In Vitroo and in Vivo Pharmacology of 4-Substituted

Home , Glucuronic acid, Oxazole, Thiazole

IN VITROO AND IN VIVO PHARMACOLOGY OF 4-SUBSTITUTED

METHOXYBENZOYL-ARYL-THIAZOLES (SMART) AND 2-

ARYLTHIAZOLIDINE-4-CARBOXYLIC ACID AMIDES (ATCAA)

Dissertation

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

in the Graduate School of The Ohio State University

Chien-Ming Li, M.S.

Graduate Program in Pharmacy

The Ohio State University

2010

Dissertation Committee:

Dr. James T. Dalton, Advisor

Dr. Robert W. Brueggemeier

Dr. Thomas D. Schmittgen

Dr. Mitch A. Phelps Copyright by

Chien-Ming Li

2010

ABSTRACT

Formation of microtubules is a dynamic process that involves polymerization and depolymerization of the αβ-tubulin heterodimers. Drugs that enhance or inhibit tubulin polymerization can destroy this dynamic process, arresting cells in the G2/M phase of the cell cycle. Although drugs that target tubulin generally demonstrate cytotoxic potency in sub-nanomolar range, resistance due to drug efflux is a common phenomenon among the antitubulin agents. We recently reported a class of 4-Substituted Methoxybenzoyl-Aryl-

Thiazoles (SMART), which exhibited great in vitro and in vivo potency.

SMART compounds effectively inhibited tubulin polymerization in dose dependent manner, suggesting that SMART compounds may bind to tubulin and subsequently hamper the polymerization. To date the only method to determine the binding of inhibitor to tubulin has been competitive radioligand binding assays. We developed a novel non-radioactive mass spectrometry (MS) binding assay to study the tubulin binding of colchicine, vinblastine and paclitaxel. The method involves a very simple step of separating the unbound ligand from macromolecules using ultrafiltration.

The unbound ligand in the filtrate can be accurately determined using a highly sensitive and specific LC-MS/MS method. The method does not require the use of radiolabeled ligands, can be applied to a wide variety of drugs that either inhibit or promote tubulin polymerization, and allows for studies to define the reversibility of the interaction with ii tubulin. This method was subsequently applied to determine the tubulin binding site of 4-

(3,4,5-trimethoxylbenzoyl)-2-phenyl-thiazole (SMART-H). The results indicated that

SMART-H specifically and reversibly bound only to the colchicine-binding site, but not to vinblastine- or paclitaxel-sites.

In metabolic stability studies, SMART-H was used to examine in the liver microsomes of four species (mouse, rat, dog, and human) revealed half-lives between < 5 and 30 min, demonstrating the inter-species variability in the metabolism of SMART-H.

The clearance predictions based on in vitro data correspond well with in vivo clearance obtained from mouse, rat, and dog in vivo pharmacokinetic studies. Four major metabolic processes were found in human liver microsomes. We designed and tested four derivatives by blocking labile sites to improve metabolic stability. The oxime and hydrazide derivatives, replacing the ketone site, demonstrated 2-3-fold improved half-life in human liver microsomes, indicating that metabolic stability of SMART-H can be extended by blocking ketone reduction. These studies led us to the next generation of

SMART compounds with greater metabolic stability.

Evaluation of the in vivo anti-cancer activities of three SMART compounds,

SMART-H (H), SMART-F (F) and SMART-OH (OH) with varying substituents at the 4- position of aryl ring, were used to perform in vivo anti-tumor efficacy in human prostate

(PC-3) and melanoma (A375) cancer xenograft models. These data indicated that

SMART-H and SMART-F showed greater than 70% tumor growth inhibition (TGI), while SMART-OH failed to inhibit tumor growth. In addition, higher dose of SMART-H

(15 mg/kg) exhibited 96% PC-3 xenograft TGI after 21-days of treatment without any apparent neurotoxicity. iii In conclusion, these studies provide the first in vivo evidence and proof-of-

concept that SMART compounds represent a potent and broad spectrum class of antitubulin agents for the treatment of cancer.

iv Dedicated to my parents

and

my wife

v ACKNOWLEDGMENTS

I would like to express my deeply appreciation to my advisor, Dr. James T.

Dalton, for his guidance, advice and also appreciate the opportunity to be at GTx Inc. for continuing my training in anticancer drug discovery and development.

I thank my committee members, Dr. Robert W. Brueggemeier, Dr. Thomas D.

Schmittgen, and Dr. Mitch A. Phelps, for their valuable advice. I truly thank Dr. Duane

D. Miller, Dr. Wei Li, Yan Lu and Jianjin Chen for helping design all the compounds and

helpful discussions.

I also thank Dr Eunju Hurh to help me continue ATCAA project. I thank Amanda

Jones and Sunjoo Ahn to be my classmates for five years.

I would like to express my thanks to my lovely wife, Huei-Ling Chang, for her

encouragement and support.

I would like to thank my parents for their support.

vi VITA

March 13, 1976…………………………Born-Tainan, Taiwan 1998……………………………………..B.S. Chemistry National Cheng Kung University, Taiwan 1998-2000………………………………M.S. Analytical Chemistry National Cheng Kung University, Taiwan Advisor: Shu-Hui Chen Thesis: Quantitative detection of N7-(2- hydroxyethyl)guanine adducts in DNA using LC-MS/MS 2001-2005………………………………Research Associate National Health Research Institutes, Taiwan 2005-2008………………………………Graduate Teaching Associate The Ohio State University 2008-present…………………………….Research Associate II GTx Inc.

PUBLICATIONS

1. Liao PC, Li CM, Hung CW, Chen SH. Quantitative detection of N(7)-(2- hydroxyethyl)guanine adducts in DNA using high-performance liquid chromatography/electrospray ionization tandem mass spectrometry. Journal of Mass Spectrometry, 36(3): 336-343, 2001.

vii 2. Liao PC, Li CM, Lin LC, Hung CW, Shih TS. An online automatic sample cleanup system for the quantitative detection of the benzene exposure biomarker S- phenylmercapturic acid in human urine by electrospray ionization tandem mass spectrometry. Journal of Analytical Toxicology, 26(4): 205-210, 2002. 3. Chang YC, Li CM, Li LA, Jong SB, Liao PC, Chang LW. Quantitative measurement of male steroid hormones using automated on-line solid phase extraction-liquid chromatography-tandem mass spectrometry and comparison with radioimmunoassay. Analyst, 128(4): 363-368, 2003. 4. Hsieh PW., Chang FR., Wu CC., Wu KY., Li CM., Wang WY., Gu LC, Wu YC. Selective Inhibition of collagen-induced platelet aggregation by a cyclic peptide from drymaria diandra. Helvetica Chimica Acta, 87(1): 57-66, 2004. 5. Hsieh PW, Chang FR, Wu CC, Wu KY, Li CM, Chen SL, Wu YC. New cytotoxic cyclic peptides and dianthramide from Dianthus superbus. Journal of Natural Products, 67(9): 1522-1527, 2004 6. Hsieh PW, Chang FR, Wu CC, Li CM, Wu KY, Chen SL, Yen HF, Wu YC. Longicalycinin A, a new cytotoxic cyclic peptide from Dianthus superbus var. longicalycinus (MAXIM.) WILL. Chemical & Pharmaceutical Bulletin (Tokyo), 53(3): 336-338, 2005. 7. Li CM, Hu CW, Wu KY. Quantification of urinary N-acetyl-S- (propionamide)cysteine using an on-line clean-up system coupled with liquid chromatography/tandem mass spectrometry. Journal of Mass Spectrometry, 40(4): 511-515, 2005. 8. Li CM, Chu RY, Hsientang Hsieh DP. An enhanced LC-MS/MS method for microcystin-LR in lake water. Journal of Mass Spectrometry, 41(2): 169-174, 2006. 9. Wang SL, Chang YC, Chao HR, Li CM, Li LA, Lin LY, Papke O. Body burdens of polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyls and their relations to estrogen metabolism in pregnant women. Environmental Health Perspectives, 114(5): 740-745, 2006.

viii 10. Huang CC, Li CM, Wu CF, Jao SP, Wu KY. Analysis of Urinary N-acetyl-S- (propionamide)-cysteine as a biomarker for the assessment of acrylamide exposure in smokers. Environmental Research. 104(3): 346-51, 2007. 11. Lu Y, Li CM, Wang Z, Ross CR, Chen J, Dalton JT, Li W, Miller DD. Discovery of 4-substituted methoxybenzoyl-aryl-thiazole as novel anticancer agents: synthesis, biological evaluation, and structure-activity relationships. J Med Chem. 52(6):1701- 11, 2009 12. Li CM, Yeh TK, Chen CP, Chuu JJ, Huang CL, Wang HS, Shen CC, Lee TY, Chang CY, Chang CM, Chao YS, Lin CT, Chang JY, Chen CT. Antitumor activities and pharmacokinetics of silatecans DB-67 and DB-91. Pharmacol Res. 61(2):108-115, 2010. 13. Lu Y, Wang Z, Li CM, Chen J, Dalton JT, Li W, Miller DD. Synthesis, in vitro structure-activity relationship, and in vivo studies of 2-arylthiazolidine-4-carboxylic acid amides as anticancer agents. Bioorg Med Chem. 18(2): 477-495, 2010.

FIELD OF STUDY Major Field: Pharmacy

ix TABLE OF CONTENTS

Abstract...... ii Dedication...... v Acknowledgments ...... vi Vita...... vii Table of Contents...... x List of Tables ...... xv List of Figures...... xvii 1. Introduction...... 1 1.1. The role of tubulin and microtuble ...... 1 1.2. The tubulin binding mode of microtubule-stabilizing and destabilizing agents ... 3 1.3. Overview of microtubule stabilizing agents (MSA)...... 3 1.4. Overview of microtubule destabilizing agents ...... 5 1.5. Resistance to tubulin-targeting drugs ...... 6 1.6. Overview of dissertation...... 7 2. Biological Activity of 4-Substituted Methoxybenzoyl-Aryl-Thiazoles (SMART): An Active Microtubule Inhibitor ...... 14 2.1. Introduction...... 14 2.2. Materials and Methods...... 16 2.2.1 In vitro microtubule polymerization assay ...... 16 2.2.2 MS competition binding assay...... 16 2.2.3 Cell culture and cytotoxicity assay of prostate and melanoma cancerc ...... 17 2.2.4 Cell cycle analysis ...... 18 2.2.5 Apoptosis detection by ELISA ...... 18 x 2.2.6 Pharmacokinetic study...... 19 2.2.7 LC-MS/MS method for measuring SMART compounds...... 20 2.2.8 PC-3 and A375 tumor xenograft studies...... 21 2.2.9 Rotarod test...... 21 2.2.10 In vivo drug resistance studied...... 22 2.3. Results...... 22 2.3.1 Modifications of “A” ring of the SMART molecules...... 22 2.3.2 Modifications of “B” ring of the SMART molecules...... 24 2.3.3 Modifications of “C” ring of the SMART molecules...... 25 2.3.4 Modifications of “linker” of the SMART molecules...... 26 2.3.5 SMARTs inhibit microtubule polymerization by binding to the colchicine binding site on tubulin ...... 26 2.3.6 SMART compounds inhibit the growth of multidrug-resistant cancer cell lines...... 27 2.3.7 SMART compounds arrest PC-3 (Prostate) and A375(Melanoma) cells in

G2/M phase of cell cycle and induce cell apoptosis...... 28 2.3.8 In vivo PK profile of SMART compounds and in vitro metabolic stability 29 2.3.9 SMART compounds inhibit prostate and melanoma xenografts growth without neurotoxicity...... 30 2.3.10 SMART-H did not develop drug-resistance in PC-3 tumor bearing mice 32 2.4. Discussion...... 32 2.5. Acknowledgements...... 36 3. A Novel Mass Spectrometry Binding Assay for Determination of Tubulin Binding Site for Small Molecule Inhibitors...... 53 3.1. Introduction...... 53 3.2. Materials and Methods...... 56 3.2.1 LC-MS/MS method for measuring colchicine, vinblastine, and paclitaxel 56 3.2.2 In Vitro tubulin polymerization assay ...... 57 3.2.3 MS binding assay using ultrafiltration ...... 57 3.2.4 Competitive MS binding assay...... 58

xi 3.2.5 Reversible binding assay ...... 59 3.3. Results and sidcussion ...... 59 3.3.1 LC-MS/MS method for colchicine, vinblastine, and paclitaxel ...... 59 3.3.2 Effects of GTP and DMSO on tubulin polymerization ...... 60 3.3.3 Study of ligand-tubulin interaction by the MS binding assay ...... 61 3.3.4 Competitive MS binding assay and its application to determine the binding site of SMART-H on tubulin ...... 62 3.3.5 Reversible binding of SMART-H on colchicine-binding site ...... 63 3.4. Conclusion ...... 64 4. Drug Metabolism and Pharmacokinetics of 4- Substituted Methoxybenzoyl-Aryl- Thiazoles (SMART)...... 70 4.1. Introduction...... 70 4.2. Materials and Methods...... 71 4.2.1 Metabolic incubations...... 71 4.2.2 Protein binding assay ...... 72 4.2.3 Prediction of the in vivo clearance of SMART-H in mouse, rat, dog, and human ...... 73 4.2.4 Pharmacokinetic study ...... 73 4.2.5 Analytical method ...... 75 4.2.6 Cell Culture and Cytotoxicity Assay of prostate cancer ...... 76 4.3. Results...... 76 4.3.1 Metabolic stability ...... 76 4.3.2 Prediction of the in vivo clearance of SMART-H in mouse, rat, dog, and human...... 77 4.3.3 Pharmacokinetic studies of SMART-H ...... 78 4.3.4 Identification of metabolites in human liver microsomes ...... 78 4.3.5 Species specific metabolism of SMART-H...... 80 4.3.6 Blockage of soft spots of SMART-H to increase the metabolic stability in human liver microsomes...... 81 4.4. Discussion...... 82

xii 4.5. Acknowledgements...... 85 5. 2-Arylthiazolidine-4-carboxylic acid amides (ATCAA) targets dual pathways in cancer cells: 5’-AMP-acticated protein kinase (AMPK)/mTOR and PI3K/Akt/mTOR pathways ...... 97 5.1. Introduction...... 97 5.2. Materials and Methods...... 99 5.2.1 Cell culture and cytotoxicity assay ...... 99 5.2.2 Cell cycle analysis ...... 100 5.2.3 FLIPR Intracellular Calcium Mobilization Assays ...... 101 5.2.4 Cell treatment and immunoblotting ...... 101 5.2.5 Nucleotide extraction and FPLC analysis ...... 102 5.2.6 AMPK activity ...... 103 5.2.7 Cell-free purified enzyme activity assay ...... 103 5.2.8 Pharmacokinetic study ...... 104 5.2.9 LC-MS/MS Analytical method ...... 105 5.2.10 A549 tumor xenograft studies ...... 106 5.2.11 Blood glucose and plasma insulin ...... 106 5.3. Results...... 107 5.3.1 ATCAA-10 inhibits the growth of human cancer cell lines and multidrug- resistant cancer cell lines ...... 107 5.3.2 ATCAA-10 induces sub-G1 phase in LNCaP and A375 cell lines ...... 107 5.3.3 ATCAA-10 did not target LPA receptors ...... 108 5.3.4 ATCAA-10 inhibits Akt phosphorylation and activates AMP-activated Protein Kinase (AMPK) ...... 109 5.3.5 ATCAA-10 activates AMPK by decreasing intracellular AMP/ATP ratio ...... 110 5.3.6 Pharmacokinetic study (PK) of ATCAA-10 in mice and rats ...... 111 5.3.7 ATCAA-10 inhibited lung cancer A549 xenograft growth ...... 112 5.3.8 ATCAA-10 did not cause hyperglycemia at a single dose in rats or repeated- doses in mice ...... 112

xiii 5.4. Discussion...... 113 5.5. Acknowledgements...... 116 6. Summary and Discussion...... 128 Reference ...... 133

xiv LIST OF TABLES

Table 2.1 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 42

Table 2.2 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 43

Table 2.3 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 44

Table 2.4 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 45 Table 2.5 Modifications on B-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 46 Table 2.6 Modifications on B-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 47 Table 2.7 Modifications on C-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 48 Table 2.8 Modifications on C-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 49 Table 2.9 Modifications on linkage and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 50 Table 2.10 Modifications on linkage and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown...... 51 Table 2.11 In vitro efficacy of SMART compounds on prostate, melanoma and drug resistant cell lines...... 52

xv Table 4.1 Prediction of in vivo hepatic clearance of SMART-H in mouse, rat, dog, and human from in vitro data...... 94 Table 4.2 Pharmacokinetic parameters of SMART-H in mouse, rat, and dog ...... 95 Table 4.3 Anti-proliferation of SMART compounds on prostate cancer cell lines...... 96 Table 5.1 Potency of ATCAA-10 in inhibiting cell growth in vitro...... 125 Table 5.2 Inhibition of pure enzyme by ATCAA-10...... 126 Table 5.3 Pharmacokinetic parameters of ATCAA-10 (10 mg/kg) in mice and rats ..... 127

xvi LIST OF FIGURES

Figure 1.1 Microtubules are made up of αβ-tubulin that bind head to tail with reversible, noncovalent interaction into 13 protofilaments that form a cylinder of 24 nm in diameter. The end with the β-subunit exposed is termed the plus end and the end with the α-subunit exposed is termed the minus end ...... 8 Figure 1.2 Polymerization dynamics and the GTP cap. Tubulin-bound GTP can be hydrolyzed to guanosine 5’-diphosphate (GDP) on polymerization. A GTP-tubulin can stabilize at their ends to avoid from disassembly or an inorganic phosphate (Pi) binding to two microtubule ends, called a GTP-cap. Ultimately, the Pi dissociates from the microtubule, leaving a microtubule core consisting of tubulin with bound GDP ...... 9 Figure 1.3 Antimitotic drugs bind to microtubules...... 10 Figure 1.4 Structures of paclitaxel, epothilones A and B, and discodermolide...... 11 Figure 1.5 Structure of inhibitors binding to vinblastine site on tubulin...... 12 Figure 1.6 Structure of inhibitors binding to colchicine site on tubulin ...... 13 Figure 2.1 SMART compounds inhibit tubulin polymerization via binding to the colchicine binding site on tubulin...... 37 Figure 2.2 SMART compounds overcome multi-drug resistance in vitro...... 38

Figure 2.3 SMART compounds arrested cells into G2/M phase and induced apoptosis .. 39 Figure 2.4 Pharmacokinetic studies of SMART-H, -F, and -OH in mice and rats...... 40 Figure 2.5 In vivo anti-cancer efficacy and neurotoxicity of SMART compounds...... 41 Figure 3.1 MS/MS of colchicine (A), vinblastine (B), and paclitaxel (C) ...... 65 Figure 3.2 GTP and DMSO effects on tubulin polymerization...... 66 Figure 3.3 MS tubulin binding studies...... 67 Figure 3.4 MS competitive binding studies ...... 68

xvii Figure 3.5 Reversible binding...... 69 Figure 4.1 Phase I metabolic stability of SMART-H in human liver microsome ...... 86 Figure 4.2 Pharmacokinetic studies of SMART-H in mouse (A), rat (B) and dog (C).... 87 Figure 4.3 Metabolite identification...... 88 Figure 4.4 Metabolite profile of SMART-H in human liver microsome...... 89 Figure 4.5 Metabolite profile in species ...... 90 Figure 4.6 Kinetics of four metabolites of SMART-H in human liver microsome (A) and rat liver microsome (B)...... 91 Figure 4.7 Blockage of soft spots ...... 92 Figure 4.8 Metabolic stability of SMART-H (•), -213 (○), -173A (▵), -176A (d), and - 329 (■) in human liver microsome ...... 93 Figure 5.1 ATCAA-10 activated AMPK and deactivated PI3K/Akt pathways ...... 117 Figure 5.2 Flow cytometric analysis of ATCAA-10 ...... 118 Figure 5.3 FLIPR intracellular calcium mobilization assay in agonist mode...... 119 Figure 5.4 FLIPR intracellular calcium mobilization assay in antagonist mode...... 120 Figure 5.5 ATCAA-10 stimulated AMPK via changing the ratio of intracellular AMP/ATP ...... 121 Figure 5.6 Pharmacokinetic studies of ATCAA-10...... 122 Figure 5.7 In vivo efficacy of ATCAA-10...... 123 Figure 5.8 Hyperglycemia and hyperinsulineia test ...... 124

xviii CHAPTER 1

1 INTRODUCTION

1.1. The role of tubulin and microtubule

Microtubules are critical components of the cytoskeleton and play pivotal roles in

normal cellular function such as spindles formation, cellular shape maintenance, and

intracellular transportation. Due to their critical role in mitosis and cell division,

microtubules are regarded as an excellent chemotherapeutic target to treat proliferative

oncogenic disorders. Some of the widely used microtubule-targeted FDA approved drugs

are paclitaxel, docetaxel, ixabepilone and vinca alkaloids such as vinblastine, vincristine

and vinorelbine. Though microtubules are one of the oldest chemotherapeutic targets,

they are still attractive and several groups are investing time and money to develop

improved drugs to target them (1, 2).

Microtubules are made up of αβ-tubulin that binds head to tail with the reversible,

noncovalent interaction into protofilaments (2, 3) (Fig 1.1). Tubulin dimers, which are 46

× 80 × 65 Å in size (width, length, and depth, respectively), are constituted of two 50-

kDa monomers: an α- and a β-subunit, sharing about 40% homology between them.

During polymerization, a hollow cylinder is made by an average of 13 protofilaments in

1 vivo. The protofilaments are aligned with the same polarity. During mitosis, the “minus- end” capped by α-monomer is generally attached to centrosome; the “plus-end” capped

by β-monomers are attached to kinetochore may explore the cell. Structure and function of microtubules are tightly regulated by endogenous proteins. The microtubule-associated

proteins (MAPs) bind to microtubules and stabilize them (4). +Tip proteins bind to (+)

end of microtubule and help for microtubule polymerization (5). Conversely, stathmin (6) induces disassembly of microtubules.

Each tubulin monomer binds to guanosine 5’-triphosphate (GTP), however, only β-

tubulin contains exchangeable GTP, which can be hydrolyzed to guanosine 5’-

diphosphate (GDP) on polymerization. A microtubule end containing tubulin bound GTP

or GDP-Pi is stable, or ‘capped’, against depolymerization. Hydrolysis of tubulin-bound

GTP and the subsequent release of Pi induce a conformational change in tubulin molecules resulting in destabilization, catastrophe and shortening of the microtubules (2)

(Fig 1.2). The complex dynamics of microtubules are extremely sensitive to manipulation. Perturbing dynamic stability, either during polymerization or depolymerization process, results in mitotic arrest at the metaphase-anaphase transition and ultimately leads to cell death. As described above, interfering with tubulin- microtubule dynamics is one of the most attractive and successful molecular strategies to treat cancer.

2 1.2. The tubulin binding mode of microtubule-stabilizing and destabilizing agents

The equilibrium between the dimeric and polymeric forms of tubulin can be altered In vitro by different effecters, such as DMSO (7), cofactors (Mg2+, GTP, GDP) (8), or small molecules (9, 10), that alter the stability of tubulin dimers or the polymerization process.

Three unique small-molecule binding sites are known in tubulin and are responsible for the interaction and pharmacologic effect of paclitaxel, vinblastine, and colchicine (11, 12)

(Fig 1.3). Paclitaxel, which preferably binds to polymeric tubulin as opposed to its dimeric form, is a classic microtubule-stabilizing agent. At high concentration, paclitaxel promotes the assembly of all available tubulins into microtubules (2, 9). At low concentrations, paclitaxel pauses the action of microtubule. Vinblastine and colchicine are microtubule-destabilizing agents, which prefer to bind to dimeric tubulin. Vinblastine binds to the β-tubulin subunit at a distinct region known as the Vinca-binding domain, while colchicine binds to the β subunit at the interface with α monomer of the same tubulin molecule. Small amounts of the tubulin-colchicine complex will copolymerize with free tubulin, with colchicine remaining bound to the microtubule. The dissociation of colchicine from tubulin is extremely slow (13). Similar to vinblastine, colchicine depolymerizes microtubules at high concentrations. Both colchicine and vinblastine effectively halt microtubule dynamics at low concentrations.

1.3. Overview of microtubule stabilizing agents (MSA)

Paclitaxel (Taxol ®) extracted from yew (Taxus baccata) is the first identified microtubule stabilizing agent (14). However, only since 1993, it is used as an anti-cancer 3 agent. Paclitaxel binds near the lateral association interface between protofilaments, reducing the average protofilaments from 13 to 12 (15). Most MSAs discovered till date are from natural sources. Taxanes come from plants; epothilones is from microbial origin; discodermolide, is from sea organism. The structures of these compounds are shown in (Fig 1.4). Paclitaxel and docetaxel share broad spectrum of antitumor activity including breast, lung, ovarian, and bladder cancer. In 2004, clinical data suggest that combination of docetaxel and estramustine or prednisone may prolong survival of patients with prostate cancer (16). Whereas MSAs bind tightly to the polymeric microtubule, they bind with non-measureable affinity to dimeric tubulin (17). The binding affinity is highly related to cytotoxicity in tumor cells from Epothilone studies

(18). The limitations of paclitaxel are low aqueous solubility and the development of clinical resistance, leading to search of new MSAs with greater or comparable efficacy to paclitaxel, but with better solubility and poorer substrates of P-glycoprotein, a known mediator of paclitaxel resistance. Discodermolide and epothilones were reported to exhibit 100- and 30-fold, respectively, more solubility than paclitaxel and both are poor substrates for P-glycoprotein (19, 20). Although discodermolide binds to paclitaxel binding site, a synergism between paclitaxel and discodermolide were observed in human carcinoma cell lines and in a ovarian cancer nude mice model (21, 22).

Major toxicities of paclitaxel and docetaxel are peripheral neuropathy, dose-limiting bone marrow suppression, nausea, vomiting, diarrhea, and alopecia. Both paclitaxel and docetaxel cause type I hypersensitivity reactions characterized by flushing, bronchospasm, dyspnea, and hypotension (23, 24).

4 1.4. Overview of microtubule destabilizing agents

Vinblastine and colchicine are both destabilizing agents due to their ability to the

formation of microtubule, subsequently triggering their disassembly. At low

(substoichiometric) concentrations of colchicine and vinblastine, microtubule remains in

a “pause” state. At high concentrations, they induce depolymerization (25, 26).

Vinblastine isolated from Madagascar periwinkles (Vinca rosea) has long been used

in traditional medicine (27). Vinblastine and its analogs have been used as anticancer agents since 1960s. Both competitive and noncompetitive of drugs binding to vinca domain has been proposed (28). The structures of the chemicals are shown in Fig 1.5.

NSC 707389 (29), which inhibits the binding of radiolabeled vinblastine to tubulin in a

noncompetitive manner (30), is currently in clinical trial as an anticancer agent.

Major toxicities of vinblastine and vincristine include dose-limiting

myelosuppression and neurotoxicity. The most frequent neurotoxicities are numbness and

tingling of extremities, loss of deep tendon reflexes, and distal muscle weakness (31).

Colchicine, extracted from meadow saffron (Colchicum autumnale), has long been

used to block mitosis since 1960s. It is being used in the treatment of acute gout for at

least two centuries (32). Colchicine contains three rings: a trimethoxyphenyl (ring A), a

seven-membered ring (ring B) and a methoxytropone (ring C). Structure-activity

relationship (SAR) studies revealed that trimethoxyphenyl and tropone rings are crucial

for the colchicine to bind to tubulin. The association of colchicine is biphasic, a fast first

step followed by a slow step of isomerization of the tubulin-colchicine complex (33).

Colchicine has never been used clinically use for cancer chemotherapy, primarily due to

5 toxicity (34); leading to search for less toxic agents acting at doses well below their maximum tolerated doses (MTDs). There are increasing number of small molecules binding to colchicine site with diverse structures and relatively simple chemical synthesis schemes compared to agents binding to paclitaxel- or vinblastine-binding sites (Fig 1.6).

Combretastatin A-4 (CA4) is one of the most interesting antitubulin agents binding to colchicine binding site and was considered for clinical trials (35).

2, 4-dichlorobenzyl thiocyanate is another type of antitubulin agent. It alkylated tubulin at multiple cysteine residues by covalent interaction, especially at Cys-239 site of

β-tubulin, resulting in cytotoxic IC50 values with 200-500 nM in cells (36). This reaction is not reversible, and provides an alternative mechanism to interfere with polymerization of tubulin.

1.5. Resistance to tubulin-targeting drugs

In the clinic, development of resistance is an impediment to the effectiveness of anticancer agents. Drug resistance is defined as a state of decreased sensitivity of cancer cells to drugs that would originally cause cell death. For antitubulin agents, some mechanisms have been proposed to be involved in drug resistance. These included but not limited to 1. P-glycoprotein (P-gp)-mediated multidrug resistance (MDR) (37), 2. different expression levels of β-tubulin isotypes (38), 3. tubulin mutation (39). 4. alternation of pharmacokinetic properties.

P-gp spans membrane 12 times and expels hydrophobic, weakly cationic compounds out of the cell (40). P-gp is widely believed for the failure of chemotherapy, and several

6 anticancer drugs such as paclitaxel and docetaxel are substrates of P-gp. However, this mechanism has been questioned in clinical observation. Rowinsky et al. (41) reported that patients resistant to paclitaxel are still sensitive to docetaxel even though both drugs are substrates of P-gp. This argument has been explained by Verschraegen et al. (42), suggesting that patients could relapse because of increased metabolism or excretion in chemotherapy. The administration of a related drug with different pharmacokinetic properties may reduce relapse.

Tubulin mutation is another potential possibility for drug resistance. In vitro β-tubulin mutation resulted in 24-fold resistance to paclitaxel (43). However, to date no mutation that changes the coding sequence of the major β-tubulin has been found in patients’ tumor samples (44, 45).

1.6. Overview of dissertation

Overall objectives of this dissertation were:

1. to study the structure-activity-relationship of 4-Substituted Methoxybenzoyl-Aryl-

Thiazoles (SMART) in prostate cancer cell lines.

2. to validate the molecular mechanism of SMART compounds.

3. to investigate drug metabolism and pharmacokinetics of SMART-H and improve

metabolic stability by blocking the liability spots.

4. to examine the in vivo efficacy of SMART compounds in tumor bearing mice and in

vivo neurotoxicity using rotarod tests.

5. to develop an orally available antitubulin agent.

Figure 1.2 Polymerization dynamics and the GTP cap. Tubulin-bound GTP can be hydrolyzed to guanosine 5’-diphosphate (GDP) on polymerization. A GTP-tubulin can stabilize at their ends to avoid from disassembly or an inorganic phosphate (Pi) binding to two microtubule ends, called a GTP-cap. Ultimately, the Pi dissociates from the microtubule, leaving a microtubule core consisting of tubulin with bound GDP (2).

Figure 1.3 Antimitotic drugs bind to microtubules. Vinblastine bound to high-affinity sites at the microtubule plus end suffice to suppress microtubule dynamics (A). Colchicine forms complexes with tubulin dimers and copolymerizes into the microtubule lattice, suppressing microtubule dynamics (B). A microtubule cut away to show its interior surface. Taxol binds along the interior surface of the microtubule, suppressing its dynamics (C) (2).

AcO O OH

O NH O O O H O OH HO O O O Me

Paclitaxel

R O S

N OH O

O OH O Epothilone A: R=H Epothilone B: R=CH3

HO H O O OH OCNH2 O OH

OH Discodermolide

Figure 1.4 Structures of paclitaxel, epothilones A and B, and discodermolide

Figure 1.5 Structure of inhibitors binding to vinblastine site on tubulin

OH O O H3CO NHCOCH3 O

H3CO O CH3O

O H3CO OCH3 OCH3 OCH3 Colchicine Podophyllotoxin

H3CO OH

H3CO H3CO OCH 3 OH HO OCH3

Combretastatin A-4 2-Methoxyestradiol

Figure 1.6 Structure of inhibitors binding to colchicine site on tubulin

13 CHAPTER 2

2 BIOLOGICAL ACTIVITY OF 4-SUBSTITUTED METHOXYBENZOYL-ARYL-

THIAZOLES (SMART): AN ACTIVE MICROTUBULE INHIBITOR

2.1. Introduction

Microtubules, composed of α- and β-tubulin heterodimers, play an important role in

cell mitosis, motility and organelle distribution (46). Formation of microtubules from

tubulin is a dynamic process that involves polymerization and depolymerization and is

crucial for mitosis (47). Interference with tubulin-microtubule dynamics is one of the

most attractive and successful molecular strategies to treat cancer (2, 35). Paclitaxel,

docetaxel, ixabepilone and three Vinca alkaloids (vinblastine, vincristine, and

vinorelbine), all FDA approved drugs, effectively inhibit tubulin action and successfully

treat cancer (48).

The taxanes and epothilones are semi-synthetic natural products that stabilize

microtubules and cause apoptosis (48). Paclitaxel and ixabepilone occupy overlapping binding sites on the surface of β-tubulin (49) and demonstrate nanomolar cytotoxicity in

a variety of cancer cell lines. Unlike paclitaxel and docetaxel, ixabepilone demonstrates reduced susceptibility to drug resistance mechanisms that limit the effectiveness of

14 taxanes, anthracyclines and a host of other anticancer agents (50). The over-expression of

ATP binding cassette (ABC) proteins, most notably P-glycoprotein (P-gp), in many cancer cells and tumors results in innate or acquired resistance to chemotherapy (51). The ability of ixabepilone to circumvent these multidrug resistance (MDR) mechanisms has provided a unique chemotherapeutic approach for patients with anthracyclines or taxanes resistant tumors. An orally bioavailable sulfonamide (ABT-751) that inhibits tubulin polymerization and maintains activity in cells expressing P-glycoprotein is currently being evaluated in phase I trials (52, 53), but is not yet available for routine clinical use.

Despite the widespread use of the Vinca alkaloids for the treatment of cancer and their susceptibility to these same MDR mechanisms, an anticancer agent that destabilizes microtubules and effectively circumvents P-gp-mediated drug resistance is not yet approved for clinical use.

We recently reported the synthesis and in vitro anticancer characterization of a novel series of 4-Substituted Methoxylbenzoyl-Aryl-Thiazoles (SMART) (54). SMART compounds were synthesized in a simple four-step procedure, providing a method for rapid and high yield synthesis of a variety of analogs. Compounds with an un-substituted aryl ring (SMART-H) or bearing a 4-fluoro substituent (SMART-F) were the most active, with IC50 values ranging from 6 to 55 nM. The studies further revealed that the presence of 3, 4, 5-trimethoxyphenyl ring is essential for the inhibition of tubulin function.

In the present study, we synthesized and characterized an additional 4-hydroxy

SMART compound (SMART-OH) to improve the solubility of the parent SMART-H.

We examined the binding of SMART-H to tubulin, demonstrated that the compounds inhibit tubulin polymerization via binding to the colchicine binding site on tubulin, show 15 that these compounds circumvent P-gp-mediated MDR, and demonstrate promising in

vivo antitumor activity in mice bearing human prostate cancer (PC-3) and melanoma

(A375) tumors. SMART-H did not develop drug resistance in PC-3 xenograft after 21- days treatment. These analogs thus represent a novel series of compounds to inhibit tubulin polymerization and circumvent P-gp-mediated drug resistance with nanomolar anticancer activity in vitro. This discovery of these SMART drugs may provide a novel pharmacologic alternative to the MDR exhibited by several cancers upon prolonged exposure to taxane, anthracycline and Vinca alkaloids.

2.2. Materials and Methods

2.2.1 In vitro microtubule polymerization assay

Bovine brain tubulin (0.4 mg) (Cytoskeleton, Denver, CO) was mixed with 10 μM of the test compound or vehicle (DMSO) and incubated in 100 μl of buffer (80 mM PIPES,

2.0 mM MgCl2, 0.5 mM EGTA, pH 6.9 and 1 mM GTP). The absorbance at 340 nm

wavelength was monitored every min for 15 min (SYNERGY 4 Microplate Reader, Bio-

Tek Instruments, Winooski, VT). The spectrophotometer was maintained at 37 °C for

tubulin polymerization.

2.2.2 MS competition binding assay

Colchicine, vinblastine, and paclitaxel (1.2 μM for each) were incubated with tubulin

(1.2 mg/mL) in the incubation buffer (80 mM PIPES, 2.0 mM MgCl2, 0.5 mM EGTA,

16 pH 6.9) at 37 °C for 1 hr. SMART-H (0.5-125 μM) was examined to individually compete with colchicine-, vinblastine-, and paclitaxel-tubulin binding. The free-form ligands were separated from tubulin or microtubule using an ultrafiltration method

(microconcentrator) (Microcon, Bedford, MA) with a molecular cutoff size of 30k Da.

Colchicine, vinblastine and paclitaxel were separated on a narrow-bore C4 column

(2.1×150 mm, 5 μm, Varian Inc, Palo Alto, CA).Multiple reaction monitoring (MRM) mode, scanning m/z 400.3→ 310.4 (colchicine), m/z 406.3→ 272.1 (vinblastine), m/z

854.6 → 286.3 (paclitaxel), and m/z 434.0 → 266.0 (internal standard, an analog of

SMART compounds) was used to obtain the most sensitive signals for these tubulin ligands. The ability of SMART-H to inhibit the binding of ligands was expressed as a percentage of control binding in the absence of any competitor. Each reaction was run in triplicate.

2.2.3 Cell culture and cytotoxicity assay of prostate and melanoma cancer

All cell lines were obtained from ATCC (American Type Culture Collection,

Manassas, VA, USA), while cell culture supplies were purchased from Cellgro

Mediatech (Herndon, VA, USA). We examined the antiproliferative activity of SMART compounds in four human prostate cancer cell lines (LNCaP, DU 145, PC-3, and PPC-1) and two human melanoma cell lines (A375 and WM-164). Human sarcoma cell line

MES-SA and its doxorubicin-resistant cell line that over-expresses P-gp, MES-SA/Dx 5, were used as MDR models. All prostate cancer cell lines were cultured in RPMI 1640, supplemented with 10% fetal bovine serum (FBS). Melanoma cells were cultured in

17 DMEM, supplemented with 5% FBS, 1% antibiotic/antimycotic mixture (Sigma-Aldrich,

Inc., St. Louis, MO, USA) and bovine insulin (5 μg/ml; Sigma-Aldrich). The cytotoxic potential of the SMART drugs was evaluated using the sulforhodamine B (SRB) assay after 96 h of treatment.

2.2.4 Cell cycle analysis

Flow cytometry was performed to study the effects of the SMART compounds on cell cycle distribution. PC-3 and A375 cells were plated in growth media at 70% confluence.

Medium was changed to 0.5% charcoal-stripped FBS (cs-FBS) for 48 h before the treatment to synchronize the cells in G0/G1 phase of the cell cycle. The cells were then treated in growth media with the indicated concentrations of SMART-H, -F, and -OH for

24 h. Cellular DNA was stained with 100 μg/mL propidium iodide and 100 μg/mL

RNase A in PBS and flow cytometry was performed to determine the cell cycle distribution of the cells.

2.2.5 Apoptosis detection by ELISA

Quantification of the enrichment of mono- and oligo-nucleosomes in the cytoplasm was used to determine the ability of the SMART compounds to induce apoptosis. Briefly,

1×104 PC-3 and A375 cells/well were seeded in 96-well plates and treated with varying concentrations of SMART-H, -F, and -OH for 24 h. The cells were lysed, centrifuged, and the supernatant was added to streptavidin-coated microplates. The cell lysate was mixed with anti-histone antibodies labeled with biotin and anti-DNA antibodies coupled 18 with peroxidase. The peroxidase substrate was added and development of color, an indication of apoptosis, was measured at 405 nm.

2.2.6 Pharmacokinetic study

Male ICR mice (n = 3 or 4 per group) 6 to 8 weeks of age were used to examine the

pharmacokinetics (PK) of the SMART compounds. All animal studies were conducted

under the auspices of a protocol approved by the Institutional Laboratory Animal Care

and Use Center of the University of Tennessee. SMART-H, -F, and –OH (15 mg/kg)

were dissolved in PEG300/DMSO (1/4) and administered by a single i.v. injection into

the tail vein. Blood samples were collected from posterior vena cava in heparinized tubes

under isoflurane anesthesia at 2, 5, 15, and 30 min, 1, 2, 4, 8, 16, and 24 hr after

administration. Blood samples were centrifuged to obtain plasma.

Male Sprague-Dawley rats (n = 4; 254 ± 4 g) were purchased from Harlan Inc.

(Indianapolis, IN). Rat thoracic jugular vein catheters were purchased from Braintree

Scientific Inc. (Braintree, MA). SMART-H and -F were administered intravenously into

the JVC at 2.5 mg/kg (in DMSO/PEG300, 1/4). Equal volume of heparinized saline was

injected to replace the removed blood, and blood samples (250 μL) were collected via the

JVC at 10, 20, 30 min, and 1, 2, 4, 8, 12, 24, 48hr. All heparinized syringes and vials were prepared before blood collection.

A protein precipitation method was used for sample preparation. An aliquot (200 μL) of acetonitrile (ACN) containing an internal standard (100 ng/mL of a SMART analog) was added to 100 µL of plasma and then was thoroughly vortexed for 15 s. After

19 centrifugation, the supernatant was transferred into a clean glass vial and directly used for

liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis. The PK

parameters were determined using Non compartment analysis (WinNonlin, Pharsight

Corporation, Mountain View, CA)

2.2.7 LC-MS/MS method for measuring SMART compounds

Aliquots (10 μL) of the supernatant were injected into the HPLC system (Model 1100

Series Chemstation, Agilent Technology Co, Santa Clara, CA). SMART-H, SMART-F,

and SMART-OH were separated on a narrow-bore C4 column (2.1×150 mm, 5 μm,

Varian Inc, Palo Alto, CA). Gradient mode was used to achieve separation of the analytes

using mixtures of mobile phase A (5% acetonitrile containing 0.1% formic acid) and

mobile phase B (95% acetonitrile containing 0.1% formic acid) at a flow rate of 300

μL/min. Mobile phase A was used at 80% from 0 to 1 min followed by a linearly

programmed gradient to 100% of mobile phase B within 3 min, 100% of mobile phase B

was maintained for 1 min before a quick ramp to 80% mobile phase A. Mobile phase A

was continued for another 8 min towards the end of analysis.

A triple-quadruple mass spectrometer (API QtrapTM Applied Biosystems/MDS

SCIEX, Concord, Ontario, Canada) operating with a TurboIonSpray source was used.

Multiple reaction monitoring (MRM) mode, scanning m/z 356.2 → 188.2 (SMART-H),

m/z 374.2→ 206.2 (SMART-F), m/z 372.2→ 204.2 (SMART-OH), and m/z 434.0 →

266.0 (IS), was used to obtain the most sensitive signals for the SMART compounds.

20 2.2.8 PC-3 and A375 tumor xenograft studies

PC-3 and A375 cells (5×107 per ml) were prepared in phenol red-free growth media

containing 10% FBS, and mixed with Matrigel (BD Biosciences, San Jose, CA) at 1:1

ratio. Tumors were established by injecting 100 μL of the mixture (2.5×106 cells per animal) subcutaneously (s.c.) into the flank of 6-8-week-old male athymic nude mice.

Length and width of tumors were measured and the tumor volume (mm3) was calculated

by the formula, π/6 × L × W2, where length (L) and width (W) were determined in mm.

When the tumor volumes reached 150 mm3, the animals bearing PC-3 tumors were

treated with vehicle [Captex200/Tween80 (1/4)], SMART-H (5 and 15 mg/kg), SMART-

F (5 and 15 mg/kg) and SMART-OH (50 mg/kg) intraperitorally for 21 days. Vinblastine

(0.5 mg/kg) was used as the positive control and dosed q2d with vehicle [DMSO/PEG300

(1/9)]. On the other hand, A375 tumor bearing mice were treated for 34 days with vehicle

[Captex200/Tween80 (1/4)], SMART-H (20 mg/kg) or SMART-F (15 mg/kg). Tumor

growth inhibition (TGI, %) was represented as antitumor efficacy. TGI, % was measured

by tumor volumes (mm3) on the first day and final day for drug-treated compared with

vehical-treated mice and calculated as {1-[(Volume, treatment, final day – Volume, treatment, day 1)

/ (Volume, vehicle, final day – Volume, vehicle, day1)]} × 100%.

2.2.9 Rotarod test

ICR mice received training three times a day for two days to enable them to stay on

the rotating rod for >120 seconds at 12 rpm. Mice were then randomized by the length of

time that they could stay on the rotating rod and divided into 7-8 mice per group. 21 SMART-H at a dose of 5 or 15 mg/kg in Captex200/Tween80 (1/4) was administered by

intraperitoneal injection. Vinblastine at a dose of 0.5 mg/kg/day was used as a positive

control under the same conditions. The rotarod test was performed twice a week. The rod

speed was increased from 5 rpm to 40 rpm over a period of 5 min. Performance was measured as the length of time that a mouse could stay on the rotating rod.

2.2.10 In vivo drug resistance studies

At the end of the PC-3 xenograft studies, solid tumors from control and SMART-H treated (15 mg/kg) groups were removed and digested with 0.1% collagenase (Type I) and 50 mg/ml DNAse (Worthington Biochemical Corp., Freehold, NJ). Dispersed cells were plated in RPMI medium + 10% FBS and incubated at 37°C and 5% CO2 for 24 hr to

allow attachment. The anti-proliferative effects of SMART-H were compared to

determine whether tumor cells remaining in PC-3 xenografts retained sensitivity to drug.

The PC-3 cells obtained from ATCC were used as in vitro control. Statistical analyses

were performed using simple t-Test.

2.3. Results

2.3.1 Modifications of “A” ring of the SMART molecules

In SAR studies of the SMART compounds, different para-substitutions, including

electron-withdrawing groups (EWG) and electron-donating groups (EDG) on “A” phenyl

ring of the SMART compounds, were examined by cytotoxicity assays (Table 2.1-2.3).

22 However, electronic effects of “A” ring phenyl substituent did not show clear influence

on antiproliferative activity. With a weak EWD, (ID45, IC50 values: 6-43 nM) or weak

EDG (ID104, IC50 values: 5-21 nM), both increased potency compared to ID100. The

replacement of para- position with strong EWG such as NO2 (ID110), CN (ID132), CF3

(ID170) or introducing strong EDG (3,4-dimethoxy) to “A” phenyl ring (ID52) exhibited comparable antiproliferative activity.

To compare the effects of ortho-, meta- and para- substitutions, a fluoro atom was introduced to different positions of “A” phenyl ring (ID129, 134 and 45). However, the various o-, m-, p- substituents showed different activities. p-fluoro substituted (ID45) has the best activity in prostate cancer cells (6-13 nM) while o-fluoro substituted (ID129) has lowest potency (IC50s: 52-114 nM) among the three substituted compounds on prostate

cancer cells. Meta-substituted compound (ID134) showed moderate inhibition on prostate

cancer cells (IC50s: 23-46 nM). Although differences were found along the position, all

three compounds still remained cytotoxicity under 120 nM. Interestingly, three fluoro

(ID249) on A-ring caused activity loss. With a steric hindrance group on the “A” phenyl

ring substituents, p-bromo (ID182, IC50s: 30-47 nM) caused a slight decrease in

antiproliferative activity relative to p-fluoro position (ID45, IC50s: 6-12 nM). When p- methyl (ID104, IC50s: 5-21 nM) was replaced with a p-ethyl group (ID188, IC50s: 17-70

nM), activity was also slightly decreased. These data suggested that para- position is the best site for substitution and the size of the substitutes favored small size. When greatly increased molecular size, such as ID203, 286, 142, 149, and 261, the potency decreased in micromolar ranges. When removed the A-ring (ID168, IC50s: 540-680 nM), the activity was lost compared to ID100, suggesting that phenyl ring played an important 23 role. When phenyl ring was replaced to pyridine ring (ID84, IC50s: >10,000 nM) or 2-

pyrimidine ring (ID122, IC50s: 1220-6540 nM), activity was dramatically decreased.

However, thiophene ring (ID105, IC50s: 13-48 nM), 2-indole ring (ID296, IC50s: 24-32 nM) and 4-indole ring (ID293, IC50s: 7-13 nM) showed great potency, providing

alternative choice for A-ring (Table 2.4).

Three compounds (ID211, ID183H, and ID287, IC50s < 150 nM) were successfully

designed to improve solubility without sacrifice in potency. ID100, ID45, and ID211,

namely SMART-H, SMART-F, and SMART-OH, respectively, were then used to

perform in vitro and in vivo pharmacology in this study.

2.3.2 Modifications of “B” ring of the SMART molecules

Variety of B-ring replacements were tested (Table 2.5 and 2.6) and some of them

exhibited comparable potency, such as furan ring (ID241, IC50s: 20-35 nM), thiophene

ring (ID247, IC50s: 12-26 nM), pyrazole ring (ID320, IC50s: 84-140 nM), and pyridine

ring (ID313, IC50s: 25-33 nM), oxazole ring (ID163, IC50s: 292-324 nM), and phenyl ring

(ID127, IC50s: 81-234 nM). However, some B-ring replacements abolished activity, such

as 4,5-dihydrooxazole ring (ID165, IC50s: 1100-1200 nM), isoxazole ring (ID304, IC50s >

10,000 nM), pyrimidine (ID199, IC50s: 3700-8300 nM), and piperidine (ID194, IC50s: >

20,000 nM). Interestingly, the ID299 (IC50s > 10,000 nM) has the same thiazole ring as

ID100, but lost its potency due to different position linked to carbonyl linker.

24 2.3.3 Modifications of “C” ring of the SMART molecules

Introducing different substituted phenyls or alkyl chain on C ring were carried out in

this study. Variation of the phenyl substituents has a remarkable change in effect on

potency. The in vitro assay as shown in Table 2.7-2.8 gave us an interesting result but only 3, 4, 5-trimethoxylphenyl in “C” ring (ID100) showed excellent inhibition against

all cancer cells (IC50= 21 -71 nM, average IC50= 41 nM). Compound ID186, with a 3, 5- dimethoxyphenyl group, showed 8-fold average cytotoxicity lower than ID100 against prostate cancer cell lines (IC50 = 313-347 nM, average IC50= 334 nM). Modifications of

ID100 by removal of one methoxy at meta-position (ID64) or two methoxy groups (ID62,

ID67 and ID76) or three methoxy groups (ID97) from ID100 led to a dramatic decrease in activity (IC50 >10 μM). Although ortho- substituted monomethoxy compound ID76

exhibited weak activity against a certain cell lines compared with meta-/para-MeO

substituted ID67/ID62 and dimethoxyphenyl compound ID64, none of them showed

significant potency in inhibition compared to ID100. Similar trends were also seen in

ID78 and ID279 with 2- and 4- fluorophenyl in “C” ring modifications. However, with 2-

, 3-, 4-, 5-, 6-fluorophenyl (ID 213, IC50s: 1300-2600 nM) exhibited weak activity. It

reduced potency if trimethoxy group was replaced to trihydroxy group (ID 184, IC50 >

2000 nM). Pyridine ring and its salt form (ID102 and ID 103, IC50s: > 20,000 nM) and

hexadecyl (ID29, IC50s: > 20,000 nM) replaced trimethoxyphenyl group did not remain

potency.

25 2.3.4 Modifications of “linker” of the SMART molecules

Carboxyl linker provided excellent potency in this class of compounds. However, it

was found the ketone group is a labile site based on our metabolites identification study

in human liver microsome in vitro (see section 4.3.5). Some linker replacements were

tested in this study (Table 2.9 and 2.10). Previously, 2-Arylthiazolidine-4-carboxylic acid

amides (ATCAA) compounds with the amide CONH linkage were developed as

anticancer agents in our laboratory. The amide CONH linkage was tested in SMART

compounds, such as ID28, ID70, ID72, and ID173b. However, their IC50s are more than

10,000 nM. When eliminated the linkage (ID329, IC50s > 10,000 nM), the activity did not exhibit. To address the labile site of ketone, ID193, ID173A, ID176A, ID176B, ID331,

ID339a, and ID339b were designed to avoid ketone reduction reaction. All compounds lost activity with IC50s > 1,000 nM except for ID 173A (IC50s: 102-189 nM). ID173A,

namely SMART-173A, also showed prolonged half-life in human liver microsomes (see

section 4.3.6).

2.3.5 SMARTs inhibit microtubule polymerization by binding to the colchicine binding

site on tubulin

Based on structure-activity relationship studies, three SMART compounds (Fig. 2.1A)

were selected for biological characterization. While SMART-H and SMART-F are highly

potent molecules with low nanomolar cytotoxic properties, SMART-OH, which was

rationally designed as a potential metabolite with improved solubility, had the least

potent anti-proliferative effects (data was shown in Table 2.11).

26 Bovine brain tubulin (>97% pure) was incubated with the individual SMART

compounds (10 μM) to test their effect on tubulin polymerization (Fig. 2.1B). While

SMART-H and SMART-F inhibited tubulin polymerization by 90%, SMART-OH

inhibited the polymerization by only 55%. Previous studies demonstrated a

concentration-dependent inhibition of tubulin polymerization by SMART-H. In addition,

under the same experimental conditions, the IC50 for SMART-H (4.23 μM) is similar to

that of colchicine (4.91 μM). These data suggest that SMART compounds exhibit strong

antitubulin polymerization activity that corresponds well with their cytotoxicity (Table

2.11).

The ability of the SMART compounds to compete for known binding sites on tubulin

was determined using a novel MS competitive binding assay, which was developed in our laboratory (data not published). Three tubulin ligands, corresponding to the three binding sites on tubulin, colchicine, vinblastine, and paclitaxel were used for these competitive binding studies. We found that, over a concentration range of 0.1-125 μM, SMART-H specifically competed with colchicine binding to tubulin, but it did not compete with either vinblastine or paclitaxel binding to tubulin (Fig. 2.1C).

2.3.6 SMART compounds inhibit the growth of multidrug-resistant cancer cell lines

The ability of SMART compounds to inhibit the growth of cancer cell lines was evaluated using the SRB assay. As shown in Table 2.11, the SMART compounds inhibited the growth of several human cancer cell lines, including four prostate cancer cell lines, and two melanoma cell lines, with IC50 values in the low nanomolar range. Out

27 of the three SMART compounds, SMART-OH was the least potent (IC50 76~116 nM).

SMART-F exhibited the best anti-proliferative effect with IC50 values between 6 and 43

nM in prostate cancer and melanoma cell lines. In addition, the effect of SMART

compounds in the doxorubicin-resistant cell line MES-SA/Dx5 was also evaluated (Table

2.1). SMART compounds were equally potent against MDR cells and the parent cells

(Fig. 2.2). Paclitaxel, vinblastine, and colchicine were significantly less effective in the

drug-resistant cell line. In addition, the slope of dose-response curve varied for paclitaxel,

vinblastine and colchicine (55), but not for the SMART compounds (Fig. 2.2). These data

indicate that the SMART compounds circumvent P-gp-mediated drug resistance.

2.3.7 SMART compounds arrest PC-3 (Prostate) and A375 (Melanoma) cells in G2/M phase of cell cycle and induce cell apoptosis

PC-3 and A375 cells were exposed to 10, 50, 200, and 1000 nM of SMART compounds for 24 h. Treatment with the SMART compounds resulted in concentration- dependent accumulation of both PC-3 and A375 cells in the G2/M phase with

concomitant decreases in the percentage of cells in G0/G1 phase (Fig. 2.3A and 2.3B).

The proportion of cells in G2/M phase significantly increased when treated with 50 to 200

nM of SMART-H, –F, and -OH.

Apoptosis was then examined by measuring the level of cytoplasmic DNA-histone

complexes in PC-3 and A375 cells after 24 h treatment. Increasing concentration of the

SMART compounds increased the level of cytoplasmic DNA-histone complexes in PC-3

and A375 cells (Fig. 2.3C). The effect was more pronounced in A375 cells than PC-3

28 cells, but apoptosis was evident in both cell types. SMART-H and SMART-F induced

moderate apoptosis at a concentration of 50 nM, while SMART-OH induced apoptosis

only at concentrations greater than or equal to 200 nM.

2.3.8 In vivo PK profile of SMART compounds and in vitro metabolic stability

A single dose bolus of each SMART compound (15 mg/kg) was administered by tail

vein injection to ICR mice to characterize their pharmacokinetics (Fig. 2.4A). SMART-H

and -F exhibited similar PK properties, but SMART-OH exhibited slightly greater AUC

than SMART-H and SMART-F, indicative of a lower clearance for SMART-OH (Fig.

2.4C). SMART-OH also had 2-3 times higher Vss than that of SMART-H and –F. The clearance values for all three SMARTs were equal to or higher than 90 ml/min/kg (the hepatic blood flow rate in mice (56)), suggesting that in addition to hepatic removal,

other degradation routes may be involved in the elimination of SMARTs.

The pharmacokinetics of SMART-H and –F (2.5 mg/kg) were also examined in rats

(Fig. 2.4B). Interestingly, low clearance values and hepatic extraction rates were obtained

by both SMART compounds, suggesting that SMART compounds exhibit species

differences in clearance. In rats, SMART-H exhibited favorable pharmacokinetic

properties, which are low clearance (6 ml/min/kg), moderate volume of distribution (7.6

L/kg), long half-life (24 hr), and high exposure (AUC, 5.8 hr*μg/ml) (Fig. 2.4C).

In vitro metabolic stability studies were conducted using 0.5 μM substrate and 1 mg/mL microsomal proteins to estimate the in vitro stability of the SMART compounds in liver microsomes isolated from five different species (Fig. 2.4D). The data

29 demonstrated that SMART compounds are extremely unstable in mice and monkey liver

microsomes, having half-lives less than 5 min. The half-life of SMART-H in human liver

microsomes (17 min) was close to that observed for dog (19 min), but less than that in the rat liver microsomes (30 min). SMART-OH exhibited slightly longer half-lives than

SMART-H and –F in all species. These in vitro data suggest that SMART compounds exhibit significant species differences in metabolism. SMART compounds were rapidly metabolized in mice liver microsomes, corresponding well with the high hepatic clearance observed during pharmacokinetic studies, and presenting a challenge for tumor xenograft studies in this species.

2.3.9 SMART compounds inhibit prostate and melanoma xenografts growth without

neurotoxicity

Prostate cancer PC-3 and melanoma A375 tumors in mice were allowed to reach a volume of 150 mm3 and then tumor-bearing mice were treated with the SMART

compounds. As shown in Fig. 2.5A, tumor volumes in the control group increased to 680

± 198 mm3 over the 21 day duration of the study. Tumor volumes in the SMART-H- treated group increased to 370 ± 103 mm3 (5 mg/kg treatment) and 176 ± 112 mm3 (15 mg/kg treatment) by day 21, indicating strong anti-tumor activity for this compound.

Tumors in the SMART-F-treated animals increased to 269 ± 177 mm3 (5 mg/kg

treatment) and 292 ± 103 mm3 (15 mg/kg treatment), while animals in the SMART-OH

(50 mg/kg) treated group had tumors of 331 ± 130 mm3 at day 21. This reduction in

tumor volume reversed upon withdrawal of SMART compounds (data not shown).

30 SMART-H tumor elicited 71% and 96% tumor growth inhibition (TGI) at 5 and 15

mg/kg treatment, respectively, whereas, SMART-F elicited TGI of 79% and 76% at 5 and

15 mg/kg treatment, respectively. The high dose of SMART-OH (50 mg/kg) exhibited

the TGI of 66%. Vinblastine, the positive control, showed TGI of 72% at day 22 in PC-3

xenografts (Fig. 2.5C). Body weight measurements, to monitor toxicity, indicated that

only 1 of 8 mice treated with SMART-H (15 mg/kg), and 2 out of 7 mice treated with

SMART-F (15 mg/kg) lost more than 15 % body weight. In addition to the antitumor

effects of the SMART compounds on PC-3 prostate tumors, SMART-H (20 mg/kg) and

SMART-F (15 mg/kg) demonstrated a significant reduction of A375 tumors. As shown in

Fig. 2.5C, the tumor volumes of control group increased to 2183 ± 279 mm3, whereas the

volumes in SMART-H and SMART-F treatment groups increased to 775 ± 107 mm3 and

722 ± 135 mm3, respectively. SMART-H and SMART-F treatment evoked 70% and 72

% TGI, respectively.

Rotarod tests were performed to examine the in vivo neurotoxic effects of SMART-H.

Based on the result of in vivo efficacy experiments, 5 or 15 mg/kg [i.p. administration,

Captex200/Tween80 (1/4)] of SMART-H was chosen to study the effect on motor

coordination. A 0.5 mg/kg treatment with vinblastine was used as the positive control

under the same conditions. As shown in Fig. 2.5D, vinblastine gradually reduced the time

(in seconds) that the mice could stay on the rotating rod, and attained significance by

days 27 and 31 (p < 0.05) compared to the vehicle group. However, no significant difference was observed in the SMART-H treatment groups, suggesting that SMART-H

did not cause neurotoxicity in ICR mice at doses that are associated with antitumor

effects. 31

2.3.10 SMART-H did not develop drug-resistance in PC-3 tumor bearing mice

We excised the PC-3 tumors from nude mice after 21 days of treatment with vehicle

(n = 3) or 15 mg/kg SMART-H (n = 3). Solid tumors were digested and dispersed into

cells as described in the methods section. PC-3 cell line from ATCC (American Type

Culture Collection, Manassas, VA, USA) was used as a control. IC50 values were 29.1 ±

1.1, 29.1 ± 0.8, and 30.4 ± 0.5 nM in PC-3 cells from ATCC, and dissociated cells from

vehicle and SMART-H treated tumors, respectively. These data demonstrate that

SMART-H did not induce drug-resistance in PC-3 tumors after 21 days of continuous

SMART-H treatment.

2.4. Discussion

The general molecular mechanism of action of antitubulin agents is to bind to

microtubules, arrest cells in G2/M phase, and cause cell apoptosis. In this study, we

examined the ability of three novel SMART compounds to bind tubulin, inhibit its

polymerization and induce apoptosis in two tumor cell types. The SMART compounds

bound to the colchicine binding site in tubulin, disturbed tubulin polymerization, arrested

cells in the G2/M phase, induced apoptosis, and displayed potent antiproliferative activities against prostate and melanoma cell lines in vitro and in vivo. Consistent with

these biological effects, the in vivo antiproliferative activity of SMART-H and SMART-F

were greater than SMART-OH.

32 Currently, there are many potential microtubule stabilizers or destabilizers in various stages of development (1). Some antimitotic compounds, such as the vinca alkaloids and taxanes, are already in clinical use. However, the vinca alkaloids and taxanes are substrates of P-gp, which pumps drugs out of cancer cells and is thought to limits their effectiveness. Compared to other drug classes, antitubulin agents have somewhat different resistance mechanisms. For example, over-expression of P-glycoprotein, changes in microtubule dynamics (57) and mutations in β-tubulin are known to limit sensitivity to the taxanes (58). Two mechanisms of drug resistance have been examined with the SMART compounds in this study: (a) the presence of drug efflux pumps and (b) development of drug resistance by tumor in vivo. We tested SMART compounds against a cell line that over-expresses P-gp. Notably, the SMART compounds demonstrated equi- potent anti-proliferative effects against the MES-SA derived MDR-positive cell line,

MES-SA/Dx5, suggesting that SMART compounds are not P-gp substrates and that they function in a P-gp-independent manner. This feature is distinct from that of paclitaxel, vinblastine, and colchicine in MES-SA/Dx5 cells. We also examined the cytotoxicity of

SMART-H in PC-3 tumor cells (in vivo) excised from mice, which were treated with 15 mg/kg SMART-H for 21 days. SMART-H equi-potently inhibited the growth of cells derived from the primary tumor cell culture and ATCC PC-3 cell line, indicating that prolonged treatment with SMART-H did not develop drug resistance in PC-3 tumors over

21 days.

Many drug candidates encounter problems during development associated with chemical instability, high metabolic clearance, and peripheral neurotoxicity (59-61). We performed PK studies with the SMART compounds in mice, and found that the SMART 33 compounds exhibited high clearance. Based on metabolic stability studies using liver

microsomes in vitro, we predicted that the high clearance observed in mice may be due to

rapid hepatic metabolism. However, the SMART compounds exhibited species

differences in metabolism, and were more stable in liver microsomes obtained from rats,

dogs and human. We also performed PK studies in rats. Total plasma clearance of

SMART-H in rats was 6 ml/min/kg, indicative of a low hepatic extraction rate (rat

hepatic blood rate is 55 ml/min/kg (56)). These PK data demonstrate that in vitro

metabolic stability studies provided a reliable prediction for in vivo clearance in animals,

and further suggest that xenograft studies in mice may underestimate the antitumor

efficacy achievable in humans due to inter-species differences in metabolism. We

developed a suitable formulation to increase drug exposure after intraperitoneal injection.

We compared the AUC (drug exposure) using two different formulations and intraperitoneal injection prior to the initiation of tumor xenograft studies. SMART-H (15

mg/kg) in Captex200/Tween80 (1/4) exhibited 4-fold higher AUC compared to the

PEG300/DMSO (1/4) formulation, indicating that the Captex200/Tween80 (1/4)

formulation facilitated a greater exposure to SMART-H (data not shown). Based on PK

data (high clearance and short half-life) from mice, we employed a high frequency dosing

regimen. The promising in vivo antitumor activity of SMART-H and SMART-F in a

species that rapidly eliminates these drugs, coupled with their improved stability in

human liver microsomes, further suggests these SMART compounds may demonstrate

anticancer activity in patients with prostate cancer or melanoma.

One of the major adverse events associated with antitubulin agents is peripheral

neurotoxicity. The rotarod assay has been used to examine in vivo neurotoxicity (62, 63). 34 In this report, the rotarod assay was conducted to test peripheral neurotoxicity of

SMART-H in ICR mice. Mice treated with SMART-H did not show significant impairment in motor coordination compared to the vehicle group. The positive control, vinblastine-treated group, demonstrated a lesser ability to stay on the rotating rod, indicative of neurotoxicity. We also extended the experiment for 4 weeks after drug treatment was stopped, and found that the ability of the vinblastine-treated group to stay on the rotating rod failed to return to baseline or to the vehicle-treated group values, indicating that neurotoxicity could only be partially recovered with time (data not shown). These data provide compelling evidence that we have developed a novel family of SMART compounds with lesser neurotoxicity.

Prostate cancer PC-3 and melanoma A375 cells were chosen to examine the activity of

SMART compounds in vivo because SMART compounds exhibited greater potency in these cancer cell lines in vitro. A375 xenografts exhibited higher tumor growth rate than

PC-3 xenografts. This corresponded well with their in vitro growth rate. In this report,

SMART-H and SMART-F both showed significant efficacy on both low and high growth rate xenograft models in vivo. Compared to the control groups, SMART-H and –F inhibited tumor growth significantly at doses between 5 and 20 mg/kg. We obtained more than 70 % TGI with SMART-H and -F treatments in both xenograft models. We also calculated % T/C for our efficacy study. By National Cancer Institute criteria, % T/C <

42% is considered to be moderately active (64). SMART-H at a dosed 15 mg/kg and

SMART-F at a dosed 5 mg/kg in the PC-3 xenograft model provided % T/C values of 26

% and 39%, respectively. In addition, 20 mg/kg of SMART-H and 15 mg/kg of SMART-

F in A375 xenograft demonstrated % T/C values of 36% and 33%, respectively. Though 35 SMART-OH, was tested at a dose higher than that of SMART-H and SMART-F, and demonstrated more favorable pharmacokinetic properties, it exhibited 10-fold less potency than SMART-H in PC-3 xenograft (Fig. 2.5A). Thus, the lower tubulin inhibitory potency of SMART-OH was not sufficiently compensated by its lower clearance and greater exposure. These data further suggest that SMART-OH, if a metabolite of SMART-H, would contribute little, if any, to the in vivo anticancer activity of SMART-H.

In summary, we have demonstrated that SMART compounds are potent antitubulin compounds that overcome P-gp-mediated drug resistance in vitro and did not develop drug-resistance after treatment in vivo. The efficacy studies revealed that SMART compounds have activities in PC-3 and A375 cancer xenografts. Importantly, SMART-H did not show neurotoxicity in vivo. The promising anticancer effect of the SMART compounds and their lesser neurotoxicity suggest that these novel compounds have therapeutic potential for cancer chemotherapy.

2.5. Acknowledgements

Thank Terrence A. Costello, Katie N. Kail and Stacey L. Barnett for providing technical support for animal studies at GTx Inc. The data of Fig 2.5 B were provided by

Sunjoo Ahn. The data of Fig 2.3 A, B and Fig 2.5 C were provided by Zhao Wang. The cytotoxic data in melanoma in Table 2.11 were provided by Zhao Wang.

Figure 2.1 SMART compounds inhibit tubulin polymerization via binding to the colchicine binding site on tubulin. Structures of SMART-H, -F, and –OH (A). Effect of SMART compounds on tubulin polymerization. Tubulin (0.4 mg) was exposed to SMART compounds (10 μM). Absorbance at 340 nm was monitored every min for 15 min (B). Ability of SMART-H to compete for colchicine, vinblastine and paclitaxel binding sites on tubulin using mass spectrometry competitive binding assay (C) (n = 3); bars, SD.

Figure 2.2 SMART compounds overcome multi-drug resistance in vitro. Dose-response curve of SMART-H, -F, and –OH in MES-SA and MES-SA/Dx cells. Paclitaxel, vinblastine, and colchicine were used as positive controls (n = 3); bars, SD.

Figure 2.3 SMART compounds arrested cells into G2/M phase and induced apoptosis. Representative graphs of cell cycle analysis after compounds treatment for 24 h on PC-3 and A375 cells (A). The changes in G2/M proportion induced by SMART-H, -F, and –OH in PC-3 and A375 cells after 24h treatment (B). Ability of SMART-H, -F, and –OH to enhance cytoplasmic DNA-Histone complex formation in 24h (C) (n = 3); bars, SD. Colchicine and vinblastine were used as positive controls.

Figure 2.4 Pharmacokinetic studies of SMART-H, -F, and -OH in mice and rats. Concentration-time curve of SMARTs in ICR mice (n = 3); bars, SD. SMARTs were administrated 15mg/kg i.v. by tail vein injection (A). Concentration-time curve of SMART-H and SMART-F in SD rats (n = 4); bars, SD. SD rats were dosed 2.5 mg/kg i.v. with the formulation DMSO/PEG300 (1/4) (B). Pharmacokinetic parameters of SMART compounds. Parameters were obtained following IV administration in mice and rats (C). In vitro microsomal stability of SMART compounds in five species liver microsomes (n = 3). 0.5 μM substrate and 1 mg/mL microsomal proteins were incubated at 37°C and 50% disappearance of parent compounds was represented as half-life (D).

Figure 2.5 In vivo anti-cancer efficacy and neurotoxicity of SMART compounds. SMARTs efficacy for PC-3 prostate tumor xenografted on nude mice (A) (n = 6-8). Vinblastine efficacy for PC-3 prostate tumor xenografted on nude mice (n = 8). This served as the positive control (B). In vivo efficacy of SMART-H and SMART-F in nude mice bearing A375 melanoma xenografts (n = 10). Nude mice were inoculated 2.5 × 106 PC-3 or A375 cells and dosed i.p. daily (SMART compounds) and q2d (vinblastine) after tumor formation (150-200 mm3). Each point represents mean tumor volume for animals in each group (C). In vivo neurotoxicity (rotarod test) of SMART-H in ICR mice (n = 7 or 8). SMART-H (5 and 15 mg/kg), vinblastine (0.5 mg/kg) and vehivle were given i.p. daily, and vinblastine was used as the positive control (D). *, p < 0.05. Bars, SE

Table 2.1 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.2 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.3 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.4 Modifications on A-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.5 Modifications on B-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

47 Table 2.6 Modifications on B-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.7 Modifications on C-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.8 Modifications on C-ring and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.9 Modifications on linkage and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

Table 2.10 Modifications on linkage and anticancer activity of SMART derivatives. IC50 values for cell growth inhibition are shown.

a Previously reported in reference (54).

Table 2.11 In vitro efficacy of SMART compounds on prostate, melanoma and drug resistant cell lines (n = 3, mean ± SE). Paciltaxel, vinblastine, and colchicine were used as positive controls.

CHAPTER 3

3 A NOVEL MASS SPECTROMETRY BINDING ASSAY FOR DETERMINATION

OF TUBULIN BINDING SITE FOR SMALL MOLECULE INHIBITORS

3.1. Introduction

Tubulins are the building blocks of microtubules, which are dynamic structural

components of the cellular cytoskeleton. Polymerization and depolymerization of

microtubules play important roles during normal cell division, while perturbing

microtubule function leads to mitotic attract and ultimately to cell death by apoptosis (46,

47). Antitubulin agents continue to arrest significant attention in drug discovery pipelines

owing to their chemotherapeutic potential (2, 35). Antitubulin agents are divided into two categories, microtubule stabilizers and destabilizers, based on their ability to bind to tubulin and change the ratio between assembled microtubules and dimeric tubulin.

Recently, we developed a class of potent antitubulin agents, 4-Substituted

Methoxylbenzoyl-Aryl-Thiazoles (SMART), which effectively inhibit tubulin polymerization, and exhibit low nanomolar cytotoxic IC50 values in a variety of cancer cell types (54).

In vitro, the equilibrium between the dimeric and polymeric forms of tubulin can be altered by different effecters, such as DMSO (7), cofactors (Mg2+, GTP, GDP)(8), or small molecules (9, 10), that alter the stability of tubulin dimers or the polymerization process. Three unique binding sites for small molecule antitubulin agents are known and are responsible for the interactions and pharmacologic effect of paclitaxel, vinblastine, and colchicine (11, 12). Paclitaxel, which preferably binds to polymeric tubulin as opposed to its dimeric form, is a classic type of microtubule-stabilizing agent. At high concentration, paclitaxel promotes the assembly of all the available tubulin into microtubules (9). Vinblastine and colchicine are microtubule-destabilizing agents, which prefer to bind to dimeric tubulin. Vinblastine binds to the β-tubulin subunit at a distinct region known as the Vinca-binding domain, while colchicine binds to the β subunit at the interface with α monomer of the same tubulin molecule. Small amounts of the tubulin- colchicine complex will copolymerize with free tubulin, with colchicine remaining bound to the microtubule (65). The dissociation of colchicine from tubulin is extremely slow

(13). As occurs with vinblastine, colchicine depolymerizes microtubules at high concentrations. Both colchicine and vinblastine effectively halt microtubule dynamics at low concentrations.

Radioligand binding assays have been used over the past two decades to characterize the binding of small molecule inhibitors to tubulin (66). Although such assays provide a sensitive method to study ligand-target interactions, radioligand assays also have inherent disadvantages, including the need for expensive radiolabeled drugs that require special handling, disposal, detection and synthesis. Issues related to isotope effects and stability may also complicate such assays, but are often overlooked or ignored (67). Mass 54

spectrometry (MS) provides an alternative technique for sensitive and selective

quantification of ligand that can be applied to the study of ligand-target interactions. MS

binding assays share all of the advantages and principles of radioligand binding assays,

but eliminate the disadvantages of radioactive use and allow for simultaneous

measurement of different ligands that cannot be accomplished when using radioligands

because of limitations associated with using and the sole availability of 3H or 14C-labeled

radioligands. There are additional differences between MS binding and radioligand

binding assays. Bound radioligand as a fraction of total radioactivity is always measured

during radioligand binding assays, while the concentration of free ligand after protein precipitation is quantified during MS analysis. Alternatively, MS binding assay can be also used to directly quantify the concentration of unbound ligand (68-70).

In this study, we developed an MS binding method to determine the binding site of a novel antitubulin drug on tubulin. The non-radiolabeled ligand was incubated with tubulin, and then the unbound fraction was isolated by ultrafiltration. The filtrate was injected into the LC-MS/MS system without further sample preparation. Three well-

known tubulin binders, colchicine, vinblastine, and paclitaxel, were used to optimize and

validate the binding study with tubulin. The optimized binder and tubulin concentrations

were then used to examine the tubulin binding characteristics of their competitors,

podophyllotoxin, vincristine, and docetaxel, respectively. Once validated, this MS

binding assay was applied to determine the binding site of a novel SMART compound

(SMART-H). The method demonstrated that SMART-H effectively and specifically bound to the colchicine-binding site, but not to vinblastine- or paclitaxel-binding sites. In addition, the ability of colchicine to displace SMART-H from its tubulin binding site was 55

assessed, a feat not possible using standard radioligand binding assays without the

synthesis of radiolabeled SMART-H. These studies provide the first evidence that MS

can be used to quickly and selectively identify the tubulin binding site of novel small

molecule inhibitors.

3.2. Materials and Methods

3.2.1 LC-MS/MS method for measuring colchicine, vinblastine, and paclitaxel

An HPLC system (Model 1100 Series Chemstation, Agilent Technology Co, Santa

Clara, CA), narrow-bore C4 column (Varian Inc, 2.1×150 mm, 5 μm, Palo Alto, CA) and

triple-quadruple mass spectrometer (API QtrapTM Applied Biosystems/MDS SCIEX,

Concord, Ontario, Canada) with a TurboIonSpray source were used to develop this

method. Gradient mode was used to achieve separation of the analytes using mixtures of

mobile phase A (5%/95% acetonitrile/H2O containing 0.1% formic acid) and mobile

phase B (95%/5% acetonitrile/H2O containing 0.1% formic acid) at a flow rate of 300

μL/min. Mobile phase A was used at 90% from 0 to 1 min followed by a linearly programmed gradient to 100% of mobile phase B within 3 min. 100% of mobile phase B was maintained for 1 min before a quick ramp to 90% mobile phase A. Mobile phase A was continued for another 10 min towards the end of analysis.

Multiple reaction monitoring (MRM) mode, scanning m/z 400.3→ 310.4 (colchicine), m/z 406.3→ 272.1 (vinblastine), m/z 854.6 → 286.3 (paclitaxel), and m/z 434.0 → 266.0

(internal standard, an analog of SMART compounds), was used to obtain the most sensitive signals for these tubulin ligands. The spraying needle voltage was set at 5000 V 56

for positive mode, Collision-Assisted-Dissociation (CAD) gas at medium, and the source

heater probe temperature at 500°C. CE were set at 40, 35, and 45 V, respectively. Data

acquisition and quantitative processing were accomplished using AnalystTM software,

Ver. 4.0 (Applied Biosystems).

3.2.2 In Vitro tubulin polymerization assay

Bovine brain tubulin (0.2 mg) (Cytoskeleton, Denver, CO) was mixed with vehicle

(DMSO, 5% v/v) and GTP (1 mM) and incubated in 100 μl of buffer (80 mM PIPES, 2.0

mM MgCl2, 0.5 mM EGTA, pH 6.9) at 37 °C for 1hr. The absorbance at 340 nm

wavelength was monitored every min (SYNERGY 4 Microplate Reader, Bio-Tek

Instruments, Winooski, VT). The spectrophotometer was maintained at 37 °C for tubulin

polymerization.

3.2.3 MS binding assay using ultrafiltration

Colchicine, vinblastine, and paclitaxel (1.2 μM concentration) were incubated with varying concentrations of tubulin (8-4000 μg/mL) in the incubation buffer (80 mM

PIPES, 2.0 mM MgCl2, 0.5 mM EGTA, pH 6.9) in a shaking incubator (37 °C) for 1 hr.

To study the colchicine and vinblastine binding sites, the incubations were done under

conditions which did not contain GTP because these ligands prefer to bind to dimeric

tubulin. For paclitaxel binding studies, pre-formed microtubules were prepared by pre-

incubating tubulin in the presence of GTP (1 mM) for 1hr. Paclitaxel was added to the

pre-formed microtubules, and incubated for an additional hour. The unbound ligands

were separated from tubulin or microtubules using an ultrafiltration method

(microconcentrator) (Microcon, Bedford, MA) with a molecular cutoff size of 30k Da.

The filtrate (50 μL) was diluted with acetonitrile/H2O (1/2) (150 μL) containing an

analog of our SMART compound as internal standard. An aliquot of 10 μL was injected

into the LC-MS/MS system.

3.2.4 Competitive MS binding assay

Competitive MS binding assay was conducted as previously described but with

slightly different conditions for colchicine, vinblastine, and paclitaxel binding.

Colchicine, vinblastine, and paclitaxel (1.2 μM for each) were incubated with tubulin (1.2

mg/mL) in the incubation buffer (80 mM PIPES, 2.0 mM MgCl2, 0.5 mM EGTA, pH

6.9) at 37 °C for 1 hr. Varying concentrations (0.1-125 μM) of podophyllotoxin,

vincristine, and docetaxel were used to compete with colchicine-, vinblastine-, and

paclitaxel-tubulin binding, respectively. After incubation, the filtrate was obtained as

previously described. SMART-H (0.5-125 μM) was examined to individually compete with colchicine-, vinblastine-, and paclitaxel-tubulin binding. The ability of the competitor or SMART-H to inhibit the binding of ligands was expressed as a percentage of control binding in the absence of any competitor. Each study was run in triplicate.

3.2.5 Reversible binding assay

SMART-H (1 μM) was incubated with tubulin (1.0 mg/mL) in the incubation buffer

(80 mM PIPES, 2.0 mM MgCl2, 0.5 mM EGTA, pH 6.9) at 37 °C for 1 hr. Varying

concentrations (0.5-500 μM) of colchicine were added to compete with SMART-H- tubulin binding. After incubation, the filtrate was obtained as described previously. The

MRM ion pair of SMART-H, m/z 356.2 → 188.2, was monitored by the LC-MS/MS method using the same chromatographic conditions as described previously. The ability of colchicine to inhibit the binding of SMART-H-tubulin was expressed as a percentage of control binding in the absence of any SMART-H. Each experiment was performed in triplicate.

3.3. Results and discussion

3.3.1 LC-MS/MS method for colchicine, vinblastine, and paclitaxel

The full-scan electrospray mass spectra of colchicine, vinblastine, and paclitaxel, were

obtained by infusion of a solution of each analyte (final concentration 2 μM) into the

mass spectrometer (data not shown). The data indicated that singly charged species were

the predominant ions for colchicine ([M+H]+, m/z 400) and paclitaxel (MH+, m/z 854 and

[M+Na]+, m/z 876) in full-scan mode by electrospray ionization (ESI). However, the

intensity of the [M+2H]2+ ion at m/z 406 was more than 5 times greater than that of the

singly charged [M+H]+ ion for vinblastine. The singly charged precursor ions of

colchicine and paclitaxel, and the doubly charged precursor ion of vinblastine were

selected for fragmentation to generate respective product ions (Fig 3.1A-C). The MS/MS

spectrum of colchicine (Fig 3.1A) was consistent with that of Emuri et al. and Yao et al.

(71, 72), who reported the major fragment ions (m/z 358, 310, 326, 282) from the singly

protonated form of colchicine. In Fig 3.1B, the predominant product ion m/z 272 of

double charged vinblastine was consistent with Dennison (73). For paclitaxel (m/z 854),

the main fragment ions, m/z 569, 551, 509, and 286 were consistent with Liia et al. (74).

In subsequent experiments using MRM modes for quantitative measurement, the MRM

ion pairs, m/z 400.3→310.4, 406.3→272.1, and 854.6→286.3, were selected to result in a highly specific and sensitive method for quantitation of colchicine, vinblastine, and paclitaxel, respectively. An internal standard, an analog of SMART compounds, was added to improve the analytical accuracy of the quantitative method.

The LC-MS/MS method reported here included C4 reversed-phase chromatography using a gradient program of ACN/water with 0.1% formic acid. The chromatograms indicate that colchicine, vinblastine, paclitaxel, and the internal standard were completely separated under these chromatographic conditions (Fig 3.1D). To our knowledge, this is the first analytical method with the capability to simultaneously quantify the three antitubulin agents in a well-separated chromatography.

3.3.2 Effects of GTP and DMSO on tubulin polymerization

Fig 3.2A shows the effects of 1 mM GTP, 5% DMSO, and the combination effects of

GTP and DMSO on tubulin polymerization, respectively. The buffer (without added GTP or DMSO) did not significantly cause tubulin aggregation. Both DMSO and GTP

promoted tubulin aggregation, while the combination produced a hyperbolic curve with a

higher plateau, suggesting that GTP and DMSO synergistically increased the extent of

tubulin polymerization. Fig 3.2B shows the polymerization curve that was obtained when

tubulin was pre-incubated with GTP (1 mM) for 1hr prior to the addition of DMSO or

buffer. The extent of tubulin polymerization increased after DMSO was added. These

data suggested that pre-incubation with GTP (Fig 3.2B) should be avoided when

performing studies to examine binding to dimeric tubulin (e.g., colchicine and

vinblastine), but would be beneficial when performing studies with paclitaxel , which preferentially binds to polymeric microtubules.

3.3.3 Study of ligand-tubulin interaction by the MS binding assay

Colchicine (1.2 μM) was incubated with varying concentrations of tubulin in the absence of GTP for 1hr. Ultrafiltration was then used to separate free colchicine and macro-molecules, such as tubulin and microtubules. The filtrate was used for quantification without further sample preparation. Fig 3.3A shows that the percentage of colchicine bound to tubulin increased with increasing tubulin concentrations, and approached 100% at tubulin concentration > 1 mg/mL. The binding of colchicine is thought to be a two-step process where initial complex formation is followed by a slow conformational change resulting in the formation of a stable complex (75). The stair-step increase in the percentage of colchicine bound observed during our studies, supports the idea that colchicine exhibits a different binding mechanism or kinetics with different ratios of colchicine/tubulin. The vinblastine-tubulin binding (Fig 3.3B) was examined

under the same conditions. For paclitaxel-tubulin binding, tubulin (4 mg/mL) was pre-

incubated in the presence of 1 mM GTP to generate partial microtubules, and then titrated to the indicated tubulin concentrations. Paclitaxel (1.2 μM) was then incubated with the pre-formed microtubules binding study (Fig 3.3C). Fig 3.3A-C show that tubulin concentrations of 1000 μg/mL or greater bound > 80% of each ligand. Thus tubulin

concentrations of 1 to 1.5 mg/mL were used for further competitive binding studies.

The tubulin binding assays were performed in a nonvolatile incubation buffer, which

is not compatible with the MS system. However, the incubation buffer can be diverted

into waste by a simple switching valve during HPLC, avoiding its entrance to the MS

system. The analytes (e.g. colchicine, vinblastine and paclitaxel) were retained longer in

the column and then were detected by LC-MS/MS. Therefore, no further sample

preparation (desalting) was needed after ultrafiltration. This approach is not limited by

the buffer system used. The longer retention times and baseline resolution obtained

during HPLC separation further served to minimize ion suppression effects.

3.3.4 Competitive MS binding assay and its application to determine the binding site of

SMART-H on tubulin

Tubulin (1-1.5 mg/mL) was incubated with colchicine, vinblastine, and paclitaxel (1.2

μM). Both colchicine- and vinblastine-tubulin binding studies were examined in the

absence of GTP without pre-incubation. Paclitaxel-tubulin binding studies were

performed after preincubation of tubulin with GTP for 1 hr. Podophylltoxin, vincristine,

and docetaxel are well-known competitors for colchicine, vinblastine, and paclitaxel,

respectively, binding to tubulin. As shown in Fig 3.4B-D, the MS competitive binding method demonstrated that podophylltoxin, vincristine, and docetaxel effectively competed for the colchicine-, vinblastine-, and paclitaxel-tubulin binding sites, respectively, indicating that competitive displacement of the known ligand can be quantified and identified using this newly developed MS binding assay. The MS competitive binding method was further employed to determine the binding site of

SMART-H (Fig 3.4A) on tubulin. SMART-H is a novel anti-tubulin agent that potently inhibits tubulin polymerization (54). Varying concentrations of SMART-H were used to compete with colchicine-, vinblastine-, and paclitaxel-tubulin binding. Fig 3.4B-D show that SMART-H competed effectively and specifically with colchicine-tubulin binding, but not to the Vinca alkaloid or paclitaxel sites, indicating that SMART-H bound to the colchicine binding site on tubulin. SMART-H was slightly less potent than podophyllotoxin, suggesting that determinations of relative affinity to tubulin can also be obtained using this method.

3.3.5 Reversible binding of SMART-H on colchicine-binding site

The ability to use non-radiolabeled ligand in the MS binding assay allowed us to also examine the ability of colchicine to displace SMART-H from tubulin binding site.

Varying concentrations of colchicine were used to compete for the binding of tubulin (1 mg/mL) by SMART-H (1 μM). The LC-MS/MS was set to detect SMART-H (m/z 356.2

→ 188.2) with the same chromatographic conditions (data not shown). Colchicine competed effectively with the SMART-H-tubulin binding (Fig 3.5), suggesting that the

binding of SMART-H to tubulin is not covalent and further corroborating the conclusion that the binding site for SMART-H overlaps with the colchicine-binding site. This

reversible binding study demonstrated that competitive MS binding assay is simple and

widely applicable to non-radiolabeled agents.

3.4. Conclusion

We developed a MS-based assay to study the interactions of small molecule inhibitors

with tubulin. The concentration of unbound ligands was determined after one-step sample

purification using ultrafiltration. The assay is broadly applicable to any small molecule

that can be measured by MS, as demonstrated by studies using colchicine,

podophyllotoxin, vinblastine, vincristine, paclitaxel, docetaxel and SMART-H. The assay

was used to demonstrate that SMART-H reversibly competes for the colchicine binding

site on tubulin with an affinity slightly less than podophyllotoxin. The assay does not

require the use of radiolabeled materials can be used to study drugs that inhibit or

promote tubulin polymerization, could be applied to drugs with unknown binding sites in

tubulin and thus represents a novel technique with broad applicability in the study of

ligand-tubulin interactions.

+MS2 (400.30) CE (40): 0.988 to 1.055 min from Sample 6 (12240804_VTC_MSMS_C400) of 122408_VTC.wiff (Turbo Spray) Max. 2.2e5 cps. +MS2 (406.30) CE (35): 1.173 to 1.223 min from Sample 2 (12260802_VTC_MSMS_V406) of 122608_VTC.wiff (Turbo Spray) Max. 5.2e5 cps. A 310.4 B 100% 100% 272.1 OH H3CO N Et 90% NHCOCH3 90% 282.4 326.3 H CO 80% 3 80% N CH3O N Et 70% O H CO 2Me H 70% OCOMe OCH 358.4 MeO N 3 H CO2Me 60% 60% Me OH colchicine 285.2 295.4 vinblastine 50% 400.6 50% 40% 311.1 340.5 40% 267.3 30% 30%

20% 20% 355.7 210.3 10% 10%

50 100 150 200 250 300 350 400 450 100 200 300 400 500 600 700 800 900 m/z, amu m/ z, amu

+MS2 (854.90) CE (25): 1.223 to 1.340 min from Sample 7 (12260806_VTC_MSMS_T855) of 122608_VTC.wiff (Turbo Spray) Max. 3.0e4 cps. TIC of +MRM (4 pairs): from Sample 8 (11150900_STD) of 11150900_VTC_STD.wiff (Turbo Spray) Max. 1.5e5 cps. 286.4 5.6 100% C AcO O OH 1.5e5 D vinblastine

90% O NH O 1.4e5 O O H colchicine 80% O OH HO O 1.2e5 O Me 70% O 509.6 1.0e5 6.2 60% 8.0e4 paclitaxel 50% paclitaxel 854.8 6.0e4 40% 439.9 I.S. 551.4 30% 4.0e4 7.3 20% 7.9 2.0e4 10% 0.0 100 200 300 400 500 600 700 800 900 2468 m/z, amu Time, min

Figure 3.1 MS/MS of colchicine (A), vinblastine (B), and paclitaxel (C). The three tubulin ligands (0.3 μM for each) were thoroughly separated using our chromatographic method (D).

Figure 3.2 GTP and DMSO effects on tubulin polymerization. Tubulin (2 mg/mL) was incubated in the presence or absence of GTP (1mM), DMSO (5%, v/v) (A) for 1hr. Pre- incubated tubulin (2 mg/mL) in the presence GTP (1mM) for 1 hr, and then added buffer or DMSO (5%, v/v) (B).

Figure 3.3 MS tubulin binding studies. The binding of vary concentrations of tubulin with three ligands (1.2 μM for each), colchicine (A), vinblastine (B), and taxol (C). Both A and B were incubated in the absence of GTP for 1hr; C, tubulin were pre-incubated in the presence of GTP for 1hr, and then add the ligand (taxol) for another hour. Vehicle, DMSO (5%, v/v) (N = 3).

Figure 3.4 MS competitive binding studies. SMART-H (A) competitively binds to tubulin (1.3 mg/mL) with three ligands, colchicine (B), vinblastine (C), and paclitaxel (D). Both B and C incubated tubulin with ligand (colchicine or vinblastine, 1.2 μM) and varying concentrations of test compound in the absence of GTP. D, pre-incubated tubulin in the presence of 1 mM GTP for 1hr, and then add the ligand (paclitaxel, 1.2 μM) and varying concentrations of test compound in the presence of 1mM GTP. Podophylltoxin, vincristine, and docetaxel were used as positive controls for competitive binding with colchicine, vincristine, and paclitaxel, respectively. Vehicle was 5% DMSO; bar, SD. (N = 3).

100

0 % SMART-H bound on tubulin

-20 0.1 1 10 100 1000

Colchicine concentration, μM

Figure 3.5 Reversible binding. Vary concentrations of colchicine competed with SMART-H (1 μM) using tubulin (1 mg/mL) in the absence of GTP. N=3; bar, SD.

CHAPTER 4

4 DRUG METABOLISM AND PHARMACOKINETICS OF 4- SUBSTITUTED

METHOXYBENZOYL-ARYL-THIAZOLES (SMART)

4.1. Introduction

Cancer is the second leading cause of death, with 500,000 people estimated to die

from the disease in 2010 in The United States (Cancer facts& figures 2010, The

American Cancer Society Website, cancer.org). Despite multiple pathways being utilized

for the development of novel therapeutics, tubulins remain an attractive target to treat

cancer (76, 77). However, most of the tubulin polymerizing and depolymerizing agents are p-glycoprotein (P-gp) substrates, leading to the development of resistance over

prolonged period of treatment (48).

Recently our group developed a series of 4-Substituted Methoxybenzoyl-Aryl-

Thiazoles (SMART) that bind to the colchicine-binding site and inhibit tubulin

polymerization and cancer cell growth at low nanomolar concentrations (54). In addition, the SMART compounds were not substrates of P-gp and retained potent anticancer activity demonstrated both in vitro and in vivo in wild-type and resistant cancer cells (Li et al., 2010).

Pharmacokinetics (PK) and metabolism play important roles in lead optimization as they affect the exposure and thus efficacy of drugs (78). Many molecules that were demonstrated to be potent in vitro, failed in vivo due to poor PK properties. A potential drug candidate is expected to be cleared slowly from the system and to have sufficient exposure to elicit its action. Identification of chemical or functional groups or so called

“soft spots” is a key first step in the design of compounds with better metabolic stability

(79). Liver microsomes, which contain several key enzymes such as cytochrome P450s

(CYPs), flavin monooxygenases and glucuronosyltransferases, all required for drug metabolism, are widely used for in vitro metabolic stability studies. In vitro microsomal stability assays not only facilitate the selection of compounds with greater metabolic stability, but are also useful to identify metabolite of the parent molecule (80).

In this study, we identified the metabolites and PK properties of one of the SMART compounds, SMART-H, using in vitro liver microsomes and PK studies. We also identified the labile sites and developed the next generation of SMART compounds having better metabolic stability with little or no impact on potency.

4.2. Materials and Methods

4.2.1 Metabolic stability studies

Incubations for metabolic stability studies were conducted in 1 ml reaction volume containing 0.5 μM (final concentration) of SMART-H or other test compounds and 1 mg/mL microsomal protein (mouse, rat, dog and human liver microsomes, Xenotech,

Kansas City, MO) in reaction buffer [0.2 M of phosphate buffer solution (pH 7.4), 1.3 71

mM NADP+, 3.3 mM glucose-6-phosphate, and 0.4 U/mL glucose-6-phosphate

dehydrogenase] at 37 °C in a shaking water bath. SMART-H, at 50 μM concentration, with the above mentioned conditions, was used for metabolite identification studies. For glucuronidation studies, 2 mM UDP-glucuronic acid (Sigma, St. Louis, MO) cofactor in deionized water was incubated with 8 mM MgCl2, 25 μg of alamethicin (Sigma, St.

Louis, MO) in deionized water, and NADPH regenerating solutions (BD Biosciences,

Bedford, MA) as described previously. The total DMSO concentration in the reaction

solution was approximately 0.5% (v/v). Aliquots (100 μL) from the reaction mixtures were sampled at 5, 10, 20, 30, 60, and 120 min, and acetonitrile (150 μL) containing 100 nM of internal standard (an analog of SMART-H) was added to quench the reaction and to precipitate the proteins. Samples were then centrifuged at 4,000 g for 15 min at room temperature, and the supernatant was analyzed directly by LC-MS/MS.

4.2.2 Protein binding assay

Plasma protein binding studies of SMART-H were conducted by ultrafiltration technique. One ml of mouse, rat, dog and human plasma (Fisher, Pittsburgh, PA) samples was spiked with 5 μl of 100 μM SMART-H and incubated at 37 °C for 30 min prior to ultrafiltration. Samples (400 μL) were transferred to Amicon centrifugal filter devices

(30-kd molecular weight cutoff, Millipore, Bedford, MA, USA) and centrifuged at 14,000 g for 20 min. An aliquot (50 μL) of the ultrafiltrate was combined with 150 μL acetonitrile containing an internal standard for LC-MS/MS analysis. Protein binding of

SMART-H in mouse, rat, dog, and human microsomes was carried out using 0.5 μM 72

SMART-H and 1 mg/mL microsomal proteins in the absence of NADPH. The incubation

condition and sample preparation were the same as described for plasma protein binding.

4.2.3 Prediction of the in vivo clearance of SMART-H in mouse, rat, dog, and human

In vivo clearance was predicted utilizing the data obtained from metabolic stability

(half-life in liver microsomes), and protein binding in plasma and liver microsomes

studies. The intrinsic hepatic clearance (Cli, in vitro) of SMART-H was determined using

the equation: Cli, in vitro = [0.693 / (t1/2, min • protein concentration, mg/mL)]. The intrinsic

clearance was then scaled to predict clearance that would occur in the liver in vivo.

Scaling factors (mg protein/g liver • g liver/kg body weight) are 2400, 1815, and 1980 for

rat, dog, and human (81, 82). In vivo intrinsic hepatic clearance (Cli,h, ml/min/kg body

weight) in liver was estimated by multiplying Cli, in vitro by the scaling factors. In vivo

hepatic clearance (Clh) was estimated by incorporating estimates of Cli,h, Qh and fu into the well-stirred model (venous equation): Clh = [Qh • fu,p • (Cli,h/ fu,m)]/(Qh + fu,p • (Cli,h/

fu,m)) (83). The Qh and fu,p and fu,m represented hepatic blood flow, fraction unbound in plasma, and fraction unbound in microsomes, respectively.

4.2.4 Pharmacokinetic studies

All animal studies were conducted under the auspices of a protocol reviewed and

approved by the Institutional Laboratory Animal Care and Use Committee of The

University of Tennessee. SMART-H (15 mg/kg) was dissolved in PEG300/DMSO (1/4)

and administered once intravenously into the tail vein of 6-8 week old ICR mice (n=3 per 73

each time point, Harlan Inc., Indianapolis, IN). Blood samples were collected in

heparinized tubes via cardiac puncture under isoflurane anesthesia at 2, 5, 15, and 30 min,

1, 2, 4, 8, 16, and 24 hr after administration. Plasma samples were collected by

centrifugation at 8,000 g for 5 min, and stored immediately at -80°C for further analysis.

SMART-H was administered intravenously into thoracic jugular vein (catheters from

Braintree Scientific Inc, Braintree, MA) of male Sprague-Dawley rats (n=4; 254 ± 4g,

Harlan Inc., Indianapolis, IN) at 2.5 mg/kg (in DMSO/PEG300, 1/4). Equal volume of

heparinized saline was injected to replace the removed blood, and blood samples (250

μL) were collected via the jugular vein catheters at 10, 20, 30 min, and 1, 2, 4, 8, 12, 24,

48hr. All syringes and vials were heparinized for blood collection. Plasma samples were

obtained as previously described.

A female beagle dog weighing 11.2 kg was used in this study. The animal was fasted

overnight and until 2 hr after drug administration. The dog was given a single intravenous

dose of SMART-H (0.25 mg/kg, in DMSO/PFG300, 1/4). Blood was drawn at 10, 20, 30

min. and 1, 2, 4, 8, 12, 24, 48, 96 hr. Plasma samples were obtained as previously

described.

SMART-H was extracted from 100 μL of plasma with 200 μL of acetonitrile

containing 100 nM internal standard. The samples were thoroughly mixed, centrifuged,

and the organic extract was transferred to autosampler for LC-MS/MS analysis. The PK

parameters were determined using non-compartmental analysis (WinNonlin, Pharsight

Corporation, Mountain View, CA)

4.2.5 Analytical method

Sample solution (10 μL) was injected into an Agilent series HPLC system (Agilent

1100 Series Agilent 1100 Chemstation, Agilent Technology Co, Ltd). SMART-H and its

metabolites were separated on a narrow-bore C4 column (Varian Inc, 2.1×150 mm, 5 μm,

Palo Alto, CA). Two gradient modes were used. For metabolic stability, gradient mode

was used to achieve the separation of analytes using mixtures of mobile phase A

(ACN/H2O (55%/45%) containing 0.1% formic acid) and mobile phase B (ACN/H2O

(95%/5%) containing 0.1% formic acid) at a flow rate of 300 μL/min. Mobile phase A

was used at 55% from 0 to 0.5 min followed by a linearly programmed gradient to 100% of mobile phase B within 2.5 min, 100% of mobile phase B was maintained for 1 min before a quick ramp to 55% mobile phase A. Mobile phase A was continued for another 8 min towards the end of analysis. For metabolite identification studies, a slower gradient mode was used to achieve the separation of analytes with the same flow rate and mobile phase A and B as described. Mobile phase A was used at 20% from 0 to 1 min followed by a linearly programmed gradient to 100% of mobile phase B within 17 min, 100% of mobile phase B was maintained for 2 min before a quick ramp to 20% mobile phase A.

Mobile phase A was continued for another 25 min towards the end of analysis.

A triple-quadruple mass spectrometer, API QtrapTM (Applied Biosystems/MDS

SCIEX, Concord, Ontario, Canada), operating with a TurboIonSpray source was used.

The spraying needle voltage was set at 5 kV for positive mode. Curtain gas was set at 10;

Gas 1 and gas 2 were set at 50. Collision-Assisted-Dissociation (CAD) gas at medium and the source heater probe temperature at 500°C. Data acquisition and quantitative

processing were accomplished using AnalystTM software, Ver. 1.4.1 (Applied

Biosystems).

4.2.6 Cell Culture and Cytotoxicity Assay of prostate cancer

We examined the antiproliferative activity of the SMART compounds in four human

prostate cancer cell lines, LNCaP, DU 145, PC-3, and PPC-1 (ATCC, American Type

Culture Collection, Manassas,VA, USA). Cells were cultured in RPMI 1640 (Cellgro

Mediatech, Inc., Herndon, VA, USA) supplemented with 10% FBS (Cellgro Mediatech)

and were maintained at 37°C in a humidified atmosphere containing 5% CO2. Depending

on cell types, 1000 to 5000 cells were plated into each well of 96-well plates and exposed to different concentrations of the compound of interest for 96 h. At the end of the

treatments, cell viability was measured using the sulforhodamine B (SRB) assay.

Percentage of cell survival was plotted against drug concentrations and the IC50 values

(concentration that inhibited cell growth by 50% of untreated control) were obtained by

nonlinear regression analysis using WinNonlin.

4.3. Results

4.3.1 Metabolic stability

Figure 4.1 shows the metabolic stability of SMART-H in the presence and absence of

NADPH and UDP-glucuronic acid. In the absence of NADPH, more than 85% of the

parent SMART-H remained after 120 min, suggesting that metabolism of SMART-H was

NADPH-dependent. In the presence of NADPH, SMART-H had a half-life of 17 min by

phase I reaction, suggesting that SMART-H was rapidly metabolized by phase I

metabolic processes. The half-life (17 min) in the presence of UDP-glucuronic acid was identical to that observed in its absence (Fig 4.1). As such, SMART-H exhibited moderate to high in vitro clearance in human liver microsomes exclusively through phase

I reaction.

4.3.2 Prediction of the in vivo clearance of SMART-H in mouse, rat, dog, and human

Table 4.1 summarizes in vitro half-lives (in liver microsomes), protein binding (in

plasma and liver microsomes), and clearance predictions in mouse, rat, dog, and human

sample. Half-lives were <5, 31, 19, and 17 min for mouse, rat, dog, and human,

respectively, indicating that SMART-H exhibited inter-species variability in its

metabolism. The protein binding results indicated that SMART-H is highly protein bound

(> 90% bound) in plasma and liver microsomes for all four species. With high protein

binding, SMART-H was predicted to have low hepatic clearances and extraction rates (≤

0.3) in rat, dog, and human. Mouse was an exception. SMART-H was extremely unstable

in mouse liver microsomes. Therefore, a high clearance and extraction rate were

predicted in this species. Encouragingly, in vitro protein binding and metabolic stability

further suggested that humans will have a low hepatic extraction ratio and low clearance.

4.3.3 Pharmacokinetic studies of SMART-H

A single dose IV bolus of SMART-H was administered to ICR mice, Sprague-Dawley rats, and a Beagle dog to characterize the pharmacokinetics in these species (Fig 4.2).

Their PK parameters are summarized in Table 4.2. In vivo clearances were 130, 5.3, and

2.7 ml/min/kg for mouse, rat, and dog, respectively (Table 4.2), indicating a high hepatic extraction ratio (> 0.7) for mouse and low extraction ratios (< 0.3) for rat and dog. These data were consistent with our predictions based on in vitro data, suggesting that prediction of in vivo clearance based on in vitro protein binding and metabolic stability was reliable. In mouse, SMART-H clearance was higher than 90 ml/min/kg (the hepatic blood flow rate in mice), suggesting that in addition to hepatic removal, other degradation routes may be involved in the elimination of SMART-H in this species. Although

SMART-H exhibited high protein binding in mouse, it did not compensate for instability in mouse liver microsomes. These data suggest that metabolic stability of SMART-H was critical for hepatic clearance. Intermediate volumes of distribution of 4.9 and 11 L/kg were obtained for mouse and rat. However, a smaller value of 0.4 L/kg was obtained in the dog, suggesting that SMART-H also exhibited species difference in volume of distribution. SMART-H exhibited long half-lives (more than 24 hr) in both rat and dog, while a short half-life was obtained in mouse due to its high clearance.

4.3.4 Identification of metabolites in human liver microsomes

In vitro metabolite identification studies were performed using human liver microsomes to identify metabolically labile sites (i.e., soft spots) in the SMART-H

pharmacophore. LC-MS/MS was used for metabolite identification based on mass shifts compared to the molecular ion [M+H]+ and retention time shifts to the parent SMART-H.

To identify the potential metabolites, the product ion scan for each peak of interest was examined to obtain structural information. Four major metabolites of SMART-H were identified in human liver microsomes. Figure 4.3A shows the multiple-reaction- monitoring (MRM) chromatography of SMART-H with its four metabolites (M1-M4) after the 2h incubation. Due to the high polarity exhibited by the hydroxyl groups, all the metabolites had shorter retention time than SMART-H. Two metabolites, M2 and M3, were present as isomers and were separated based on the gradient chromatographic condition. The product ion spectrum of SMART-H (Fig 4.3B) had an abundant product ion m/z 188 (loss of trimethoxy benzene). The major and most abundant metabolite, M1

(m/z 358, mass shifts +2 Da, ketone reduction) (Fig 4.3C), resulted in the fragment ion m/z 340 (loss of H2O). This metabolite was synthesized and the retention time and product ion spectrum were confirmed (data not shown). This metabolite demonstrated an altered fragment ion spectra compared to SMART-H and thus could not be detected by either precursor ion or neutral loss scans. M2 (m/z 342, mass shifts -14 Da, demethylation) (Fig 4.3D) was the second most abundant metabolite of SMART-H in human liver microsomes with the same fragment ion pathway and product ion m/z as the parent SMART-H. M2 was present as both 3- or 4- demethylated metabolite of SMART-

H, which were separated by our chromatographic conditions. However, we were unable to distinguish which of the methoxy groups were demethylated. M3 (m/z 344, mass shift

-12 Da, ketone-reduction and demethylation) (Fig 4.3E) demonstrated a fragmentation pattern similar to M1, and was apparently formed after M1. The major fragment ion of 79

M3 was m/z 326 (loss of H2O). M3 exhibited the shortest retention time due to the presence of two hydroxyl groups. M4 (m/z 374, mass shift +16 Da, hydroxylation) (Fig

4.3F) was an oxidized metabolite, having a hydroxyl group on the B-ring and

fragmentation pattern similar to M1. Demethylated metabolite (M2) was the only one to

be detected by the precursor ion scan (precursor of m/z 188). On the contrary, other

metabolites exhibited different fragmentation patterns, and could be detected by neither

precursor ion (precursor of 188) nor by natural loss scan (loss of m/z 170) based on the

SMART-H’s fragmentation pattern. The metabolite pathway of SMART-H is

summarized in Fig 4.4.

4.3.5 Species specific metabolism of SMART-H

We determined the metabolite kinetics of SMART-H (0.5 μM) in the presence of liver

microsomes from mouse, rat, dog, and human. The parent, SMART-H, and the four

metabolites identified in human liver microsomes were simultaneously monitored in each

sample (Fig 4.5). The half-lives of SMART-H in microsomes from different species

varied over a broad range from < 5 min to 30 min (Fig 4.5 and Table 4.1). Interestingly,

we found that the metabolite arising from ketone-reduction (M1) was dominant only in

human liver microsomes. We then increased the SMART-H concentration from 0.5 μM

to 50 μM and incubated with 1 mg/mL human and rat liver microsomes to saturate the

enzymes and maximize the metabolite signals (Fig 4.6). The demethylated metabolite,

M2, was found in microsomal incubation using both rat and human liver microsomes. A

significant difference in M1 and M3 levels was observed between the two species. In rat

liver microsomes (Fig 4.6B), the primary metabolite was the demethylated metabolite

(M2). M1 could be found in rat liver microsomes only when a higher concentration of

SMART-H was used. However, the amount of M1 in rat liver microsomes was much less

than that in human liver microsomes (Fig 4.6). In addition, the relative amount of the demethylated metabolite (M2) in rat liver microsomes was greater than that in the human liver microsomes. These studies confirmed that M1 was a dominant metabolite only in

human liver microsomes.

4.3.6 Blockage of soft spots of SMART-H to increase the metabolic stability in human

liver microsomes

The two major metabolites in human liver microsomes, M1 (ketone reduction) and M2

(demethylation), identified the carbonyl and methoxy groups as the most labile sites (i.e.,

soft spots) to metabolic conversion. We thus designed the next generation of SMART

compounds to include substitutions aimed at improving the metabolic stability of these

soft spots. Three compounds (SMART-329, 173A, and 176A) with modified or direct

linkage between the trimethoxy phenyl ring and thiazole ring were made and tested while

a single compound (SMART-213) with a pentafluoro phenyl ring was made to replace

the trimethoxy substituents (Fig 4.7). The inclusion of a penta-fluoro phenyl ring

(SMART-213) failed to increase metabolic stability in human liver microsomes (Fig 4.8),

suggesting that demethylation may not be critical for metabolic stability or that SMART-

213 was susceptible to other new metabolic pathways. We also examined the metabolic stability of analogs containing electron-withdrawing substituents such as 4-fluorophenyl

or 3, 4, 5-trifluorophenyl to replace the trimethoxyphenyl ring. All of these derivatives

failed to increase the metabolic stability (data not shown). A dramatic increase in the metabolic stability was observed with compound SMART-176A which contains C=N-

NH2 instead of C=O found in SMART-H. This compound exhibited three-fold longer

stability than that of our lead compound, SMART-H, in human liver microsomes.

SMART-173A and SMART-329 also demonstrated an increase in metabolic stability by

2-fold. These data indicate that enhanced metabolic stability observed in human liver

microsomes correlated with the modification of the labile ketone site of SMART-H.

Table 4.3 shows IC50 values of SMART-213, -173A, -176A, and -329 for anti-

proliferative effects on four prostate cancer cell lines, LNCaP, PC-3, Du145, and PPC-1.

SMART-173A retained potent activity with an average IC50 value of 143 nM. SMART-

176A had less potent ability (IC50 value of 1250 nM) to inhibit cell proliferation. With the

removal of the labile functional group (C=O), SMART-329 demonstrated increased metabolic stability; however, it lost the ability to inhibit cell growth (IC50 value was >

10,000 nM). SMART-213 failed to either inhibit cell growth or improve metabolic

stability. These results indicate that blocking ketone-reduction increased metabolic

stability in human liver microsomes, but that in vitro anticancer activity could only be

partially retained by substituting with the oxime (SMART-173A).

4.4. Discussion

In vitro microsomal stability assays represent a high-throughput method to predict in

vivo hepatic clearance. Here, we used mice, rats, and dog to study in vitro-in vivo

relationship of SMART-H metabolism. The in vitro metabolic stability of SMART-H was extremely poor in mouse liver microsomes, resulting in the high clearance value obtained during in vivo pharmacokinetic studies in mice. The in vivo total clearance in rats was 5.3 ml/min/kg, which is close to our prediction based on in vitro microsomal stability studies.

The in vivo total clearance in dog was 2.7 ml/min/kg, which was somewhat smaller than our prediction. In mouse, rat, and dog pharmacokinetic studies, the distributional kinetics, not the long terminal phase, seem to define most of the AUC (Fig 4.2). The in vitro value for dogs may have been overestimated due to errors in the scaling factor or determination of fraction unbound in plasma and liver microsomes, while the predicted clearance value for mice slightly underestimated the in vivo clearance apparently due to extra-hepatic drug clearance. The in vivo clearance values of rat and dog were lower than expected based on their in vitro metabolic stability without considering protein binding. These low clearance data are reasonable when considering the high protein binding of SMART-H, demonstrating that high protein binding of SMART-H reduced its in vivo clearance.

However, high protein binding did not preclude high clearance when the metabolic stability was poor as seen in mice.

Although SMART-H demonstrated reasonable metabolic stability during incubation with rat, dog and human liver microsomes, notable differences in the relative quantities and identity of the metabolites formed in different species were observed (Figure 6). As such differences would complicate future toxicological assessment and inter-species comparisons, we sought to define these differences and evaluate structurally optimized

SMART compounds with improved metabolic stability and potent anticancer activity.

In metabolite identification studies in rat liver microsomes, the O-demethylated metabolite (M2) was the major metabolite. M2 was found in a time-dependent manner when 50 μM substrate was used. The level of O-demethylated metabolite was much greater than that of ketone reduced (M1) metabolite. This phenomenon was confirmed by in vivo experiments. In rat PK (dosed 2.5 mg/kg i.v.) studies, we compared the metabolite levels in plasma samples (see supplemental data). Only the O-demethylated (M2) and ketone reduced metabolites (M1) could be detected in rat plasma. The M2 metabolite exhibited about 10-fold higher exposure than M1 metabolite in rat plasma. These data further corroborate the conclusion that in vitro metabolic stability is a good predictor of in vivo pharmacokinetics.

Our ultimate goal is to develop an anti-cancer drug for use in humans with appropriate

PK properties. In this study, SMART-H exhibited a half-life of 17 min in human liver microsomes, suggesting that additional measures could be taken to improve its metabolic stability. Interestingly, the metabolic kinetic profiles in rat and human suggested that ketone reduction was an important metabolic pathway in humans than in rats, due to the apparent presence of M1 in humans (Fig 4.6). We speculated that modifying the ketone functional group would improve the metabolic stability in human liver microsomes.

Blocking or removing labile sites is a common strategy to increase metabolic stability. In this study, we synthesized three analogs (SMART-173A, -176A, and -329) to prevent the ketone-reduction reaction. The half-lives were longer (35, 55, and 32 min, respectively) in human liver microsomes, indicating that the ketone site is the critical soft spot for improving metabolic stability in humans. The in vitro metabolic stability of SMART-

176A, a hydrazine derivative, was not tested using conditions suitable for acetylation 84

(perhaps resulting in an over-estimate of its stability), but failed to retain anticancer

activity. SMART-173A retained its potency, with an average IC50 value of 143 nM in

four prostate cancer cell lines.

In summary, the studies presented here predict that SMART-H will exhibit low

clearance in humans based on in vitro data. However, our studies also suggest that the

metabolic stability of SMART-H and its analogs can be further optimized to obtain prolonged half-life and increased exposure in humans. Metabolism studies demonstrate that SMART-H was rapidly metabolized under phase I reactions in human liver

microsomes. Ketone-reduction and O-demethylation reactions were the primary

metabolic pathways for SMART-H. The metabolic stability (half-life) was successfully

improved 2-3 times in vitro, by blocking the ketone group. Although SMART-173A, the

oxime derivative with C=N-OH instead of C=O of SMART-H, demonstrated greater

metabolic stability, the in vitro potency was compromised as evidenced by a nearly equal

increase in IC50 in prostate cancer cell lines. Studies to identify additional analogs with

improved metabolic stability that retain high potency are ongoing in our laboratory.

4.5. Acknowledgements

Thank Terrence A. Costello, Katie N. Kail and Stacey L. Barnett for providing

technical support for animal studies at GTx Inc.

120

100

40 % remaining

0 20406080100120

minutes

Figure 4.1 Phase I metabolic stability of SMART-H in human liver microsome in the presence of NADPH (•) and in the absence of NADPH (○). Phase I + II metabolic stability in the presence of NADPH (d). Values represent the mean ± SD of triplicates.

Figure 4.2 Pharmacokinetic studies of SMART-H in mouse (A), rat (B) and dog (C). Single intravenous bolus administration was given with 15, 2.5, and 0.25 mg/kg for mouse (n = 3), rat (n = 4), and dog (n = 1), respectively.

Figure 4.3 Metabolite identification. 50 μM SMART-H was incubated with 1 mg/mL human liver microsomes for 2 hr. Chromatography (A). MS/MS spectrum of parent SMART-H (B), and four major MS/MS spectrum were identified in M1 (C), M2 (D), M3 (E), and M4 (F).

Figure 4.4 Metabolite profile of SMART-H in human liver microsome.

Figure 4.5 Metabolite profile in species. 0.5 μM substrate was incubated with 1 mg/mL microsomal protein of human (A), rat (B), mice (C), and dog (D). Four metabolites with the parent were measured by LC-MS/MS (n=3).

Figure 4.6 Kinetics of four metabolites of SMART-H in human liver microsome (A) and rat liver microsome (B). 50 μM of SMART-H was incubated with 1 mg/mL microsomal proteins. M1 (♦), M2 (○), M3 (□), and M4 (▲).

Figure 4.7 Blockage of soft spots.

Figure 4.8 Metabolic stability of SMART-H (•), -213 (○), -173A (▵), -176A (d), and - 329 (■) in human liver microsome.

NA, not available.

Table 4.1 Prediction of in vivo hepatic clearance of SMART-H in mouse, rat, dog, and human from in vitro data. Hepatic blood flow values were referenced from (56)

Table 4.2 Pharmacokinetic parameters of SMART-H in mouse, rat, and dog. The half-life was represented as harmonic mean ± pseudo SD in rat.

Table 4.3 Anti-proliferation of SMART compounds on prostate cancer cell lines (n = 3).

CHAPTER 5

2-ARYLTHIAZOLIDINE-4-CARBOXYLIC ACID AMIDES (ATCAA) TARGETS

DUAL PATHWAYS IN CANCER CELLS: 5’-AMP-ACTICATED PROTEIN

KINASE (AMPK)/MTOR AND PI3K/AKT/MTOR PATHWAYS

5.1 Introduction

The phosphatidylinositol 3-kinase/Akt (PI3K/Akt) pathway regulates many cellular functions and is aberrantly activated in many human malignancies. Eight isoforms of

PI3K exist, which are divided into class I, class II, and class III PI3Ks. Class I PI3Ks have drawn the most attention due to their roles in cell survival (84). Class I PI3K is further divided into classes Ia and Ib; with class Ia activated by receptor tyrosine kinases and class Ib activated by G-protein-couple receptors (GPCRs). Once activated, class I

PI3Ks rapidly phosphorylate phosphatidylinositol (4, 5)-bisphosphate (PIP2) to phosphatidylinositol-3, 4, 5-trisphosphate (PIP3). Signaling proteins, such as PDK1 and

Akt, bind to PIP3 via their Pleckstrin Homology domain (PH domain) with subsequent phosphorylation of Akt by PDK1 or other kinases (85, 86). PI3K can be inactivated by the tumor suppressor protein, phosphatase and tensin homolog (PTEN), which converts

PIP3 back to PIP2, hampering the lipid signaling. Loss or mutation of PTEN constitutively activates Akt leading to increased cellular proliferation and cancer (87). 97

Thus, inhibition of PI3K/Akt signaling provides a promising strategy to prevent and treat cancers, such as prostate cancer (88). There are many substrates of Akt involved in various physiological roles. While glycogen synthase kinase 3 (GSK3) regulates glucose metabolism and cell-cycle-regulatory proteins (89, 90), mTOR, another substrate, inhibits tuberous sclerosis complex 2 (TSC2), resulting in increased mTORC1 activity (91).

Another proliferative protein, S6 ribosomal protein functions downstream of mTOR.

Inhibition of the PI3K/Akt/mTOR signaling cascade is therefore viewed as a logical means to inhibit cell growth and treat cancer (92, 93).

AMP-activated protein kinase (AMPK), a serine/threonine kinase, consists of a heterotrimeric complex containing a catalytic α subunit and regulatory β and γ subunits

(94). AMP binds to AMPK allosterically to facilitate phosphorylation on Thr-172, an effect that is mediated by the tumor suppressor protein, LKB1 (95). Previous studies have shown that the allosteric activation of AMPK is antagonized by ATP, suggesting that both AMP and ATP bind to the same site (96) and that AMPK can be activated by ATP depletion. Overall, activation of AMPK is associated with an increased intracellular

AMP/ATP ratio (97). AMPK activation, by synthetic molecules, such as 2-deoxyglucose

(2DG) or 5-aminoimidazole-4-carboxyamide ribonucleotide (AICAR), has been shown to activate TSC2, which in turn inhibits the downstream target of mTOR, S6K (98, 99).

Activated AMPK also positively regulates p53 and p21, which are important proteins that control growth arrest and apoptosis. In addition, AMPK regulates acetyl-CoA carboxylase (ACC) and glycerol phosphate acyltransferase for fatty acid and triglyceride synthesis, respectively (100). mTOR is the downstream target of the PI3K/Akt and

AMPK pathways, suggesting that both PI3K/Akt/mTOR and AMPK/mTOR pathways

are potential therapeutic targets for cancer (101, 102).

We previously studied the structure-activity relationship (SAR) of a series of 2-

arylthiazolidine-4-carboxylic acid amides (ATCAA) in prostate and melanoma cancer

cells (103). ATCAA-10, our most potent compound, exhibited sub-micromolar IC50 values in cytotoxicity assays and induced severe sub-G1 phase arrest. In this study, we demonstrate that ATCAA-10 exhibited dual effects on PI3K/Akt/mTOR and

AMPK/mTOR pathways. ATCAA-10 inactivated PI3K/Akt and activated AMPK, resulting in synergistic inactivation of mTOR/S6K. We also studied the effects of

ATCAA-10 on the upstream and downstream pathways of AMPK. Pharmacokinetics, toxicity and in vivo efficacy (human lung cancer A549 cells) were also examined in this report.

5.2 Materials and Methods

5.2.1 Cell culture and cytotoxicity assay

Cell lines were obtained from ATCC (American Type Culture Collection,

Manassas,VA, USA), while cell culture supplies were purchased from Cellgro Mediatech

(Herndon, VA, USA). We examined the antiproliferative activity of ATCAA-10 in four

human prostate cancer cell lines (LNCaP, DU 145, PC-3, and PPC-1), two human lung

cell lines (A549 and H1299) and one cervical cancer cell line (HeLa). Human sarcoma cell line MES-SA and its doxorubicin-resistant cell line that over-expresses P-

glycoprotein (P-gp), MES-SA/Dx 5, were used as MDR models. All prostate cancer cell

lines were cultured in RPMI 1640 medium. Lung cancer cells and sarcoma cells were cultured in Dulbecco's Modified Eagle Medium (DMEM). HeLa cells were grown in

Eagle's MEM medium. MES-SA and MES-SA/Dx 5 cells were grown in McCoy's 5A medium. All cell media were supplemented with 10% fetal bovine serum (FBS). The cytotoxic potential of ATCAA-10 was evaluated using sulforhodamine B (SRB) assay after 96 h after treatment. Percentage of cell survival was plotted against drug concentrations and the IC50 values (concentration that inhibited cell growth by 50% of

untreated control) were obtained by nonlinear regression analysis using WinNonlin

(Pharsight Corporation, Mountain View, CA).

5.2.2 Cell cycle analysis

Flow cytometry was performed to study the effects of the ATCAA-10 on cell cycle distribution. PC-3 and A375 cells were plated in growth media at 70% confluence.

Medium was changed to 0.5% charcoal-stripped FBS (cs-FBS) for 48 h before the

treatment to synchronize the cells in G0/G1 phase of the cell cycle. The cells were then

treated in growth media with the indicated concentrations of ATCAA-10 for 24 h.

Cellular DNA was stained with 100 μg/mL propidium iodide and 100 μg/mL RNase A in

PBS and flow cytometry was performed to determine the cell cycle distribution of the cells.

100

5.2.3 FLIPR Intracellular Calcium Mobilization Assays

Chem-1 cells (Chemicon Int Inc., Billerica, MA) were stable-transfected with LPA1,

LPA3, and LPA5, respectively. Cells are maintained in DMEM/10 % fetal bovine serum/100 U/mL penicillin. Cells were passaged by washing with Ca2+ and Mg2+-free

HBSS (10 mL/T75), incubating with 0.05 % trypsin/0.2 g/L EDTA (1 mL/T75) for 5-10 minutes at 37°C. Cells are loaded with Fluo-4 NW and assay was read for 180 seconds

TETRA 2+ using the FLIPR . The signal is proportional to [Ca ]i and thus can be used to

quantify differences in calcium mobilization in ATCAA-10-treated and control cells.

Percentage activations were determined upon initial addition of ATCAA-10 followed by

10 minute incubation at 25ºC. To study antagonism of ATCAA-10 on LPAs, cells were incubated with ATCAA-10 for 10 min at 25ºC. Following compound incubation, reference agonists (Oleoyl-LPA) were added at EC80 to determine percentage inhibition.

ATCAA-10 was plated in an eight-point, four-fold serial dilution series with a top

concentration of 10 μM. The concentration described here reflects the final concentration

of the compound during the antagonist assay.

5.2.4 Cell treatment and immunoblotting

A549 and HeLa cells were grown in DMEM and Eagle's MEM medium containing

10% FBS, respectively, in humidified, 37 °C chambers with 5% CO2. A549 cells were

treated with 10 μM ATCAA-10 for 2, 4 and 6 hr. Varying concentrations of ATCAA-10

were used to treat A549 and HeLa cells for 6 and 4 hr, respectively. Cells were washed

twice by cold phosphate-buffered saline (PBS) and lysed in lysis buffer [20 mM Tris-HCl 101

(pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na3VO4, 1 µg/ml leupeptin]

incubated on ice for 30 min. The lysates were centrifuged at 14,000 × g for 20 min and

supernatants were collected. Proteins (50 μg) were resolved in a 4-20% gradient Tris-

Glycine gel, and transferred to nitrocellulose. Blots were probed with antibodies to

phospho-Thr172-AMPK, total-AMPK, phospho-Ser473-AKT, total-AKT, phospho-Ser9-

GSK3β, total-GSK3β, phospho- Ser235/236-S6 ribosomal protein, or total-S6 ribosomal

protein. All antibodies were purchased from Cell Signaling (Beverly, MA). Immunoblots

were developed using ECL Plex CyDye conjugated with goat anti-rabbit IgG-Cy5 (GE

Healthcare, Waukesha, WI), followed by detection with fluorescence imaging system

(Typhoon 8600, GE Healthcare, Waukesha, WI).

5.2.5 Nucleotide extraction and FPLC analysis

HeLa cells were treated with vehicle, ATCAA-10 (10 μM) and PD98059 (100 μM)

for 4hr. Following treatments, cells were washed quickly in PBS and the cell pellets were

obtained by centrifugation at 14,000 × g for 20 min. Perchloric acid (HClO4, 400 μL of

0.2M) was added and incubated on ice for 10 min. KOH (100 μL of 1N) was added to neutralize the pH to about 7.0. Macromolecules were removed using an ultrafiltration method (microconcentrator) (Microcon, Bedford, MA) with a molecular cutoff size of

30k Da. Nucleotides were separated by ion-exchange chromatography on a Mono Q 5/50

GL column run on Monitor UPC-900 system (GE Healthcare, Waukesha, WI). The column was equilibrated with 10 mM Tris-HCl, pH 8.0, and developed with a linear

102

gradient to 10 mM Tris-HCl with 1.5 M NaCl at a flow rate 1 mL/min. Nucleotides were

detected by their absorance at 254 nm. Peak heights were used to quantify and calculate

the AMP: ATP ratios.

5.2.6 AMPK activity

Active AMPK (α1/β1/γ1) enzyme and commercial AMPK kinase assay kit were

purchased from CycLex Co., Ltd (Nagano Japan) for measuring AMPK activity. Pure

AMPK enzyme (40 ng) was incubated with ATCAA-10 (10 μM) in kinase buffer at 30°C for 40 min. AMP (10 μM) was used as the positive control. The absorbance at wavelength 450 nm was used to represent the AMPK activity compared to the vehicle

control.

5.2.7 Cell-free purified enzyme activity assay

ATCAA-10 (20 μM) was used to examine the inhibition of enzyme activity. Potency

(IC50) was determined using full concentration range of ATCAA-10. Human receptor

tyrosines kinases (RTKs), including IGF-1R, EGFR, FGFR, and PDGFR, were

examined. DMSO (0.5 %) was used as negative control, and staurosporine was used as

positive control for all RTKs. Each RTK (h) (10 mU) was incubated with 8 mM 3-(N-

morpholino)propanesulfonic acid (MOPS), 0.2 mM EDTA, 10 mM MnCl2, 0.1 mg/mL

poly(Glu:Tyr, 4:1), 10 mM MgAcetate, [γ-33P-ATP] (activity at 500 cpm/pmol,

concentration as required), and ATCAA-10. The reaction was initiated by adding

Mg/ATP mix in a final reaction volume of 25 μL. After 40 minutes incubation at room 103

temperature, the reaction was stopped by adding 5 μL of 3% phosphoric acid solution.

Ten microliters of the reaction was then spotted onto a Filtermat A and washed three times for 5 minutes in 75 mM phosphoric acid and once in methanol prior to drying and scintillation counting. Percentage of kinase activities were compared with vehicle.

5.2.8 Pharmacokinetic study

All animal studies were conducted under the auspices of a protocol reviewed and approved by the Institutional Laboratory Animal Care and Use Committee of either The

University of Tennessee or The Ohio State University. Male ICR mice (n = 3, each time point) 6 to 8 weeks of age were purchased from Harlan Inc. (Indianapolis, IN). ATCAA-

10 (10 mg/kg) was dissolved in PEG300/DMSO (1/4) and administered by a single i.v. injection into their tail vein. Blood samples were collected in heparinized tubes via cardiac puncture under isoflurane anesthesia at 2, 5, 15, and 30 min, 1, 2, 4, 8, 16, and 24 hr after administration. Plasma samples were collected by centrifuge at 8,000 g for 5 min.

All plasma samples were stored immediately at -80°C until analyzed.

Male Sprague-Dawley rats (n = 4; 254 ± 4 g) were purchased from Harlan Inc.

(Indianapolis, IN). Rat thoracic jugular vein catheters (JVC) were purchased from

Braintree Scientific Inc. (Braintree, MA). On arrival to the animal facility, the animals were acclimated for 3 days in a temperature-controlled room (20–22°C) with a 12-h light/dark cycle before any treatment. ATCAA-10 was administered intravenously into the JVC at 2.5 mg/kg (in DMSO/PFG300, 1/4). Equal volume of heparinized saline was injected to replace the removed blood and rinse the catheter. Blood samples (250 μL)

104

were collected via the JVC at 10, 20, 30 min, and 1, 2, 4, 8, 12, 24hr. Plasma samples

were obtained as previously described. An aliquot of plasma was used to measure plasma

insulin levels using the rat insulin ELISA kit (Crystal Chem Inc., Downers Grove, IL)

following the manufacturer’s instructions. Whole blood samples (5 μL) were also

collected and used to measure blood glucose levels using a commercial glucose meter

(ACCU-CHEK, Hamilton, New Zealand)

ATCAA-10 was extracted from 100 μL of plasma with 200 μL of acetonitrile (ACN) containing 100 nM internal standard (ATCAA analog). The samples were thoroughly mixed, centrifuged, and the organic extract was transferred to autosampler for LC-

MS/MS analysis. The PK parameters were determined using non-compartment analysis

(WinNonlin, Pharsight Corporation, USA)

5.2.9 LC-MS/MS Analytical method

Aliquots (10 μL) of the supernatant were injected into the HPLC system (Model 1100

Series Chemstation, Agilent Technology Co, Santa Clara, CA). A narrow-bore C4

column (2.1×150 mm, 5 μm, Varian Inc, Palo Alto, CA) was used to separate ATCAA-

10 from plasma matrix. Gradient mode was used to achieve separation of the analyte

using mixtures of mobile phase A (5%/95% acetonitrile/H2O containing 0.1% formic

acid) and mobile phase B (95%/5% acetonitrile/H2O containing 0.1% formic acid) at a

flow rate of 300 μL/min. Mobile phase A was used at 80% from 0 to 1 min followed by a

linearly programmed gradient to 100% of mobile phase B within 3 min, 100% of mobile

105

phase B was maintained for 1 min before a quick ramp to 80% mobile phase A. Mobile

phase A was continued for another 8 min towards the end of analysis.

A triple-quadruple mass spectrometer (API QtrapTM Applied Biosystems/MDS

SCIEX, Concord, Ontario, Canada) operating with a TurboIonSpray source was used.

Multiple reaction monitoring (MRM) mode, scanning m/z 523.0 → 196.3 (ATCAA-10)

and m/z 434.0 → 266.0 (IS) was used for quantitation.

5.2.10 A549 tumor xenograft studies

A549 cells (8×107 per ml) were prepared in phenol red-free growth media containing

10% FBS, and mixed with Matrigel (BD Biosciences, San Jose, CA) at 1:1 ratio. Tumors

were established by injecting 100 μL of the mixture (4×106 cells per animal)

subcutaneously (s.c.) into the flank of 6-8-week-old male athymic nude mice. Length and

width of tumors were measured and the tumor volume (mm3) was calculated by the

formula, π/6 × L × W2, where length (L) and width (W) were determined in mm. When

the tumor volumes reached 200 mm3, the animals bearing A549 tumors were treated with vehicle [Captex200/Tween80 (1/4)], ATCAA-10 (5, 10, 15 and 20 mg/kg)

intraperitorally twice a week for 31 days.

5.2.11 Blood glucose and plasma insulin

When collected whole blood samples from rat PK study, 5 μL of whole blood was

used to measure the blood glucose levels by glucose meter. The rat plasma PK plasma

samples as previously described were used to measure plasma insulin levels. Insulin 106

levels were determined by the rat insulin ELISA kit (Crystal Chem Inc., Downers Grove,

IL) following the manufacturer’s instruction.

5.3 Results

5.3.1 ATCAA-10 inhibits the growth of human cancer cell lines and multidrug-resistant cancer cell lines

The ability of ATCAA-10 (Fig 5.1A) to inhibit the growth of cancer cell lines was

evaluated using the SRB assay. As shown in Table 5.1, ATCAA-10 inhibited the growth

of several human cancer cell lines, including four prostate cancer cell lines and two lung

cancer cell lines, with IC50 values in the low micro-molar range. In addition, the effect of

ATCAA-10 in the doxorubicin resistant cell line MES-SA/Dx5 was also evaluated (Table

5.1). ATCAA-10 exhibited a smaller resistance factor (1.7-fold) compared to the positive

control, vinblastine, a P-gp substrate (9.3-fold). In addition, the slope of dose-response

curve varied for vinblastine, but not for ATCAA-10 (data not shown). These data indicate

that ATCAA-10 may overcome P-gp mediated drug resistance.

5.3.2 ATCAA-10 induces sub-G1 phase in LNCaP and A375 cell lines

To examine the effect of the new synthetics on cell cycle progression, flow cytometric

analysis was performed on human prostate cancer LNCaP cells and human melanoma

A375 cells. Analysis of the DNA content of control cells (Fig 5.2) showed typical

distribution of peaks corresponding to cells in G1/G0 phase (66-67 %), S-phase (24-28

107

%) and G2/M-phase (6.0-8.9 %). After treatments of tumor cells with compounds

ATCAA-10 at 1, 5, 10 and 20 μM concertration, flow cytometric analysis showed that tested compounds significantly induced apoptosis (sub-G1 phase) in the two tumor cell lines in a dose-dependent manner. On LNCaP and A375 cells, sub-G1 phase started to increase at 5 μM of ATCAA-10 treatment groups.

5.3.3 ATCAA-10 did not target LPA receptors

The initial design of ATCAA was derived from contriving mimics of lysophosphatidic

acid (LPA), a lipid mediator generated via the regulated breakdown of membrane

phospholipids that are known to stimulate G-protein-coupled receptors (GPCR) signaling.

GPCR activation could be detected through changes in intracellular calcium

concentrations. To examine if ATCAA compounds inhibit cancer cell growth via GPCR

pathway, A FLIPR assay (Ca2+ flux response) was conducted to screen ATCAA-10 for

dose dependent agonist and antagonist activities on the LPA1, LPA3, and LPA5-

transfected Chem-1 cells. Oleoyl-LPA and a known LPA antagonist compound Ki16425

(104) were used as a positive control and negative control, respectively. In agonist assay

mode, Oleoyl-LPA provided EC50 values of 31 nM, 120 nM, and 12 nM for LPA1,

LPA3, and LPA5, respectively. 15% activation will be considered as an agonist when

referencing Emax of Oleoyl-LPA. However, ATCAA-10 did not exhibit any dose-

dependent agonism on LPA1, LPA3 or LPA5 (Fig 5.3). We also tested antagonism of

ATCAA-10, and 50% inhibition will be considered as an antagonist when referencing

EC80. The negative control, Ki16425, has an IC50 value of 46 nM on LPA1. However,

108

ATCAA-10 did not exhibit any dose-dependent antagonist activity on the LPA1, LPA3 or LPA5 (Fig 5.4). The FLIPR assay results suggest that ATCAA-10 did not display cytotoxicity due to antagonism of the LPA receptors. ATCAA compounds exhibit cytotoxicity though other mechanisms.

5.3.4 ATCAA-10 inhibits Akt phosphorylation and activates AMP-activated Protein

Kinase (AMPK)

We then examined the effect of ATCAA-10 on the PI3K/Akt pathway. Fig 5.1B shows that ATCAA-10 (10 μM) decreased phospho-Akt levels without changing total-

Akt levels in A549 cells in a time-dependent manner. At 6 hr, the ratio of phospho-Akt to total-Akt was lower compared to the control. Simultaneously phospho-AMPK was induced by ATCAA-10 at 6 hr (Fig 5.1B), suggesting that ATCAA-10 impinges on both

PI3K/Akt and AMPK pathways. The treatment time (6 hr) was fixed for future dose- response studies and the dose-dependent effects of ATCAA-10 were examined (Fig 5.1C) in A549 cells.. Phosphorylation of a downstream target of Akt, phospho-Ser235/236-S6 ribosomal protein, was also greatly reduced following ATCAA-10 treatment (Fig 5.1C).

The total protein level of S6 ribosomal protein remained unchanged. Modulation of phospho-Ser9-GSK3β, another downstream target of Akt, was also examined in A549 cells; however, the signal was under the detection limit (data not shown).

HeLa cells, which are known to be LKB1-null, were then used to confirm the mechanism of action of ATCAA-10 via PI3K/Akt and AMPK pathways. As shown in Fig

5.1D, ATCAA-10 exhibited the same effect on Akt and AMPK in HeLa cells as observed

109

in A549 cells, with phosphor AMPK levels induced more than 2-fold and a concomitant decrease in phosphor-Akt levels. Phospho-Ser9-GSK3β and phospho-Ser235/236-S6 ribosomal protein levels were also decreased in a dose-dependent manner, indicating that the effects of ATCAA-10 on Akt and AMPK were not cell-type dependent.

In previous studies, we demonstrated that ATCAA-10 failed to target LPA receptors

(103). In an attempt to identify the molecular target, we examined the interaction of

ATCAA-10 with several receptor tyrosine kinases (RTKs), including IGF-1R, EGFR,

FGFR and PDGFR in this study. ATCAA-10 exhibited IC50 values more than 20 μM for all these kinases (Table 5.2), suggesting that ATCAA-10 did not appreciably interact with these potential targets. ATCAA-10 did not target PDK1, which is the direct upstream of

Akt. In addition, ATCAA-10 did not reduce activities on PI3K isoforms (β, γ, δ) and Akt isoforms (α, β, γ) (Table 5.2). The major target causing deactivation of Akt by ATCAA-

10 remains unclear.

5.3.5 ATCAA-10 activates AMPK by decreasing intracellular AMP/ATP ratio

An FPLC method was developed to measure intracellular AMP, ADP and ATP levels

(Fig 5.5A). ATCAA-10 (10 μM) treatment reduced AMP and ATP levels, while ADP levels remained the same compared to the control (Fig 5.5B). ATCAA-10 reduced ATP levels more than AMP levels, resulting in an increase in the ratio of intracellular

AMP/ATP (Fig 5.5C). PD98059 was used as a positive control; however, PD98059 increased the ratio of AMP/ATP by reducing ATP levels without altering the AMP levels. Pure and active AMPK enzyme was used to understand the direct effect of

110

ATCAA-10 on AMPK, if any. Interestingly, ATCAA-10 failed to stimulate AMPK (Fig

5.5D), suggesting that ATCAA-10 induces phosphorylation of AMPK indirectly by increasing the ratio of intracellular AMP/ATP, but not by directly stimulating AMPK.

5.3.6 Pharmacokinetic study (PK) of ATCAA-10 in mice and rats

Since ATCAA-10 had an interesting in vitro pharmacologic profile, we determined its

PK properties before proceeding to tumor xenograft studies. The plasma concentration vs. time profiles in mice and rats given a single i.v. bolus dose of 10 mg/kg ATCAA-10 are shown in Fig 5.6. The PK parameters of ATCAA-10 in mice and rats are summarized in Table 5.3, indicating that ATCAA-10 exhibited favorable PK properties in both mice and rats. In mice, ATCAA-10 had a long elimination half-life (311 min), moderate volume of distribution at steady state (2.9 L/kg), but high total plasma clearance (34 ml/min/kg, hepatic extraction rate is 0.68, if the hepatic plasma flow rate is 50 ml/min/kg in mice). The plasma concentration vs. time profile was also examined in mice given single dose by intraperitoneal (i.p.) route (Fig 5.6A). Compared to the area under the curve (AUC) by i.v. administration, the bioavailability of i.p. administration was 93%, indicating that the i.p. route provided an efficient and easy method of administration for mice. Most importantly, the PK studies by i.p. injection also demonstrated that plasma concentrations above the in vitro effective concentration obtained in the cytotoxicity assays in A549 cancer cells could be achieved.

Similar to mice PK, ATCAA-10 also exhibited favorable PK properties in rats (Fig

5.6B), exhibiting a long half-life (477 min), moderate volume of distribution at steady

111

state (4.1 L/kg), and low clearance (8.3 ml/min/kg), with low hepatic extraction ratio

(ER= 0.28, if the hepatic plasma flow rate is 30 ml/min/kg in rats).

5.3.7 ATCAA-10 inhibited lung cancer A549 xenograft growth

A549 xenografts were allowed to reach a volume of 200 mm3 and the tumor-bearing

mice were treated with ATCAA-10 by i.p. administration at the indicated doses. As

shown in Fig. 5.7A, tumor volumes in the control group increased to 1615 ± 244 mm3 over the 31 day study duration. ATCAA-10 elicited 46 % tumor growth inhibition (TGI) at 20 mg/kg treatment, decreasing tumor volumes to 968 ± 358 mm3. Body weight measurements (Fig 5.7B), to monitor toxicity, indicated that ATCAA-10 did not significantly reduce body weights at the effective dose.

5.3.8 ATCAA-10 did not cause hyperglycemia at a single dose in rats or repeated-doses in mice

Inhibitors affecting the PI3K/Akt/mTOR pathway are often associated with hyperglycemia and hyperinsulinemia due to their integral role in insulin signaling (105).

To test if hyperglycemia or hyperinsulinemia was caused by ATCAA-10, whole blood glucose and plasma insulin were measured when the rat PK studies were performed. Fig

5.8A shows that a single dose (10 mg/kg) of ATCAA-10 in rats did not cause a significant change in whole blood glucose levels. The same trend was also observed in plasma insulin levels (Fig 5.8B), suggesting that ATCAA-10 did not induce hyperglycemia and hyperinsulinemia by single dose in a short-term study (24 hr). Though 112

single dose failed to induce hyperglycemia, cancer patients are treated for a prolonged duration. Blood glucose levels in nude mice treated long-term with ATCAA-10 (twice a week, 31 days) were 194 ± 47, 198 ± 53, 199 ± 44, 211 ± 44, and 223 ± 61 mg/dl in control, 5, 10, 15, and 20 mg/kg, respectively. These data indicate that ATCAA-10 did not induce hyperglycemia after repeated doses over 31 days in mice.

5.4 Discussion

ATCAA-10 exhibited potent antitumor activities against a spectrum of cancer cell lines with IC50 values in the range of 0.4 to 1.8 μM (Table 5.1). Since resistance to prolonged treatment due to over-expression of P-gp is a major problem, studies were also conducted to determine whether ATCAA-10 is a substrate of P-gp. Our data and other publications demonstrated that the overexpression of P-gp in MES-SA/Dx5 results in a high resistance factor and varied dose-response curve for vinblastine (55). However,

ATCAA-10 showed similar activity in both MES-SA and MES-SA/Dx5 cell lines and did not cause a varied dose-response curve, suggesting that ATCAA-10 may circumvent P- gp-mediated drug resistance. These data suggest that resistance may not be a major problem for ATCAA-10.

We found that dephosphorylation of Akt by ATCAA-10 occurred not only in A549 and HeLa cells (in this study), but also in PTEN-mutated prostate cancer cell lines, such as LNCaP and PC-3 (data not shown). These data confirmed that ATCAA-10 consistently, regardsless of the cancer type, targets PI3K/Akt pathway. The mechanism for the dephosphorylation of Akt by ATCAA-10 was investigated in this study. In our

113

earlier studies (103), we speculated that ATCAA-10 target lysophosphatidic acid (LPA) receptors to inhibit Akt, as the ATCAA were originally designed as mimics of LPA.

However, ATCAA-10 did not target the LPA receptors leading us to examine other upstream RTKs and GPCRs. We excluded IGF-1R, EGFR, FGFR, and PDGFR as targets of ATCAA-10. Future studies will include more receptor tyrosine kinases and GPCRs.

PDK1 is the direct upstream activation of Akt, and ATCAA-10 did not inhibit PDK1.

The PI3K and Akt themselves were also tested, but ATCAA-10 failed to effectively inhibit their activities, suggesting that the target of ATCAA-10 may be obscure.

Hyperglycemia has been reported as an on-target side-effect for agents targeting the

PI3K/Akt pathway in animal models (105) and has also been observed in patients treated with triciribine phosphate (106). Although ATCAA-10 is involved in PI3K/Akt pathway, a hyperglycemic effect was not observed after a signal dose (10 mg/kg) of ATCAA-10 in rats or repeated dosing in mice (twice a week for 31 days). These data suggest that

ATCAA-10 may circumvent this side effect.

ATCAA-10 stimulated phospho-AMPK in both A549 and HeLa cell lines. We confirmed that ATCAA-10 induced phospho-AMPK by depleting ATP and changing the intracellular AMP/ATP ratio, but by directly targeting AMPK. The upstream of AMPK,

LKB1 kinase, can phosphate and activate AMPK. However, both A549 and HeLa cells are LKB1-null (107), indicating that ATCAA-10 stimulated phospho-AMPK independent of its upstream kinase LKB1. Since AMPK is considered as a negative regulator of proliferation that attenuates cancer, we hypothesize that part of ATCAA-10’s antiproliferative effects emanate from AMPK activation. AICAR and phenformin are often experimentally examined for stimulation of AMPK and subsequent inactivation of Akt 114

(108). However, the concentration of AICAR or phenformin needed to induce these effects is in millimolar range (i.e., 1000-fold higher than ATCAA-10). It remains controversial whether Akt regulates AMPK activity or vice versa. Recent evidence indicates that activated AMPK induces the expression of p21, p27, and p53 proteins and suppresses the PI3K/Akt pathway (109). Conversely, Akt is claimed to regulate the intracellular ATP levels and act as a negative regulator of AMPK (110). Both activation of AMPK and deactivation of Akt may contribute to tumor growth inhibition observed after treatment with ATCAA-10. In further studies, the use of Akt or AMPK antisense experiments combining ATCAA-10 treatment may help to elucidate the relationship between Akt and AMPK.

ATCAA-10 exhibited acceptable PK properties in mice and rats. By the i.p. route in mice, plasma concentrations exceeded 574 ng/mL (IC50 value in A549 cell line in vitro) for more than two hours at a dose of 10 mg/kg, providing proof that sufficient systemic exposure could be and was achieved during tumor xenograft studies in mice. Since

ATCAA-10 exhibited favorable PK properties in rats, we also tested the oral bioavailability of ATCAA-10. Unfortunately, the oral bioavailability was only 4.4% after a dose at 10 mg/kg in rats (data not shown), suggesting that ATCAA-10 may have low permeability, be susceptible to intestinal metabolism or have limited solubility.

Permeability studies using Caco-2 cells confirmed that ATCAA-10 has an extremely low

-6 permeability (Papp is << 1 × 10 cm/s, A→B, data not shown), suggesting further modifications on the chemical scaffold of ATCAA-10 will be necessary to achieve acceptable oral bioavailability.

115

These studies showed that ATCAA-10 inhibited the growth of A549 lung tumor

xenograft via inhibition of PI3K/Akt pathway and activation of AMPK pathway. We also found that ATCAA-10 exhibited acceptable PK properties for tumor growth inhibition. In efficacy, ATCAA-10 treatments were able to reduce the growth of tumor in animals in dose manner, and achieved 46% of TGI at 20 mg/kg group. These studies provide initial in vitro and in vivo proof of concept that ATCAAs or optimized derivatives thereof modulate the activity of the PI3K/Akt and AMPK pathways and hold promise as a novel approach to the treatment of drug-resistant cancer.

5.5 Acknowledgements

We thank Terrence A. Costello, Katie N. Kail and Stacey L. Barnett for providing technical support for animal studies at GTx Inc.

116

Figure 5.1 ATCAA-10 activated AMPK and deactivated PI3K/Akt pathways. The structure of ATCAA-10 (A). Effects of ATCAA-10 (10 μM) in A549 cells at indicated time points (B). Dose-response of ATCAA-10 in A549 cells after 6 hr treatment (C). Dose-response of ATCAA-10 in HeLa cells after 4 hr treatment. The number under blots indicated the ratio of phosphor-protein and total-protein related to the control (D).

117

LNCaP A375 Control Control

1 M 1 M

5 M 5 M

10 M 10 M

20 M 20 M

Figure 5.2 Flow cytometric analysis of ATCAA-10. Sub-G1 phase accumulation was induced by ATCAA-10 in LNCaP and A374 cells after 24h at the indicated concentrations.

118

LPA1 LPA3

Oleoyl-LPA LPA5 ATCAA-10 Ki16425

Compound ID EC50 on LPA1 EC50 on LPA3 EC50 on LPA5

Oleoyl-LPA 31nM 120nM 12nM ATCAA-10 - -- Ki16425 ---

Figure 5.3 FLIPR intracellular calcium mobilization assay in agonist mode. Calcium flux in LPA1, LPA3, LPA5-expressing Chem-1 cell lines induced by Oleoyl-LPA. Oleoyl- 2+ LPA increased [Ca ]i in a dose-dependent manner, ATCAA-10 and LPA antagonist Ki15425 did not show agonist effect.

119

LPA1 LPA3

LPA5 ATCAA-10 Ki16425

Compound ID IC50 on LPA1 IC50 on LPA3 IC50 on LPA5

Ki16425 46nM ATCAA-10 ---

Figure 5.4 FLIPR intracellular calcium mobilization assay in antagonist mode Response of ATCAA-10 and Oleoyl-LPA on LPA1, LPA3, and LPA5 kinetic data. Based upon these kinetic traces exhibited by ATCAA-10, the relatively small Ca2+ flux is non-LPA receptor mediated (A). Antagonist data of ATCAA-10, which did not show antagonism on LPA1, LPA3, and LPA5. Ki16425 had an IC50= 46 nM on LPA1 (B).

120

Figure 5.5 ATCAA-10 stimulated AMPK via changing the ratio of intracellular AMP/ATP. Chromatogram of AMP, ADP and ATP by using FPLC (A). Intracellular AMP, ADP and ATP levels of control, ATCAA-10 (10 μM), and PD98059 (100 μM) in HeLa cells (n = 3); bars, SD. PD98059 was used as a positive control (B). The ratio of intracellular AMP/ATP of ATCAA-10 based on Fig. 2B (C). AMPK enzyme activity assay in cell-free system. ATCAA-10 (10 μM) was used to test the induction of AMPK. AMP (10 μM) was used as a positive control (D).

121

Figure 5.6 Pharmacokinetic studies of ATCAA-10. Concentration-time curve of ATCAA-10 in ICR mice (A) (n = 3); bars, SD. ATCAA-10 was administrated 10 mg/kg by i.v.and i.p. Concentration-time curve of ATCAA-10 in SD rats (B) (n = 4); bars, SD SD rats were dosed 10 mg/kg i.v. Dash line indicated that cytotoxic IC50 value obtained in A549 cancer cell line in vitro. The IC50 value was 1.1 μM (574 ng/mL). 122

Figure 5.7 In vivo efficacy of ATCAA-10. Tumor cells were s.c. implanted in nude mice. ATCAA-10 was administrated after tumor formation (150-200 mm3, n = 6-8) Human lung A549 xenografts were treated with 5, 10, 15 and 20mg/kg i.p (A); bars, SE. Body weights were measured twice a week. The percentage of body weights were determined by comparing with the body weight on day 1 (B); bars, SD.

123

Figure 5.8 Hyperglycemia and hyperinsulineia test. Blood glucose levels of ATCAA-10 with 10 mg/kg i.v. in SD rats (A). Plasma insulin levels of ATCAA-10 treatment. Plasma insulin was determined using ELISA kit (B). Vehicle was DMSO/PEG300 (1/4). N = 4; bars, SD.

124

Table 5.1 Potency of ATCAA-10 in inhibiting cell growth in vitro. Cell growth inhibition was measured by SRB assay after 96-hr treatment. The numbers in parenthesis are folds of resistance factor based on the IC50 ratio when compared with the parent cell line (MES-SA). Values represent the mean ± SD of triplicates.

125

Table 5.2 Inhibition of pure enzyme by ATCAA-10. ATCAA-10 (20 μM) was used to test inhibition of enzyme activity in cell free system. Receptor tyrosine kinases were tested in duplicate, and mean values were used to represent. PDK1, PI3K and Akt were tested in triplicate (mean ± SD).

126

In vivo, pharmacokinetic parameters of ATCAA-10

Parameter AUC t1/2 Vss CL

Unit hr * mg/mL min L/kg mL/min/kg

Mice 4.9 311 2.9 34

Rats 17.3 477 4.1 8.3

Table 5.3 Pharmacokinetic parameters of ATCAA-10 (10 mg/kg) in mice and rats.

127

CHAPTER 6

SUMMARY AND DISCUSSION

Tubulin and microtubules exhibit complex dynamic instability and they are essential

for cell division. Disturbing their dynamic process can cause cell death in nanomolar

range and cell apoptosis. Current research focused on elucidating mechanism of action of our anti tubulin agents, characterizing their in vitro and in vivo anticancer efficacy, and developing an orally available antitubulin agent for chemotherapy. Key findings from this study are summaried below.

1. We proposed that (SMART-H) was our lead compound which exhibited potent

cytotoxicity and inhibited growth of a broad spectrum of cancer cells. The molecular

mechanism of action was studied in (see chapter 2), showing SMART-H bound to

colchicine binding site on tubulin, and inhibited tubulin polymerization, subsequently

induced apoptosis. The study also revealed that SMART-H exhibited different

metabolic stabilities in species in vitro using liver microsome systems. SMART-H

was relatively stable in rat, dog, and human liver microsomes compared to mouse

liver microsome although SMART-H. Therefore, in vivo pharmacokinetic studies

128

demonstrated that high clearance was obtained in mice, but low clearance in rats,

suggesting that mice model is a poor model to perform efficacy; however, it is the

easiest and gold standard to perform efficacy for anticancer drug. To overcome this

problem, the formulation (Tween80/Captex200, 1/4) was developed to improve

exposure via i.p. administration, and the data showed it could increase four-fold AUC

compared to the formulation with (DMSO/PEG300, 1/4). Frequent dosing with daily

i.p. injection was used to compensate the high clearance in mice. With the

improvement of formulation and dose schedule, SMART-H could be able to inhibit

tumor growth in prostate and melanoma tumor models. The data also indicated that

SMART-H is an active agent (T/C ≤ 42%) based on Nation Cancer Institutes

requirement. In addition, SMART-H did not show neurotoxicity by the effective

doses in rotarod assay in vivo. The drug resistance was also examined in this study.

We demonstrated that SMART-H may circumvent P-glycoprotein (Pgp) mediated

drug resistance in vitro, and may not developing drug resistance after a long term

treatment in vivo.

2. We also developed a novel mass spectrometry binding assay for determination of

tubulin binding site for small molecule inhibitors (see chapter 3). This methodology

is characterized by LC-MS/MS quantification of a non bound marker after separation

from target bound marker by ultrafiltration. Binding assays are performed by titration

of a marker, such as colchicine, vinblastine and paclitaxel in a fixed concentration

with increasing target concentrations. This way the (known) binding sites for

podophylltoxin, vincristine, and docetaxel could be confirmed. The method does not 129

require the use of radiolabeled ligands, can be applied to a wide variety of drugs that

either inhibit or promote tubulin polymerization, and allows for studies to define the

reversibility of the interaction with tubulin. SMART-H was applied and determined to

reversely bind to colchicine binding site on tubulin by this novel method.

3. We further studied in vitro and in vivo correlation of SMART-H (see chapter 4),

combining in vitro protein and liver microsome binding and metabolic stability in

liver microsomes to predict in vivo clearance in species. Overall, the in vitro data

provided valuable and reasonable predictions to predict in vivo clearance in many

species. Encouragingly, the prediction further suggested that humans will have a low

hepatic extraction ratio and low clearance. Metabolite identification was also

characterized in this study, and revealed that SMART-H exhibited unique ketone

reduction in human liver microsomes. Metabolites were characterized using LC-

MS/MS and used to identify labile sites in order to improve metabolic stability. We

could successfully be able to improve 2-3 times by blocking the labile site in human

liver microsomes. Interestingly, SMART-173A exhibited two-fold longer half life

than SMART-H, and it still remained great potency with an averaged IC50 value 143

nM in four prostate cancer cell lines. This study guided our design of second

generation of SMART compounds with greater metabolic stability and potency.

4. One of thiazolidine analog, namely ATCAA-10, was selected to examine its in vitro

and in vivo characterizes. First of all, LPA receptor, which was our previous

hypothesis, was excluded in this study. Based on the study we also excluded IGF-1R, 130

EGFR, FGFR, and PDGFR, PI3K, PDK1, and Akt as targets of ATCAA-10. We

found that ATCAA-10 involved in PI3K/Akt/mTOR and AMPK/mTOR pathways.

ATCAA-10 effectively dephosphorylated Akt, with concomitant phosphorylation of

AMPK. Determination of intracellular ATP and AMP concentrations revealed that

ATCAA-10 activated AMPK by altering the intracellular AMP/ATP ratio. By

pharmacokinetic studies, ATCAA-10 exhibited favorable pharmacokinetic properties

in both mice and rats, including low clearance, low hepatic extraction rate, moderate

volume of distribution and long half-life. In addition, ATCAA-10 inhibited A549

tumor xentograft growth with 46% tumor growth inhibition (TGI) at 20 mg/kg dose.

Perspective

The current antitubulin cancer drug, such as paclitaxel, exhibits high potency; however, the limitations of paclitaxel are low aqueous solubility, low bioavailability and the development of clinical resistance. Our SMART compounds have comparable potency with paclitaxel, but with a poorer substrate of P-glycoprotein, a known mediator of paclitaxel resistance. It remains to be determined if SMART exhibits great bioavailability for oral use.

Based on rat pharmacokinetic studies, SMART-H exhibited favorable pharmacokinetic properties with low clearance, suggesting that SMART-H may avoid hepatic first-pass effect in rats. However, SMART-H had a poor bioavailability (less than

5% F) with DMSO/PEG300 formulation. Although some formulations, including

Captex200/Tween80 (1/4) and Tween80/H2O (1/4), were examined, all of their bioavailability was still under 5%. We also performed Caco-2 assay to exclude 131

permeability of SMART-H. However, we found that SMART-H showed low recovery

rate in Caco2 assay, suggesting that SMART-H either stuck on the membrane or exhibited low solubility. Further solubility assay demonstrated that SMART-H only exhibited 1.1 μg/mL water-solubility. These data indicated that SMART-H had

solubility-limited absorption, resulting in a poor bioavailability. To improve

bioavailability of SMART-H, either develop a better formulation, or modify chemical

structure with a polar function group to improve solubility. However, the potency should

not be sacrificed.

We further introduced an imidazole ring instead of thiazole ring of SMART

compounds to be 2-aryl-4-benzoyl-imidazoles (ABI) compounds. ABI compounds

reduced log P values and increases aqueous solubility; in addition, they exhibited

comparable potency with SMART compounds. ABI-274 exhibited greater potency in

four prostate cancer cell lines. ABI-274 exhibited averaged cytotoxic IC50 value of 12

nM, and was selected to examined pharmacokinetic studies to see whether ABI

compounds could improve bioavailability. Encouragingly, ABI-274 exhibited near 25%

bioavailability at the dose of 10 mg/kg. ABI-274 exhibited higher clearance (29

mL/min/kg) compared to SMART-H (6 mL/min/kg), suggesting that the bioavailability

could be possibly refined by improving metabolic stability and reducing clearance. To

date, the metabolites were characterized by LC-MS/MS, and we proposed some ways to block labile sites to improve metabolic stability. Lead optimization of ABI-274 is underway in our laboratory.

132

REFERENCE

1. Jordan, M. A. and Wilson, L. Microtubules as a target for anticancer drugs. Nat Rev Cancer, 4: 253-265, 2004. 2. Jordan, M. A. and Kamath, K. How do microtubule-targeted drugs work? An overview. Curr Cancer Drug Targets, 7: 730-742, 2007. 3. Nogales, E., Whittaker, M., Milligan, R. A., and Downing, K. H. High-resolution model of the microtubule. Cell, 96: 79-88, 1999. 4. Desai, A. and Mitchison, T. J. Microtubule polymerization dynamics. Annu Rev Cell Dev Biol, 13: 83-117, 1997. 5. Akhmanova, A. and Hoogenraad, C. C. Microtubule plus-end-tracking proteins: mechanisms and functions. Curr Opin Cell Biol, 17: 47-54, 2005. 6. Jourdain, L., Curmi, P., Sobel, A., Pantaloni, D., and Carlier, M. F. Stathmin: a tubulin-sequestering protein which forms a ternary T2S complex with two tubulin molecules. Biochemistry, 36: 10817-10821, 1997. 7. Sharma, S., Ganesh, T., Kingston, D. G., and Bane, S. Promotion of tubulin assembly by poorly soluble taxol analogs. Anal Biochem, 360: 56-62, 2007. 8. Na, G. C. and Timasheff, S. N. Interaction of vinblastine with calf brain tubulin: effects of magnesium ions. Biochemistry, 25: 6222-6228, 1986. 9. Kumar, N. Taxol-induced polymerization of purified tubulin. Mechanism of action. J Biol Chem, 256: 10435-10441, 1981. 10. Edler, M. C., Fernandez, A. M., Lassota, P., Ireland, C. M., and Barrows, L. R. Inhibition of tubulin polymerization by vitilevuamide, a bicyclic marine peptide, at a site distinct from colchicine, the vinca alkaloids, and dolastatin 10. Biochem Pharmacol, 63: 707-715, 2002. 11. Gigant, B., Wang, C., Ravelli, R. B., Roussi, F., Steinmetz, M. O., Curmi, P. A., Sobel, A., and Knossow, M. Structural basis for the regulation of tubulin by vinblastine. Nature, 435: 519-522, 2005. 12. Nogales, E., Wolf, S. G., and Downing, K. H. Structure of the alpha beta tubulin dimer by electron crystallography. Nature, 391: 199-203, 1998. 13. Hastie, S. B. Interactions of colchicine with tubulin. Pharmacol Ther, 51: 377- 401, 1991. 14. Schiff, P. B., Fant, J., and Horwitz, S. B. Promotion of microtubule assembly in vitro by taxol. Nature, 277: 665-667, 1979. 15. Andreu, J. M., Diaz, J. F., Gil, R., de Pereda, J. M., Garcia de Lacoba, M., Peyrot, V., Briand, C., Towns-Andrews, E., and Bordas, J. Solution structure of Taxotere- induced microtubules to 3-nm resolution. The change in protofilament number is linked to the binding of the taxol side chain. J Biol Chem, 269: 31785-31792, 1994. 133

16. Petrylak, D. P., Tangen, C. M., Hussain, M. H., Lara, P. N., Jr., Jones, J. A., Taplin, M. E., Burch, P. A., Berry, D., Moinpour, C., Kohli, M., Benson, M. C., Small, E. J., Raghavan, D., and Crawford, E. D. Docetaxel and estramustine compared with mitoxantrone and prednisone for advanced refractory prostate cancer. N Engl J Med, 351: 1513-1520, 2004. 17. Diaz, J. F., Menendez, M., and Andreu, J. M. Thermodynamics of ligand-induced assembly of tubulin. Biochemistry, 32: 10067-10077, 1993. 18. Buey, R. M., Diaz, J. F., Andreu, J. M., O'Brate, A., Giannakakou, P., Nicolaou, K. C., Sasmal, P. K., Ritzen, A., and Namoto, K. Interaction of epothilone analogs with the paclitaxel binding site: relationship between binding affinity, microtubule stabilization, and cytotoxicity. Chem Biol, 11: 225-236, 2004. 19. Balachandran, R., ter Haar, E., Welsh, M. J., Grant, S. G., and Day, B. W. The potent microtubule-stabilizing agent (+)-discodermolide induces apoptosis in human breast carcinoma cells--preliminary comparisons to paclitaxel. Anticancer Drugs, 9: 67-76, 1998. 20. Mann, J. Myxobacterial bounty. Nature, 385: 117, 1997. 21. Martello, L. A., McDaid, H. M., Regl, D. L., Yang, C. P., Meng, D., Pettus, T. R., Kaufman, M. D., Arimoto, H., Danishefsky, S. J., Smith, A. B., 3rd, and Horwitz, S. B. Taxol and discodermolide represent a synergistic drug combination in human carcinoma cell lines. Clin Cancer Res, 6: 1978-1987, 2000. 22. Huang, G. S., Lopez-Barcons, L., Freeze, B. S., Smith, A. B., 3rd, Goldberg, G. L., Horwitz, S. B., and McDaid, H. M. Potentiation of taxol efficacy and by discodermolide in ovarian carcinoma xenograft-bearing mice. Clin Cancer Res, 12: 298-304, 2006. 23. Cortes, J. E. and Pazdur, R. Docetaxel. J Clin Oncol, 13: 2643-2655, 1995. 24. Rowinsky, E. K., Chaudhry, V., Cornblath, D. R., and Donehower, R. C. Neurotoxicity of Taxol. J Natl Cancer Inst Monogr: 107-115, 1993. 25. Wilson, L., Panda, D., and Jordan, M. A. Modulation of microtubule dynamics by drugs: a paradigm for the actions of cellular regulators. Cell Struct Funct, 24: 329-335, 1999. 26. Jordan, M. A. Mechanism of action of antitumor drugs that interact with microtubules and tubulin. Curr Med Chem Anticancer Agents, 2: 1-17, 2002. 27. Noble, R. L. The discovery of the vinca alkaloids--chemotherapeutic agents against cancer. Biochem Cell Biol, 68: 1344-1351, 1990. 28. Bai, R. L., Pettit, G. R., and Hamel, E. Binding of dolastatin 10 to tubulin at a distinct site for peptide antimitotic agents near the exchangeable nucleotide and vinca alkaloid sites. J Biol Chem, 265: 17141-17149, 1990. 29. Towle, M. J., Salvato, K. A., Budrow, J., Wels, B. F., Kuznetsov, G., Aalfs, K. K., Welsh, S., Zheng, W., Seletsk, B. M., Palme, M. H., Habgood, G. J., Singer, L. A., Dipietro, L. V., Wang, Y., Chen, J. J., Quincy, D. A., Davis, A., Yoshimatsu, K., Kishi, Y., Yu, M. J., and Littlefield, B. A. In vitro and in vivo anticancer activities of synthetic macrocyclic ketone analogues of halichondrin B. Cancer Res, 61: 1013-1021, 2001. 30. Bai, R. L., Paull, K. D., Herald, C. L., Malspeis, L., Pettit, G. R., and Hamel, E. Halichondrin B and homohalichondrin B, marine natural products binding in the 134

vinca domain of tubulin. Discovery of tubulin-based mechanism of action by analysis of differential cytotoxicity data. J Biol Chem, 266: 15882-15889, 1991. 31. Zhou, J. and Giannakakou, P. Targeting microtubules for cancer chemotherapy. Curr Med Chem Anticancer Agents, 5: 65-71, 2005. 32. Peterson, J. R. and Mitchison, T. J. Small molecules, big impact: a history of chemical inhibitors and the cytoskeleton. Chem Biol, 9: 1275-1285, 2002. 33. Engelborghs, Y. General features of the recognition by tubulin of colchicine and related compounds. Eur Biophys J, 27: 437-445, 1998. 34. Nakagawa-Goto, K., Chen, C. X., Hamel, E., Wu, C. C., Bastow, K. F., Brossi, A., and Lee, K. H. Antitumor agents. Part 236: Synthesis of water-soluble colchicine derivatives. Bioorg Med Chem Lett, 15: 235-238, 2005. 35. Jordan, A., Hadfield, J. A., Lawrence, N. J., and McGown, A. T. Tubulin as a target for anticancer drugs: agents which interact with the mitotic spindle. Med Res Rev, 18: 259-296, 1998. 36. Abraham, I., Dion, R. L., Chi, D. M., Gottesman, M. M., and Hamel, E. 2,4- Dichlorobenzyl thiocyanate, an antimitotic agent that alters microtubule morphology. Proc Natl Acad Sci U S A, 83: 6839-6843, 1986. 37. Ling, V. Charles F. Kettering Prize. P-glycoprotein and resistance to anticancer drugs. Cancer, 69: 2603-2609, 1992. 38. Kavallaris, M., Kuo, D. Y., Burkhart, C. A., Regl, D. L., Norris, M. D., Haber, M., and Horwitz, S. B. Taxol-resistant epithelial ovarian tumors are associated with altered expression of specific beta-tubulin isotypes. J Clin Invest, 100: 1282- 1293, 1997. 39. Hari, M., Loganzo, F., Annable, T., Tan, X., Musto, S., Morilla, D. B., Nettles, J. H., Snyder, J. P., and Greenberger, L. M. Paclitaxel-resistant cells have a mutation in the paclitaxel-binding region of beta-tubulin (Asp26Glu) and less stable microtubules. Mol Cancer Ther, 5: 270-278, 2006. 40. Ambudkar, S. V., Dey, S., Hrycyna, C. A., Ramachandra, M., Pastan, I., and Gottesman, M. M. Biochemical, cellular, and pharmacological aspects of the multidrug transporter. Annu Rev Pharmacol Toxicol, 39: 361-398, 1999. 41. Rowinsky, E. K. and Donehower, R. C. Paclitaxel (taxol). N Engl J Med, 332: 1004-1014, 1995. 42. Verschraegen, C. F., Sittisomwong, T., Kudelka, A. P., Guedes, E., Steger, M., Nelson-Taylor, T., Vincent, M., Rogers, R., Atkinson, E. N., and Kavanagh, J. J. Docetaxel for patients with paclitaxel-resistant Mullerian carcinoma. J Clin Oncol, 18: 2733-2739, 2000. 43. Giannakakou, P., Sackett, D. L., Kang, Y. K., Zhan, Z., Buters, J. T., Fojo, T., and Poruchynsky, M. S. Paclitaxel-resistant human ovarian cancer cells have mutant beta-tubulins that exhibit impaired paclitaxel-driven polymerization. J Biol Chem, 272: 17118-17125, 1997. 44. Lamendola, D. E., Duan, Z., Penson, R. T., Oliva, E., and Seiden, M. V. beta tubulin mutations are rare in human ovarian carcinoma. Anticancer Res, 23: 681- 686, 2003. 45. Sale, S., Sung, R., Shen, P., Yu, K., Wang, Y., Duran, G. E., Kim, J. H., Fojo, T., Oefner, P. J., and Sikic, B. I. Conservation of the class I beta-tubulin gene in 135

human populations and lack of mutations in lung cancers and paclitaxel-resistant ovarian cancers. Mol Cancer Ther, 1: 215-225, 2002. 46. Hadfield, J. A., Ducki, S., Hirst, N., and McGown, A. T. Tubulin and microtubules as targets for anticancer drugs. Prog Cell Cycle Res, 5: 309-325, 2003. 47. Honore, S., Pasquier, E., and Braguer, D. Understanding microtubule dynamics for improved cancer therapy. Cell Mol Life Sci, 62: 3039-3056, 2005. 48. Perez, E. A. Microtubule inhibitors: Differentiating tubulin-inhibiting agents based on mechanisms of action, clinical activity, and resistance. Mol Cancer Ther, 8: 2086-2095, 2009. 49. Heinz, D. W., Schubert, W. D., and Hofle, G. Much anticipated--the bioactive conformation of epothilone and its binding to tubulin. Angew Chem Int Ed Engl, 44: 1298-1301, 2005. 50. Lee, F. Y., Borzilleri, R., Fairchild, C. R., Kamath, A., Smykla, R., Kramer, R., and Vite, G. Preclinical discovery of ixabepilone, a highly active antineoplastic agent. Cancer Chemother Pharmacol, 63: 157-166, 2008. 51. Polgar, O. and Bates, S. E. ABC transporters in the balance: is there a role in multidrug resistance? Biochem Soc Trans, 33: 241-245, 2005. 52. Michels, J., Ellard, S. L., Le, L., Kollmannsberger, C., Murray, N., Tomlinson Guns, E. S., Carr, R., and Chi, K. N. A phase IB study of ABT-751 in combination with docetaxel in patients with advanced castration-resistant prostate cancer. Ann Oncol, 21: 305-311. 53. Fox, E., Maris, J. M., Widemann, B. C., Goodspeed, W., Goodwin, A., Kromplewski, M., Fouts, M. E., Medina, D., Cohn, S. L., Krivoshik, A., Hagey, A. E., Adamson, P. C., and Balis, F. M. A phase I study of ABT-751, an orally bioavailable tubulin inhibitor, administered daily for 21 days every 28 days in pediatric patients with solid tumors. Clin Cancer Res, 14: 1111-1115, 2008. 54. Lu, Y., Li, C. M., Wang, Z., Ross, C. R., 2nd, Chen, J., Dalton, J. T., Li, W., and Miller, D. D. Discovery of 4-substituted methoxybenzoyl-aryl-thiazole as novel anticancer agents: synthesis, biological evaluation, and structure-activity relationships. J Med Chem, 52: 1701-1711, 2009. 55. Traunecker, H. C., Stevens, M. C., Kerr, D. J., and Ferry, D. R. The acridonecarboxamide GF120918 potently reverses P-glycoprotein-mediated resistance in human sarcoma MES-Dx5 cells. Br J Cancer, 81: 942-951, 1999. 56. Davies, B. and Morris, T. Physiological parameters in laboratory animals and humans. Pharm Res, 10: 1093-1095, 1993. 57. Goncalves, A., Braguer, D., Kamath, K., Martello, L., Briand, C., Horwitz, S., Wilson, L., and Jordan, M. A. Resistance to Taxol in lung cancer cells associated with increased microtubule dynamics. Proc Natl Acad Sci U S A, 98: 11737- 11742, 2001. 58. Monzo, M., Rosell, R., Sanchez, J. J., Lee, J. S., O'Brate, A., Gonzalez-Larriba, J. L., Alberola, V., Lorenzo, J. C., Nunez, L., Ro, J. Y., and Martin, C. Paclitaxel resistance in non-small-cell lung cancer associated with beta-tubulin gene mutations. J Clin Oncol, 17: 1786-1793, 1999.

136

59. Lin, J. H. and Lu, A. Y. Role of pharmacokinetics and metabolism in drug discovery and development. Pharmacol Rev, 49: 403-449, 1997. 60. Fukuda, D. S., Mabe, J. A., and Brannon, D. R. Production of substituted L- tryptophans by fermentation. Appl Microbiol, 21: 841-843, 1971. 61. Windebank, A. J. Chemotherapeutic neuropathy. Curr Opin Neurol, 12: 565-571, 1999. 62. Nakai, R., Iida, S., Takahashi, T., Tsujita, T., Okamoto, S., Takada, C., Akasaka, K., Ichikawa, S., Ishida, H., Kusaka, H., Akinaga, S., Murakata, C., Honda, S., Nitta, M., Saya, H., and Yamashita, Y. K858, a novel inhibitor of mitotic kinesin Eg5 and antitumor agent, induces cell death in cancer cells. Cancer Res, 69: 3901- 3909, 2009. 63. Bacher, G., Nickel, B., Emig, P., Vanhoefer, U., Seeber, S., Shandra, A., Klenner, T., and Beckers, T. D-24851, a novel synthetic microtubule inhibitor, exerts curative antitumoral activity in vivo, shows efficacy toward multidrug-resistant tumor cells, and lacks neurotoxicity. Cancer Res, 61: 392-399, 2001. 64. Gourdeau, H., Leblond, L., Hamelin, B., Dong, K., Ouellet, F., Boudreau, C., Custeau, D., Richard, A., Gilbert, M. J., and Jolivet, J. Species differences in troxacitabine pharmacokinetics and pharmacodynamics: implications for clinical development. Clin Cancer Res, 10: 7692-7702, 2004. 65. Skoufias, D. A. and Wilson, L. Mechanism of inhibition of microtubule polymerization by colchicine: inhibitory potencies of unliganded colchicine and tubulin-colchicine complexes. Biochemistry, 31: 738-746, 1992. 66. Sweetnam, P. M., Caldwell, L., Lancaster, J., Bauer, C., Jr., McMillan, B., Kinnier, W. J., and Price, C. H. The role of receptor binding in drug discovery. J Nat Prod, 56: 441-455, 1993. 67. Schramm, V. L. Binding isotope effects: boon and bane. Curr Opin Chem Biol, 11: 529-536, 2007. 68. Niessen, K. V., Hofner, G., and Wanner, K. T. Competitive MS binding assays for dopamine D2 receptors employing spiperone as a native marker. Chembiochem, 6: 1769-1775, 2005. 69. Zepperitz, C., Hofner, G., and Wanner, K. T. Expanding the scope of MS binding assays to low-affinity markers as exemplified for mGAT1. Anal Bioanal Chem, 391: 309-316, 2008. 70. de Jong, L. A., Jeronimus-Stratingh, C. M., and Cremers, T. I. Development of a multiplex non-radioactive receptor assay: the benzodiazepine receptor, the serotonin transporter and the beta-adrenergic receptor. Rapid Commun Mass Spectrom, 21: 567-572, 2007. 71. Jiang, Y., Wang, J., Wang, Y., Li, H., Fawcett, J. P., and Gu, J. Rapid and sensitive liquid chromatography-tandem mass spectrometry method for the quantitation of colchicine in human plasma. J Chromatogr B Analyt Technol Biomed Life Sci, 850: 564-568, 2007. 72. Abe, E., Lemaire-Hurtel, A. S., Duverneuil, C., Etting, I., Guillot, E., de Mazancourt, P., and Alvarez, J. C. A novel LC-ESI-MS-MS method for sensitive quantification of colchicine in human plasma: application to two case reports. J Anal Toxicol, 30: 210-215, 2006. 137

73. Dennison, J. B., Renbarger, J. L., Walterhouse, D. O., Jones, D. R., and Hall, S. D. Quantification of vincristine and its major metabolite in human plasma by high-performance liquid chromatography/tandem mass spectrometry. Ther Drug Monit, 30: 357-364, 2008. 74. Vainchtein, L. D., Thijssen, B., Stokvis, E., Rosing, H., Schellens, J. H., and Beijnen, J. H. A simple and sensitive assay for the quantitative analysis of paclitaxel and metabolites in human plasma using liquid chromatography/tandem mass spectrometry. Biomed Chromatogr, 20: 139-148, 2006. 75. Garland, D. L. Kinetics and mechanism of colchicine binding to tubulin: evidence for ligand-induced conformational change. Biochemistry, 17: 4266-4272, 1978. 76. Zhao, Y., Fang, W. S., and Pors, K. Microtubule stabilising agents for cancer chemotherapy. Expert Opin Ther Pat, 19: 607-622, 2009. 77. Carlson, R. O. New tubulin targeting agents currently in clinical development. Expert Opin Investig Drugs, 17: 707-722, 2008. 78. Pelkonen, O., Turpeinen, M., Uusitalo, J., Rautio, A., and Raunio, H. Prediction of drug metabolism and interactions on the basis of in vitro investigations. Basic Clin Pharmacol Toxicol, 96: 167-175, 2005. 79. Zhang, H., Zhang, D., Li, W., Yao, M., D'Arienzo, C., Li, Y. X., Ewing, W. R., Gu, Z., Zhu, Y., Murugesan, N., Shyu, W. C., and Humphreys, W. G. Reduction of site-specific CYP3A-mediated metabolism for dual angiotensin and endothelin receptor antagonists in various in vitro systems and in cynomolgus monkeys. Drug Metab Dispos, 35: 795-805, 2007. 80. Huang, M. and Ho, P. C. Identification of metabolites of meisoindigo in rat, pig and human liver microsomes by UFLC-MS/MS. Biochem Pharmacol, 77: 1418- 1428, 2009. 81. Baarnhielm, C., Dahlback, H., and Skanberg, I. In vivo pharmacokinetics of felodipine predicted from in vitro studies in rat, dog and man. Acta Pharmacol Toxicol (Copenh), 59: 113-122, 1986. 82. Smith, R., Jones, R. D., Ballard, P. G., and Griffiths, H. H. Determination of microsome and hepatocyte scaling factors for in vitro/in vivo extrapolation in the rat and dog. Xenobiotica, 38: 1386-1398, 2008. 83. Chiba, M., Ishii, Y., and Sugiyama, Y. Prediction of hepatic clearance in human from in vitro data for successful drug development. Aaps J, 11: 262-276, 2009. 84. Engelman, J. A., Luo, J., and Cantley, L. C. The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolism. Nat Rev Genet, 7: 606-619, 2006. 85. Wymann, M. P., Zvelebil, M., and Laffargue, M. Phosphoinositide 3-kinase signalling--which way to target? Trends Pharmacol Sci, 24: 366-376, 2003. 86. Feng, J., Park, J., Cron, P., Hess, D., and Hemmings, B. A. Identification of a PKB/Akt hydrophobic motif Ser-473 kinase as DNA-dependent protein kinase. J Biol Chem, 279: 41189-41196, 2004. 87. Vivanco, I. and Sawyers, C. L. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer, 2: 489-501, 2002.

138

88. Grunwald, V., DeGraffenried, L., Russel, D., Friedrichs, W. E., Ray, R. B., and Hidalgo, M. Inhibitors of mTOR reverse doxorubicin resistance conferred by PTEN status in prostate cancer cells. Cancer Res, 62: 6141-6145, 2002. 89. Cohen, P. and Frame, S. The renaissance of GSK3. Nat Rev Mol Cell Biol, 2: 769-776, 2001. 90. Frame, S. and Cohen, P. GSK3 takes centre stage more than 20 years after its discovery. Biochem J, 359: 1-16, 2001. 91. Dan, H. C., Sun, M., Yang, L., Feldman, R. I., Sui, X. M., Ou, C. C., Nellist, M., Yeung, R. S., Halley, D. J., Nicosia, S. V., Pledger, W. J., and Cheng, J. Q. Phosphatidylinositol 3-kinase/Akt pathway regulates tuberous sclerosis tumor suppressor complex by phosphorylation of tuberin. J Biol Chem, 277: 35364- 35370, 2002. 92. Mita, M. M., Mita, A., and Rowinsky, E. K. The molecular target of rapamycin (mTOR) as a therapeutic target against cancer. Cancer Biol Ther, 2: S169-177, 2003. 93. Gunn, R. M. and Hailes, H. C. Insights into the PI3-K-PKB-mTOR signalling pathway from small molecules. J Chem Biol, 1: 49-62, 2008. 94. Hardie, D. G. and Carling, D. The AMP-activated protein kinase--fuel gauge of the mammalian cell? Eur J Biochem, 246: 259-273, 1997. 95. Woods, A., Johnstone, S. R., Dickerson, K., Leiper, F. C., Fryer, L. G., Neumann, D., Schlattner, U., Wallimann, T., Carlson, M., and Carling, D. LKB1 is the upstream kinase in the AMP-activated protein kinase cascade. Curr Biol, 13: 2004-2008, 2003. 96. Hardie, D. G., Salt, I. P., Hawley, S. A., and Davies, S. P. AMP-activated protein kinase: an ultrasensitive system for monitoring cellular energy charge. Biochem J, 338 (Pt 3): 717-722, 1999. 97. Kemp, B. E., Stapleton, D., Campbell, D. J., Chen, Z. P., Murthy, S., Walter, M., Gupta, A., Adams, J. J., Katsis, F., van Denderen, B., Jennings, I. G., Iseli, T., Michell, B. J., and Witters, L. A. AMP-activated protein kinase, super metabolic regulator. Biochem Soc Trans, 31: 162-168, 2003. 98. Inoki, K., Zhu, T., and Guan, K. L. TSC2 mediates cellular energy response to control cell growth and survival. Cell, 115: 577-590, 2003. 99. Kimura, N., Tokunaga, C., Dalal, S., Richardson, C., Yoshino, K., Hara, K., Kemp, B. E., Witters, L. A., Mimura, O., and Yonezawa, K. A possible linkage between AMP-activated protein kinase (AMPK) and mammalian target of rapamycin (mTOR) signalling pathway. Genes Cells, 8: 65-79, 2003. 100. Luo, Z., Saha, A. K., Xiang, X., and Ruderman, N. B. AMPK, the metabolic syndrome and cancer. Trends Pharmacol Sci, 26: 69-76, 2005. 101. Hadad, S. M., Fleming, S., and Thompson, A. M. Targeting AMPK: a new therapeutic opportunity in breast cancer. Crit Rev Oncol Hematol, 67: 1-7, 2008. 102. Motoshima, H., Goldstein, B. J., Igata, M., and Araki, E. AMPK and cell proliferation--AMPK as a therapeutic target for atherosclerosis and cancer. J Physiol, 574: 63-71, 2006. 103. Lu, Y., Wang, Z., Li, C. M., Chen, J., Dalton, J. T., Li, W., and Miller, D. D. Synthesis, in vitro structure-activity relationship, and in vivo studies of 2- 139

arylthiazolidine-4-carboxylic acid amides as anticancer agents. Bioorg Med Chem, 18: 477-495. 104. Ohta, H., Sato, K., Murata, N., Damirin, A., Malchinkhuu, E., Kon, J., Kimura, T., Tobo, M., Yamazaki, Y., Watanabe, T., Yagi, M., Sato, M., Suzuki, R., Murooka, H., Sakai, T., Nishitoba, T., Im, D. S., Nochi, H., Tamoto, K., Tomura, H., and Okajima, F. Ki16425, a subtype-selective antagonist for EDG-family lysophosphatidic acid receptors. Mol Pharmacol, 64: 994-1005, 2003. 105. Crouthamel, M. C., Kahana, J. A., Korenchuk, S., Zhang, S. Y., Sundaresan, G., Eberwein, D. J., Brown, K. K., and Kumar, R. Mechanism and management of AKT inhibitor-induced hyperglycemia. Clin Cancer Res, 15: 217-225, 2009. 106. Hoffman, K., Holmes, F. A., Fraschini, G., Esparza, L., Frye, D., Raber, M. N., Newman, R. A., and Hortobagyi, G. N. Phase I-II study: triciribine (tricyclic nucleoside phosphate) for metastatic breast cancer. Cancer Chemother Pharmacol, 37: 254-258, 1996. 107. Hurley, R. L., Anderson, K. A., Franzone, J. M., Kemp, B. E., Means, A. R., and Witters, L. A. The Ca2+/calmodulin-dependent protein kinase kinases are AMP- activated protein kinase kinases. J Biol Chem, 280: 29060-29066, 2005. 108. King, T. D., Song, L., and Jope, R. S. AMP-activated protein kinase (AMPK) activating agents cause dephosphorylation of Akt and glycogen synthase kinase-3. Biochem Pharmacol, 71: 1637-1647, 2006. 109. Rattan, R., Giri, S., Singh, A. K., and Singh, I. 5-Aminoimidazole-4- carboxamide-1-beta-D-ribofuranoside inhibits cancer cell proliferation in vitro and in vivo via AMP-activated protein kinase. J Biol Chem, 280: 39582-39593, 2005. 110. Hahn-Windgassen, A., Nogueira, V., Chen, C. C., Skeen, J. E., Sonenberg, N., and Hay, N. Akt activates the mammalian target of rapamycin by regulating cellular ATP level and AMPK activity. J Biol Chem, 280: 32081-32089, 2005.

140