-uronium-hexafluoro- phosphate (HBTU), 4-dimethylaminopyridine (DMAP), N,N-diisopropylethylamine (DIPEA), and horseradish peroxidise (HRP) were purchased from Sigma-Aldrich and used without further purification. The chlorin, pyropheophorbide a (17), was synthesized according to literature procedures (296). The fluorescence quantum yield (ΦF) standard, Tookad (18), was synthesized according to literature procedures (153). Deuterated NMR solvents, DMSO-d6 and CDCl3, were purchased from Cambridge Isotope Laboratories. Amplex® UltraRed was purchased from Invitrogen. The MTT reagent, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, was purchased from Invitrogen-Gibco.

Equipment

Compound 12 was purified using a Biotage Isolera One flash chromatography system with a 30 g Biotage C-18 cartridge under reverse-phase conditions (0 – 100% acetonitrile in 0.1% TFA (aq)

184 over 12 minutes at a 40 mL/ minute flow) while monitoring the product at 357 nm. The purity of 12 was assessed by analytical HPLC-MS using a Waters 2695 HPLC with a Waters Delta Pak C18, 5 μm 3.9 x 150 mm column under reverse-phase conditions (0 – 100% acetonitrile in 0.1%

TFA (aq) over 15 minutes at a 0.8 mL/min flow) while monitoring the product with a Waters 2996 photodiode array detector and a Waters Micromass ZQ mass spectrometer. The UV-Vis spectrum of 12 was recorded using a Varian 50 Bio UV-Visible spectrophotometer. The fluorescence spectra of 12, 16 and 18 were recorded in MeOH using a Horiba Jobin Yvon

Fluoromax-4 spectrofluorometer to calculate the ΦF‘s of 12 and 16, relative to 18 (OD525nm =

0.1; λex = 525 nm, λem = 700 nm – 875 nm). HRMS was performed using an AB/Sciex QStar mass spectrometer. NMR was performed using a Bruker Ultrashield 400 Plus 400 MHz NMR. A Molecular Devices Spectra Max M5 plate reader was used to measure fluorescence intensities for antioxidant assays and absorbencies for MTT assays.

One-step synthesis of 12 from bacteriopheophorbide a (16)

The precursor, 16, was synthesized according to literature procedures (219, 220). Compound 16 (60 mg, 98.1 µmoles or 1.0 equiv), choline chloride (54 mg, 490.2 μmoles or 5.0 equiv), HBTU (76.2 mg, 196.2 μmoles or 2.0 equiv), DMAP (25.6 mg, 196.2 μmoles or 2.0 equiv) and DIPEA (85 μL, 490.2 μmoles or 5.0 equiv) were dissolved in 2.9 mL of DMSO. The solution was purged with Ar (g) for 2 minutes and was sealed under Ar (g). The solution was stirred, and the reaction proceeded at ambient temperature for 3 days in the dark. The crude was purified by flash chromatography under reverse-phase conditions. The eluted product was concentrated by rotary evaporation and was dried using a speed-vac. This reaction yielded 36.0 mg (61.6 µmoles) of 12 (62% yield) with > 95% purity, assessed by HPLC-MS (Figure S4.8). The structure elucidation of 12 was discussed in details in the Results section.

Investigation of reaction conditions for the one-step synthesis of 12 from bacteriopheophorbide a (16)

With the knowledge that the combination of all of the aforementioned reagents produced 12 from the precursor, 16, compound 16 (10 mg, 16.4 µmoles or 1.0 equiv) was incubated with various combinations of choline chloride (9.0 mg, 82 μmoles or 5.0 equiv), HBTU (12.7 mg, 32.8 μmoles or 2.0 equiv), DMAP (4.3 mg, 32.8 μmoles or 2.0 equiv) and DIPEA (14.2 μL, 82 μmoles or 5.0 equiv) in 972 μL of DMSO to determine the effect of each reagent, which reagents were required for the 132-173-cyclization of 16, and which reagents enhanced the isolated yield

185 of 12. Compound 16 was first incubated with either choline chloride, HBTU, or with DMAP and DIPEA to determine the affect of choline chloride, the carboxylic acid-activating reagent, and bases on the precursor, 16. Compound 16 was then incubated with all reagents except DMAP, DIPEA, DMAP and DIPEA, HBTU, choline chloride, and choline chloride and DMAP to determine which reagents were required for the 132-173-cyclization of 16, and which combination of reagents enhanced the isolated yield of 12 compared to the use of all of the aforementioned reagents. The solutions were prepared and rotated at ambient temperature in the dark for 3 days to allow each reaction to potentially proceed to completion. The progress of these reactions was monitored by HPLC-MS after the allotted 3 day reaction time. If the reaction produced the desired product (12), the crude was purified by flash chromatography under reverse-phase conditions. The eluted product was then concentrated by rotary evaporation and was dried using a speed-vac. The isolated yield of 12 was quantified gravimetrically, and the purity was assessed by HPLC analysis by integrating spectra at 750 nm. Compound 17 (5.0 mg, 9.35 μmoles or 1.0 equiv), which contained a secondary carbon at the 132 position, was incubated with the described molar equivalents of all of the aforementioned reagents under the described reaction conditions for 8 days to determine if the 132-173-cyclization reaction required the substrate to contain a tertiary carbon at the 132 position (assessed by HPLC using integrations at 663 nm). Compound 16 (10 mg, 16.4 µmoles or 1.0 equiv) was then reacted with either all of the aforementioned reagents or with all reagents except choline chloride at ambient temperature, at 80 °C, and at 80 °C using one quarter of the described volume of DMSO (243 μL) for 1 hour in the dark. These reactions were meant to assess the effect of choline chloride, increasing temperature, and increasing concentration on the isolated yield and purity of 12. After each reaction proceeded for 1 hour, the product (12) was purified, dried, quantified and assessed for purity as described. Finally, the chemical stability of the purified product, 12 (5.0 mg, 8.4 µmoles or 1.0 equiv), was assessed by incubating the compound with DIPEA (7.3 μL, 42 μmoles or 5.0 equiv) at 80 °C for 1 hour, with DIPEA (7.3 μL, 42 μmoles or 5.0 equiv) at ambient temperature for 3 days, and by heating it at 80 °C in DMSO for 1 hour. After the allotted incubation times, the purity of 12 was assessed by HPLC as described. These investigations were used to gain insights into the mechanism of the 132-173-cyclization reaction that produced 12 from 16.

186 Investigation of the antioxidant activity of 12

Two concentrated stock solutions of 12 (20 mM and 2 mM) were prepared by dissolving 12 in DMSO. These stock solutions (0.5 μL – 2.5 μL) were used to prepare two sets of dilutions of 12 in 200 μL of 70% DMSO (aq) (v/v) on 96-well plates to have 0 μM, 5 μM, 10 μM, 25 μM, 50 μM and 100 μM concentrations of 12. A stock solution (10 mM) of the H2O2 and hydroxyl radical indicator, AUR (229-233), was prepared by dissolving 1 mg of AUR in 340 μL of DMSO. A 1 μL volume of the AUR stock solution was aliquoted to each of the solutions containing the aforementioned concentrations of 12 (50 μM AUR/well). A stock solution of HRP was prepared by dissolving 1 mg HRP (167 U/mg) in 1.667 mL of PBS. A 2 μL volume of the HRP stock solution was aliquoted to each of the solutions containing 12 and AUR (0.2 U HRP/well). The baseline fluorescence of AUR was recorded at 581 nm (λex = 550 nm) using a plate reader. Two stock solutions of H2O2 (50 μM and 500 μM) were prepared by diluting 10 M H2O2 in H2O. A 2

μL volume of these stock solutions of H2O2 were added to the aforementioned solutions containing 12, AUR and HRP so that AUR was exposed to either 0.5 μM or 5.0 μM H2O2 in solutions containing various (0 μM – 100 μM) concentrations of 12. These solutions were incubated at ambient temperature for 30 minutes to allow H2O2 and hydroxyl radicals to oxidize AUR. The fluorescence of AUR was again recorded using a plate reader. The antioxidant activity of 12 was investigated by observing if increasing concentrations of 12 could reduce the oxidation of AUR by H2O2 and hydroxyl radicals (N = 3). Once the baseline fluorescence of AUR (before

H2O2 and hydroxyl radical exposure) was subtracted, the fluorescence of AUR was plotted versus increasing concentrations of 12 in solutions containing 0.5 μM and 5.0 μM of H2O2 (with HRP).

PDT activity of 12

U87 glioblastoma cancer cells were cultured in DMEM cell culture media containing 10% FBS (v/v) at 37 ˚C until reaching ~70% confluence. The cells were treated with trypsin, counted, and seeded (5 x 104 cells/well) on sterile black-sided, clear-bottomed Costar 96-well plates. Cells were incubated overnight at 37 ˚C to allow them to adhere to the bottom of wells. The media was removed, and the cells were incubated with 12 (0 μM, 1 μM, 5 μM, 25 μM, 50 μM, and 100 μM; in PBS with 0.5% DMSO (v/v)) for 30 minutes at 37 ˚C). Cells were then washed with PBS, and fresh DMEM (containing 10% FBS, v/v) was added to the cells. Cells were then either irradiated with a 740 nm LED light box (4.55 mW/cm2) for 14 minutes and 39 seconds to deliver a 4.0 J/cm2 light dose, or were left in the dark. The cells were then incubated overnight at 37 ˚C. The

187 media was removed, and cells were incubated with a 0.5 mg/mL solution of MTT reagent (200 μL/well) at 37 ˚C for 2 hours until purple formazan precipitates formed. After washing away the excess MTT reagent with PBS, the formazan was dissolved in (1:1) 70% isopropanol (v/v) in 0.1

M HCl (aq): DMSO using a plate shaker. The absorbance of all sample wells was recorded at 570 nm using a plate reader. These values were blanked, and were normalized to the control left in the dark without being treated with 12 (N = 3). GraphPad PRISM software was then used to calculate the LD50‘s of 12 due to PDT activity and dark toxicity.

Acknowledgements

We thank Canadian Institute of Health Research, Canadian Cancer Society Research Institute, Canadian Foundation for Innovation, Joey and Toby Tanenbaum/Brazilian Ball Chair in Prostate Cancer Research for funding. We also thank Dr. Juan Chen for assistance with experimental planning, and Dr. Matthew Forbes at the AIMS Mass Spectrometry Laboratory in the Department of Chemistry at the University of Toronto for performing HRMS.

188 Supporting Information

189

190

Figure S4.3. HPLC absorbance chromatograms showing that a) incubating 12 with DIPEA for 3 days at ambient temperature (Table 4.1, entry 18) slowly degraded 12 (integrations excluded broad shoulder after 10 min), b) incubating 12 with DIPEA for 1 hour at 80 °C (Table 4.1, entry 19) degraded 12 and lead to the formation of 16, and c) heating 12 for 1 hour at 80 °C (Table 4.1, entry 20) had no effect on 12 (% of compounds in crude calculated by integrations at 750 nm). Reproduced with permission from (221). Copyright © 2013, World Scientific Publishing Co.

191

192

193

194

Proton 17 8b a

1 2 -CH3 3.36 (s) 3.38 (s)

31-CH 7.90 (dd) 7.98 (dd)

2 3 -CH2 6.22 (dd), 6.12 (dd) 6.28 (dd), 6.19 (dd)

5-CH 9.23 (s) 9.39 (s)

1 7 -CH3 3.14 (s) 3.23 (s)

1 8 a/b-CH2 3.58 (m) 3.66 (q)

2 8 -CH3 1.64 (t) 1.69 (t)

10-CH 9.32 (s) 9.49 (s)

1 12 -CH3 3.56 (s) 3.66 (s)

132a/b-CH 5.23 (d), 5.09 (d) -

17-CH 4.26 (d) 3.83 (ddd)

171a/b-CH 2.36 (m), 2.19 (m) 3.70 (dddd), 2.67 (dddd)

172a/b-CH 2.66 (m) 3.86 (ddd), 2.97 (ddd)

18-CH 4.44 (q) 4.77 (dq)

1 18 -CH3 1.80 (d) 2.21 (d)

20-CH 8.50 (s) 8.54 (s)

21-NH -1.75 (br. s) -1.50 (br. s)

23-NH 0.89 (br. s) 0.98 (br. s) a. From ref. (276).

195

196

Chapter 5 Future Directions

Preamble

This chapter presents the results of additional experiments that were conducted in the fields previously addressed in Chapters 2, 3 and 4. They are meant to provide preliminary data that can be used to help future studies investigate various extensions of the research that was completed in this thesis. I was the primary contributor to these studies. Gang Zheng and Juan Chen helped design some of the studies, while Lili Ding conducted the Western Blot analysis.

Introduction

Chapters 2, 3, and 4 of this thesis presented the synthesis and characterization of new aziridine aldehyde-cyclized integrin-targeting peptides and bacteriochlorin PSs. Briefly, chapter 2 described a versatile conjugation strategy that was used to generate aziridine aldehyde-cyclized RGD-based integrin-targeting peptides. Most notably, we found that the geometry offered by the macrocyclization chemistry can be tuned by modifying the length of the peptide sequences to modulate RGD-integrin binding. Chapter 3 revealed that taurine-driven exocyclic E-ring opening and the inclusion of anionic substituents generally enhanced the ROS photogeneration of free- base bacteriopheophorbide a derivatives in aqueous solutions. Finally, chapter 4 presented the synthesis and characterization of a new natural product analogue with a distinct 7-membered exocyclic F-ring that displayed photosensitizing properties. Each of these projects have made modest strides towards optimizing strategies for creating superior cancer-targeting peptides and bacteriochlorin photosensitizers. As a continuation of the work presented in chapters 2, 3, and 4, this chapter presents the results of additional unpublished studies that were aimed to either solve challenges of the aforementioned major projects or to provide preliminary data that is intended to inspire future efforts to pursue extensions of the work described in this thesis (see Appendices for Figures, Schemes and Tables). It is envisioned that the studies described herein will contribute to achieving the end goal of efficient targeted PDT with cyclic aziridine aldehyde-cyclized RGD- peptides and bacteriochlorin PSs.

197

198 Cancer-targeting aziridine aldehyde-cyclized RGD peptides

The aziridine aldehyde-driven macrocyclization chemistry, initially reported by Yudin et al. (170), is essentially a modified Ugi reaction. It involves a two-step mechanism whereby the N- and C-termini of linear peptides, an isocyanide, and an aziridine aldehyde dimer react to form a head-to-tail bridged macrocycle. Because this macrocyclization chemistry requires a secondary amine, a carboxylic acid, an isocyanide and an aziridine aldehyde, an immense library of cyclic peptides can be created for a wide variety of biomedical applications.

The report presented in chapter 2 provides methods for achieving sulfhydryl-aziridine ring opening (169). We chose to work with an unmodified aziridine aldehyde because we predicted this would allow facile sulfhydryl-aziridine ring opening for our proposed conjugation strategy. Indeed, the chosen aziridine aldehyde proved to be quite useful for our purposes. However, we wanted to explore the compatibility of our sulfhydryl-aziridine ring opening reaction (using cysteamine and ZnCl2) with modified aziridine aldehyde dimers. If successful, future studies could simply buy commercially-available modified aziridine aldehydes and achieve sulfhydryl- aziridine ring opening without the need of synthesizing the unmodified aziridine aldehyde dimer or devising new methods of conjugating bifunctional linkers to the aziridine ring within these macrocycles.

Using the described linear peptide, HN-PR(Pbf)GD(t-butyl)A-OH, the macrocyclization from chapter 2 was repeated with either an isobutyl-aziridine aldehyde dimer or a phenyl-aziridine- aldehyde dimer (Figure A1.1). Both of these reagents were purchased from Sigma-Aldrich. Though the desired macrocyclization was achieved, our previously reported methods did not provide sulfhydryl-aziridine ring opening. The most likely explanation for this latter result is that these modified aziridines imposed a restricting degree of steric hindrance that prevented the desired ring opening reaction. Therefore, new synthetic methods must be developed for conjugating bifunctional linkers (via sulfhydryl-aziridine ring opening) to modified aziridines within these macrocycles.

Besides being currently limited to using the unmodified aziridine aldehyde dimer, our sulfhydryl-aziridine ring opening reaction generated a distinct side-product. HPLC-MS revealed the formation of macrocycles that underwent amino-aziridine ring opening and contained two cysteamine moieties as a result of disulfide bonding (Figure A1.2A and B). Upon NMR analysis

199 of our cysteamine reagent, we found that a portion of this starting material had become dimerized via disulfide bonding. Since these cysteamine dimers contained two free primary amines and no reactive sulfhydryl group, the presence of this dimer would have enhanced the undesired amino-aziridine ring opening reaction. To remedy this, we protected the primary amines of cysteamine using Fmoc-Cl (Figure A1.2C) (297). This resulted in the complete dimerization of Fmoc-cysteamine and required a subsequent disulfide reduction with tris-(2- carboxyethyl)phosphine (TCEP) (298) to yield the desired Fmoc-cysteamine monomer. As a proof-of-concept, we used Fmoc-cysteamine to achieve sulfhydryl-aziridine ring opening within the model macrocycles cPR(Pbf)DG(t-butyl)A (Figure A1.2D) and cPR(Pbf)GD(t-butyl)AA (Figure A1.2E). These experiments showed that Fmoc-cysteamine can be used to avoid the noted side-products. Future studies can continue using this reagent to improve synthetic yields.

We found that the RGD-containing peptides cPRGDA and cPRGDAA successfully targeted αVβ3 integrin receptors on U87 cells (Figure 2.5C – F). This raised the question of whether these peptides would specifically bind to other integrin variants that have also been associated with cancers. In particular, we were interested in targeting αVβ5 (299), αVβ6 (300, 301), and α5β1 (302- 304) integrins. We investigated the integrin-binding specificity of the fluorescein conjugates of cPRGDA and cPRGDAA using fluorescence polarization spectroscopy. Briefly, as a fluorescent molecule binds to a macromolecules (ex. proteins), the free rotation of the fluorophore will be reduced. And as a result, the fluorescent molecule will emit polarized light upon illumination when bound to these complimentary macromolecules (305). While we only have preliminary data (N = 1), these suggest that i) cPRGDA may show specificity toward αVβ5 compared to cPRGDAA (Figure A1.3A), that ii) cPRGDA and cPRGDAA may both bind to αVβ6 (Figure

A1.3B) and that iii) cPRGDA and cPRGDAA may not specifically bind to α5β1 compared to cPRDGA (Figure A1.3C). Future studies should continue to explore the use of cPRGDA and cPRGDAA to target αVβ6 integrins. In addition, future efforts can investigate if the γ–like-turn of cPRGDA (described in chapter 2) can afford enhanced binding specificity towards αVβ5 integrins receptor compared to the less-constrained cPRGDAA macrocycle.

In chapter 2, we showed that our conjugation strategy could be used to synthesize aziridine aldehyde-cyclized macrocycles for radioimaging, radiotherapy, fluorescence imaging and PDT applications. Specifically, we conjugated the radiometal chelator, DOTA, to the cyclic PR(Pbf)GD(t-butyl)A macrocycle through the primary anime of cysteamine. While we have not yet radiolabeled our aziridine aldehyde-cyclized peptides, we were able to adapt known copper-

200 insertion methods (293) to label the model amide-cyclized macrocycle, cPRGDK(DOTA) , with both ‗cold‘ 63Cu(II) (Figure A1.4A) and radioactive 64Cu(II) (Figure A1.4B). HPLC results 64 showed that a >95% radiochemical yields was achieved by incubating Cu(OAc)2 (100 μCi) with the amide-cyclized cPRGDK(DOTA) (100 nmoles) in 0.1 M TEAA (pH 5.5) for 1 h at 60 °C. Future studies should apply this 64Cu-radiolabeling procedure using the aforementioned aziridine aldehyde-cyclized cPRGDA-DOTA conjugate to evaluate the in vivo fate of this macrocycle and determine its value as a PET imaging agent. If these studies are successful, cPRGDA(DOTA) may be radiolabeled with other radionuclides (ex. 111In, 68Ga, etc.) and investigated for additional in vivo radioimaging or radiotherapeutic applications.

Promising new bacteriochlorin photosensitizers

Photosensitizers should satisfy several criteria to be considered useful for PDT. One particularly important requirement is efficient ROS photogeneration. The studies described in chapter 3 showed that bacteriochlorins synthesized by taurine-driven exocyclic E-ring opening generally produced more ROS compared to E-ring containing derivatives (163). In addition, the inclusion of anionic substituents also enhanced ROS photogeneration in aqueous solutions. However, efficient ROS photogeneration is only one important aspect of PDT. Useful PSs should also exhibit potent PDT-activity, show minimal dark toxicity, have high water-solubility, be readily uptaken by cells, and localize at ROS-sensitive intracellular organelles (97-102). Future studies should continue to evaluate the described bacteriopheophorbide a derivatives for the aforementioned properties. At this point, however, we can provide some insights into which of these bacteriochlorins may serve as promising new PSs.

The photosensitizing ability of a portion of the described bacteriochlorin series (1a, 2a, 2b, 4a, 4b, 5a, and 5b) was screened using MTT-based cell viability assays. These were performed using four model cancer cell lines (22Rv1 prostate cancer, A549 non-small cell lung cancer, HT29 colon cancer, and U87 glioblastoma) and high concentrations (50 μM) of the bacteriochlorin that were available at the time (Figure A2.1). Surprisingly, these experiments revealed that efficient ROS photogeneration did not guarantee potent photosensitization. It was clear that compounds 4a and 5a, which produce relatively low levels of ROS, actually displayed better PDT activity compared to their respective derivatives synthesized by E-ring opening (4b and 5b) in all of the tested cell lines. Since both 4b and 5b contained two anionic substituents, it is possible that these compounds displayed poor PDT activity due to limited cell uptake

201 compared to derivatives that contained fewer anionic substituents. In contrast, the zwitterionic bacteriochlorin, 2b, that produced relatively high levels of ROS, exhibited PDT activity comparable to the positive control, 1a, in all of these tested cell lines. Even the cationic compound (2a), that generated relatively little ROS, consistently displayed efficient PDT activity. These results suggest that cationic moieties on bacteriochlorins may facilitate efficient PDT, and possibly compensate for poor ROS photogeneration, by enhancing cell uptake. This is not a novel suggestion. In fact, several groups have proposed that PDT activity may become hindered as the number of anionic or cationic substituents increase because this may render the PSs too polar to efficiently cross cell membranes by diffusion (306-309). Therefore, it would be quite valuable if future studies continued to evaluate the proposed design parameters to determine i) if cationic moieties generally enhance cell uptake to improve PDT activity, ii) if anionic moieties perturb cell uptake, and iii) if simple charged functional groups can deliver bacteriochlorin PSs to specific ROS-sensitive organelles (ex. mitochondria, nucleus, etc.).

Several previous reports have found links between water-solubility and the PDT activity of PSs. It is well-established that PS aggregation can severely reduce the efficiency of photophysical properties related to PDT (ex. ΦISC, ΦT, and ΦΔ) (97, 98, 308, 310, 311). While PSs should avoid aggregation, previous reports have also shown that increasing the hydrophobicity of PSs can generally enhance cell uptake and PDT efficacy (106, 146, 164, 308, 312). With this in mind, we proposed that water-solubility may have contributed to the aforementioned cell viability assay results. The partitioning coefficients (log P) of 1a, 2a, 2b, 4b, and 5b were calculated by either water: 1-octanol partitioning experiments (313) or by computer modeling (ACD software) to determine if 4b or 5b displayed especially different water-solubilities compared to 1a, 2a, and 2b. Unfortunately, these investigations did not find any relationship between log P and PDT activity for these compounds (Table A2.1). This adds support for the hypothesis that the charge of bacteriochlorin PSs may be a valuable contributor to photosensitizing activity. Therefore, future studies should continue to investigate the uptake patterns of charged bacteriochlorins and determine which moieties can afford efficient cell uptake of bacteriochlorins to ROS-sensitive intracellular sites.

In addition to modulating cell uptake patterns, charged moieties may also contribute to the type of PDT mechanism that is employed by PSs upon photoirradiation. Previous reports have suggested that the sulfonic acid (-SO3-) (141) and sulfenato (-SO-) (168) groups may contribute to enhancing the excited-state electron transfer pathway (type 1 PDT) of PSs. In this light, it is

202 possible that WST11 (1b from chapter 3) may predominantly exhibit type 1 PDT (162) partially due to electron transfer at the 135-sulfonic acid group. With this in mind, and the PDT efficacy screening results in hand, we proposed that the promising new zwitterionic bacteriochlorin, 2b, may also exhibit type 1 PDT activity.

The type 1 PDT pathway is not limited to the presence of O2. Therefore, if PSs undergo excited- state electron transfer, these should cause cell toxicity when illuminated under hypoxic or anoxic conditions. With the type 1 PDT negative control PS 1a (WST09 or Tookad) and the positive control 1b (WST11) in hand, we began optimizing clonogenic survival assay methods for eventually testing if the new bacteriochlorin, 2b, would display type 1 PDT activity under hypoxia. We chose to evaluate PDT activity using the RIF-1 (murine radiation-induced fibrosarcoma) cell line. Since this cell line was first reported in 1980 by Twentyman et al. (314), RIF-1 cells have been found to withstand harsh exposures to hypoxia (315) and were suggested to be an excellent choice for assaying the effects of both ionizing (ex. x-ray) and non-ionizing (ex. PDT) radiation therapy using in vitro and in vivo clonogenic assays (316). Briefly, RIF-1 cells were cultured on 12-well tissue culture plates, incubated with each of the three PSs (1 μM and 5 μM) in RPMI-1640 media (1 % DMSO, 0.05% cremophor, and 10 % FBS, v/v/v/) for 24 h at 37 °C, were irradiated at 740 nm (7.5 J/cm2) and were seeded at 50 and 100 cells/well (each in triplicate) on 6-well tissue culture plates (N=3). Under normoxic conditions, this experiment revealed that these PSs did not display significant dark toxicity, and showed that 1a was clearly the best PS between the three PSs that were tested (Figure A2.2). Interestingly, the new PS (2b) showed significantly better PDT activity compared to the control, 1b, at 5 μM concentrations.

Our western blot experiments found that a 24 h hypoxia (0.2 % O2) exposure significantly increased the expression of the hypoxia marker, HIF-1 α (Figure A2.3) (317). Therefore, future studies should use the described methods to test the PDT activity of the promising new PS, 2b, under hypoxia (after 24 h exposures) and determine if it operates via the type 1 PDT pathway. If this is the case, these experiments can provide additional evidence to suggest that the anionic sulfonic acid group can promote the excited-state electron transfer pathway in bacteriochlorin PSs. Because PSs that cause cell toxicity predominantly by type 1 PDT are elusive, confirming this design parameter would greatly contribute to future efforts that wish to discover superior new bacteriochlorin PSs for efficient PDT under hypoxia.

203 Future bacteriochlorophyllones

The new natural product analog described in chapter 4, 132-173-bacteriochlorophyllone a (12), was serendipitously synthesized while attempting to produce the 173-choline ester derivative (2a) of bacteriopheophorbide a (221). The reagents chosen for the initially-intended esterification (for synthesizing compound 1a in chapter 3) included the coupling reagent, HBTU, and the mild bases, DMAP and DIPEA. These were intended to activate the 173-carboxylic acid of bacteriopheophorbide a to drive the esterification with choline. However, as HBTU activated this group, it formed a hydroxybenzotriazole ester at the 173-position. With this strong leaving group at the 173-ester and the proximal 132-methyl ester, the mild basic conditions were suitable to drive an intramolecular Claisen condensation that formed the new F-ring. Future studies can consider two synthetic strategies for expanding the library of this class of bacteriochlorins. This facile reaction can be attempted using other bacteriochlorin precursors (and indeed chlorin and porphyrin analogues) with the aforementioned structures (i.e. 173-carboxylic acid, 132-methyl ester, and tertiary 132-carbon). In addition, compound 12 may be modified at the peripheral 31-, 131-, 173-positions and/or at the central 23-/24-positions. With an expanded library of these distinct F-ring containing macrocycles, researchers may discover key structural features that can tune the redox (i.e. stability, antioxidant potential, etc.) and photophysical properties (ex. ΦF,

ΦICS, ΦT, ΦΔ, etc.) of bacteriochlorophyllones.

Compound 12 clearly possessed photosensitizing activity. Yet, it is only the second reported exocyclic F-ring containing bacteriochlorin, and there are no studies that have evaluated the photosensitizing ability of its counterpart, 11. Because so little is known about this class of bacteriochlorins, future research efforts may explore several avenues. Two especially interesting options are to i) create metalated derivatives of 12 to enhance PDT-efficacy, and to ii) conjugate cancer biomarker-targeting peptides for targeted-PDT applications. Diamagnetic metal such as Pd, In, and Zn may be easily inserted to afford new derivatives capable of efficient ROS photogeneration. In addition, reductive amination may be performed at the 31- or the 131- ketone of bacteriopheophorbide a before the 132-173 cyclization to conjugate cancer-targeting peptides through the N-terminus or the ε–NH2 of lysine residues. It is envisioned that these two strategies can be used to create highly photoactive bacteriochlorophyllones with great specificity towards biomarker-expressing cancer cells or to cancer-associated neovasculature.

204 Targeted photodynamic therapy with RGD peptides and bacteriochlorin photosensitizers

The overarching goal of this thesis was to create new cancer-targeting moieties (i.e. cyclic peptides) and potent light-activated drugs (i.e. bacteriochlorin PSs) that could be combined to develop novel cancer-specific PDT agents. In chapter 2, two new aziridine aldehyde-cyclized integrin-targeting peptides (cPRGDA and cPRGDAA) were shown to display binding-specificity towards αVβ3. In chapter 3, four bacteriochlorin PSs (2b – 5b) were found to efficiently generate ROS in situ upon illumination in aqueous solutions. With these results in hand, we now propose that future studies should i) evaluate the in vivo biodistribution, clearance, and stability of the aziridine aldehyde-cyclized RGD peptides, ii) evaluate the in vivo fate and PDT activity of 2b, iii) and evaluate the in vitro and in vivo properties of conjugates composed of aziridine aldehyde- cyclized peptides (ex. cPRGDA) and conjugatable bacteriochlorin PSs that efficiently generate ROS (ex. 1b and 4b).

With these proposed animal studies in mind, we first chose to evaluate the in vitro performance of peptide-PS conjugates to help guide future in vivo studies. Specifically, we chose to test the PDT activity of a peptide-PS conjugate composed of the conjugatable PS, 4b, and a simple model RGD-based linear peptide, Ac-PRGDK. The fluorescein-derivative of this linear RGD- peptide was characterized in vitro (using the same conditions described in chapter 2) and it was found to bind to integrin-overexpressing U87 glioblastoma cells (IC50 = 2.6 ± 0.7 µM) (Figure A3.1A) while avoiding binding to HT29 colon cancer cells that express negligible amounts of extracellular integrin receptors (Figure A3.1B). The PS 4b was chosen because, while it efficiently generated ROS in situ compared to Tookad (1a) (Figure A3.2A and B), it displayed relatively poor PDT activity towards three of the four tested cell lines we chose for the aforementioned in vitro PDT screening studies (Figure A3.2 C – F). Therefore, of the bacteriochlorin PSs described in chapter 3, 4b would benefit the most from a peptide-based targeting strategy. If successful, these reports would provide supporting data for the continued development of aziridine aldehyde-cyclized peptide (cPRGDA)-4b conjugates for integrin- targeted PDT.

A multistep synthetic strategy was designed and used to create a new peptide-PS conjugate composed of the model RGD-peptide and the PS, 4b (Ac-PRGDK(4b)). Briefly, upon R-group deprotection, the N-acetylated Ac-PRGDK-OH peptide was conjugated to 4b at the primary

205 amine of the lysine residue by NHS-chemistry (Scheme A3.1). An identical synthetic strategy was used to create a positive control peptide-PS conjugate composed of 1a and this peptide sequence (Ac-PRGDK(1a)) (Scheme A3.2). The ROS-activated probe, SOSG, was then used to quantify the relative ROS photogeneration of 1a, 4b, Ac-PRGDK(1a), and Ac-PRGDK(4b) in DMEM cell culture media (with 2% DMSO and 0.5% cremophor, v/v) using methods similar to those described in chapter 3. These experiments reiterated that 4b generated relatively more ROS compared to 1a, and found that peptide conjugation had little effect on the ROS-photogenerating ability of the PS components within Ac-PRGDK(1a) and Ac-PRGDK(4b) (Figure A3.3A).

We hypothesized that RGD-integrin binding would enhance the PDT activity of the chosen PSs. To this end, we first attempted to quantify the cell uptake of the four aforementioned compounds. However, this proved to be challenging. The ΦF of 1a (0.004 in ethanol) (222) and 4b (0.04 in methanol) (163) are quite low, and this made it difficult to accurately quantify cell uptake using fluorescence-based methods. While ICP-MS would accurately quantify the uptake of the Pd-containing 1a and Ac-PRGDK(1a), these methods would not be compatible with 4b or 5 -1 Ac-PRGDK(4b). The molar absorption extinction coefficients of 1a (ε763nm = 1.086 x 10 M -1 4 -1 -1 cm in CHCl3) (318) and 4b (ε747nm = 6.3 x 10 M cm in MeOH) (208) are sufficient to make cell uptake quantification feasible by absorbance-based methods. After incubating these compounds with U87 cells, the cells were treated with lysis buffer, and these samples were dried. We attempted to redissolved the compounds in organic solvent for spectrophotometric quantification, however, little or no absorbance signals were detected. While we were unsuccessful in our efforts, the PDT activities of these four compounds were still evaluated using MTT-based cell viability assays.

Standard MTT protocols were used to assess the dose-dependent dark toxicity and the PDT activity of Ac-PRGDK (control), 1a, 4b, Ac-PRGDK(1a), and Ac-PRGDK(4b) with αVβ3 integrin-overexpressing U87 cells. Briefly, these compounds were incubated with the cells for 4 h‘s at 37 ºC. After changing the cell media, plates were either left in the dark or were irradiated 2 with a 740 nm light box (4 J/cm ). As expected, 1a (LD50 = 0.8 µM) was more photoactive compared to 4b (LD50 = 10.2 µM) (Figure A3.3C and D). Surprisingly though, these results clearly showed that this model RGD-peptide abolished the PDT efficacy of both Ac-PRGDK(1a)

(LD50 > 100 µM) and Ac-PRGDK(4b) (LD50 > 100 µM) (Figure A3.3C and D). These preliminary results suggest that in vitro RGD-integrin binding can hinder the PDT activity of PSs. A possible explanation for this is that the RGD-motif may have enhanced extracellular

206 integrin-binding, but that this strategy reduced the cell uptake of the PSs. A previous report by Sancey et al. showed that cell uptake via integrin-targeting is most efficient when multimeric RGD-containing conjugates simultaneously bind to adjacent integrin receptors at the cell surface (63). Once several adjacent integrins are activated by a single molecule, the integrins will cluster and efficiently internalizes the bound molecule (319). Therefore, it is still possible that peptide- PS conjugates composed of multiple aziridine aldehyde-cyclized RGD macrocycles (i.e. dimers, trimmers, etc.) may successfully enhance the PDT activity of the conjugatable PSs described in this thesis (i.e. 1b and 4b). It is also important to note that integrin receptors can be overexpressed on tumour neovasculature (320). So it may still be possible for aziridine aldehyde- cyclized RGD monomers (and their respective multimers) to enhance the tumour-specific in vivo PDT activity of the described PSs by vascular-targeted photodynamic therapy (VPT) (321). In light of these previous reports and our recent preliminary results, we suggest that future studies should evaluate the in vitro and in vivo properties of conjugates composed of aziridine aldehyde- cyclized RGD-peptide monomers (or multimers) and conjugatable bacteriochlorin PSs (ex. 1b or 4b) to pursue the goal of peptide-targeted PDT.

Closing comments

Aziridine aldehyde-cyclized peptides can be synthesized by facile modular methods and hold immense potential for cancer biomarker-targeting. Yet, the biological properties of this class of cyclic peptides have not been extensively evaluated compared to conventional amide-cyclized and disulfide-cyclized cancer-targeting peptides. Similarly, bacteriochlorins display inherent photonic properties (i.e. NIR-Qy, high ε, efficient ΦT , etc.) that make them among the best classes of PSs for PDT applications. However, the bacteriochlorin scaffold has yet to be rationally modified in a manner that imparts ideal photophysical properties for PDT, as intended. Therefore, much is unknown about how these two classes of chemicals will behave in vivo; separately and as conjugates. The projects presented in this thesis, and the future directions that have been suggested, each provide knowledge and insights to the field of cancer targeted-PDT. It is envisioned that future studies will continue to take advantage of the attributes of the described classes of cyclic peptides and PSs to one day develop aziridine aldehyde-cyclized peptide- bacteriochlorin conjugates that are biocompatible, biologically/chemically stable, cancer- specific, and clinically relevant. Achieving this goal would have great benefits; both on the fundamental scientific level and possibly in clinical settings.

207 References

1. Á. Roxin, G. Zheng, Flexible or fixed: a comparative review of linear and cyclic cancer- targeting peptides. Future Medicinal Chemistry 4, 1601-1618 (2012).

2. R. M. J. Liskamp, D. T. S. Rijkers, S. E. Bakker, in Modern Supramolecular Chemistry. (Wiley-VCH Verlag GmbH & Co. KGaA, 2008), pp. 1-27.

3. L. D. D'Andrea et al., Targeting angiogenesis: Structural characterization and biological properties of a de novo engineered VEGF mimicking peptide. Proceedings of the National Academy of Sciences of the United States of America 102, 14215-14220 (2005).

4. H. Kessler, Conformation and Biological Activity of Cyclic Peptides. Angewandte Chemie International Edition in English 21, 512-523 (1982).

5. C. Ramakrishnan, P. K. C. Paul, K. Ramnarayan, Cyclic peptides — Small and big and their conformational aspects. Journal of Biosciences 8, 239-251 (1985).

6. C. M. Deber, V. Madison, E. R. Blout, Why cyclic peptides? Complementary approaches to conformations. Accounts of Chemical Research 9, 106-113 (1976).

7. J. Grünewald, M. A. Marahiel, Chemoenzymatic and template-directed synthesis of bioactive macrocyclic peptides. Microbiology and molecular biology reviews : MMBR 70, 121-146 (2006).

8. O. H. Aina, T. C. Sroka, M.-L. Chen, K. S. Lam, Therapeutic cancer targeting peptides. Biopolymers 66, 184-199 (2002).

9. E. R. Blout, Cyclic peptides: Past, present, and future. Biopolymers 20, 1901-1912 (1981).

10. M. W. Deem, J. S. Bader, A configurational bias Monte Carlo method for linear and cyclic peptides. Molecular Physics 87, 1245-1260 (1996).

11. M. S. Lee, B. Gardner, M. Kahn, H. Nakanishi, The three-dimensional solution structure of a constrained peptidomimetic in water and in chloroform observation of solvent induced hydrophobic cluster. FEBS Letters 359, 113-118 (1995).

12. K. Uma, R. Kishore, P. Balaram, Stereochemical constraints in peptide design: Analysis of the influence of a disulfide bridge and an α-aminoisobutyryl residue on the conformation of a hexapeptide. Biopolymers 33, 865-871 (1993).

13. D. Lau, L. Guo, R. Liu, J. Marik, K. Lam, Peptide ligands targeting integrin alpha3beta1 in non-small cell lung cancer. Lung cancer (Amsterdam, Netherlands) 52, 291-297 (2006).

14. K. D. Kopple, Synthesis of cyclic peptides. Journal of Pharmaceutical Sciences 61, 1345-1356 (1972).

208 15. J. N. Lambert, J. P. Mitchell, K. D. Roberts, The synthesis of cyclic peptides. Journal of the Chemical Society, Perkin Transactions 1, 471-484 (2001).

16. J. S. Davies, The cyclization of peptides and depsipeptides. Journal of Peptide Science 9, 471-501 (2003).

17. C. J. White, A. K. Yudin, Contemporary strategies for peptide macrocyclization. Nature chemistry 3, 509-524 (2011).

18. A. Kumar et al., Synthesis and Structure−Activity Relationships of Linear and Conformationally Constrained Peptide Analogues of CIYKYY as Src Tyrosine Kinase Inhibitors. Journal of Medicinal Chemistry 49, 3395-3401 (2006).

19. N. Assa-Munt, X. Jia, P. Laakkonen, E. Ruoslahti, Solution Structures and Integrin Binding Activities of an RGD Peptide with Two Isomers†. Biochemistry 40, 2373-2378 (2001).

20. M. I. García-Aranda et al., Disulfide and amide-bridged cyclic peptide analogues of the VEGF₈₁₋₉₁ fragment: synthesis, conformational analysis and biological evaluation. Bioorganic & medicinal chemistry 19, 7526-7533 (2011).

21. M. A. Gallop, R. W. Barrett, W. J. Dower, S. P. A. Fodor, E. M. Gordon, Applications of Combinatorial Technologies to Drug Discovery. 1. Background and Peptide Combinatorial Libraries. Journal of Medicinal Chemistry 37, 1233-1251 (1994).

22. C. P. Scott, E. Abel-Santos, A. D. Jones, S. J. Benkovic, Structural requirements for the biosynthesis of backbone cyclic peptide libraries. Chemistry & biology 8, 801-815 (2001).

23. O. H. Aina, J. Marik, R. Liu, D. H. Lau, K. S. Lam, Identification of novel targeting peptides for human ovarian cancer cells using ―one-bead one-compound‖ combinatorial libraries. Molecular Cancer Therapeutics 4, 806-813 (2005).

24. O. H. Aina et al., From Combinatorial Chemistry to Cancer-Targeting Peptides. Molecular Pharmaceutics 4, 631-651 (2007).

25. S. Bonetto, L. Spadola, A. G. Buchanan, L. Jermutus, J. Lund, Identification of cyclic peptides able to mimic the functional epitope of IgG1-Fc for human FcγRI. The FASEB Journal 23, 575-585 (2009).

26. J.-R. Hsiao et al., Cyclic αvβ6-targeting peptide selected from biopanning with clinical potential for head and neck squamous cell carcinoma. Head & Neck 32, 160-172 (2010).

27. K. E. Duncan et al., Discovery and characterization of a nonphosphorylated cyclic peptide inhibitor of the peptidylprolyl isomerase, Pin1. Journal of medicinal chemistry 54, 3854-3865 (2011).

28. A. B. Pomilio, S. M. Battista, A. A. Vitale, 1 . Quantitative structure-activity relationship ( QSAR ) studies on bioactive cyclopeptides. (2010), vol. 661, pp. 1-34.

209 29. M. J. I. Andrews et al., Design, synthesis, biological activity and structural analysis of cyclic peptide inhibitors targeting the substrate recruitment site of cyclin-dependent kinase complexes. Organic & biomolecular chemistry 2, 2735-2741 (2004).

30. Y. Zhou, S. Chakraborty, S. Liu, Radiolabeled Cyclic RGD Peptides as Radiotracers for Imaging Tumors and Thrombosis by SPECT. Theranostics 1, 58-82 (2011).

31. M. Pfaff et al., Selective recognition of cyclic RGD peptides of NMR defined conformation by alpha IIb beta 3, alpha V beta 3, and alpha 5 beta 1 integrins. Journal of Biological Chemistry 269, 20233-20238 (1994).

32. R. Haubner et al., Structural and Functional Aspects of RGD-Containing Cyclic Pentapeptides as Highly Potent and Selective Integrin αVβ3 Antagonists. Journal of the American Chemical Society 118, 7461-7472 (1996).

33. A. C. Bach et al., Type II‗ to Type I β-Turn Swap Changes Specificity for Integrins. Journal of the American Chemical Society 118, 293-294 (1996).

34. J. Wermuth, S. Goodman, A. Jonczyk, H. Kessler, Stereoisomerism and biological activity of the selective and superactive αvβ3 integrin inhibitor cyclo (-RGDfV-) and its retro-inverso peptide. Journal of the American Chemical Society 119, 1328-1335 (1997).

35. C. Gilon, D. Halle, M. Chorev, Z. Selincer, G. Byk, Backbone cyclization: a new method for conferring conformational constraint on peptides. Biopolymers 31, 745-750 (1991).

36. S. Bogdanowich‐Knipp, D. Jois, T. Siahaan, The effect of conformation on the solution stability of linear vs. cyclic RGD peptides. The Journal of peptide research 53, 523-529 (1999).

37. S. J. Bogdanowich‐Knipp, S. Chakrabarti, T. J. Siahaan, T. D. Williams, R. K. Dillman, Solution stability of linear vs. cyclic RGD peptides. The Journal of peptide research 53, 530-541 (1999).

38. L. Yang et al., Synthesis and biological activities of potent peptidomimetics selective for somatostatin receptor subtype 2. Proceedings of the National Academy of Sciences 95, 10836-10841 (1998).

39. F. Diel, H. Borck, S. Hosenfeld, Effects of somatostatin on ethanol-induced gastric erosions in the rat: role of mast cells. Agents and actions 18, 273-275 (1986).

40. L. Vecsei, I. Bollok, J. Varga, B. Penke, G. Telegdy, The effect of somatostatin, its fragments and an analog on electroconvulsive shock-induced amnesia in rats. Neuropeptides 4, 137-143 (1984).

41. J. M. Rosenberg, Octreotide: a synthetic analog of somatostatin. The Annals of Pharmacotherapy 22, 748-754 (1988).

42. B. M. Evers, D. Parekh, C. M. Townsend, J. C. Thompson, Somatostatin and analogues in the treatment of cancer. A review. Annals of surgery 213, 190-198 (1991).

210 43. L. Anthony, P. U. Freda, From somatostatin to octreotide LAR: evolution of a somatostatin analogue. Current Medical Research & Opinion 25, 2989-2999 (2009).

44. D. Besser et al., Synthesis and characterization of octapeptide somatostatin analogues with backbone cyclization: comparison of different strategies, biological activities and enzymatic stabilities. Journal für praktische Chemie 342, 537-545 (2000).

45. M. Pakkala et al., Activity and stability of human kallikrein‐2‐specific linear and cyclic peptide inhibitors. Journal of Peptide Science 13, 348-353 (2007).

46. X. Ma, C. Wu, W. Wang, X. Li, Peptides from Plants : A New Source for Antitumor Drug Research.

47. S. Namjoshi, H. a. E. Benson, Cyclic peptides as potential therapeutic agents for skin disorders. Biopolymers 94, 673-680 (2010).

48. A. Aneiros, A. Garateix, Bioactive peptides from marine sources: pharmacological properties and isolation procedures. Journal of Chromatography B 803, 41-53 (2004).

49. Z. Liu, F. Wang, X. Chen, Integrin αvβ3-targeted cancer therapy. Drug development research 69, 329-339 (2008).

50. P. C. Oyston, M. A. Fox, S. J. Richards, G. C. Clark, Novel peptide therapeutics for treatment of infections. Journal of medical microbiology 58, 977-987 (2009).

51. G. Schwartsmann, A. B. da Rocha, R. G. Berlinck, J. Jimeno, Marine organisms as a source of new anticancer agents. The lancet oncology 2, 221-225 (2001).

52. L. Zilberberg et al., Structure and inhibitory effects on angiogenesis and tumor development of a new vascular endothelial growth inhibitor. The Journal of biological chemistry 278, 35564-35573 (2003).

53. J. Haier, U. Goldmann, B. Hotz, N. Runkel, U. Keilholz, Inhibition of tumor progression and neoangiogenesis using cyclic RGD-peptides in a chemically induced colon carcinoma in rats. Clinical & experimental metastasis 19, 665-672 (2002).

54. F. Eskens et al., Phase I and pharmacokinetic study of continuous twice weekly intravenous administration of Cilengitide (EMD 121974), a novel inhibitor of the integrins αvβ3 and αvβ5 in patients with advanced solid tumours. European Journal of Cancer 39, 917-926 (2003).

55. J.-D. Raguse, H. J. Gath, J. Bier, H. Riess, H. Oettle, Cilengitide (EMD 121974) arrests the growth of a heavily pretreated highly vascularised head and neck tumour. Oral Oncology 40, 228-230 (2004).

56. M. R. Gilbert et al., Cilengitide in patients with recurrent glioblastoma: the results of NABTC 03-02, a phase II trial with measures of treatment delivery. Journal of neuro- oncology 106, 147-153 (2012).

211 57. R. Haubner et al., Glycosylated RGD-containing peptides: tracer for tumor targeting and angiogenesis imaging with improved biokinetics. Journal of Nuclear Medicine 42, 326- 336 (2001).

58. S. Achilefu et al., Synergistic effects of light-emitting probes and peptides for targeting and monitoring integrin expression. Proceedings of the National Academy of Sciences 102, 7976-7981 (2005).

59. S. M. Okarvi, Peptide-based radiopharmaceuticals: future tools for diagnostic imaging of cancers and other diseases. Medicinal research reviews 24, 357-397 (2004).

60. H. Lundqvist, V. Tolmachev, Targeting peptides and positron emission tomography. Peptide Science 66, 381-392 (2002).

61. R. R. P. Warner, T. M. O'Dorisio, Radiolabeled peptides in diagnosis and tumor imaging: clinical overview. Seminars in nuclear medicine 32, 79-83 (2002).

62. M. Fani et al., Comparative evaluation of linear and cyclic 99mTc-RGD peptides for targeting of integrins in tumor angiogenesis. Anticancer research 26, 431-434 (2006).

63. S. Liu, Radiolabeled cyclic RGD peptides as integrin αvβ3-targeted radiotracers: maximizing binding affinity via bivalency. Bioconjugate chemistry 20, 2199-2213 (2009).

64. J. Chen et al., Evaluation of an 111In-DOTA–Rhenium Cyclized α-MSH Analog: A Novel Cyclic-Peptide Analog with Improved Tumor-Targeting Properties. Journal of Nuclear Medicine 42, 1847-1855 (2001).

65. S. Raha, T. Paunesku, G. Woloschak, Peptide-mediated cancer targeting of nanoconjugates. Wiley interdisciplinary reviews. Nanomedicine and nanobiotechnology 3, 269-281 (2011).

66. M. J. Kogan et al., Peptides and metallic nanoparticles for biomedical applications. Nanomedicine (London, England) 2, 287-306 (2007).

67. J. B. Delehanty et al., Peptides for specific intracellular delivery and targeting of nanoparticles: implications for developing nanoparticle-mediated drug delivery. Therapeutic Delivery 1, 411-433 (2010).

68. A. H. Negussie et al., Synthesis and in vitro evaluation of cyclic NGR peptide targeted thermally sensitive liposome. Journal of Controlled Release 143, 265-273 (2010).

69. M. C. Morris, S. Deshayes, F. Heitz, G. Divita, Cell-penetrating peptides: from molecular mechanisms to therapeutics. Biology of the Cell 100, 201-217 (2008).

70. V. Kersemans, K. Kersemans, B. Cornelissen, Cell penetrating peptides for in vivo molecular imaging applications. Current pharmaceutical design 14, 2415-2427 (2008).

71. V. Kersemans, B. Cornelissen, Targeting the tumour: Cell penetrating peptides for molecular imaging and radiotherapy. Pharmaceuticals 3, 600-620 (2010).

212 72. B. G. Bitler, J. A. Schroeder, Anti-cancer therapies that utilize cell penetrating peptides. Recent Patents on Anti-Cancer Drug Discovery 5, 99-108 (2010).

73. K. N. Sugahara et al., Coadministration of a tumor-penetrating peptide enhances the efficacy of cancer drugs. Science (New York, N.Y.) 328, 1031-1035 (2010).

74. D. Mandal, A. Nasrolahi Shirazi, K. Parang, Cell-Penetrating Homochiral Cyclic Peptides as Nuclear-Targeting Molecular Transporters. Angewandte Chemie International Edition, n/a-n/a (2011).

75. G. Colombo et al., Structure-activity relationships of linear and cyclic peptides containing the NGR tumor-homing motif. Journal of Biological Chemistry 277, 47891- 47897 (2002).

76. G. J. Mizejewski, M. Muehlemann, M. Dauphinee, Update of alpha fetoprotein growth- inhibitory peptides as biotherapeutic agents for tumor growth and metastasis. Chemotherapy 52, 83-90 (2006).

77. G. J. Mizejewski, R. MacColl, α-Fetoprotein growth inhibitory peptides: Potential leads for cancer therapeutics. Molecular Cancer Therapeutics 2, 1243-1255 (2003).

78. F. B. Mesfin, S. Zhu, J. A. Bennett, T. T. Andersen, H. I. Jacobson, Development of a synthetic cyclized peptide derived from α-fetoprotein that prevents the growth of human breast cancer. The Journal of Peptide Research 58, 246-256 (2001).

79. E. A. Woltering et al., Somatostatin analogs: angiogenesis inhibitors with novel mechanisms of action. Investigational new drugs 15, 77-86 (1997).

80. W. R. Baumbach et al., A linear hexapeptide somatostatin antagonist blocks somatostatin activity in vitro and influences growth hormone release in rats. Molecular pharmacology 54, 864-873 (1998).

81. Y.-U. Kwon, T. Kodadek, Quantitative comparison of the relative cell permeability of cyclic and linear peptides. Chemistry & biology 14, 671-677 (2007).

82. P. Fischer et al., Structure–activity relationship of truncated and substituted analogues of the intracellular delivery vector Penetratin. The Journal of peptide research 55, 163-172 (2000).

83. J. Zhong, Y. Chau, Antitumor activity of a membrane lytic peptide cyclized with a linker sensitive to membrane type 1-matrix metalloproteinase. Molecular Cancer Therapeutics 7, 2933-2940 (2008).

84. M. F. Giblin, N. Wang, T. J. Hoffman, S. S. Jurisson, T. P. Quinn, Design and characterization of α-melanotropin peptide analogs cyclized through rhenium and technetium metal coordination. Proceedings of the National Academy of Sciences 95, 12814-12818 (1998).

85. A. M. Colangelo et al., A new nerve growth factor-mimetic peptide active on neuropathic pain in rats. The journal of neuroscience 28, 2698-2709 (2008).

213 86. L. A. Greene, A. S. Tischler, Establishment of a noradrenergic clonal line of rat adrenal pheochromocytoma cells which respond to nerve growth factor. Proceedings of the National Academy of Sciences 73, 2424-2428 (1976).

87. M. V. Chao, Neurotrophins and their receptors: a convergence point for many signalling pathways. Nature Reviews Neuroscience 4, 299-309 (2003).

88. E. Huang, L. Reichardt, Trk receptors: Roles in neuronal signal transduction. Annual Review of Biochemistry 72, 609-642 (2003).

89. E. Adriaenssens et al., Nerve growth factor is a potential therapeutic target in breast cancer. Cancer research 68, 346-351 (2008).

90. V. Steiner et al., Analysis of synthetic peptides using matrix-assisted laser desorption ionization mass spectrometry. Peptide research 5, 25-29 (1992).

91. T. W. Liu, J. Chen, G. Zheng, Peptide-based molecular beacons for cancer imaging and therapy. Amino acids 41, 1123-1134 (2011).

92. V. C. Blank, C. Peña, L. P. Roguin, A cyclic chimeric interferon-α2b peptide induces apoptosis in tumor cells. Cancer biology & therapy 6, 1787-1793 (2007).

93. C. Peña, V. C. Blank, V. J. Marino, L. P. Roguin, Synthesis and biological properties of chimeric interferon-α2b peptides. Peptides 26, 1144-1149 (2005).

94. E. J. Witsch et al., Generation and characterization of peptide mimotopes specific for anti ErbB-2 monoclonal antibodies. International immunology 23, 391-403 (2011).

95. S. F. Brady et al., Design and synthesis of a pro-drug of vinblastine targeted at treatment of prostate cancer with enhanced efficacy and reduced systemic toxicity. Journal of Medicinal Chemistry 45, 4706-4715 (2002).

96. P. Gomes, N. Vale, R. Moreira, Cyclization-activated prodrugs. Molecules 12, 2484-2506 (2007).

97. M. C. DeRosa, R. J. Crutchley, Photosensitized singlet oxygen and its applications. Coordination Chemistry Reviews 233, 351-371 (2002).

98. I. J. Macdonald, T. J. Dougherty, Basic principles of photodynamic therapy. Journal of Porphyrins and Phthalocyanines 5, 105-129 (2001).

99. R. R. Allison et al., Photosensitizers in clinical PDT. Photodiagnosis and photodynamic therapy 1, 27-42 (2004).

100. R. R. Allison, C. H. Sibata, Oncologic photodynamic therapy photosensitizers: a clinical review. Photodiagnosis and photodynamic therapy 7, 61-75 (2010).

101. R. Weissleder, A clearer vision for in vivo imaging. Nature biotechnology 19, 316-316 (2001).

214 102. P. Agostinis et al., Photodynamic therapy of cancer: an update. CA: a cancer journal for clinicians 61, 250-281 (2011).

103. C. Sibata, V. Colussi, N. Oleinick, T. Kinsella, Photodynamic therapy: a new concept in medical treatment. Brazilian Journal of Medical and Biological Research 33, 869-880 (2000).

104. A. S. Brandis, Y. Salomon, A. Scherz, in Chlorophylls and Bacteriochlorophylls. (Springer, 2006), pp. 485-494.

105. M. R. Detty, Photosensitisers for the photodynamic therapy of cancer and other diseases. Expert Opinion on Therapeutic Patents 11, 1849-1860 (2001).

106. M. Ochsner, Photophysical and photobiological processes in the photodynamic therapy of tumours. Journal of Photochemistry and Photobiology B: Biology 39, 1-18 (1997).

107. M. Oertel et al., Novel bacteriochlorine for high tissue-penetration: photodynamic properties in human biliary tract cancer cells in vitro and in a mouse tumour model. Journal of Photochemistry and Photobiology B: Biology 71, 1-10 (2003).

108. J. Hasegawa, Y. Ozeki, K. Ohkawa, M. Hada, H. Nakatsuji, Theoretical study of the excited states of chlorin, bacteriochlorin, pheophytin a, and chlorophyll a by the SAC/SAC-CI method. The Journal of Physical Chemistry B 102, 1320-1326 (1998).

109. L. Huang et al., Stable synthetic cationic bacteriochlorins as selective antimicrobial photosensitizers. Antimicrobial agents and chemotherapy 54, 3834-3841 (2010).

110. P. Mroz et al., Stable synthetic bacteriochlorins overcome the resistance of melanoma to photodynamic therapy. The FASEB Journal 24, 3160-3170 (2010).

111. E. Yang et al., Photophysical properties and electronic structure of stable, tunable synthetic bacteriochlorins: Extending the features of native photosynthetic pigments. The Journal of Physical Chemistry B 115, 10801-10816 (2011).

112. H. Ali, J. E. Van Lier, Metal complexes as photo-and radiosensitizers. Chemical Reviews 99, 2379-2450 (1999).

113. C. Ruzié, M. Krayer, T. Balasubramanian, J. S. Lindsey, Tailoring a bacteriochlorin building block with cationic, amphipathic, or lipophilic substituents. The Journal of organic chemistry 73, 5806-5820 (2008).

114. S.-i. Sasaki, H. Tamiaki, Synthesis and optical properties of bacteriochlorophyll-a derivatives having various C3 substituents on the bacteriochlorin π-system. The Journal of organic chemistry 71, 2648-2654 (2006).

115. M. Taniguchi et al., Accessing the near-infrared spectral region with stable, synthetic, wavelength-tunable bacteriochlorins. New Journal of Chemistry 32, 947-958 (2008).

116. Z. Yu, M. Ptaszek, Multifunctional bacteriochlorins from selective palladium-coupling reactions. Organic letters 14, 3708-3711 (2012).

215 117. Z. Yu, M. Ptaszek, Near-IR Emissive Chlorin–Bacteriochlorin Energy-Transfer Dyads with a Common Donor and Acceptors with Tunable Emission Wavelength. The Journal of organic chemistry 78, 10678-10691 (2013).

118. M. Kunieda, T. Mizoguchi, H. Tamiaki, Syntheses and optical properties of stable 8- alkylidene-bacteriochlorins mimicking the molecular structures of natural bacteriochlorophylls-b and g. Tetrahedron 60, 11349-11357 (2004).

119. E. Yang et al., Photophysical Properties and Electronic Structure of Bacteriochlorin– Chalcones with Extended Near‐Infrared Absorption. Photochemistry and photobiology 89, 586-604 (2013).

120. R. K. Pandey et al., Nature: A rich source for developing multifunctional agents. tumor‐imaging and photodynamic therapy. Lasers in Surgery and Medicine 38, 445-467 (2006).

121. P. Joshi et al., Synthesis, Spectroscopic, and in Vitro Photosensitizing Efficacy of Ketobacteriochlorins Derived from Ring-B and Ring-D Reduced Chlorins via Pinacol– Pinacolone Rearrangement. The Journal of organic chemistry 76, 8629-8640 (2011).

122. M. Kunieda, H. Tamiaki, Self-aggregation of synthetic zinc oxo-bacteriochlorins bearing substituents characteristic of chlorosomal chlorophylls. The Journal of organic chemistry 70, 820-828 (2005).

123. X.-F. Wang, O. Kitao, H. Zhou, H. Tamiaki, S.-i. Sasaki, Efficient dye-sensitized solar cell based on oxo-bacteriochlorin sensitizers with broadband absorption capability. The Journal of Physical Chemistry C 113, 7954-7961 (2009).

124. A. Gorman et al., In vitro demonstration of the heavy-atom effect for photodynamic therapy. Journal of the American Chemical Society 126, 10619-10631 (2004).

125. S. Hirohara et al., Sugar and Heavy Atom Effects of Glycoconjugated Chlorin Palladium Complex on Photocytotoxicity. Bioconjugate chemistry 23, 1881-1890 (2012).

126. N. Picard, H. Ali, J. E. van Lier, K. Klarskov, B. Paquette, Bromines on N-allyl position of cationic porphyrins affect both radio-and photosensitizing properties. Photochemical & photobiological sciences 8, 224-232 (2009).

127. B. Roeder, D. U. Nather, in Berlin-DL tentative. (International Society for Optics and Photonics, 1991), pp. 377-384.

128. A. P. Castano, T. N. Demidova, M. R. Hamblin, Mechanisms in photodynamic therapy: part one—photosensitizers, photochemistry and cellular localization. Photodiagnosis and Photodynamic Therapy 1, 279-293 (2004).

129. J. Leanne B, B. Ross W, Photodynamic therapy and the development of metal-based photosensitisers. Metal-based drugs 2008, (2008).

130. M. Ethirajan, Y. Chen, P. Joshi, R. K. Pandey, The role of porphyrin chemistry in tumor imaging and photodynamic therapy. Chemical Society Reviews 40, 340-362 (2011).

216 131. M. Obata et al., In vitro heavy-atom effect of palladium (II) and platinum (II) complexes of pyrrolidine-fused chlorin in photodynamic therapy. Journal of medicinal chemistry 52, 2747-2753 (2009).

132. C.-Y. Chen et al., Synthesis and physicochemical properties of metallobacteriochlorins. Inorganic chemistry 51, 9443-9464 (2012).

133. E. Yang et al., Molecular electronic tuning of photosensitizers to enhance photodynamic therapy: Synthetic dicyanobacteriochlorins as a case study. Photochemistry and photobiology 89, 605-618 (2013).

134. C. Musewald et al., Time-resolved spectral investigation of bacteriochlorophyll a and its transmetalated derivatives [Zn]-bacteriochlorophyll a and [Pd]-bacteriochlorophyll a. The Journal of Physical Chemistry B 102, 8336-8342 (1998).

135. S. Fukuzumi et al., Metal bacteriochlorins which act as dual singlet oxygen and superoxide generators. The Journal of Physical Chemistry B 112, 2738-2746 (2008).

136. R. Bonnett, G. Mart nez, Photobleaching of sensitisers used in photodynamic therapy. Tetrahedron 57, 9513-9547 (2001).

137. M. R. Detty, S. L. Gibson, S. J. Wagner, Current clinical and preclinical photosensitizers for use in photodynamic therapy. Journal of medicinal chemistry 47, 3897-3915 (2004).

138. M. Pineiro, R. Gonsalves, M. M. Pereira, S. J. Formosinho, L. G. Arnaut, New halogenated phenylbacteriochlorins and their efficiency in singlet-oxygen sensitization. The Journal of Physical Chemistry A 106, 3787-3795 (2002).

139. M. M. Pereira et al., Synthesis and photophysical characterization of a library of photostable halogenated bacteriochlorins: an access to near infrared chemistry. Tetrahedron 66, 9545-9551 (2010).

140. E. l. G. Azenha et al., Heavy-atom effects on metalloporphyrins and polyhalogenated porphyrins. Chemical physics 280, 177-190 (2002).

141. E. F. Silva et al., Mechanisms of Singlet‐Oxygen and Superoxide‐Ion Generation by Porphyrins and Bacteriochlorins and their Implications in Photodynamic Therapy. Chemistry-A European Journal 16, 9273-9286 (2010).

142. J. M. Dąbrowski et al., Combined effects of singlet oxygen and hydroxyl radical in photodynamic therapy with photostable bacteriochlorins: evidence from intracellular fluorescence and increased photodynamic efficacy in vitro. Free Radical Biology and Medicine 52, 1188-1200 (2012).

143. K. R. Reddy et al., Synthetic bacteriochlorins with integral spiro-piperidine motifs. New Journal of Chemistry 37, 1157-1173 (2013).

144. K. R. Reddy et al., Palette of lipophilic bioconjugatable bacteriochlorins for construction of biohybrid light-harvesting architectures. Chemical Science 4, 2036-2053 (2013).

217 145. M. Krayer et al., Synthesis and photophysical characterization of stable indium bacteriochlorins. Inorganic chemistry 50, 4607-4618 (2011).

146. Y.-Y. Huang et al., In vitro photodynamic therapy and quantitative structure− activity relationship studies with stable synthetic near-infrared-absorbing bacteriochlorin photosensitizers. Journal of medicinal chemistry 53, 4018-4027 (2010).

147. T. Fukuda, E. A. Makarova, N. Kobayashi, Synthesis and Spectroscopic and Electrochemical Studies of Novel Benzo‐or 2, 3‐Naphtho‐Fused Tetraazachlorins, Bacteriochlorins, and Isobacteriochlorins. Chemistry-A European Journal 10, 117-133 (2004).

148. L. P. Samankumara et al., Expanded bacteriochlorins. Angewandte Chemie International Edition 51, 5757-5760 (2012).

149. S. Pervaiz, M. Olivo, Art and science of photodynamic therapy. Clinical and experimental pharmacology and physiology 33, 551-556 (2006).

150. F. Stewart, P. Baas, W. Star, What does photodynamic therapy have to offer radiation oncologists (or their cancer patients)? Radiotherapy and oncology 48, 233-248 (1998).

151. G. Laustriat, Molecular mechanisms of photosensitization. Biochimie 68, 771-778 (1986).

152. E. Sternberg, D. Dolphin, Pyrrolic photosensitizers. Current medicinal chemistry 3, 239- 272 (1996).

153. A. Scherz, Y. Salomon, A. Brandis, H. Scheer. (Google Patents, 2003).

154. Z. Huang et al., Studies of a vascular‐acting photosensitizer, Pd‐bacteriopheophorbide (Tookad), in normal canine prostate and spontaneous canine prostate cancer. Lasers in surgery and medicine 36, 390-397 (2005).

155. R. A. Weersink et al., Assessment of Cutaneous Photosensitivity of TOOKAD (WST09) in Preclinical Animal Models and in Patients. Photochemistry and Photobiology 81, 106- 113 (2005).

156. Z. Huang et al., Effects of Pd-bacteriopheophorbide (TOOKAD)-mediated photodynamic therapy on canine prostate pretreated with ionizing radiation. Radiation research 161, 723-731 (2004).

157. N. V. Koudinova et al., Photodynamic therapy with Pd‐bacteriopheophorbide (TOOKAD): Successful in vivo treatment of human prostatic small cell carcinoma xenografts. International journal of cancer 104, 782-789 (2003).

158. Y. Vakrat-Haglili et al., The microenvironment effect on the generation of reactive oxygen species by Pd-bacteriopheophorbide. Journal of the American Chemical Society 127, 6487-6497 (2005).

159. L. F. Agnez-Lima et al., DNA damage by singlet oxygen and cellular protective mechanisms. Mutation Research/Reviews in Mutation Research 751, 15-28 (2012).

218 160. A. Scherz, Y. Salomon, O. Mazor, H. Scheer. (US Patent 20,060,142,260, 2006).

161. A. Brandis et al., Novel Water‐soluble Bacteriochlorophyll Derivatives for Vascular‐targeted Photodynamic Therapy: Synthesis, Solubility, Phototoxicity and the Effect of Serum Proteins¶. Photochemistry and Photobiology 81, 983-992 (2005).

162. I. Ashur et al., Photocatalytic generation of oxygen radicals by the water-soluble bacteriochlorophyll derivative WST11, noncovalently bound to serum albumin. The Journal of Physical Chemistry A 113, 8027-8037 (2009).

163. Á. Roxin, J. Chen, A. S. Paton, T. P. Bender, G. Zheng, Modulation of reactive oxygen species photogeneration of bacteriopheophorbide a derivatives by exocyclic E-ring- opening and charge modifications. Journal of Medicinal Chemistry, (2013).

164. J. Dandler, B. Wilhelm, H. Scheer, Photochemistry of Bacteriochlorophylls in Human Blood Plasma: 1. Pigment Stability and Light‐induced Modifications of Lipoproteins. Photochemistry and photobiology 86, 331-341 (2010).

165. M. Hoebeke et al., Electron spin resonance evidence of the generation of superoxide anion, hydroxyl radical and singlet oxygen during the photohemolysis of human erythrocytes with bacteriochlorin a. Photochemistry and photobiology 66, 502-508 (1997).

166. T. Y. Wang, J. R. Chen, J. S. Ma, Diphenylchlorin and diphenylbacteriochlorin: synthesis, spectroscopy and photosensitising properties. Dyes and pigments 52, 199-208 (2002).

167. D. E. Dolmans, D. Fukumura, R. K. Jain, Photodynamic therapy for cancer. Nature Reviews Cancer 3, 380-387 (2003).

168. R. Rajagopalan et al., in 12th World Congress of the International Photodynamic Association. (International Society for Optics and Photonics, 2009), pp. 738027-738027- 738028.

169. A. r. Roxin et al., Conformational Modulation of in Vitro Activity of Cyclic RGD Peptides via Aziridine Aldehyde-Driven Macrocyclization Chemistry. Bioconjugate chemistry 23, 1387-1395 (2012).

170. R. Hili, V. Rai, A. K. Yudin, Macrocyclization of linear peptides enabled by amphoteric molecules. Journal of the American Chemical Society 132, 2889-2891 (2010).

171. B. H. Rotstein, R. Mourtada, S. O. Kelley, A. K. Yudin, Solvatochromic Reagents for Multicomponent Reactions and their Utility in the Development of Cell‐Permeable Macrocyclic Peptide Vectors. Chemistry-A European Journal 17, 12257-12261 (2011).

172. B. H. Rotstein, D. J. Winternheimer, L. M. Yin, C. M. Deber, A. K. Yudin, Thioester- isocyanides: versatile reagents for the synthesis of cycle–tail peptides. Chemical Communications 48, 3775-3777 (2012).

219 173. Q.-J. Zhang, Y. Lindquist, V. Levitsky, M. G. Masucci, Solvent-exposed side chains of peptides bound to HLA A* 1101 have similar effects on the reactivity of alloantibodies and specific TCR. International immunology 8, 927-938 (1996).

174. X. Zhang et al., Quantitative PET imaging of tumor integrin αvβ3 expression with 18F- FRGD2. Journal of Nuclear Medicine 47, 113-121 (2006).

175. K. E. Ratcliffe, H. M. Fraser, R. Sellar, J. Rivier, R. P. Millar, Bifunctional gonadotropin- releasing hormone antagonist-progesterone analogs with increased efficacy and duration of action. Endocrinology 147, 571-579 (2006).

176. A. Isidro-Llobet, M. Alvarez, F. Albericio, Amino acid-protecting groups. Chemical reviews 109, 2455-2504 (2009).

177. K. M. Harris, S. Flemer, R. J. Hondal, Studies on deprotection of cysteine and selenocysteine side‐chain protecting groups. Journal of Peptide Science 13, 81-93 (2007).

178. F. Guzmán, S. Barberis, A. Illanes, Peptide synthesis: chemical or enzymatic. Electronic Journal of Biotechnology 10, 279-314 (2007).

179. Y. Ye, X. Chen, Integrin targeting for tumor optical imaging. Theranostics 1, 102-126 (2011).

180. E. Ruoslahti, RGD and other recognition sequences for integrins. Annual review of cell and developmental biology 12, 697-715 (1996).

181. X. Montet, K. Montet-Abou, F. Reynolds, R. Weissleder, L. Josephson, Nanoparticle imaging of integrins on tumor cells. Neoplasia (New York, N.Y.) 8, 214-222 (2006).

182. Y. Zhou, S. Chakraborty, S. Liu, Radiolabeled cyclic RGD peptides as radiotracers for imaging tumors and thrombosis by SPECT. Theranostics 1, 58 (2011).

183. M. A. Dechantsreiter et al., N-Methylated cyclic RGD peptides as highly active and selective αvβ3 integrin antagonists. Journal of Medicinal Chemistry 42, 3033-3040 (1999).

184. M. R. Gilbert et al., Cilengitide in patients with recurrent glioblastoma: the results of NABTC 03-02, a phase II trial with measures of treatment delivery. Journal of neuro- oncology 106, 147-153 (2012).

185. S. Hariharan et al., Assessment of the biological and pharmacological effects of the ανβ3 and ανβ5 integrin receptor antagonist, cilengitide (EMD 121974), in patients with advanced solid tumors. Annals of oncology 18, 1400-1407 (2007).

186. J.-P. Xiong et al., Crystal structure of the extracellular segment of integrin αVβ3 in complex with an Arg-Gly-Asp ligand. science 296, 151-155 (2002).

187. W. Wang, Q. Wu, M. Pasuelo, J. S. McMurray, C. Li, Probing for integrin αvβ3 binding of RGD peptides using fluorescence polarization. Bioconjugate chemistry 16, 729-734 (2005).

220 188. C. Rüegg, M. Hasmim, F. J. Lejeune, G. C. Alghisi, Antiangiogenic peptides and proteins: from experimental tools to clinical drugs. Biochimica et Biophysica Acta (BBA)- Reviews on Cancer 1765, 155-177 (2006).

189. A. Geyer, G. Mueller, H. Kessler, Conformational analysis of a cyclic RGD peptide containing a ψ[CH2-NH] bond: a positional shift in backbone structure caused by a single dipeptide mimetic. Journal of the American Chemical Society 116, 7735-7743 (1994).

190. M. Zhang et al., Pyropheophorbide 2-deoxyglucosamide: a new photosensitizer targeting glucose transporters. Bioconjugate chemistry 14, 709-714 (2003).

191. B. H. Rotstein, V. Rai, R. Hili, A. K. Yudin, Synthesis of peptide macrocycles using unprotected amino aldehydes. Nature protocols 5, 1813-1822 (2010).

192. F. Mohamadi et al., MacroModel—an integrated software system for modeling organic and bioorganic molecules using molecular mechanics. Journal of Computational Chemistry 11, 440-467 (1990).

193. W. L. Jorgensen, J. Tirado-Rives, Potential energy functions for atomic-level simulations of water and organic and biomolecular systems. Proceedings of the National Academy of Sciences of the United States of America 102, 6665-6670 (2005).

194. V. Tsui, D. A. Case, Theory and applications of the generalized Born solvation model in macromolecular simulations. Biopolymers 56, 275-291 (2000).

195. W. K. Hastings, Monte Carlo sampling methods using Markov chains and their applications. Biometrika 57, 97-109 (1970).

196. N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, E. Teller, Equation of state calculations by fast computing machines. The journal of chemical physics 21, 1087 (1953).

197. Z. Liu et al., Tumor uptake of the RGD dimeric probe 99mTc-G3-2P4-RGD2 is correlated with integrin αvβ3 expressed on both tumor cells and neovasculature. Bioconjugate chemistry 21, 548-555 (2010).

198. K. Stefflova, J. Chen, G. Zheng, Using molecular beacons for cancer imaging and treatment. Frontiers in bioscience: a journal and virtual library 12, 4709-4721 (2006).

199. K. Stefflova, J. Chen, G. Zheng, Killer beacons for combined cancer imaging and therapy. Current medicinal chemistry 14, 2110-2125 (2007).

200. J. F. Lovell, T. W. Liu, J. Chen, G. Zheng, Activatable photosensitizers for imaging and therapy. Chemical reviews 110, 2839-2857 (2010).

201. A. E. O‘Connor, W. M. Gallagher, A. T. Byrne, Porphyrin and Nonporphyrin Photosensitizers in Oncology: Preclinical and Clinical Advances in Photodynamic Therapy. Photochemistry and Photobiology 85, 1053-1074 (2009).

221 202. K. Plaetzer, B. Krammer, J. Berlanda, F. Berr, T. Kiesslich, Photophysics and photochemistry of photodynamic therapy: fundamental aspects. Lasers in medical science 24, 259-268 (2009).

203. D. Kelleher, O. Thews, A. Scherz, Y. Salomon, P. Vaupel, Combined hyperthermia and chlorophyll-based photodynamic therapy: tumour growth and metabolic microenvironment. British journal of cancer 89, 2333-2339 (2003).

204. J. Zhao, W. Wu, J. Sun, S. Guo, Triplet photosensitizers: from molecular design to applications. Chemical Society Reviews, (2013).

205. R. A. Weersink et al., Assessment of Cutaneous Photosensitivity of TOOKAD (WST09) in Preclinical Animal Models and in Patients¶. Photochemistry and Photobiology 81, 106-113 (2005).

206. M. Berdugo et al., Evaluation of the new photosensitizer Stakel (WST-11) for photodynamic choroidal vessel occlusion in rabbit and rat eyes. Investigative ophthalmology & visual science 49, 1633-1644 (2008).

207. S. Chevalier et al., Preclinical study of the novel vascular occluding agent, WST11, for photodynamic therapy of the canine prostate. The Journal of urology 186, 302-309 (2011).

208. L. Goldshaid et al., Novel design principles enable specific targeting of imaging and therapeutic agents to necrotic domains in breast tumors. Breast Cancer Res 12, R29 (2010).

209. O. Mazor et al., WST11, A Novel Water‐soluble Bacteriochlorophyll Derivative; Cellular Uptake, Pharmacokinetics, Biodistribution and Vascular‐targeted Photodynamic Activity Using Melanoma Tumors as a Model¶. Photochemistry and Photobiology 81, 342-351 (2005).

210. J. Llano, J. Raber, L. A. Eriksson, Theoretical study of phototoxic reactions of psoralens. Journal of Photochemistry and Photobiology A: Chemistry 154, 235-243 (2003).

211. R. C. Guedes, L. A. Eriksson, Photophysics, photochemistry, and reactivity: Molecular aspects of perylenequinone reactions. Photochemical & Photobiological Sciences 6, 1089-1096 (2007).

212. L. Shen, H.-F. Ji, H.-Y. Zhang, A TD-DFT study on triplet excited-state properties of curcumin and its implications in elucidating the photosensitizing mechanisms of the pigment. Chemical Physics Letters 409, 300-303 (2005).

213. B. Tian, E. S. Eriksson, L. A. Eriksson, Can range-separated and hybrid DFT functionals predict low-lying excitations? A Tookad case study. Journal of Chemical Theory and Computation 6, 2086-2094 (2010).

214. R. Vargas, M. Galván, A. Vela, Singlet-triplet gaps and spin potentials. The Journal of Physical Chemistry A 102, 3134-3140 (1998).

222 215. M. E. Alberto, C. Iuga, A. D. Quartarolo, N. Russo, Bisanthracene Bis (dicarboxylic imide) s as Potential Photosensitizers in Photodynamic Therapy: A Theoretical Investigation. Journal of chemical information and modeling 53, 2334-2340 (2013).

216. E. S. Eriksson, L. A. Eriksson, Predictive power of long-range corrected functionals on the spectroscopic properties of tetrapyrrole derivatives for photodynamic therapy. Physical Chemistry Chemical Physics 13, 7207-7217 (2011).

217. G. Mazzone, N. Russo, E. Sicilia, Theoretical Investigation Of The Absorption Spectra And Singlet-Triplet Energy Gap Of Positively-Charged Tetraphenylporphyrins As Potential Photodynamic Therapy Photosensitizers. Canadian Journal of Chemistry.

218. I. Corral, P. Pérez, O. Tapia, M. Yáñez, Oxygenation of the phenylhalocarbenes. Are they spin-allowed or spin-forbidden reactions? Journal of molecular modeling 18, 2813-2821 (2012).

219. A. N. Kozyrev et al., Syntheses of stable bacteriochlorophyll-a derivatives as potential photosensitizers for photodynamic therapy. Tetrahedron letters 37, 6431-6434 (1996).

220. R. K. Pandey et al., Synthesis, photophysical properties, in vivo photosensitizing efficacy, and human serum albumin binding properties of some novel bacteriochlorins. Journal of Medicinal Chemistry 40, 2770-2779 (1997).

221. Á. Roxin, T. D. MacDonald, G. Zheng, Synthesis and characterization of a new natural product analog, 132-173-bacteriochlorophyllone a. Journal of Porphyrins and Phthalocyanines, 1-12 (2013).

222. K. Teuchner et al., Fluorescence and excited state absorption in modified pigments of bacterial photosynthesis a comparative study of metal-substituted bacteriochlorophylls a. Journal of luminescence 72, 612-614 (1997).

223. C. Gettinger, A. Heeger, J. Drake, D. Pine, The effect of intrinsic rigidity on the optical properties of PPV derivatives. Molecular Crystals and Liquid Crystals 256, 507-512 (1994).

224. P. Abbyad, W. Childs, X. Shi, S. G. Boxer, Dynamic Stokes shift in green fluorescent protein variants. Proceedings of the National Academy of Sciences 104, 20189-20194 (2007).

225. M. Henary, M. Mojzych, M. Say, L. Strekowski, Functionalization of Benzo (c, d indole System for the Synthesis of Visible and Near-Infrared Dyes. ChemInform 40, i (2009).

226. R. A. Chica, M. M. Moore, B. D. Allen, S. L. Mayo, Generation of longer emission wavelength red fluorescent proteins using computationally designed libraries. Proceedings of the National Academy of Sciences 107, 20257-20262 (2010).

227. J. O. Escobedo, O. Rusin, S. Lim, R. M. Strongin, NIR dyes for bioimaging applications. Current opinion in chemical biology 14, 64-70 (2010).

228. K. D. Piatkevich et al., Extended Stokes Shift in Fluorescent Proteins: Chromophore– Protein Interactions in a Near-Infrared TagRFP675 Variant. Scientific reports 3, (2013).

223 229. C. A. Cohn, S. R. Simon, M. A. Schoonen, Comparison of fluorescence-based techniques for the quantification of particle-induced hydroxyl radicals. Particle and fibre toxicology 5, 2 (2008).

230. P. R. Castello, D. A. Drechsel, B. J. Day, M. Patel, Inhibition of mitochondrial hydrogen peroxide production by lipophilic metalloporphyrins. Journal of Pharmacology and Experimental Therapeutics 324, 970-976 (2008).

231. I. Šnyrychová, F. Ayaydin, E. Hideg, Detecting hydrogen peroxide in leaves in vivo–a comparison of methods. Physiologia plantarum 135, 1-18 (2009).

232. L. Fruk, C. M. Niemeyer, Covalent hemin–DNA adducts for generating a novel class of artificial heme enzymes. Angewandte Chemie International Edition 44, 2603-2606 (2005).

233. C. C. Widmer et al., Hemoglobin Can Attenuate Hydrogen Peroxide–Induced Oxidative Stress by Acting as an Antioxidative Peroxidase. Antioxidants & redox signaling 12, 185- 198 (2010).

234. R. B. Fox, R. N. Harada, R. M. Tate, J. E. Repine, Prevention of thiourea-induced pulmonary edema by hydroxyl-radical scavengers. Journal of Applied Physiology 55, 1456-1459 (1983).

235. R. B. Fox, Prevention of granulocyte-mediated oxidant lung injury in rats by a hydroxyl radical scavenger, dimethylthiourea. Journal of Clinical Investigation 74, 1456 (1984).

236. A. Levine, R. Tenhaken, R. Dixon, C. Lamb, H2O2 from the oxidative burst orchestrates the plant hypersensitive disease resistance response. Cell 79, 583-593 (1994).

237. C. Greenstock, R. Miller, The oxidation of tiron by superoxide anion. Kinetics of the reaction in aqueous solution and in chloroplasts. Biochimica et Biophysica Acta (BBA)- Bioenergetics 396, 11-16 (1975).

238. R. Miller, F. MacDowall, The tiron free radical as a sensitive indicator of chloroplastic photoautoxidation. Biochimica et Biophysica Acta (BBA)-Bioenergetics 387, 176-187 (1975).

239. É. Hideg, C. Spetea, I. Vass, Superoxide radicals are not the main promoters of acceptor- side-induced photoinhibitory damage in spinach thylakoids. Photosynthesis Research 46, 399-407 (1995).

240. M. Price, J. J. Reiners, A. M. Santiago, D. Kessel, Monitoring singlet oxygen and hydroxyl radical formation with fluorescent probes during photodynamic therapy. Photochemistry and Photobiology 85, 1177-1181 (2009).

241. M. Bancirova, Sodium azide as a specific quencher of singlet oxygen during chemiluminescent detection by luminol and Cypridina luciferin analogues. Luminescence 26, 685-688 (2011).

224 242. J. M. Dąbrowski et al., New Halogenated Water‐Soluble Chlorin and Bacteriochlorin as Photostable PDT Sensitizers: Synthesis, Spectroscopy, Photophysics, and in vitro Photosensitizing Efficacy. ChemMedChem 5, 1770-1780 (2010).

243. L. Bezdetnaya et al., Spectroscopic and Biological Testing of Photobleaching of Porphyrins in Solutions*. Photochemistry and Photobiology 64, 382-386 (1996).

244. A. Blum, L. I. Grossweiner, Singlet oxygen generation by hematoporphyrin IX, uroporphyrin I and hematoporphyrin derivative at 546 nm in phosphate buffer and in the presence of egg phosphatidylcholine liposomes. Photochemistry and Photobiology 41, 27-32 (1985).

245. X. Ragàs, L. P. Cooper, J. H. White, S. Nonell, C. Flors, Quantification of photosensitized singlet oxygen production by a fluorescent protein. ChemPhysChem 12, 161-165 (2011).

246. J. S. Dysart, G. Singh, M. S. Patterson, Calculation of Singlet Oxygen Dose from Photosensitizer Fluorescence and Photobleaching During mTHPC Photodynamic Therapy of MLL Cells¶. Photochemistry and Photobiology 81, 196-205 (2005).

247. L. J. Martinez, R. H. Sik, C. F. Chignell, Fluoroquinolone antimicrobials: singlet oxygen, superoxide and phototoxicity. Photochemistry and Photobiology 67, 399-403 (1998).

248. J. S. Dysart, M. S. Patterson, Characterization of Photofrin photobleaching for singlet oxygen dose estimation during photodynamic therapy of MLL cells in vitro. Physics in medicine and biology 50, 2597 (2005).

249. A. Gollmer et al., Singlet Oxygen Sensor Green®: Photochemical behavior in solution and in a mammalian cell. Photochemistry and Photobiology 87, 671-679 (2011).

250. C. Flors et al., Imaging the production of singlet oxygen in vivo using a new fluorescent sensor, Singlet Oxygen Sensor Green®. Journal of experimental botany 57, 1725-1734 (2006).

251. H. Lin et al., Feasibility study on quantitative measurements of singlet oxygen generation using singlet oxygen sensor green. Journal of fluorescence 23, 41-47 (2013).

252. Y. Shen et al., Indirect imaging of singlet oxygen generation from a single cell. Laser Physics Letters 8, 232 (2011).

253. K. Nakamura et al., Reevaluation of analytical methods for photogenerated singlet oxygen. Journal of clinical biochemistry and nutrition 49, 87 (2011).

254. C. J. Monteiro, J. Pina, M. M. Pereira, L. G. Arnaut, On the singlet states of porphyrins, chlorins and bacteriochlorins and their ability to harvest red/infrared light. Photochemical & Photobiological Sciences 11, 1233-1238 (2012).

255. H. Ding et al., Photoactivation switch from type II to type I reactions by electron-rich micelles for improved photodynamic therapy of cancer cells under hypoxia. Journal of Controlled Release 156, 276-280 (2011).

225 256. P. A. Firey, W. E. Ford, J. R. Sounik, M. E. Kenney, M. A. Rodgers, Silicon naphthalocyanine triplet state and oxygen. A reversible energy-transfer reaction. Journal of the American Chemical Society 110, 7626-7630 (1988).

257. I. Georgakoudi, T. H. Foster, Singlet Oxygen‐Versus Nonsinglet Oxygen‐Mediated Mechanisms of Sensitizer Photobleaching and Their Effects on Photodynamic Dosimetry. Photochemistry and Photobiology 67, 612-625 (1998).

258. R. Bonnett, B. D. Djelal, P. A. Hamilton, G. Martinez, F. Wierrani, Photobleaching of 5,10,15,20-tetrakis (m-hydroxyphenyl) porphyrin (m-THPP) and the corresponding chlorin (m-THPC) and bacteriochlorin (m-THPBC). A comparative study. Journal of Photochemistry and Photobiology B: Biology 53, 136-143 (1999).

259. J. S. Dysart, M. S. Patterson, T. J. Farrell, G. Singh, Relationship Between mTHPC Fluorescence Photobleaching and Cell Viability During In Vitro Photodynamic Treatment of DP16 Cells¶. Photochemistry and Photobiology 75, 289-295 (2002).

260. M. T. Jarvi, M. S. Patterson, B. C. Wilson, Insights into photodynamic therapy dosimetry: simultaneous singlet oxygen luminescence and photosensitizer photobleaching measurements. Biophysical journal 102, 661-671 (2012).

261. H. Falk, G. Hoornaert, H. Isenring, A. Eschenmoser, Enol derivatives in chlorophyll series-preparation of 132,173-cyclopheophorbide enols. Helvetica Chimica Acta 58, 2347-2357 (1975).

262. H. P. Isenring et al., Über enolisierte Derivate der Chlorophyllreihe. 132‐Desmethoxycarbonyl‐173‐desoxy‐132,173‐cyclochlorophyllid a‐enol und eine Methode zur Einführung von Magnesium in porphinoide Ligandsysteme unter milden Bedingungen.(Vorläufige Mitteilung). Helvetica Chimica Acta 58, 2357-2367 (1975).

263. C. J. Fookes, Identification of a homologous series of nickel (II) 15, 17-butanoporphyrins from an oil shale. Journal of the Chemical Society, Chemical Communications, 1474- 1476 (1983).

264. G. A. Wolff, M. Murray, J. R. Maxwell, B. K. Hunter, J. K. Sanders, 15, 17-Butano-3, 8- diethyl-2, 7, 12, 18-tetramethylporphyrin-a novel naturally occurring tetrapyrrole. Journal of the Chemical Society, Chemical Communications, 922-924 (1983).

265. P. Sundrraraman et al., Temporal changes in the distribution of C33cycloheptenoDPEP and 17-nor C30 DPEP in rocks. Organic geochemistry 21, 1051-1058 (1994).

266. R. Ocampo, J. P. Sachs, D. J. Repeta, Isolation and structure determination of the unstable 132, 173-Cyclopheophorbide a enol from recent sediments. Geochimica et cosmochimica Acta 63, 3743-3749 (1999).

267. B. Keely et al., Porphyrin and chlorin distributions in a Late Pliocene lacustrine sediment. Geochimica et cosmochimica Acta 58, 3691-3701 (1994).

268. W. Prowse, M. Chicarelli, B. Keely, S. Kaur, J. Maxwell, Characterisation of fossil porphyrins of the ―di-DPEP‖ type. Geochimica et cosmochimica Acta 51, 2875-2877 (1987).

226 269. P. Schaeffer, R. Ocampo, H. J. Callot, P. Albrecht, Structure determination by deuterium labelling of a sulfur-bound petroporphyrin. Geochimica et cosmochimica Acta 58, 4247- 4252 (1994).

270. P. G. Harris, G. E. Pearce, T. M. Peakman, J. R. Maxwell, A widespread and abundant chlorophyll transformation product in aquatic environments. Organic geochemistry 23, 183-187 (1995).

271. M. I. Chicarelli, J. R. Maxwell, A novel fossil porphyrin with a fused ring system: Evidence for water column transformation of chlorophyll? Tetrahedron letters 27, 4653- 4654 (1986).

272. P. Karuso et al., 132, 173-Cyclopheophorbide enol, the first porphyrin isolated from a sponge. Tetrahedron letters 27, 2177-2178 (1986).

273. N. Aydin, S. Daher, F. O. Gülaçar, On the sedimentary occurrence of chlorophyllone a. Chemosphere 52, 937-942 (2003).

274. R. Goericke, S. L. Strom, M. A. Bell, Distribution and sources of cyclic pheophorbides in the marine environment. Limnol. Oceanogr 45, 200-211 (2000).

275. L. William, J, J. W. Loitz, D. T. Rudnick, E. W. Baker, Early diagenetic alteration of chlorophyll-a and bacteriochlorophyll-a in a contemporaneous marl ecosystem; Florida Bay. Organic geochemistry 31, 1561-1580 (2000).

276. N. Watanabe et al., New chlorophyll-a-related compounds isolated as antioxidants from marine bivalves. Journal of natural products 56, 305-317 (1993).

277. K. Sakata et al., Chlorophyllone-a, a new pheophorbide-a related compound isolated from Ruditapes philippinarum as an antioxidative compound. Tetrahedron letters 31, 1165-1168 (1990).

278. Y. Kashiyama et al., Ubiquity and quantitative significance of detoxification catabolism of chlorophyll associated with protistan herbivory. Proceedings of the National Academy of Sciences 109, 17328-17335 (2012).

279. P. G. Harris et al., Identification of chlorophyll transformation products in zooplankton faecal pellets and marine sediment extracts by liquid chromatography/mass spectrometry atmospheric pressure chemical ionization. Rapid Communications in Mass Spectrometry 9, 1177-1183 (1995).

280. X. F. D. Chillier, F. O. Gülaçar, A. Buchs, A novel sedimentary lacustrine chlorin: characterisation and geochemical significance. Chemosphere 27, 2103-2110 (1993).

281. X. F. D. Chillier, G. J. Van Berkel, F. O. Gülaçar, A. Buchs, Characterization of chlorins within a natural chlorin mixture using electrospray/ion trap mass spectrometry. Organic mass spectrometry 29, 672-678 (1994).

282. V. Cariou-Le Gall, A. Rosell-Mele, J. Maxwell, in Proceedings of the Ocean Drilling Program, Scientific Results. (1998), vol. 160, pp. 297-302.

227 283. G. Kowalewska, B. Winterhalter, H. Talbot, J. Maxwell, J. Konat, Chlorins in sediments of the Gotland Deep (Baltic Sea). Oceanologia 41, 81-97 (1999).

284. J. Villanueva, D. W. Hastings, A century-scale record of the preservation of chlorophyll and its transformation products in anoxic sediments. Geochimica et cosmochimica Acta 64, 2281-2294 (2000).

285. R. L. Airs, J. E. Atkinson, B. J. Keely, Development and application of a high resolution liquid chromatographic method for the analysis of complex pigment distributions. Journal of Chromatography A 917, 167-177 (2001).

286. L. Ma, D. Dolphin, Asymmetric hydroxylation of chlorophyll derivatives: A facile entry to both diastereomers of chlorophyllone a. Tetrahedron: Asymmetry 6, 313-316 (1995).

287. L. Ma, D. Dolphin, Stereoselective synthesis of new chlorophyll a related antioxidants isolated from marine organisms. The Journal of organic chemistry 61, 2501-2510 (1996).

288. L. Ma, D. Dolphin, Synthesis and spectral studies of natural chlorins having antioxidative activity. Pure and applied chemistry 68, 765-769 (1996).

289. L. Ma, D. Dolphin, The metabolites of dietary chlorophylls. Phytochemistry 50, 195-202 (1999).

290. J. F. Lovell et al., Porphysome nanovesicles generated by porphyrin bilayers for use as multimodal biophotonic contrast agents. Nature materials 10, 324-332 (2011).

291. W. Cao et al., Synthesis and evaluation of a stable bacteriochlorophyll-analog and its incorporation into high-density lipoprotein nanoparticles for tumor imaging. Bioconjugate chemistry 20, 2023-2031 (2009).

292. D. E. Marotta et al., Evaluation of bacteriochlorophyll-reconstituted low-density lipoprotein nanoparticles for photodynamic therapy efficacy in vivo. Nanomedicine 6, 475-487 (2011).

293. T. W. Liu et al., Multimodal bacteriochlorophyll theranostic agent. Theranostics 1, 354 (2011).

294. G. Zheng, Y. Chen, X. Intes, B. Chance, J. D. Glickson, Contrast-enhanced near-infrared (NIR) optical imaging for subsurface cancer detection. Journal of Porphyrins and Phthalocyanines 8, 1106-1117 (2004).

295. D. S. Gottfried, S. G. Boxer, Studies of crystalline aggregates of pheophorbide and bacteriopheophorbide by electro-absorption (Stark effect) spectroscopy and powder X- ray diffraction. Journal of luminescence 51, 39-50 (1992).

296. G. Zheng et al., Low-Density Lipoprotein Reconstituted by Pyropheophorbide Cholesteryl Oleate as Target-Specific Photosensitizer. Bioconjugate chemistry 13, 392- 396 (2002).

228 297. S. P. Raillard, A. D. Mann, T. A. Baer, An efficient procedure for the preparation of Fmoc-amino acids. Organic preparations and procedures international 30, 183-186 (1998).

298. J. A. Burns, J. C. Butler, J. Moran, G. M. Whitesides, Selective reduction of disulfides by tris (2-carboxyethyl) phosphine. The Journal of organic chemistry 56, 2648-2650 (1991).

299. L. Bello et al., αvβ3 and αvβ5 integrin expression in glioma periphery. Neurosurgery 49, 380-390 (2001).

300. S. Hazelbag et al., Overexpression of the αvβ6 integrin in cervical squamous cell carcinoma is a prognostic factor for decreased survival. The Journal of pathology 212, 316-324 (2007).

301. R. C. Bates et al., Transcriptional activation of integrin β6 during the epithelial- mesenchymal transition defines a novel prognostic indicator of aggressive colon carcinoma. Journal of Clinical Investigation 115, 339-347 (2005).

302. E. Danen et al., Emergence of α5β1 fibronectin‐and αvβ3 vitronectin‐receptor expression in melanocytic tumour progression. Histopathology 24, 249-256 (1994).

303. M. Adachi et al., Significance of integrin α5 gene expression as a prognostic factor in node-negative non-small cell lung cancer. Clinical cancer research 6, 96-101 (2000).

304. D. Pramanik et al., Lipopeptide with a RGDK tetrapeptide sequence can selectively target genes to proangiogenic α5β1 integrin receptor and mouse tumor vasculature. Journal of medicinal chemistry 51, 7298-7302 (2008).

305. R. Blue, M. Murcia, C. Karan, M. Jiroušková, B. S. Coller, Application of high- throughput screening to identify a novel αIIb-specific small-molecule inhibitor of αIIbβ3- mediated platelet interaction with fibrinogen. Blood 111, 1248-1256 (2008).

306. K. Berg, A. Western, J. Bommer, J. Moan, Intracellular localization of sulfonated meso- tetraphenylporphines in a human carcinoma cell line. Photochemistry and photobiology 52, 481-487 (1990).

307. S. Bonneau, P. Morlière, D. Brault, Dynamics of interactions of photosensitizers with lipoproteins and membrane-models: correlation with cellular incorporation and subcellular distribution. Biochemical pharmacology 68, 1443-1452 (2004).

308. H. Mojzisova, S. Bonneau, D. Brault, Structural and physico-chemical determinants of the interactions of macrocyclic photosensitizers with cells. European Biophysics Journal 36, 943-953 (2007).

309. K. Woodburn, N. Vardaxis, J. Hill, A. Kaye, D. Phillips, Subcellular localization of porphyrins using confocal laser scanning microscopy. Photochemistry and photobiology 54, 725-732 (1991).

310. P.-C. Lo, C. M. Chan, J.-Y. Liu, W.-P. Fong, D. K. Ng, Highly photocytotoxic glucosylated silicon (IV) phthalocyanines. Effects of peripheral chloro substitution on the

229 photophysical and photodynamic properties. Journal of medicinal chemistry 50, 2100- 2107 (2007).

311. X. Damoiseau, F. Tfibel, M. Hoebeke, M. Fontaine-Aupart, Effect of aggregation on bacteriochlorin a triplet-state formation: a laser flash photolysis study. Photochemistry and photobiology 76, 480 (2002).

312. H. Dummin, T. Cernay, H. Zimmermann, Selective photosensitization of mitochondria in HeLa cells by cationic Zn (II) phthalocyanines with lipophilic side-chains. Journal of Photochemistry and Photobiology B: Biology 37, 219-229 (1997).

313. G. Oenbrink, P. Jurgenlimke, D. Gabel, Accumulation of porphyrins in cells: influence of hydrophobicity aggregation and protein binding. Photochemistry and photobiology 48, 451-456 (1988).

314. P. R. Twentyman et al., A new mouse tumor model system (RIF-1) for comparison of end-point studies. Journal of the National Cancer Institute 64, 595-604 (1980).

315. M. Kiani, B. Fenton, Inherent cellular differences may explain the dissimilar survival of RIF-1 and KHT tumour cells under aerobic and hypoxic conditions. International journal of radiation biology 67, 449-452 (1995).

316. J. M. Brown, P. R. Twentyman, S. S. Zamvil, Response of the RIF-1 tumor in vitro and in C3H/Km mice to X-radiation (cell survival, regrowth delay, and tumor control), chemotherapeutic agents, and activated macrophages. Journal of the National Cancer Institute 64, 605-611 (1980).

317. Q. Ke, M. Costa, Hypoxia-inducible factor-1 (HIF-1). Molecular pharmacology 70, 1469-1480 (2006).

318. K. W. Woodburn, C. J. Engelman, M. S. Blumenkranz, Photodynamic therapy for choroidal neovascularization: a review. Retina 22, 391-405 (2002).

319. L. Sancey et al., Clustering and internalization of integrin alphavbeta3 with a tetrameric RGD-synthetic peptide. Molecular therapy : the journal of the American Society of Gene Therapy 17, 837-843 (2009).

320. K. Temming, R. M. Schiffelers, G. Molema, R. J. Kok, RGD-based strategies for selective delivery of therapeutics and imaging agents to the tumour vasculature. Drug resistance updates 8, 381-402 (2005).

321. H. Lepor, Vascular targeted photodynamic therapy for localized prostate cancer. Reviews in urology 10, 254 (2008).

Appendices Cancer-targeting aziridine aldehyde-cyclized RGD-based peptides

Figure A1.1. Investigating modified aziridine aldehydes to assess their compatibility with the sulfhydryl-aziridine ring opening-based conjugation strategy described in chapter 2 (169).

230

231

232

Figure A1.3. Preliminary exploration of the binding specificity of cPRGDA, cPRGDAA and the negative control cPRDGA into buffered solutions of purified cancer-related integrin variants A)

αVβ5, B) αVβ6 and C) α5β1, using fluorescence polarization spectroscopy (N = 1).

233

Figure A1.4. HPLC spectra showing the successful labeling of the amide-cyclized cPRGDK(DOTA) macrocycle using A) ‗cold‘ 63Cu(II) and B) radioactive 64Cu(II).

234 Promising new bacteriochlorin photosensitizers

Table A2.1. The charge and water-solubility of selected bacteriochlorin photosensitizers from Chapter 3.

Photosensitizers Net charge Log P 1a -1 1.72 a 1b -2 0.09 a 2a +1 -0.66 a 2b 0 c -2.64 a 4b -2 0.1 b 5b -2 -1.87 a a, Calculated experimentally used 1-octanol: water partitioning. b, Calculated theoretically using ACD modeling software. c. Zwitterionic compound that contains both an anionic and a cationic moiety.

235

236 Towards targeted photodynamic therapy with RGD peptides and bacteriochlorin photosensitizers

Figure A3.1. Competitive cell adhesion assay results showing the concentration-dependent displacement of integrin-expressing U87 cells from vitronectin-coated wells using increasing concentrations of Ac-PRGDK(Fluorescein) after 30 minute co-incubations at 37 °C (A). Confocal microscopy images showing the specificity of Ac-PRGDK(Fluorescein) (10 μM, 2 h incubations at 37 °C) to integrin-expressing U87 cells (B and C) compared to the negative control, HT29 cells (D and E).

237

238

Scheme A3.1. Synthesis of the integrin-targeting peptide-photosensitizer conjugate, Ac- PRGDK(4b).

239

Scheme A3.2. Synthesis of the integrin-targeting peptide-photosensitizer conjugate, Ac- PRGDK(1a).

240

Á. Roxin, and G. Zheng. Flexible or fixed: a comparative review of linear and cyclic cancer- targeting peptides. Future Medicinal Chemistry 4(12), 1601-1618 (2012). Copyright © 2012, Future Sciences Ltd.

N. Assa-Munt, X. Jia, P. Laakkonen, E. Ruoslahti. Solution structures and integrin binding activities of an RGD peptide with two isomers. Biochemistry 40, 2373-2378 (2001). Copyright © 2001, American Chemical Society.

A. H. Negussie, J. L. Miller, G. Reddy, S. K. Drake, B. J. Wood, M. R. Dreher. Synthesis and in vitro evaluation of cyclic NGR peptide targeted thermally sensitive liposome. Journal of Controlled Release 143, 265-273 (2010). Copyright © 2010, Elsevier Publishers.

M. F. Giblin, N. Wang, T. J. Hoffman, S. S. Jurisson, T. P. Quinn. Design and characterization of α-melanotropin peptide analogs cyclized through rhenium and technetium metal coordination. Proceedings of the National Academy of Sciences 95, 12814-12818 (1998). Copyright © 1998, Proceedings of the National Academy of Sciences.

Á. Roxin, J. Chen, C. C. G. Scully, B. H. Rotstein, A. K. Yudin, and G. Zheng. Conformational modulation of in vitro activity of cyclic RGD peptides via aziridine aldehyde-driven macrocyclization chemistry. Bioconjugate Chemistry 23(7), 1387-1295 (2012). Copyright © 2012, American Chemistry Society.

Á. Roxin, J. Chen, A. S. Paton, T. P. Bender, and G. Zheng. Modulation of reactive oxygen species photogeneration of bacteriopheophorbide a derivatives by exocyclic E-ring opening and charge modification. Journal of Medicinal Chemistry 57(1), 223-237 (2014). Copyright © 2014, American Chemistry Society.

Á. Roxin, T. D. MacDonald, and G. Zheng. Synthesis and characterization of a new natural product analogue, 132-173-bacteriochlorophyllone a. Journal of Porphyrins and Phthalocyanines, (2013). DOI: 10.1142/S1088424613501058. Copyright © 2013, World Scientific Publishing Co.

241