Pharmacokinetics, Pharmacodynamics and Metabolism Of

Home , Hybridization assay, Nuclease S1, Nucleic acid methods

PHARMACOKINETICS, PHARMACODYNAMICS AND METABOLISM OF

GTI-2040, A PHOSPHOROTHIOATE OLIGONUCLEOTIDE TARGETING

R2 SUBUNIT OF RIBONUCLEOTIDE REDUCTASE

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the

Graduate School of The Ohio State University

Xiaohui Wei, M.S.

* * * * * *

The Ohio State University 2006

Approved by Dissertation Committee:

Dr. Kenneth K. Chan, Adviser Adviser Dr. Guido Marcucci, Co-adviser Graduate Program in Pharmacy Dr. Thomas D. Schmittgen

Dr. Robert J. Lee Co-Adviser Graduate Program in Pharmacy ABSTRACT

Over the last several decades, antisense therapy has been developed into a promising gene-targeted strategy to specifically inhibit the gene expression.

Ribonucleotide reductase (RNR), composing of subunits R1 and R2, is an important

enzyme involved in the synthesis of all of the precursors used in DNA replication. Over-

expression of R2 has been found in almost every type of cancer studied. GTI-2040 is a

20-mer phosphorothioate oligonucleotide targeting the coding region in mRNA of the R2

component of human RNR. In this project, clinical pharamcokinetics (PK),

pharmacodynamics (PD) and metabolism of this novel therapeutics were investigated in

patients with acute myeloid leukemia (AML).

A picomolar specific hybridization-ligation ELISA method has been developed

and validated for quantification of GTI-2040. GTI-2040 and neophectin complex was

found to enhance drug cellular uptake and exhibited sequence- and dose-dependent

down-regulation of R2 mRNA and protein in K562 cells.

Robust intracellular concentrations (ICs) of GTI-2040 were achieved in

peripheral blood mononuclear cells (PBMC) and bone marrow (BM) cells from treated

AML patients. GTI-2040 concentrations in the nucleus of BM cells were found to

correlate with the R2 mRNA down-regulation and disease response. In treated patients

ii from 18-60 years old, down-regulations of R2 protein were found in responders, while the non-responders were found to exhibit up-regulation of R2 proteins.

Plasma PK of GTI-2040 in AML patients were characterized and found to fit a two-compartment infusion model with mean body clearance of 10.1 L/hr, mean t1/2α of

0.81 hr and mean t 1/2β of 27 hr. Population PK analysis of GTI-2040 identified BM

cellularity, white blood cell counts and gender to be important covariates of PK

parameters. A PK/PD model has been developed to characterize the dynamics of down-

regulation of R2 mRNA following drug perturbation.

The 3’ end chain-shortened metabolites of GTI-2040 were identified in various

specimens from human and animals using a novel ion-pair reverse-phase high

performance liquid chromatography/mass spectrometry (HPLC/MS) method. Sequential

metabolism and protein binding were found to complicate the enzyme kinetics of GTI-

2040 in human liver microsomes.

The results from these studies provided valuable insights to the evaluation and

utilization of GTI-2040 in the clinics.

iii Dedicated to my parents and to my friends.

iv ACKNOWLEDGMENTS

I would like to thank my adviser, Dr. Kenneth K. Chan, for his intellectual guidance, constant encouragement, stimulation and support during my graduate studies. I also would like to thank my co-adviser, Dr. Guido Macucci, for his support in the clinical study and his invaluable suggestions and comments on this project. Appreciation also goes to my committee members, Dr. Thomas D. Schmittgen, Dr. Rober J. Lee and Dr.

William L. Hayton for their time, constructive suggestions and comments on this project.

The work with GTI-2040 project would not be possible without the significant contribution from the laboratory of Dr. Guido Marcucci in the Division of Hematology-

Oncology, College of Medicine. A special acknowledgement goes to Dr. Guowei Dai for his invaluable suggestions and help on the entire project. I also thank Dr. Shujun Liu for his assistance on western blotting, transfection experiments and stimulating discussions on this project. A special thank is also extended to Ms. LeNguyen Huynh for the many hours she spent collecting patients’ samples and her technical assistance in the extraction and quantitation of R2 mRNA. Additional thanks go to Dr. Kui Liu for his help in cloning R2 cDNA standards and Dr. Jin Sun for his assistance in flow cytometry

v analysis. Thanks also goes to Dr. Zhongfa Liu for his help on operation of mass

spectrometer and valuable scientific discussions on the metabolism studies, to Dr. Jim

Xiao for his discussion of modeling and continuous encouragement, to Dr. Duxin Sun

and his laboratory for their support of cell culture facilities and valuable comments on

this project, and to Drs. Hao Cheng and Zhiliang Xie for their help on tissue handling and sample preparation in metabolism study and their friendship.

I am also grateful to Ms. Joy Scott and Ms. Kathy Brooks for their administrative assistance, encouragement and friendship. I am further indebted to all of my fellow graduate students and colleagues in the College of Pharmacy, especially my roommate

Ms. Yan Xin, for their warmest friendship, invaluable encouragement and scientific discussions.

Finally, I would like to express special thanks to my parents and sisters who have been giving me unconditional support and have encouraged me to strive for excellence in this long and challenging endeavor.

vi VITA

1989-1993…………………………...B.S. in Pharmacy Beijing University of Chinese Traditional Medicine

1993-1996…………………………...Instructor Guangxi University of Chinese Traditional Medicine

1996-1999……………………………M.S. in Pharmacy West China University of Medical Sciences

1999-2000……………………………Associate Researcher Institute of Materia Medica Chinese Academy of Medical Sciences

2000-Present…………………………Graduate Teaching and Research Associate, The Ohio State University

PUBLICATIONS

1. Wei X, Chen W, Liu H. Evaluation of dissolution rate of baicalin in yinhuang tablet. Chinese Medicine Patent. 1996, 18(1): 3-6.

2. Liu H, Wei X, Tang D, Zhong D. Pharmacokinetics of baicalin in Yinzhi injection in rabbit. Journal of Guanxi Traditional Chinese Medical University. 1998, 1:21-25.

3. Yuan Y, Wei X, Yang J. Zhang L, Jiang X. Study on the optimum technology of extraction for Danqi Chinese medicine preparation. West China Journal of Pharmaceutical Sciences. 2002, 17(1):10-13.

4. Dai G, Wei X, Liu Z, Liu S, Marcucci, G, Chan KK. Characterization and vii quantification of Bcl-2 antisense G3139 and major metabolites in plasma and urine by ion-pair reverse phase HPLC coupled with electrospray ion-trap mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci. 2005, 825(2): 201-13.

5. Otterson GA, Lavelle J, Villalone-Calero MA, Shah M, Wei X, Chan K, Fischer B, Zwiebel J, Grever M. A phase I clinical and pharmacokinetic study of fenretinide combined with paclitaxel and cisplatin for refractory solid tumors. Invest New Drugs. 2005, 23(6): 555-562.

6. Wei X, Dai G, Marcucci G, Lui S, Hoyt D, Chan KK. A specific picomolar hybridization-based ELISA assay for the determination of phosphorothioate oligonucleotides in plasma and cellular matrices. Pharmaceutical Research. In press, 2006

7. O’Connor OA, Smith EA, Toner LE, Feldstien J, Frankel S, Rolfe M, Wei X, Lui S, Marcucci G, Chan KK, Chanan-Khan A. The combination of the proteasome inhibitor Bortezomib and the Bcl-2 antisense molecule Oblimersen sensitizes human B-cell lymphomas to cyclophosphamide. Clinical Cancer Research. Accepted, 2006.

FIELDS OF STUDY

Major Field: Pharmacy/Pharmaceutics

-with studies on bioanalytical method development, clinical pharmacokinetics, pharmacodynamics and drug metabolism.

viii

TABLE OF CONTENTS

Page

ABSTRACT...... ii ACKNOWLEDGMENTS ...... v VITA...... vii LIST OF TABLES...... xv LIST OF FIGURES ...... xvii LIST OF SCHEMES.……………………………………………………………….xxiii

CHAPTER 1 ...... 1 BACKGROUND AND INTRODUCTION...... 1 1.1 Background...... 1 1.2 Introduction...... 4 1.2.1 Principle of antisense therapy...... 4 1.2.1.1 Mechanisms of action of antisense oligonucleotides...... 4 1.2.1.2 Antisense oligonucleotides modifications and other anti-mRNA strategies ………………………………………………………………………….6 1.2.1.2.1 First generation of antisense oligonucleotides...... 7 1.2.1.2.2 The second generation of antisense oligonucleotides...... 8 1.2.1.2.3 Other anti-mRNA strategies ...... 9 1.2.2 Pharmacokinetics and pharmacodynamics of phosphorothioate oligonucleotides...... 11 1.2.2.1 Cellular uptake of PS-ODNs...... 11 1.2.2.2 Pharmacokinetics of phosphorothioate oligonucleotides ...... 13 1.2.2.3 Metabolism of phosphorothioate oligonucleotides...... 16 1.2.2.4 Pharmacodynamics of phosphorothioate oligonucleotides...... 17 1.2.3 Inhibition of ribonucleotide reductase as a strategy in anticancer therapy18 1.2.4 GTI-2040-a specific inhibitor of R2 subunit of RNR...... 20 1.2.5 Acute myeloid leukemia and cytarabine...... 23 1.2.5.1 Acute myeloid leukemia ...... 23 1.2.5.2 Cytarabine...... 25 1.2.6 Rationale for combination therapy of GTI-2040 with cytarabine ...... 26 1.3 Specific aims...... 27

ix 1.4 Dissertation Overview...... 28

CHAPTER 2 ...... 42 A SPECIFIC PICOMOLAR HYBRIDIZATION-BASED ELISA ASSAY FOR THE DETERMINATION OF PHOSPHOROTHIOATE OLIGONUCLEOTIDES IN PLASMA AND CELLULAR MATRICES...... 42 2.1 Introduction...... 42 2.2 Materials and methods ...... 44 2.2.1 Antisense and Reagents ...... 44 2.2.2 Hybridization ELISA Assay Procedures ...... 46 2.2.2.1 A One-step Hybridization ELISA...... 46 2.2.2.2 A Two-step Hybridization-ligation ELISA Assay...... 47 2.2.3 Evaluation of Cutting Efficiency of S1 Nuclease...... 48 2.2.4 Specificity and Selectivity of Hybridization-Ligation ELISA ...... 49 2.2.4.1 Effect of S1 Nuclease on the Assay Selectivity...... 49 2.2.4.2 Specificity and Selectivity of Hybridization-Ligation ELISA ...... 50 2.2.5 Method Validation...... 50 2.3 Results...... 51 2.3.1 Cutting Efficiency of S1 Nuclease as Evaluated by the One-step Hybridization ELISA...... 51 2.3.2 Effect of S1 Nuclease on Selectivity of the Hybridization-Ligation ELISA ……………………………………………………………………………51 2.3.3 Method Validation...... 52 2.4 Discussion...... 54

CHAPTER 3 ...... 70 CELLULAR UPTAKE AND PHARMACOLOGY OF GTI-2040 IN LEUKEMIA CELL LINES ...... 70 3.1 Introduction...... 70 3.2 Materials and Methods...... 73 3.2.1 Drugs and reagents...... 73 3.2.2 Cell cultures...... 74 3.2.3 Transfection cells with GTI-2040 delivered by cationic reagents...... 75 3.2.4 Fractionation of nucleus and cytoplasm from cells ...... 77 3.2.5 Uptake of GTI-2040 in mononuclear cells and red blood cells following blood incubation ...... 78 3.2.6 Determination of intracellular GTI-2040 concentrations ...... 79 3.2.7 Flow cytometric analysis for studying of cellular uptake...... 80 3.2.8 Laser Scanning Microscopy analysis of cellular uptake of GTI-2040 ....81 x 3.2.9 Quantification of R2 mRNA levels...... 81 3.2.10 Measurement of R2 protein levels ...... 82 3.2.11 Cytotoxicity studies by MTS assay ...... 84 3.3 Results...... 85 3.3.1 Cellular uptake of GTI-2040 without or with transfection agents...... 85 3.3.2 Subcellular distribution of GTI-2040 ...... 87 3.3.3 Effect of temperature on the cellular uptake of GTI-2040 ...... 88 3.3.4 Cellular uptake of GTI-2040 from the human blood...... 88 3.3.5 Pharmacological effects of GTI-2040 on K562 cells ...... 89 3.3.5 Effect of GTI-2040 on the cytotoxicity of cytarabine ...... 90 3.4 Discussion...... 91

CHAPTER 4 ...... 106 CLINICAL PHARMACOKINETICS AND PHARMACODYNAMICS OF GTI-2040 IN PATIENTS WITH ACUTE MYELOID LEUKEMIA AND ITS PK/PD CORRELATIONS ...... 106 4.1 Introduction...... 106 4.2 Materials and Methods...... 108 4.2.1 Drugs and reagents...... 108 4.2.2 Patient Characteristics and Dosing Schedules...... 109 4.2.3 Plasma Protein Binding ...... 110 4.2.4 Sample Collection and Preparation Procedures...... 110 4.2.4.1 Separation of plasma, PBMC and RBC from patients’ blood samples ………………………………………………………………………..110 4.2.4.3 Bone Marrow Samples Preparation ...... 111 4.2.4.4 Separation of the nucleus and cytoplasm from the frozen bone marrow cells ………………………………………………………………………..113 4.2.4.2 Urine Sample Collection...... 113 4.3 Assay methods ...... 113 4.3.1 GTI-2040 and protein determinations...... 113 4.3.2 Quantitation of R2 mRNA levels...... 114 4.3.3 Measurement of R2 Protein Levels ...... 115 4.4 Data Analysis...... 115 4.4.1 Pharmacokinetic (PK) analysis...... 115 4.5 Results...... 116 4.5.1 Plasma Protein Binding ...... 116 4.5.2 Plasma Pharmacokinetics...... 117 4.5.3 Urinary clearance and hepatic intrinsic clearance ...... 117 4.5.4 Intracellular uptake and distribution of GTI-2040 in peripheral blood and

xi bone marrow mononuclear cells ...... 118 4.5.5 Pharmacodynamic results...... 120 4.5.6 Disease Response...... 122 4.5.7 Correlations among pharmacokinetics, pharmacodynamics, and response …………………………………………………………………………..122 4.6 Discussion...... 124

CHAPTER 5 ...... 156 POPULATION PHARMACOKINETICS AND PK/PD MODELING OF GTI-2040 IN PATIENTS WITH ACUTE MYELOID LEUKEMIA...... 156 5.1 Introduction...... 156 5.2 Clinical trial and Methods...... 161 5.2.1 Clinical Trial...... 161 5.2.2 Software ...... 162 5.3 Results and Discussions...... 163 5.3.1 Population PK of GTI-2040 in plasma ...... 163 5.3.2 Population pharmacokinetics of intracellular GTI-2040 in PBMC...... 166 5.3.3 PK/PD modeling of GTI-2040 and down-regulation of R2 mRNA in patients …………………………………………………………………………..170 5.3.4 Simulations based on the PK/PD indirect response model...... 175

CHAPTER 6 ...... 201 IN VIVO AND IN VITRO METABOLISM OF GTI-2040, A PHOSPHOROTHIOATE OLIGONUCLEOTIDE ANTISENSE TARGETING RIBONUCLEOTIDE REDUCTASE, USING ION-PAIR REVERSED PHASE HPLC COUPLED WITH ELECTROSPRAY ION-TRAP MASS SPECTROMETRY...... 201 6.1 Introduction...... 201 6.2.1 Drugs and Chemicals ...... 203 6.2.2 Instrumentation ...... 204 6.2.3 HPLC and LC-MS Conditions...... 204 6.2.4 Sample Preparation--Solid Phase Extraction (SPE) ...... 206 6.2.5 Identification of GTI-2040 Metabolites in Plasma from AML Patients.....206 6.2.6 In vitro Enzymatic Reaction ...... 207 6.2.7 In vitro Incubation of GTI-2040 with Fresh Human Blood...... 207 6.2.8 In vitro Incubation of GTI-2040 with Mouse Tissue Homogenates...... 207 6.2.9 In vitro Incubation of GTI-2040 with Pooled Human Liver Microsomes.208 6.2.10 Urinary Excretion of GTI-2040 and Metabolites in the Rat...... 208 6.3 Data Analysis...... 209 6.4 Results...... 209 xii 6.4.1 Identification of GTI-2040 Metabolites in Plasma obtained from Treated Patients...... 209 6.4.2 Role of 3’-Exonuclease in GTI-2040 Metabolism ...... 212 6.4.3 In vitro Metabolism Profiles of GTI-2040 in Tissue-Related Samples...... 213 6.4.4 Metabolites in Urine ...... 214 6.5 Discussion...... 215 6.6 Conclusion ...... 220

CHAPTER 7 ...... 235 ENZYME KINETICS OF GTI-2040, A PHOSPHOROTHIOATE OLIGONUCLEOTIDE TARGETING RIBONUCLEOTIDE REDUCTASE, TREATED WITH 3’ EXONUCLEASE AND HUMAN LIVER MICROSOMES..235 7.1 Introduction...... 235 7.2 Materials and Methods...... 237 7.2.1 Drugs and Chemicals ...... 237 7.2.2 HPLC Conditions...... 238 7.2.3 Sample Preparation--Solid Phase Extraction...... 239 7.2.4 Protein binding...... 239 7.2.5 Enzyme Kinetics of GTI 2040 in the Solution containing 3’ Exonuclease240 7.2.6 Enzyme Kinetics of GTI-2040 in Pooled Human Liver Microsomes ...... 241 7.2.7 In vivo Clearance ...... 241 7.2.8 Quantitation ...... 242 7.3 Data analysis ...... 242 7.3.1 Non-specific binding in HLM...... 242 7.3.2 Calculation of Vmax and Km...... 243 7.3.3 Calculation of CLint in human patients...... 243 7.3.4 Scaling of in vitro CLint to in vivo CLint...... 244 7.4 Results...... 244 7.4.1 Protein binding...... 244 7.4.2 Kinetics of Metabolite Formation from GTI-2040 with 3’ Exonuclease .245 7.4.3 Kinetics of Metabolites Formation of GTI-2040 in Pooled Human Liver Microsomes...... 246 7.4.4 Correlation of the in vitro CLint from HLM with the In vivo CLint.....248 7.5 Discussion...... 248

CHAPTER 8 ...... 266 CONCLUSION AND FUTURE PERSPECTIVES ...... 266

xiii APPENDIX A...... 272 CONTROL STREAMS IN NONMEM PROGRAM USED FOR POPULATION PHARMACOKINETIC MODELING OF PLASMA GTI-2040 IN AML PATIENTS WITH COVARIATES...... 272

APPENDIX B ...... 276 CONTROL STREAMS IN NONMEM PROGRAM USED FOR POPULATION PHARMACOKINETIC MODELING OF PBMC GTI-2040 IN AML PATIENTS WITH COVARIATES...... 276

APPENDIX C ...... 281 COMPUTER PROGRAMS USED FOR FITTING OF PK/PD MODEL OF R2 MRNA IN PATIENTS USING ADAPT II ...... 281

APPENDIX D...... 288 DATA RELATED TO CHAPTER 5: PK/PD FITTING RESULTS IN DOWN- REGULAION OF R2 MRNA FROM PATIENTS BM SAMPLES ...... 288

RFERENCE ...... 290

xiv LIST OF TABLES

Table 1.1 Examples of antisense oligonucleotides that were approved or under clinical trials...... 30

Table 2.1 Cutting efficiency of S1 nuclease toward one-step GTI-2040 capture probe with S1 at levels of 60 U/well and 100 U/well...... 60

Table 2.2 Cross-reactivity of 3’ and 5’ end putative metabolites, mismatched and scrambled ODNs toward GTI 2040...... 61

Table 2.3 Assay accuracy and reproducibility of GTI 2040 plasma samples following dilution...... 62

Table 2.4 Within-day and between-day accuracy and reproducibility of GTI 2040 in plasma, and within-day accuracy and precision in K562 cell lysates, RBC and urine...... 63

Table 4.1 Characteristics of the AML patients enrolled in the phase I trial with GTI-2040 (OSU 0304)...... 130

Table 4.2 Dose schedule escalation for cohort 1: patients <60 yrs ...... 131

Table 4.3 Dose schedule escalation cohort 2: patients ≥60 yrs ...... 132

Table 4.4 Results of plasma protein binding of GTI-2040 in plasma...... 133

Table 4.5 Relevant PK parameters of GTI-2040 in AML patients < 60yrs old given GTI- 2040 with 3.5 or 5.0 mg/kg/day...... 134

Table 4.6 Relevant PK parameters of GTI-2040 in AML patients ≥ 60yrs old given GTI- 2040 at 3.5, 5.0 or 7.0 mg/kg/day, respectively...... 135

Table 4.7 Renal clearance and hepatic intrinsic clearance from AML patients (n=20) treated with GTI-2040 ...... 136

Table 4.8 Comparison of cellular uptake of GTI-2040 from plasma into PBMC with its uptake to RBC...... 137 xv Table 4.9 Clinical response in patients treated with GTI-2040 combined with high....138

Table 5.1 Demographic and physiological parameters of 31 AML patients in the Phase I trial of GTI-2040...... 180

Table 5.2 Results of plasma pharmacokinetic parameters, inter- and intra- individual variance of GTI-2040 in AML patients (n=31) ...... 181

Table 5.3 Model predicted and Bootstrap results of pharmacokinetic parameters from182

Table 5.4 Results of PBMC pharmacokinetic parameters, inter- and intra- individual variance of GTI-2040 in AML patients (n=31) ...... 183

Table 5.5 Observed and the PK/PD model predicted GTI-2040 plasma concentrations (A) and R2 mRNA down-regulations (B) with time in patient 0304-26...... 184

Table 5.6 Mean values of the PK/PD parameters (Mean ± SD) from 5 patients estimated from an indirect response PK/PD model linked with an effect compartment ...... 185

Table 6.1 Structures and molecular weights of GTI-2040 and its metabolites in human plasma...... 221

Table 6.2 Assignment of fragment ions obtained from the MS2 spectrum following the collision-induced dissociation of the ion with m/z 1004.0 ([M-6H]-6) from 3’N-1 standard and M1 peak...... 222

Table 6.3 Assignment of the fragment ions obtained from the MS2 spectrum following the collision-induced dissociation on the ion with m/z 1004.0 ([M-6H]-6) from the 5’ N-1 standard...... 223

Table 7.1 Free fraction of GTI-2040 in human liver microsomes (0.2 mg protein/ml).256

Table 7.2 Kinetic parameters for the formation of 3’N-1 and 3’N-2 in 3’ exonuclease257

xvi LIST OF FIGURES

Figure 1.1 Mechanistic pathways of antisense oligonucleotides...... 31

Figure 1.2 Mechanisms of RNase H mediated antisense effect...... 32

Figure 1.3 Sites for chemical modifications of antisense oligonucleotides...... 33

Figure 1.4 Chemical structures of the first, second and the third generations of antisense oligonucleotides. B denotes one of the bases, adenine, guanine, cytosine or thymine...... 34

Figure 1.5 Possible mechanisms and pathways in cellular uptake and subcellular trafficking of PS-ODNs. 1: Passive diffusion; 2: Receptor mediated endocytosis; 3: Fluid phase pinocytosis 4: Adsorptive endocytosis...... 35

Figure 1.6 Possible degradation pathways (A) and cleavage positions (B) of PS-ODNs by the action of nuclease. a: Hydrolysis by 3’ exonuclease. b: Hydrolysis by 5’ exonuclease...... 36

Figure 1.7: Sp and Rp diastereoisomers generated from the phosphorothioate chiral center in the backbone of PS-ODNs...... 37

Figure 1.8 The function of ribonucleotide reductase (RNR). RNR reduces the ribonucleoside diphosphate to deoxyribonuceoside diphosphate...... 38

Figure 1.9 Sequence and structure of GTI-2040...... 39

Figure 1.10 Chemical structure of cytarabine (1-β-D-arabionfuranosylcytosine, ara-C).40

Figure 1.11 Metabolic pathway of cytarabine (ara-C)...... 41

Figure 2.1 Effect of S1 nuclease on assay selectivity...... 66

Figure 2.2 Effect of S1 nuclease on linearity of GTI 2040 in plasma...... 67

Figure 2.3 Cross-reactivity of the putative 3’ end metabolites (3’N-1, 3’N-2 and 3’N-3), scrambled and mismatched GTI-2040...... 68 xvii Figure 2.4 Representative calibration curves of GTI 2040 in (A) 10% plasma from the rat, mouse, human, and 100% human plasma and K562 cell lysates and (B) in human urine and 10% plasma, RBC, and K562 cell lysates...... 69

Figure 3.1 Uptake and efllux of GTI-2040 in K562 cells transfected with neophectin. .95

Figure 3.2 Flow cytometry analysis of cellular uptake of FITC-GTI-2040 in K562 cells in the absence or the presence of transfection reagents...... 96

Figure 3.3 Laser scanning microscopy analysis of cellular uptake of FITC-GTI-2040 in K562 cells...... 97

Figure 3.4 Subcellular distribution of GTI-2040 with or without the delivery of neophectin...... 99

Figure 3.5 Effect of temperature on the uptake of GTI-2040 complexed with neophectin...... 100

Figure 3.6 ICs of GTI-2040 in human peripheral blood mononuclear cells (PBMC) and red blood cells (RBC) after 0.5μM of free GTI-2040 were incubated in human blood from two subjects for 2hrs...... 101

Figure 3.7 ICs of GTI-2040 in the nucleus of K562 cells after cells were exposed to different dose of GTI-2040 complexed with neophectin for 24 hrs at 37 ◦C...... 102

Figure 3.8 Expression of R2 mRNA in K562 cells with treatments of complexed GTI- 2040, free GTI-2040 and scrambled or mismatch GTI-2040...... 103

Figure 3.9 Expression of R2 protein in K562 cells as measured by western blotting...104

Figure 3.10 Effect of GTI-2040 on cytotoxicity of cytarabin in K562 cells...... 105

Figure 4.1 Simultaneous separation of the plasma, PBMC and RBC from patient whole blood using BD Vacutainer cell preparation tube...... 140

Figure 4.2 A composite concentration-time profile of GTI-2040 in younger patients (<60yr) with AML after 144 hrs continuous infusion of GTI-2040 at 5mg/kg/day...... 141

xviii Figure 4.3 A composite concentration-time profile of GTI-2040 in older patients (≥60yr) with AML after 144 hrs continuous infusion of GTI-2040 at 5mg/kg/day...... 142

Figure 4.4 The entire concentration-time profile (A) and expanded post-infusion profile (B) of GTI-2040 in plasma, PBMC and RBC from 15 AML patients given GTI-2040 of 5.0 mg/kg/day as a 144 hr continuous i.v. infusion...... 143

Figure 4.5 Average concentrations of GTI-2040 in plasma (during day 1 , day 2, day 3, day 7, and day 8hr respectively), PBMC (during day 1 , day 2, day 3, day 7, and day 8hr respectively) and in BM (during day 1 and day 5) from younger patients (<60yr, A) and older AML patients (>60 yr, B) treated with GTI-2040...... 144

Figure 4.6 Correlation of plasma t1/2,β of GTI-2040 and IC in BM on day 5...... 145

Figure 4.7 Comparison of IC of GTI-2040 in CD34 + bone marrow blast cells and in the unmanipulated BM cells...... 146

Figure 4.8 Subcellular distribution of GTI-2040 in the nucleus, cytoplasm and in whole cell lysis from patients bone marrow cell samples...... 147

Figure 4.9 Scatter plots of the changes of R2 mRNA levels in AML patients sorted in two age groups at 24 hrs (A) and 120 hrs (B) following continuous infusion of GTI- 2040. The horizontal bars show the median values of the fold change of R2 mRNA...... 148

Figure 4.10 R2/GAPDH quantification by immunobotting in bone marrow cells from patients treated with GTI-2040 for 1,5 and 32 days...... 149

Figure 4.11 Scatter plots of R2 protein levels in AML patients sorted in two age groups at 24 hrs (A) and 120 hrs (B) following continuous infusion of GTI-2040. The horizontal bars show the median values of % R2/GAPDH at each time point. ....150

Figure 4.12 Comparison of the scatter plots of nuclear IC of GTI-2040 in patients with R2 mRNA down-regulation to those in the up-regulated patients after 24 hr treatment...... 151

Figure 4.13 Correlation of R2 protein changes with change of R2 mRNA levels following 24 hr infusion of GTI-2040 in BM cells from AML patients...... 152

Figure 4.14 Comparison of R2 mRNA levels at 24 and 120 hrs after GTI-2040 treatment xix in the responders and in non-responders from patients <60 years old...... 153

Figure 4.15 Comparison of R2 protein levels at 24 and 120 hrs after GTI-2040 treatment in the responders and in non-responders from patients <60 years old...... 154

Figure 4.16 Comparisons of baseline copy numbers of R2 mRNA in two age group of patients (A), in patients with down- or up-regualltion of R2 mRNA (B) and in responsed or unresponsed patients...... 155

Figure 5.1 Models used in GTI-2040 population PK studies in AML patients...... 186

Figure 5.2 Goodness-of-fit diagnostic plots of base plasma population pharmacokinetics...... 187

Figure 5.3 Plots of residuals of plasma PK parameter versus the correlated covariates of GTI-2040 in AML patients...... 188

Figure 5.4 Goodness-of-fit diagnostic plots of plasma GTI-2040 population pharmacokinetics model incorporated with covariates...... 189

Figure 5.5 Individual fittings of GTI-2040 plasma concentration-time profiles in two patients using the Bayes estimated individual PK parameters from the plasma final popPK model...... 190

Figure 5.6 Goodness-of-fit diagnostic plots of base PBMC population pharmacokinetic model of GTI-2040...... 191

Figure 5.7 Plots of residuals of PBMC PK parameter versus the correlated covariates of GTI—2040 in AML patients...... 192

Figure 5.8 Goodness-of-fit diagnostic plots of population pharmacokinetic model of GTI- 2040 in PBMC in AML patients...... 193

Figure 5.9 Individual fittings of GTI-2040 concentration-time profiles in PBMC on two patients using the Bayes estimated individual PK parameters from the final popPK model of GTI-2040 in PBMC...... 194

Figure 5.10 An indirect response PK-PD linked model used for the fitting of R2 mRNA down-regulation by GTI-2040...... 195

Figure 5.11 PK/PD model fitting results of GTI-2040 plasma concentration-time profiles and the time course of R2 mRNA levels using a two-stage method...... 196

Figure 5.12 Effects of dose on GTI-2040 plasma concentrations and the change of R2 mRNA levels...... 197

Figure 5.13 Effects of change of PK/PD model parameters on GTI-2040 concentration and R2 mRNA levels...... 198

Figure 5.14 Effects of PK/PD model parameters on the change of R2 mRNA levels. .199

Figure 5.15 Effects of model parameters on the change of R2 mRNA levels...... 200

Figure 6.1 IP-RP-HPLC-UV chromatogram (A) and LC-MS total ion (TIC) chromatogram (B) of standards of GTI-2040, 3’ N-1, 3’ N-2, and 3’N-3. UV was set at 260 nm...... 224

Figure 6.2 Total ion chromatogram and mass spectra of GTI-2040 and major metabolites obtained from the plasma extract of a patient treated with GTI-2040...... 225

Figure 6.3 The ESI LC-MS mass spectrum (A) and the corresponding deconvoluted mass spectrum (B) of GTI-2040...... 226

Figure 6.4 Fragment ions assignments following collision induced dissociation of ions with m/z 1004 from the mass spectrum of (A) 3’N-1 standard and M1 peak, and from the mass spectrum of (B) 5’ N-1 standard...... 227

Figure 6.5 HPLC-UV chromatogram (A) and total ion LC-MS chromatogram (B) of GTI- 2040 and metabolites after 5μM of GTI-2040 was incubated in SVP (0.5 U/mL) for 10 hrs...... 228

Figure 6.6 HPLC-UV chromatograms of the sequential metabolism profiles of GTI-2040 (A), 3’N-1 (B) or 3’N-2 (C) after each of them was incubated in 0.5 U/mL SVP solution...... 229

Figure 6.7 HPLC-UV chromatogram of GTI-2040 and metabolites after 5μM of GTI- 2040 was incubated in human blood for 24 hrs...... 230

Figure 6.8 Metabolism profiles of GTI-2040 in mouse liver homogenates (A,B) and in xxi human liver microsomes (C,D)...... 231

Figure 6.9 Metabolism profiles of GTI-2040 in mouse kidneys homogenates (A, B) and in rat urinary excretion (C, D) ...... 233

Figure 7.1 Non-specific binding of GTI-2040 in human liver microsomes (HLM) (0.2 mg/mL) ...... 258

Figure 7.2 HPLC chromatograms of GTI-2040 and its metabolites in Mobile Phase A (MPA) containing 3’ exonuclease (SVP) ...... 259

Figure 7.3 Plot of formation rate of 3’N-1 versus GTI-2040 substrate concentration (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in 0.3 U/ml of SVP solution...... 260

Figure 7.4 Plot of total formation rate of 3’N-1 and 3’N-2 versus GTI-2040 substrate concentration (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in 0.3 U/ml of SVP...... 261

Figure 7.5 HPLC chromatograms of GTI-2040 and its metabolites in human liver microsomes (HLM)...... 262

Figure 7.6 Plot of formation rate of 3’N-1 versus total GTI-2040 substrate concentrations (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in H Each data point represents the mean of duplicate determinations LM (0.2mg protein/ml)...... 263

Figure 7.7 Plot of total formation rate of 3’N-1 and 3’N-2 versus total GTI-2040 substrate concentrations (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in HLM (0.2mg protein/ml)...... 264

Figure 7.8 Plot of total formation rate of free 3’N-1 and 3’N-2 versus free substrate concentrations of GTI-2040 (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in HLM (0.2mg protein/ml)...... 265

xxii

LIST OF SCHEMES

Scheme 2.1: Illustrations of hybridization ELISA...... 64

Scheme 2.2: The working procedure of hybridization ELISA assay to measure ...... 65

Scheme 4.1: Protocol of GTI-2040 combined with cytarabine in phase I trial in the treatment of acute myeloid leukemia...... 139

xxiii CHAPTER 1

BACKGROUND AND INTRODUCTION

1.1 BACKGROUND

With the completion of the Human Genome Project in 2003, about 30,000 gene

sequences and 100,000 mRNA are now available as tools to validate candidate genes and

for therapeutic purposes [Human Genome Project, link:

http://www.ornl.gov/sci/techresources/Human_Genome; (1)]. Antisense therapy has

emerged as a promising gene-targeting strategy to specifically inhibit the gene

expressions. Antisense oligonucleotides are short, synthetic DNA molecules, which have

sequences that are complementary to their target mRNAs. Once hybridized with its

mRNA complement, the oligonucleotides can interfere with the transcription or the

translation of mRNA, thereby preventing the target genes from conversion to the proteins

(2,3). Similar to the action of small molecular drugs working by binding to their

receptors, mRNA can be considered as a receptor for the antisense oligoucleotide and

they interact through the Watson-Crick hybridization. Post-binding

1 events in blocking the function of receptor mRNA can be considered to be pharmacological effects. The sequence-based hybridization specificity of antisense oligonucleotides has made them attractive as therapeutics to selectively inhibit the action of disease-related genes (4). Although natural phosphodiester oligonucleotides are capable to suppress gene expressions, their activities were limited by their susceptibility to degradation by cellular nucleases (5). Chemical modifications in sugar, base and backbone have been developed to improve the stability of antisense oligonucleotides.

A large percentage of antisense oligonucleutides have been developed in cancer therapy. Although chemotherapy has shown its significance in anticancer treatment, the high toxicity and inefficient timelines in the discovery of conventionally chemotherapeutic drugs has prompted alternative strategies in cancer therapy (6). Since the discovery of oncogenes over 25 years ago, there has been tremendous progress in identifying genes that are important in the regulation of neoplastic transformation and tumor growth (6). Elucidation of the roles of cancer-related genes in tumor development has created a wealth of potential targets for molecular intervention strategies including antisense therapy. The most promising targets are those involved in cell proliferation, apoptosis, angiogenesis and metastasis (1). Antisense therapy represents a promising genetic intervention strategy for the treatment of cancer. Compared to the conventional anticancer treatment, antisense oligonucleotides show high selectivity and less side effects. In addition, the timeline for the identification of an antisense oligonucleotide drug is rather short and many nuclear-encoded genes could be targeted (7). To date, over

50 antisense oligonucleotides are being evaluated in clinical trials, and more than 12 are

2 used for the treatment of cancers (1). And there is one antisense drug, Vitravene

(fomivirsen) has been approved by US Food and Drug Administration for the treatment

of cytomegalovirus (CMV) retinitis (8). Examples of some ODNs that were approved or

are under clinical trials are listed in Table 1.1. Among them, an anti-Bcl-2 PS-ODN,

G3139, showed encouraging responses for the treatment of solid tumors, multiple

myeloma, malignant melanoma, leukemia, non-small-cell and small-cell lung cancer and

prostate cancer (9). Recently, a PS-ODN targeting the R2 subunit of ribonucleotide

reductase (RNR) named GTI-2040 has been developed for the treatment of solid tumors

and leukemia and is currently under clinical evaluation (10).

Although there is an increasing number of clinical trials with antisense

compounds directed at various targets in various diseases including cancer, viral disease and inflammatory disorders, questions remain on how to better translate its in vitro

antisense activity into a clinical setting, especially the understanding of their

pharmacokinetics (PK), pharmacodynamics (PD) and intracellular kinetics. It has been reported that PK/PD properties of AS-ODN and its intracellular uptake and subcellular trafficking processes are important factors to evaluate the antisense effects in patients

(11,12). The main focus of this project is to investigate the clinical PK/PD, cellular uptake and metabolism of a phosphorothioate oligonucleotide, GTI-2040 in patients with

AML.

In this chapter, mechanisms of antisense activity, PK/PD and metabolism of phosphorothioate oligonucleotides will be reviewed. Using ribonucleotide reductase as a

3 target in the treatment of acute myeloid leukemia will be discussed. Discovery and the

preclinical experiences with GTI-2040, as a specific RNR inhibitor, will be reviewed.

1.2 INTRODUCTION

1.2.1 Principle of antisense therapy

Antisense oligonucleotides (AS-ODNs) are short single strand DNA molecules usually containing 16-25 nucleotides that hybridize with specific mRNA strands corresponding to target genes (3). By binding to the mRNA, the antisense oligonucleotides prevent the sequence of the target gene being translated into proteins, thereby blocking the action of the genes. Antisense ODNs are specifically designed to bind to the complementary mRNA by Watson-Crick base-pairing. In the human genome,

there are 30,000 genes and 3.16 × 109 nucleotide. The specificity of antisense approach is

based on the estimate that, assuming a random distribution of four nucleotide (A, T, G,

C), a sequence of 17 nucleotide is expected to occur only once (frequency is 1 out of 1.7

× 1010) within the human genome (13)

1.2.1.1 Mechanisms of action of antisense oligonucleotides

There are multiple mechanisms by which AS-ODNs down regulate the target

mRNAs (Figure 1.1). The major mechanisms include: 1) cleavage of RNA/DNA hybrids

by recruitment of RNase H (14) (Figure 1.1B), 2) induction of DNA translation arrest by

a steric blockade interaction of mRNA with ribosome (3,15) (Figure 1.1C), 3) modulation

4 of mRNA splicing (16) (Figure 1.1E), and 4) formation of DNA triplexes by binding to

DNA, resulting in the prevention of transcription (Figure 1.1A) (17).

Degradation of target mRNA by endogenous RNase H is probably the most

important mechanism of antisense action, thus underlying the activity of almost all of the oligonucleotides being evaluated in clinical trials thus far (4). RNase H is a ubiquitous enzyme that cleaves RNA strand in the form of a DNA/RNA heteroduplex (18).

Degradation of RNA causes the release of ODN, which will bind to the next target mRNA and triggers a new round of RNase H action (Figure 1.2). This process could continue before the inactivation of ODNs. Therefore, a sustained decrease in target mRNA translation would ultimately result in a reduced synthesis of the corresponding disease-related protein (4). RNase H is found both in the cytoplasm and the nucleus with higher levels in the nucleus. The precise recognition mechanisms for RNase H are unknown. However, a duplex forming with mRNA and DNA-like antisense molecule appears necessary to activate RNase H (19). Substitution of the nonbridge oxgen atom with sulfur in the phosphate backbone (phosphorothioate oligonucleotides) retains the ability to activate RNase H, while modifications of the 2’-H group in the ribose structure of ODNs which become RNA-like oligonucleotides abrogates their ability to activate

RNase H.

Antisense oligonucleotides that do not induce RNase H cleavage may elicit their antisense activity through translational arrest by blocking the assembly of the translational machinery or impeding the progress of translation in the coding region (8).

In addition, regulation of RNA processing is another possible mechanism for AS-ODNs

5 to interfere gene expression. For example, pre-mRNA must remove the introns and

assemble exons into a finished strand of mRNA that is ready for translation. AS-ODNs

can target to aberrant splice sites generated by the mutations and restore correct splicing

and therefore increase the levels of correctly spliced pre-mRNAs (16). AS-ODNs binding to the pre-mRNA may also affect the regulation in 5’-capping and 3’-adenylation and result in the altered mRNA processing (20).

1.2.1.2 Antisense oligonucleotides modifications and other anti-mRNA strategies

Since unmodified oligonucleotides are rapidly degraded in biological fluids by

nucleases, a vast number of chemically modified nucleotides have been developed to

improve the stability of ODNs. In general, three types of modifications of ribonucleotides

can be distinguished: analogs with unnatural bases, modified ribose (especially the 2’

position of the sugar) or altered phosphate backbones (Figure 1.3). Furthermore, a variety

of modifications in the heterocycles have been introduced into AS-ODNs to strengthen

base-pairing, thus the binding affinity between the AS-ODNs and their target mRNAs

(21). Examples of the three generations of structure modifications in AS-ODNs are

shown in Figure 1.4. In general, a potent modified antisense oligonucleotide should have

resistance to nuclease, high affinity for the target mRNA and the capability to cleave the

nucleic acid target. In addition, the modifications should also improve the cellular uptake

and distribution and provide favorable pharmacokinetic properties for tissue distribution,

metabolism and clearance (22).

6 1.2.1.2.1 First generation of antisense oligonucleotides

Phosphorothioate oligodeoxynucleotides (PS-ODNs) are the best known first

generation DNA analogs and are the most widely used AS-ODNs to date. In this class of

ODNs, one of the nonbridging oxygen atoms in the phosphodiester bond is substituted

with a sulfur and form such called phosphorothioate linkage. The sulfuration in the

phosphate backbones greatly enhances their nuclease resistance and prolongs their human

serum half-life to approximately 9-30 hr compared to ~ 1hr for the unmodified

oligodeoxynucleotides (13). In addition, PS-ODNs form regular Watson-Crick base pair

and are capable to activate RNase H for the catalysis of the RNA/DNA hybrids. The negatively charged property of PS-ODNs makes them prone to protein binding, which has proven to be advantageous for their pharmacokinetic profile. PS-ODNs also bind to a wide variety of proteins particularly heparin-binding proteins, which may relate to their sequence-independent side effects or non-antisense effect such as complement activation, thrombocytopenia, inhibition of cell-matrix interaction, and reduction of cell proliferation

(4,13). Due to its favorable PK properties and high antisense activity, PS-ODNs represent the most widely used class of antisense compounds among the AS-ODNs currently under clinical trials (10,12,23-25). Some examples of the PS-ODNs being evaluated in clinical trials are shown in Table 1.1. Over 3000 patients have been exposed to PS-ODNs in 12 clinical trials and demonstrated tolerable toxicity without any indication of activation of the alternative complement cascade (6). As an example, an 18- mer PS-ODN, Oblimersen (G3139), represents an attractive therapy for leukemia through targeting antiapoptotic mRNA, Bcl-2. In one of the Phase I clinical trials conducted at

The Ohio State University, combination of G3139 with fludarabine, cytarabine, and 7 granulocyte colony-stimulating factor was found feasible in patients with refractory or

relapsed acute leukemia and demonstrated in vivo down-regulation of the Bcl-2 target and disease response with manageable toxicity (12,26). The concentration of G3139 in cells from bone marrow was found to correlate with the disease responses (11,12).

1.2.1.2.2 The second generation of antisense oligonucleotides

The second generation of antisense oligonulceotides focused on the sugar

modifications. The most promising ribose modifications involve the introduction of a

methyl at the 2’ position of sugar (2’-O-methoxy-ethyl or 2’-O-MOE) and form a RNA-

like oligonucleotide. AS-ODNs made from alkylation modification are less toxic than

PS-ODNs and have a significantly enhanced affinity towards their complementary RNA

(27). In mice, rat, and monkeys, the elimination half-life of 2’-O-MOE ODNs is nearly

30 days in plasma and several organ tissues (28). The substantial stability of 2’-O-MOE,

however, is counterbalanced by the shortcoming that this class of compound cannot

induce RNase H cleavage of the target RNA. Their antisense effects are mainly due to the

steric blockage of translation or splicing modulation (29). Since for most antisense

approaches, it is desirable to stimulate RNase H in order to increase antisense potency,

the use of 2’-O-MOE chimeras (a gapmer approach) has been suggested as a solution to

the loss of RNase H stimulation effect (30). This gapmer was designed to contain mixed-

backbone that only has the 2-MOE modified nucleotides at both ends, but leaving the

unmodified PS-ODNs in the center. The DNA-like gap in the middle will retain the

capability to recruit RNase H, while the MOE nucleotides at both terminals provide

enhanced stability and binding affinity of the chimeric compound (13). For example, in 8 ISIS 104838, a 2’-O-MOE chimeras, wings of five 2’-O-MOE nucleotides at both

terminal block prevent nucleolytic degradation and the contiguous stretch of 10 PS-

ODNs at the center was reported to be sufficient for activation of RNase H (23).

Moreover, their dramatically prolonged half-life and the substantial stability make them feasible of monthly dosing and oral administration.

In recent years, novel technologies in modified nucleotides have been developed

to further improve properties such as target affinity and nuclease resistance. With this in

mind, a completely different chemical moiety modification in phosphate linkages or

ribose as well as nucleotides has been made, which has led to the advanced generation of

antisense oligonucleotides. For examples, peptide nucleic acids (PNAs) are designed by

replacement of deoxyribose phosphate backbone by polyamide linkage, which result in

favorable hybridization properties and high biological stability (31). Other novel

chemical modifications in ribose or phosphate backbones, such as locked nucleic acid

(LNA), morpholino oligonucleotides improved the nuclease resistance and stabilized the

duplex between the AS-ODNs and the mRNA (32). However, a major inherent

disadvantage of all of the novel modifications in the ribose is their inability to activate

efficiently RNase H cleavage to the target RNA.

1.2.1.2.3 Other anti-mRNA strategies

One of the other RNase-dependent antisense mechanisms that have recently

received much attention is interference RNA or RNAi. RNAi is initiated by long double–

strand RNA (ds-RNA), which are processed into 21-23 nucleotide long RNAs called

small interference RNA (siRNA) by the endogenous Dicer enzyme. These siRNAs are 9 then incorporated into the RISC (RNA-induced silencing complex), RISC then separates the two strands of RNA molecules and guides the antisense strand to bind to the targeted mRNA. The RISC also contains endonuclease activity, which hydrolyzes the target RNA in the form of the hybrid duplex with antisense strand. siRNA acts on a mechanism similar to antisense oligonucleotides, which binds to the target RNA by Watson-Crick base pairing rules and use ribonuclease to degrade the target RNA (33,34). However, siRNA is distinct from the antisense mechanism in that single strand RNAs are used as antisense molecules which activate the double strand RNase, when binding to the target mRNA. siRNA is much more potent in the down-regulation of target mRNA than AS-

ODN and needs a significantly lower concentration than the AS-ODNs (100 fold lower in

IC50) to achieve the comparable knock-down effect in cells (35). siRNA has already become a powerful tool in genetic silencing due to its ability from transient to lost-of- function knock-down on the target mRNA. However, since this technology is new and has only emerged in the last 5 years, there are many unknown issues that need to be answered before reaching the in vivo application. One of the main challenges is its in vivo delivery as a duplex RNA. The mass of a synthetic duplex RNAs is twice that of a traditional AS-ODNs, which makes their cellular uptake more difficult. Furthermore, the presence of a second nucleic acid strand increases the potential for off-target effects (36).

Therefore, until now it is not clear whether or not siRNA possesses advantages over the traditional AS-ODNs for in vivo experiment and in therapeutics. Its pharmacological and pharmacokinetic studies, which have been extensively studied on AS-ODN, must be performed.

10 MicroRNAs (miRNA) are a recently discovered family of 22 nt short non-protein-

coding RNAs that negatively regulate gene expression endogenously (37). Most

microRNAs in animals are thought to inhibit the effective mRNA translation through

imperfect base pairing with the target mRNAs. Currently, it is estimated that there are at

least 200 microRNA genes in the human genome (38,39). The finding that microRNAs

are misexpressed in cancers has suggested their regulatory roles in human disease and

may imply a new avenue in gene therapy development (37).

1.2.2 Pharmacokinetics and pharmacodynamics of phosphorothioate oligonucleotides

Since the phosphorothioate oligonucleotides are the most widely used AS-ODNs

in the clinic to date, and in this dissertation, I will focus on one of the first generation of

antisense oligonucleotides, GTI-2040, in the introduction chapter, I will briefly review

the PK/PD characteristics of phosphorothioate oligonucleotides.

1.2.2.1 Cellular uptake of PS-ODNs

To fulfill the antisense function, antisense drugs have to reach the cytoplasm and

finally the nucleus to efficiently interact with their mRNA targets. Uptake of PS-ODNs in cell lines has been found to be dose-, time-, and temperature- dependent (40,41). Some studies observed receptor mediated or adsorptive endocytosis for PS-ODNs at relatively

low concentrations (<1μM) and fluid phase pinocytosis at high concentrations (42,43).

Possible mechanisms and pathways for cellular uptake and subcellular trafficking of PS-

ODNs are illustrated in Figure 1.5. Most PS-ODNs are rather hydrophilic with an anionic

backbone, resulting in low cell internalization in cell cultures (44). Transfection reagents 11 are usually used to enhance the cellular uptake of PS-ODNs in cell cultures. The most

commonly and successfully used delivery systems are liposomes and cationic lipids,

which can either encapsulate nucleic acids within their aqueous core or through lipid-

nucleic acid complexes as a result of charge neutralization (45,46). Theses complexes are

usually internalized by endocytosis and then undergo dissociation inside the cells to

release the PS-ODNs (45,47).

Interestingly, in animal models and in patients, all therapeutically active ODNs

have been administered as naked (noncomplexed) compounds, which appear to have

sufficient uptake of drugs at target locations. For example, down-regulation of Bcl-2

mRNA and robust intracellular uptake of free G3139 were achieved in mononuclear cells

from bone marrow in treated AML patients, while only low intracellular concentrations of G3139 were found and no suppression of Bcl-2 mRNA was observed when free

G3139 was exposed to K562 cells (11). Hypothesis to explain the observed discrepancy between the in vitro and in vivo settings are provided here: 1) In vivo, certain factors, e.g.

extracellular matrix and/or cell membrane binding proteins may interact with ODNs and

facilitate its endocytosis (41,48); in vivo, albumin is the most abundant protein bound to

ODN but with low binding affinity (Kd ~20-200 μM) (49). Uptake of albumin-ODN

complex has been reported through active endocytosis (50). ODNs may shuttle from low-

affinity high-capacity binding sites in circulation to higher affinity, lower capacity sites in tissues or cells (48). This shuttling process may involve the uptake of ODNs into the cells in vivo. 2) A two-dimensional culture environment in cell lines may not reflect the real situation in tissues in vivo, where cells are engaged in an appropriate interaction with a 3-dimensional extracellular matrix. This may result in fundamental differences in 12 oligonucleotide uptake and transport between in vitro and in vivo (51,52). 3) In vivo, the

prolonged period of drug administration may allow build-up of oligonucleotide in

cytoplasm and the nucleus, resulting in a sustained effect on the down-regulation of

targets. In cell culture, the relatively short exposure of free oligonucleotides by cells may

not be sufficient to permit significant uptake and production of antisense action, and the

long-term drug treatment on cells may affect the cell viability (52).

Cellular uptake of PS-ODNs shows a large variability, dependent on cell types

(53). Therefore, for a specific PS-ODN, examination of its uptake in target cells and

study of its distribution pattern in cell culture are necessary to evaluate the antisense

effect.

1.2.2.2 Pharmacokinetics of phosphorothioate oligonucleotides

In general, pharmacokinetics of modified ODNs appears to be largely influenced

by their plasma protein binding property. PS-ODNs bind nonspecifically to a wide

variety of proteins in plasma with greater ability and capacity than their phosphodiester

counterparts (48,54). At clinically relevant doses with steady-state concentration at

higher than 100 nM levels, >96% of PS-ODNs in plasma are bound to proteins in mice,

monkeys, and humans (48,54). Albumin and α2-macroglobulin are the primary partners for PS-ODN binding in plasma (4,55). The apparent binding affinity for albumin was determined to be 17.7 μM for a PS-ODN drug, ISIS 2302, in the rat plasma. It has been reported that total clearance in plasma and the extent of urinary excretion correlate with plasma protein binding of PS-ODN (48). Results from the studies by Gear RS et al showed that the plasma protein binding for an ODN with full PO-ODNs (ISIS 16952) or 13 a mixture of PS/PO backbones (ISIS 18268) were determined to be 79.6 or 88%, respectively, which corresponded to 45% and 16% urinary excretion (48). While an ODN with full phosphorothioate linkages (ISIS 2302) were 99% bound to proteins and showed

< 0.3% of urinary excretion (48). Therefore, binding to serum proteins prevents a rapid clearance of PS-ODNs by glomerular filtration, thus providing long residence times for these drugs to allow for tissue distribution (44). Also the urinary excretion of PS-ODNs is considered to be a minor elimination pathway.

Pharmacokinetics of PS-ODNs has been studied in mouse, rat, monkey and human (48,56-59). Plasma pharmacokinetic parameters scale well across species as a function of the body weight alone (57). Following intravenous administration, the distribution phase of PS-ODN is rapid, with an initial half-life of <0.5 hr. Plasma concentration-time profile is multi-exponential with a terminal elimination half-life of

10-60 hr, depending on the sensitivity of the bioanalytical methods used in measuring the elimination phase (55,59-61). The disappearance of PS-ODNs from plasma occurs as a result of distribution to tissue and to a lesser extent the metabolism and urinary excretion

(55,62). Dose-dependent pharmacokinetics was observed in high dose PS-ODNs (20 mg/kg) in mice and monkeys (63). The observed non-linear increase in AUC and decrease in CL may result from a saturation process under the high dose including the saturated albumin binding, saturated tissue distribution and enzyme metabolism (63).

However, the non-linear pharmacokinetics was not observed in patients at the clinically relevant doses (12).

PS-ODNs are distributed widely in the body and accumulate in most tissues.

Liver, kidney, spleen, bone marrow and lymph nodes are the major organs that contain 14 the highest amounts of PS-ODNs (53,55). PS-ODNs do not cross the blood-brain barrier.

The distribution of PS-ODNs into the liver has been studied extensively. Non- parenchymal cells such as the endothelial cells and Kupffer cells contain more PS-ODNs than the hepatocytes. Drug accumulation in the hepatocytes became significant when the non-parenchymal cells were saturated with PS-ODNs at high doses (64,65). Kidney is another organ retaining high amounts of PS-ODNs. Urinary excretion is usually <1% and increases when the protein binding becomes saturated at high doses (60,66). It has been observed that PS-ODNs are filtered by the glomerulus and then reabsorbed by the proximal tubule epithelial cells. In the kidney, proximal tubule cells accumulate more PS-

ODNs (67-69). Elimination of PS-ODNs from organ tissues is much slower than plasma with the elimination half lives from the liver and kidney prolonged to 2-3 days (56).

Based on these previous studies, it is expected that different types of cells will have different affinities and uptake capacities for PS-ODNs. It is important to investigate the cellular uptake and subcellular distribution of PS-ODNs from the in vivo samples and this information may help the evaluation of the antisense effect. To determine the intracellular concentrations and to define the terminal elimination phase of PS-ODNs in plasma, a highly sensitive and specific assay is required. However, most of the conventional analytic methods such as HPLC, CGE and radio labeled ODNs are not feasible due to their insufficient sensitivity and lack of specificity (55,70-72). An specific ELISA method based on the hybridization between PS-ODNs and their sense template has been developed in this dissertation and has been applied to quantify PS-ODNs in various biological matrices with a low limit of quantification at 50 pM (11,73,74).

1.2.2.3 Metabolism of phosphorothioate oligonucleotides

Degradation of oligonucleotide is usually mediated through the hydrolysis of the phosphate bond by the action of exonucleases or endonucleases (Figure 1.6).

Exonucleases cleave single nucleotide one by one from either the 3’ or 5’ end of the molecule (Figure 1.6a, 1.6b) (75). Metabolism by endonuclease would be expected to produce a profile characterized by an irregular distribution of shortened metabolites

(Figure 1.6 c) (76). Studies with various PS-ODNs using LC/MS, MALDI-TOFMS, CGE and end-labeling techniques indicate the predominantly 3’ exonucleolytic cleavage, generating a metabolic pattern of progressively chain-shortened metabolites (64,77-79).

Cleavages of PS-ODNs by 5’exonuclease and endonuclease are considered minor metabolic pathways (77). An example of the strategy of PS-ODN metabolism is illustrated on the metabolism study of a 18-mer PS-ODNs, G3139, in plasma and urine using LC/ESI-MS (78). In this study, identification of metabolites was accomplished by the individual masses and possible nucleotide composition of metabolites in each HPLC peak as determined by deconvolution. The identity of metabolites was further confirmed by the fragment ion patterns generated by LC/MS/MS (78,80). A number of 3’ chain- shortened metabolites of G3139 were identified in plasma and urine from treated patients and rats (78).

Hydrolytic cleavage of PS-ODNs has been shown to be stereoselective with respect to the phosphorothioate center. The sulfurization process of ODN creates a random distribution of R and S chiral centers at each of the phosphorothioate linkage

(Figure 1.7), which result in the stereoselectivity of 3’-exonuclease degradation of PS 16 dimers (76,81). It has been shown that when synthesized 3’ pure Sp or Rp nucleotides were incubated with 3’ exonuclease, PS-ODNs with the Rp configuration were degraded

more than 300 times faster than the Sp isomers (81,82). Additionally, an in vitro study

using rat liver homogenates showed that the rat liver nuclease preferentially digested

phosphorothiate in the Rp and Rp/Sp configuration, while the Sp isomer was barely

metabolized (77). The significant resistance of Sp diastereoisomer of PS-ODNs toward 3’

exonuclease may suggest an approach for stabilizing the antisense oligonucleotide based

on a diastereoisomer control of the 3’–end linkage. However, it has been found that the

Sp isomers may be more sensitive toward the endonuclease, which will limit the utility of an all S-phosphorothioate oligonucleotide in practice (77).

Oxidative products were also reported in PS-ODN metabolism (77). Sulfur oxidation at the phosphorothioate linkage by the CYP450s activity was suggested (83).

However, this finding was not consistently observed in other PS-ODNs and may be artificially generated during in sample processing and handling (77,83,84).

1.2.2.4 Pharmacodynamics of phosphorothioate oligonucleotides

In vivo antisense effect would be best evaluated by the characterization of pharmacodynamics of the target mRNA and its correlations to the drug level and the therapeutic effects (52,85). To date, the level of understanding the pharmacodynamics of

the target mRNA in antisense therapy is inadequate compared to the pharmacokinetics of

ODNs, although progresses have been made in this area recently (52). For example, time

course of MAO-B mRNA expression in the rat striatum following antisense perturbation

was measured by RT-PCR and showed a time-dependent mRNA down-regulation (86). 17 In a phase I trial using a chimeric antisense drug OGX-011 for the treatment of prostate

cancer, a decrease of the targeting clusterin expression was observed in prostate tumor

tissue and lymph nodes (87). Additionally the downregulation of Bcl-2 mRNA was observed in leukemia patients treated with G3139 (11,12,26).

Moreover, the level and the duration of drug at the target site are important to

correlate to the antisense activity (52,85). For example, pharmacodynamics in the down-

regulation of Fas mRNA after administration of ISIS 22023 in mice correlated better with

the drug concentrations in hepatocytes, rather than its plasma concentrations. Using an

indirect response model, concentrations of ISIS 22023 in hepatocytes reasonablly

predicted the time course of change of Fas mRNA and in concordance with the

observations (58). In AML patients treated with G3139, disease response was found to

correlate to IC of G3139 in cells from bone marrow rather than its plasma concentrations

(12,26). Therefore, determination of intracellular drug concentrations is desirable to

evaluate the PK/PD correlation.

1.2.3 Inhibition of ribonucleotide reductase as a strategy in anticancer therapy

Ribonucleotide reductase (RNR) plays a unique and important role in nucleotide

metabolism, as it is the only highly regulated enzyme involved in the de novo synthesis

of all of the precursors used in DNA synthesis. RNR catalyzes the reaction in which 2’-

deoxyribonucleotides (dADP, dGDP, dUDP, and dCDP) are converted from the

corresponding ribonucleoside 5’-diphosphates (ADP, GDP, UDP, and CDP) (Figure 1.8)

(88-90). This step is the rate-limiting reaction in the production of 2’-

deoxyribonucleoside 5’-triphosphate required for DNA replication (89). This enzyme is 18 closely linked to proliferative status in tumor cells and it becomes a target for anticancer

drug development. Human RNR consists of two protein components, R1 and R2

subunits. R1 is a Mr 160,000 dimer that contains effector-binding sites, and R2 is a Mr

78,000 dimer that contains a non-heme iron that participates in catalysis through forming a free radical on the aromatic ring of a tyrosine residue (91). Expression of both the R1

and R2 genes is required for the enzyme activity of RNR. The level of R1 protein

remains relatively stable throughout the cell cycle, whereas R2 is only expressed during

late G1/early S phase, when DNA replication occurs (92). An R2 analog, p53R2 was a

newly identified subunit of RNR that is induced by DNA damage and supply precursors

for DNA repair in a p53-dependent manner. It appears that R2 provides normal

maintenance of dNTPs for DNA replication during the S-phase in a cell cycle-dependent

manner, while p53R2 is responsible for the production of dNTPs for DNA repair in G0-

G1 cells in a p53-dependent pathway (93,94).

Over-expression of R2 has been found in almost every type of cancer studied

(95). The R2 protein also appears to relate to the malignant status of tumor cells via the

cooperation with a number of activated oncogenes. For example, enhanced expressions of

oncogenes, such as, v-fms, v-src, A-raf, v-fes, c-myc in tumor cells are associated with

over-expressed R2 (96). In addition, over-expression of R2 mRNA and RNR enzymatic

activity contributed to the 3-fold increase in tumor’s invasiveness in R2 gene transfected

cells compared to the KB wild-type cells (97). Several agents possessing inhibitory

activity on RNR are currently clinically used. These include hydroxyurea, gemcitabine

and fludarabine (89). However, gemcitabine and fludarabine are not specific RNR

inhibitors, and treatment with these drug results in significant side effects (95). 19 Hydroxyurea is a reversible inhibitor of RNR that requires relatively high concentrations and may develop chemoresistance (98). An increased expression of R2 has been also

found to be associated with drug resistance of cancer cells. An enhanced R2 subunit

alone was sufficient to increase resistance of hydroxyurea in tumor cells (99), which

makes R2 a good target of anticancer drugs. This resistance could be reversed by

suppression of the R2 expression using interference RNA, resulting in the enhanced

chemosensitivity of gemcitabine to pancreatic adenocarcinoma cells (100).

Taken together, these studies indicate that inhibition of R2 expression would

likely provide antiproliferative and antineoplastic benefits for neoplastic diseases, and for

future design and development of specific R2 inhibitors is desirable.

1.2.4 GTI-2040-a specific inhibitor of R2 subunit of RNR

GTI 2040 is a 20-mer oligonucleotide that is complementary to the coding region

of R2 mRNA with sequence of 5’-GGCTAAATCGCTCCACCAAG-3’ (Figure 1.9). It is

selected from 102 AS-ODNs complementary to RNR R2 subunit mRNA through in vitro

screening. GTI 2040 demonstrated sequence- and target- specific down-regulation effects

of mRNA and the R2 protein in several tumor cell lines including human H460 lung

carcinoma cells, human T24 bladder cancer cells and murine L cells. A GTI-2040 analog

containing a four base mismatch (mismatch control) and an analog with the same ratio of

ATCG but with scrambled sequence (scrambled control) did not produce down-

regulation of the R2 mRNA or protein therefore verifying the sequence-specific antisense

effect of GTI-2040. The antitumor activity of GTI-2040 has been shown in tumor bearing 20 animal models. In nude and SCID mice, activity of GTI-2040 was tested against a wide

range of human xenograft tumors, including colon, pancreatic, liver, lung, breast, renal, prostate cancers. In these animal models, GTI-2040 acted in a sequence-specific, dose-

dependent manner to down-regulate R2 expression with a concomitant decrease in tumor

growth, metastasis and an increase in animal survival (95). GTI-2040 was also found to

increase survival in animals with lymphoma and leukemia. In mice implanted with

erythroleukemia, GTI-2040 in normal saline was administered by tail vein injections

every second day at a dose of 10 mg/kg for a total of 71 days. All untreated mice died as

a consequence of tumor progression within 36 days. All GTI-2040 treated animals except

one survived beyond day 71. The antitumor activity of GTI-2040 was also investigated in

combination with standard chemo-therapeutic agents against the growth of tumor-bearing

nude/SCID mice. These chemotherapeutic compounds included mitomycin-C, CPT –11,

5-FU, vinblastine, IL-2, mitoxantrone, dacarbazine and paclitaxel. All combination

therapies showed enhanced antitumor effects (101).

Preclinical pharmacokinetic and tissue distribution properties of GTI-2040 have

been characterized in rats and monkeys. GTI-2040 was administered to Sprague-Dawley

rats via an intravenous bolus injection every second day up to 21 days, with a 21-day

recovery period at different doses. The plasma half-life for the 50 mg/kg dose group was

determined to be 1.77 hours. Clearance was calculated to be 687 mL/kg/hr, but was

probably over-estimated due to an underestimation of AUC. The accumulation ratio,

based on the concentration at 4 hours after 50 mg/kg dosing was 1.90. GTI-2040 was

administered to monkeys by continuous intravenous infusion for 21 days followed by a

21-day recovery period. Twenty-eight monkeys were administered one of the following 21 dose levels; vehicle control, 2, 10, or 50 mg/kg/day. Plasma concentration appeared to

reach a steady state by 8 hours after dosing. The Cmax was 2.94 and 23.6 μm/mL, Css

1.88 and 14.3 μm/mL, and clearance 250 and 160 mL/hr/kg for the 10 mg/kg and the 50 mg/kg dose groups, respectively. However, due to the limited sensitivity and the insufficient specificity of the quantitation methods (capillary electrophoresis, CGE), the elimination phase might not be adequately defined and the values of the clearance from these data were only rough estimates (101).

Tissues samples from the rat and monkey, such as kidney, liver, spleen, heart, lung, bone marrow (rat), lymph nodes (monkey) and brain, were extracted and analyzed by capillary electrophoresis to determine the amounts of GTI-2040 and presumed metabolites. Tissue biodistribution following continuous infusion of GTI-2040 in the rat and monkey are consistent with observations reported for other phosphorothioate

oligonucleotides such as a 20- mer PS-ODN, ISIS 2503 (61,63,64,102). The

concentrations of GTI-2040 were found in the descending order: kidney> liver> spleen>

bone marrow> heart > lung. GTI-2040 concentrations in the brain were not detected

(101).

The promising results in the preclinical studies suggest that GTI 2040 could be

developed as a specific RNR inhibitor effective in cancer therapy as a single agent or in combination with chemotherapeutic drugs. Two Phase I studies of GTI-2040 were

initiated in patients with advanced solid tumors, lymphoma or leukemia.

In a Phase I study using GTI-2040 for the treatment of advanced solid tumors, 36

patients were administered GTI-2040 as a 21-day continuous infusion with a dose range

from 18.5 mg/m2/day to 222.0 mg/m2/day. The maximum tolerated dose of GTI-2040 22 was determined to be 222.0 mg/m2/day as a single agent in this schedule. The

recommended Phase II dose was 185.0 mg/m2/day. GTI-2040 appears to have a manageable toxicity profile and is generally well-tolerated as a single agent. Fatigue and

anorexia were the most common toxicities. Hematologic toxicity was relatively mild.

Pharmacokinetics was characterized in this trial. The mean elimination half-life of GTI

2040 was <3 hrs at a dose of 185.0 mg/m2/day as assayed by CGE (10). Plasma

clearances did not differ significantly among the tested dose groups suggesting the dose

independent PK of GTI-2040. In addition, large inter-individual variations were observed

in PK parameters among patients. Attempts to evaluate R2 mRNAs from peripheral blood leukocytes of the patients were made; however, large variations of the data precluded a meaningful interpretation (10).

Currently, a phase I clinical trial of GTI-2040 in combination with high dose cytarabine is being evaluated in patients with acute myeloid leukemia (AML) at The

Ohio State University James Cancer Hospital (Protocol number: OSU 0304).

1.2.5 Acute myeloid leukemia and cytarabine

1.2.5.1 Acute myeloid leukemia

Acute myeloid leukemia (AML) is a heterogeneous disease of the hematopoietic system characterized by a clonal expansion of immature myeloid cells in the bone marrow BM (103). The results of the uncontrolled accumulation of leukemic blasts block the production of normal marrow cells, leading to a deficiency of red cells (anemia), and platelets (thrombocytopenia) and normal white cells, especially neutrophils (neutropenia) in the blood (103). Acute myeloid leukemia is the most common type of leukemia in 23 adults, with an estimated 10,000 new cases reported each year. Older patients are more

likely to develop AML than children. In fact, the risk for developing the disease increase

about ten-fold from age 30 (1 case per 100,000) to age 70 (1 case per 10,000) (104).

Non-random chromosomal abnormalities, identified at the cytogenetic level in

approximately 55% of all adult primary AML patients, have long been recognized as one

of the most important independent prognostic indicators for achieving of complete

remission (CR). Other indicators include the duration of response and survival following

intensive chemotherapy treatment (104). AML patients are currently classified into

groups with variable prognosis based on cytogenetics and molecular analyses. For

instance, the chromosomal translocations in v(16) and t(8;21), which result in the

expression of the fusion proteins CBFB/MYH11 and AML1/ETO, respectively,

characterize AML with relatively favorable prognosis. Patients with acute promyelocytic

leukemia, an AML subtype characterized by the accumulation of t(15;17)-containing

promyelocytes, are particularly sensitive to all trans-retinoic acid (ATRA), and hence

have a good treatment outcome. Thus, t(8;21), inv(16) and t(15;17) are established

molecular indicators for favorable treatment outcome in AML (105).

Patients with AML usually receive chemotherapeutic drugs as soon as possible

after diagnosis. Common chemotherapeutic drugs given during induction therapy include

daunorubicin, doxorubicin and cytarabine (106). Despite major recent advances in the understanding of the molecular biology of the disease, the treatment of AML in adults remains challenging. Seventy-five % of AML patients older than 60 years exhibit significant toxicity with overall poor response rates and survival rates using the currently

24 available treatment. In younger patients, standard regimens using cytarabine and an

anthracycline for induction followed by some form of intensive post-remission therapy

produces response rates of 70% with 5-year relapse-free survival rates from 25% to 40%.

Overall, 20% to 30% of AML patients never achieve CR. Among patients who achieved

CR, 50 to 80 % of them would experience disease relapse (107). Further, in patients >60 years, a long-term remission is achieved in only <10% of the treated population. The lack

of significant advancements in the treatment of this disease in adults, therefore, calls for

development of novel therapeutic strategies (108).

1.2.5.2 Cytarabine

Cytosine arabinoside (cytarabinl; ara-C) is widely used as an initial or salvage treatment for AML. Cytarabine is a nucleoside analog of deoxycytidine, in which the ribose sugar has been replaced with an arabinose where the 2’ hydroxy group is oriented in the trans position (89) (Figure 1.10). Intracellular access of cytarabine is dependent upon the specific membrane nucleoside transporters such as hENT1 protein, which becomes saturated at concentrations at >10 μM, above which the uptake is via passive diffusion (109,110). Once inside the cell, ara-C is activated to ara-C triphosphate (ara-

CTP) by 3 sequential enzymes, deoxycytidine kinase, deoxycytidine monophosphate kinase, and nucleotide-disphosphate kinase. Competing with active phosphorylation, ara-

C and ara-CMP were inactivated by two other enzymes, cytidine deaminase and dCMP deaminase, respectively, to the corresponding uridine metabolites (89,110). Activation by the first anabolic enzyme deoxycytidine kinase represents the overall rate-limiting step in conversion of ara-C to its active metabolite (110). Figure 1.11 shows the metabolic 25 pathway of cytarabine in cells. The activated ara-CTP competes with the natural

deoxycytidine triphosphate for incorporation in DNA by DNA polymerase. Once

incorporated in the DNA, ara-CTP inhibits the DNA polymerases resulting in termination

of the strand elongation important for DNA synthesis (110-112). A relationship between

the intracellular levels of ara-CTP and antileukemic effect has been demonstrated. High

dose cytarabine (>2g/m2) is recommended in patients to overcome the drug resistance associated with a decrease in influx by nucleoside transporter, decreased activity of

activation enzyme and increased activity of inactivation enzymes (112,113). Greater than

100 μM plasma concentration is expected to gain effective intracellular concentrations of

active metabolites (114). Cytarabine has a potent myelosuppressive action and it is used primarily in combination with daunorubicin to treat acute nonlymphocytic leukemia (89).

Cytarabine has not been found to be active against solid tumor (89).

1.2.6 Rationale for combination therapy of GTI-2040 with cytarabine

Combination chemotherapy has been proven to have a great advantage over the single agent in cancer treatment. Experience from trials with other antisense oligonucleotides suggested that this class of compound may be best utilized in combination with chemotherapeutic agents for the treatment of cancers. In a Phase I clinical trial at The Ohio State University (Protocol number: OSU 0304), GTI-2040 was proposed to combine with high dose cytarabine in patients with refractory or relapsed

AML. We hypothesized that combination with an RNR inhibitor, such as GTI-2040, will produce an imbalance in the dNTP pool therefore limiting the competition with the endogenous dCTP for DNA incorporation of Ara-CTP and result in enhanced 26 cytotoxicity. Cytarabine not only acts as an inhibitor of DNA polymerase but also is a non-specific inhibitor of RNR and also could potentiate the activity of GTI-2040.

The dosing schedule in protocol OSU 0304 is modified from an earlier version of

the trial for patients younger than 60 years old, with GTI 2040 at 3.5 mg/kg and

cytarabine at 2000 mg/m2 q12 hours x 12 doses (101). To emphasize the safety, in OSU

0304, the frequency of cytarabine was modified to escalate the cytarabine dose from an

alternating day regimen for 6 doses to up to 10 total doses. In this trial, patients were

stratified into two different cohorts of cytarabine dose and schedule, based on the

increased toxicity seen in the elderly patients (>60 years) treated with high-dose

cytarabine. Therefore, 3.5 mg/kg/day of GTI-2040 with cytarabine dose level of 2.0 g/m2 q12 x 5 days was selected as the starting dose for patient <60 years, while for patients

>60 years a dose at 1.5 g/m2 cytarabine and 3.5 mg/kg/day GTI-2040 were initially used.

Cytarabine dosed with 2.0 g/m2 q12 x 5 was already used in combination with another

PS-ODN, G3139, and showed manageable toxicity in both younger and older patients.

1.3 SPECIFIC AIMS

The purpose of this dissertation project was to investigate the clinical pharmacokinetics, pharmacodynamics and metabolism of GTI-2040 and includes the following Specific Aims:

1. To develop and validate an ultra-sensitive and specific hybridization ELISA-

based fluorogenic assay for GTI-2040 quantification in a variety of biological

matrices.

2. To evaluate cellular uptake, intracellular distribution and pharmacology of GTI- 27 2040 in cultured leukemia cells.

3. To determine the intracellular concentration and localization of GTI-2040 in

mononuclear cells obtained from blood and bone marrow of AML patients

4. To characterize the clinical PK and PD of GTI-2040, to investigate the PK/PD

correlation and correlations with clinical response.

5. To perform population pharmacokinetics of GTI-2040 in plasma and in PBMC

and to identify PK relevant covariates.

6. To investigate the clinical PK/PD correlation through PK/PD modeling, to

identify barriers for antisense effectiveness by simulation study.

7. To identify in vivo and in vitro metabolites of GTI-2040 using an ion-pair LC/ESI

MS method developed in our laboratory.

8. To study the enzyme kinetics of GTI-2040 using appropriate metabolic enzyme

systems and with human liver microsomes.

1.4 DISSERTATION OVERVIEW

This dissertation consists of eight chapters.

Chapter 1 describes the principle and molecular mechanism of antisense therapy, the development of structure modifications in antisense oligonucleotides, and PK/PD properties of phosphorothioate oligonucleotides. Current knowledge on ribonucleotide reductase and its role in cancer therapy was reviewed. The antisense activity of GTI-2040 and its preliminary PK and PD studies were discussed. The objectives of this study were outlined.

Chapters 2-7 present the main body of this dissertation. 28 In Chapter 2, the development and validation of a sensitive and specific

hybridization ELISA assay for GTI-2040 in various biological matrices was presented.

This assay provided a powerful tool in the PK/PD study of GTI-2040.

In Chapter 3, cellular uptake of GTI-2040 in cell lines was described. Antisense

effects of GTI-2040 on leukemia cells was evaluated.

In Chapter 4, clinical pharmacokinetics of GTI-2040 was presented. Bone marrow

cellular uptake and distribution of GTI-2040 was described. Clinical activity, disease

response and their association with PK were evaluated.

In Chapter 5, population PK study was performed to identify the covariates that

could explain the variability in PK parameters of GTI-2040. An indirect response PK/PD

model was proposed to describe the dynamic perturbation of R2 mRNA in patients

treated with GTI-2040.

In Chapter 6, a novel ion-pair reverse phase ESI LC/MS/MS method was applied for the identification of GTI-2040 metabolites from in vivo and in vitro biological samples.

In Chapter 7, enzyme kinetics of GTI-2040 was characterized in solutions containing 3’ exonuclease and in human liver microsomes. Complications from the protein binding and sequential metabolism were discussed.

In Chapter 8, the results from these studies were summarized, and implications for future clinical studies were proposed.

Examples of antisense oligonucleotides that were approved or under clinical trials.

Table 1.1 Adapted from J.Kurreck Eur.J.Biochem.270, 1628-1644, 2004.

RNase H

Modified from The Antisense Research Group at the University of Liverpool http://www.liv.ac.uk/~giles/theory/theory.htm

Figure 1.1 Mechanistic pathways of antisense oligonucleotides. A. Forming the triplex with double strand DNA and preventing the transcription of mRNA. B. Forming the duplex with mRNA and activating RNase H to cleave the mRNA strand. C. Modulation of mRNA splicing. D, E. Blocking the assembly of ribosome or the progress of translation and inducing the translational arrest.

Adapted from Arenz C. et.al. Naturwissenschaften 90:345-359, 2003.

Figure 1.2 Mechanisms of RNase H mediated antisense effect. Binding of PS-ODNs with target mRNA recruits RNase H which degrades the mRNA strand in the form of DNA-RNA duplex. PS-ODNs and RNase H could be recycled to a new round of antisense action. (blue butterfly: RNase H)

Adapted from Aboul-Fadl, Current Med. Chem. 12:2193-2214, 2005.

Figure 1.3 Sites for chemical modifications of antisense oligonucleotides.

Adapted from J.Kurreck Eur.J.Biochem.270, 1628-1644, 2004

Figure 1.4 Chemical structures of the first, second and the third generations of antisense oligonucleotides. B denotes one of the bases, adenine, guanine, cytosine or thymine.

Adapted from The Antisense Research Group at the University of Liverpool http://www.liv.ac.uk/~giles/theory/theory.htm

5’ 3’ a. A b. c. 5’ endonuclease exonuclease 3’ exonuclease

3 ’

5’ G

HO O H H

a H H O H A

O P O O S- H H N= 18 H H A O H c O P O O - H H S B H H O H b G

O P O O - H H S H H 3’ OH H

Figure 1.6 Possible degradation pathways (A) and cleavage positions (B) of PS-ODNs by the action of nuclease. a: Hydrolysis by 3’ exonuclease. b: Hydrolysis by 5’ exonuclease. c:. Random cleavage by endonuclease.

O Base

O S - Sp P Base O O O

O O Base P O Rp S - O

Figure 1.7: Sp and Rp diastereoisomers generated from the phosphorothioate chiral center in the backbone of PS-ODNs.

ADAPTED FROM FONTECAVE M, CELL. MOL; LIFE SCI. 54:684- 695, 1998

Figure 1.8 The function of ribonucleotide reductase (RNR). RNR reduces the ribonucleoside diphosphate to deoxyribonuceoside diphosphate.

A Human (R2) mRNA

194CODING 1364 2475

5’UTR 3’UTR Poly A

626 645 3’ GAACCACCTCGCTAAATCGG 5’ GTI-2040

B HO O H H

H O H A O P O O S- H H H H A O H

O P O O n=18 n S- H H H H O H G

O P O O S- H H H H OH H

Figure 1.9 Sequence and structure of GTI-2040. A. Sequence and coding region of GTI-2040; B. Chemical structure of GTI-2040 (only shows part of sequences).

NH2

N O

O H OH

H H OH

Figure 1.10 Chemical structure of cytarabine (1-β-D-arabionfuranosylcytosine, ara-C).

Inactive metabolites

ara-U ara-UMP

Cytidine Deoxycytidine deaminase deaminase Active metabolites

Cytarabine ara-CMP ara-CDP ara-CTP Deoxycitidine Nucleoside kinase diphosphate kinase Deoxycytidine kinase

Cell death

Figure 1.11 Metabolic pathway of cytarabine (ara-C). Ara-C is activated to ara-C triphosphate (ara-CTP) by 3 sequential enzymes, deoxycitidine kinase, deoxycitidine monophosphate kinase, and nucleotide-disphosphate kinase. Competing with the active phosphorylation, ara-C and ara-CMP were inactivated by two other enzymes, cytidine deaminase and dCMP deaminase, respectively, to the corresponding uridine metabolites. Ara-CMP: cytarabine 5’-monophosphate; Ara-CDP: cytarabine 5’-diphosphate; Ara-CTP: cytarabine 5’-triphosphate; Ara-U: uracil arabinoside; Ara-UMP: uracil arabinoside 5-monophosphate.

41 CHAPTER 2

A SPECIFIC PICOMOLAR HYBRIDIZATION-BASED ELISA ASSAY FOR THE DETERMINATION OF PHOSPHOROTHIOATE OLIGONUCLEOTIDES IN PLASMA AND CELLULAR MATRICES

2.1 INTRODUCTION

Nucleic acid hybridization has long been known to play a vital role in biological processes and its precise complementary association with mRNA has recently been recognized as a potential therapeutic strategy. Specific segments of DNA-like or modified DNA known as antisense oligonucleotides (AS ODNs) capable of hybridizing with specific mRNA or regions of mRNA have now been developed to aim at disruption of the expression of genes that associate with malignancy transformation. The first generation of antisense compounds, such as phosphorothioate oligonucleotides (PS

ODNs), offer significantly improved stability and biological activities and several are now being evaluated clinically in various molecular targets including cancer, viral diseases and inflammatory disorders (1,3,6,14,18,115). GTI-2040, a 20-mer PS ODN complementary to the mRNA of the R2 subunit of ribonucleotide reductase (RNR) has demonstrated down-regulation effects on mRNA and protein levels of R2 in several

42 human tumor cell lines in a sequence- and target-specific manner (95). In nude and SCID

mice, GTI 2040 significantly inhibits growth of a wide range of xenografted human

tumors (95). Results from the phase I clinical studies of GTI-2040 in advanced solid tumors showed manageable toxicity profile and it is well tolerated as a single agent (10).

This agent is currently undergoing clinical evaluation in patients with acute myeloid leukemia (AML) in combination with high dose cytarabine at the James Comprehensive

Cancer Center and Solove Research Institute at The Ohio State University.

To support the clinical evaluation of pharmacokinetic behaviors of antisense drugs and its relationship with efficacy, toxicity and disease response, it is essential to

develop a specific, sensitive and accurate quantification method of these agents in

biological matrices. Conventional approaches including anion-exchange HPLC, LC/MS, capillary gel electrophoresis (CGE) and chromatography of radiolabeled oligonucleotides have been used in biological sample analysis (49,55,70-72,79,116,117). However, insufficient sensitivity (lower limit of quantification or LLOQ >10 nM) and/or lack of selectivity toward metabolites has compromised their ability to either fully characterize the plasma pharmacokinetics or to measure intracellular drug concentrations, which may reveal its relationship with the target intervention (PK/PD), after drug administration

(74,118). Moreover, these methods demand extensive sample preparation prior to analysis, and may be subjected to matrix effects. More recently, hybridization between two single strand oligonucleotides has attracted considerable attention in their isolation and analysis based on the specific high binding affinity of nucleotide base-pairing.

Coupled with enzyme-linked immuno-reaction, pM levels assay sensitivity could be

43 achieved. Thus, hybridization based assays were proposed as quantification methods for antisense ODNs and a number of template designs have been reported for polynucleotide drugs including PS-ODNs and ribozymes (23,73,74,118-121). Among them, a one step homogeneous competitive hybridization method showed good sensitivity, but its inability to differentiate from metabolites (48% cross-reactivity with the 3’N-2) limits its extensive use (120). In a more recent hybridization assay, a capture probe design using a locked nucleic acid without 5’ overhang gave a super assay sensitivity in the femtomole range, but the assay was still unable to distinguish parent drug from chain-shortened oligomers (121). We and the others have shown that the hybridization-ligation based

ELISA for the analysis of PS ODNs could be rendered highly specific by using a capture probe with a 5’ overhang for isolation and a detection probe for signaling (11,23,74). In addition, this method provided good accuracy, a wide range of linearity and high throughput with minimal sample clean-up and was used in pharmacokinetic studies of two antisense ODNs (11,74,118). However, few of these reports investigated factors that influence the assay or only with limited information.

In this chapter, we further refined this hybridization procedure by examining the major factors that can be optimized to enhance the assay selectivity and accuracy on quantificaiton of GTI-2040 in biological samples.

2.2 MATERIALS AND METHODS

2.2.1 Antisense and Reagents

The 20mer phosphorothioate oligonucleotide, GTI-2040, with sequence of 5'-

GGC TAA ATC GCT CCA CCA AG-3' was provided by the National Cancer Institute 44 (Bethesda, MD) and used without further purification. Both the 3’ end (3' N-1, 3' N-2, 3'

N-3) and the 5’ end (5' N-1, 5' N-2, 5' N-3) putative metabolites of GTI-2040 were

purchased from Integrated DNA Technologies (Coralville, Iowa). The scrambled control

of GTI-2040 (5'-ACG CAC TCA GCT AGT GAC CA-3') and mismatched GTI-2040 (5'-

ACG CAC TCA GCT AGT GAC CA-3') were all obtained from Integrated DNA

Technologies (Coralville, Iowa). The purity and identity of each oligomer were verified

by HPLC/UV/Mass spectrometry (Ion trap mass spectrometer Model: LCQ, Finnigan,

San Jose, CA). The capture probe for one-step hybridization method of GTI-2040 was

synthesized as a 3’ end biotin attached and 5’ end digoxigenin attached 20 mer DNA

with a sequence complementary to GTI-2040. The capture probe for GTI-2040 used in

two-step hybridization ELISA was designed as a 29mer DNA oligonucleotide with the

first 20mer sequence from the 3’-end complementary to GTI-2040. The 3’-end was attached to a NeutrAvidin-coated 96-well plate via biotin and the 9mer overhang (5’-

TAA CTA GTG-3’) served as a template for the detection probe. A 9-mer DNA with phosphate at the 5’- end and digoxigenin at the 3’- end with sequence complement to the

5’-end 9mer overhang of the capture probes of GTI-2040 was used as an appropriate detection probe for the hybridization-ligation ELISA assay. All of these probes were custom synthesized by Integrated DNA Technologies (Coralville, Iowa).

Reacti-Bind NeutrAvidin-coated polystyrene plates were purchased from Pierce

(Rockford, IL). GTI-2040 standards were diluted in TE (Tris-HCl and EDTA) buffer containing 10 mM Tris-HCl and 1 mM EDTA (pH=8.0). The hybridization buffer used in preparation of capture probe solution contained 60 mM sodium phosphate, pH 7.4, 1.0 M

NaCl, 5 mM EDTA and 0.2% Tween 20. The ligation buffer was prepared as a mixture 45 of 66 mM Tris-HCl, pH 7.6, 10 mM MgCl2, 10 mM DTT, 1 mM ATP, 5U/ml T4 DNA ligase and 100 nM detection probe oligonucleotide. T4 DNA ligase and ATP were purchased from Amersham Biosciences (Piscataway, New Jersey). The washing buffer used throughout the assay contained 25 mM Tris-HCl, pH=7.2, 0.15 M NaCl, and 0.2%

Tween 20. The anti-digoxigenin-AP was obtained from Roche (Indianapolis, IN).

Attophos and its reconstitution solution were purchased from Promega (Madison, WI).

Blank human plasma was obtained from Red Cross (Columbus, OH). Detection was accomplished using a Gemini XS plate reader (Molecular Devices, CA).

2.2.2 Hybridization ELISA Assay Procedures

2.2.2.1 A One-step Hybridization ELISA

The principle of a one-step hybridization ELISA is shown in Scheme 1A.

Basically, the capture and detection of the analyte was accomplished on a one-strand capture probe, which is complementary to the analyte without an overhang and contains biotin at the 3’ end and digoxigenin at the 5’ end. After analyte-capture probe hybridization and the attachment of the duplex to the 96-well plate, S1 nuclease was added followed by 2 h incubation at 37°C. Then anti-Digoxigenin-Alkaline phosphatase

(AP) with super BSA block buffer was added into the wells, followed by incubation for

0.5 h at room temperature. After washing, an appropriate volume of Attophos® substrate solution in diethanolamine buffer was added into each well and the plate was incubated at

37°C for 0.5 h. The generated fluorescence was measured at Ex 430/Em 560 (filter=550 nm) using a Gemini XS fluorescence microtiter plate reader (Molecular Devices, CA).

46 2.2.2.2 A Two-step Hybridization-ligation ELISA Assay

The hybridization-ligation ELISA principle is shown in Scheme 1B. This method

is based on a two-step hybridization, first by base pairing of analytes to the capture probe

with an overhang, followed by hybridization with a detection probe, which is ligated to

the analyte. The detail of the procedure is described in Scheme 2. Basically, 200 nM

capture probe in hybridization buffer solution was first heated at 95°C for 5 min in a

heating block (Van Water and Rogers, PA) to disrupt the possible secondary structure.

Then 100 μL of the capture probe solution was added to 100 μL plasma or other

biological matrices containing GTI-2040 and the solution was mixed in a polypropylene

96-well plate. Five μL of 10% Triton X-100 was added to the hybrid complex to a final

concentration of 0.25% of Triton X-100. The mixture was incubated at 42°C for 2.5 h for hybridization. Then 150 μL of the solution was transferred to a NeutrAvidin-coated 96- well plate, which was incubated at 37°C for 30 min to allow the attachment of biotin labeled capture probe to NeutrAvidin-coated wells. The plate was washed six times with washing buffer, and 150 μL ligation solution containing 5 U/ml T4 DNA ligase and 100 nM detection probe was added to each well. The plate was incubated overnight at 18°C.

The plate was washed 3x with washing buffer and 3x with HPLC water to remove the

unligated detection probe. Following addition of 60 U/well S1 nuclease solution in 100 mM NaCl, the plate was incubated at 37°C for 2 h to cleave the truncated duplex. After 6

times washing, the plate was blocked with 1:1 Superblock buffer in TBS (Tris Buffered

Saline) (Pierce, IL). Then 150 μL anti-Digoxigenin-Alkaline phosphatase (AP) diluted with 1:2500 super BSA block buffer in TBS (Roche, Germany) was added into each well,

47 followed by incubation for 0.5 h at room temperature with gentle shaking. After 6x washing, 150 μL substrate solution (36 mg Attophos in 60 ml diethanolamine buffer) was added into each well and the plate was incubated at 37°C for 30 min. Finally, the fluorescence intensity was measured at Ex 430/Em 560 (filter=550nm) using a Gemini

XS fluorescence microtiter plate reader (Molecular Devices, CA).

2.2.3 Evaluation of Cutting Efficiency of S1 Nuclease

The one-step hybridization ELISA was used to evaluate the cutting efficiency of

S1 nuclease and GTI-2040 was used for evaluation. A control standard curve of fluorescence response versus concentration of intact one step capture ODN in human plasma was first established as follows. To each of 100 μL standard solutions containing the one step capture ODN of GTI-2040 at 0.025, 0.050, 0.1, 0.2, 0.5, 1, 2, 5, 10, 20 and

50 nM, 100 μL of hybridization buffer was added and the solution was incubated for 2.5 h at 42°C. Then, 150 μL of mixture was transferred into a Neutr-Avidin coated 96-well plate, which was incubated at 37°C for 0.5 h, facilitating biotin-avidin binding. After incubation with 100 mM NaCl solution for 2 h at 37°C, the plate was treated with anti-

Dig-AP antibody followed by the addition of Attophos substrate with a washing step between each steps, as described in the assay procedure of a one-step hybridization

ELISA. Fluorescent response from the one-step template ODN of GTI-2040 was then measured. The standard curve of fluorescence responses versus concentrations of digoxigenin-labeled one step capture ODN was established and the linear regression equation was obtained.

48 Cutting efficiencies at two levels of S1 nuclease, namely 60U/well and 100U/well were evaluated on each of 4 concentrations of one step ODN, i.e., 50 nM, 100 nM, 200 nM and 250 nM. The procedure was the same as above except that 100 mM NaCl solution was replaced by 60U/well or 100U/well of S1 nuclease in 100 mM NaCl solution. The remaining concentrations of one step capture ODN after the treatment with

S1 nuclease were measured and calculated from the regression equation derived from the standard curve. The level of one step capture ODN cleaved by S1 nuclease was thus obtained by subtracting the residual concentration of the capture ODN from the initial values. The cutting efficiency of S1 nuclease on GTI-2040 one step capture ODN was calculated as:

Cutting efficiency(%) = Added concentration of caputureODN − Measured concentration of captureODN ×100% Added concentration of captureODN

(Equation 1)

2.2.4 Specificity and Selectivity of Hybridization-Ligation ELISA

2.2.4.1 Effect of S1 Nuclease on the Assay Selectivity

Effect of S1 nuclease on cross-reactivity of the 3’N-1 metabolite toward GTI-

2040 in hybridization-ligation ELISA was evaluated at the concentration range of GTI-

2040 from 0.1 nM to 500 nM in human plasma. Four levels of S1 nuclease in addition to the blank, i.e. 0, 15, 30, 60, and 100 U/well, were used for evaluation of its optimal level.

Procedure for hybridization-ligation ELISA was described previously.

49 2.2.4.2 Specificity and Selectivity of Hybridization-Ligation ELISA

Using 60 U/well S1 nuclease, cross-reactivity of 3’ end metabolites, 5’N-1 and

control oligomers (scrambled or mismatched oligonucleotides) on GTI-2040 at the

concentration range between 50 pM to 1 μM in 10% human plasma in TE buffer were

evaluated. The observed cross-reactivity profiles of these oligonucleotides were

compared with those of the parent drugs. The cross-reactivity of each metabolite toward

GTI-2040 was determined as the percentage of their Emax values to that of GTI-2040. The

Emax values were calculated by nonlinear regression model in SigmaPlot (SPSS, Chicago,

IL).

2.2.5 Method Validation

The hybridization-ligation ELISA method for GTI-2040 in human plasma, RBC,

urine and K562 cell lysate was validated. Linearity at the concentration range between

50 and 10000 pM was evaluated for both analytes in 10% human and 10% mouse plasma

in TE buffer 100% human plasma and K562 cell lysates. Within-day and between-day

accuracy and precision were determined at 100 pM (low quality control, QC), 500 pM

(medium low quality control, QC), 1 nM (median QC) and 5 nM (high QC) for both

compounds with 6 replicates in each matrix. Since the drug concentration ranges in

animal and clinical samples are likely to be higher than the upper limit of the calibration

curve, extension of the dynamic range was evaluated by dilution. Human plasma spiked

with 10, 50 and 200 nM GTI-2040 was diluted with 10% human plasma at 10, 50 and

200-folds, respectively (n=4). These standards were assayed and dilution recovery was

calculated by the dilution factors. 50

2.3 RESULTS

2.3.1 Cutting Efficiency of S1 Nuclease as Evaluated by the One-step Hybridization

ELISA

Fluorescence response of the one step capture ODN for GTI-2040 using a one- step hybridization ELISA was found to be linear from 0.025 nM to 50 nM. The cutting efficiency as estimated by Equation 1 is shown in Table 2.1. As shown, <1% of ODN remained intact after treatment of S1 nuclease at two levels, 60U/well and 100U/well.

Thus, based on this result, the cutting efficiency of S1 nuclease toward template ODN of

GTI 2040 was nearly quantitative.

2.3.2 Effect of S1 Nuclease on Selectivity of the Hybridization-Ligation ELISA

As shown in Figure 2.1, the use of S1 nuclease in hybridization-ligation ELISA dramatically decreased the interference from the 3’N-1 putative metabolite without significant suppression of fluorescence signal on GTI-2040. Cross-reactivities of 3’N-1 metabolites on GTI-2040 were determined to be 89%, 15%, 11%, 7.8%, and 8.3% at levels of S1 nuclease 0, 15, 30, 60, and 100 U/well, respectively. At the same time, the maximal response of GTI-2040 decreased 2.2%, 5%, 7%, and 11% with treatment of 15,

30, 60, 100 U/well of S1 nuclease, respectively. The slight decrease in sensitivity for

GTI 2040 was offset by a large gain in selectivity toward 3’N-1 metabolites. Since there was no further improvement in selectivity when 100 U/well of S1 nuclease was used,

60U/well was chosen as the final concentration of S1 nuclease in the assay procedure.

Additionally, a further increase in S1 nuclease may destabilize the double-stranded DNA 51 (Manual of S1 nuclease, Invitrogen). NaCl (100 mM) was found to stabilize both ends of

double-stranded DNA toward attack by S1 nuclease (Manual of S1 nuclease, Invitrogen)

and was used as reaction buffer of S1 nuclease.

The use of S1 nuclease not only improved the assay selectivity, but also improved

the linearity of the calibration curve of GTI-2040 below 10 nM. Figure 2.2 shows the

fluorescence-response of GTI-2040 at concentrations ranging from 0.1 to 10 nM at

different levels of S1 nuclease. Good linearity was observed with S1 nuclease treatment.

Without S1 nuclease treatment, GTI-2040 calibration curve was not linear even after addition of detergent. In such a case, a log-log transformation was necessary to generate linearity (74,120). Another advantage of using S1 nuclease was to decrease the background fluorescence. It was found that 60 U/well of S1 nuclease reduced background signal intensity to the noise levels generated from the microtiter wall.

2.3.3 Method Validation

Specificity. Potential interferences from endogenous matrix and from scrambled or mismatched oligonucleotides were first examined. The fluorescence responses from blank matrices evaluated were negligible and showed no difference from the PBS control.

Scrambled GTI-2040, if present in the biological sample, did not seem to compete with the analytes in plasma as shown as their negligible fluorescence signal (Figure 2.3).

Mismatched ODN that contain four mismatch nucleotides for GTI-2040 exhibited a cross-reactivity of only 5.77% (Table 2.2). The slopes and intercepts of the calibration curves in 10% and 100% human plasma, 10% mouse plasma, 10% rat plasma and cell

52 lysates as shown in Figure 2.4 were very close, indicating that the matrix effects were negligible for the hybridization ELISA assay.

Cross-Reactivity with Chain-Shortened Metabolites. Hybridization-ligation

ELISA method demonstrated its selectivity toward the parent analyte from their putative

3’ N-1 to N-3 metabolites (Figure 2.3). Compared to the concentration-response curve of parent ODNs, 3’N-1 to 3’N-3 metabolites of GTI-2040 all gave significantly lower fluorescence intensity. The Cross-reactivity values for 3’N-1 toward GTI 2040 was found to be 5.9% (Table 2.2). For both GTI-2040, a general trend shows that their cross- reactivities with 3’N-2 and 3’N-3 metabolites were lower than with the 3’N-1 metabolites and were considered negligible (Table 2.2).

Linearity. The hybridization-ligation ELISA was validated in a variety of biological matrices. The regression coefficients were all >0.990. A lower limit of detection (LLOD, defined as 5 times signal to background noise) was found to be 25 pM and a lower limit of quantification (LLOQ) to be 50 pM for GTI-2040. As shown in typical calibration curves (Figure 2.4), the assay was found to be linear from 50 to 10000 pM for GTI-2040. Beyond this range, fluorescence responses were found to reach a plateau in plasma or cell lysates. Therefore, 10 nM was set as the upper limit of the calibration curve for GTI 2040. However, at concentrations ≥5 nM, serial dilution with

10% plasma in TE buffer could extend the dynamic range to 200 nM, and the effect of dilution on the accuracy and precision of the method was evaluated (Table 2.3). The nominal accuracy values were 105.4%, 110.4% and 113.3% at 10, 50, 200- fold dilutions, respectively. Coefficients of variation (CVs) for each corresponding dilution factors were estimated as 6.6%, 6.5% and 7.1%, respectively (Table 2.3). These results indicated that 53 the linear concentration range of the assay could be extended from 50 pM to 200 nM with acceptable precision and accuracy.

Accuracy and precision. The within-day and between-day accuracy and precision of GTI 2040 in human 10% plasma, 100% human plasma, human red blood cell lysate (RBC) and K562 cell lysate were summarized in Table 2.4. The within-day precision (%CV) in human plasma at concentration levels of 100, 500, 1000 and 5000 pM were found to be 11.1, 7.4, 4.9 and 4.1 % (all n=6), respectively. The corresponding accuracy values were 103.1, 95.0, 95.7, and 101.5% (all n=6) based on the nominal concentrations. The inter-day CVs were found to be 19.9, 12.9, 10.9, and 10.5%, respectively, with the corresponding accuracy values of 114, 93, 102.7, and 94.2% (all n=6). The within-day CVs of the assay in K562 cell lysates were found to be 10.8%,

2.3%, 2.9 and 3.2% for the 100, 500, 1000 and 5000 pM samples, respectively, with the corresponding accuracy values of 103.3, 103.5, 100.1 and 97.3 % (Table 2.4).

2.4 DISCUSSION

Recently, several hybridization-based ELISA methods have been successfully developed for the determination of polynucleotides in biological matrices

(11,23,73,74,118-121). Because of simplicity, we initially developed a one-step hybridization ELISA method on quantification of GTI 2040 in biological matrix.

Although good linearity was observed, the high cross-reactivity with the 3’ end metabolites precludes their practical uses. Herein, an ultra-sensitive and selective

54 hybridization-ligation ELISA was developed and validated for determination of GTI

2040 in biological matrices including human plasma, PBMC, RBC and cell lysates.

The hybridization ELISA is based on the Watson-Crick base-pairing and anti-

Digoxingenin (Dig) detection system, which provides high selectivity and sensitivity to

the assay. To enhance the assay selectivity, especially the ability of discriminating from

the 3’ end chain-shorten metabolites, a customed designed 9 mer detection probe with its sequence complementary to the 5’ overhang of capture probe and 5’ end Dig-label was

employed. In the presence of T4 ligase and ATP, ligation only occurs between the

analyte and detection probe, forming well-bound duplexes (29 mer duplex for GTI 2040)

which are detected by the Digoxingenin detection system. The 3’end chain-shortened

metabolites, when present, will bind to the capture probe but fails to complete ligation

with the detection probe due to their chain-shortened gap. The excess amount of capture

probe would still bind to detection ODN, generating a short duplex, which will give

signal and interfere with the assay. It was reported that ddH2O washing could remove

unligated detection ODN (74). This is probably due to the higher binding affinity of the

ligated analyte-detection probe complex with capture ODN (e.g. 29-mer), which has a

50°C difference in Tm values with the short duplex (9-mer). However, we have found

that the unligated detection ODN could not be completely eliminated by washing only

which was evidenced by the poor linearity of GTI-2040 in plasma (Figure 2.2) and non-

selectivity toward its 3’N-1 metabolite (89% cross-reactivity) (Figure 2.1). Subsequently,

we have found that after hybridization and ligation, incubation of the solution with S1

nuclease for 2 h at 37°C, followed by the washing step dramatically, reduced the

background fluorescence and decreased the cross-reactivity with the 3’ end deletion 55 metabolites (Figure 2.1). S1 nuclease cuts the single strand region in DNA duplex with ≥

99% efficiency (Table 2.1) and the single strand region of an imperfectly formed DNA

duplex, generating fragments that can be removed by washing. Therefore, the cross- reactivity of GTI 2040 with its 3’N-1 metabolite was considered to be minimal. We have also found that accessibility of S1 nuclease may be influenced by steric factors. In the one-step hybridization ELISA with GTI-2040, the cross-reactivity with 3’N-1 was 26%.

This cross reactivity value did not result from the background of the excess single- stranded capture probe, since fluorescence from the control capture probe (without analyte) was minimal (<10%). We speculate that in the one-step hybridization ELISA, the bulky structure of digoxigenin next to the nick may pose steric hindrance for the S1 nuclease. In addition, in the one step hybridization ELISA, cross reactivity with 3’N-2 and 3’N-3 metabolites were 18% and 10%, respectively, consistent with the lesser degree of steric hindrance with more nucleotides deletion from the 3’end. On the other hand, in the hybridization-ligation ELISA, the single strand regions in duplex were more open, facilitating the attack by the S1 nuclease. We observed that the cross-reactivity of 3’N-1 dropped from 89% to 5.5% after using S1 nuclease in the two-step hybridization-ligation

ELISA.

This possibility may also explain the high cross-reactivity with the 5’N-1 putative metabolite. The 5’ end deletion metabolites could still completely ligate with the detection probe with their intact 3’ end. On the other hand, the deleted region in the

5’end in the duplex, although could be cut by the S1 nuclease, may be difficult to be removed due to the steric hindrance, since the single strand gap is located close to the plate wall. Therefore, the observed cross-reactivity with 5’N-1 was 78%. Fortunately, 56 this cross reactivity may not pose a significant problem in practice, since in biological

samples, there was little 5’-end metabolites detected for these two antisense compounds

(78).

Our results also showed great improvement of linearity of GTI-2040 in plasma

with addition of S1 nuclease as illustrated in Figure 2.2. Although the exact mechanism is

not known, it may attribute to the reduction of nonspecific background fluorescence

signal. The overall evaluation led us to select the 60U/well of S1 nuclease as the working

level in our study of these two oligonucleotides.

During the development of the one-step hybridization ELSIA assay, it was found

that the calibration curves in human and mouse plasma were not linear from 25 to 2500 pM, with lower than expected fluorescence signals at high concentrations. A log/log

transformation improved the linearity, and is commonly used for ELISA based assays

(73,74). We hypothesized that some plasma or cellular protein (possibly albumin) may

interfere with the hybridization of antisense ODNs with capture ODNs at the first

hybridization step, therefore causing non-linearity. It was subsequently found that an

addition of 0.25 % Triton X-100 (w/v) improved the assay linearity. Detergent is known

to disrupt the nonspecific interaction between analyte and plasma protein, which may

contribute to the nonlinearity. Additionally, disruption of the interaction between analyte

and matrix not only extended the linear range to 5000 pM in plasma and cell lysate for

both antisense compounds, but also improved the analyte recovery, thus the assay precision.

The assay format and procedure of hybridization-ligation ELISA described in this chapter could possibly apply to general oligonucleotide antisense compounds in 57 biological samples. For different oligonucleotides, it is necessary to design specific capture probe based on the sequence of the test antisense and a 9 mer overhang at 5' terminal complementary to the sequence of the detection probe. Sequence design of the detection probe is based on two principles: (1) no similar match with the sequence of

5’overhang of the capture probe present for endogenous substances. The possible endogenous matches could be examined from BLAST database

(http://www.ncbi.nlm.nih.gov/BLAST). (2) as little base pairing occurs as possible between detection ODN and capture ODN in case the detection probe shifts to the gap created by the chain-shortened metabolites. Poly A or other poly nucleotides are the extreme examples resistant to the loss of base pairing resulting from frame-shift and should be avoided using. The sequence of detection probe ODNs does not have to be the same for different analytes, since two different sequences for G3139 (11) and GTI-2040 in terms of assay specificity or sensitivity worked equally well. However, poly A or other poly nucleotides as detection probe ODNs is not recommended, since we have found that these probes caused the assay to lose a single nucleotide resolution toward the 3'-end deletion oligomers (data not shown). When differentiation of 5’-end metabolites becomes necessary, the signal interference from 5'-end metabolites could be markedly reduced by using reversed sequence of capture and probe ODNs presented in this paper.

In conclusion, we have developed and validated a non-radioactive hybridization - ligation ELISA assay for determination of an antisense drug, GTI-2040, in a variety of biological matrixes. The major advantages of this assay are: (a) ultra-sensitive with a

LOQ of 50 pM, (b) highly selective toward 3' end deletion metabolites, (c) negligible matrix effect, and (d), simple sample preparation and may be adapted to high throughput 58 application. Moreover, the ultra-sensitivity of the ELISA assays allow more accurate characterization of pharmacokinetics of these antisense compounds, especially the terminal half-lives and clearance values, when compared with the less sensitive HPLC method. Importantly, this assay provided feasible analytical method to measure intracellular concentrations of GTI-2040 in cells from peripheral blood and bone marrow obtained from AML patients, which are valuable in the evaluation of drug effect and disease response.

Nominal template conc. Cutting efficiency of S1 nuclease (n=5) (nM) 60 U/well 100 U/well 50 99.3 ± 0.10% 99.8 ± 0.03% 100 99.6 ± 0.02% 99.9 ± 0.01% 200 99.7 ± 0.03% 99.8 ± 0.01% 250 99.7 ± 0.02% 99.8 ± 0.01%

Table 2.1 Cutting efficiency of S1 nuclease toward one-step GTI-2040 capture probe with S1 at levels of 60 U/well and 100 U/well.

Name Sequence Cross-reactivity (From 5’ to 3’) (%) GTI 2040 GGCTAAATCGCTCCACCAAG 100 3’-N-1 GGCTAAATCGCTCCACCAA 5.9 3’-N-2 GGCTAAATCGCTCCACCA 0.15 3’-N-3 GGCTAAATCGCTCCACC 0.12 5’-N-1 GCTAAATCGCTCCACCAAG 78.4 Mismatched GTI 2040 GGCTAAACTCGTCCACCAAG 5.77 Scrambled GTI 2040 ACGCACTCAGCTAGTGACAC <0.02

Table 2.2 Cross-reactivity of the 3’ and 5’ end putative metabolites, mismatched and scrambled ODNs toward GTI 2040.

Nominal concentration Mean Measured Accuracy (%) CV (%) (nM) (dilution factor) concentration (nM)

10 ( ×10) 10.54 105.4 6.6 50 (× 50) 55.18 110.4 6.5 200 (× 200) 226.3 113.3 7.1

Table 2.3 Assay accuracy and reproducibility of GTI-2040 plasma samples following dilution.

GTI conc. Accuracy (%) Reproducibility (CV%) 100 500 100 500 1 matrix 1 nM 5 nM 5 nM pM pM pM pM nM Plasma 103.1 95 95.7 101.5 11.1 7.4 4.9 4.1 (within day) Plasma 114 93 102.7 94.2 19.9 12.9 10.9 10.5 (between day) K562 103.3 103.5 100.1 97.3 10.8 2.3 2.9 3.2 RBC NA 108.3 108.1 91.3 NA 5.7 6.9 4.9 Urine 98.4 92.3 100.9 96.4 18.7 5.1 11.5 7.4

Table 2.4 Within-day and between-day accuracy and reproducibility of GTI 2040 in plasma, and within-day accuracy and precision in K562 cell lysates, RBC and urine.

63 D Analyte (GTI) Template D D B B B hybridization

Ap D E-Ab D Ap B Single strand specific B Nuclease (S1) D D B SI B

Scheme 1A

5’ 3’ Analyte antisense 3’ GTG ATC AAT 5’ B 5’ 3’ Capture probe with 5’ overhang 3’ chain-shorten metabolites

1) hybridization 2) Apply to avidin-coated plate

3’ GTG ATC AAT 5’ D B P 5’CAC TAG TTA3’ 3’ GTG ATC AAT 5’ B 3)T4 ligase, ATP GTG ATC AAT 5’ B 3’ Ap D E-Ab 3’ Ap B P CAC TAG TTA D 4)S1 nuclease D GTG ATC AAT 5’ 5) wash B 3’ GTG ATC AAT 5’ P 5’CAC TAG TTA3’ B

Sheme 1B

Scheme 2.1: Illustrations of hybridization ELISA 1A. A one-step hybridization ELISA 1B. A two-step hybridization-ligation ELISA

Scheme 2.2: The working procedure of hybridization ELISA assay to measure antisense drugs in plasma and cell lysate.

70000

60000

50000 GTI2040 (S1 0U/well) 3'N-1 (S1 0U/well) 40000 GTI2040 (S1 15U/well) 3'N-1 (S1 15U/well) GTI2040 (S1 30U/well) 30000 3'N-1 (S1 30U/well) GTI2040 (S1 60U/well) 3'N-1 (S1 60U/well)

fluorescence 20000 GTI2040 (S1 100U/well) 3'N-1 (S1 100U/well) 10000

-10000 0 100 200 300 400 500 600 concentration (nM)

Figure 2.1 Effect of S1 nuclease on assay selectivity. Treatment of S1 nuclease decreased cross-reactivity of 3’N-1 dramatically from 89% without S1 nuclease to 15%, 11%,7.8% and 8.3% using 15, 30, 60 and 100 U/well S1 nuclease, respectively.

70000 S1 0U/well S1 15U/well 60000 S1 30U/well S1 60U/well S1 100U/well 50000

40000

30000

fluorescence 20000

10000

024681012

concentration (nM)

Figure 2.2 Effect of S1 nuclease on linearity of GTI 2040 in plasma. With the addition of S1 nuclease, linearity of GTI-2040 in the concentration range of 0.1 to 10 nM was improved.

40000

30000

16000 14000 12000 20000 10000 8000 6000

4000

fluorescnece signal 2000

0 10000 -2000 024681012

Fluorescence concentration (nM)

-10000 0 200 400 600 800 1000 1200

GTI 2040 Concentration (nM) 3' N-1 3' N-2 3' N-3 scrambled GTI mismatched GTI

Figure 2.3 Cross-reactivity of the putative 3’ end metabolites (3’N-1, 3’N-2 and 3’N-3), scrambled and mismatched GTI-2040. The small insert showed the cross-reactivity at low concentrations.

14000

12000 10% rat plasma 10% mouse plasma 10000 10% human plasma 100% human plasma K562 cell lysates 8000

6000

Fluorescence 4000

2000

0 0 1000 2000 3000 4000 5000 GTI-2040 Concentration (nM)

25000

20000

15000

10000

Fuorescence 10% plasma K562 cell lysates urine 10% RBC 5000

0 0 2000 4000 6000 8000 10000 GTI-2040 Concentration (pM) B

Figure 2.4 Representative calibration curves of GTI 2040 in (A) 10% plasma from the rat, mouse, human, and 100% human plasma and K562 cell lysates and (B) in human urine and 10% plasma, RBC, and K562 cell lysates.

69 CHAPTER 3

CELLULAR UPTAKE AND PHARMACOLOGY OF GTI-2040 IN LEUKEMIA

CELL LINES

3.1 INTRODUCTION

Over the last decade, antisense oligonucleotides (ODNs) have emerged as an

important tool to inhibit the expression of target mRNA and protein products in a

sequence-specific manner, and are used for functional genomics, target validation and

therapeutic purpose (2,3,8). Phosphorothioate oligonucleotides (PS-ODNs) are the most

widely used oligonucleotides in therapeutics in which one of the non-bridge oxygen atom

is replaced by sulfur. The sulfuration on backbones renders PS-ODNs increased nuclease

resistance, ability in RNase H recruitment and tolerable toxicity, which makes them more suitable for the in vivo administration (4,6,13). For successful antisense inhibition, specific Watson–Crick base pairing between an oligonucleotide and its target mRNA is essential. However, the selection of effective antisense agents is difficult because RNAs fold to form secondary and tertiary structures, resulting in the inaccessibility of the most

70 AS-ODNs to pair with the complementary nucleic acids (122,123). In an initial high

through-put screening, in vitro hybridization examinations in cell-free medium is used to

speed up the identification of the effective antisense compounds (123,124).

Oligonucleotide scanning arrays and oligomer library/ribonucleae H digestion-based

screen are two methods used in the empirical screening of ODNs in cell-free settings

(124-126). Usually from hundreds of synthesized anstisense candidates, only 2–5% of the

oligonucleotides are found to effectively hybridize with the target mRNA using oligonucleotide scanning arrays (126,127). For the ODNs to demonstrate favorable hybridization efficiency with the target mRNA in the screening assay, it is necessary to verify that a decrease in target mRNA or protein occurs in cell culture and in animal models in order to validate the proposed antisense mechanism. Mismatch and scrambled

ODNs are important controls used to identify the pharmacological or toxic effects not related to the antisense mechanisms (non-antisense effects). In addition, cellular factors, such as the membrane penetration of ODNs, cellular trafficking of the ODNs, interference from the RNA-binding proteins for target and ODNs hybridization, recruitment of RNase H to the target site (7,21), also determine the potency and the

specificity of the antisense drugs. Cell cultures not only serves as a powerful tool in

validation of the antisense activity, but also provides valuable information on cellular

uptake and distribution of AS-ODNs. Therefore, it is important to evaluate the

pharmacological effect and cellular uptake of ODNs in cell cultures before their in vivo

application.

71 GTI-2040, a PS-ODN, was selected from 102 ODNs complementary to R2

mRNA following their screening for the ability of down regulation of R2 mRNA levels

in vitro. In a number of human solid tumor cell lines such as bladder and lung carcinoma

cells and melanoma cells GTI-2040 demonstrated down-regulation of R2 mRNA and protein expression in a sequence- and target- specific manner (95). To explore its therapeutic potential in leukemia, in this study, we further evaluated the effects of GTI-

2040 on the expression of R2 mRNA and protein in cell lines derived from leukemia.

In order to exert their pharmacological and biological effects, antisense molecules must enter into cells and interact with mRNA in the nucleus and cytoplasm. However, antisense oligonucleotides are hydrophilic molecules with multiple anionic charges resulting in an inefficient cellular uptake (44,128). Hence, in cell cultures, cationic lipids or liposomes are used as delivery reagents to enhance the cellular uptake of ODNs as well as to demonstrate their pharmacological effect. Paradoxically, while the evaluation of the pharmacological effects requires the use of complexation agents, most of the therapeutically active ODNs including GTI-2040 have been administered clinically as free compounds. Despite the encouraging pharmacological effects of ODNs in vivo,

questions concerning their cellular trafficking and subcellular distribution remain to be answered.

In treatment of cancers, antisense oligonucleotides are usually combined with chemotherapeutic agents in order to enhance the cytotoxicity of the chemotherapeutic drugs in tumor cells. As a result of the inhibition of R2 expression, GTI-2040 may produce an imbalance of the dNTP pool, which could or may enhance the antitumor

72 effect of cytarabine (ara-C). The triphosphate form of the cytarabine, ara-CTP, competes

with the natural dCTP for DNA incorporation by DNA polymerases, resulting in the termination of DNA chain elongation (111-113). As a result of the inhibition of the expression of R2 mRNA and protein by GTI-2040, synthesis of the natural dCTP could be suppressed. With less competition from the endogenous deoxycytidine, DNA incorporation of ara-CTP may be enhanced, leading into an increase in irreversible cell damage (110-112).

In this chapter, we investigated the cellular uptake and distribution behaviors of

GTI-2040 in cultured leukemia cells with or without the use of cationic reagent. Effect of

GTI-2040 on the R2 mRNA and protein was investigated in leukemia derived cell lines.

In addition, a possible additive or synergistic effect in combination of GTI-2040 with cytarabine was evaluated in leukemia cell line.

3.2 MATERIALS AND METHODS

3.2.1 Drugs and reagents

GTI-2040, a 20-mer phosphorothioate oligonucleotide with the sequence of 5'-

GGC TAA ATC GCT CCA CCA AG-3', was provided by the National Cancer Institute

(Bethesda, MD) and used without further purification. The scrambled control of GTI-

2040 (5'-ACG CAC TCA GCT AGT GAC CA-3'), mismatched GTI-2040 (5'-ACG CAC

TCA GCT AGT GAC CA-3') were all obtained from Integrated DNA Technologies

(Coralville, Iowa). PS-dC 28, a 28 mer polycytidine phosphorothioate oligonucleotide was obtained from Integrated DNA technologies (Coralville, Iowa). Oligofectamine was

73 purchased from Invitrogen (Rockville, MD), neophectine was purchased from Neopharm

(Waukegan, IL). The sequence of R2 mRNA forward primer used is 5’ GCC TGG CCT

CAC ATT TTC TAA T 3’ and for the R2 mRNA reversed primer is 5’ GAA CAT CAG

GCA AGC AAA ATC A 3’. The sequence of R2 mRNA probe is 5' CAG AGA TGA

GGG TTT ACA C 3' with a reporter dye of FAM (FAM: 5-carboxyfluorescein) and a

quencher dye of the minor groove binder (MGB) (MGB: Minor Groove Binder, eg.

Dystamycin). Sequences of ABL mRNA forward primer and reversed primers are 5'

TGG AGA TTA ACA CTC TAA GCA TAA CTA AAG GT 3' and 5' GAT GTA GTT

GCT TGG GAC CCA 3', respectively. Sequence of ABL mRNA probe is 5’ CCA TTT

TTG GTT TGG GCT TCA CAC CAT T 3’ with reporter dye of FAM and quencher dye

of TAMRA. (TAMRA: 5-Carboxytetramethylrhodamine). The primers and probes are all purchased from Applied Biosystems (Foster City, CA). Goat anti-human R2 antibody (E-

16) and GAPDH antigoat second antibody were obtained from Santa Cruz Biotechnology

(Santa Cruz, CA.) RIPA lysis buffer was purchased from Cell Signal Technology Inc

(Danvers, MA) and PMSF (Phenylmethanesulfonyl fluoride) is purchased from Sigma-

Aldrich Co. (St. Louis, MO). Protease inhibitor cocktail is purchased from Calbiochem

(San Diego, CA). BD Vacutainer® cell preparation tubes were obtained from BD Science

(Woburn, MA). Nuclease preparation kit was obtained from the Molecular Motif

(Woburn, MA).

3.2.2 Cell cultures

The leukemia cell line K562 (ATCC, Manassa) was used in the study of cellular

uptake and antisense activity of GTI-2040. K562 cells were cultured in RPMI 1640 74 medium supplemented with antibiotics (Penicillin-Streptomycin) and 10% fetal bovine

serum (FBS) (Invitrogen, Rockville, MD). Cultures were maintained at 37 ºC in a

humidified environment with 5% CO2. Viability and cell counts were determined using

trypan blue dye exclusion and hemocytometer under the microscope.

3.2.3 Transfection cells with GTI-2040 delivered by cationic reagents

Cells were cultured until 70-80% confluence and then media were replenished

with the fresh complete medium and incubated at 37 ºC in a humidified environment with

5% CO2 overnight. The cells were then preceded to transfection experiments. Before the transfection, cells were washed with Opti-MEM (serum-free medium) (Invitrogen,

Rockville, MD) and seeded into a 6 well plate at a density of 1-2× 106 per well in Opti-

MEM. All transfections were performed in Opti-MEM medium, since low transfection

efficiency was observed in the presence of serum. Three transfection agents, i.e.

protamine sulfate, oligofectamine and neophectin depending on different purposes were

used. Oligofectamine and neophectin were used to examine the cellular uptake of GTI-

2040 in K562 cells. Neophectin was also used to transfect GTI-2040 into K562 cells in

order to evaluate the inhibition effect of GTI-2040 on the expression of R2 mRNA and

protein. When cells were transfected with oligofectamine, 6 μL/mL and 15 μL/mL were

used to complex with 0.2 µM and 0.500 µM of GTI-2040, respectively. To avoid

cytotoxicity due to oligofectamine, its final concentration was kept below 20 μL/mL.

Neophectin was used to transfect GTI-2040 using a 5:1 molar ratio to the drug. Protamine

sulfate is more clinically relevant and has being applied as a heparin antidote. In the

cytotoxicity study, 5: 1 (w/w) of protamine sulfate was used to deliver GTI-2040 into 75 K562 cells due to its low in vivo toxicity. Stock solutions of transfection agents and GTI-

2040 were prepared using Opti-MEM as the diluent. Drug solutions and transfection solutions were then mixed and incubated at room temperature for 30 min to allow lipid-

ODN complex formation. An appropriate volume of each of these complex solutions was overlaid on the cells to a final volume of 1 mL. Cells were incubated with lipid-ODN complex at 37 ºC for 6 hrs. Following the end of transfection, 1 mL of medium containing 20% FBS was added to each well and the content was gently mixed. The

mixture was incubated for certain duration for different purposes e.g. cytotoxicity (72

hrs), mRNA (24 hrs) or protein extraction (48 hrs) and GTI-2040 quantitation. Cells

without any treatment or cells treated with GTI-2040 alone, transfection reagents alone or

scrambled, mismatch GTI-2040 were used as controls in the indicated experiments.

The time course for cellular uptake of GTI-2040 (0.2 μM) in K562 cells

transfected by neophectin was examined at 0.5, 1, 2, 4, 6, 12, 24 hrs after transfection.

Examination of efflux of GTI-2040 from K562 cells was conducted after 6 hrs

transfection of 0.2 μM GTI-2040. Opti-MEM medium containing the drug was

replenished with the drug-free RPMI medium with 10% FBS. Aliquot of medium was

collected 0.167, 0.5, 1, 2, 4, 13 hrs post transfection for assaying GTI-2040

concentrations.

Effect of temperature on the cellular uptake of GTI-2040 in K562 cells was

evaluated using neophectin as the delivery agent. GTI-2040 at concentrations of 0.1, 0.2,

0.5 and 1.0 μM were transfected and incubated with 1 × 106 K562 cells for 24 hr at 4 ºC

and 37 ºC, respectively. Cells were then harvested and washed to remove the membrane

76 bound drugs and the intracellular concentration of GTI-2040 was assayed using the

previously validated ELISA method as described in 3.2.6.

Each of transfection studies was conducted three times and the results were

reported as the mean ± SD of triplicates.

3.2.4 Fractionation of nucleus and cytoplasm from cells

K562 cells (2×106) were treated with GTI-2040 in the presence or absence of

neophectin and incubated for 6 hrs at 37 ºC. To separate cytoplasm and nucleus, cell

pellets were suspended in 1 × hypotonic buffer and incubated on ice for 15 min. Ten μL of detergent per 1 million cells was added and the cell suspension was vortexed for 10 seconds. After centrifugation for 30 seconds at 14,000g at 4 ºC, the supernatant as the cytoplasm lysate was transferred to a new tube. The gel-like pellets, which contains the nucleus, were washed once with 1 × hypotonic buffer and to it was added 40 μL complete lysis buffer per 1 million cells. After mixing by vortex for 15 min at the highest speed setting, it was centrifuged at 14,000g for 10 min at 4 ºC. The supernatant fraction which contained the nucleus lysates was removed and stored in a new tube. In preparation for the whole cell lysates, the whole cell pellets were resuspended in 100 μL complete cell

lysis buffer and, after vortexing at the highest speed setting for 10 seconds, the mixture

was incubated on ice for 10 min. After centrifugation at 14,000g for 10 min at 4 ºC, the

supernatant was transferred to a new tube as the whole cell lysate, which was used for

determinations of GTI-2040 and protein. Upon damage of the cell membrane, lactate dehydrogenate (LDH), that is a stable cytoplasmic enzyme present in all cells, will be rapidly released from the cytosol. Colorimetric assay of LDH activity in the cell culture 77 supernatant is used for the quantitation of cell death and cell lysis [manual of

Cytotoxicity Detection Kit (LDH), Roche, Penzberg, Germany]. Herein, LDH

concentrations in the nucleus and cytoplasm fractions were measured to test the

contamination of cytoplasm in the nucleus fraction.

3.2.5 Uptake of GTI-2040 in mononuclear cells and red blood cells following blood

incubation

Five mL of fresh blood was drawn from two human donors and incubated with

500 nM of GTI-2040 at 37 ºC for 2 hrs. EDTA was added to a final concentration of 5 mM to retard drug degradation by nuclease in blood. Peripheral blood mononuclear cells

(PBMC) and red blood cells (RBC) were isolated from the whole blood using the BD

vacutainer CPT tube by a single centrifugation step. Briefly, the whole blood was

centrifuged in the tube with Ficoll gradient for 30 min at 1400 g at room temperature.

After centrifugation, plasma located in the upper layer was aspirated without disturbing

the PBMC and plateletes, which was located as a whitish layer underneath the plasma.

RBC is trapped in the tube bottom by gel barrier. Cell layer was then collected with

pipette and transferred to a 15 mL conical centrifuge tube with cap. Process of PBMC

cells then proceeded to the next step for “determination of intracellular GTI-2040 concentrations”. After collection of cell layer, RBCs below the gel barrier were also removed for GTI-2040 determination.

78 3.2.6 Determination of intracellular GTI-2040 concentrations

Following transfection and incubation, cells were centrifuged and the cell pellets

were washed with cold PBS. Cell pellet was then incubated with 200 μL, 0.1 μM

phosphorothioate 28 mer polycytidine (PS-dC28) for 2 min on ice and washed with PBS

to remove membrane-bound ODN (129,130). The number of viable cells was counted

using trypan blue exclusion. This assay is based on the loss of cell membrane integrity in

the dead cells, which allows the binding of acidic dye, trypan blue, to the intracellular

protein. Trypan blue is usually excluded by the intact membrane of the alive cells.

Following addition of 200 μL lysis buffer (10 mM Tris-HCl, pH=8.0, 0.5 mM EDTA, 1%

Triton X-100) and incubation on ice for 10 min, cells were lysed by vortexing and

sonication. The cell lysate was then centrifuged at 10,000 g, and the supernatant was

transferred to a new tube for the ELISA and protein assays (BioRad protein assay kit,

Hercules, CA). The intracellular levels of GTI-2040 were measured using the ELISA

assay as described in Chapter 2.

Concentration of GTI-2040 in cells or in mononuclear cells from human

peripheral blood was expressed as the intracellular concentrations after the cell volume

was determined as follows (131,132). The cell volume was determined using the Samba

Image Analyzer 4000 (Imaging Products International, Inc., Chantilly, VA). The Samba

4000 quantified the distance as the number of pixels, i.e., 164 pixels per 200×

microscopic field. The maximum (L) and minimum (W) diameters of a cell, which were

determined by counting the number of pixels and converting the value to micrometers,

were used to calculate the cell volume. Following equation describes the calculation of the volume (V) of ellipsoid cells. 79 π V = × L×W 2 6

Mononuclear cells in the peripheral blood were to found to have an average volume of 1.09 ± 0.431 µL/2 × 106 cells (mean ± S.D., n = 152), or normalized by the protein amount as 1 μL/0.035μg protein. K562 cells were found to have an average volume of 2.11 ± 0.563 µL/2 × 106 cells (mean ± S.D., n = 209).

GTI-2040 in PBMC was converted to an intracellular concentration (nM) by dividing intracellular GTI-2040 amount of nmole/mg protein to calculated cell volume

(0.035 μg protein equals to 1 μL cell volume or 2 × 106 cell number equals to 1 μL cell volume). Intracellular concentrations of GTI-2040 in K562 cells were also converted to nM by using the cellular volume of 2.1 µL/2 × 106 cells.

3.2.7 Flow cytometric analysis for studying of cellular uptake

K562 cells (0.5 × 106) were exposed to 0.5 µM FITC-GTI-2040 (IDT, Coralville,

Iowa) in the presence or absence of delivery vehicle (oligofectamine or neophectin) at 37

ºC for 4 hrs. Cells without treatment were used as controls. The cells were then harvested, washed with cold PBS, and analyzed by flow cytometry on a FACScalibur flow cytometer (Becton-Dickinson, San Diego, CA). Histograms were drawn to compare difference in mean fluorescence intensity (MFI) of FITC positive cell population between treatments.

80 3.2.8 Laser Scanning Microscopy analysis of cellular uptake of GTI-2040

Cells transfection with FITC-GTI-2040 was conducted using oligofectamine, neophectin and protamine sulfate as the delivery agent. Following 6 hrs transfection each in Opti-MEM, the cells were harvested. To the harvested cell pellet was added 200 μL,

0.1 μM PS-dC 28 and incubated on ice for 2 min. Following FBS washing, the harvested cells were fixed with 3.7% paraformaldehyde in PBS at room temperature for 10 min.

The cells were then transferred onto slides by centrifugation (200 g, 5 min) with a

Shandon Cytospin3 (Thermo Shandon, Pittsburgh, PA) and stained with 600 nM DAPI

(4’-6-diamidino-2-phenylindole) (Molecular Probes, Eugene, OR) for 5 min for nuclear counterstaining. The slides were mounted and examined with a Zeiss 510 META Laser

Scanning microscope (Carl Zeiss Inc., Germany). Images were captured with a CCD camera and processed in Zeiss LSM image Browser.

3.2.9 Quantification of R2 mRNA levels

Quantification of R2 mRNA was performed by Real Time RT-PCR.

2 × 106 K562 cells were treated with controls or neophectin complexed GTI-2040 in Opti-MEM for 6 hrs, followed by continuous incubation in RPMI medium with 10%

FBS for 24 hrs. Cells were then harvested. Total cellular RNA was extracted using

Absolutely RNA® RT-PCR Miniprep Kit (Stratagene, La Jolla, CA) according to the instruction from the manual. For complementary (cDNA) synthesis, total RNA (2 μg) from each sample was mixed with 1.5 μL of 20 μM random hexamer primer (Perkin

Elmer, Boston, MA) and heated to 70 ºC for 2 min followed by cooling on ice for 5 min.

Seventeen μL of a Master Mixture containing M-Murine leukemia virus reverse 81 transcriptase (Invitrogen, CA), 5 × reaction buffer (Invitrogen, CA), 100 mM DTT, 10 mM of each dNTP and RNAsin (Promega, Madison, WI) were added into each sample.

The samples were then incubated in a Thermal Cycler (Applied Biosystem, Foster City,

CA) at 40 ºC for 60 min then at 94 ºC for 5 min. Following cDNA synthesis, real time

PCR was performed on an ABI Prism 7700 Sequence Detection System (Applied

Biosystem, Foster City, CA). Each cDNA was used as a template in a PCR amplification reaction containing (1) a set of primers and a probe with dual labeled of 3’ FAM and 5’

MGB for the R2 transcripts, and (2) primers and a dual labeled probe with 3’ FAM and

5’ TAMRA for the housekeeping gene ABL. For each reaction, the critical value CT for the target transcript was determined. R2 and ABL calibration standards were custom- synthesized by cloning R2 and ABL cDNA using the TA cloning kit (Invitrogen, CA).

Diluted standards were amplified to create separate calibration curves (CT versus log copy number) for the R2 and ABL. The amounts (copy numbers) of R2 mRNA and housekeeping transcript (ABL) in each sample were calculated against the calibration curves run at the same time. Expression of R2 mRNA at each time point was then reported as the copy number of R2 normalized to ABL by dividing the copy numbers of

R2 to that of ABL. Fold change of R2 mRNA after treatment was expressed as the ratio of the R2 expression to that of pretreatment.

3.2.10 Measurement of R2 protein levels

Western blot analysis was used to measure R2 protein expression levels after

K562 cells were transfected with GTI-2040-neophectin complexation or controls in Opti-

82 MEM for 6 hrs and followed by a continuous incubation in RPMI medium with 10% FBS for a total of 48 hrs. Cells were washed twice with phosphate-buffer saline (PBS), collected in serum free media, and lysed in cell lysis buffer (50 mM PH 7.6 Tris-HCl,

250 mM NaCl, 5 mM EDTA, 2 mM Na3VO4, 50 mM NaF) with 1% protease inhibitor cocktail (P8340, Sigma) for 30 min on ice, and sonicated three times for 20 seconds at each round. Protein concentration was determined using BCA protein assay method

(Pierce, Rockford, IL). The cell lysates were incubated with 6x SDS loading buffer (100 mM, pH 6.8 Tris, 200 mM DTT, 4% SDS, 20% glycerol, and 0.015% bromphenol blue) and boiled for 5 min. Then, 30 μg of protein was fractionated on 15% SDS-PAGE and transferred to nitrocellulose membranes. The R2 protein was detected with a goat antihuman R2 polyclonal antibody (E-16) (Santa Cruz Biotechnology, Santa Cruz, CA), followed by a HRP (horseradish peroxidase) conjugated antigoat IgG secondary

antibody. The Mr 45,000 R2 protein was visualized by ECL (Amersham, Arlington

Heights, IL). For internal loading control, the membrane was stripped for 10 min with stripping buffer (100 mM 2-mercaptoethanol, 2% SDS, 62.5 mM Tris-HCl pH 6.7) and then was re-probed with goat anti-GAPDH antibody. Results were quantified by densitometry. Changes in R2 protein levels after treatment were expressed as the percentage of the GAPDH normalized R2 density to that from the pretreatment.

3.2.11 Cytotoxicity studies by MTS assay

K562 cells were grown in log phase and suspended at 5 × 104 cells/mL in fresh

RPMI medium with 10% FBS or in Opti-MEM, when used in the GTI-2040 transfection.

Cell suspensions (100 μL) were plated into each well of a 96-well tissue culture plate.

The wells on the four outer sides of the plate were placed with medium alone to prevent evaporation of the inside wells. For single treatment with GTI-2040, serials GTI-2040 solutions (0.01-300 μM) in Opti-MEM with or without protamine sulfate were added into each well. After transfection for 6 hrs, RPMI medium with 20% FBS was added and the incubation was continued for a total of 72 hr. Cell viability was also evaluated with protamine sulfate alone in a concentration range of 0.1 to 30 μM. For sequential drug exposure of GTI-2040 followed by cytarabine, cells were transfected with 0.5 or 1.0 μM of GTI-2040 in Opti-MEM for 6 hrs. Then the solutions of cytarabine in the concentration range 0.001-10 μM diluted in RPMI medium with 20% FBS were added, and the plates were incubated continuously for a total of 72 hrs. The cell viability was determined by the MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- sulfopheyl)-2H-tetrazolium), which is bioreduced by cells into a formazan product that is soluble in tissue culture medium. Briefly, 20 μL of MTS/PMS (phenazine methosulfate)

(ratio 20:1) mixture was added into each well and plates were incubated for 1-4 hrs at 37

ºC in a humidified, 5% CO2 incubator. Absorbance was read at 490 nm on a microplate reader Germini XS (Molecular devices, CA). Three replicates were performed at each drug concentration. Data were plotted and IC50 values were calculated using inhibitory

84 effect sigmoidal Emax model in WinNonLin software (version 4.0, pharsight, Mountain

View, CA).

3.3 RESULTS

3.3.1 Cellular uptake of GTI-2040 without or with transfection agents

A previously validated ELISA method (Chapter 2) was used to measure the intracellular concentration of GTI-2040 in the presence or absence of transfection agents.

Exposure of 0.2 μM of naked GTI-2040 for 6 hrs in K562 cells resulted in cellular concentration of 13.5 ± 5.6 nM. In contrast, when 0.2 μM of GTI-2040 was transfected to

K562 cells for 6 hrs in the presence with cationic lipids such as oligofectamine or neophectin, intracellular concentrations of GTI-2040 in K562 cells were found to increase by about 10 and 13 folds, respectively. Therefore, cellular uptake of GTI-2040 was enhanced significantly after complexed with the cationic delivery agents.

Maximum uptake of GTI-2040 (0.2 μM) was achieved with IC at 158 ± 34 nM at

4-6 hrs after delivery with neophectin, as shown in Figure 3.1A. In the first 6 hrs, GTI-

2040 and neophectin complexation was carried out in the serum-free Opti-MEM medium, then the medium was replaced with RPMI medium containing 10% FBS. It was found that intracellular concentrations were reduced slightly at the time when serum containing medium was added and then was returned to the peak concentration in 24 hrs, as shown in Figure 3.1A. Based on these results, 6 hr was chosen as the transfection duration in the subsequent cell transfection experiments. As shown in Figure 3.1B, after 6 hrs transfection with 0.2 μM of GTI-2040 in K562 cells complexed with neophectin, 85 efflux of cellular GTI-2040 reached a plateau in 3 hrs. The cumulative efflux of GTI-

2040 at 13 hr in the drug-free medium was 33.4 ± 5.6 nM, approximating 20% of intracellular GTI-2040. Therefore, efflux of GTI-2040 from K562 cells occurred rapidly, and only a small fraction of IC was released from the cells.

To visualize the enhanced cellular uptake of GTI-2040 delivered by cationic lipids, we used a fluorescein (FITC) labeled GTI-2040 to treat K562 cells with or without the transfection agents (oligofectamine or neophectin). Flow cytometry (Figure 3.2) results showed that without transfection agents, K562 cells accumulated FITC-GTI-2040 with a mean fluorescence intensity of 19.72 FL1-Height. When the cells were treated with FITC-GTI-2040 complexed with oligofectamine or neophectin, the mean fluorescences in cells were determined to be 67.97 and 433.07 FL1-Height, respectively, which was 3.4 and 20 folds higher than the cells without cationic lipid transfection, respectively. The right shifts of the curves (C and D in the Figure 3.2) indicates the increase of the fluorescence intensity of cellular FITC-GTI-2040 with the delivery of oligofectamine and neofectin, respectively (Figure 3.2C, 3.2D). Therefore, cationic lipids greatly enhanced the cellular uptake of GTI-2040.

Intracellular uptake of FITC-GTI-2040 in K562 cells with or without transfection reagents was also investigated using laser scanning fluorescence microscope. After transfection and fixation, Cells were treated with DAPI for nuclear staining. As shown in the left panel of Figure 3.3A, cells treated with GTI-2040 alone showed a faint fluorescence in the outlining area mainly from residual membrane bound FITC-GTI-

2040. In contrast, FITC-GTI-2040 complexed with cationic lipids i.e. neophectin,

86 oligofectamine or protamine sulfate resulted in a bright and diffuse cytoplasm or nucleus distribution after 6 hrs transfection (Figures 3.3B, 3.3C and 3.3D). Of interest, green fluorescence from FITC-GTI-2040 was observed in a higher percentage of cells transfected with neophectin than with oligofectamine or protamine sulfate. This may indicate a higher transfection efficiency of neophectin on GTI-2040 in K562 cells. In addition, neophectin was reported to have a low cytotoxicity (Manual of Neophectin,

Neopharm). Therefore, in the subsequent evalualtion of GTI-2040 on the expression of

R2 mRNA and protein in K562 cells, neophectin was chosen.

3.3.2 Subcellular distribution of GTI-2040

As both flow cytometry and fluorescence microscopy did not reveal the GTI-2040 localization in cytoplasm and the nucleus in a quantitative manner, cell subcellular fractionation was performed to examine differential GTI-2040 distribution in K562 cells administered alone or complexed with neophectin. Internalization of 0.5 µM free GTI-

2040 was quite low with its total IC at 3.6 ± 1.2 pmole/mg protein, which is only 12% of the IC (30.6 ± 6.2 pmole/mg protein) of the drug delivered by neophectin (Figure 3.4).

Approximately 2-fold higher intracelluar GTI-2040 (mean nucleus/cytoplasm 1.2/0.7 pmole/mg protein) was found in the cytoplasm than that in the nucleus for the noncomplexed GTI-2040, while greater than 4-fold of GTI-2040 was found accumulated in the nucleus (nucleus/cytoplasm drug ratio 20.9/4.3 pmole/mg protein), when delivered with neophectin (Figure 3.4). To test a possible contamination to the nucleus with cytoplasm during the cell fractionation, LDH, a cytosolic enzyme, was measured in both

87 fractions. Less than 10% of LDH concentration was detected in the nuclear fraction compared to the cytoplasm lysate, indicating a minimal contamination.

3.3.3 Effect of temperature on the cellular uptake of GTI-2040

Figure 3.5 shows that ICs of GTI-2040 in K562 cells were 2-6 folds higher when transfected at 37 ºC than at 4 ºC at all concentrations evaluated. When cells were exposed to 1 µM GTI-2040 complexed with neophectin, IC was observed to be 387.8 ± 31.2 nM at 37 ºC, as compared to 60.2 ± 10.8 nM at 4 ºC. These results may imply that the cellular uptake of GTI-2040 complexed with neophectin in K562 cells was temperature dependent.

3.3.4 Cellular uptake of GTI-2040 from the human blood

Cellular uptake of GTI-2040 in PBMC and RBC obtained from two human donors was investigated and the results are shown in Figure 3.6. As shown, appreciable uptake in PBMC (12.3 nM and 14.4 nM) and RBC (4.7 nM and 8.5 nM) following incubation of free GTI-2040 (0.5 μM) for 2 hrs was observed, with the former being at higher levels.

3.3.5 Pharmacological effects of GTI-2040 on K562 cells

As shown in Figure 3.7, ICs of GTI-2040 in the nucleus were determined to be

12.1 ± 4.1, 15.4 ± 2.3, 25.4 ± 9.4 and 39.0 ± 10.3 pmole/mg protein, when cells were treated with 0.1, 0.2, 0.5 and 1.0 µM of GTI-2040 complexed with neophectin, respectively. Thus, ICs was found to increase with an increase in exposure concentrations. In GTI-2040-neophectin complex treated K562 cells, R2 mRNA levels were determined and the results are shown in Figure 3.8. As shown, the expression of R2 mRNA was reduced 26%, 35% and 52% after treatment with 0.2, 0.5 and 1.0 µM of complex, respectively. Therefore, down-regulation of R2 mRNA in K562 cells by GTI-

2040-neophectin complex occurred in a concentration dependent manner. This down- regulation effect was also found to be sequence-dependent, since the inhibition of R2 mRNA was not observed in the scrambled and mismatch controls, also delivered with neophectin. Interestingly, cells treated with 0.5 µM free GTI-2040 without cationic lipid did not show down-regulation of R2 mRNA, possibly because of low nuclear IC (28 ±

5.1 nM).

Parallel experiments to examine R2 protein expression were also carried out in

K562 cells and the results are shown in Figure 3.9. As shown, after 48 hrs incubation with GTI-2040 complexed with neophectin, the expression of R2 protein was suppressed compared with the controls (neophectin transfected scrambled and mismatch GTI-2040).

No significant alteration of R2 protein expression was observed in the controls relative to the untreated cells. Treatment with free GTI-2040 (0.5 µM) also did not affect the 89 expression of R2 protein. Following treatment with GTI-2040 neophectin complex, down-regulation of R2 protein was observed (Figure 3.9). Thus, sequence-specific inhibition of R2 protein was also observed in K562 cells treated by GTI-2040 complexed with neophectin. Expression of R2 protein diminished when GTI-2040 was increased from 0.1 to 1.0 µM (Figure 3.9A), indicating that the down regulation was exposure concentration dependent. Although down-regulation of R2 protein was most pronounced following 48 hr incubation, the inhibition became evident at 24 hrs, and the effect sustained to 72 hrs after treatment (Figure 3.9B).

3.3.5 Effect of GTI-2040 on the cytotoxicity of cytarabine

Effect of GTI-2040 on cytotoxicity of cytarabine in K562 cells was examined at two levels of the PS-ODN, 0.5 and 1.0 μM transfected with protamine sulfate. As shown in Figure 3.10, no cytotoxicity was observed in K562 cells with the treatment of GTI-

2040, protamine sulfate or GTI 2040-protamine sulfate complex alone. Cytotoxicity of cytarabine was observed with an estimated IC50 of 0.51 μM. Following pre-incubation with GTI-2040-protamine containing GTI-2040 at 0.5 μM or 1.0 μM, IC50 values of cytarabine were reduced to 0.14 μM (a 3.6-fold increase in cytotoxicity) and 0.02 μM (a

25.5-fold increase in cytotoxicity), respectively. This effect may need to be further verified in other cell lines using other delivery agents such as neophectin.

90 3.4 DISCUSSION

The development of antisense oligonucleotides as a novel gene therapeutic strategy has progressed rapidly in recent years. However, the level of understanding of their biodistribution and target interactions is still far from adequate. Since the targets of antisense compounds reside inside the cells, analysis of cellular uptake as well as the subcellular trafficking is important in order to correlate its pharmacological effect. Until now, the conventional analytical methods for oligonucleotides were neither specific nor adequately sensitive for the analysis of intracellular levels of GTI-2040. To solve this problem, we developed and validated a sensitive and specific ELISA method to quantify

GTI-2040 in different biological matrices. The high sensitivity and selectivity of this assay have made the assessment of cellular uptake possible (11,133).

Most studies have found that ODNs are initially taken up by cells through endocytosis and accumulate in an endosomal-lysosomal compartment, a pharmacologically non-productive site (40,43,134,135). When some of the ODNs are released from endosomes and enters the cytoplasm by an undefined mechanism, the

ODNs can rapidly distribute into the nucleus, permitting interactions with nuclear RNAs

(136). Our data showed low uptake and no antisense activity of GTI-2040 in K562 cells in the absence of delivery reagents, similar to another antisense G3139 (11). In vivo, the prolonged exposure of free antisense drugs may result in substantial drug distribution in the cytoplasm and nucleus (52). Other factors such as the in vivo 3-dimensional extracellular matrix may also contribute to the favorable cellular uptake of ODNs in vivo

(48,51). In K562 cells, transfection with cationic lipids (oligofectamine or neophectin) tremendously increased the cellular uptake of GTI-2040 as illustrated by flow cytometry 91 and microscopy, with at least 10-fold enhancement in ICs of GTI-2040 after complexed with neophectin.

Cationic lipids not only enhanced the amount of cellular uptake of GTI-2040 in

K562 cells, but also modified the intracellular distribution compared with the free drug.

Our results showed that >4 fold increase in the nuclear accumulation of GTI-2040 was found after transfection with neophectin when compared with the free drug. Recent work has shown that the lipid-oligonucleotide complex is internalized by endocytosis (45), thereafter, the complex probably induces a flip-flop of anionic phospholipid in the endosome membrane, leading to the neutralization of the cationic lipid charge, followed by the dissociation between the ODNs and the cationic lipid, and the release of the ODNs from the endosomes (47). Efflux of the drug from the cell may have little effect on the subcellular distribution, since only a small portion of GTI-2040 was released outside of the cells (Figure 3.1B).

Our data show that cellular uptake of GTI-2040 in K562 cells in combination with neophectin is temperature-dependent. This is consistent with the previous reports that cellular uptake of PS-ODNs was mediated by a dose-, time-, and temperature-dependent processes through the mechanisms of receptor mediated endocytosis or fluid phase pinocytosis (43,134-137). In addition, a number of papers have reported the findings of several cell membrane proteins involved in the binding and localization of PS-ODNs. For examples, cell membrane proteins ILT7 and M11S1 were identified binding with a 20- mer CpG-ODN in glioblastoma cells (41) and a 66 kDa membrane ODN receptor for a 25 mer PS-ODN was purified from HepG2 cells (138). To determine whether or not an active transport was involved in the uptake of GTI-2040, more studies are required, such 92 as examination the effects of co-treatment with sodium azide, an inhibitor of active transport or with other competitive polynucleotides.

For the first time we have shown the uptake of non-complexed GTI-2040 in

PBMC and RBC following in vitro human blood incubation (Figure 3.6). In vitro cellular uptake characteristics may be different between cell lines and cells obtained from patients. For example, in patient’s leukemia blasts, ICs of G3139 was found 2- to 7-fold higher than that in K562 cells (11,139). However, the in vitro uptake of free GTI-2040 in

PBMC and RBC did not exceed the ICs in K562 cells, which may be due to the different incubation time used in the two cell cultures. It is interesting to observe that PBMC demonstrated higher ICs than that in RBC in both of the two subjects.

When complexed with neophectin, sequence-specific and dose-dependent down- regulation of target R2 mRNA and protein were observed in K562 cells treated with GTI-

2040. These results verify the antisense effect of GTI-2040 in leukemia cells and provide the rationale for its use in the treatment of leukemia. Good correlations of dose, nuclear

ICs and target inhibition were also found in K562 cells. With an increase in dose, enhanced accumulation of GTI-2040 in the nucleus was found (Figure 3.7), resulting in a more significant down-regulation of mRNA and proteins (Figures 3.8, 3.9). Therefore, for PS-ODNs that exert their antisense effect through the recruitment of RNase H, it suggests that distribution of ODNs in the nucleus would be more suitable in correlation with the inhibition of target mRNA, since RNase H is enriched in the nuclei.

Cytarabine (cytosine arabinoside, ara-C) is widely used as the initial or salvage treatment for AML. After activation to its anabolite ara-C triphosphate (ara-CTP) by 3 sequential enzymes, the anabolite competes with the natural dCTP for DNA 93 incorporation by DNA polymerases, resulting in termination of DNA chain elongation

(111-113). Our preliminary results showed that after pretreatment with 0.5 and 1.0 μM of

GTI-2040, cytotoxicity of cytarabine increased 3.6 and 25.5 fold in a dose-dependent manner. Therefore, GTI-2040 enhanced the cytotoxicity of cytarabine in leukemia cells.

Such a beneficial effect could be a result of the inhibition of R2 by GTI-2040, which may produce an imbalance in the dNTP pool and enhance the DNA incorporation of ARA-

CTP thereby lead to more irreversible cell damage. Therefore, synergistic effect could be possible using the combination of GTI-2040 and cytarabine in the treatment of leukemia.

In the future study, the dose sequence of the combination of GTI-2040 and cytarabine could be examined in order to obtain an optimal synergistic effect.

In summary, the cellular uptake of GTI-2040 was characterized in K562 cells.

Cationic lipids or liposomal reagents were found to increase the cellular uptake of GTI-

2040 in a temperature-dependent manner and alter the intracellular distribution of GTI-

2040. GTI-2040 complexed with neophectin down-regulated R2 expression in a sequence specific and concentration-dependent manner. Concentration of GTI-2040 in the nucleus was found to correlate with the exposure dose and to the inhibition of R2 mRNA and protein expression. Combination of GTI-2040 with cytarabine produced an increase in cytotoxicity.

200

180

160

140

120

100

IC of GTI-2040 (nM) 60 40

0 0 5 10 15 20 25 30 tim e (h r)

40 35 30 25 20 15 10 IC of GTI-2040 (nM) 5 0 0 5 10 15 tim e (hr)

Figure 3.1 Uptake and efllux of GTI-2040 in K562 cells transfected with neophectin.

IC of GTI-2040 was measured by ELISA method. A. Time course of cellular uptake of GTI-2040 after 0.2 μM of GTI-2040 was incubated with neophectin in K562 cells for 0.5,1,2,4,6,8,12,24 hrs. B. Cumulative efflux of GTI-2040 from K562 cells. GTI-2040 (0.2 μM) was transfected with neophectin for 6 hrs. Cells were resuspended in drug-free medium. An aliquot of medium was collected for GTI-2040 quantification at times of 0.16, 0.5, 1, 2, 3, 13 hrs.

A B

C M1 D

Figure 3.2 Flow cytometry analysis of cellular uptake of FITC-GTI-2040 in K562 cells in the absence or the presence of transfection reagents. Histograms were drawn to compare difference in fluorescence intensity between treatments. Y-axis is cell counts and X-axis represents the logarithm of fluorescence intensity (FL-1 height). A: autofluorescence intensity of untreated K562 cells B: free 0.5 μM FITC-GTI-2040; C: 0.5 μM FITC-GTI-2040 complexed with oligofectamine D: 0.5 μM FITC-GTI-2040 complexed with neophectin

Figure 3.3 Laser scanning microscopy analysis of cellular uptake of FITC-GTI-2040 in K562 cells. 0.5 μM of FITC-GTI-2040 was incubated in K562 cells in the presence or absence of the transfection reagents for 6 hrs. Cells were washed and analyzed under microscope. The left panels show the fluorescence intensity of internalized FITC-GTI-2040 (green). The right panels show the counter-staining of cell nucleus by DAPI (blue). A. Cells were transfected with FITC-GTI-2040 alone. B. Cells were transfected with FITC-GTI-2040 and neophectin. C. Cells were transfected with FITC-GTI-2040 and oligophectamine. D. Cells were transfected with FITC-GTI-2040 and protamine sulfate.

30 cytoplasm 25 nucleus whole-cell lysate 20

IC of GTI-2040 (pmole/mg protein) 5

0 Free GTI-2040 GTI-2040 neophectine

Figure 3.4 Subcellular distribution of GTI-2040 with or without the delivery of neophectin. ICs of GTI-2040 in subcellular fractions were assayed after free GTI-2040 (0.5 µM) and GTI-2040 (0.5 µM) complexed with neophectin were incubated in K562 cells at 37 ºC for 6 hrs. Values represent the Mean ± SD (n=3 per group). Open bars: total cell lysates; left diagonally lashed bars: cytoplasm fractions; right diagonally lashed bars: nuclear fractions; Values represent the Mean ± SD (n=3 per group).

450 4 ºC 400 350 37 ºC 300

250 200 150 Intracelllular

GTI-2040 (nM) 100 50 0 0.1 0.2 0.5 1 GTI-2040-neophectin (µM)

Figure 3.5 Effect of temperature on the uptake of GTI-2040 complexed with neophectin. ICs of GTI-2040 in K562 cells were measured after 0.1,0.2,0.5 and 1 µM of GTI-2040 complexed with neophectin was incubated in K562 cells for 24 hrs at 4 °C and 37 °C, respectively. Values represent the Mean ± SD (n=3 per group). Open bars: ICs of GTI- 2040 at 4 °C; Left diagonally hatched bars: ICs of GTI-2040 at 37 °C.

100

14 RBC 12 PBMC

8 6

IC of GTI-2040 (nM) 4

2 0 Subject 1 Subject 2

Figure 3.6 ICs of GTI-2040 in human peripheral blood mononuclear cells (PBMC) and red blood cells (RBC) after 0.5μM of free GTI-2040 were incubated in human blood from two subjects for 2hrs.

101

39.0 50

40 25.4

30 15.4

20 12.1

Intracellular Nuclear GTI-2040(pmole/mg protein) 0 0.00.20.40.60.81.01.21.4 GTI-2040 complexed with neophectin (µM)

Figure 3.7 ICs of GTI-2040 in the nucleus of K562 cells after cells were exposed to different dose of GTI-2040 complexed with neophectin for 24 hrs at 37 ◦C. Values represent the Mean ± SD (n=3 per group).

102

0.8

0.6

0.4

0.2

R2 mRNA fold change to cell control 0

. t. e o n eo eo e e N N fr Neo Neo cont + +N + ll o co -G + 0.5 M ce e N C is-G+ G .2uM S M 0.5u G 0 G G 1.0uM

Figure 3.8 Expression of R2 mRNA in K562 cells with treatments of complexed GTI- 2040, free GTI-2040 and scrambled or mismatch GTI-2040. R2 mRNA levels were examined by real time PCR. ABL was used as internal standard. Sequence-specific and dose-dependent down-regulation of R2 mRNA was shown in the treatments of GTI-2040 complexed with neophectin. Values represent the Mean ± SD (n=3 per group). G: GTI-2040; Neo: neophectin; SC-G: scrambled GTI- 2040; Mis-G: mismatch GTI-2040.

103

R2 (E-16)

GAPDH

1 2 3 4 5 6 7 8 9

R2 (E-16) B

GAPDH

1 2 3 4 5

Figure 3.9 Expression of R2 protein in K562 cells as measured by western blotting. A: Change of R2 protein levels after K562 cells were exposed to the following treatment for 48 hrs. 1: cell control 2: neophectin control 3: Free GTI-2040 (0.5 μM) 4: scrambled GTI-2040 (0.5 μM) complexed with neophectin 5: mismatch GTI-2040 (0.5 μM) complexed with neophectin 6: 0.1 μM GTI-2040-neophectin complex 7: 0.2 μM GTI-2040-neophectin complex 8: 0.5 μM GTI-2040-neophectin complex 9: 1.0 μM GTI-2040-neophectin complex B. Western blotting results of R2 protein expression after K562 cells were treated with 0.5 µM GTI-2040-neophectin complex for 0 (1), 12 (2), 24 (3), 48 (4) and 72 (5) hrs.

104

120 100

80 protamine 60

GTI 2040-protamine 40 A Cells viability (%) 20 GTI 2040

0 0 10203040 Concentration (micromole)

120

100 80

60 B

viabilitycell (%)l ARA-C single 20 0.5uM GTI+ARA-C 1.0uM GTI+ARA-C 0 0.001 0.01 0.1 1 10 concentration (m icrom ole)

Figure 3.10 Effect of GTI-2040 on cytotoxicity of cytarabin in K562 cells. Cells were transfected with 0.5μM or 1.0 μM of GTI-2040 in the presence of protamine sulfate for 6 hrs and then treated with various concentrations of cytarabine for a total of 72 hrs. A. GTI-2040 and protamine sulfate or the combination did not show cytotoxicity to K562 cells. B. Pretreatment of GTI-2040 and protamine sulfate decrease the IC50 of cytarannin.

105 CHAPTER 4

CLINICAL PHARMACOKINETICS AND PHARMACODYNAMICS OF GTI- 2040 IN PATIENTS WITH ACUTE MYELOID LEUKEMIA AND ITS PK/PD CORRELATIONS

4.1 INTRODUCTION

Recently, more and more PS-ODNs have been advanced to clinical evaluations based on their favorable efficacy and safety in animal studies. There are over 50 antisense drugs currently in clinical trials for the treatment of cancers, Crohn’s disease, and inflammatory disorders with a majority of them being PS-ODNs (6). Safety evaluation, pharmacokinetic and pharmacodynamic studies were performed in a number of phase I trials on antisense oligonucleotides (10,23-25,87,140,141). This class of compounds has generally demonstrated tolerable toxicity in more than 3000 patients thus far with doses up to 6 mg/mg/day for weeks by continuous i.v. infusion (6). For many antisense compounds, the maximum plasma concentrations usually generated the pharmacological effective concentrations for mRNA inhibition on the basis of the in vitro

IC50. In a phase I trial using a chimeric antisense drug OGX-011 for the treatment of prostate cancer, a decrease of the targeting clusterin expression was observed in prostate tumor tissue and 106 lymphnodes (87). However, many clinical studies failed to show consistent inhibition of the target mRNA using peripheral blood mononuclear cells (PBMC) as a surrogate tissue

(10,24). It is difficult to obtain a reliable surrogate biomarker for the evaluation of antisense effect in a clinical trial.

It has been stated that for an antisense oligonucleotide to elicit a pharmacological response, the administered agent must distribute to its target organ and be taken up by the target cells, where it can bind to its cognate mRNA (52). Although pharmacokinetics of

PS-ODNs in plasma has been extensively studied, plasma drug concentrations may not correlate with the target down regulation and disease response as demonstrated on our previous experience with a PS-ODN, G3139, for the treatment of AML. For G3139, we

have found that AUC and Css failed to correlate with the down regulation of target mRNA/protein or disease response. However, intracellular concentration of G3139 in bone marrow mononuclear cells appeared to correlate with the disease response. A higher median intracellular concentration of G3139 was observed in the patients achieving complete remission (CR) compared with the patients showing no response (NR) (11,12).

Therefore, determination of the effective drug concentrations at the sites of action and investigation of the intracellular localization of PS-ODNs are important to evaluate the antisense activity of PS-ODNs. Additionally, intracellular drug concentrations may help to assess the correlations in PK/PD, and clinical response.

GTI-2040, designed to hybridize with the coding region of R2 mRNA of ribonucleotide reductase, has been shown to down-regulate R2 mRNA and protein levels in several human tumor cell lines and in animal xenograft tumor models in a sequence-

107 and target- specific manner (95). Results from the phase I clinical studies of GTI-2040 in advanced solid tumor or lymphoma has demonstrated manageable toxicity and promising efficacy (10). We explore its therapeutic potential on refractory or relapsed AML combined with cytarabine in an ongoing phase I clinical trial. The primary goal of this trial is to assess the maximum tolerated dose and the secondary goal is to evaluate the clinical pharmacokinetics and pharmacodynamics of the drug. In order to perform PK/PD studies of GTI-2040 in AML patients, mononuclear cells in BM were procured and processed for the determination of drug concentrations and the target expression at the same time. Plasma pharmacokinetics was also determined in these patients. A previously developed ultra-sensitive and specific hybridization assay (See Chapter 2) was used to quantify the drug levels in all of the biological matrices. All of the information was integrated and correlations were sought.

4.2 MATERIALS AND METHODS

4.2.1 Drugs and reagents

GTI-2040 (5'-GGC TAA ATC GCT CCA CCA AG-3') was supplied by the

National Cancer Institute (Bethesda, MD) in glass vials as a concentrated sterile solution

(100 mg/ml) and was further diluted with sterile normal saline (0.9% sodium chloride,

USP) for the clinical use. Cytarabine is commercially available in 100 mg, 500 mg, 1 g, 2 g, multi-dose vials in 0.68% NaCl solutions, and was diluted with sterile water for clinical use containing no preservatives.

108 PS-dC 28, a 28-mer polycytidine phosphorothioate oligonucleotide was obtained from Integrated DNA Technologies (Coralville, Iowa). Ultrafree-MC filter (cut-off MW

30,000 Da) was purchased from Millipore (Billerica, MA). BD vacutainer® cell preparation tube was obtained from BD Science (Woburn, MA). Nuclease preparation kit was obtained from the Molecular Motif (Billerica, MA).

4.2.2 Patient Characteristics and Dosing Schedules

Thus far, a total of 31 AML patients were enrolled into this study. Demographic and clinical features upon diagnosis are listed in Table 4.1. The median age was 67 years

(range, 60 to 80 years) and 13 patients were female. Of the 31 patients, 13 patients had refractory AML and 18 patients had relapsed AML. Of the enrolled patients, 16 had an intermediate-risk and 15 had poor-risk cytogenetics according to the Cancer and

Leukemia Group B (CALGB) classification for achievement of CR.

Because increased toxicity is usually observed in elderly patients (>60 years old), when using the high-dose of cytarabine, patients were divided into two Cohorts and treated with different dose regimen designs. Fifteen patients with age ranging from 18-60 yrs old were enrolled in Cohort 1, receiving the dose escalation schedule as listed in

Table 4.2. Sixteen patients, older than 60 yrs old were enrolled in Cohort 2 and received a dose escalation schedule as listed in Table 4.3. Protocols of dose regimens in these two cohorts are shown in Scheme 1. Treatment was initiated with a continuous i.v. infusion

(CIVI) of GTI-2040 as a single agent for the first 24 hrs. Patients were then co- administrated with high dose cytarabine from day 2 as i.v. bolus to the end of treatment as shown in Scheme 1. 109

4.2.3 Plasma Protein Binding

Plasma protein binding of GTI-2040 was determined by the ultrafiltration method at concentrations of 50 nM, 200 nM, 500 nM, 1 μM, 10 μM and 100 μM in human plasma. EDTA (5 mM) was added to inhibit nuclease activity in plasma and to prevent the degradation of GTI 2040. After incubation for 30 min at 37ºC in a shaking water bath, drug-matrix mixture was placed into a disposable Ultrafree-MC (MW cutoff 30,000) filter and was centrifuged at 1500g for 30 min. An aliquot of protein-free filtrate and the initial samples were analyzed using a previously validated ELISA method. The fractions of unbound GTI-2040 were determined from the ratio of the concentrations in the filtrate to the initial concentrations. Samples prepared in saline were processed similarly and used as controls for filter binding.

4.2.4 Sample Collection and Preparation Procedures

4.2.4.1 Separation of plasma, PBMC and RBC from patients’ blood samples

Blood samples were drawn from patients at the following time points, pretreatment, 2, 4, 6, 12, 24, 48 and 72 hr during the infusion and post infusion at 144

(dose removal), 144.25, 144.5 146, 148, 150, 156, 168 and 192 hrs. Simultaneous isolation of plasma, red blood cells (RBC) and peripheral blood mononuclear cells

(PBMC) from the whole blood was performed in BD vacutainer CPT tube by a single centrifugation step as shown in Figure 4.1. Briefly, the whole blood was centrifuged in the tube with Ficoll gradient for 30 min at 1400 g at room temperature. After 110 centrifugation, the upper layer containing the plasma was aspirated without disturbing the

PBMC and platelet cell layer (underneath the plasma) and stored in a -70ºC freezer. The cell layer was then carefully collected with a pipette and transferred into a 15 mL conical centrifuge tube with cap. The cells were washed twice with 15 mL cold PBS. Following centrifugation, the cell pellet was incubated with 200 μL, 0.1μM phosphorothioate 28mer polycytidine (PS-dC28) for 2 min on ice and washed with PBS to remove membrane- bound ODN. The number of viable cells was counted by hemocytometer following dilution with trypan blue. Following addition of 200 μL lysis buffer (10 mM Tris-HCl, pH=8.5, 0.5 mM EDTA, 1% Triton X-100) and incubation on ice for 10 min, the cells were lysed by vortexing and sonication. The homogenate was then centrifuged at

10,000g, and the supernatant was transferred to a new tube and stored in a –70ºC freezer for ELISA and protein assay. After collection of the cell layer, RBC below the gel barrier was removed and stored in –70 ºC.

4.2.4.3 Bone Marrow Samples Preparation

Procurement of bone marrow leukemia blasts from AML patients was in accord with an Institutional Review Board (IRB)-approved protocol at The Ohio State

University Hospital. Bone marrow samples were aspirated from GTI-2040 treated patients before treatment and on Day 1 and Day 5 post treatment, and for some patients also about one month after treatment. In the aspirated bone marrow samples, 15 mL PBS was added and the diluted solution was transferred to a tube with 20 mL Ficoll-Paque plus. Following centrifugation for 15 min at 1200g at 25ºC, the white layer containing the cells was removed and to it was added a mixture of PBS, EDTA and FBS to a final 111 volume of 50 mL. The cell density was determined and aliquots with an appropriate cell numbers were placed into different tubes for different experiments. The tubes were centrifuged for 10 min at 1200 g and cells were harvested. To prepare cell lysate for mRNA determination, the cell pellet was lysed with Trizol and stored at – 80 ºC. To prepare the cell lysate for protein analysis, the cell pellet was added with protein lysis buffer (1 × RIPA buffer (100 mM Tris pH 7.4, 500 mM NaCl, 0.1% Triton X-100), 1 mM PMSF, 1 × protease inhibitor cocktail). The cell lysates were pipetted up and down 5 times and stored at – 80 ºC until analysis. The cells were frozen by suspending the cell pellets with the cell buffer and transferred to liquid nitrogen before analysis. To select the

CD 34+ cells, cell pellets were resuspended into a mixture of PBS, EDTA and FBS.

Blocking buffer was then added into the cell suspension to block the selection of non- specific cells (CD 34- cells). CD 34 Hapten (first antibody) was added and the cell suspension was incubated on ice for 15 min. Following washing by a mixture of PBS,

EDTA and FBS, CD 34 micro beads (second antibody) was added and incubated on ice for 15 min. The cells were then washed and resuspended in a mixture of PBS, EDTA and

FBS, and cell solution was loaded onto a Mac column and filters (BD Bioscience,

Woburn, MA). The column was rinsed 3 times with the mixture of PBS, EDTA and FBS under the MAC magnet to wash out the non-specific cells. The columns were then placed distal from the MAC magnet. The CD 34 + cells were eluted with the mixture of PBS,

EDTA and FBS and stored at – 80 ºC.

112

4.2.4.4 Separation of the nucleus and cytoplasm from the frozen bone marrow cells

The frozen BM cells obtained from patients were placed on ice. At the semi- thawed state, 1 mL of FBS was added and the content was pipetted up and down until completely thawed. Then the sample was centrifuged at 1400 g for 5 min and the cell pellet was harvested. The cells were washed with PBS and viable cells were counted after diluted with trypan blue. After centrifugation, cell pellets were collected and incubated with 200 μL, 0.1 μM PS-dC28 for 2 min on ice followed by washing with PBS. To examine the subcellular distribution of GTI-2040 in BM cells from GTI-2040 treated patients, nuclease preparation kit (Molecular Motif) was used to separate cytoplasm and nucleus according to the detailed procedure as described in the Chapter 3.

4.2.4.2 Urine Sample Collection

Pretreatment and post-treatment 24 hr cumulative urine samples from GTI-2040 treated patients were collected and stored at -70 ºC until analysis.

4.3 ASSAY METHODS

4.3.1 GTI-2040 and protein determinations

Concentrations of GTI-2040 in plasma, urine, PBMC, RBC and bone marrow cell lysates were determined using the previously validated ELISA method. Protein amounts in the cells were measured by the BCA kit (Pierce, Rockfold, IL) and used to normalize the cellular concentrations of GTI-2040. LDH kit was used to detect the contamination of

113 nucleus with cytoplasm by measurement the LDH content in cytoplasm and in nuclear fraction.

GTI-2040 levels in PBMC and BM were converted to intracellular concentrations

(nM) by dividing intracellular GTI 2040 amounts by the cell volume as previously described in Chapter 3 (0.035μg protein equal to 1μL cell volume or 2 × 106 cell number equal to 1μL cell volume).

4.3.2 Quantitation of R2 mRNA levels

Quantitation of R2 mRNA was performed by Real Time RT-PCR. Total cellular

RNA was extracted in Trizol according to the manufacturer’s manual instruction

(Invitrogen, Calsbad, CA). Contaminated DNA was removed by a DNA clean-up kit

(Qiagen, Valencia, CA). The RNA concentration was determined using a UV spectrophotometer (PerkinElmer, Boston, MA). cDNA was synthesized as previously described in Chapter 3. The ABI Prism 7700 Sequence Detection System (Perkin Elmer

Applied Biosystem, Foster City, CA) was used for real-time monitoring of PCR amplification of the cDNA following the Taqman Universal PCR Master Mix protocol as shown in Chapter 3. The previously described primers and probes for R2 and ABL were used for the amplification of R2 cDNAs and the internal control of ABL cDNAs.

Expression of R2 mRNA at each time point was normalized to the copy number of R2 to

ABL. The fold change of R2 mRNA after treatment was expressed as the ratio of the R2 expression to that of pretreatment.

114 4.3.3 Measurement of R2 Protein Levels

Western blotting analysis was used to measure the expression of R2 protein in bone marrow blasts obtained from patients. Bone marrow cell lysates that prepared in protein lysis buffer were thawed and sonicated to fully rupture the cell membrane. After centrifugation at 2000g for 5 min, the supernatant was collected and the amount of protein in the extract was measured by the BCA protein assay (Pierce, Rockford, IL).

Cell extracts were subjected to Western analysis using the antibody against R2 and

GAPDH following the procedures as described in Chapter 3. Results were quantified by densitometry. Changes in R2 protein levels after treatment were expressed as the percentage of the GAPDH normalized R2 density to that of the pretreatment.

4.4 DATA ANALYSIS

4.4.1 Pharmacokinetic (PK) analysis

WinNonLin (Pharsight, Mountain View, CA) was used to fit the plasma concentration-time profiles and estimate the PK parameters. Image J (version 1.34s) software was used to determine the density of protein bands from the western blotting.

4.4.2. Calculation of CLR and CLint in AML patients.

Renal clearance of GTI-2040 in patients was estimated by dividing the cumulative amount of GTI-2040 in 24 hr urine to plasma AUC0-24, as shown in Equation 4. Hepatic clearance from patients was obtained by subtracting renal clearance from the total plasma clearance with the assumption that liver and kidneys are two major pathways for the drug elimination. 115 Aurine0−24hr CLR = Eq. 2 AUC plasma0−24hr

CLH = CL − CLR Eq. 3

Based on the well-stirred (venous equilibrium) liver model, CLH is given by the following equation:

QH × fu ×CLint CLH = Eq. 4 QH + fu ×CLint

which can be transformed into

CLH in vivo CLint = Eq. 5 CLH fu (1− ) QH

where fu is the free fraction of the drug in plasma and QH is the hepatic plasma flow rate (49 liters/hr for a 70-kg human).

4.5 RESULTS

4.5.1 Plasma Protein Binding

GTI-2040 samples prepared in saline exhibited no binding with the filter. GTI-

2040 displayed concentration-dependent high protein binding in plasma. When total GTI-

2040 in plasma was <500 nM, free drug in ultrafiltrate was below the LLOQ of the

ELISA assay, and the protein binding was probably higher than 99.9%. As shown in

Table 4.4, with GTI-2040 concentration increased to 1, 10, 100 μM, mean plasma bindings were found to be 99.8%, 97.6% and 93.5%, respectively, as obtained in triplicate determinations.

116 4.5.2 Plasma Pharmacokinetics

The composite plot of plasma concentration-time profile of GTI 2040 for patients receiving 5mg/kg/day in Cohort 1 (<60 yrs) is shown in Figure 4.2 and for patients receiving the same dose in Cohort 2 (>60 yrs) is shown in Figure 4.3. In both of these two groups, the steady-state concentrations (Css) were achieved within 4 hrs and remained so until the end of CIVI. GTI-2040 plasma concentrations were measurable up to 48 hr post infusion and a bi-exponential GTI-2040 plasma concentration-time profile was observed (Figures 4.2 and 4.3). Therefore, the profiles were fitted to a two- compartment infusion model and the relevant PK parameters were computed as shown in

Table 4.5 and 4.6. As shown, no dose-dependent PK behavior was observed in the dose range from 3.5 to 7.0 mg/kg/day, when dose normalized PK parameters were compared

(Css, AUC). Among PK parameters, no statistical difference in CL and Vss between the two age groups was found, although there was a weak statistical difference in the β value

(elimination rate constant) (p=0.066, Mann Whitney test) between these two groups.

When comparing the PK parameters of GTI-2040 in patients co-treated with different doses groups of cytarabine, no difference was found. Therefore co-administration of cytarabine do not affect the PK behavior of GTI 2040.

4.5.3 Urinary clearance and hepatic intrinsic clearance

Probably due to the high protein binding of GTI-2040, urinary excretion was low but detectable. Cumulative amount of GTI-2040 in 24 hr urine in 21 patients was determined to be in the range of 22-566.5 μg with a dose recovery between 0.007-0.13% 117 (0.034 ± 0.008)%. In 21 patients with AML, total plasma clearances were determined to be in the range of 4.5 to 30.5 L/hr with a mean value of 13 L/hr in 3.5 mg/kg/day dose group and 11 L/hr in the 5.0 mg/kg/day dose group. Renal clearances values ranged from

0.00026 to 0.019 L/hr with a mean value of 0.032±0.0012 L/hr (Mean ± SD, n=21).

Assuming that the hepatic and renal clearances are the two major elimination pathways for GTI-2040, the hepatic clearance may be calculated from the total clearance minus the renal clearance. Additionally, since the renal clearance of GTI-4020 in patients was significantly less than the total plasma clearance value, the total clearance may be used as an approximation of the hepatic clearance in patients. At steady state, plasma concentrations of GTI-2040 in patients were less than 500 nM, and at this concentration,

99.9% of the drug was bound to the plasma proteins based on our measurement.

Therefore, the intrinsic clearance was calculated from Eqution 9 and the values varied from 238.7 to 2846.4 L/hr with a mean value of 758.7 L/hr (SD 526 L/hr, n=20), as shown in Table 4.7.

4.5.4 Intracellular uptake and distribution of GTI-2040 in peripheral blood and bone marrow mononuclear cells

Simultaneous measurements of the time course of GTI-2040 concentrations in plasma and in PBMC and RBC were available in 15 patients. Figure 4.4A shows the average concentration-time profiles of GTI-2040 in plasma, PBMC and RBC, in the entire monitored time period. The post infusion profiles in plasma, PBMC and RBC are shown in an expanded plot with mean ± SD values for each data point (Figure 4.4B). As shown, the steady state concentrations in plasma, PBMC and in RBC were achieved 118 rapidly within 6 hrs. The mean Css values in plasma ranged from 149 to 728 nM with a mean value of 392 nM (SD=179, n=15). The mean Css in PBMC ranged from 3.1 – 61.1 nM with a mean value of 28.0 nM (SD=10.6, n=15) and the mean Css in RBC ranged from 1.9 – 10.1 nM with a mean value of 5.1 nM (SD=2.8, n=15). Compared to plasma,

RBC uptake of GTI 2040 from plasma was <5% of the mean Css in plasma and remained lower than the concentrations in plasma to the levels declined in parallel with those of plasma. In PBMC, the intracellular concentrations (ICs) were much lower than plasma concentration during infusion. However, with the cessation of infusion, the ICs in PBMC remained relatively flat over time and became 2.7-fold higher than that in plasma from 4 hr post-infusion (Figure 4.4B). Compared to the RBC, it appeared that PBMC has a larger capacity to retain GTI-2040, as shown by a 14-fold increase in Css between PBMC and plasma from Day 1 to Day 7, while the fold change in RBC was only 1.91 (Table

4.8). ). In addition, t1/2,β appeared to relate to the ICs in BM; higher ICs were observed in patients with longer t1/2,β (Figure 4.6), which may suggest that a longer retention of the drug in vivo might benefit the cellular uptake of GTI-2040.

ICs of GTI-2040 were measured in BM samples collected at 24 hr and 120 hours after GTI-2040 treatment in only 20 patients because of the limited availability of

samples. The 24 hr mean Css of GTI-2040 in BM was 21.6 nM (1.38-61.4 nM), the 120 hr mean Css of GTI-2040 in BM was 113.6 nM (3.9-324.8 nM) (Figure 4.5). Therefore, it appeared that GTI-2040 accumulated in BM blasts following drug infusion. No difference was observed in ICs of BM either between groups or with different doses (3.5 mg/kg/day and 5.0 mg/kg/day).

119 GTI-2040 appeared to be preferentially taken up in the CD 34+ BM blasts (Figure

4.7). After 24 hrs of drug treatment, in 9 patients with the cell samples available, ICs in

CD 34+ BM blasts ranged from 3.1 to 142.8 nM with a median value at 53.1 nM. While, in the same population of patients, unmanipulated BM cells contained GTI-2040 in the range of 2.5-76.1 nM with a median at 27.9 nM. It is of clinical interest that a more favorable drug distribution in CD 34+ BM blasts exists, since most leukemia blast crisis are found with higher CD 34+ expression (142).

Subcellular distributions of GTI-2040 in the cytoplasm and the nucleus were examined in BM cells obtained from 12 patients. Nucleus contamination with cytoplasm was found to be minimal during the cell fractionation procedure with <10% of the LDH content found in the nucleus fraction. As shown in Figure 4.8, GTI-2040 was found in the nucleus of BM cells following a 24 hrs CIVI of GTI-2040. In 12 patients, median IC in the nucleus was found to be at 1.9 pmole/mg protein (0.2-9.6 pmole/mg protein) and a median IC in cytoplasm was 1.3 pmole/mg protein (0.04-6.4 pmole/mg protein). As shown in Figure 4.8, from the 12 patients with the assessable samples, 7 patients showed a higher IC in the nucleus than the IC in cytoplasm, and 5 patients showed a higher IC in cytoplasm than in the IC in nucleus. The differential subcellular distribution in patients may have implications on the drug effects and disease response.

4.5.5 Pharmacodynamic results

To evaluate the effects of drug perturbation on the target, R2 mRNA levels were measured in BM samples collected at pretreatment, 24 and 120 hrs from the start of GTI-

2040 infusion. Results are shown in Figure 4.9. In younger patients, the median values of 120 the fold change in R2 mRNA compared with pretreatment were 1.08 (n=14, range 0.3-

1.92) on Day 1 and 1.00 (n=14, range 0.1-2.65) on Day 5. Compared to the baseline level of R2 mRNA, among 14 younger patients in Cohort 1, down-regulation of R2 mRNA was found in 4 patients both on Day 1 and Day 5. Six patients were found to have up-regulation of R2 mRNA on both Day 1 and Day 5, although in 4 patients, no change was observed. In Cohort 2, the median values of the fold of changes in R2 mRNA to the pretreatment were 0.75 (n=12, range 0.34-9.7) on Day 1 and 0.29 (n=12, range 0.2-3.5) on Day 5. Among 12 older patients, 6 were found to have down-regulation of R2 mRNA both on Day 1 and Day 5, and 4 patients were found to have up-regulation of R2 mRNA on both days with 2 patients showing no change. Changes in R2 proteins from the pretreatment values in 19 assessable patients were evaluated in BM samples 24 hrs and

120 hrs after drug administration. In younger patients, the median R2 protein change was

105% (n=9, range 34-309%) on Day 1 and 69% (n=9, range 10-985%) on Day 5. Among the 9 patients younger than 60 yrs old, 5 patients showed down-regulation and 4 patients showed up-regulation of R2 protein. In older patients, the median % of R2 protein change was 83% (n=10, range 40-287%) on Day 1 and 87% (n=9, range 10-640%) on Day 5. Of the 10 older patients, 5 showed down-regulation, 3 showed up-regulation of R2 protein and 2 showed no change.

In 8 patients with CD 34+ BM cells, 5 showed down-regulation at 24 hrs following GTI-2040 infusion, one patient showed up-regulation of R2 protein, and the other two showed no apparent change, compared to the pretreatment. Figure 4.10 illustrates a representative result from the western blotting. Figures 4.10A and 4.10B show the examples of down-regulation of R2 protein on Day 1, Day 5 or Day 32 121 following drug treatment, C shows an example of the up-regulation, and D shows an example of down-regulation in BM CD 34+ cells 24 hr after drug infusion. Figure 4.11 shows an example of the change in R2 proteins after 24 hrs and 120 hrs treatment with

GTI-2040 in one patient each in two groups of patients with different age.

4.5.6 Disease Response

Clinical responses in patients in two populations after receiving the treatment of

GTI-2040 combined with high dose of cytarabine are summarized in Table 4.9. In patients younger than 60 yrs old, complete remission was noted in 6 of 14 treated patients. Among them, 3 had relapsed AML and 3 had refractory AML. All 14 older patients except one, who achieved a partial response, failed to show disease improvement following treatment.

4.5.7 Correlations among pharmacokinetics, pharmacodynamics, and response

No correlation of pharmacokinetic parameters of GTI-2040, including AUC and

Css values, with pharmacodynamic endpoints or disease response were found.

ICs in whole BM cell lyates were also found not to correlate with PD endpoints and disease response. However, the nuclear ICs appeared to correlate with the changes in mRNA. Following 24 hrs CIVI with GTI-2040, six patients showed down-regulation of

R2 mRNA, ranging from 0.3 to 0.77 fold relative to the baseline line with a median value of 0.64. Five patients showed up-regulation of R2 mRNA (median 1.4, range 1.1-1.7).

Interestingly, in patients showing mRNA down regulation, nuclear drug ICs (range 0.5-

122 9.6 pmole/mg protein, median 2.78 pmole/mg protein) were higher than those showing up regulation (range 0.2-3.3 pmole/mg protein, median 0.75 pmole/mg protein) (Figure

4.12). Additionally, subcelluar GTI-2040 levels also showed differential distribution patterns between responders (CRs) and nonresponders (NRs). Higher IC of GTI-2040 was found in the nucleus of CRs (n=3) than in NRs (n=5), 62% vs. 20.3%, respectively, while 21.2% of IC of the drug was found in cytoplasm in CR patients and 53.5% in NRs.

However, the overall IC drug levels did not correlate with the change in R2 protein levels.

In the 9 patients with assessable BM CD 34+ samples, only two showed down- regulation of R2 mRNA with one also showing CR. The IC of GTI-2040 in BM CD34+ of this responder (117.8 nM) was higher than the median IC value of 53.4 nM.

There was a good correlation between the changes in R2 mRNA and R2 protein levels both on the 24 hrs and 120 hrs samples, following GTI-2040 infusion in patients

(Figure 4.13). Patients showing down-regulation of R2 mRNA (median 0.55 fold from baseline, n=7) also exhibited a general trend of down regulation of R2 protein (median

75% of pretreatment value, n=6). In contrast, in patients having up-regulated R2 mRNA expression (median 2.55 fold, n=8), their R2 protein levels showed a trend of up- regulation (median 156%, n=8).

In patients younger than 60 yrs old, mRNA results were assessable in 12 patients.

Among them, 6 patients achieved CR while 6 did not show response. As shown in Figure

4.14, in younger patients, there was no significant difference in changes of R2 mRNA levels between the responders and non-responders. Therefore, changes in R2 mRNA may

123 not be a good biomarker as a prognosis predictor for the effect of GTI-2040 in AML patients.

Excellent correlations were observed between the changes in R2 protein levels and the disease responses in patients younger than 60 yrs old. As shown in Figure 4.15, among 9 patients whose clinical response and R2 protein levels were both assessable, 4 showed complete response (CR) and had down-regulated R2 protein levels on both 24 hrs and 120 hrs following GTI-2040 infusion. On the other hand, in the remaining 5 patients who showed no response, the R2 protein levels were all found to be up-regulated at both

24 hrs and 120 hrs. In the CR patients, the median R2 level normalized to the GAPDH protein at 24 hr was 79% (range 33-88%) and 32% at 120 hrs (range 10-48%) from the pretreatment level. In the NR patients, the median normalized R2 level was 210.8%

(range 111-285%) at 24 hr and 257% at 120 hrs (range 163-984%), when compared to the pretreatment level. The difference between the CR and NR patients in changes of the

R2 protein expression showed a statistical difference at both 24 hrs (p= 0.036, Mann-

Whitney test) and 120 hrs (p= 0.016, Mann-Whitney test). Therefore, in patients younger than 60 yrs old in this trial, the changes in R2 protein may be a better predictor for the effect of GTI-2040 on AML patients.

4.6 DISCUSSION

In this clinical trial, our study provides detailed PK/PD data of GTI-2040 administered by CIVI in AML patient stratified as younger (18- 60yr) and older (> 60yr) populations. With the assistance of a highly sensitive ELISA-based hybridization method

(LLOQ=50pM), plasma concentrations of GTI-2040 remained detectable up to 48 hours 124 following termination of the infusion, allowing us to capture the true elimination phase with a t½β >17hrs. This differs from the previously reported half-life of 3 hr (5), and the discrepancy may be due to the limited assay sensitivities used in the previous studies. In fact, in our data we found that the short intermediate half-life of 3 hr may be relate to the tissue distribution phase of the drug. Using this highly sensitive method, it was possible to monitor the intracellular concentration of GTI-2040 in samples of peripheral blood and bone marrow from AML patients, which is valuable in correlation with the drug effect.

The long terminal half-life of GTI-2040 may in part be due to its high protein binding. The high protein binding may also reduce its renal clearance, which only accounts for 0.02% of the total clearance. Therefore, the major clearance pathway of

GTI-2040 was thought to be due to metabolism and such has been investigated elsewhere

(See Chapters 6 and 7). The longer residence time of GTI-2040 in vivo also favors its cellular uptake as shown in the higher ICs of GTI-2040 in patients with longer t ½ β

(Figure 4.6).

Pharmacokinetics of GTI-2040 did not seem to be affected by a concomitant administration with cytarabine. Additionally, plasma pharmacokinetics of GTI-2040 was found not to correlate with toxicity or disease response, and this underscores the importance of monitoring the intracellular drug concentration.

GTI-2040 was found to be taken up by PBMC and RBC. Notably, intracellular concentrations remained higher in PBMC than in RBC during the entire monitored time course. Previously, a preferential uptake of GTI-2040 in PBMC rather than in RBC was also observed from in the in vitro incubation in whole blood (Chapter 3); therefore, there may exist an unknown mechanism for GTI-2040 uptake in selective cell-type. Two 125 scenarios were hypothesized: 1) expression of carrier proteins on the membrane of

PBMC facilitated the internalization of GTI-2040 through an efficient active transport.

Recently, a number of papers have reported the findings of proteins involved in the binding and localization of the PS-ODNs in cells (40-42). The cellular uptake of PS-

ODNs has been characterized as a dose-, time- and temperature-dependent processes through the mechanisms of receptor mediated endocytosis or fluid phase pinocytosis

(43,134-137). Temperature dependent cellular uptake of GTI-2040 was also observed in the K562 cells (Chapter 3). Therefore, perhaps via an active drug transport, a higher and sustained ICs levels were attained in PBMC than in RBC, in which an efficient transport pathway may be absent. In fact, the concentration-time profile of RBC is essentially in parallel to that in plasma, suggesting the involvement of a passive diffusion of the drug to

RBC. 2) GTI-2040 bound to specific intracellular factors in the PBMC, probably some nuclear factors, and was subsequently trapped in the PBMC with a decreased efflux rate.

This hypothesis is based on the fact that RBC dose not contain a nucleus and the drug may have a higher efflux rate in RBC.

Robust cellular uptake of GTI-2040 was observed in bone marrow mononuclear cells. It is interesting to note that GTI-2040 was preferentially taken up in BM CD 34+ cells as compared to the un-manipulated ones. This trend was also observed for antisense drug G3139 (139). Although the mechanisms for the cell-type selective uptake of PS-

ODNs in BM are unclear, it is possible that certain transporter proteins specifically expressed on the membrane of BM CD 34+ cells might be responsible for the uptake of

GTI-2040. This finding, if further confirmed in patients with other PS-ODNs, would be of clinical importance, since most leukemia blasts are associated with CD 34+ cells 126 (142). Hence the PS-ODNs compounds like GTI-2040 and G3139 could selectively target to the tumor cells in bone marrow, therefore generate less effect on the normal cells.

ICs in the BM cell lysates did not appear to correlate with PD endpoints and disease response, while good correlations between the nuclear ICs with changes in R2 mRNA levels and with disease responses were observed. In order for PS-ODNs to disrupt the targeting mRNA through the activation of RNase H, accessibility of drugs in nucleus is necessary, since most RNase H is located in the nucleus. Therefore, drug concentrations unrelated to the nucleus such as in plasma are not expected to well- describe the interaction of the drug with mRNA and RNas H. On the other hand, GTI-

2040 concentration in the nucleus should provide a better correlation with the down- regulation of the target mRNA. This also suggests that identification of the drug subcellular distribution pattern and the measurement of drugs in the effective organelles or compartments are important to estimate the biological effects. However, nuclear IC of

GTI-2040 did not appear to correlate to the changes of R2 protein level. It is possible that the dynamics of protein was also affected by the complexity in the post-translational process.

Correlation between the changes in R2 mRNA and R2 protein was observed.

Additionally, in younger patients, excellent correlation between the changes of R2 protein and disease response was found, suggesting that GTI-2040 may have a better pharmacological effect in this patient population. We have found that it is more important to monitor the change in target protein, since the changes in the precursor mRNA did not predict the disease response. Additionally, it appears that proteins directly participate in 127 the disease-related biological changes, whereas mRNA may not. Furthermore, an accurate and reproducible quantification method is needed to monitor the expression of the R2 proteins. Since no commercially developed ELISA method was available for the quantification of R2 protein expression, in this study western blotting, which is a semi- quantification method was used to estimate the R2 protein expression. In the future study, development of a more reliable ELISA kit may facilitate an accurate, rapid measurement of expression of R2 protein.

Older patients with AML usually have a poor prognosis. In patients >60 years old, long-term remission is achieved only in <10% of the treated population

(104,107,108). In this trial, the disease response is disappointing in patients >60 years old with no CR observed. The down regulations of R2 mRNA and protein were detected in patients >60 years old. However, this inhibition of targets did not correspond to their poor clinical response. Additionally, examination of the pretreatment copy number of R2 mRNA revealed that in older patients the baseline copy number of R2 mRNA is much less than that in the younger patients with a statistical difference (P=0.0015, Mann-

Whitney test), as shown in Figure 4.16. The median value of baseline copy number of R2 mRNA in older patients was 0.35 (range 0.06-19.4) and was 8.3 in younger patients

(range 0.2-2736). It may suggest that expression of R2 mRNA in older patients was very low to begin with, and therefore unstable and may not relate to the disease status. In addition, in younger patients copy numbers of baseline R2 mRNA seemed higher in CR patients (median copy number 24.2, range 8.2-69.5) than in the NR patients (median copy number 6.9, range 2.3-273.4) (p=0.0513, Mann-Whitney test). Moreover, baseline levels of R2 mRNA were found higher in patients with R2 mRNA down-regulation (median= 128 26.5, n=10) than patients with up-regulation of R2 mRNA (median= 5.9, n=10) with a statistical difference (P=0.0039, Mann-Whitney test). This observation suggested that overexpression of R2 mRNA is important to manifest an antisense effect in GTI-2040 therapy and the expression of R2 mRNA should be examined during patients enrollment.

Probably AML patients older than 60 yrs with small R2 expression may not be expected to achieve good effect with the treatment of GTI-2040.

Because the number of patients and tissue samples tested in this trial were small, the correlation results are preliminary in nature. This underscores the complexity of the clinical evaluation of the antisense activity GTI-2040 and manifestation of an antisense effect may rely on many factors, including cellular drug uptake and distribution, endogenous levels of endonucleases (RNase H), and positive response of the leukemia cells to target down regulation.

In summary, intracellular concentrations of GTI-2040, especially its subcellular distribution patterns are important in evaluation of antisense effect and disease response in patients. PK/PD analysis demonstrated target alterations after drug perturbation and their correlations with IC and response. Effect of GTI-2040 treatment would achieve better response in patients with over-expressed R2 mRNA.

129

parameters Pt <60 yr (n=15) Pt > 60 yr (n=16) Age (years)

Meidan (range) 47 (21-59) 68 (60-79) Gender (No. female) 6 7 Disease status, No, refractory 7 6 relapsed 8 10 No. of prior therapies Median (range) 1 (1-3) 1 (1-3) Prior therapy No. HDAC 6 5 Antisense 0 3 Cytogenetics, No. Favorable 0 0

Intermediate 8 8 Adverse 7 8

Table 4.1 Characteristics of the AML patients enrolled in the Phase I trial with GTI-2040 (OSU protocal # 0304).

130

GTI-2040 CIVI Dose of cytarabine Dose level Pt numbers days 1 to 6 iv q 12 hr 2 1 3.5 mg/kg/day 2.0 g/m days 2 to 7 3 1A 3.5 mg/kg/day 2.5 g/m2 days 2,4,6 3 2A 5.0 mg/kg/day 2.5 g/m2 days 2,4,6 3 3A 5.0 mg/kg/day 3.0 g/m2 days 2,4,6 3 4A 3.5 mg/kg/day 3.0 g/m2 days 2,3,4,6 3

Table 4.2 Dose schedule escalation in Cohort 1: patients <60 yrs old.

131

GTI-2040 CIVI Cytarabine days 2-6 Dose level Pt numbers days 1 to 6 iv over 4hr 1 3.5 mg/kg/day 1.5 g/m2 days 2 to 7 3 2 5.0 mg/kg/day 1.5 g/m2 days 2,4,6 3 3 5.0 mg/kg/day 2.0 g/m2 days 2,4,6 6

4 7.0 mg/kg/day 2.0 g/m2 days 2,4,6 4

Table 4.3 Dose schedule escalation in Cohort 2: patients ≥60 yrs old.

132

GTI-2040 concentration (μM) % protein bound 0.5 99.95 ± 0.02 1 99.91 ± 0.05

10 97.80 ± 2.4 100 94.50 ± 2.5

Table 4.4 Plasma protein binding of GTI-2040. Values represent Mean ± SD from a triplicate.

133

Cohort 1 (<60yr) Dose levels parameters 3.5mg/kg/day (8 pts) 5.0mg/kg/day (6 pts) Css (nM) 224.2 ± 83.4 353.0 ± 69.2 AUC 0 to inf(μM × hr) 30.0 ± 12.5 45.4 ± 4.6

CL (L/hr) 13.0 ± 4.0 15.0 ± 8.8 Vss (L) 19.3 ± 5.3 26.0 ± 11.3 t1/2 α (hr) 0.73 ± 0.39 1.13 ± 0.84 t1/2 β (hr) 29.7 ±20.3 38.3 ± 15.1

Table 4.5 Relevant PK parameters of GTI-2040 in AML patients < 60yrs old given GTI- 2040 at 3.5 or 5.0 mg/kg/day.

134

Cohort 2 (>60yr) Dose levels parameters 3.5mg/kg/day 5.0mg/kg/day 7.0mg/kg/day (3 pts) (9 pts) (3 pts) Css (nM) 268.7 ± 111.5 396.1 ± 200.1 430.3 ± 44.4 AUC 0 to inf 46.1 ± 20.5 54.0 ± 29.2 76.9 ± 10.9 (μM × hr) CL (L/hr) 4.27 ± 0.60 9.3 ± 7.1 8.9 ± 1.4 Vss (L) 4.67 ± 0.62 19.8 ± 17.1 14.6 ± 4.0 t1/2 α (hr) 0.77 ± 0.19 0.66 ± 0.26 0.77 ± 0.16 t1/2 β (hr) - 25.0 ± 10.8 16.7 ± 4.4

Table 4.6 Relevant PK parameters of GTI-2040 in AML patients ≥ 60yrs old given GTI- 2040 at 3.5, 5.0 or 7.0 mg/kg/day.

135

Urine dose aRenal CL cProtein dCLint bCrCL(L/hr) recovery% (L/hr) bound (%) (L/hr) average 0.048 0.0032 6.5 99.94 758.7

SD 0.034 0.0044 2.5 0.1 526.7

aRenal CL: renal clearance was calculated using the equation 2. bCrCL: measured creatinine clearance from patients cProtein bound: In vivo plasma protein binding was estimated from our in vitro plasma protein binding assay. dCLint: Hepatic intrinsic clearance was calculated using Equation 5.

Table 4.7 Renal clearance and hepatic intrinsic clearance from AML patients (n=20) treated with GTI-2040

136

aMean: Ratio of the mean Css or AUC in the cells to those in plasma. bRatio: Accumulation ratio was calculated by dividing the ratio on day 7 (d) by the ratio on day 1(c).

Table 4.8 Comparison of cellular uptake of GTI-2040 from plasma into PBMC with its uptake to RBC. PBMC accumulated more drugs than RBC as shown by a 14 and a 13.3 fold increase in Css and AUC from day 1 to day 7, respectively.

137

Pts < 60 yrs (n=14) Pts >= 60 yrs (n=15) CR/CRi 6 (3 relapsed, 3 refractory) 0 PR/IR 0 1 (relapsed) NR 8 14

Table 4.9 Clinical response in patients treated with GTI-2040 combined with high Dose cytarabin.

138

GTI2040 continuous IV infusion Patients <60 years D1 D2 D 6

Refractory, High -dose cytarabine q12 iv x 1 cycle Relapsed AML

GTI2040 continuous IV infusion Patients >60 years D1 D2 D 6

High -dose cytarabine q24 iv x 1 cycle

Scheme 4.1: Treatment Scheme of GTI-2040 combined with cytarabine in the Phase I trial in patients with acute myeloid leukemia.

139

Figure 4.1 Simultaneous separation of the plasma, PBMC and RBC from patient whole blood using BD Vacutainer cell preparation tube.

140

10000.0

1000.0

100.0

Observed 10.0 Predicted

1.0

0.1 0 50 100 150 200 time (hr)

Figure 4.2 A composite concentration-time profile of GTI-2040 in younger patients (<60yr) with AML after 144 hrs continuous infusion of GTI-2040 at 5mg/kg/day. A two-compartment infusion model in WinNonLin was used to fit the PK profile.

141

5000

1000

100 Observed Predicted 10

1 0 20 40 60 80 100 120 140 160 180 200 time (hr)

Figure 4.3 A composite concentration-time profile of GTI-2040 in older patients (≥60yr) with AML after 144 hrs continuous infusion of GTI-2040 at 5mg/kg/day. A two-compartment infusion model in WinNonLin was used to fit the PK profile

142

10000

1000 A

100

GTI-2040 concentration (nM) 0.1 plasma PBMC RBC 0.01 0 50 100 150 200 250 time(hr)

10000

B 1000 plasma PBMC RBC 100

(nM) concentration GTI-2040 0.1

0.01 140 150 160 170 180 190 200 time (hr)

143

1000

plasma PBMC BM 100 A

10 concentration (nM)

1 day 1 day 2 day 3 day 5 day 7 day 8

1000

plasma (n=9) PBMC (n=6) 100 BM (n=7)

10 concentration (nM)

1 day 1 day 2 day 3 day 5 day 7 day 8

144

700 600 500

400

300

200 R2 = 0.6685 100

IC of GTI-2040 in BM at day 5 (nM) 0 0 20 40 60 80 100 120 -100 beta half lives of GTI-2040 in plasma (hr)

Figure 4.6 Correlation of plasma t1/2,β of GTI-2040 and IC in BM on day 5.

ICs of GTI-2040 appeared to be greater in the patients with a longer plasma t1/2,β.

145

150

100

GTI-2040 IC (nM) IC GTI-2040

unmanipulated BM CD34+ Blasts

Figure 4.7 Comparison of IC of GTI-2040 in CD34 + bone marrow blast cells and in the unmanipulated BM cells. GTI-2040 was preferentially taken up in CD34 + bone marrow blast cells (median IC 53.1 nM, n=9) compared to the unmanipulated BM cells (median IC 27.9 nM, n=9). (Mann Whitney test P=0.38) The horizontal bars show the median values of the IC of GTI-2040.

146

25 nucleus cytoplasm 20 whole cell lysate

protein) 10

GTI-2040 concentration (pmole/mg 0 6 1924252627321014202131

pt codes

Figure 4.8 Subcellular distribution of GTI-2040 in the nucleus, cytoplasm and in whole cell lysis from patients bone marrow cell samples.

red bars: Nuclear fraction, yellow bars: Cytoplasm fraction, green bars: Whole cell lysates. Patients 6, 19, 24, 25, 26, 32 showed higher ICs in nuclei than the ICs in cytoplasm. Patients 10, 14, 20, 21 and 31 showed higher ICs in cytoplasm than the ICs in nuclei.

147

8.5

5.0 A

after 24 treatment 1

fold change of R2 mRNA R2 of change fold 0

pt < 60 yr pt > 60yr

3.5 3.0

2.5 B 2.0 1.5 1.0 0.5

fold of mRNA change R2 a 0.0

fter 120 hr treatment of GTI-2040 pt < 60 yr pt > 60 yr

Figure 4.9 Scatter plots of the changes of R2 mRNA levels in AML patients sorted in two age groups at 24 hrs (A) and 120 hrs (B) following continuous infusion of GTI-2040. The horizontal bars show the median values of the fold change of R2 mRNA.

148

GAPDH

Days after 0 1 5 0 1 5 32 dosing Pt 0304-25 Pt 0304-24

A B

GAPDH

Days after 0 1 dosing 0 1 5

Pt 0304-06 BM CD34+ Pt 0304-06

D C

Figure 4.10 R2/GAPDH quantification by immunobotting in bone marrow cells from patients treated with GTI-2040 for 1,5 and 32 days. A and B show examples of the down-regulation of R2 protein in patients (totally 10 pts showed down regulation of R2 protein from 19 pts with samples available). C shows an example of the up-regulation of R2 protein in a patient (totally 7 pts showed up regulation of R2 protein from 19 pts with samples available). D shows an example of R2 protein down-regulation in BM CD 34+ cells (totally 5 pts showed down regulation of R2 protein from 8 pts with samples available).

149

1100 600 300 200 100 75 50

% level R2/GAPDH 25 0

0h 24h 120h 0h 24h 120h

<60 years (n=9) >60 years (n=10)

150

10.0 P=0.1255 * 7.5

5.0

2.5 (pomle/mg protein) Nuclear IC of GTI-2040

following hr 24 infusion 0.0

pts with R2 mRNA down pts with R2 mRNA up

Figure 4.12 Comparison of the scatter plots of nuclear IC of GTI-2040 in patients with R2 mRNA down-regulation to those in the up-regulated patients after 24 hr treatment. Nuclear IC of GTI-2040 was found higher in the R2 mRNA down-regulated patients (Median 2.78 pmole/mg protein, n=6) than the up-regulated patients (Median 0.75 ,pmole/mg protein n=5) (p=0.1255, Mann-Whitney test). The line across the scatter points show the median values of the nuclear IC of GTI-2040.

151

(n=8)

2.5 R2 mRNA

2.0

1.5 (n=8) 1.0 R2 Protein 0.5 0.0 Untreated Baseline Untreated Fold Changesfrom -0.5 R2 mRNA R2 Protein -1.0 (n=6) (n=6)

Figure 4.13 Correlation of R2 protein changes with change of R2 mRNA levels following 24 hr infusion of GTI-2040 in BM cells from AML patients. In patients showed down-regulation of R2 mRNA, their median value of R2 protein was reduced compared to the pretreatment, and in the R2 mRNA up-regulated patients, their R2 proteins were also found upregulated.

152

3 0 h 24 h

2 120h

1 R2in mRNA Median fold Change

Responder Non-responder n=6 n=6

Figure 4.14 Comparison of R2 mRNA levels at 24 and 120 hrs after GTI-2040 treatment in the responders and in non-responders from patients <60 years old.

No difference of R2 mRNA changes between responder and non-responders after drug treatment.

153

1250 * P=0.0159 0h 1000 24h 750 * P=0.0357 120h 500 400 300

Median change 200 in R2 protein (%) 100 0 Responder Non-responder n=4 n=5

Figure 4.15 Comparison of R2 protein levels at 24 and 120 hrs after GTI-2040 treatment in the responders and in non-responders from patients <60 years old.

Excellent correlations were observed between the changes in R2 protein levels and the disease response in patients <60 years. Significant differences were found in the R2 protein levels both at 24 hrs (p=0.0357) and 120 hrs (p=0.0159) post drug infusion between the responders and non-responders using the Mann-Whitney test.

154

ber 70 60 p=0.0015 50

A 40 30 20

line base num copy A

N 10 R 0

2 m R P t < 60 yrs P t > 60 yrs

ber 75 p=0.0039 50 B

baseline num copy A N

R 0

2 m R down-regulation up-regulaltion

ber 70 p=0.0513 60 50 C 40 30

baseline num copy 20 A

N 10 R 0 2 m R Responders Non-responders

Figure 4.16 Comparisons of baseline copy numbers of R2 mRNA in two age group of patients (A), in patients with down- or up-regualltion of R2 mRNA (B) and in responsed or unresponsed patients.

Significant differences were found in the baseline R2 copy numbers between younger patients and older patients (p=0.0015), between patients with down-regulated R2 mRNA and up-regulated R2 mRNA after 24 hr infusion of GTI-2040 (p=0.0039), and between the responders and non-responders (p=0.0513), using the Mann-Whitney test.

155

CHAPTER 5

POPULATION PHARMACOKINETICS AND PK/PD MODELING OF GTI-2040 IN PATIENTS WITH ACUTE MYELOID LEUKEMIA

5.1 INTRODUCTION

A phase I clinical trial of GTI-2040 (OSU 0304) is currently in progress at The

Ohio State University for the treatment of patients with acute myeloid leukemia (AML).

Concentration-time profiles of GTI-2040 in both plasma and in cells obtained from peripheral blood in patients have been characterized, as shown in Chapter 4. Large variations were observed in the pharmacokinetic (PK) parameters of plasma GTI-2040 and in its intracellular concentrations among patients. In this chapter a population analysis for the plasma PK and intracellular PK of GTI-2040 were performed aiming to probe the reasons for the variability and to identify the covariates related to the variations in PK parameters. Results from the modeling will provide useful information to facilitate dose regimen optimization and ultimately support the purpose of individualized dosing for patients based on each patient’s own PK parameters that could

156 be estimated from the related covariates. As a primary pharmacodynamic (PD) endpoint, expression of target R2 mRNA was monitored during and following the treatment of

GTI-2040 in patients in this trial. To better evaluate the antisense activity of GTI-2040 in patients and to understand its clinical PK/PD relationship, an integrated PK/PD model was developed and the critical factors determining the antisense effects of GTI-2040 were evaluated using a simulation study, based on the developed PK/PD model.

The following descriptions give a very brief overview of NONMEM guided population PK analysis and the related methods used in the development and validation of the PopPK model of GTI-2040 (143,144). Population pharmacokinetic analysis using non-linear mixed effects modeling (NONMEM) has become an important tool for the analysis of pharmacokinetic and pharmacodynamic (PK/PD) data. Traditional modeling approaches focuses on individual parameter estimates. In contrast, population analysis pools the data from all individuals and aims to estimate the population parameters, the magnitude of inter-individual variability in the parameter estimates, and the intra- individual errors in the predictions. The population pharmacokinetic model consists of four basic components: 1) a structural model, which describes the concentration-time profiles and defines the PK parameters, 2) a covariate model, which describes the influence of fixed effects (e.g. demographic factors such as sex and body weight) on the

PK parameters, 3) a random inter-individual error model, which relates the population parameters to individual parameters and accounts the inter-individual variability in PK parameters, and 4) a random residual error, which describes the random errors contributing to the differences between the predicted and the observed concentrations,

157 such as assay errors, model misspecification and within subject variability. The inter- individual random error (symbol “η”, ETA) is assumed to follow an approximately log normal distribution with a mean at “0” and a variance at “ω2”. The random residual error

(symbol “ε”, EPS) is assumed to follow an approximately normal (or log normal) distribution with a mean at “0” and a variance at “σ2”. The first order conditional estimation (FOCE) method is a strongly recommended method to gain accurate predictions of the PK parameters in population type of data. FOCE needs to be used in the estimation procedure, when the inter-individual error model is defined to be a log normal model or the residual random error is defined as a constant coefficient of variation model. Evaluation of goodness-of-fit should be based on the multi-evaluations of the performance from the model, which includes changes in the objective function value, a decrease in unexplained variability and the improvements in the diagnostic plots.

Objective function value (OFV) calculated as minus twice the log likelihood of the data is a global measurement of goodness of fit. Decrease in the objective function value indicates the improvement of data fitting. In addition, elaboration of the model should be accompanied by a decrease in the estimates in the variance in “ω2” or “σ2”, especially with the addition of covariates to explain the kinetic differences among individuals. The other useful evidence confirming the improvement of a model is the pattern in the diagnostic plots including IPRED vs DV (individual prediction outcome vs dependent variable, i.e. observed outcome) and weighted residual plots. When a model is relatively fully specified, all weighted residual plots should show no patterns; the “unexplained” part of the data should display featureless random noise. Covariates are selected into the base model using a forward inclusion significant level of 0.05 (3.84 fall in OFV) and a 158 backward deletion criterion of 0.005 (10.8 fall in OFV). Identifying the explanatory covariates can be a time-consuming task, when working with a long list of candidate parameters. GAM (generalized additive modeling), a specified program in the Xpose can be used to speed up the covariates search (145,146). GAM tests the single or the combination covariates models in a stepwise fashion and retains the covariate models that decrease the Akaike Information Criteria (AIC) most. Each relevant covariate is tested in a hierarchy of possible functions: no model, a linear model and a non-linear model. The search is terminated when no covariate model can decrease the AIC further. Final covariate models identified from GAM need to be further justified by their physiological contributions to the PK parameters. For a final PopPK model, model validation needs to be performed to evaluate the stability of the model and the importance of the covariates.

The internal method and external method are two strategies used in the model validation

(147-149). The external method uses a separate data set to test the estimates from the tested data set and usually needs a large sample size. A number of methods are classified into the internal method that includes: 1) Data splitting, which split the whole data into model building and testing sub-dataset, 2) Cross validation, which contains multiple splits, and 3) Bootstrapping. For data sets with small to middle size of patients, bootstrapping is the most often used validation method. The bootstrapped data sets are constructed by randomly sampling with replacement from the original data set of individual parameter values and corresponding covariate vectors. After a certain number of bootstraps, data sets (hundreds to thousands) are created, and an estimation run with final model on each of the replicate data sets is performed. Distribution of bootstrapped parameters is then evaluated and the mean value and 90% confidence interval (CI) of the 159 bootstrapped parameters are calculated. The model predicted parameters are compared to the bootstrapped mean and 90% CI, good correspondence indicates the stability of the model.

We conducted PopPK model development and validation on plasma GTI-2040 from 30 patients. Since the cellular concentration of GTI-2040 is more important to correlate with the PD changes, population PK of GTI-2040 was also evaluated in PBMC.

In order to understand and predict the pharmacological effect of drugs, it is important to characterize the time course of pharmacological responses in relation to the drug concentrations. For drugs that produce response with a time lag between the effective drug concentration and the pharmacological effect, an indirect response model is usually used according to Jusko WJ et al. (150-152). Indirect response models assume that the drug produces its action by stimulating or inhibiting the production or dissipation of a pharmacological response. For PS-ODNs, the most efficient antisense mechanism is to recruit the RNase H to cleave the mRNA strand in drug-mRNA duplex. Therefore, binding of PS-ODNs with target mRNA could be assumed to stimulate the RNase H activity and induce the mRNA degradation. An indirect response model that controlling the dissipation of the response was therefore used to describe the pharmacodynamics of

R2 mRNA of GTI-2040. In addition, since PS-ODNs have to be internalized in order to reach the target cells and bind to its cognate mRNA, there should be a delay between the plasma drug concentration and the site of action. A hypothetical effect-compartment can be used to link the plasma compartment with the PD response, which accounts for the delay (153-155). This effect compartment in the PK/PD model of PS-ODNs could be

160 viewed as an intracellular compartment that reflects the dynamics of the intracellular trafficking of PS-ODNs. Therefore, an effect compartment linked indirect response

PK/PD model was designed for GTI-2040. After internalization, PS-ODNs must escape endosomal/lysosomal degradation, avoid nuclease attack and efflux, find their complementary mRNA, hybridize and remain associated with mRNA long enough to mediate RNase H cleavage. Simulation study is a good strategy in evaluation of these determinants and barriers, limiting the effectiveness of PS-ODNs and was employed in this PK/PD modeling study (156,157). Results from the simulation study may provide insights and guidelines for improving the success of the antisense technology.

5.2 CLINICAL TRIAL AND METHODS

5.2.1 Clinical Trial

An open-label, single institution, Phase I study of GTI-2040 combined with high dose cytarabine (OSU 0304) on patients with acute myeloid leukemia (AML) is currently underway at The Ohio State University in accordance with an Institutional Review Board

(IRB)-approved protocol at the Ohio State University Hospital.

Briefly, the enrolled Patients must have an unequivocal histological diagnosis of

AML according to the World Health Organization (WHO) classification. Patients also must have refractory AML or relapsed AML. Thus far, 31 patients were enrolled and used in the pharmacokinetic evaluations. The examined dose levels of GTI-2040 were

3.5, 5.0, 7.0 mg/kg/day given as a 144 hr continuous i.v. infusion. Plasma population PK analysis was applied on all 30 patients with 499 plasma concentration-time profiles

161 available and 281 concentration-time profiles in PBMC were available from 15 patients to conduct the population PK in PBMC. Of patients showing the down-regulation of R2 mRNA, 5 had R2 mRNA expression available in the following sampling times: pretreatment, 24 hr (day 1), 120 hr (day 5) and 720 hr (day 30) after drug treatment and were used in the two step PK/PD modeling study. Baseline demographic and physiological parameters of patients were summarized in Table 5.1 and were used for the covariates selection.

Plasma and PBMC GTI-2040 concentrations were determined using the previously validated ELISA method. R2 mRNA expression was quantified using real time RT-PCR using ABL as the internal control.

5.2.2 Software

NONMEM (version V, level 1.1, University of California, San Francisco) was used for the population PK study running under a Windows professional XP platform and a Visual Fortran standard edition 6.0 compiler (Microsoft Corporation). Xpose (version

3.1, Uppsala University, Sweden), an S-plus based program assisting in analysis of the results from NONMEM, was used in NONMEN data output and graphic visualization, model diagnostics and candidate covariate identification. Screening of the influential covariates on PK-parameters was assisted by GAM (generalized additive modeling) program in the Xpose program. The stability of the covariate model was tested by a bootstrap re-sampling procedure run by a bootstrap command in NONMEM.

ADAPT II (version 4) (David Z. D’Argenio and Alan Schumitzky, University of

Southern California) was used in the PD model development and simulation. Nonlinear 162 parametric regression method of Maximum Likelihood was used in the estimation of

PK/PD parameters.

5.3 RESULTS AND DISCUSSIONS

5.3.1 Population PK of GTI-2040 in plasma

As shown in Chapter 4, GTI-2040 in plasma decayed bi-exponentially after the infusion ended and its concentration-time profiles were well fitted by a two-compartment infusion model with a WinNonLin Computer Program. To perform the population PK of

GTI-2040 in plasma, a two compartment model as depicted in Figure 5.1A, was used and coded with the NONMEM built-in model (ADVAN 3) to gain a faster running speed. In this model, CLp is the plasma clearance from the central compartment, Q is the distribution clearance between the central and peripheral compartments, V1 and V2 are volume distributions in the central and peripheral compartments, respectively. The inter- individual random effects model was an exponential error model (ETA) that assumed a log normal distribution for the differences in the PK parameters among patients. The random residual error model was a combination of proportional and additive error terms

[W= (F^2 +THETA^2]^0.5, where F is the predicted concentration and THETA is the estimation of additive error. Since the plasma concentrations of GTI-2040 span several orders of magnitude, a combination residual error model was used to accommodate the errors from both low and high concentrations. The First-order conditional estimation

(FOCE) with interaction was used to obtain the final population parameter estimates. The empirical Bayes estimates of individual PK parameters were obtained with the 163 NONMEM POSTHOC step. Running of the base model resulted in an objective function value (OFV) of 3697.814. Estimated PK parameters and the inter-individual variances and residual errors from the base model (without covariates) are summarized in Table

5.2A. Figure 5.2 shows the diagnostic plots from the base model and Figure 5.2A shows the individual predicted concentrations versus observed concentrations. The closer of the scattered points to the diagonal unit line indicates the better fitting. Figures 5.2B and

5.2C show the residual plots.

Using this model, percentage cell in the bone marrow blasts (cellularity, CELL) was identified as a covariate of the plasma clearance (CLp) and the gender was found to relate to both central and peripheral volumes distribution of GTI-2040 after running the

GAM. As shown in Figure 5.3A, a positive correlation was observed between the cellularity and the residuals of CLp, indicating that an increase of CLp is associated with greater cellularity. Figure 5.3B shows that a larger volume of distribution in the central compartment was found in the male patients than in the female patients. The same trend was observed in the peripheral volume of distribution. A covariate model was therefore established as:

CELL CL = θ (1)× ( ) p 70

V1 = θ (2)× (SEX −1) +θ (3)× (2 − SEX )

V2 = θ (4)× (SEX −1) +θ (5)× (2 − SEX )

Where, θ (1) represents the population mean of the plasma clearance of GTI-2040, 70 was the median value of the cellularity in 30 patients. Sex was set to 2 for female and to 1 for male patients. θ (2 ) and θ (3) represent the population mean of the volumes distribution 164 in the central compartment in female and male patients, respectively. θ (4 ) andθ (5 ) represent the population mean of the volume distribution in the peripheral compartment in the female and male patients, respectively. The control streams of NOMEM codes for

PopPK of GTI-2040 in plasma with covariates incorporated are attached in Appendix A.

After running this model, the objective function value of the final model was 3654.360, a decrease of 43.45 as compared to that of the base model. Stepwise inclusion of CELL in

CLp, SEX in V2 and SEX in V1 resulted in decrease of OFV by 25.002, 10.260 and 8.19

, respectively, in each step. The decreases of OFV in each step were greater than 3.84 and is therefore statistically significant (p<0.05). The backward exclusion of the covariates also validates the significance of their correlation among the PK parameters. As shown in

Table 5.2B, variances in CLp, V1 and V2 were all decreased after the incorporation of the covariates, suggesting that these covariates could to some extent explain the variation of the PK parameters among patients. Cellularity was positively correlated to CLp, which may suggest that the larger number of cells in bone marrow resulted in a higher drug uptake and therefore the larger plasma clearance. Additionally, a difference in the volume distribution between genders was found and this may be associated with gender-related difference in some unknown physiological parameters. Stability of the parameter estimates from the final model was examined using the nonparametric bootstrap based on

100 re-sampled data sets. Out of the 100 resample data sets, 94 converged successfully.

Table 5.3 presents the results of the bootstrap analysis. As shown, low bias between the final model estimates and the bootstrap mean values was observed and the estimated PK parameters were covered in the 90% confidence interval. This indicates the accuracy and stability of this covariate PK model. In addition, representative individual fittings of 165 concentration-time profile on two patients using the Bayesian estimated individual PK parameters were performed, and the results showed a good agreement with the observed profiles (Figure 5.5) indicating the good fitting of the model.

However, the diagnostic plots, including the individual prediction versus observed concentrations and the residual plots in the final model (Figure 5.4) did not show much improvement over the diagnostic plots from the base model. Additionally, this model did not predict the fluctuations in the steady state concentrations, especially near the end of infusion (e.g. 144 hr) in weighted residues versus time plot (Figure 5.4B).

The maximum predicted concentration of GTI-2040 could not reach the similar levels of the observed. This problem may probably be due to the large variations in the observed steady state concentrations during the long-term continuous infusion (6 days). However, the final residual error of 50% is still tolerable in NONMEM and may indicate a reasonable estimation of the random residuals from concentrations in the elimination phase. In the future, inclusion of between occasion variance in the steady state concentration may be considered to improve the estimation of the variations of the steady state concentrations.

5.3.2 Population pharmacokinetics of intracellular GTI-2040 in PBMC

GTI-2040 was detected in PBMC from treated patients and the levels still sustained after drug removal. Since the plasma protein binding of GTI-2040 was rather high, it would retard the celluar passive permeation. However, a number of in vitro experiments suggested an active endocytosis mechanism for the uptake of PS-ODNs in 166 cells (40,43,134,135), based on the observed of dose-, time-, and temperature-dependent uptakes of PS-ODNs and the uptake inhibition by sodium azide, an active transport inhibitor(41,42,137). Therefore, we speculate that the cellular uptake of GTI-2040 was via an active transport process. Saturable distribution parameters (Vm/Km) were used to describe the distribution of GTI-2040 into PBMC from the central compartment. To better describe the concentration-time profile of GTI-2040 in PBMC, a third compartment connected to the central compartment was proposed, as shown in Figure

5.1B. The differentiation equations are listed as follows:

dA1 CLp Q Vm Q = R0 − ( + + )× A1+ × A2 Central compartment dt V1 V1 K m + A1/V1 V 2 dA2 Q Q = × A1− × A2 Peripheral compartment dt V1 V 2 dA3 V CL3 = m × A1− × A3 PBMC compartment dt K m + A1/V1 V 3

where Vm is the maximum uptake rate of GTI-2040 into cells and Km is the substrate concentration that reaches the half maximum uptake rate. V3 is the volume of distribution in PBMC compartment and CL3 is the clearance of GTI-2040 from the

PBMC compartment. Efflux of GTI-2040 from PBMC was assumed to be small in comparison with its plasma concentrations, so that its reverse distribution to plasma is negligible. The error models and the estimation method used in the PopPK analysis of

GTI-2040 in PBMC were similar to those used in the plasma PK model. To avoid the instability of the model by over-parameterization, a two-step strategy for the simulation of the PK parameters in PBMC was employed. Individual PK parameters of CLp, Q, V1 and V2 obtained from the POSTHOC step in the plasma PopPK model for each patient 167 were fixed in the PBMC PK model, and the parameters related to the PBMC compartment including Vm, Km, V3 and CL3 were estimated.

Results from NONMEM showed that the concentration-time profile in PBMC was well fitted by this model (Figure 5.9). Objective function value after running the base model was 1425.820. Results of the parameter estimates and variances from the base model are summarized in Table 5.4A. Goodness-of-fit plots and the residual plots from the base model are shown in Figures 5.6A and 5.6B. After running GAM, white blood cell counts (WBC, × 109) and genders were found to relate to the Km and Vm values, respectively, as shown in the residual correlative plots (Figure 5.7). The covariate models are described as:

1.6 K = θ (1)× ( ) m WBC

Vm = θ (2)× (SEX −1) +θ(3)× (2 − SEX ) where θ (1) is the population mean value of Km, 1.6 is the median value of white blood cell counts (× 109) in 15 patients, SEX was set to 2 for female and to 1 for male patients,θ (2 ) is the population mean value of the maximum uptake rate (Vm) for female patients, and θ (3 ) is the population mean value of Vm for male patients. The control streams with covariates model are attached in Appendix B. After incorporation of the covariates, objective function value was 1315.738, a decrease of 110.082 compared to the base model. As shown in Table 5.4B, inter-individual variances in Vm and Km decreased

12% and 28%, respectively, indicating that the covariates explained in part the variations of these parameters among patients. The intra-individual variance decreased by 31%, suggesting a better prediction with the addition of these covariates. However, the residual 168 error was still 76% and the diagnostic plots (Figure 5.8) did not show a major improvement. Other sources contributed to variations in concentration and the fluctuations in Css need to be identified. Using the Bayes estimated individual PK parameters, good individual fittings of observed concentration-time profile of GTI-2040 in PBMC were observed as shown in the two patients in Figure 5.9.

In the final PK model of GTI-2040 in PBMC, the white blood cell counts appeared to negatively correlate with the Km value (Figure 5.7A). This correlation may suggest that in patients with more white blood cells (WBC), a smaller Km value, i.e. less

GTI-2040 is needed to achieve the half maximum cellular uptake, which means that the uptake of GTI-2040 in WBC may occur more efficiently than in other cell types. If this was true, some specific transport factors may exist in WBC, which favor the intracellular uptake of GTI-2040. In addition, maximum uptake rate of GTI-2040 (Vm) was found to display a gender difference with a large Vm value observed in female patients. Since plasma clearance of GTI-2040 also showed gender dependence, the drug distribution and cellular uptake of drug may behave differently between male and female patients, and may affect their PK/PD outcomes and disease response. This trend needs to be verified in larger population in order to assess the necessity of dose adjustment with genders.

Attempts were also made to extract PK parameters of intracellular GTI-2040 in cells from bone marrow. However, since we only had 3 concentration measurements on each of 20 assessable patients, we were unable to obtain reliable results with this small sample size.

169 5.3.3 PK/PD modeling of GTI-2040 and down-regulation of R2 mRNA in patients

Activation of RNase H is one of the major mechanisms of PS-ODNs in down regulating of their target mRNAs. GTI-2040 specifically hybridizes to R2 mRNA mostly in the nucleus. The resulting duplex triggers the activity of RNase H, which cleaves the mRNA strand in the duplex. Therefore, binding of GTI-2040 is assumed to stimulate

RNase H activity and enhances the degradation of R2 mRNA. Since the indirect response models are usually used in the characterization of the inhibition of the production or dissipation of a pharmacological response(150), it was proposed to define the pharmacodynamics of R2 mRNA down-regulation as an effect of GTI-2040, which is described by the following equation:

n d(mRNA) Emax ×C = Rin − Kdeg × (1+ n n )× (mRNA) dt EC50 + C

We assume a zero-order rate constant (Rin) for the synthesis of mRNA and a first-order degradation rate (Kdeg) for enzymatic cleavage of mRNA. Recruitment of

RNase H by GTI 2040 and R2 mRNA hybrid is considered to increase Kdeg. Emax is the maximum reduced level of R2 mRNA, EC50 is the concentration of GTI-2040 required for half maximal reduction of R2 mRNA, C is the effective concentration of GTI-2040 binding to mRNA, n is the sigmoidicity factor, and mRNA is the normalized R2 mRNA level at time t.

In Chapter 4, we reported that GTI-2040 concentrations in plasma could not predict the change in R2 mRNA expression in bone marrow cells, instead, the nuclear

GTI-2040 concentrations in bone marrow blast cells showed a correlation with the change in R2 mRNA levels in AML patients. Therefore, an effect compartment was used

170 to link to the central compartment that received the drug. The effect compartment is a hypothetical compartment, representing the site of drug action, here most likely being the nucleus in bone marrow cells. Concentrations of GTI-2040 in the effect compartment were defined as the effective concentrations bound to the R2 mRNA. Endocytosis mediated cellular uptake was assumed to be the transport process of GTI-2040 from the central compartment to the effect compartment. The saturable terms (Vm, Km) were used to define the distribution. The integrated PK/PD model as proposed is shown in Figure

5.10. Derived differential equations are described as follow: dA1 Vm,b = R − (K + + K ) × A1+ K × A2 Plasma central compartment dt 0 10 Km,b + C1 12 21 dA2 = K × A1− K × A2 Peripheral compartment dt 12 21 dA3 Vm,b = × A1− K × A3 Effect compartment dt Km,b + C1 30 dA4 E max×C3n = K − K × (1+ )× A4 R2 mRNA compartment dt 04 40 EC50n + C3n where A1, A2, A3 represent GTI-2040 amounts in the central, peripheral and effect compartments, respectively. A4 represents R2 mRNA amount and was normalized to the pretreatment. C1 and C3 are GTI-2040 concentrations in the central and effect compartments and were obtained by dividing A1 and A3 with V1 and V3, respectively.

R0 is the infusion rate of GTI-2040, K10 is the elimination rate constant of GTI-2040 from the central compartment, K12 is the distribution rate constant of GTI-2040 from the central to the peripheral compartments, and K21 is the distribution rate constant of GTI-

2040 from the peripheral to the central compartment. Vm,b is the maximal uptake rate of

171 GTI-2040 from the central compartment to the effect compartment, and Km,b is the concentration of GTI-2040 to reach the half maximal uptake rate. K30 is the elimination rate constant of GTI-2040 from the effect compartment and represents loss of the drug available in binding to the target, which includes the efflux of the drug, drug degradation in the cells or entrapment in the endosomes. R2 mRNA is assumed to be synthesized with a zero-order rate coefficient (K04) and degrades with a first-order rate coefficient (K40).

Without perturbation of drugs, there is a steady state between the synthesis and degradation of R2 mRNA, in which the rate of synthesis equals to the rate of decay, i.e.

K04=K40×Base, where Base is the baseline level of R2 mRNA before drug treatment and is set at 100%. Therefore K40 was replaced as K04/100. Emax is the maximum stimulation fraction of mRNA degradation, and EC50 is the concentration required for half-maximal stimulation.

Since the plasma PK parameters were not influenced by the implementation of the effect compartment, parameters K10, K12, K21 and V1 were fixed in the PK/PD model with their individually estimated values obtained from the plasma PK model.

PK/PD data fitting and parameter estimation were performed with ADAPT II.

Maximum likelihood was used to obtain the final parameter estimates. ADAPT II codes are supplied in Appendix C. Better data fitting was observed, when the sigmoidal factor

(n) was set at 1 rather than set as a variable. Therefore, in the final model n was fixed to

1. In the initial modeling, an over-estimation of down-regulation of R2 mRNA was observed in patients using the proposed model (data not shown). It should be noted that the expression of R2 mRNA is cell cycle dependent and is only expressed in the late

G1/early S phase (92,95). Therefore, only the cells in the S phase provide a stable 172 expression of R2 mRNA, and during this period of time, the cells are subject to antisense activity. It is documented that the length of S phase in mammalian cells represents approximates 30% in one cell cycle (158). Some studies on the cell cycle in cells from leukemic marrow showed wide variation of the percentage of S phase cells in patients and reported the length of S phase to be in the range of 14-35% (159,160). In our study, we assumed that 35% of the cells in bone marrow from AML patients were in the S phase and that the GTI-2040 uptake and intracellular stability (e.g. intracellular degradation and/or efflux) behave equally in each cell. Therefore 35% of the effective GTI-2040 concentration was considered to be be taken up in the S phase-cells and produced the antisense effect. Changes of R2 mRNA with time is thus modified as follows, dA4 E max× (C3× 0.35) = K − K × (1+ )× A4 dt 04 40 EC50 + (C3× 0.35)

After the modification with effective concentrations, the model predictions were in better agreement with the observed data points. Two examples of the fitting results on the time course of R2 mRNA changes in patients are illustrated in Figure 5.11.

Simulations of GTI-2040 concentration profiles in plasma were also demonstrated with the fixed PK parameters. The raw data and the results of the parameter estimates on patient 0306-26 were listed in Table 5.5. As shown in Figure 5.11, the observed R2 mRNA levels agreed closely to the predicted curve (dashed line). Down-regulation of R2 mRNA occurred rapidly after treatment with the drug and the R2 mRNA levels gradually returned to the baseline, when the infusion stopped. From the predicted data, about 2-4 days after drug removal, R2 mRNA recovered to the normal level. The rapid onset and

173 slow recovery of R2 mRNA may be consistent with the rapid uptake and the sustained levels of GTI-2040 in the cells.

This PK/PD model was applied to five patients with R2 mRNA measurements available on 4 collected time points. The average results of estimated PK/PD parameters are summarized in Table 5.6. As shown, the estimated average degradation rate constant

-1 for R2 mRNA is 0.432 hr (K40), which is close to the reported half-life of R2 mRNA

-1 (1.5 hr, Kdeg=0.462 hr ) (161). Cellular uptake parameters Vm, Km were estimated to be

1.13 ± 1.03 nM/hr and 10.2 ± 5.14 nM, respectively. Little information concerning the cellular uptake kinetics of PS-ODNs, especially in hematological derived cells, is available in the literature. Km values, obtained from the in vitro cell lines such as renal epithelial cells, colorectal adenocarcinoma cells, varied from 0.1 nM to 100 nM for PS-

ODNs (41,134,137). The Km value predicted in our PK/PD model is 1.13 ± 1.03 nM/hr and is in the range of the observed data as shown above, indicating a reasonable prediction. In our current data, large variations were observed on the PK/PD parameters among patients. The data on the changes in R2 mRNA were limited, especially during the onset period and the recovered phase of the mRNA. This adds uncertainty to the modeling. However, it may be difficult to increase the sampling number, since it is not practical for intensive bone marrow sampling from patients. In this case, R2 mRNA levels in the PBMC may still be an alternative surrogate marker, but the correlation of the drug concentration and R2 mRNA expression in PBMC in comparison with that in BM cells needs to be validated.

174 5.3.4 Simulations based on the PK/PD indirect response model

Using GTI-2040 as a model drug based on the final PK/PD model and the average

PK/PD parameters of GTI-2040 shown in Table 5.6, the influence of the key PK/PD parameters of the antisense drugs targeting R2 mRNA were evaluated by model simulation. The key PK/PD parameters included are dose, Vm, Km, K30, EC50 and % cells in the S phase. Simulations were conducted by varying the values of the tested parameters, while keeping the values of the other parameters fixed. Since the model predictions showed that R2 mRNA levels returned to the baseline 240 hrs after drug treatment, the time course used for the simulation was 240 hr. Following the identification of the contribution of each parameter to the inhibition effect of R2 mRNA, strategies to optimize the atisense effect of GTI-2040, such as the use of gapmer technology and/or application of liposomal delivery system, were proposed.

Figure 5.12 shows the influence of dose on the plasma concentrations of R2- antisense drugs and the perturbation of R2 mRNA levels. As shown in Figure 5.12A, doubling the dose led to a 2-fold increase in drug plasma concentrations as expected.

However, a dose increase did not result in significant increases in both onset rate and the extent of R2 mRNA down-regulation, except that there was a slightly slower recovery rate of R2 mRNA at the higher dose (Figure 5.12B). Therefore, based on the current dose regimen of GTI-2040 in the treatment of AML patients, dose escalation may not translate into a higher antisense effect and response.

The effect of Vm values on drug plasma concentrations and on the down- regulation of R2 mRNA was evaluated. As shown in Figure 5.13A, drug plasma concentrations showed a slight variation over a 100 fold change in Vm values. In 175 contrast, R2 mRNA was much more sensitive to the changes in Vm values. R2 mRNA was inhibited almost 2 fold with a 10 fold increase in the Vm value (Figure 5.13B).

Therefore, the cellular uptake rate of the antisense drugs plays a critical role in the inhibition of the target R2 mRNA.

The role of Km value on the inhibition of R2 mRNA was also examined by simulations. The results are shown in Figure 5.14A. The extent of R2 mRNA down- regulation was not sensitive to the Km values in the range from 0.2 to 200 nM. Only a

10% decrease in down-regulation was observed with Km at 200 nM (a 1000 fold increase). However, a substantial difference was observed for the recovered rate of R2 mRNA with a change in Km. R2 mRNA returned to the baseline much faster with a higher Km. In this scenario, since the Vm was held constant, a lower Km value would produce a lower threshold for the drug uptake to achieve the maximum rate (Vm). When the substrate concentrations in blood were all higher than the tested value of Km, no difference in the extent of down-regulation was observed, since the drugs were already taken up at the same maximum rate (Vm fixed). When the plasma drug concentrations decayed in the elimination phase, drugs with lower Km value would maintain their maximum cellular uptake longer and sustain their effects on the target inhibition, as shown in the prolonged recovery phase of target mRNA (Figure 5.14A). Therefore, strategies to lower the Km value of GTI-2040 thereby to enhance its cellular uptake efficiency, such as the stucture modification and using the liposomes or target transporters, would improve the down-regulation effect on R2 mRNA. We have found that GTI-2040 plasma concentrations were not affected by a change in Km values (data not shown). 176 Elimination rate of the intracellular drug from the effect compartment (K30) exhibited a profound influence on the mRNA down-regulation. As shown in Figure

-1 5.14B, when the half-life of the drug is very short (K30=1 hr , t1/2 ~40 min), the model predicted a low inhibition of mRNA expression; If the drug half-life increase to hours

-1 (K30=0.1 hr , t1/2 ~6.9 hrs), the extent of inhibition could increase significantly. A drug

-1 with an exceedingly long half-life (K30=0.01 hr , t1/2 ~69 hrs) could result in a profound and sustained inhibition of target mRNA. The intracellular half-lives of PS-ODNs have been estimated to be in the order of several hours, depending on the delivery method and cell type (128,157). Therefore, stabilization of ODNs in cells in order to prolong their half-lives will be able to improve the antisense effect. In the effect compartment, , K30 reflects processes leading to the loss of intracellular drug including efflux, intracellular degradation and endosome entrapment following drug internalization. Strategies that prevent the drug decay in cells could be considered for optimizing the use of antisense drug. Structure modification such as alkylation on the sugar to 2’-MOE (2’-O-methoxyethyl) ODN has been reported to enhance the exonuclease resistance significantly and therefore could be proposed as a potential method (48,57,162,163). For examples, tissue half-life was observed for multiple days for ISIS 104838, a 2’-MOE ODN (57).

Additionally, the MOE PS-ODNs may bind to the target mRNA with higher affinity (27), which may result in an improved antisense effect. Therefore, 2’-MOE modification on

GTI-2040 is suggested to be used in order to improve the down-regulation of R2 mRNA.

In addition, liposomes or other delivery systems, that would allow antisense drugs to escape the endosomal degradation (45-47), may provide higher intracellular drug levels for R2 mRNA down regulation and may be used for GTI-2040. 177 Significant effect of EC50 on down-regulation of the target mRNA was observed as illustrated in Figure 5.15A. With other parameters fixed, a ten-fold decrease in the

EC50 value (in the range of 1-100 nM) resulted in about two-fold increase in the target inhibition. Binding affinity between ODNs and target mRNAs as well as the efficiency in recruitment of RNase H are assumed to affect the value of EC50 and influence the antisense activity. Chimeric ODN (gapmer) could be designed to enhance the binding affinity in the drug-mRNA duplex while retaining the ability to activate RNase H

(23,30). For example, in ISIS 104838, a 2’-O-MOE chimeras, five 2’-O-MOE nucleotides at both terminal block prevent nucleolytic degradation and the contiguous stretch of 10 PS-ODNs at the center is able to activate RNase H (23). Therefore, antisense activity of GTI-2040 could be improved if employing the gapmer technology to lower its EC50.

Finally, simulation of dynamics of mRNA inhibition was explored on the effect of fractions of S phase in cells. As shown in Figure 5.15B, percentages of the cells in the S phase exhibited a modest impact on the inhibition of targetR2 mRNA. With the % of S phase varied from 35% to 100%, inhibition of mRNA increased by 20%. Therefore, pretreatment with cell cycle sensitive agents to arrest the cells in the S phase would benefit the antisense effect of GTI-2040.

In summary, an integrated PK/PD model with an indirect response has been developed and successfully applied to evaluate the dynamics of R2 mRNA after treatment with GTI-2040. Model simulation was performed to evaluate the influence of several parameters, including dose, Vm, Km, K30, EC50 and % of S phase in cells, on the target mRNA down regulation. Our results showed that dose escalation would not 178 improve the down-regulation of the target mRNA, instead, issues on the cellular uptake of the drug, intracellular drug stability and cleavage efficiency of RNase H are the major barriers to optimize the antisense activity. Applications of drug delivery systems such as liposomes or nanoparticles, which may enhance the cellular uptake of ODNs, are among the promising strategies. New antisense technology (chimeric antisense) and the new generation of ODNs (2-MOE), which possess substantial stability and enhanced binding affinity, are good candidates to improve the antisense effect.

179

Number (%) of male 60 Median (range) age 59.5 (20 ~ 77)

Median (range) body weight (kg) 72 (52 ~ 117) Median (range) BSA (m2) 1.85 (1.46 ~ 2.45)

Median (range) CrCL (mL/min) 86.5 (59.0 ~ 200.3) Median (range) WBC (× 109/L) 3.3 (0.4 ~ 137)

Median (range) Platelets (× 109/L) 51.5 (3.3 ~ 250.0) Median (range) Hemoglobin (g/dL) 9.0 (7.0 ~ 13.1)

Median (range) BM cell% 72.5 (20 ~ 90) Median (range) BM blast % 53 (6 ~ 91)

Median (range) PB blast % 34.5 (0 ~ 99.0)

Table 5.1 Demographic and physiological parameters of 31 AML patients in the Phase I trial of GTI-2040.

180

CLp(L/hr) V1(L) V2(L) Q (L/hr) value 7.0 6.61 4.43 0.113 Variance (%) 34 51 86 53 Residue variance (%) 50

B CL(L/hr) aV1F(L) bV1M(L) cV2F(L) dV2M(L) Q(L/hr) Value 3.54 4.07 9.36 2.34 10.2 0.115 Variance% 29 31 27 63 Residue variance 48 %

C CELL CL = θ (1 ) × ( ) p 70 V = θ ( 2 ) × ( SEX − 1 ) + θ ( 3 ) × ( 2 − SEX ) 1 V = θ ( 4 ) × ( SEX − 1 ) + θ ( 5 ) × ( 2 − SEX ) 2 M a le : S E X = 1 F e m a le : S E X = 2

Table 5.2 Results of plasma pharmacokinetic parameters, inter- and intra- individual variance of GTI-2040 in AML patients (n=31).

A shows the parameters and variances from the base plasma population PK model . B shows the parameters and variances from the final population PK model (B). C shows the covariate models in the final population PK model. After incorporated covariates, the objective function values of the final model decreased 43.5 compared to the base model; in addition, variance in CLp, V1 and V2 decreased 6%, 20% and 59% respectively.

181

Final model Bootstrap Boostrap 90% Parameter estimate mean CI for mean CLp (L/hr) 3.52 3.54 3.48-3.56 Q (L/hr) 0.115 0.117 0.113-0.121 V1 (female) 4.07 4.09 3.99 – 4.20 (L) V1 (male) (L) 9.36 9.13 8.83 –9.43 V2 (female) 2.34 2.41 2.30 – 2.52 (L) V2 (male) (L) 10.2 9.81 9.43 – 10.4 Variance 0.29 0.23 0.15 – 0.31 (CLp) Variance (Q) 0.63 0.68 0.61 – 0.73 Variance (V1) 0.31 0.26 0.21 – 0.31 Variance (V2) 0.27 0.33 0.25 – 0.41

Table 5.3 Model predicted and bootstrap results of pharmacokinetic parameters from a final plasma population PK model with covariance.

182 A Km(nM) V3 (L) CL3 (L/hr) Vm (nM/hr)

values 2.95 0.112 0.0133 0.71 Variance 44 10 35 63 % Residue variance % 107

B CL3 aVmf bVmm Km(nM) V3 (L) (L/hr) (nM/hr) (nM/hr) values 3.22 0.094 0.0232 1.12 0.476 Variance 16 31 28 51 % Residue 76 variance %

C 1 .6 K = θ (1) × ( ) m WBC

V m = θ ( 2 ) × ( SEX − 1) + θ (3 ) × ( 2 − SEX )

M ale: SE X =1 Fem ale: SE X =2

Table 5.4 Results of PBMC pharmacokinetic parameters, inter- and intra- individual variance of GTI-2040 in AML patients (n=31). A shows the parameters and variances from the base PBMC population PK model, B shows the parameters and variances from the final population PK model, C shows the covariate models in the final population PK model. After incorporated covariates, the objective function values of the final model decreased 110.1compared to the base model; in addition, variance in Km, CL3 and Vm decreased 28%, 7% and 12% respectively. Intra-individual variance decreased 31%.

183

A B Tim e (hr) C onc .(nM ) M odel E st . Time (hr) Obs.R2 mRNA% Model Est.(% 00 0 100 100 218.5267.4 2 97.21 4 229.7 353.4 4 87.85 6 259.2 381 6 75.36 12 263.7 394 12 45.49 24 232.5 395 24 30 29.19 48 263.3 396.1 48 24.91 72 325.8 396.8 72 24.29 120 397.8 120 25.2 24.14 144 257.4 398.2 144 24.14 144.2 559.8 346 144.2 24.14 144.5 458.2 300.8 144.5 24.14 145 272.3 227.4 145 24.15 146 113.4 130.7 146 24.25 148 33.2 44.73 148 24.89 150 16.7 17.11 150 26.16 156 4.43 4.289 156 32.35 168 3.38 3.365 168 48.27 192 76.24 192 2.58 2.552 240 94.02 240 1.468 720 100 99.98 720 5.81E-03

C Parameters values DOSE (nM/hr) 2670 K10 (1/hr) 0.5613 K12 (1/hr) 0.007495 K21 (1/hr) 0.01168 V1 (L) 11.91 VM (nM/hr) 0.8075 KM (nM) 15.19 K30 (1/hr) 0.25 V3 (L) 11.76 K04 (% mRNA/hr) 5.41 K40 (1/hr) 0.0541 Emax (% mRNA/hr) 6.43 EC50 (nM) 2.2

Table 5.5 Observed and the PK/PD model predicted GTI-2040 plasma concentrations (A) and R2 mRNA down-regulations (B) with time in patient 0304-26.

Estimated PK/PD parameters of this patient are listed in C.

184

Parameters Values (mean ± SD)

Vm (nM/hr) 1.13 ± 1.03 Km (nM) 10.2 ± 5.14 K30 (1/hr) 0.12 ± 0.08

V3 (L) 4.62 ± 4.04

K04 (% mRNA/hr) 43.2 ± 25.6

K40 (1/hr) 0.432 ± 0.22

Emax (% mRNA/hr) 5.5 ± 3.9

EC50 (nM) 15.0 ± 8.8

Table 5.6 Mean values of the PK/PD parameters (Mean ± SD) from 5 patients estimated from an indirect response PK/PD model linked with an effect compartment.

185

Q R0 Plasma V1 Peripheral V2 Q

CLp

PBMC V3

CL3 Vm/Km

Q R0 Plasma V1 Peripheral V2 Q

CLp

Figure 5.1 Models used in GTI-2040 population PK studies in AML patients. 5.1A: A two-compartment plasma population PK Model of GTI-2040 5.1B: An intracellular linked Plasma PK Model for GTI-2040 population PK analysis in PBMC.

186 IPred vs DV

IPRE

0 100 200 300 400 500

0 200 400 600 800 1000 WRES vs TIMEDV (run 80)

1 6 20

4 30 11 26 23 9 620 18 10 22 255 26 6 25 1 B 5 22 12 255 17 147 2 1426 251715154 15 1145 23 5 22 19 7 6 22 4 15 5 2111097 18 2727 69 10 13 26 18171811533 1 1621 15 24 1097 3 4 22 22 121918142330 15 15 22 92517 2 WRES 6 25 4 15 1211324212 2226 225 191310179 28 12 2417 126 6 132712195416161520 13 22 27 5 251830881324 2 8 201920 17 12319 20 14 13 30 25 2712 17 17 21 11 24 16716 127 30 1310168314216 11 203 9 2192 3011 2811 811 21 0 2411 191021 19 30174 5 17101630436 1320 181211239 2417 9 242116164625241210 16 13172112 16203 2 121910144 231321 2416 2415213 132419 8 229 2226106784321177 20 3 1318 242173 11 26232261214 264 12211513 16 3019 108 14 231428 108 1421 2 142621781523182023 195 1214 22 4181488 1323 111932 3 12301720 1 14 4 9 2619 1 162014 18218 222443711 24237 1316 30212491918226 6 237 18 27 189 1685 27 19 5 11 1 131620 242227 25 23 17 27 23 251811 7 20 15 107 4 15274 18 10 2 5 1722 30 14 26 15 9 25 6 186 101 179 -2 12 1 25 5 20

-4 20

0 50 100 150 200 TIME

WRES vs PRED (run 80)

1 6 20

30 4 11 23 26 18 20 9 6 22 10

25 2526 5 6 1 22 5 12 17 25 5 14 7 2 265 14 14 1 25 151515 4 17 22 5 2219 6 23 4 7 5 27 15 10 9 7 1 18 11596 27 18 26 17 10 2113 18 103973 3 1 16 15 24 15 22 21 1730 23 14 12 2218 19 4 225 9 15 22

WRES 132 122248 9 13101721 25 22 15 6 4 5 19 1 121516 27 12 2624 22 66 17 1320168138 45 8 30 19 12518 13 14242 1923 27 25 22 1 20 20 17 19 24201121 27 25 17 12 3 1 30 1121614 16 3 8 13 10 71612 167 17 21118830 2 9 1321 9 20 3 30 0 2013112 6 244 30 1010 511 17 21 1919 C 1012 25 24 16 616 2424 164 499 123 23 21 1811 17 30 16 17 241516213 13 232 1 2324 1022242 12 161317 1114 20 3 1919 207 11 7 4 10 26 223 7 8 6 22 9 813 15191626301213214 26 1412 26 23 24 2 7 3 11 13 183 21 22121423192018235 15 26 21 107 10822 8 1423 14141488 21 9 20 2 8811 13 183 1441 17 4 12 23 141430 19 318 19 237 1122 1819 247 22 26 24 24 4 716 8 20 3 121 6 65 9 1623 13 18 21 30 30 2723195 27 27 916 16 8 18 1725711 22 24 23 11 2013 1 1018 25 27 15 20 10274187 4 15 5 221730 5 2 9 2615 15 14 181769 25 10 1 -2 25 12 1 5 20

-4 20

0 100 200 300 400 500 600 PRED

Figure 5.2 Goodness-of-fit diagnostic plots of base plasma population pharmacokinetics. A: dependent variable (observed concentrations) vs individual predictions B. Weighted residues of concentration vs time. C. Weighted residues of concentration vs predicted concentrations.

187

GAM results for CLp

4 2

19 3 2 24 23 16 14 18 11 21 22 13 4 0 7 27 5 30 15 25 20 17 6 12 26

Residuals -2 9 10 1

-40 -20 0 20

CELL

GAM results for V1

2627 6 12 2 25 15 24 16 10 5 7

SEX

23 3 17 22 19 1 21 13 14 1 184 30 20 11 9 82

-6 -4 -2 0 2 4 Residuals

Figure 5.3 Plots of residuals of plasma PK parameter versus the correlated covariates of GTI-2040 in AML patients. A. Correlation of blast cell% with residuals of plasma clearance (CLp). B. Correlation of sex with residuals of V1 in plasma PK.1=male, 2=female.

188

IPred vs DV

IPRE

0 100 200 300 400 500

0 200 400 600 800 1000 DV

WRES vs TIME (run 82)

6 18 26 20 1 30 4 23 6 12 10 15 17 22 10 9 26 6 720 14 17 14 6 19 264 1 14 23 22 18 25 12 2 25575 11 21 75 4 18 6 25 12 1 1718223 5 6 7 25 22 24 14715 22101541516 13 15225 181926 5 24 102221 221216 917 4 26 191227 22 15 5 38 B 2224 17 1812 92 19 15 27 5 16 17 161923243082130 20 203 25 6 6 13171 1621 1526 75 15 1610178124913 20 15 3 2 61631 3019 1219 5 172713231312 220 20 25 WRES 20 17 2667 211715 1617187 3025 4 1310121 162411 5 13117 0 1124 23 21121724 12 10241348112010201911 132411 8 13219 1093 30 14192254 2123309943219231 82 11242 202716 2 278 211415 4 13 2184 21 3121311189 219 10 222613962511142511 1521249 16 4 13 73 129 26233072721 23 14 12 81814 142382 248 1421 112 1482637 3 14 19148 8 101323 1932 3 232442711 16 9 11 2118 48 2718 17 149 10 1 142016 8 21123024211735 23241897 1613 30229 26 18 271 13 162581819 25 12 11 2016 23 152415 22 19 4 27 20 2222 10 5 67 15 46 23 2 251820 23 19 26 1 2026 1 14 1 25 18 30 -2 12 10 7 17 30 5 22 7 18 6 17 20 10 -4 0 50 100 150 200 TIME

WRES vs PRED (run 82)

10 186 26 1 20 4 30 23 6 12 1510 2217 10 26 9 6 20 7 14 17 19 14 6 14 26 4 1 22 23 25 7 12 18 2 11 25 4 18 5 5 6 25 7 21 5 12 1 57223 18 17 6 157 14 24 2225 24526 22 1822 19 4 10 15151516 513 16 22 12 22 10 21 27 1226 4 19 17 9 8 3 5 22 15 2 912 24 22 18 17 16530 1 27 15 19 2482 1630 23 19 20 17 1621 16 25 17 66 133 20 1 152420122313 9178 116 10 26 157 15 5 202 1323 13 27 25 166 17 3 12 195 1 30 WRES 21 7 6 17 20 1316715517 26 10 21 25 74171816 1 30 2424111013191111 20 11 8 24 413 131123 17 1212 21 0 20828 191 2 3 9 30 24244 9 10103 2114 19 9 30 23 144 22 27 2 22 4 13 32111195 C 21151348411 1511 24 6 27 2 10 16 89 20 11 9 9 24219 22 18 161225142527 71326 23 71 3 312131213 9 21 14233 7 6 26 2 23 243 8 4 2 2 308 21 14 11 24 423 28 1014 141423 818 3 19 11 161418279 82 4 14238198 18 3 1311 17279 5 17 24 21 10 14 3 20 21 1 22 24 7 9 1623 8 12 13 9 21 30 30 26 16 18 30 2512 182519 27 16 8 18 13 1 22194 15 24 151623 20 11 10 22 22 27 20 76 5 5 236 4 15 15 19232018 25 2 261 20 26 1 1830 25 14 1 -2 17307 10 12 22 5 187 176 20

10 -4

0 200 400 600 PRED

Figure 5.4 Goodness-of-fit diagnostic plots of plasma GTI-2040 population pharmacokinetics model incorporated with covariates. A. Dependent variable (observed concentrations) vs individual predictions. B. Weighted residues of concentrations vs time. C. Weighted residues of concentrations vs predicted concentrations.

189

Patient 0304-15

1000

100

(nM) plasma GTI-2040 in concentration

0.1 0 50 100 150 200 250 time (hr)

Patient 0304-20

1000

100

GTI-2040 concentration in plasma (nM) 1 0 50 100 150 200 250 time (hr)

Figure 5.5 Individual fittings of GTI-2040 plasma concentration-time profiles in two patients using the Bayes estimated individual PK parameters from the plasma final popPK model. (Black dots: observed concentrations, solid line: predicted concentrations vs time)

190

IPred vs DV A

IPRE

0 20406080100

0 20406080100 DV

WRES vs TIME (run 35)

4 B 15 4 5 15 15 3 6 9 10 15 5

2 5 76 12 6 10 4 103 9

4 15 1 9 125 WRES 5 1 1 1 4 12 75 7 13 3 4 511 2 6 4 1 2 12 7 5 1 13 3 11 5 9 7 7 7 11 2 115 97 7 10 121 11 89 10 4 15 13 713 14 8 0 7 11 13 8 4 14151211139112 23 13131 8 9 3 14 14 32 8 14 11111011 1410 1411815 19 2138 32 3 141528 111411132 2 10 8214 3 9 148148 14 14 4 2 9 8 128108 11 6 4 8 9 5 4 1011 10 12 14 148 8 92 102 4 14 13 3 51213 12 10 1341211 12 10 12 6 4 79 7 2 1 13 136 1515 11 4 1274 103123911 106 6 12 7 13 6 2 -1 7 12 6 1 6 1591 4 4 1 6 3 7 6 5 10 15 5 1 15 10 15 5 15 15 -2 0 50 100 150 200 TIME

WRES vs PRED (run 35)

4 15 C 4 5

15 15 3 6 9 10 15 5

2 5 7 6 6 12 10 4 10 3

15 4 1 9 5 12 WRES 5 1 1 4 1 5 7 7 12 13 5 3 2 4 11 6 4 1 12 2 5 7 113 7 3 5 9 11 7 7 2 11 7 5 7 9 11 9 8 12 1 10 11 10 4 15 7 8 13 13 14 0 7 4 8 13 11 141513111291 12 2 3 1 13 8 9 14 3 8 2 3 14 14 15 11 9 10 111 11 10 8 1411 2 1310 815 2 2 132 3 2 3 8 11141114 2 8 14 14 9 3 88 1414 4 12 14 11 10 9 2 8 8 6 12 45 10 489 11 14 13 10 4 10 8 14 14 22 8 3 5 12 121312 10 9 13 11 10 12 127 6 4 9 7 2 115 6 11 4 9 13 133 4 10 7 12 12 4 10 3 11 6 6 12 7 13 6 2 -1 7 6 12 1 6 15 1 9 4 6 1 4 7 6 3 15 5 10 5 15 1 5 15 10 15 15 -2 0 20406080 PRED

Figure 5.6 Goodness-of-fit diagnostic plots of base PBMC population pharmacokinetic model of GTI-2040. A. Dependent variable (observed concentrations) vs individual predictions. B. Weighted concentration residues vs time. C. Weighted concentration residues vs predicted concentrations.

191

GAM results for KM (run 35)

15 1.0 5

11 0.5 10

0.0 14 1 128 3 7 13 9

Residuals 2 -0.5 6

-1.0

4 -1012345 WBC

GAM results for VM (run 35)

12 7 2 5 10 15 6

SEX

2 8 3 4 1 11 9 1 14 13

0.0 0.5 1.0 Residuals

Figure 5.7 Plots of residuals of PBMC PK parameter versus the correlated covariates of GTI—2040 in AML patients. A. Correlation between WBC counts (× 109) and the residuals of Km. B. Correlation between gender and the residues of Vm. 1=male, 2=female.

192 IPred vs DV

IPRE

0 20406080100

0 20406080100 DV

WRES vs TIME (run 38)

4 9 B 4 3 3 15 5 4 15 15 9 4 1 2 9 1 15 6 121013 13 1 4 111 5 4 76 9 35 4 1015 1 211 9 1 12 12 10 52 5 12 5 1 87 WRES 11 594 7 11 7 11 11 6 3 7 5 10 56 131 811 8 2 11 133 14 75 12 1313 10 13 3 7 8 0 793 103 3 7211 72 14 14 14 1421111813 10 91 10 2 3 141514415 11 15117 109 14814 42 14 12 1111 9 148854 2 8 14104 92 1284108 11 12 131282 2 68 104212 13 1484 14 126 109 1513157 14 1513111412916 8 8 8 6 7 109 14 2 12 7 12 11 6 14656 13 39 10101312126 7 1512 3 7 6 7 13126 13 10 -1 21 4 2 5 15 10 5 4 1 15 3 9 15 31 1 15 4 4 15 -2 5 1 0 50 100 150 200 TIME

WRES vs PRED (run 38)

4 9 C 4 3 3 15 5 15 4

15 9 4 1 2 15 1 9 6 13 12 13 10 1 4 1 11 5 9 4 7 5 4 6 9 11 3 151 10 2 12 1 12 5 2 5 1210 8 51

WRES 7 7 5 497 11 11 6 1111 10 3 5 7 5 6 8 11 8 2 1 13 313 11 14 10 5 12 7 13 13 8 3 7 0 9 7 9 7 3 7 33 11 2 10 1 2 81013 11 11 2 1414 11 15 4 3 2 14 10 15 7 1211 15 144 9 82 1414 10 11 5 11 49 2 8 1414 8 4 9 12414 11 2 810 10 13 8 122 8 8 6 22 10 12 13 8 141515 12 49 7 10 111312141591 4 14 8 12 141386 14 26 12 11 9 8 7 6 10 6 14 6 12 6 5 1210 13 1215 9 7 3 133 10 67 12 6 7 12 6 13 13 10 -1 21 4 2 4 5 5 15 10 15 1 3 15 9 1 1 3 15 4 4 15 -2 5 1 0 1020304050 PRED

Figure 5.8 Goodness-of-fit diagnostic plots of population pharmacokinetic model of GTI-2040 in PBMC in AML patients. A. Dependent variable (observed concentrations) vs. individual predictions. B. Weighted residues of concentrations vs time. C. Weighted residues of concentrations vs predicted concentrations

193

Patient 0304-09

100

GTI-2040 concentration in PBMC(nM)

1 0 50 100 150 200 250 time (hr)

Patient 0304-27

1000

100

(nM) PBMC in concentration GTI-2040 1 0 50 100 150 200 250 time (hr)

194

K10 K12 R0 C1, V1 C2, V2

Peripheral CMT Iv infusion Central CMT K21

Vm,Km K04 K40 mRNA

C3,V3 PD CMT

Effect CMT K40(1+EMAX*C3/(EC50+C3))

K30

Figure 5.10 An indirect response PK-PD linked model used for the fitting of R2 mRNA down-regulation by GTI-2040. In this model, plasma GTI-2040 is linked to the change of R2 mRNA by an effect compartment. GTI-2040 in the effect compartment induced the decay of mRNA by increasing the degradation rate constant of R2 mRNA (K40).

195

1000 1000

100 100

observed mRNA level 10 observed plasma conc 10 predicted mRNA level predicted plasma conc

% of R2 mRNA level to pretreatment to level mRNA R2 of % (nM) plasma in concentration GTI-2040 1 1 0 200 400 600 800 time (hr)

10000 10000

1000 1000

100 100

10 observed GTI--2040 plasma conc. 10 observed R2 mRNA level predicted GTI-2040 plasma conc. predicted R2 mRNA level 1 1 % of R2 mRNA level to pretreatment GTI-2040 plasma concentration (nM)

0.1 0.1 0 200 400 600 800 time (hr)

Figure 5.11 PK/PD model fitting results of GTI-2040 plasma concentration-time profiles and the time course of R2 mRNA levels using a two-stage method. A shows the fitting results in patients #26, B shows the fitting results in patient #13. Left Y axis represents the plasma concentration of GTI-2040, right Y axis represents the % of R2 mRNAlevels compared to the pretreatment (100%).

196 A

10000 mg/kg/day 15 1000 A 7.0 3.5 100

conc.(nM) plasma GTI-2040 1

0.1 0 50 100 150 200 250 300 time (hr)

B 110

100 mg/kg/day 3.5 90 7 80

15 70 R2of mRNA% 60

40 0 50 100 150 200 250 300 time (hr)

Figure 5.12 Effects of dose on GTI-2040 plasma concentrations and the change of R2 mRNA levels. A. Effect of dose on GTI-2040 plasma concentrations. B. Effect of dose on down-regulation of R2 mRNA.

197

A 1000

100

Vm=0.1 1

plasma conc.(nM)GTI-2040 Vm=1 Vm=10

0.1 0 50 100 150 200 250 300

time (hr)

B 120

100 Vm=0.1

80 Vm=1

60 Vm=10

of R2 mRNA % 40

0 0 50 100 150 200 250 300 time (hr)

Figure 5.13 Effects of change of PK/PD model parameters on GTI-2040 concentration and R2 mRNA levels. A. Effect of Vm values on GTI-2040 plasma concentration B. Effect of Vm values on down-regulation of R2 mRNA

198

110 A

100 Km=200

90 Km=20 80

% of R2 mRNA of % Km=2 60

50 Km=0.2

40 0 50 100 150 200 250 300

time (hr)

120 B K30=1 100

K30=0.1 60

% of R2 mRNA 40 K30=0.01 20

0 0 50 100 150 200 250 300 time (hr)

Figure 5.14 Effects of PK/PD model parameters on the change of R2 mRNA levels. A. Effect of Km values on the down-regulation of R2 mRNA B. Effect of K30 values on the down-regulation of R2 mRNA

199

A 120

100 EC50=100

60 EC50=10

R2 mRNA of % 40

20 EC50=1

0 0 50 100 150 200 250 300 time (hr)

B 120

S phase 100

35% 60% 80 100%

mRNA R2 of %

20 0 50 100 150 200 250 300

time (hr)

Figure 5.15 Effects of model parameters on the change of R2 mRNA levels. A. Effect of EC50 values on the down-regulation of R2 mRNA B. Effect of % of S phase on the down-regulation of R2 mRNA

200 CHAPTER 6

IN VIVO AND IN VITRO METABOLISM OF GTI-2040, A PHOSPHOROTHIOATE OLIGONUCLEOTIDE ANTISENSE TARGETING RIBONUCLEOTIDE REDUCTASE, USING ION-PAIR REVERSED PHASE HPLC COUPLED WITH ELECTROSPRAY ION-TRAP MASS SPECTROMETRY

6.1 INTRODUCTION

Phosphorothioate oligonucleotides (PS-ODNs) have been investigated extensively in recent years as antisense drugs for the treatment of a variety of diseases including viral infections, inflammatory disorders and cancers (4). With the substitution of one of the non-bridge oxygen atom by sulfur, PS-ODNs exhibit improved resistance to nuclease, enhanced antisense activity and better pharmacokinetic properties, which may contribute to their therapeutic potential (4,13). Despite their stability, their in vitro and in vivo metabolism of PS-ODNs was still observed. Most of the metabolic studies on oligonucleotides have been performed using radiolabeled oligonucleotides, capillary gel electrophoresis (CGE) and high performance liquid chromatography (HPLC)

(49,55,164,165). Use of the radiolabeled technique is sensitive but nonspecific and is only

201 useful for mass balance assessment. CGE separates oligonucleotides based on the chain length and is sensitive to nanomolar levels. However, this method does not provide the structural information. Recently, a novel ion-pair reversed phase HPLC coupled with electrospray ion-trap mass spectrometry (ESI MS) has been published, which demonstrated its utility in identification of metabolites of antisense oligonucleotides (AS-

ODNs) (72,79,166). This method not only permits separation of AS-ODNs at single- nucleotide resolution but also largely suppresses cation adducts formation, thus providing a more accurate mass measurement (167,168). The molecular weights of the metabolites can be determined via the deconvolution technique and the sequence and identity of metabolites can be confirmed from the fragmentation pattern generated by the MSn mass spectra (72,79,80). Thus the in vitro and in vivo metabolism of a number of PS-ODNs such as G3139, ISIS 2302 have been successfully characterized by LC-MS/MS (77-

79,169).

Major metabolites of PS-ODNs isolated from plasma and tissues following their systemic administration in mice, rats and monkeys have been characterized as the 3’ chain-shortened products presumably formed by the action of 3’ exonuclease (170-173).

Minor 5’ chain shortened metabolites and metabolites possiblely formed by endonuclease were also observed (77,79,102,174). PS-ODNs metabolism studies have also been performed in other biological fluids such as tissue culture medium, serum, urine, human and animal plasma, solutions containing purified bacterial enzymes (64,81,82,173,175).

Although as a class of compound metabolism profiles among PS-ODNs retain similar, sequence- and length dependence in metabolism may exist and these differences may

202 affect their PK and PD responses. Thus extrapolation of the metabolism of PS-ODNs to predict the metabolism of a new PS-ODN may not be accurate.

GTI-2040 is a phosphorothioate oligonucleotide targeting the R2 subunit of ribonucleotide reductase, a crucial enzyme in DNA synthesis (95). GTI-2040 is currently under the clinical evaluations in phase I trials for the treatment of solid tumors and acute myeloid leukemia (AML) as a single agent or in combination therapy (10). Till now, no metabolism stuy of GTI-2040 has been reported. Herein, a highly specific ion-pair RP-

HPLC/ESI MS was employed to investigate the metabolism of GTI-2040 in a variety of biological matrices. These biological samples include plasma from treated patients, solutions containing pure enzyme, mouse liver/kidney homogenates, human liver microsomes and urine from treated rat.

6.2 Materials and METHODS

6.2.1 Drugs and Chemicals

GTI-2040, a 20-mer phosphorothioate oligonucleotide with sequence of 5'-GGC

TAA ATC GCT CCA CCA AG-3', was provided by the National Cancer Institute

(Bethesda, MD) and used without further purification. Putative 3’ end (3' N-1, 3' N-2, 3'

N-3, here-to-fore GTI-2040 is omitted) and 5’ end (5' N-1, 5’N-2, 5’N-3) metabolites of

GTI 2040, from which 1-3 nucleotides were deleted from the 3' or 5' end, respectively, were all purchased from Integrated DNA Technologies (Coralville, Iowa). The purity and identity of each oligomer were verified by HPLC-UV-Mass spectrometry (Model: LCQ,

Finnigan, San Jose, CA). Phosphodiesterase I (EC 3.1.4.1), a 3’ to 5’-exonuclease, from

203 snake (Crotalus Adamanteus) venom (SVP), was obtained from USB (Cleveland, OH).

Phosphodiesterase II (EC 3.1.16.1), a 5’ to 3’-exonuclease, from bovine spleen was purchased from Sigma-Aldrich Co. (St. Louis, MO). Pooled human liver microsomes from 18 individuals (10 male and 8 female donors) were obtained as a stock emulsion (20 mg of protein/ml in 250 mM Sucrose) from BD Biosciences (Woburn, MA). HPLC-grade methanol, triethylamine (TEA, 99.5%), triethylammonium bicarbonate (TEAB) and

1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, 99.8%) were purchased from Sigma-Aldrich

Co. (St. Louis, MO). HPLC-grade water was generated by an E-pure water purification system (Barnstead, Dubuque, IA).

6.2.2 Instrumentation

LC-MS/MS analysis was performed on a Finnigan LCQ ion trap mass spectrometer (ThermoFinnigan, San Jose, CA) coupled with a Shimadzu HPLC system

(Shimadzu, Columbia, MD) and SPD-M10A PDA detector (Shimadzu, Columbia, MD).

The Shimadzu HPLC system consisted of two LC-10AT vp pumps and a SIL-10AD vp autosampler.

6.2.3 HPLC and LC-MS Conditions

Previously reported HPLC and LC/MS conditions (78) were used in the separation and identification of GTI-2040 and its chain-shortened metabolites. Briefly, the separation was accomplished on a 2.5 μm Waters Xterra MS18 column (50 × 2.1 mm) coupled to a MSC18 10 × 2.1 mm guard column (Waters Corp., Milford, MA). The mobile phase was prepared as Mobile Phase A consisting of 100 mM HFIP buffered to 204 pH 8.3 by 8.4 mM TEA and Mobile Phase B consisting of 100 mM HFIP and 8.6 mM

TEA (pH 8.3) in methanol (50:50, v/v). Gradient elution was used for the oligomers separation at a flow rate of 0.2 ml/min. The elution was initiated with 30% Mobile Phase

B followed by a linear increase to 45% in 30 min, returning to 30% B in 2 min and maintaining as such for 8 min before the next run. Column temperature was set to 50 °C throughout the analysis using a column heater (Keystone, Woburn, MA). Autosampler temperature was kept at 4 °C throughout the sample run. The PDA was operated to give the spectra from 200 to 600 nm. The LC-MS was operated with an ESI source in the negative ion mode. The LC effluent was introduced into the ion source without split. The electrospray high voltage was set at 2.0-2.2 kV, the tube lens offset was set to 31 V, and the inlet capillary was heated to 190 °C. The LCQ ion trap mass spectrometer was operated with a background helium pressure of 1.75 × 10-3 torr, a typical electrospray needle voltage of 4.5 kV, a sheath gas flow of 80 (arbitrary unit), and a auxiliary nitrogen gas flow of 30 (arbitrary unit). Triple play mode (full scan, zoom scan and MS/MS)

(78,176,177) was used for the identification of major metabolites of GTI-2040. Full mass scan was in the range of 600 to 2000 Da, zoom scan was based on the most intense peak from the full scan mass spectrum, and data dependent tandem MS/MS was collected from the most abundant ion. Collision energy was set to 25-35% with isolation width of 3.0

Da. The mass spectrometer was tuned to its optimum sensitivity of charge states from

[M-3H]3- to [M-7H]7- by infusion of either GTI-2040 or 3’N-1 through 3’N-3. All of the operations were controlled by the Fininigan Xcaliber software on a Windows NT 4.0 system.

205 6.2.4 Sample Preparation--Solid Phase Extraction (SPE)

Samples containing GTI-2040 and metabolites were thawed and centrifuged at

1000 g for 5 min. One mL of the supernatant was then mixed with 2 mL of 0.1 M TEAB and allowed to stand at room temperature for 30 min for ion pair formation between the oligonucleotides and TEA. GTI-2040 and related metabolites were extracted and isolated on an Oasis HLB cartridge packed with 60 mg material (Waters Corp., Milford, MA).

The extraction cartridges were pre-conditioned with 1 mL acetonitrile followed by 1 mL of 0.1 M TEAB (pH 8.0). Plasma or urine samples mixed with 0.1 M TEAB, or 0.1M

TEAB reconstituted tissue homogenate or microsomal extract were loaded onto the conditioned solid phase columns. The proteins and salts were removed by sequential washing with 3 mL of 0.1 M TEAB, 3 mL of distilled water, and 3 mL of 10% acetonitrile in 0.1 M TEAB by gravity flow. Then the ODNs were eluted with 3 mL 50% acetonitrile and the eluant was evaporated to dryness under N2. The residue was reconstituted with 150 μL of Mobile Phase A and 50 μL aliquot was analyzed by LC-MS.

6.2.5 Identification of GTI-2040 Metabolites in Plasma from AML Patients

Patients with acute myeloid leukemia (AML) were treated with GTI-2040 at 5 mg/kg/day as a continuous i.v. infusion for a total of 144 hr. Blood was drawn during and post infusion. Plasma was separated by centrifugation at 1400 g for 20 min. GTI-2040 and metabolites were isolated from the plasma samples using SPE procedure as described above and then analyzed by LC-MS.

206 6.2.6 In vitro Enzymatic Reaction

To prepare 10 U/mL stock solutions of Phosphodiesterase I (SVP) and

Phosphodiesterase II, each of the lyophilized enzymes was reconstituted in Tris-salts buffer:glycerol (50:50, v/v) solution containing 110 mM Tris-HCl, pH 8.9, 110 mM

NaCl, 15 mM MgCl2, and stored at –20 °C before use. Five μM of GTI-2040 was spiked in Mobile Phase A (100 mM HFIP titrated to pH 8.3 by 8.4 mM TEA) containing 0.5

U/mL SVP or phosphodiesterase II. After incubation at 37 °C with gentle shaking for 10 hrs, the reaction was stopped by freezing and the mixture was stored at -80 °C until analysis. For analysis, the sample was thawed at 4 °C and an aliquot was analyzed by

LC-MS without further sample preparation.

6.2.7 In vitro Incubation of GTI-2040 with Fresh Human Blood

Fresh human blood from a donor drawn within 1 hr was spiked with 5 μM GTI-

2040 and the blood was incubated at 37 °C for 24 hrs with frequent inversed mixing. The whole blood was then centrifuged at 1400 g for 10 min and the plasma was separated.

GTI-2040 and its metabolites were isolated from plasma by solid phase extraction (SPE) as described before. The extract was then analyzed by LC-MS for metabolites identification.

6.2.8 In vitro Incubation of GTI-2040 with Mouse Tissue Homogenates

The liver and kidneys were excised from mice following CO2 anesthesia and were frozen in liquid N2 immediately after rinsing in PBS and weighing. Liver or kidney homogenate was prepared with addition of 1:10 (w/v) ice-cold nuclease buffer (100 mM 207 Tris-HCl, 1 mM Magnesium chloride, pH 8.0) and homogenized on ice with a tissue homogenizer (VIRTIS, Gardiner, NY). Five μM of GTI-2040 was incubated in 1 mL of liver or kidney homogenate at 37 °C for 10 hrs with gentle shaking and frequent inversed mixing. To the homogenates, 0.5 mL of 3 mg/mL proteinase K in buffer containing Tris-

HCl (pH 8.0), 10 mM EDTA were then added and the mixture was incubated for 8 hrs to digest tissue components. After centrifugation, the supernatant was collected from the digestion mixture and was extracted with phenol-chloroform-isoamylalcohol (25:24:1, pH 8.0, Invitrogen, Carlsbad, CA) to remove both proteins and lipids. Oligonucleotides remained in the aqueous phase, while proteins and lipids stayed in the organic phase. The aqueous fraction was evaporated to dryness. The residue was reconstituted with 3 mL of

0.1 M TEAB and proceeded with solid phase extraction as described previously.

6.2.9 In vitro Incubation of GTI-2040 with Pooled Human Liver Microsomes

Five μM of GTI-2040 was spiked in 1 mL nuclease buffer containing diluted human liver microsomes (0.5 mg of protein/mL). The reaction was carried out at 37 °C and incubated for 10 hrs in a water bath with gentle shaking. At the end of incubation,

0.5 mM EDTA was added to stop the reaction. The mixture was then centrifuged at 1400 g for 10 min and the supernatant was separated and stored at –80 °C until analysis. GTI-

2040 and its metabolites were isolated by SPE and analyzed by LC-MS.

6.2.10 Urinary Excretion of GTI-2040 and Metabolites in the Rat

Two female Sprague Dawley rats weighing approximately 340-380 g were purchased from Harlan (Indianapolis, IN). GTI-2040 in sterile saline was injected into the 208 tail vein at a single dose at 20 mg/kg. Each of the rats was housed in a metabolism cage and urine samples were collected at intervals of 0-7, 7-12 and 12-24 hours after dosing.

The total urine volume was measured and the urin was immediately frozen at –80 °C until analysis. Following solid phase extraction, identification of metabolites was accomplished by LC-MS. The animal study was performed under a protocol adhered to the “Principle of Laboratory Animal Care” by NIH and approved by The Ohio State

University and the Institute and Laboratory Animal Care and Use Committee.

6.3 DATA ANALYSIS Raw spectra were processed with Xcalibur (Version 1.2, Finnigan, CA).

Molecular weights of the oligonucleotides were obtained by automatic deconvolution using the BioMass program (Xcalibur 2.1). The Simple Oligonucleotide Sequencer software (SOS, Version1.10 Jef Rozenski, University of Utah, Salt Lake City, UT, 2002) was used to derive the oligonucleotide sequences from product ion mass spectra.

6.4 RESULTS 6.4.1 Identification of GTI-2040 Metabolites in Plasma obtained from Treated Patients

Previously, a highly specific LC-MS analytical method was developed in our laboratory (78) for antisense G3139 and metabolites with single nucleotide resolution and this method was applied to GTI-2040 and metabolites in patient plasma.

Representative LC-UV/MS chromatograms showed the separation of GTI-2040 and its 3’ metabolites (Figure 6.1). A total ion chromatogram (TIC) and the corresponding mass 209 spectra obtained from the extract of patients plasma collected at 72 hr in 6-day i.v. infusion are shown in Figure 6.2. Except for the peak of GTI-2040, five peaks, denoted as M1, M2, M3, M4, M5, were found from the TIC chromatogram (Figure 6.2A). GTI -

2040 was identified at the peak with the retention time of 15.8 min and its mass spectrum is shown in Figure 6.3A. As shown, an ion envelop with a distribution of multiple ion peaks possessing different charge states was found for GTI 2040, which is typical for ESI mass spectra of oligonucleotides (72,83,172). Major ions at m/z 1061.5, 1274.1, and

1593.1 were assigned with 6, 5, and 4 negative charges, respectively. The charged states could also be revealed by zoom scan, and the data dependent scan provided the MS/MS information for specific ions, which are helpful in sequence determination. Based on this information, the molecular weight of GTI-2040 was readily obtained by deconvolution and is shown in Figure 6.3B.

An ion envelop including major ions at m/z 1004.0 ([M-6H]6-), 1205.0 ([M-5H]5-) and 1506.6 ([M-4H]4-) was observed for peak M1 (Figure 6.2B). Peaks M2 and M3 also showed multiple charged ions of different intensities with the most abundant ions at m/z

1898.9 ([M-3H]3-) for M2 peak and m/z at 1789.5 ([M-3H]3-) for M3 peak (Figures 6.2B,

6.2C and 6.2D). Similar strategy for identification of GTI 2040 was applied to these unknown peaks. Table 6.1 summarized the calculated and observed molecular weights of

GTI-2040 and its metabolites. As shown, following deconvolution, the GTI-2040 peak yielded [MH]- at 6375.5, which matched the theoretical molecular weight of GTI-2040 with ± 0.004% mass accuracy. Similarly, molecular weights of M1, M2, M3, M4 and M5 were determined to be 6031.3, 5699.4, 5370.2, 5064.9 and 4758.3. With the exception of

M1, the observed molecular weights were found to be close to the theoretical values for 210 3’N-2, 3’N-3, 3’N-4 and 3’N-5 GTI-2040, rather than the corresponding 5’ shortenmers.

Less than 0.07% differences to the respective theoretical molecular weights of 3’ N-2,

3’N-3, 3’N-4 and 3’N-5 standards were found in all cases (Table 6.1). Additionally, according to the sequence of GTI-2040, the nucleotide sequences from N-2 to N-5 at the

3’ end are different from those at the 5’ end, the molecular weights and the total ion spectra of N-2 to N-5 chain-shortened metabolites from 3’ end of GTI 2040 were found different from those of the synthesized 5’ end shortenmer standards (data not shown).

Therefore, peaks of M2-M5 in Figure 6.1A were assigned as 2 to 5 nucleotide-deletion metabolites from the 3’ end of GTI-2040, respectively, according to their mass spectra patterns as compared to the 3’ chain-shortened standards.

While the observed molecular weight of M1 is similar to 3’N-1 GTI-2040, it cannot be differentiated from that of 5’N-1, since the molecular weights of both 3’N-1 and 5’N-1 are identical. To distinguish between these two possibilities, we constructed the oligonucleotide sequences of 3’N-1 and 5’-N-1 GTI-2040 based on the MS2 fragment ion spectra with the aid of the SOS software (Simple Oligonucleotide Sequencer)

(80,178). MS2 spectra of 3’N-1 and 5’N-1 were obtained using the most abundant ion

[M-6H]-6 at m/z 1004.0 and compared to the MS2 spectra of M1 on the same ion. The most diagnostic ions for the determination of the 3’→5’ sequence are the w series ions and for the determination of the 5’→3’ sequence are the a-B ions (179,180). Seven w ions and 5 a-B ions generated from MS2 of M1 peak were identified and assigned (Table

6.2, Figure 6.4A). Using the same method, the w and a-B ions ladders were constructed and the sequences were aligned for 3’N-1 and 5’N-1 standards (Table 6.3, Figure 6.4 B).

The MS2 mass spectrum of M1 was essentially identical to that of 3’N-1 but not 5’N-1, 211 and these fragment ions are consistent with the fragmentation assignment for 3’N-1. As shown in Figure 6.4, MS2 mass spectrum of 5’N-1 (Figure 6.4B) showed a distinct pattern of w ion and a-B ion series from those of 3’N-1 and M1 peaks (Figure 6.4A). For

-1 -1 instance, w2 at m/z 674.9 and w4 at m/z 1285.1 were assigned as sequence of 3’ AA

-1 -2 and 3’ AACC in 3’N-1 and M1, while the w2 (m/z 691.1) and w4 (m/z 691.1) ions correspond to sequence of 3’ GA and 3’ GAAC in 5’N-1 fragmentation, respectively.

-2 Similarly, the fragment ion of a5-B5 (m/z 871.0) corresponding to the sequence of 5’

-2 GGCTA was identified in M1 and 3’N-1, while the a5-B5 ion (m/z 862.6) was identified as 5’GCTAA in 5’N-1. Therefore, M1 ion was identified as 3’N-1 metabolite of GTI-

2040 in plasma from AML patients.

In summary, five metabolites were identified in AML patients’ plasma extract and were determined as progressively chain shortened oligomers from the 3’ end of GTI-

2040. Minor metabolites from 5’ exonuclease or endonuclease activity may potentially exist, but were not detected because of the sensitivity limitation.

6.4.2 Role of 3’-Exonuclease in GTI-2040 Metabolism

We then speculated that 3’ exonuclease was the major enzyme responsible for

GTI-2040 metabolism in vivo. The role of 3’ exonuclease in GTI-2040 metabolism was examined by in vitro enzymatic reaction, using snake venom phosphodiesterase (SVP).

As shown in Figure 6.5A, a typical right-triangle shaped pattern was observed in the

HPLC chromatogram after 10 hr incubation of 5 μM GTI-2040 in MPA (100 mM HFIP buffered to pH 8.3 by 8.4 mM TEA) containing 0.5 U/ml SVP at 37 °C. GTI-2040 and its 3’N-1, 3’N-2 and 3’N-3 were identified by LC-MS and MS/MS, using the same 212 strategies as before (Figure 6.5B). Under the same condition, GTI-2040 was stable in

MPA without the enzyme and no interference was found in the SVP control (data not shown). In contrast, GTI-2040 exhibited resistance to the phosphodiesterase II under the same reaction condition with almost no degradation products detected (Figure 6.5C).

Using a very high level of phosphodiesterase II (5U/ml), GTI-2040 only showed slight degradation after 24 hr incubation (Figure 6.5D). It is well-known that SVP is a 3’ specific enzyme, which hydrolyzes 5’-mononucleotides from 3’-hydroxy terminated

DNA, and phosphodiesterase II is a 5’ specific enzyme, which hydrolyzes 5’- mononucleotides from 5’-hydroxy terminated DNA. Based on the above results, GTI-

2040 is expected to be a good substrate for 3’ exonuclease but not 5’ exonuclease.

Consistent with the in vivo results, the in vitro enzymatic incubation indicated the sequential nucleotide deletion from the 3’ terminus in degradation of GTI-2040. Similar patterns of sequential nucleotide degradation from 3’ terminus were also observed using

3’N-1 or 3’N-2 of GTI-2040 as substrate in SVP (Figure 6.6).

6.4.3 In vitro Metabolism Profiles of GTI-2040 in Tissue-Related Samples

As shown in Figure 6.7, the HPLC results from the in vitro incubation of GTI-

2040 with fresh human blood displayed the same degradation pattern as that in SVP. The peaks eluted earlier than GTI-2040 were identified as the 3’ chain-shorten metabolites.

Thus, blood is a potential source for metabolism of GTI-2040. Not surprisingly, after incubation GTI-2040 with mouse liver homogenate, many chain-shortened oligomers of

GTI-2040 were observed. In liver homogenate, 3’N-1, 3’N-2 and 3’N-3 were identified by both HPLC-UV and LC-MS (Figures 6.8A, 6.8B). Similar to the mouse liver 213 homogenate, the 3’ end deleted metabolites were observed after incubation with human liver microsomes (Figures 6.8C and 6.8D); both 3’N-1 and 3’N-2 were detected and identified by both LC-UV and LC-MS systems. In addition, 3’N-1 and 3’N-2 were identified in mouse kidney homogenate (Figures 6.9A and 6.9B). In the LC-MS metabolism profiles of GTI-2040 with mouse kidney homogenate, an earlier eluting peak with a relatively high intensity was also detected and was assigned as 3’N-5 GTI-2040, according to its deconvoluted molecular weight and the ion fragmentation pattern from its MS2 spectrum. No 5’ related chain-shortened metabolites were observed in liver and kidney homogenates within the detection sensitivity limit.

6.4.4 Metabolites in Urine

A 24-hr urine sample was collected from AML patients after receiving 5 mg/kg/day GTI-2040 as a continuous i.v. infusion for 144 hr. After 250-fold enrichment by lyophilization and SPE extraction, no metabolites were detected in the extract by LC-

MS. In rat urine, extensive metabolites of GTI-2040 were observed in 0-7 hr urine collection after an iv bolus with GTI-2040 at 20 mg/kg, as shown in Figures 6.9C and

6.9D. The 3’N-1, 3’N-2, 3’N-3, 3’N-4 and 3’N-5 GTI-2040 were identified as the major metabolites in both HPLC-UV and LC-MS systems. Some small peaks were eluted before 10 min, and they may possibly be the progressively n-1 chain-shortened metabolites resulting from the action of exonuclease, but the possibility of being the shortenmers generated from endonuclease might exist. Concentrations of GTI-2040 and its metabolites in rat urine collected after 7 hr were too low to be detected by HPLC-UV and LC-MS. 214

6.5 DISCUSSION Ion-pair reversed phase HPLC is an efficient method for separation of oligonucleotides. In the LC-MS system, using TEA-HFIP as an ion-pairing buffer provides a single nucleotide resolution for the phosphorothioate oligonucleotides. TEA provides the counter ion TEA+ in the formation of ion pair with the negatively charged phosphorothioate oligonucleotides (181). The use of the ion-pair buffer helps to decrease the interference from sodium adducts and improves the performance of deconvolution and the mass accuracy. HFIP, being a highly volatile weak acid, enhances the ionization efficiency and increases the detection sensitivity (181). In our study, we found that 100 mM HFIP combined with 8.6 mM TEA (pH 8.3) shifted the ion envelope to the low charge states and offered good ion pair efficiency and higher MS sensitivity (78).

Therefore, using an optimized buffer system containing TEA and HFIP with methanol gradient maximizes both chromatographic separation and mass spectrometric sensitivity in the analysis of oligonucleotides.

A careful examination on the HPLC chromatograms of the GTI-2040 metabolism profiles reveals a pattern of uneven spacing of peaks, for example, 3’N-1 was closely eluted with parent GTI-2040 (Figure 6.8A), while 3’N-2 eluted much slower than 3’N-3, giving rise to a wider space between the two peaks. The similar separation pattern between nucleosides was also observed in other oligonucleotides (166). As reported by

G. Martin et al (166), the separation selectivity seems to be enhanced, when cleavage occurred between CA from the 3’ end, while the separation selectivity decreased, when cleavage occurred between AG. This is consistent with our observations, a close elution

215 of 3’N-1 and the parent GTI-2040 with an AG cleavage, and a longer retention time between peaks of 3’N-2 and 3’N-3 resulting from a CA cleavage. It has been suggested that oligonucleotide sequence (nucleobase composition) affects the separation of oligonucleotide ladders (166). The different retention behaviors of oligonucleotide ladders in ion-pair RP-HPLC may result from the relative difference in the hydrophobicity of nucleobases (182,183). A loss of the more hydrophobic A or T mononucleotides from the 3’ end results in a pronounced, shorter retention time than loss of rather hydrophilic C and G mononucleotides (166). Other reasons may exist. For example, the differences in the polarity of nucleosides may affect their chemical bonds to the column materials, and result in different behaviors in their elution. Such chromatographic characteristics of nucleotides may be useful serving as supplemental evidences in the confirmation of the sequence of oligonucleotides.

Similar to the right-triangle ladder in the metabolism of other PS-ODNs

(64,172,173), our in vivo and in vitro data also showed a progressive n-1 chain shortened pattern in the hydrolysis of GTI-2040 by the action of exonuclease. Metabolism by endonuclease would be expected to produce a profile characterized by an irregular distribution of shortened metabolites, since endonuclease cleaves nucleic acids at internal sites and produces nucleotide fragments of various sizes (76). Metabolism of PS-ODNs with endonuclease activity has not been observed (56,77,172). Furthermore, sequential metabolism appears to be the characteristics with PS-ODNs including GTI-2040. This is supported by our observed pattern of gradual hydrolysis of the terminal phosphorothioate linkage was also observed in the degradation of 3’N-1 or 3’N-2, when they were used as the substrates (Figure 6.6). A major issue in the metabolite identification of PS-ODNs is 216 to discriminate the 3’ terminal sequential degradation from the 5’ terminal degradation.

The strategies used involved the analysis of the total ion spectra and the fragment ion patterns from MS2. Progressive chain-shortened metabolites with different nucleobase compositions yield different molecular weights and different total ion spectra. The ion trap mass spectrometer used in our study provides with a high accuracy in mass measurements with ± 0.004% mass accuracy for GTI-2040. Nucleotide sequences of each

N-2 to N-5 GTI-2040 deletion from the 3’ end are different from the corresponding 2-5 shortenmers from the 5’ end. This gives rise to the distinct ion envelop patterns as well as the different convoluted molecular masses of 3’N-2, 3’N-3, 3’N-4 and 3’N-5 compared to the corresponding 5’N-2 to 5’N-5 shortenmers. Using this method, M2 to M5 unknown peaks in patient’s plasma extract were identified to be 3’ N-2 to 3’N-5 metabolites of GTI-2040 (Table 6.1). In the case when metabolites formed from the 3’ and 5’-terminals deletion possesses the same nucleotide composition, for example in

3’N-1 and 5’N-1 of GTI-2040, their ion envelop and molecular masses were identical and they can not be differentiated. This issue could be solved by examination of the oligonucleotide sequencing based on the MS2 fragment ion spectrum. Based on the ion intensities, charge states and isotope distribution in fragment ion spectrum, the sequence ladder could be constructed from both sides by the 3’ w the 5’ a-B ion extensions with the assistance of Simple Oligonucleotide Sequencer software. The fragmentation assignment of the unknown M1 product of GTI-2040 was then compared to the fragmentation patterns of standards of 3’N-1 and 5’N-1 and was found to be essentially identical to that of 3’N-1 standard rather to the 5’N-1 standard (Figure 6.4, Tables 6.2 and 6.3). The 3’ end degradation of GTI-2040 was also suggested that GTI-2040 was a 217 good substrate of 3’ exonuclease but not 5’ exonuclease (Figure 6.5). Using the above methods, major metabolites of GTI-2040 in patients were elucidated to be the hydrolytic products by the 3’ exonuclease in vivo.

Following the identification of metabolites from the in vivo plasma samples, it is of interest to evaluate the stability of GTI-2040 in blood, since the stability of PS-ODNs in circulation is important for its distribution to the target sites. The 3’ chain-shortened metabolites were detected after GTI-2040 was incubated in fresh blood (Figure 6.7), which suggested that GTI-2040 was susceptible to the 3’ exonuclease in circulation.

However, the rapid distribution of PS-ODNs (t 1/2 α ~ 0.5 hr) to tissues and the very slow efflux rate from tissues (days) following systemic (55,56,184) administration may counteract its metabolism in circulation, and the degradation of GTI-2040 in blood may not markedly affect its distribution to the target tissues and cells.

As reported in the animal studies, liver and kidneys are the two major organs that accumulate and metabolize the PS-ODN (53,55,62,64). For GTI-2040, sequential 3’ end chain shortened metabolites were detected from both mouse liver and kidney homogenates (Figures 6.8A, 6.8B and 6.9A, 6.9B). While the HPLC-UV chromatograph from the mouse liver homogenate showed a typical right-triangle pattern of the metabolites ladder (Figure 6.8A), in the mouse kidney homogenate it appeared that more shorter oligmers were generated. (Figure 6.9A). Same experimental conditions were used in these studies. We rationalize this difference in metabolic patterns as follows. 1) n-1 progressive chain shortened degradation occurred more extensively in mouse kidney homogenate than in the liver homogenate. 2) Mouse kidney homogenate contained high endonuclease activity and generated more shortenmers. In addition, tissue biodistribution 218 studies following the infusion of GTI-2040 in the rat and monkey demonstrated that the highest accumulation of GTI-2040 was in kidneys (101). It would be likely that more chain shortened metabolites would accumulate in the kidneys. It has been reported that acute renal failure resulting from renal cortical necrosis or acute tubular necrosis were observed in mice when administered with very high dose (≥ 100 mg/kg) of a 20-mer PS-

ODN (69). Therefore, caution of the renal side effect in patients should be considered when a high dose of GTI-2040 is given. Consistent to the metabolism profile in mouse kidney homogenate, abundant progressive 3’N-1 to 3’N-5 chain shortened metabolites were detected in rat urine from the first 7 hr collection (Figures 6.9C, 6.9D). This may indicate that significantly accumulated renal metabolites were filtered out from the kidneys as the result of the saturation of the protein binding of oligonucleotides at a very high i.v. injection dose. In human liver microsomes, the metabolism profiles were associated with 3’ end cleavage and displayed the similar pattern observed in mouse liver homogenate (Figure 6.8C, 6.8D) and in SVP as well. This may imply that human liver microsomes retain the exonuclease activity and could be used as a human-related enzyme system in testing the metabolism of PS-ODNs.

Due to the limited sensitivity of MS, many earlier eluted metabolites with lengths shorter than 3’N-3 in tissue homogenates could not get detected and identified from mass spectrometer. These peaks could be derived from the sequential hydrolysis by the exonuclease, but possibility of the shortermers generated from the endonuclease activity may exist, since endonucleases cleave oligonucleotides into small fragments directly. Our enzyme reaction studies showed that GTI-2040 was highly resistant to 5’ exonulcease

(phosphodiesterase II); therefore, 5’ degradation, if present, would be a minor 219 degradation pathway. In fact, within the detection sensitivity of our method, we did not observe the 5’ end chain shortened oligomers from both in vivo and in vitro studies.

Oxidative products were also not found as the metabolites of GTI-2040.

6.6 CONCLUSION

Using a highly specific ion-pair reversed HPLC/ESI-MS, metabolism of GTI-

2040 was found to be mediated mainly by the 3’ exonuclease and progressive 3’ end chain shortened metabolites were detected. 5’ end related degradation and oxidative products were not observed in metabolites of GTI-2040. Enzyme kinetic studies of GTI-

2040 in SVP and in human liver microsomes are currently in progress.

220

Table 6.1 Structures and molecular weights of GTI-2040 its metabolites in human plasma.

221

m/z Ion assignment Sequence (from 5’ to 3’) -1 346.0 w1 A -1 675.1 w 2 AA -1 981.5 w 3 CAA -1 1285.1 w 4 CCAA -2 802.6 w 5 ACCAA -2 959.1 w 6 CACCAA -2 1271.6 w 8 TCCACCAA -3 1165.8 w 11 CGCTCCACCAA -3 1272.4 w 12 TCGCTCCACCAA -4 1036.3 w 13 ATCGCTCCACCAA -5 894.7 w 14 AATCGCTCCACCAA -5 960.5 w 15 AAATCGCTCCACCAA -5 853.6 w 16 TAAATCGCTCCACCAA -6 1019.4 w 19 GGCTAAATCGCTCCACCAA -1 442.1 a1-B1 G -1 787.1 a2-B2 GG -1 1412.1 a4-B4 GGCT -2 1741.2 a5-B5 GGCTA -3 689.4 a6-B6 GGCTAA -3 798.0 a7-B7 GGCTAAA -3 1007.4 a9-B9 GGCTAAATC -3 1122.4 a10-B10 GGCTAAATCG -3 1224.1 a11-B11 GGCTAAATCGC -5 920.1 a14-B14 GGCTAAATCGCTCC -6 872.24 a16-B16 GGCTAAATCGCTCCAC

Table 6.2 Assignment of fragment ions obtained from the MS2 spectrum following the collision-induced dissociation of the ion with m/z 1004.0 ([M-6H]-6) from 3’N-1 standard and M1 peak.

222

m/z Ion assignment Sequence (from 5’ to 3’) -1 362.0 w1 G -1 691.1 w 2 AG -1 1325.1 w 4 CAAG -1 1630.1 w 5 CCAAG -2 981.1 w 6 ACCAAG -2 1284.1 w 8 CCACCAAG -2 1444.1 w 9 TCCACCAAG -3 1280.8 w 12 CGCTCCACCAAG -4 1122.6 w 14 ATCGCTCCACCAAG -5 1093.5 w 17 TAAATCGCTCCACCAAG -5 1154.5 w 18 CTAAATCGCTCCACCAAG -6 1019.4 w 19 GCTAAATCGCTCCACCAAG -1 442.1 a1-B1 G -1 747.1 a2-B2 GC -1 1396.1 a4-B4 GCTA -2 862.1 a5-B5 GCTAA -2 1026.6 a6-B6 GCTAAA -2 1511.6 a9-B9 GCTAAATCG -3 1215.8 a11-B11 GCTAAATCGCT -4 1146.4 a14-B14 GCTAAATCGCTCCA -4 1298.9 a16-B16 GCTAAATCGCTCCACC -4 1381.1 a17-B17 GCTAAATCGCTCCACCA -5 1170.5 a18-B18 GCTAAATCGCTCCACCAA

Table 6.3 Assignment of fragment ions obtained from the MS2 spectrum following the collision-induced dissociation on the ion with m/z 1004.0 ([M-6H]-6) from the 5’ N-1 standard.

223

600 3’N-2 3’N-1 600 A 3’N-3 GTI-2040 400 400

mAu mAu 200 200

0 0

5 10 15 20 25 30 Minutes

100 3’N-2 B 90

80 3’N-3 70 60 3’N-1 50 40 GTI-2040 30 20

10 0 6 8 10 12 14 16 18 20 22 24 Time (min)

Figure 6.1 IP-RP-HPLC-UV chromatogram (A) and LC-MS total ion (TIC) chromatogram (B) of standards of GTI-2040, 3’ N-1, 3’ N-2, and 3’N-3. UV was set at 260 nm.

224

100 M2 M1 90 80

GTI 70 M3 60 A 50 M4 40

Relative Abundance Relative 30 M5 20

10 0 6 8 10 12 14 16 18 20 22 24 Time (min)

1004.0 [M -5H ]5- 100 [M-6H]6- 90 1205.0 80

4- 70 [M -4H ] B 60 1032.0 50 1506.6 883.7 40 1058.6 1086.7 Relative Abundance 30 1225.0 882.4 1535.1 20 895.0 1237.8 1549.0 1894.2 10 803.8 1693.7 692.5 1497.7 0 600 800 1000 1200 1400 1600 1800 2000 m/z

1898.9 100 [M -3H]3- 90 80 C

70 1899.9 60 50 40 1931.8

Relative Abundance Relative 30 1937.4 1139.2 1424.1 20 949.4 10 1116.8 1171.4 1449.1 635.6 813.4 1343.9 1646.3 1854.6 0 600 800 1000 1200 1400 1600 1800 2000 m/z

1789.5 100 [M -3H]3- 90 80 D

70 60 50 40 1822.5

Relative Abundance Relative 30 1978.2 803.5 635.7 894.1 1687.0 1964.0 20 1073.0 693.1 921.6 1365.9 1665.9 10 941.4 1093.5 1267.9 1421.0 0 600 800 1000 1200 1400 1600 1800 2000 m/z

Figure 6.2 Total ion chromatogram and mass spectra of GTI-2040 and major metabolites obtained from the plasma extract of a patient treated with GTI-2040 (A) TIC of GTI-2040 and metabolites. The corresponding mass spectra of (B) 3’N-1 metabolite at retention time (RT) of 14.8 min and (C) 3’N-2 metabolite at RT 14.2 min and (D) 3’N-3 metabolite at RT 12.9 min.

225

[M-5H]5-

[M-6H]6- 1274.1 100

90 1061.5 A 80

70 [M-4H]4- 60 50 1593.1 40

Relative Abundance Relative 30 1089.3 20 1307.4 1506.4 909.5 1205.0 10 1334.2 1639.0 610.6 803.6 933.8 1780.4 1859.2 0 600 800 1000 1200 1400 1600 1800 2000 m/z

6375.5

100 B 90

7630.7 8720.8 9810.8 6540.6 30 Relative Abundance 4360.4 6031.4 5450.5 3270.3 7479.5 2180.2 7771.8 20 4825.1 2616.9 3893.0 10

2000 3000 4000 5000 6000 7000 8000 9000 10000

mass

Figure 6.3 The ESI LC-MS mass spectrum (A) and the corresponding deconvoluted mass spectrum (B) of GTI-2040.

226 5' 3' w w 17 w 16 w 2 1 G O G C C A HO O O T O O O A O P O O P O O P O O O P O O P O O P O - OH S - - S S - S - S S - n=13 a 2 -B 2 a 3 -B 3 a 17-B 17 a 18-B 18

981.3 100 -1 w 3 95 90 85 80 75 70 65 A 60 55 50 -2 817.9 a -B -1 45 6 6 w a -B -2 -2 4 Relative Abundance Relative 40 4 4 a5-B5 -4 978.6 w 35 705.5 14 -1 870.4 1036.6 a -B 30 1170.8 1285.1 4 4 803.1 -2 25 -1 -1 1118.6 w -1 w 958.4w 6 20 1 a -B 2 1 1 1413.0 w 15 1518.8 6 674.9 513.0 1602.7 -1 10 442.1 1661.9 1784.2 628.2 5 346.1 1919.1 0 400 600 800 1000 1200 1400 1600 1800 2000 m/z

5' 3' w w 17 w 16 w 2 1 G O C T A A HO O O A O O O G O P O O P O O P O O O P O O P O O P O - OH S - - S S - S - S S - n=13 a 2 -B 2 a -B a -B a -B 3 3 17 17 18 18 981.5 100 w -2 95 6 90 85 80 75 70 65 -1 a3-B3 60 B 55 50 1067.0 45 -2 -1 a -B Relative Abundance Relative 40 5 5 w 2 -4 35 a14-B14 -1 -2 a4-B4 30 w 5 978.7 862.6 25 -1 691.1 940.4 a -B 1147.0 1 1 697.8 1396.0 20 -2 w 817.8 15 -1 4 w 802.9 1307.3 1397.9 442.1 10 1 664.0 1461.0 525.0 1630.3 5 1707.8 1855.1 1933.0 362.1 0 400 600 800 1000 1200 1400 1600 1800 2000 m/z

227

250 GTI 2040 250

200 A 200 3’N-1 150 150 mAu mAu 100 3’N-2 100 3’N-3 50 50 0 0

0 5 10 15 20 25 30 35 40 Minutes

RT: 0.00 - 29.99 23.03 NL: 100 GTI 20409.77E7 TIC F: - p ES I 90 F ull m s [ 3’N-1 600.00-2000.00] 22.83 MS GTI 10uM 80 SVP 1U 20hr B 3’N-2 70 20.47 3’N-3 60 17.66 23.82 28.37 28.53 5.95 6.156.90 9.61 12.31 14.58 16.11 50 5.20

40 Relative Abundance 30 20

0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 Time (min)

250 250 400 400 C D 200 200

150 150 200 200 u A mAu mAu mAu m 100 100

50 50 0 0

0 0

0 5 10 15 20 25 30 35 40 0 10 20 30 40 Minutes Minutes

Figure 6.5 HPLC-UV chromatogram (A) and total ion LC-MS chromatogram (B) of GTI-2040 and metabolites after 5μM of GTI-2040 was incubated in SVP (0.5 U/mL) for 10 hrs. HPLC-UV chromatogram of GTI-2040 metabolism after incubation with phosphodiesterase II (0.5U/mL) for 10 hrs (C) and with phosphodiesterase II (5U/mL) for 24 hrs (D).

228

22.083 29.117 75 IS 75 GTI-2040 A A mVolts 50 50 mV olt s

25 20.925 25 18.083

0 0

0 5 10 15 20 25 30 35 40 Minutes

100 20.908 GTI-2040 3’N-1 100 B mVolts

18.225 mV olt 50 50 s 12.975

0 0

0 5 10 15 20 25 30 35 40 Minutes

17.633 C GTI-2040 3’N-2 60 60

40 40 mV olt s 6.533 8.175 20 20

0 0

0 5 10 15 20 25 30 35 40 Minutes

Figure 6.6 HPLC-UV chromatograms of the sequential metabolism profiles of GTI-2040 (A), 3’N-1 (B) or 3’N-2 (C) after each of them was incubated in 0.5 U/mL SVP solution.

229

150 150 GTI 2040 3’N-1

100 100

mAu mAu 3’N-2 50 50

0 0

0 5 10 15 20 25 30 35 40 Minutes

Figure 6.7 HPLC-UV chromatogram of GTI-2040 and metabolites after 5μM of GTI- 2040 was incubated in human blood for 24 hrs.

230

Figure 6.8 Metabolism profiles of GTI-2040 in mouse liver homogenates (A,B) and in human liver microsomes (C,D) GTI-2040 (5μM) was incubated in mouse liver homogenate (0.1g/ml) for 10 hrs. GTI- 2040 and metabolites from mouse liver homogenate extracts were detected in HPLC-UV chromatogram (A) and in LC-MS TIC (B) GTI-2040 (5μM) was incubated in human liver microsomes (HLM) (0.5 mg of protein/mL) for 10 hrs. GTI-2040 and metabolites from HLM were detected in HPLC- UV chromatogram (C) and in LC-MS TIC (D).

231

600 GTI 2040 600 A 3’N-1 400 400 mAu mAu 3’N-2 200 200

0 0

5 10 15 20 25 30 35 M inutes

RT: 0.00 - 29.99 22.56 100 3’N-1 95 GTI 2040 90 B 22.07 85 3’N-2 22.84 80 21.88 75 unknown 20.51 70 5.45 18.16 65 11.38 18.05 60 11.67 5.16 24.17 55 11.20 10.97 17.84 26.36 50 15.51 26.54 6.58 45 6.71 5.10 10.74 40 7.07 10.28 35 30

Relative Abundance Relative 7.52 25 7.69 20 15 3’N-3 10 5 0 0 5 10 15 20 25 Time (min)

500 GTI 500 C 400 400

300 3’N-1 300 mAu mAu 200 3’N-2 200 100 100

0 0

RT: 0.000 - 35.00 5 10 15 20 25 30 35 40 25.19 25.56 100 Minutes GTI 95 90 85 80 75 3’N-1 24.38 70 D 25.75 65 60 3’N-2 22.29 55 50 26.16 45 18.19 22.05 40 13.64 26.68 11.97 17.50 21.27 35 11.15 27.22 30 29.39 Relative Abundance Relative 25 20 15 10 5 0 0 5 10 15 20 25 30 Time (min)

232

Figure 6.9 Metabolism profiles of GTI-2040 in mouse kidneys homogenates (A, B) and in rat urinary excretion (C, D) GTI-2040 (5μM) was incubated in mouse kidney homogenates for 10 hrs. GTI-2040 and metabolites from the kidneys homogenates were detected in HPLC-UV chromatogram (A) and in LC-MS TIC (B). 0-7 hr rat urine was collected after an i.v. injection of GTI-2040 at 20mg/kg was given to the rat. GTI-2040 and metabolites in the rat urine were detected in HPLC-UV chromatogram (C) and in LC-MS TIC (D).

233

600 600 GTI 2040 A 400 3’N-1 400 mAu mAu 3’N-2 200 3’N-5 200

0 0 5 10 15 20 25 30 35 40 Minutes 23.22 100 95 GTI 2040 90 85 3’N-1 80 B 3’N-2 23.62 75 22.72 70 3’N-5 20.74 65 24.96 8.75 27.01 60 6.03 7.75 15.22 5.43 9.73 11.80 15.83 55 14.60 19.87 50 45 40 35 30 Relative Abundance Relative 25 20 15 10 5 0 0 5 10 15 20 25 Time (min)

C 3’N-1 600 GTI 2040 600 3’N-2 400 400 3’N-3 mAu mAu 3’N-4 200 200 3’N-5

0 0

5 10 15 20 25 30 35 40 M inutes 3’N-1 20.62 100 3’N-2 20.34 95 19.54 GTI 2040 90 3’N-3 17.88 20.89 85 D 80 15.58 75 3’N-4 15.27 70 16.05 65 60 3’N-5 12.98 5.76 9.15 55 12.32 50 7.45 8.40 9.41 21.51 11.09 22.96 7.27 45 23.25 40 23.75 35 25.54 26.29 30

Relative Abundance Relative 25 20 15 10 5 0 6 8 10 12 14 16 18 20 22 24 26 28 Time (min)

234 CHAPTER 7

ENZYME KINETICS OF GTI-2040, A PHOSPHOROTHIOATE OLIGONUCLEOTIDE TARGETING RIBONUCLEOTIDE REDUCTASE, TREATED WITH 3’ EXONUCLEASE AND HUMAN LIVER MICROSOMES

7.1 INTRODUCTION

Antisense oligonucleotides (ODNs) are short single strand DNA molecules designed to hybridize with specific mRNA strands thereby to selectively inhibit production of disease-related gene products (2,3). For successful therapeutic purpose, the in vivo stability of the antisense compounds is crucial. Unmodified ODNs are rapidly degraded in biological fluids by nucleases therefore limiting their clinical use. By substitution of one of the non-bridge oxygen atom with sulfur, the resulting phosphorothioate analogs are more resistant to exonucleases than the unmodified phosphodiester ODNs (4,8,13). Thus far, phosphorothioate oligonucleotides (PS-ODNs) have become the most widely used class of antisense drugs under clinical developments, due to their improved stability and their capability to activate the RNase H

(1,6,13,15,115). Various sequences of PS-ODNs are being evaluated in clinical trials for a number of diseases, including cancer, viral infections and inflammatory disorder.

235 Therefore, it has become increasingly important to obtain a fundamental understanding of the metabolism and degradation kinetics of PS-ODNs. Significant efforts have been invested in the identification of the in vivo and in vitro metabolites on this class of compound and 3’ exonuclease mediated hydrolysis of phosphorothioate linkage has been identified as the predominant pathway for metabolism of PS-ODNs (48,171,173). The 3’ end progressively deleted metabolites were detected in various PS-ODNs

(61,64,77,79,83). Following metabolite identification, it will be important to evaluate kinetics of metabolism in order to provide parameters for extrapolation from the in vitro metabolism to the in vivo clearance. However, till now no enzyme kinetics of PS-ODNs have ever been reported. This is probably due to the complex metabolism processes of

PS-ODNs and/or the lack of a proper analytical tool in characterization of the enzyme kinetics of PS-ODNs. First, possible sequential metabolism of PS-ODNs may complicate its enzyme kinetics. Selective monitoring of the formation rate of the major metabolite may differ from other metabolites or the total metabolites. In addition, PS-ODNs are highly protein bound in biological samples (54,172,185). The free fraction may be low; therefore, a sensitive and specific analytical is needed and such only recently becomes available. Additionally, the dose-dependent protein binding of PS-ODNs in biological matrices may cause the free fraction of the substrate varied across the substrate concentration used, which brings much difficulty in defining its kinetic models (186-

188).

GTI-2040 is a 20-mer phosphorothioate oligonucleotide that inhibits the production of the R2 subunit of ribonucleotide reductase (RNR), which is essential for the DNA synthesis (92,95). GTI-2040 showed a promising response and tolerable toxicity in phase 236 I clinical trials for the treatment of advanced solid tumors and acute myeloid leukemia

(10). In our previous study (unpublished data), using a highly specific ion-pair

HPLC/MS/MS method (78,79,168), metabolism of GTI-2040 was elucidated as a sequential nucleotide deletion via the 3’ exonuclease. A series of progressively 3’chain- shortend metabolites were identified in several biological media including plasma from

AML patients, solutions containing 3’ exonuclease and in human liver microsomes

(HLM). In this study, we investigated the enzyme kinetics of GTI-2040 in media containing 3’ exonuclease and in human liver micromses (HLM). Formation rates of the

3’ end metabolites were monitored and used to characterize the enzyme kinetics of GTI-

2040. The in vitro intrinsic clearance was determined and used to predict the in vivo value, which is verified by the intrinsic clearance determined from patients.

7.2 MATERIALS AND METHODS 7.2.1 Drugs and Chemicals

GTI-2040, a 20mer phosphorothioate oligonucleotide with the sequence 5'-GGC

TAA ATC GCT CCA CCA AG-3' was provided by the National Cancer Institute

(Bethesda, MD) and used without further purification. Putative 3’ end metabolites of GTI

2040, 3' N-1, 3' N-2, 3' N-3 (here-to-fore GTI-2040 is omitted), from which 1-3 nucleotides were deleted from the 3' end, respectively, and the internal standard (IS), PS- dC 28, a 28mer polycytidine phosphorothioate oligonucleotide were purchased from

Integrated DNA Technologies (Coralville, Iowa). The purity and identity of all of the oligomers were verified by HPLC/UV/Mass spectrometry (Finnigan LCQ, San Jose,

CA). Phosphodiesterase I (EC 3.1.4.1) from snake (Crotalus Adamanteus) venom (SVP), 237 a 3’ to 5’-exonuclease, was obtained from USB (Cleveland, OH). Pooled human liver microsomes from 18 individuals (10 males and 8 female donors) were obtained as a stock emulsion (20 mg of protein/ml in 250 mM sucrose) from BD Biosciences (Woburn, MA).

HPLC-grade methanol, triethylamine (TEA, 99.5%), triethylammonium bicarbonate

(TEAB) and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP, 99.8%) were purchased from

Sigma-Aldrich Co. (St. Louis, MO). HPLC-grade water was generated by an E-pure water purification system (Barnstead, Dubuque, IA).

7.2.2 HPLC Conditions

A Previously reported HPLC conditions (78) were used for the separation of GTI-

2040 and its chain-shortened metabolites. Briefly, the separation was achieved on a 2.5

μm Waters Xterra MS18 column (50 × 2.1 mm) coupled to a MS C18 10×2.1 mm guard column (Waters Corp., Milford, MA). The mobile phase was prepared as Mobile Phase A consisting of 100 mM HFIP buffered to pH 8.3 by 8.4 mM TEA and Mobile Phase B consisting of 100 mM HFIP and 8.6 mM TEA (pH 8.3) in methanol (50:50, v/v).

Gradient elution was used for the oligomer separation at a flow rate of 0.2 mL/min. The elution was initiated with 30% Mobile Phase B followed by a linear increase to 45% in

30 min, and returned to 30% in 2 min, which was maintained for 8 min before the next run. The column temperature was set to 50 °C throughout the analysis using a column heater (Keystone, Woburn, MA). The autosampler temperature was kept at 4 °C throughout the sample running. A Shimadzu HPLC system consisting of two LC-10AT vp pumps, a SIL-10AD vp autosampler, and a SPD-10A vp UV-VIS detector was used for quantification. The detected wavelength was set at 260 nm. 238

7.2.3 Sample Preparation--Solid Phase Extraction

Samples containing GTI-2040 and its metabolites were thawed and centrifuged at

1000 g for 5 min. One mL of the supernatant was mixed with 2 mL of 0.1 M TEAB and the mixture was allowed to stand at room temperature for 30 min for ion pair formation between the oligonucleotides and TEA. GTI-2040 and related metabolites were extracted with an Oasis HLB cartridge packed with 60 mg material (Waters Corp., Milford, MA).

The extraction tubes were pre-conditioned with 1 mL of acetonitrile followed by 1 mL of

0.1 M TEAB (pH 8.0). Samples mixed with 0.1 M TEAB were loaded onto these pre- conditioned solid phase columns. The protein and salts were removed by sequential washings with 3 mL of 0.1 M TEAB, 3 mL of distilled water, and 3 mL of 10% acetonitrile in 0.1 M TEAB by gravity flow. Then, the ODNs were eluted with 3 mL of

50% acetonitrile and the eluant was evaporated to dryness under N2. The residue was reconstituted with 150 μL of Mobile Phase A and a 50 μL aliquot was analyzed by

HPLC.

7.2.4 Protein binding

Plasma protein binding of GTI-2040 in human plasma was determined by ultrafiltration at a concentration range between 0.05-100 μM. The extent of binding of

GTI- 2040 to human liver microsomes was examined at the protein concentration of 0.2 mg/mL and a GTI-2040 concentration range between 0.1-20 μM, which was evaluated in the kinetic studies. EDTA (5 mM) was added to inhibit the nuclease activity in plasma and in microsomes. After incubation in a 37 º C shaking water bath for 30 min, the drug- 239 matrix mixture was placed into a disposable Ultrafree-MC (MW cutoff 30,000) filtration system (Millipore, Billerica, MA), which was centrifuged at 1500g for 30 min. An aliquot of protein-free filtrate and the initial samples were analyzed using a previously validated ELISA method (133). The fractions of unbound GTI-2040 were estimated from the percentage of the concentrations in the filtrate compared to the initial concentrations.

Samples prepared in saline were processed under the identical condition and was used as controls for non-specific binding of drug to the filter.

7.2.5 Enzyme Kinetics of GTI 2040 in the Solution containing 3’ Exonuclease

To prepare the 10 U/ml stock solution of Phosphodiesterase I (SVP), the lyophilized enzyme was reconstituted in Tris-salt buffer:glycerol (50:50, v/v) containing

110 mM Tris-HCl, pH 8.9, 110 mM NaCl, 15 mM MgCl2, and stored at –20 °C before use. To conduct the enzyme kinetic study, various concentrations of GTI-2040 ranged from 0.1 to 20 μM were incubated with 0.3 U/ml of SVP in Mobile Phase A (100 mM

HFIP titrated to pH 8.3 by 8.4 mM TEA) at 37 °C for 0.5 hr. The time and unit of enzyme used were both within the linear region in the reaction condition. The reactions were terminated by the addition of 5 mM EDTA. Then, an appropriate amount of PS-dC

28 was added and used as the internal standard and the samples were immediately frozen until analysis. For analysis, samples were thawed at 4 °C and analyzed by HPLC without further preparation. Solutions spiked with the initial levels of GTI-2040 in 0.3 U/ml SVP containing appropriate amounts of PS-dC 28 and 5mM EDTA without incubation were prepared as the controls for the calculation of metabolite concentrations.

240 7.2.6 Enzyme Kinetics of GTI-2040 in Pooled Human Liver Microsomes

Incubations of GTI- 2040 were carried out under initial linear conditions with respect to time (30 min) and microsomal protein concentration (0.2 mg protein/mL).

Substrates at the concentration range between 0.1 and 20 μM were incubated with human liver microsomes in a 1 mL tube in a shaking water bath at 37 °C with frequent inversed mixing. All incubations were performed in duplicate. At the end of incubation, 5 mM

EDTA was added to stop the reaction. Appropriate amounts of the IS PS-dC 28 were added into each sample. The mixture was then centrifuged at 3000 g for 2 min and the supernatant was separated and stored at –80 °C until analysis. GTI-2040 and its metabolites were extracted with SPE and analyzed by HPLC-UV. Another set of EDTA treated plasma samples containing GTI-2040 and PS-dC 28 without incubation was processed with SPE and. HPLC separation. These enzyme untreated samples were used as the controls with known concentrations in the calculation of the metabolite concentrations.

7.2.7 In vivo Clearance

Patients with acute myeloid leukemia (AML) were treated with GTI-2040 at 3.5 or 5 mg/kg/day as a continuous i.v. infusion for a total of 144 hr. Blood was drawn at various time points during and post infusion. Plasma was separated from the whole blood by centrifugation at 1400g for 20 min. Pretreatment and 24 hr cumulative urine was also collected. GTI-2040 concentrations in plasma and urine were determined by the previous reported ELISA method (11,133).

241 7.2.8 Quantitation

Quantitation of metabolites was performed using the intact GTI-2040 at each initial level in kinetic studies as the controls with known concentrations. Peak ratios of

GTI-2040 or each formed metabolites to IS in each HPLC chromatogram were obtained by dividing the peak areas of GTI-2040 or metabolites to the corresponding IS peak areas. The concentrations of formed metabolites were then calculated using the following equation:

R m ,I EG Cm = × CG,I × Eq. 1 RG,I Em

Cm is the concentration of the formed metabolites in the reaction tube. CG,I is the known concentrations of the initial substrate GTI-2040. Rm,I is the peak area ratio of metabolites to the internal standard. RG,I is the peak area ratio of the initial GTI-2040 to the internal standard. EG is the molar extinction coefficient of GTI-2040, Em is the molar extinction coefficient of the metabolites. EG/E3’N-1 was determined to be 1.053 in the calculation of

3’N-1 concentrations, and EG/ E3’N-2 was determined to be 1.128 in the calculation of

3’N-2 concentrations.

7.3 DATA ANALYSIS 7.3.1 Non-specific binding in HLM

The non-specific binding of GTI-2040 in HLM is described by Equation 2. KD and Bmax were calculated from the Scatchard equation (Equation 3), which is a linear transformation of Equation 2. 242 Bmax ×CF CB = Eq. 2 K D + CF

C B C B = max − B Eq.3 CF K D K D

CB is the concentration of bound drug, CF is the free drug concentration, Bmax is the maximal binding capacity and KD is the dissociation constant.

7.3.2 Calculation of Vmax and Km

The rate-substrate concentration profiles were fitted by the Michaelis-Menten equation (Eq 4) or Hill equation (Eq5) using the non-linear regression model in

SigmaPlot (Systat Inc. Point Richmond, CA).

V × S v = max Eq. 4 K m + S

n Vmax × S v = n n Eq. 5 S50 + S

where v and Vmax are the observed and maximal rates of metabolism, K m is the

concentration at half of Vmax in Michaelis-Menten equation, S50 is the concentration at

half of Vmax in the Hill equation, n is the Hill coefficient, and S is the substrate concentration.

7.3.3 Calculation of CLint in human patients.

Estimation of patients’ renal clearances and the calculation of hepatic CLint were described in the section 4.4.2 of Chapter 4. Briefly, renal clearance of GTI-2040 in

243 patients was estimated by dividing the cumulative amount of GTI-2040 in 24 hr urine collections to the 24 hr plasma AUC. Hepatic clearances from patients were then obtained by subtracting renal clearance from the total plasma clearance. Hepatic CLint was then calculated from the values of hepatic plasma clearance, the free fraction of the drug and the hepatic plasma flow rate, using the equations listed in the section 4.4.2.

7.3.4 Scaling of in vitro CLint to in vivo CLint

Microsomal protein amount was used as a scaling factor to transform the in vitro

CLint in microsomes to an in vivo CLint value. Assuming that 1 g of liver contains approximately 50 mg of microsomal protein and the liver of a 70-kg human weighs approximately 1,400 g, a scaling factor (SF) of 70,000 mg is obtained (189).

Multiplication of the in vitro CLint by this scaling factor yields the scaled in vivo CLint, expressed as liters per hour.

Scaled in vivo CLint = SF × in vitro CLint

7.4 RESULTS

7.4.1 Protein binding

GTI-2040 samples prepared in saline did not exhibit non-specific binding with filter after centrifugation. GTI-2040 was found to display concentration-dependent high plasma protein binding. When total GTI-2040 in plasma was < 0.5 μM, free drugs in the ultrafiltrate were below the LOQ of the ELISA assay (0.05 nM) and their protein bindings were probably higher than 99.99%. With GTI-2040 concentrations increased to 244 1, 10, 100 μM, mean plasma protein bindings were found to be 99.8%, 97.6% and 93.5%, respectively, based on triplicate determinations. In HLM, a concentration-dependent binding of GTI-2040 was also observed as shown in Table 7.1. With concentrations of

GTI-2040 increased from 0.1 μM to 20 μM, the free fraction of GTI-2040 in HLM (0.2 mg protein/ml) was found to enhance from 7.2 % to 63.9 %. The binding appears to be saturated, when the substrate concentrations were higher than 10 μM, as shown in the

Figure 7.1A. Two linear segments with different slopes were derived from the Scatchard plot (Figure 7.1B), indicating more than one binding site present in HLM for GTI-2040 with different affinities. A higher binding affinity with a dissociation constant KD = 0.03

μM, and maximum binding capacity Bmax=0.49 μM and a lower binding affinity with KD

= 3.8 μM and Bmax=8.9 μM were calculated from the Scatchard equation (Eq.3).

7.4.2 Kinetics of Metabolite Formation from GTI-2040 with 3’ Exonuclease

Figure 7.2C illustrated a representative HPLC chromatogram for the separation of GTI-2040 and metabolites in 0.3 U/ml of SVP. In the incubation period (0.5 hr), 3’N-1 and 3’N-2 were the two major metabolites identified in all levels of substrate concentrations. In the evaluated concentration range of substrate, the dose recoveries calculated as the percentage of the sum of the unchanged GTI-2040 and metabolites to the added amounts of GTI-2040 were found to be in the range of 85.8-109.8% with a mean value at 98.8%.

As the 3’N-1 is the most abundant metabolite of GTI-2040, we first examined its formation kinetics of 3’N-1. The plot of formation rate versus substrate concentration displayed a sigmoidal curve as shown in Figure 7.3A. This data were reprocessed using 245 an Eadie-Hofstee plot as shown in Figure 7.3B. This plot exhibits a curvilinear curve instead of a commonly found straight line. A curved Eadie-Hofstee plot indicated an atypical kinetics and deviates from the classical Michaelis-Menten model (190,191). In this case, the Hill equation (Equation 5) is commonly used for estimation of the parameters for the compound displaying atypical kinetics (190,192). Using the Hill equation, the S50 was calculated to be 1.21 ± 0.35 μM, Vmax to be 0.42 ± 0.16 nmol/hr, and Hill coefficient to be 1.41.

Enzyme kinetics of GTI-2040 in 0.3 U/ml of SVP was also evaluated using the sum of formation rates of all detected metabolites, i.e. sum of formation rates of 3’N-1 and 3’N-2. As shown in Figure 7.4A, a hyperbolic saturation curve was observed. In addition, the Eadie-Hofstee plot became linear (Figure 7.4B) and therefore, the

Michaelis-Menten equation was used to fit the total formation kinetics of metabolites. As determined by Equation 3, the Km was found to be 1.28 ± 0.42 μM and the Vmax 0.73 ±

0.06 nmol/hr. CLint of GTI-2040 in 0.3 U/ml of SVP, as the ratio of Vmax to Km, was calculated to be 0.60 ± 0.19 ml/hr.

7.4.3 Kinetics of Metabolites Formation of GTI-2040 in Pooled Human Liver

Microsomes

A representative HPLC chromatogram of the separation of GTI-2040 and metabolits in HLM is shown in Figure 7.5B. In the initial incubation period (0.5 hr), 3’N-

1 and 3’N-2 were found to be the major metabolites in all of the substrate concentrations used. The drug recoveries as obtained by comparison of the sum of the unchanged GTI-

246 2040 and metabolites to the added amount of GTI-2040 were in the range of 89.6-110.0% with a mean value at 98.9%.

When the formation rates of only the 3’N-1 metabolite were plotted against the

GTI-2040 concentrations, a sigmoidal shape curve was obtained as shown in Figure

7.6A. In addition, its Eadie-Hofstee plot also shows an atypical curvilinear shape, similar to that as in the case of GTI-2040 with SVP (Figure 7.6B), indicating a non-Michaelis-

Menten behavior. Using the Hill equation, the S50 was determined to be 7.65 ± 2.6 μM, the Vmax 13.3 ± 5.9 nmol/hr/mg protein, and the Hill coefficient 1.40 ± 0.1.

The sum of the formation rates of the major metabolites 3’N-1 and 3’N-2 was then plotted against the initial GTI-2040 concentrations and also shows a sigmoidal shape curve (Figure 7.7A). The Eadie-Hofstee plots, however, deviate from a linear relationship

(Figure 7.7B) and its kinetics was therefore better estimated by the Hill equation. The relevant kinetic constants as estimated are listed in Table 7.2. The S50 was found to be

12.1 ± 8.6 μM, the Vmax 16.8 ± 8.0 nmol/hr/mg protein, and the Hill coefficient 1.28 ±

0.4.

To examine the impact of the non-specific binding of GTI-2040 in HLM on the sigmoidal effect, unbound substrates (free fractions) were employed to determine the rate constants in metabolism kinetics (Figure 7.8A). As shown in Figure 7.8B, the Eadie-

Hofstee plots became linear using the free concentrations of substrate and the metabolites. Only the lowest concentration deviated from linearity. Therefore, after using the unbound fraction for the substrate and metabolites, metabolism kinetics of GTI-

2040 in HLM was found to follow the classical Michaelis-Menten behavior, and

Equantion 4 was used to calculate the kinetic constants. The free Km,f was estimated to be 247 4.4 ± 0.5 μM, the Vmax,f was 8.4 ± 0.4 nmol/hr/mg protein, and the CLint,f (Vmax,f/Km,f) was 1.91 ± 0.08 ml/hr.

7.4.4 Correlation of the in vitro CLint from HLM with the In vivo CLint

As described in the Chapter 4, the renal clearance values were estimated to be in the range from 0.00026 to 0.019 L/hr with a mean value of 0.032 L/hr in 21 evaluated patients. The in vivo intrinsic clearance was found to vary from 238.7 L/hr to 2846.4 L/hr with a mean value at 758.7 L/hr.

The in vitro CLint calculated from the unbound GTI-2040 in HLM was therefore used to estimate the in vivo CLint using the microsomal protein scaling factor. In this fashion, the predicted in vivo CLint from the in vitro HLM CLint was determined to be

133.5 L/hr. Compared to the in vivo CLint from patients as shown above, the predicted in vivo CLint represented 17.6% of the observed.

7.5 DISCUSSION

Following our previous identification of the 3’ end chain-shortened nucleotides being the major metabolites of GTI-2040 from both in vivo and in vitro experiments

(unpublished data), here we present the results of enzyme kinetics of GTI-2040 with 3’ exonuclease and in human liver microsomes based on the measurement of the formation rates of 3’ end metabolites.

248 In order to accurately characterize the kinetic property of a drug, it is important to apply an appropriate kinetic model to the in vitro data, especially when extrapolating the in vitro results to those of the in vivo. The Eadie-Hofstee plot ([V / s ]-[V ] plot) is a valuable tool in characterization of an enzyme kinetic model that defines the substrate concentration-velocity functions (193). Enzyme kinetics of most drugs can be adequately described by the Michaelis-Menten model which assumes independent single binding site per enzyme (193). A linear Eadie-Hofstee plot is usually expected from the Michaelis-

Menten model. However, when the enzyme-substrate interaction involves multiple sites resulting cooperative binding for the drug substrate, the Eadie-Hofstee plot will deviate from linearity and the use of a non-Michaelis-Menten model is necessary to adequately process the data (190,192,194). A positive cooperative binding between an enzyme and a substrate results in an initial lag in the rate-substrate concentration profile, which generates a sigmoidal rate-substrate concentration curve and a characteristic curvilinear

Eadie-Hofstee plot (190,191). In such a case, the Hill equation is usually used to characterize the sigmoidal kinetics (191,195-197). However, it must be noted that some artificial sources involving more complex factors from the drug properties and the metabolism systems could result in sigmoidal kinetics in the absence of enzyme autoactivation(188,190).

When interpreting the enzyme kinetics of GTI-2040, sequential metabolism should be taken into consideration, since progressively chained shortened metabolites have been identified. As shown in Figure 7.3, when the generation of only the 3’N-1 metabolite was considered with SVP, the Eadie-Hofstee plot is not linear in the range of evaluated substrate concentrations and the Hill equation was needed to fit the data. 249 When the sum of amounts of 3’N-1 and 3’N-2 metabolites (approximately the total GTI-

2040 metabolites) was considered, the Eadie-Hofstee plot became linear and the kinetics appeared to follow the Michaelis-Menten behavior (Figure 7.4). Therefore, it appears that in the sequential metabolism the characteristics of the enzyme kinetics of a single metabolite was different from the one observed with the formation of total metabolites.

We do not have a full explanation at present. However, one speculation would be due to an underestimation of the initial velocity if only a single metabolite was measured, while the sequential metabolism may occur juxtaposition with the same enzyme. For a compound generating multiple metabolites from the sequential metabolism, using the formation rate of the total metabolites may provide better estimation of its enzyme kinetics. In our experiments, 3’N-1 and 3’N-2 were found to be the major metabolites in the sequential metabolism of GTI-2040. Dose recoveries of GTI-2040 degradation were in the range of 85% to 110% when the total metabolites were estimated from the sum of

3’N-1 and 3’N-2, indicating the full quantitation of metabolites. Hence, enzyme kinetics of GTI-2040 calculated from the formation rate of 3’N-1 and 3’N-2 may be more accurate.

In PS-ODNs, the sulfurization process creates a random distribution of R and S configurations at each of the phosphorothioate linkage. 3’-exonuclease degradation of PS nucleotides is found to be stereoselective (82). Metabolism of Rp configuration was found to be >10 times faster than the Sp isomer via 3’exonuclease and in rat liver homogenate (77,81). The impact of 3’-exonuclease stereoselectivity on the kinetics of

GTI-2040 would be worthy to be evaluated. In our current metabolism study, no attempt was made to resolve any potential stereoisomers from neither the parent GTI-2040 nor 250 the metabolites, because of technical difficulty. It is possible, however, under the reaction conditions, the Rp isomers might be degraded first within the initial period, since this isomer undergoes a significantly more rapid degradation than the Sp isomers, and possibly all of the cleavage products would be the results from the hydrolysis of the Rp isomers. Therefore, the stereoselectivity in the metabolism of GTI-2040 may not contribute to its atypical enzyme kinetics.

In addition to the issue of sequential metabolism, non-specific binding of GTI-2040 in HLM was also found to confound the characterization of its enzyme kinetics. Similar in the kinetics in SVP, when the formation kinetics of only the 3’N-1 metabolite was examined, a curvilinear Eadie-Hofstee plot was obtained (Figure 7.6) (190). When the sum of all the detected metabolites was taken into consideration, a non-linear Eadie-

Hofstee plot was still obtained; however, the curvature significantly decreased (Figure

7.7). However, when the free fraction of the substrate in HLM was used, a linear Eadie-

Hofstee plot was essentially observed, except the point at the lowest substrate concentration (Figure 7.8), and the data satisfactorily conformed to the Michaelis-Menten kinetics. When the nonspecific drug binding in HLM is consistent over the range of substrate concentrations, the apparent measured Km,app value is usually converted to the free Km by multiplication of the free fraction of drug in HLM (fu(mic), Km=Km,app*fu(mic)), and no apparent complex effect would be predicted in its enzyme kinetics (187,198,199).

In GTI-2040, a dose-dependent protein binding in HLM was observed. It is possible that the sigmoidal kinetics of GTI-2040 in HLM, as expressed as the total substrate concentration, may be due in part to a variation of fu(mic) over the substrate concentration range used. Others have also showed the occurrence of sigmoidal kinetics as a result of 251 concentration-dependent binding of the substrate in the microsomes with the absence of enzyme autoactivation (188). When the protein binding of drugs is dose-dependent, and if the substrate concentrations used for in vitro kinetics are in the saturable binding range

(Km value or the substrate levels higher than the KD), sigmoidal kinetics and curvilinear

Eadie-Hofstee plots could arise as the consequence of the various fraction of the free substrate available to the enzyme across the substrate concentration used (188). McLure et al. (188) showed that the sigmoidicity becomes greater if KD is significantly less than

Km via a simulation studies. As shown in Figure 7.1, GTI-2040 displayed a saturable dose-dependent binding in HLM. A biphasic Schatchard plot suggested two binding sites of GTI-2040 in HLM, a higher affinity (KD at 0.03 μM) and a lower affinity (KD at 3.8

μM) sites. Most concentrations of the substrate GTI-2040 (0.1-20 μM) used in the kinetics as well as the determined Km value are greater than both of the KD values, which resulted in a variation of fu(mic) and the sigmoidal kinetics when expressed in terms of total substrate. The sigmoidicity is more significant to the lower concentrations of GTI-

2040 with their KD (0.03 μM) being much smaller than the Km value. After using the unbound fraction for the substrate, complications with the protein binding were largely removed and the actual enzymatic reactive substrate concentrations could be defined. The

Eadie-Hofstee plot in the free GTI-2040 became linear and the Michaelis-Menten kinetics resumed. Therefore, in the case of saturable binding of drugs in HLM, unbound drugs should be used to determine the relationship between rate and substrate concentrations.

As discussed above, properties of GTI-2040 including its sequential metabolism and high protein binding were found to complicate its metabolism and played critical 252 roles in revealing its true metabolic kinetics. Ignoring the complications from these two processes would result in a lower estimation of the velocity, especially in the lower substrate levels, which may lead to an artificial autoactivation kinetics, and an inaccurate estimation of in vitro CLint (190). The impact would be concentration-dependent, more significant at the lower concentrations and diminish at higher concentrations (190).

Other artificial sources of sigmoidicity include analytical sensitivity in quantification of the formation of metabolites (190). A Poorly defined limit of quantification may underestimate the formation rates of metabolites in the lower range of substrates, which could be presented as the initial lag in the rate-substrate concentration profile and resembled the sigmoidal curve. In our study, the sensitivity of HPLC-UV in the quantification the metabolites of GTI-2040 extracted from HLM was insufficient to detect the 3’N-2 metabolite at substrate concentrations lower than 0.1 μM. Therefore, the metabolites formation rate at 0.1 μM was much lower than the other levels and could not fit into the linear Eadie-Hofstee plot even after correction for protein binding.

In vivo intrinsic clearance from patients was compared to the predicted CLint that was scaled up from the free CLint of GTI-2040 in HLM. The predicted in vivo CLint only accounted for 17.6% of the observed in vivo CLint. This might indicate that human liver microsomes only partially contributed to the metabolism of GTI-2040. The underestimation of the in vivo CLint is likely due to the multiple metabolism sites of PS-

ODNs in vivo (61,62,64,164,172). In vivo, PS-ODNs extensively distribute to different organ tissues from the blood with the majority of drugs accumulating within the liver and the kidneys (79,102,200). Besides liver, significant amounts of chain-shortened metabolites were detected in the kidneys, especially in the proximal tubular cells 253 (68,79,102). Circulation system and other organ tissues would also contribute to the metabolism of GTI-2040 (75,82). Even in the liver, the suborgan distribution of PS-

ODNs and the metabolites is not homogenous. PS-ODNs were found to rapidly distributes to the nonparenchymal cells, including Kupffer cells and endothelial cells followed by a slow uptake into hepatocytes (65,174). Therefore, liver microsmes that derived from the hepatocytes may not be the sole system for the full prediction of the in vivo metabolism of GTI-2040. Since no enzyme kinetics of PS-ODN has been published, we selected human liver microsomes in this study, since liver is the major, relevant metabolism system for GTI-2040 (201,202), recognizing that there are other organ tissues that may also contribute significantly to the total intrinsic clearance of GTI-2040.

Since the primary metabolic pathway of GTI-2040 was mediated by 3' exonuclease, enzyme kinetics of GTI-2040 was also conducted in a solution containing

3'-exonuclease. MPA was chosen as the reaction matrix, because it provided good chromatographic separation of parent compound and metabolites. Buffers containing

NaCl and MgCl2 were found to cause a chromatographic peak distortion (data not shown) probably through interference with ion-pair formation. Only little information on the in vivo relevant expression level of 3’ exonuclease is available in the literature (75,81). In our study, 0.3 U/ml was used for SVP since SVP of this level provided a similar metabolism pattern of GTI-2040 as observed in vivo. Degradation of GTI-2040 could not be observed at a level lower than 0.05 U/ml of SVP in MPA. This limit is higher than the level used in the previously reported literature (0.001 U) (75). This discrepancy may be due to the use of different source of enzyme and the different reaction matrix.

Investigation of enzyme kinetics in SVP helped us to discern the complex factors from 254 the sequential metabolism in the kinetics of GTI-2040. Since the protein binding of GTI-

2040 in 0.3 U/ml of SVP was negligible (data not shown), complications from protein binding on the enzyme kinetics was probably not important and the identification of complex effect from sequential metabolism became easier in this metabolism system.

Impact of saturable protein binding on metabolism kinetics in HLM was subsequently identified with the recognition of the complication from sequential metabolism. Substrate concentrations used in this study (0.1 to 0.5 μM) covered the steady state plasma concentrations of GTI-2040 in patients. Therefore, results from the metabolism kinetics in SVP and HLM may possibly provide insights for the in vivo metabolism of GTI-2040 in patients.

In summary, enzyme kinetics of GTI-2040 was studied in solution containing 3’ exonuclease and in human liver microsomes. Michaelis-Menten kinetics of GTI-2040 was found with 3’ exonuclease for the formation of all detected sequential metabolites and in human liver microsomes using the free drug fraction. Careful considerations must be made for the interpretation of the enzyme kinetics of PS-ODNs with the complex effect from the sequential metabolism and the saturable binding in the metabolism matrix. Scaled intrinsic clearance from the in vitro HLM only partially predicted the in vivo intrinsic clearance.

255

Substrate concentration (μM) Free fraction (%) 0.1 7.2 ± 0.9 0.2 10.4 ± 0.3 0.5 22.4 ± 0.9 1 31.2 ± 7.1 2 32.9 ± 3.1 5 35.4 ± 6.3 10 63.5 ± 1.4 20 63.9 ± 2.7

Table 7.1 Free fraction of GTI-2040 in human liver microsomes (0.2 mg protein/mL). Results are expressed as Mean ± SD from triplicate measurements.

256

Values 0.3 U/ml SVPa HLM (total HLM (free drug)d drug)c b Vmax 0.73 ± 0.06 16.8 ± 8.0 8.4 ± 0.4 (nmole/hr/mg)

Km (μM) 1.28 ± 0.42 12.1 ± 8.6 4.4 ± 0.5 ne 0.97 ± 0.05 1.28 ± 0.4 1.12 ± 0.07 aSVP, cHLM: enzyme kinetic parameters were calculated from the plot of formation rate of sum of 3’N-1 and 3’N-2 vs total added substrate concentration b: with unit of nmole/hr dfree drug: enzyme kinetic parameters were calculated from the plot of formation rate of sum of 3’N-1 and 3’N-2 vs free fraction of the added substrate concentration en:Hill coefficient

Table 7.2 Kinetic parameters for the formation of 3’N-1 and 3’N-2 in 3’ exonuclease (SVP) AND IN HUMAN LIVER MICROSOMES Results are Mean ± SD from duplicate measurements

257

40 A

10 Free fraction of GTI-2040 in HLM(%)

0 0 5 10 15 20 25 GTI-2040 (µM)

12 B 10 KD =0.03μM

Bound/Free 4 K =3.8μM D 2

02468 Bound (µM)

Figure 7.1 Non-specific binding of GTI-2040 in human liver microsomes (HLM) (0.2 mg/mL) A. Plot of free fraction of GTI-2040 in HLM against the added GTI-2040 substrate concentrations. B. Scatchard plot of binding of GTI-2040 in human liver microsomes.

258

A 3’N-1 3’N-2 GTI-2040 40 3’N-3 IS 40 mVolts mV olt 20 20 s

21.0 22.2 0 0 13.6 18.4 28.5

0 5 10 15 20 25 30 35 40 B Minutes

10 10

0 0 mAu mAu

-10 -10

-20 -20

0 5 10 15 20 25 30 35 40 Minutes

C GTI-2040 IS 40 40

mVolts mV olt 20 20 s 3’N-1 3’N-2 0 21.1 22.3 29.2 0 18.24

0 5 10 15 20 25 30 35 40 Minutes

Figure 7.2 HPLC chromatograms of GTI-2040 and its metabolites in Mobile Phase A (MPA) containing 3’ exonuclease (SVP) A: Mixture of standards of GTI-2040 and its 3’N-1, 3’N-2 and 3’N-3 metabolites in MPA. B: 0.3 U/ml SVP alone in MPA after incubation at 37ºC for 30 min. C: GTI- 2040 and 3’N-1 and 3’N-2 metabolites after GTI-2040 (1 μM) was incubated with SVP (0.3 U/ml) in MPA for 30 min at 37ºC. 259

0.5

0.4 A

nM/hr) 0.3

0.2

0.1

Formation rate of3'N--1 ( 0.0

0.1 1 10 GTI-2040 conc. (µM)

0.5 B 0.4

0.3

0.2

V(nmole/hr) 0.1

0.0

0.00 0.05 0.10 0.15 0.20 0.25 V/s (nmole/hr/µM)

Figure 7.3 Plot of formation rate of 3’N-1 versus GTI-2040 substrate concentration (A) and the corresponding Eadie-Hofstee plot (B) of the kinetics of GTI-2040 in 0.3 U/ml of SVP solution

Each data point represents the mean of duplicate determinations.

260

0.7

0.6

A 0.5

0.4

0.3

0.2

0.1

0.0 Formation rate of sum of 3'N-1 and 3'N-2(nmole/hr) and 3'N-1 sum of of rate Formation 0.1 1 10

GTI-2040 conc. (µM)

0.8

0.6 B

0.4

0.2

V(nmole/hr)

0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

V/s (nmole/hr/µM)

261

IS 75 A 75

mVolts 50 50 mV olt s

25 25

0 28.3 0

0 5 10 15 20 25 30 35 40 Minutes

GTI-2040 100 B 100

mVolts IS mV olt s 50 50 3’N-1 3’N-2

0 0 18.7 21.3 22.4 28.8

0 5 10 15 20 25 30 35 40 Minutes

Figure 7.5 HPLC chromatograms of GTI-2040 and its metabolites in human liver microsomes (HLM) A: 0.2 mg/ml HLM alone after incubation at 37 °C for 30 min. B: GTI-2040 and 3’N-1, 3’N-2 metabolites after GTI-2040 (1 μM) was incubated in 0.2 mg/ml of HLM at 37 °C for 30 min.

262

A 8

0 Formation rate of 3'N-1 (nmole/hr/mg) 3'N-1 rate of Formation

0.1 1 10 GTI-2040 concentration (µM)

10 B

V (nmole/hr/mg) 2

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 V/[s] (nmole/hr/mg/µM)

263

10 A

Formation rate of sum 2

3'N-1 and 3'N-2 in HLM(nmole/hr/mg) 0

0.1 1 10 GTI-2040 concentration (µM)

10 B

4 V(nmole/hr/mg)

0.2 0.4 0.6 0.8 1.0 1.2 V/s (nmole/hr/mg/µM)

264

8 A 6

of sum rateof Formation 0 free 3'N-1 and 3'N-2 (nmole/hr/mg)

-2 0.001 0.01 0.1 1 10 100 free GTI-2040 concentration (µM)

(nmole/hr/mg) V 2

0.40.60.81.01.21.4 V/s (nmole/hr/mg/µM)

265 CHAPTER 8

CONCLUSION AND FUTURE PERSPECTIVES

In order to support the clinical evaluation and development of the phosphorothioate oligonucleotide GTI-2040 for the treatment of acute myeloid leukemia, its in vivo and in vitro cellular uptake, metabolism, and PK/PD correlations have been extensively investigated in this dissertation project.

Prior to this project, a suitable analytical tool was needed. Thus, a novel hybridization-ligation ELISA method has been developed and validated for the quantification of GTI-2040 in a variety of matrices. The picomolar sensitivity of the

ELISA assays allowed the accurate characterization of pharmacokinetics of GTI-2040 and the determination of intracellular drug concentrations in PBMC and in bone marrow cell lysates.

The sequence-specific and concentration-dependent down-regulation of R2 mRNA and protein in GTI-2040 were confirmed in treated K562 cells. The manifestation of R2 mRNA down regulation by GTI-2040 in K562 cells required the intracellular delivery of the antisense by cationic lipids, as low cellular uptake of the drug was found, when K562 cells were treated with noncomplexed GTI-2040. Additionally, low nuclear

266 accumulation of GTI 2040 was found and no suppression of R2 mRNA was observed with the non-complexed drug. The transfection agent not only enhanced the ICs of GTI-

2040 to more than 10 fold compared to the noncomplexed drug, but also resulted in more than 4 fold higher nuclear accumulation than the IC of GTI-2040 in cytoplasm. Robust in vivo ICs of GTI-2040 were achieved in bone marrow (BM) and peripheral blood cells obtained from GTI-2040 treated AML patients. GTI-2040 concentrations in the nucleus of bone marrow cells rather than the total ICs were found to correlate with the change in

R2 mRNA expression and the disease response. Therefore, in future studies, subcellular kinetics, the intracellular disposition and trafficking in endosomes/lysosomes, cytosol, and nucleus of GTI-2040 need to be investigated. Moreover, efficient nuclear targeting of

GTI-2040 could be explored using nuclear localization signal peptides (NLS) (203). By conjugation with a NLS peptide via peptide nucleic acid (PNA) linkage, nuclear targeted translocation of the PNA-NLS/GTI-2040 complex would be possible (203,204).

Cellular uptakes of PS-ODNs have been shown to be time-, dose- and temperature dependent probably via a receptor mediated endocytosis (43,134-136). In addition, a number of papers have reported the findings of several cell membrane proteins involved in the binding and localization of PS-ODNs, for example, a 66 kDa membrane ODN receptor for a 25 mer PS-ODN was purified from HepG2 cells (138). In vivo, GTI-2040 showed a preferential uptake to PBMC over RBC in peripheral blood and bone marrow cells with CD 34+ sorted over unmanipulated cells. Based on our data and other previous reports, we hypothesized that cell uptake and internalization of GTI-2040 may also undergo an active transport mediated by certain membrane-bound proteins. Also there may exist some specific transporters expressed on PBMC or CD 34+ cells membrane, 267 and these transporters may bind to GTI-2040 with a high affinity and result in an efficient drug uptake. In future studies, possibility of the active transport of GTI-2040 could be examined by determination its intracellular uptakes using the inhibitors (sodium azide or chloroquine) for the energy-dependent endocytosis or with the competence of other polyanions. Additionally, in order to characterize the cellular membrane transport proteins of GTI-2040, one suggested method would be to use streptavidin microbeads coated with biotin labeled GTI-2040 to extract the GTI-2040 bound proteins after incubation of GTI-2040 coated magnetic beads in cell cultures. Proteins could then be separated by 10% polyacrylamide gel and identified using MALDI-TOF and/or Q-TOF following tryptic digestion (41,138). All of the above cellular uptake studies will be proposed to perform in leukemia cell lines in vitro in culture or as primary tumor cells obtained from leukemia patients. The results of these studies may provide insights in improving the cellular uptake of GTI-2040 by a rational design of drug delivery systems.

As shown in Chapter 3, the preliminary cytotoxicity evaluation of GTI-2040 in combination with cytarabine in K562 cells showed a promising synergistic effect. Since the drug effect of GTI-2040 and cytarabine are both cell cycle dependent (89,91), the drug effects in different dosing sequences and at different cell cycles should also be examined. In the current dose regimen, AML patients were pre-dosed with GTI-2040 followed by co-administration of cytarabine. It has been reported that low dose cytarabine causes S phase arrest in murine stem cells (205). Since the expression of R2 mRNA is mainly at the late G1 phase/early S phase, pretreatment with low dose cytarabine may recruit more cells into the S phase and may potentiate the antisense effect

268 of GTI-2040. Synchronization of cells to the S phase in vitro by cell starvation may also be useful to examine the cell cycle effect on the antisense activity of GTI-2040.

Correlation of the ICs of GTI-2040 in bone marrow cells, R2 mRNA/protein down-regulation and disease response was observed in this phase I clinical trial (Chapter

4). Over-expression of the baseline levels of R2 mRNA was found to be important to manifest antisense activity of GTI-2040. Due to the limited population of patients and the small assessable sample size in this phase I clinical trial, statistically significant linear correlations were not obtainable in the PK/PD and response relationships. There may exist a threshold value of baseline expression of R2 mRNA in order for the antisense to down-regulate and this needs to be established in order to use it as a criteria in patients enrollment in future clinical studies. In addition, the change in R2 protein showed an excellent correlation to the disease response in the patients younger than 60 yrs old.

Therefore, R2 protein would be an important biomarker for the GTI-2040 treatment and need to be accurately determined. A more reliable ELISA quantification method than the current western blotting method is necessary. In addition, RNR enzyme activity should be monitored and correlated with the antisense effect and disease response in future clinical trials. Our experience with the PK/PD and response correlation studies in this phase I trial have provided a useful procedure and an assay platform, which will be useful for future phase II and phase III studies.

Additionally, in this clinical trial, population pharamcokinetics of GTI-2040 in plasma and in PBMC were studied in AML patients using NONMEM. Bone marrow cellularity was found to correlate to patients’ plasma clearance, and the white blood cells counts were found to relate to the cellular uptake of GTI-2040 in PBMC. However, this 269 result needs to be confirmed in a larger patient population in order to guide the future dose design. A PK/PD integrated indirect response model was used to characterize the dynamic of down-regulation of R2 mRNA following drug perturbation. Due to the limited sampling information, the onset phase and the recovery phase of R2 mRNA down-regulation may not be accurately defined using this model. More frequent bone marrow sampling will be desirable to measure the changes of R2 mRNA and proteins during and post drug treatment. However, due to the practical ethical issues during the frequent procurement of bone marrow samples in each patient, different sampling times for each patient should be designed so that the entire patient population will provide an adequate coverage of PD profile. Such design is based on the premise that population

PK/PD study supports the analysis of scattered information from large patient sizes and provide a complete PK/PD description. With a better-designed PK/PD study, drug effects in patients will be evaluated more precisely and the clinical applications of antisense therapy can be improved.

3’ end chain-shortened metabolites of GTI-2040 were identified in various specimens including patients’ plasma, rat urine, mouse liver and kidney homogenate, human liver microsomes and 3’ exonuclease solution. The pharmacological importance of the 3’ metabolites has yet to be evaluated in the future study. In addition, enzyme kinetics was conducted in a solution containing 3’ exonuclease and in human liver microsomes. The in vitro CLint obtained from human liver microsomes was found to contribute in part to the in vivo CLint. In future studies, a similar enzyme kinetic study could be proposed to perform in blood, liver biopsy homogenate and kidney biopsy homogenate. Effects of the sequential metabolism and protein binding in enzyme kinetics 270 could be further examined in these systems. The extent of metabolism in different tissues could be compared using the calculated CLint from each tissue. Scaled in vitro CLint will better predict the in vivo CLint when in vitro CLint are available from multiple tissues.

From these future studies, the exact nature of the metabolism of GTI-2040 will be better understood.

In summary, GTI-2040 cellular uptake, PK/PD correlation and metabolism were investigated in this dissertation. Results from these studies provided valuable information in the evaluation and utilization of GTI-2040 in the clinics. Future studies as proposed above will broaden our understanding in PS-ODNs and optimize its clinical application.

271 APPENDIX A

CONTROL STREAMS IN NONMEM PROGRAM USED FOR POPULATION

PHARMACOKINETIC MODELING OF PLASMA GTI-2040 IN AML PATIENTS

WITH COVARIATES

272 $PROB GTI PLASMA POPULATION DATA 31pts with covariates

$INPUT ID AMT RATE DUR TIME DV MDV WT BSA AGE SEX CRCL

HGB PLTS WBC ANC PB CELL BB

$DATA GTIplasma_for_all_COV_EX2829_3.CSV IGNORE=#

$SUBROUTINES ADVAN3 TRANS4

$PK

TVCL=THETA(1)*(1+CELL/70)

CL=TVCL*EXP(ETA(1))

TVV1=THETA(2)*(SEX-1)+THETA(6)*(2-SEX)

V1=TVV1*EXP(ETA(2))

S1=V1

TVV2=(THETA(3)*(SEX-1)+THETA(7)*(2-SEX))

V2=TVV2*EXP(ETA(3))

Q=THETA(4)*EXP(ETA(4))

K=CL/V1

K12=Q/V1

K21=Q/V2

$THETA (0,5,100) ;CL 273 $THETA (0,5,500) ;V1F female

$THETA (0,3,500) ;V2F female

$THETA (0,0.1,5) ;Q

$THETA (0,1) ;SD

$THETA (0,10,500) ;V1M male

$THETA (0,10,500) ;V2M male

$ERROR

cALLFL=0

IPRED=F

IF (IPRED.EQ.0) IPRED=EXP(-9)

W=(F**2+THETA(5)**2)**0.5

IRES=DV-IPRED

IWRES=IRES/W

Y=IPRED+W*EPS(1)

$OMEGA 0.25 0.5 0.5 0.5

$SIGMA 0.25

$EST SIG=3 MAXEVAL=9999 POSTHOC METHOD=1 INTER

$cov

$TABLE ID AMT RATE DUR TIME NOPRINT ONEHEADER FILE=mytab82 274 $TABLE ID TIME IPRED IWRES

NOPRINT ONEHEADER FILE=sdtab82

$TABLE ID CL Q V1 V2 ETA1 ETA2 ETA3 ETA4 NOPRINT ONEHEADER

FILE=patab82

$TABLE ID WT BSA AGE CRCL HGB PLTS WBC ANC PB CELL BB

NOPRINT ONEHEADER FILE=cotab82

$TABLE ID SEX NOPRINT ONEHEADER FILE=catab82

$SCAT DV VS PRED UNIT

275 APPENDIX B

CONTROL STREAMS IN NONMEM PROGRAM USED FOR POPULATION

PHARMACOKINETIC MODELING OF PBMC GTI-2040 IN AML PATIENTS WITH

COVARIATES

276 $PROB GIT PLASMA PK FIXED PBMC SIMULATION 15PT COVARIATES

$INPUT ID AMT RATE DUR TIME CMT DV MDV CL1 CL4 V1 V3 WT SEX HGB

PLTS WBC ANC PB

$DATA GTIPBMCONLYPP_COV_2.csv IGNORE=#

$SUBROUTINES ADVAN9 TRANS1 TOL=3

$MODEL COMP=(CENTRAL,DEFDOSE) COMP=(PERIPH1,DEFOBS) cOMP(PERIPH2)

$PK

CL3=THETA(1)*EXP(ETA(1)) ; PBMC to central

TVV2=THETA(2)*(HGB/8.6)

V2=TVV2*EXP(ETA(2)) ; PBMC volume

TVVM=THETA(3)*(SEX-1)+THETA(5)*(2-SEX)

VM=TVVM*EXP(ETA(3))

TVKM=THETA(4)*(1.6/WBC)

KM=TVKM*EXP(ETA(4))

K10=CL1/V1 277 K21=CL3/V2

K13=CL4/V1

K31=CL4/V3

$THETA (0,0.01,1) ; CL3

$THETA (0,0.1,5) ;V2

$THETA (0,1,200) ;VMF female (nM/hr)

$THETA (0,5,200) ; KM (nM)

$THETA (0,0.5,200);VMM male

$THETA (0,0.5,) ; POWER ERROR

$DES

C1=A(1)/V1

DADT(1)=RATE-K10*A(1)-A(1)*VM/(KM+C1)+K21*A(2)-K13*A(1)+K31*A(3)

DADT(2)=A(1)*VM/(KM+C1)-K21*A(2)

DADT(3)=K13*A(1)-K31*A(3)

$ERROR

cALLFL=0

IPRED=F 278 IF (IPRED.EQ.0) IPRED=EXP(-9)

W=IPRED**THETA(6)

IRES=DV-IPRED

IWRES=IRES/W

Y=IPRED+W*EPS(1)

$OMEGA 0.5 0.5 0.5 0.5

$SIGMA 1

$EST SIG=3 MAXEVAL=9999 POSTHOC MSFO=msfb38 POSTHOC METHOD=1

INTER NOABORT

$TABLE ID AMT RATE DUR NOPRINT ONEHEADER FILE=mytab38

$TABLE ID CMT TIME IPRED IWRES

NOPRINT ONEHEADER FILE=sdtab38

$TABLE ID V1 V2 V3 CL1 CL3 CL4 KM VM ETA1 ETA2 ETA3 ETA4

NOPRINT ONEHEADER FILE=patab38

$TABLE ID WT HGB PLTS WBC ANC PB

NOPRINT ONEHEADER FILE=cotab38

$TABLE ID SEX NOPRINT ONEHEADER FILE=catab38 279 $SCAT DV VS PRED BY CMT UNIT

280 APPENDIX C

COMPUTER PROGRAMS USED FOR FITTING OF PK/PD MODEL OF R2 MRNA

IN PATIENTS USING ADAPT II

281 C**********************************************************************

C ADAPT II

C Release 4

C**********************************************************************

C MODEL-R2 mRNA in patients

C Xiaohui Wei

C**********************************************************************

Subroutine DIFFEQ(T,X,XP)

Implicit None

Include 'globals.inc'

Include 'model.inc'

Real*8 T,X(MaxNDE),XP(MaxNDE)

C------C

C 1. Enter Differential Equations Below {e.g. XP(1) = -P(1)*X(1) } C

C----c------C

Real*8 VV 282

XP(1) = R(1)-(P(1)+P(4)/(P(5)+X(1)/P(10))+P(2))*X(1)+P(3)*X(2)

XP(2) = P(2)*X(1)-P(3)*X(2)

XP(3) = (P(4)/(P(5)+X(1)/P(10)))*X(1)-P(6)*X(3)

VV=(X(3)/P(11))*0.35

XP(4) = P(7)-(P(7)/IC(4))*(1+P(8)*VV/(P(9)+VV))*X(4)

C------C

C######################################################################

Subroutine AMAT(A)

Implicit None

Include 'globals.inc'

Include 'model.inc'

Integer I,J

Real*8 A(MaxNDE,MaxNDE)

DO I=1,Ndeqs

Do J=1,Ndeqs

A(I,J)=0.0D0

End Do 283 End Do

C------C

C 2. Enter non zero elements of state matrix {e.g. A(1,1) = -P(1) } C

C----c------C

C######################################################################

Subroutine OUTPUT(Y,T,X)

Implicit None

Include 'globals.inc'

Include 'model.inc'

Real*8 Y(MaxNOE),T,X(MaxNDE)

C------C

C 3. Enter Output Equations Below {e.g. Y(1) = X(1)/P(2) } C

C----c------C

Y(1) = X(1)/P(11) 284 Y(2) = X(4)

C######################################################################

Subroutine SYMBOL

Implicit None

Include 'globals.inc'

Include 'model.inc'

C------C

C 4. Enter as Indicated C

C------C

NDEqs = 4 ! Enter # of Diff. Eqs.

NSParam = 11 ! Enter # of System Parameters.

NVparam = 4 ! Enter # of Variance Model Parameters.

NSecPar = 0 ! Enter # of Secondary Parameters.

NSecOut = 0 ! Enter # of Secondary Outputs (not used).

Ieqsol = 1 ! Model type: 1 - DIFFEQ, 2 - AMAT, 3 - OUTPUT only.

Descr = 'GTI plasma and mRNA pt simulation'

C------C 285 C 4. Enter Symbol for Each System Parameter (eg. Psym(1)='Kel') C

C----c------C

Psym(1) = 'K10'

Psym(2) = 'K12'

Psym(3) = 'K21'

Psym(4) = 'VM'

Psym(5) = 'KM'

Psym(6) = 'K30'

Psym(7) = 'K04'

Psym(8) = 'EMAX'

Psym(9) = 'EC50'

Psym(10)= 'V1'

Psym(11)= 'V3'

C------C

C 4. Enter Symbol for Each Variance Parameter {eg: PVsym(1)='Sigma'} C

C----c------C

PVsym(1) = 'SDinter1'

PVsym(2) = 'SDslope1'

PVsym(3) = 'SDinter2'

PVsym(4) = 'SDslope2' 286

C------C

C 4. Enter Symbol for Each Secondary Parameter {eg: PSsym(1)='CLt'} C

C----c------C

C######################################################################

Subroutine VARMOD(V,T,X,Y)

Implicit None

Include 'globals.inc'

Include 'model.inc'

Real*8 V(MaxNOE),T,X(MaxNDE),Y(MaxNOE)

287 APPENDIX D

DATA RELATED TO CHAPTER 5: PK/PD FITTING RESULTS IN DOWN-

REGULAION OF R2 MRNA FROM PATIENTS BM SAMPLES

288 120

observed R2 mRNA level 100 predicted R2 mRNA level

80 Pt 0304-05 60

40 pretreatment pretreatment to to mRNA mRNA R2 R2 of of %

20 0 50 100 150 200 250 300 time (hr)

110

100 observed R2 mRNA level 90 predicted R2 mRNA level 80

70 Pt 0304-19 60 50 % of R2 mRNA to pretreatment 40

30 0 50 100 150 200 250 300 time (hr)

120

100 observed R2 mRNA level predicted R2 mRNA level

80 Pt 0304-18 60

% of R2mRNA topretreatment 20

0 0 50 100 150 200 250 300 time (hr)

289

RFERENCE

1. Jansen B, Zangemeister-Wittke U. Antisense therapy for cancer--the time of truth Lancet Oncol 2002; 3: 672-683.

2. Stein CA, Cheng YC. Antisense oligonucleotides as therapeutic agents--is the bullet really magical? Science 1993; 261: 1004-1012.

3. Dias N, Stein CA. Antisense oligonucleotides: basic concepts and mechanisms Mol Cancer Ther 2002; 1: 347-355.

4. Crooke ST. Progress in antisense technology Annu Rev Med 2004; 55: 61-95.

5. Helene C, Toulme JJ. Specific regulation of gene expression by antisense, sense and antigene nucleic acids Biochim Biophys Acta 1990; 1049: 99-125.

6. Dean NM, Bennett CF. Antisense oligonucleotide-based therapeutics for cancer Oncogene 2003; 22: 9087-9096.

7. Bennett CF. Efficiency of antisense oligonucleotide drug discovery Antisense Nucleic Acid Drug Dev 2002; 12: 215-224.

8. Aboul-Fadl T. Antisense oligonucleotides: the state of the art Curr Med Chem 2005; 12: 2193-2214.

9. Shangary S, Johnson DE. Recent advances in the development of anticancer agents targeting cell death inhibitors in the Bcl-2 protein family Leukemia 2003; 17: 1470-1481.

10. Desai AA, Schilsky RL, Young A, Janisch L, Stadler WM, Vogelzang NJ, Cadden S, Wright JA, Ratain MJ. A phase I study of antisense oligonucleotide GTI-2040 given by continuous intravenous infusion in patients with advanced solid tumors Ann Oncol 2005; 16: 958-965.

290

11. Dai G, Chan KK, Liu S, Hoyt D, Whitman S, Klisovic M, Shen T, Caligiuri MA, Byrd J, Grever M, Marcucci G. Cellular uptake and intracellular levels of the bcl- 2 antisense g3139 in cultured cells and treated patients with acute myeloid leukemia Clin Cancer Res 2005; 11: 2998-3008.

12. Marcucci G, Stock W, Dai G, Klisovic RB, Liu S, Klisovic MI, Blum W, Kefauver C, Sher DA, Green M, Moran M, Maharry K, Novick S, Bloomfield CD, Zwiebel JA, Larson RA, Grever MR, Chan KK, Byrd JC. Phase I study of oblimersen sodium, an antisense to Bcl-2, in untreated older patients with acute myeloid leukemia: pharmacokinetics, pharmacodynamics, and clinical activity J Clin Oncol 2005; 23: 3404-3411.

13. Kurreck J. Antisense technologies. Improvement through novel chemical modifications Eur J Biochem 2003; 270: 1628-1644.

14. Furdon PJ, Dominski Z, Kole R. RNase H cleavage of RNA hybridized to oligonucleotides containing methylphosphonate, phosphorothioate and phosphodiester bonds Nucleic Acids Res 1989; 17: 9193-9204.

15. Agarwal N, Gewirtz AM. Oligonucleotide therapeutics for hematologic disorders Biochim Biophys Acta 1999; 1489: 85-96.

16. Sazani P, Kole R. Therapeutic potential of antisense oligonucleotides as modulators of alternative splicing J Clin Invest 2003; 112: 481-486.

17. Maher LJ, III, Wold B, Dervan PB. Inhibition of DNA binding proteins by oligonucleotide-directed triple helix formation Science 1989; 245: 725-730.

18. Walder RY, Walder JA. Role of RNase H in hybrid-arrested translation by antisense oligonucleotides Proc Natl Acad Sci U S A 1988; 85: 5011-5015.

19. Lima WF, Mohan V, Crooke ST. The influence of antisense oligonucleotide- induced RNA structure on Escherichia coli RNase H1 activity J Biol Chem 1997; 272: 18191-18199.

20. Baker BF, Miraglia L, Hagedorn CH. Modulation of eucaryotic initiation factor- 4E binding to 5'-capped oligoribonucleotides by modified anti-sense oligonucleotides J Biol Chem 1992; 267: 11495-11499.

21. Gallo M, Montserrat JM, Iribarren AM. Design and applications of modified oligonucleotides Braz J Med Biol Res 2003; 36: 143-151.

291 22. Manoharan M. Oligonucleotide conjugates as potential antisense drugs with improved uptake, biodistribution, targeted delivery, and mechanism of action Antisense Nucleic Acid Drug Dev 2002; 12: 103-128.

23. Sewell KL, Geary RS, Baker BF, Glover JM, Mant TG, Yu RZ, Tami JA, Dorr FA. Phase I trial of ISIS 104838, a 2'-methoxyethyl modified antisense oligonucleotide targeting tumor necrosis factor-alpha J Pharmacol Exp Ther 2002; 303: 1334-1343.

24. Stewart DJ, Donehower RC, Eisenhauer EA, Wainman N, Shah AK, Bonfils C, MacLeod AR, Besterman JM, Reid GK. A phase I pharmacokinetic and pharmacodynamic study of the DNA methyltransferase 1 inhibitor MG98 administered twice weekly Ann Oncol 2003; 14: 766-774.

25. Adjei AA, Dy GK, Erlichman C, Reid JM, Sloan JA, Pitot HC, Alberts SR, Goldberg RM, Hanson LJ, Atherton PJ, Watanabe T, Geary RS, Holmlund J, Dorr FA. A phase I trial of ISIS 2503, an antisense inhibitor of H-ras, in combination with gemcitabine in patients with advanced cancer Clin Cancer Res 2003; 9: 115- 123.

26. Marcucci G, Byrd JC, Dai G, Klisovic MI, Kourlas PJ, Young DC, Cataland SR, Fisher DB, Lucas D, Chan KK, Porcu P, Lin ZP, Farag SF, Frankel SR, Zwiebel JA, Kraut EH, Balcerzak SP, Bloomfield CD, Grever MR, Caligiuri MA. Phase 1 and pharmacodynamic studies of G3139, a Bcl-2 antisense oligonucleotide, in combination with chemotherapy in refractory or relapsed acute leukemia Blood 2003; 101: 425-432.

27. Yoo BH, Bochkareva E, Bochkarev A, Mou TC, Gray DM. 2'-O-methyl-modified phosphorothioate antisense oligonucleotides have reduced non-specific effects in vitro Nucleic Acids Res 2004; 32: 2008-2016.

28. Zhang H, Cook J, Nickel J, Yu R, Stecker K, Myers K, Dean NM. Reduction of liver Fas expression by an antisense oligonucleotide protects mice from fulminant hepatitis Nat Biotechnol 2000; 18: 862-867.

29. Zamaratski E, Pradeepkumar PI, Chattopadhyaya J. A critical survey of the structure-function of the antisense oligo/RNA heteroduplex as substrate for RNase H J Biochem Biophys Methods 2001; 48: 189-208.

30. Baker BF, Lot SS, Condon TP, Cheng-Flournoy S, Lesnik EA, Sasmor HM, Bennett CF. 2'-O-(2-Methoxy)ethyl-modified anti-intercellular adhesion molecule 1 (ICAM-1) oligonucleotides selectively increase the ICAM-1 mRNA level and inhibit formation of the ICAM-1 translation initiation complex in human umbilical vein endothelial cells J Biol Chem 1997; 272: 11994-12000.

292 31. Nielsen PE. Antisense properties of peptide nucleic acid Methods Enzymol 2000; 313: 156-164.

32. Elayadi AN, Corey DR. Application of PNA and LNA oligomers to chemotherapy Curr Opin Investig Drugs 2001; 2: 558-561.

33. Vickers TA, Koo S, Bennett CF, Crooke ST, Dean NM, Baker BF. Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents. A comparative analysis J Biol Chem 2003; 278: 7108-7118.

34. Elbashir SM, Lendeckel W, Tuschl T. RNA interference is mediated by 21- and 22-nucleotide RNAs Genes Dev 2001; 15: 188-200.

35. Miyagishi M, Hayashi M, Taira K. Comparison of the suppressive effects of antisense oligonucleotides and siRNAs directed against the same targets in mammalian cells Antisense Nucleic Acid Drug Dev 2003; 13: 1-7.

36. Paroo Z, Corey DR. Challenges for RNAi in vivo Trends Biotechnol 2004; 22: 390-394.

37. Alvarez-Garcia I, Miska EA. MicroRNA functions in animal development and human disease Development 2005; 132: 4653-4662.

38. Ambros V. The functions of animal microRNAs Nature 2004; 431: 350-355.

39. Xu P, Guo M, Hay BA. MicroRNAs and the regulation of cell death Trends Genet 2004; 20: 617-624.

40. Loke SL, Stein CA, Zhang XH, Mori K, Nakanishi M, Subasinghe C, Cohen JS, Neckers LM. Characterization of oligonucleotide transport into living cells Proc Natl Acad Sci U S A 1989; 86: 3474-3478.

41. Zhang Z, Weinschenk T, Schluesener HJ. Uptake, cellular distribution and novel cellular binding proteins of immunostimulatory CpG oligodeoxynucleotides in glioblastoma cells Mol Cell Biochem 2005; 272: 35-46.

42. Yakubov LA, Deeva EA, Zarytova VF, Ivanova EM, Ryte AS, Yurchenko LV, Vlassov VV. Mechanism of oligonucleotide uptake by cells: involvement of specific receptors? Proc Natl Acad Sci U S A 1989; 86: 6454-6458.

43. Stein CA, Tonkinson JL, Zhang LM, Yakubov L, Gervasoni J, Taub R, Rotenberg SA. Dynamics of the internalization of phosphodiester oligodeoxynucleotides in HL60 cells Biochemistry 1993; 32: 4855-4861.

44. Wang L, Prakash RK, Stein CA, Koehn RK, Ruffner DE. Progress in the delivery of therapeutic oligonucleotides: organ/cellular distribution and targeted delivery of oligonucleotides in vivo Antisense Nucleic Acid Drug Dev 2003; 13: 169-189. 293 45. Lebedeva I, Benimetskaya L, Stein CA, Vilenchik M. Cellular delivery of antisense oligonucleotides Eur J Pharm Biopharm 2000; 50: 101-119.

46. Lysik MA, Wu-Pong S. Innovations in oligonucleotide drug delivery J Pharm Sci 2003; 92: 1559-1573.

47. Marcusson EG, Bhat B, Manoharan M, Bennett CF, Dean NM. Phosphorothioate oligodeoxyribonucleotides dissociate from cationic lipids before entering the nucleus Nucleic Acids Res 1998; 26: 2016-2023.

48. Geary RS, Watanabe TA, Truong L, Freier S, Lesnik EA, Sioufi NB, Sasmor H, Manoharan M, Levin AA. Pharmacokinetic properties of 2'-O-(2-methoxyethyl)- modified oligonucleotide analogs in rats J Pharmacol Exp Ther 2001; 296: 890- 897.

49. Crooke ST, Graham MJ, Zuckerman JE, Brooks D, Conklin BS, Cummins LL, Greig MJ, Guinosso CJ, Kornbrust D, Manoharan M, Sasmor HM, Schleich T, Tivel KL, Griffey RH. Pharmacokinetic properties of several novel oligonucleotide analogs in mice J Pharmacol Exp Ther 1996; 277: 923-937.

50. Wartlick H, Spankuch-Schmitt B, Strebhardt K, Kreuter J, Langer K. Tumour cell delivery of antisense oligonuclceotides by human serum albumin nanoparticles J Control Release 2004; 96: 483-495.

51. Roskelley CD, Desprez PY, Bissell MJ. Extracellular matrix-dependent tissue- specific gene expression in mammary epithelial cells requires both physical and biochemical signal transduction Proc Natl Acad Sci U S A 1994; 91: 12378- 12382.

52. Juliano RL, Alahari S, Yoo H, Kole R, Cho M. Antisense pharmacodynamics: critical issues in the transport and delivery of antisense oligonucleotides Pharm Res 1999; 16: 494-502.

53. Butler M, Stecker K, Bennett CF. Cellular distribution of phosphorothioate oligodeoxynucleotides in normal rodent tissues Lab Invest 1997; 77: 379-388.

54. Brown DA, Kang SH, Gryaznov SM, DeDionisio L, Heidenreich O, Sullivan S, Xu X, Nerenberg MI. Effect of phosphorothioate modification of oligodeoxynucleotides on specific protein binding J Biol Chem 1994; 269: 26801-26805.

55. Cossum PA, Sasmor H, Dellinger D, Truong L, Cummins L, Owens SR, Markham PM, Shea JP, Crooke S. Disposition of the 14C-labeled phosphorothioate oligonucleotide ISIS 2105 after intravenous administration to rats J Pharmacol Exp Ther 1993; 267: 1181-1190.

294 56. Geary RS, Leeds JM, Fitchett J, Burckin T, Truong L, Spainhour C, Creek M, Levin AA. Pharmacokinetics and metabolism in mice of a phosphorothioate oligonucleotide antisense inhibitor of C-raf-1 kinase expression Drug Metab Dispos 1997; 25: 1272-1281.

57. Geary RS, Yu RZ, Watanabe T, Henry SP, Hardee GE, Chappell A, Matson J, Sasmor H, Cummins L, Levin AA. Pharmacokinetics of a tumor necrosis factor- alpha phosphorothioate 2'-O-(2-methoxyethyl) modified antisense oligonucleotide: comparison across species Drug Metab Dispos 2003; 31: 1419- 1428.

58. Yu RZ, Zhang H, Geary RS, Graham M, Masarjian L, Lemonidis K, Crooke R, Dean NM, Levin AA. Pharmacokinetics and pharmacodynamics of an antisense phosphorothioate oligonucleotide targeting Fas mRNA in mice J Pharmacol Exp Ther 2001; 296: 388-395.

59. Zhang R, Yan J, Shahinian H, Amin G, Lu Z, Liu T, Saag MS, Jiang Z, Temsamani J, Martin RR, . Pharmacokinetics of an anti-human immunodeficiency virus antisense oligodeoxynucleotide phosphorothioate (GEM 91) in HIV-infected subjects Clin Pharmacol Ther 1995; 58: 44-53.

60. Agrawal S, Temsamani J, Galbraith W, Tang J. Pharmacokinetics of antisense oligonucleotides Clin Pharmacokinet 1995; 28: 7-16.

61. Grindel JM, Musick TJ, Jiang Z, Roskey A, Agrawal S. Pharmacokinetics and metabolism of an oligodeoxynucleotide phosphorothioate (GEM91) in cynomolgus monkeys following intravenous infusion Antisense Nucleic Acid Drug Dev 1998; 8: 43-52.

62. Iversen P. In vivo studies with phosphorothioate oligonucleotides: pharmacokinetics prologue Anticancer Drug Des 1991; 6: 531-538.

63. Yu RZ, Geary RS, Leeds JM, Watanabe T, Moore M, Fitchett J, Matson J, Burckin T, Templin MV, Levin AA. Comparison of pharmacokinetics and tissue disposition of an antisense phosphorothioate oligonucleotide targeting human Ha- ras mRNA in mouse and monkey J Pharm Sci 2001; 90: 182-193.

64. Graham MJ, Crooke ST, Monteith DK, Cooper SR, Lemonidis KM, Stecker KK, Martin MJ, Crooke RM. In vivo distribution and metabolism of a phosphorothioate oligonucleotide within rat liver after intravenous administration J Pharmacol Exp Ther 1998; 286: 447-458.

65. Nolting A, DeLong RK, Fisher MH, Wickstrom E, Pollack GM, Juliano RL, Brouwer KL. Hepatic distribution and clearance of antisense oligonucleotides in the isolated perfused rat liver Pharm Res 1997; 14: 516-521.

295 66. Phillips JA, Craig SJ, Bayley D, Christian RA, Geary R, Nicklin PL. Pharmacokinetics, metabolism, and elimination of a 20-mer phosphorothioate oligodeoxynucleotide (CGP 69846A) after intravenous and subcutaneous administration Biochem Pharmacol 1997; 54: 657-668.

67. Rappaport J, Hanss B, Kopp JB, Copeland TD, Bruggeman LA, Coffman TM, Klotman PE. Transport of phosphorothioate oligonucleotides in kidney: implications for molecular therapy Kidney Int 1995; 47: 1462-1469.

68. Masarjian L, de Peyster A, Levin AA, Monteith DK. Distribution and excretion of a phosphorothioate oligonucleotide in rats with experimentally induced renal injury Oligonucleotides 2004; 14: 299-310.

69. Sawai K, Mahato RI, Oka Y, Takakura Y, Hashida M. Disposition of oligonucleotides in isolated perfused rat kidney: involvement of scavenger receptors in their renal uptake J Pharmacol Exp Ther 1996; 279: 284-290.

70. Bourque AJ, Cohen AS. Quantitative analysis of phosphorothioate oligonucleotides in biological fluids using fast anion-exchange chromatography J Chromatogr 1993; 617: 43-49.

71. Chen SH, Gallo JM. Use of capillary electrophoresis methods to characterize the pharmacokinetics of antisense drugs Electrophoresis 1998; 19: 2861-2869.

72. Gaus HJ, Owens SR, Winniman M, Cooper S, Cummins LL. On-line HPLC electrospray mass spectrometry of phosphorothioate oligonucleotide metabolites Anal Chem 1997; 69: 313-319.

73. Brown-Augsburger P, Yue XM, Lockridge JA, McSwiggen JA, Kamboj D, Hillgren KM. Development and validation of a sensitive, specific, and rapid hybridization-ELISA assay for determination of concentrations of a ribozyme in biological matrices J Pharm Biomed Anal 2004; 34: 129-139.

74. Yu RZ, Baker B, Chappell A, Geary RS, Cheung E, Levin AA. Development of an ultrasensitive noncompetitive hybridization-ligation enzyme-linked immunosorbent assay for the determination of phosphorothioate oligodeoxynucleotide in plasma Anal Biochem 2002; 304: 19-25.

75. Gilar M, Belenky A, Smisek DL, Bourque A, Cohen AS. Kinetics of phosphorothioate oligonucleotide metabolism in biological fluids Nucleic Acids Res 1997; 25: 3615-3620.

76. Stanley T.Crooke. Antisense Drug Technology. Richard S Geary, Rosie Z.Yu, Janet M.Leeds, Tanya A.Watanabe, Scott P.Henry, and Arthur A.Levin. Pharmacokinetic properties in animals. 139-141. 2001. Ref Type: Generic 296 77. Crooke RM, Graham MJ, Martin MJ, Lemonidis KM, Wyrzykiewiecz T, Cummins LL. Metabolism of antisense oligonucleotides in rat liver homogenates J Pharmacol Exp Ther 2000; 292: 140-149.

78. Dai G, Wei X, Liu Z, Liu S, Marcucci G, Chan KK. Characterization and quantification of Bcl-2 antisense G3139 and metabolites in plasma and urine by ion-pair reversed phase HPLC coupled with electrospray ion-trap mass spectrometry J Chromatogr B Analyt Technol Biomed Life Sci 2005; 825: 201- 213.

79. Griffey RH, Greig MJ, Gaus HJ, Liu K, Monteith D, Winniman M, Cummins LL. Characterization of oligonucleotide metabolism in vivo via liquid chromatography/electrospray tandem mass spectrometry with a quadrupole ion trap mass spectrometer J Mass Spectrom 1997; 32: 305-313.

80. Ni J, Pomerantz C, Rozenski J, Zhang Y, McCloskey JA. Interpretation of oligonucleotide mass spectra for determination of sequence using electrospray ionization and tandem mass spectrometry Anal Chem 1996; 68: 1989-1999.

81. Gilar M, Belenky A, Budman Y, Smisek DL, Cohen AS. Impact of 3'- exonuclease stereoselectivity on the kinetics of phosphorothioate oligonucleotide metabolism Antisense Nucleic Acid Drug Dev 1998; 8: 35-42.

82. Koziolkiewicz M, Wojcik M, Kobylanska A, Karwowski B, Rebowska B, Guga P, Stec WJ. Stability of stereoregular oligo(nucleoside phosphorothioate)s in human plasma: diastereoselectivity of plasma 3'-exonuclease Antisense Nucleic Acid Drug Dev 1997; 7: 43-48.

83. Cohen AS, Bourque AJ, Wang BH, Smisek DL, Belenky A. A nonradioisotope approach to study the in vivo metabolism of phosphorothioate oligonucleotides Antisense Nucleic Acid Drug Dev 1997; 7: 13-22.

84. Nicklin PL, Bayley D, Giddings J, Craig SJ, Cummins LL, Hastewell JG, Phillips JA. Pulmonary bioavailability of a phosphorothioate oligonucleotide (CGP 64128A): comparison with other delivery routes Pharm Res 1998; 15: 583-591.

85. Kamiya H, Akita H, Harashima H. Pharmacokinetic and pharmacodynamic considerations in gene therapy Drug Discov Today 2003; 8: 990-996.

86. Sprague JE, Worst TJ, Haynes K, Mosler CR, Nichols DE, Kane MD. The pharmacodynamic characterization of an antisense oligonucleotide against monoamine oxidase-B (MAO-B) in rat brain striatal tissue Cell Mol Neurobiol 2001; 21: 53-64.

87. Chi KN, Eisenhauer E, Fazli L, Jones EC, Goldenberg SL, Powers J, Tu D, Gleave ME. A phase I pharmacokinetic and pharmacodynamic study of OGX- 297 011, a 2'-methoxyethyl antisense oligonucleotide to clusterin, in patients with localized prostate cancer J Natl Cancer Inst 2005; 97: 1287-1296.

88. Eklund H, Uhlin U, Farnegardh M, Logan DT, Nordlund P. Structure and function of the radical enzyme ribonucleotide reductase Prog Biophys Mol Biol 2001; 77: 177-268.

89. Pratt WB, Ruddon RW, Ensminger WD, Maybaum J. The Anticancer Drugs. Antimetabolites. 97-99. 1994. Ref Type: Generic

90. Devlin T. Biochemistry with Clinical Correlations. Cory JG. Purine and pyrimidine nucleotide metabolism. [4th edition], 489-523. 1997. Ref Type: Generic

91. Fontecave M. Ribonucleotide reductases and radical reactions Cell Mol Life Sci 1998; 54: 684-695.

92. Nocentini G. Ribonucleotide reductase inhibitors: new strategies for cancer chemotherapy Crit Rev Oncol Hematol 1996; 22: 89-126.

93. Tanaka H, Arakawa H, Yamaguchi T, Shiraishi K, Fukuda S, Matsui K, Takei Y, Nakamura Y. A ribonucleotide reductase gene involved in a p53-dependent cell- cycle checkpoint for DNA damage Nature 2000; 404: 42-49.

94. Yamaguchi T, Matsuda K, Sagiya Y, Iwadate M, Fujino MA, Nakamura Y, Arakawa H. p53R2-dependent pathway for DNA synthesis in a p53-regulated cell cycle checkpoint Cancer Res 2001; 61: 8256-8262.

95. Lee Y, Vassilakos A, Feng N, Lam V, Xie H, Wang M, Jin H, Xiong K, Liu C, Wright J, Young A. GTI-2040, an antisense agent targeting the small subunit component (R2) of human ribonucleotide reductase, shows potent antitumor activity against a variety of tumors Cancer Res 2003; 63: 2802-2811.

96. Fan H, Villegas C, Huang A, Wright JA. The mammalian ribonucleotide reductase R2 component cooperates with a variety of oncogenes in mechanisms of cellular transformation Cancer Res 1998; 58: 1650-1653.

97. Zhou BS, Tsai P, Ker R, Tsai J, Ho R, Yu J, Shih J, Yen Y. Overexpression of transfected human ribonucleotide reductase M2 subunit in human cancer cells enhances their invasive potential Clin Exp Metastasis 1998; 16: 43-49.

98. Engstrom Y, Eriksson S, Jildevik I, Skog S, Thelander L, Tribukait B. Cell cycle- dependent expression of mammalian ribonucleotide reductase. Differential regulation of the two subunits J Biol Chem 1985; 260: 9114-9116.

298 99. Zhou BS, Hsu NY, Pan BC, Doroshow JH, Yen Y. Overexpression of ribonucleotide reductase in transfected human KB cells increases their resistance to hydroxyurea: M2 but not M1 is sufficient to increase resistance to hydroxyurea in transfected cells Cancer Res 1995; 55: 1328-1333.

100. Duxbury MS, Ito H, Zinner MJ, Ashley SW, Whang EE. RNA interference targeting the M2 subunit of ribonucleotide reductase enhances pancreatic adenocarcinoma chemosensitivity to gemcitabine Oncogene 2004; 23: 1539- 1548.

101. Lorus Therapeutics. GTI-2040 Investigator Brochure. 1999. Ref Type: Generic

102. Noll BO, McCluskie MJ, Sniatala T, Lohner A, Yuill S, Krieg AM, Schetter C, Davis HL, Uhlmann E. Biodistribution and metabolism of immunostimulatory oligodeoxynucleotide CPG 7909 in mouse and rat tissues following subcutaneous administration Biochem Pharmacol 2005; 69: 981-991.

103. Lowenberg B GJTM. Hematology. Acute myeloid leukemia and acute promyelocytic leukemia. 82-101. 2003. Am Soc Hematol Educ Program. Ref Type: Generic

104. Stone RM. Treatment of acute myeloid leukemia: state-of-the-art and future directions Semin Hematol 2002; 39: 4-10.

105. Bullinger L, Valk PJ. Gene expression profiling in acute myeloid leukemia J Clin Oncol 2005; 23: 6296-6305.

106. Kimby E, Nygren P, Glimelius B. A systematic overview of chemotherapy effects in acute myeloid leukaemia Acta Oncol 2001; 40: 231-252.

107. Bloomfield CD, Lawrence D, Byrd JC, Carroll A, Pettenati MJ, Tantravahi R, Patil SR, Davey FR, Berg DT, Schiffer CA, Arthur DC, Mayer RJ. Frequency of prolonged remission duration after high-dose cytarabine intensification in acute myeloid leukemia varies by cytogenetic subtype Cancer Res 1998; 58: 4173- 4179.

108. Stone RM. The difficult problem of acute myeloid leukemia in the older adult CA Cancer J Clin 2002; 52: 363-371.

109. Cros E, Jordheim L, Dumontet C, Galmarini CM. Problems related to resistance to cytarabine in acute myeloid leukemia Leuk Lymphoma 2004; 45: 1123-1132.

110. Galmarini CM, Mackey JR, Dumontet C. Nucleoside analogues: mechanisms of drug resistance and reversal strategies Leukemia 2001; 15: 875-890.

299 111. Grant S. Ara-C: cellular and molecular pharmacology Adv Cancer Res 1998; 72: 197-233.

112. Hamada A, Kawaguchi T, Nakano M. Clinical pharmacokinetics of cytarabine formulations Clin Pharmacokinet 2002; 41: 705-718.

113. Burk M, Heyll A, Arning M, Volmer M, Fartash K, Schneider W. Pharmacokinetics of high-dose cytarabine and its deamination product--a reappraisal Leuk Lymphoma 1997; 27: 321-327.

114. Bolwell BJ, Cassileth PA, Gale RP. High dose cytarabine: a review Leukemia 1988; 2: 253-260.

115. Tamm I, Dorken B, Hartmann G. Antisense therapy in oncology: new hope for an old idea? Lancet 2001; 358: 489-497.

116. Bres JC, Morvan F, Lefebvre I, Vasseur JJ, Pompon A, Imbach JL. Kinetics study of the biotransformation of an oligonucleotide prodrug in cells extract by matrix- assisted laser desorption-ionization time-of-flight mass spectrometry J Chromatogr B Biomed Sci Appl 2001; 753: 123-130.

117. Raynaud FI, Orr RM, Goddard PM, Lacey HA, Lancashire H, Judson IR, Beck T, Bryan B, Cotter FE. Pharmacokinetics of G3139, a phosphorothioate oligodeoxynucleotide antisense to bcl-2, after intravenous administration or continuous subcutaneous infusion to mice J Pharmacol Exp Ther 1997; 281: 420- 427.

118. Yu RZ, Geary RS, Levin AA. Application of novel quantitative bioanalytical methods for pharmacokinetic and pharmacokinetic/pharmacodynamic assessments of antisense oligonucleotides Curr Opin Drug Discov Devel 2004; 7: 195-203.

119. de Serres M, McNulty MJ, Christensen L, Zon G, Findlay JW. Development of a novel scintillation proximity competitive hybridization assay for the determination of phosphorothioate antisense oligonucleotide plasma concentrations in a toxicokinetic study Anal Biochem 1996; 233: 228-233.

120. Deverre JR, Boutet V, Boquet D, Ezan E, Grassi J, Grognet JM. A competitive enzyme hybridization assay for plasma determination of phosphodiester and phosphorothioate antisense oligonucleotides Nucleic Acids Res 1997; 25: 3584- 3589.

121. Efler SM, Zhang L, Noll BO, Uhlmann E, Davis HL. Quantification of oligodeoxynucleotides in human plasma with a novel hybridization assay offers greatly enhanced sensitivity over capillary gel electrophoresis Oligonucleotides 2005; 15: 119-131. 300 122. Coppelli FM, Grandis JR. Oligonucleotides as anticancer agents: from the benchside to the clinic and beyond Curr Pharm Des 2005; 11: 2825-2840.

123. Branch AD. Antisense drug discovery: can cell-free screens speed the process? Antisense Nucleic Acid Drug Dev 1998; 8: 249-254.

124. Sohail M, Hochegger H, Klotzbucher A, Guellec RL, Hunt T, Southern EM. Antisense oligonucleotides selected by hybridisation to scanning arrays are effective reagents in vivo Nucleic Acids Res 2001; 29: 2041-2051.

125. Dickson G, Hill V, Graham IR. Screening for antisense modulation of dystrophin pre-mRNA splicing Neuromuscul Disord 2002; 12 Suppl 1: S67-S70.

126. Sohail M, Southern EM. Selecting optimal antisense reagents Adv Drug Deliv Rev 2000; 44: 23-34.

127. Sohail M, Souothern EM. Oligonucleotide scanning arrays: application to high- throughput screening for effective antisense reagents and the study of nucleic acid interactions Adv Biochem Eng Biotechnol 2002; 77: 43-56.

128. Maksimenko AV, Gottikh MB, Helin V, Shabarova ZA, Malvy C. Physico- chemical and biological properties of antisense phosphodiester oligonucleotides with various secondary structures Nucleosides Nucleotides 1999; 18: 2071-2091.

129. Wang W, Chen HJ, Warshofsky M, Schwartz A, Stein CA, Rabbani LE. Effects of S-dC28 on vascular smooth muscle cell adhesion and plasminogen activator production Antisense Nucleic Acid Drug Dev 1997; 7: 101-107.

130. Tonkinson JL, Stein CA. Patterns of intracellular compartmentalization, trafficking and acidification of 5'-fluorescein labeled phosphodiester and phosphorothioate oligodeoxynucleotides in HL60 cells Nucleic Acids Res 1994; 22: 4268-4275.

131. Jang SH, Wientjes MG, Au JL. Interdependent effect of P-glycoprotein-mediated drug efflux and intracellular drug binding on intracellular paclitaxel pharmacokinetics: application of computational modeling J Pharmacol Exp Ther 2003; 304: 773-780.

132. Jang SH, Wientjes MG, Au JL. Kinetics of P-glycoprotein-mediated efflux of paclitaxel J Pharmacol Exp Ther 2001; 298: 1236-1242.

133. Wei X, Dai G, Marcucci G, Liu Z, Hoyt D, Chan KK. A specific picomolar hybridizatio-based ELISA assay for the determination of phosphorothioate oligonucleotides in plasma and cellular matrices Pharm Res 2006.

301 134. Nakai D, Seita T, Terasaki T, Iwasa S, Shoji Y, Mizushima Y, Sugiyama Y. Cellular uptake mechanism for oligonucleotides: involvement of endocytosis in the uptake of phosphodiester oligonucleotides by a human colorectal adenocarcinoma cell line, HCT-15 J Pharmacol Exp Ther 1996; 278: 1362-1372.

135. Wu-Pong S, Weiss TL, Hunt CA. Antisense c-myc oligodeoxyribonucleotide cellular uptake Pharm Res 1992; 9: 1010-1017.

136. Wu-Pong S, Weiss TL, Hunt CA. Antisense c-myc oligonucleotide cellular uptake and activity Antisense Res Dev 1994; 4: 155-163.

137. Takakura Y, Oka Y, Hashida M. Cellular uptake properties of oligonucleotides in LLC-PK1 renal epithelial cells Antisense Nucleic Acid Drug Dev 1998; 8: 67-73.

138. de Diesbach P, Berens C, N'Kuli F, Monsigny M, Sonveaux E, Wattiez R, Courtoy PJ. Identification, purification and partial characterisation of an oligonucleotide receptor in membranes of HepG2 cells Nucleic Acids Res 2000; 28: 868-874.

139. Dai G. Pharmacokinetics, Pharmacodynamics and Metabolism of Bcl-2 Antisense Phosphorothioate Oligonucleotide G3139 (Genasense). 95-114. 2005. Ref Type: Thesis/Dissertation

140. Davis AJ, Gelmon KA, Siu LL, Moore MJ, Britten CD, Mistry N, Klamut H, D'Aloisio S, MacLean M, Wainman N, Ayers D, Firby P, Besterman JM, Reid GK, Eisenhauer EA. Phase I and pharmacologic study of the human DNA methyltransferase antisense oligodeoxynucleotide MG98 given as a 21-day continuous infusion every 4 weeks Invest New Drugs 2003; 21: 85-97.

141. Glover JM, Leeds JM, Mant TG, Amin D, Kisner DL, Zuckerman JE, Geary RS, Levin AA, Shanahan WR, Jr. Phase I safety and pharmacokinetic profile of an intercellular adhesion molecule-1 antisense oligodeoxynucleotide (ISIS 2302) J Pharmacol Exp Ther 1997; 282: 1173-1180.

142. Tang X, Wu D, Xue Y, Zhu M, Lu D, Ruan C. [The morphological, immunophenotypical and cytogenetic characteristics study of blast crisis in chronic myeloid leukemia] Zhonghua Nei Ke Za Zhi 2002; 41: 685-687.

143. Beal SL, Sheiner LB. Part I-VIII. In: NONMEM Project Group C255, ed. NONMEM Users Guides, University of California at San Francisco, 1988.

144. Aarons L. Population pharmacokinetics: theory and practice Br J Clin Pharmacol 1991; 32: 669-670.

145. Jonsson N, Karlsson M. Xpose 3.1 User's Manual. Department of Pharmacy, Uppsala University, Sweden, 1998.

302 146. Jonsson EN, Karlsson MO. Xpose--an S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM Comput Methods Programs Biomed 1999; 58: 51-64.

147. Williams PJ, Ette EI. The role of population pharmacokinetics in drug development in light of the Food and Drug Administration's 'Guidance for Industry: population pharmacokinetics' Clin Pharmacokinet 2000; 39: 385-395.

148. Guidance for industry on Population Pharmacokinetics; availability. Food and Drug Administration, HHS. Notice Fed Regist 1999; 64: 6663-6664.

149. Sun H, Fadiran EO, Jones CD, Lesko L, Huang SM, Higgins K, Hu C, Machado S, Maldonado S, Williams R, Hossain M, Ette EI. Population pharmacokinetics. A regulatory perspective Clin Pharmacokinet 1999; 37: 41-58.

150. Jusko WJ, Ko HC. Physiologic indirect response models characterize diverse types of pharmacodynamic effects Clin Pharmacol Ther 1994; 56: 406-419.

151. Sharma A, Jusko WJ. Characteristics of indirect pharmacodynamic models and applications to clinical drug responses Br J Clin Pharmacol 1998; 45: 229-239.

152. Sharma A, Jusko WJ. Characterization of four basic models of indirect pharmacodynamic responses J Pharmacokinet Biopharm 1996; 24: 611-635.

153. Gentilini A, Rossoni-Gerosa M, Frei CW, Wymann R, Morari M, Zbinden AM, Schnider TW. Modeling and closed-loop control of hypnosis by means of bispectral index (BIS) with isoflurane IEEE Trans Biomed Eng 2001; 48: 874- 889.

154. Nava-Ocampo AA, Shafer SL, Velazquez-Armenta Y, Ruiz-Velazco S, Toni B. Mathematical analysis of a pharmacodynamic model without plasma concentrations to extend its applicability Med Hypotheses 2003; 60: 453-457.

155. Visser SA, Smulders CJ, Reijers BP, van der Graaf PH, Peletier LA, Danhof M. Mechanism-based pharmacokinetic-pharmacodynamic modeling of concentration-dependent hysteresis and biphasic electroencephalogram effects of alphaxalone in rats J Pharmacol Exp Ther 2002; 302: 1158-1167.

156. Fennell DA, Cotter FE. A dynamical systems model to simulate the perturbation kinetics of gene expression by antisense oligonucleotides J Theor Biol 2001; 209: 103-112.

157. Roth CM. Molecular and cellular barriers limiting the effectiveness of antisense oligonucleotides Biophys J 2005; 89: 2286-2295.

303 158. Alberts B, Bray D, Lewis J, Raff M, Robert K, Watson JD. The control of cell division. In: Molecular biology of the cell, Garland Publishing Inc., 1998: 610- 620.

159. Look AT, Melvin SL, Williams DL, Brodeur GM, Dahl GV, Kalwinsky DK, Murphy SB, Mauer AM. Aneuploidy and percentage of S-phase cells determined by flow cytometry correlate with cell phenotype in childhood acute leukemia Blood 1982; 60: 959-967.

160. Baer MR, Bernstein SH, Brunetto VL, Heinonen K, Mrozek K, Swann VL, Minderman H, Block AW, Pixley LA, Christiansen NP, Fay JW, Barcos M, Rustum Y, Herzig GP, Bloomfield CD. Biological effects of recombinant human granulocyte colony-stimulating factor in patients with untreated acute myeloid leukemia Blood 1996; 87: 1484-1494.

161. Amara FM, Chen FY, Wright JA. A novel transforming growth factor-beta 1 responsive cytoplasmic trans-acting factor binds selectively to the 3'-untranslated region of mammalian ribonucleotide reductase R2 mRNA: role in message stability Nucleic Acids Res 1993; 21: 4803-4809.

162. Kurreck J, Wyszko E, Gillen C, Erdmann VA. Design of antisense oligonucleotides stabilized by locked nucleic acids Nucleic Acids Res 2002; 30: 1911-1918.

163. Sazani P, Astriab-Fischer A, Kole R. Effects of base modifications on antisense properties of 2'-O-methoxyethyl and PNA oligonucleotides Antisense Nucleic Acid Drug Dev 2003; 13: 119-128.

164. Agrawal S, Temsamani J, Tang JY. Pharmacokinetics, biodistribution, and stability of oligodeoxynucleotide phosphorothioates in mice Proc Natl Acad Sci U S A 1991; 88: 7595-7599.

165. Leeds JM, Graham MJ, Truong L, Cummins LL. Quantitation of phosphorothioate oligonucleotides in human plasma Anal Biochem 1996; 235: 36-43.

166. Gilar M. Analysis and purification of synthetic oligonucleotides by reversed- phase high-performance liquid chromatography with photodiode array and mass spectrometry detection Anal Biochem 2001; 298: 196-206.

167. Lichtenwalter KG, Apffel A, Bai J, Chakel JA, Dai Y, Hahnenberger KM, Li L, Hancock WS. Approaches to functional genomics: potential of matrix-assisted laser desorption ionization--time of flight mass spectrometry combined with separation methods for the analysis of DNA in biological samples J Chromatogr B Biomed Sci Appl 2000; 745: 231-241.

304 168. Gilar M, Bouvier ESP. Purification of crude DNA oligonucleotides by solid- phase extraction and reversed-phase high-performance liquid chromatography J Chromatogr A 2000; 890: 167-177.

169. Murphy AT, Brown-Augsburger P, Yu RZ, Geary RS, Thibodeaux S, Ackermann BL. Development of an ion-pair reverse-phase liquid chromatographic/tandem mass spectrometry method for the determination of an 18-mer phosphorothioate oligonucleotide in mouse liver tissue Eur J Mass Spectrom (Chichester, Eng) 2005; 11: 209-215.

170. Geary RS, Leeds JM, Henry SP, Monteith DK, Levin AA. Antisense oligonucleotide inhibitors for the treatment of cancer: 1. Pharmacokinetic properties of phosphorothioate oligodeoxynucleotides Anticancer Drug Des 1997; 12: 383-393.

171. Temsamani J, Tang JY, Padmapriya A, Kubert M, Agrawal S. Pharmacokinetics, biodistribution, and stability of capped oligodeoxynucleotide phosphorothioates in mice Antisense Res Dev 1993; 3: 277-284.

172. Temsamani J, Roskey A, Chaix C, Agrawal S. In vivo metabolic profile of a phosphorothioate oligodeoxyribonucleotide Antisense Nucleic Acid Drug Dev 1997; 7: 159-165.

173. Gilar M, Belenky A, Budman Y, Smisek DL, Cohen AS. Study of phosphorothioate-modified oligonucleotide resistance to 3'-exonuclease using capillary electrophoresis J Chromatogr B Biomed Sci Appl 1998; 714: 13-20.

174. Graham MJ, Crooke ST, Lemonidis KM, Gaus HJ, Templin MV, Crooke RM. Hepatic distribution of a phosphorothioate oligodeoxynucleotide within rodents following intravenous administration Biochem Pharmacol 2001; 62: 297-306.

175. Crooke RM, Graham MJ, Cooke ME, Crooke ST. In vitro pharmacokinetics of phosphorothioate antisense oligonucleotides J Pharmacol Exp Ther 1995; 275: 462-473.

176. Liu Z, Minkler PE, Lin D, Sayre LM. Derivatization of amino acids with N,N- dimethyl-2,4-dinitro-5-fluorobenzylamine for liquid chromatography/electrospray ionization mass spectrometry Rapid Commun Mass Spectrom 2004; 18: 1059- 1065.

177. Liu Z, Floss HG, Cassady JM, Chan KK. Metabolism studies of the anti-tumor agent maytansine and its analog ansamitocin P-3 using liquid chromatography/tandem mass spectrometry J Mass Spectrom 2005; 40: 389-399.

305 178. Rozenski J, McCloskey JA. SOS: a simple interactive program for ab initio oligonucleotide sequencing by mass spectrometry J Am Soc Mass Spectrom 2002; 13: 200-203.

179. McLuckey SA, Goeringer DE, Glish GL. Collisional activation with random noise in ion trap mass spectrometry Anal Chem 1992; 64: 1455-1460.

180. McLuckey SA, Glish GL, Van Berkel GJ. Charge determination of product ions formed from collision-induced dissociation of multiply protonated molecules via ion/molecule reactions Anal Chem 1991; 63: 1971-1978.

181. Apffel A, Chakel JA, Hancock WS, Souders C, M'Timkulu T, Pungor E Jr. Application of multidimensional affinity high-performance liquid chromatography and electrospray ionization liquid chromatography-mass spectrometry to the characterization of glycosylation in single-chain plasminogen activator. Initial results J Chromatogr A 1996; 750: 35-42.

182. Huber CG, Oefner PJ, Bonn GK. High-resolution liquid chromatography of oligonucleotides on nonporous alkylated styrene-divinylbenzene copolymers Anal Biochem 1993; 212: 351-358.

183. R.Haefele, D.Gjerde. Quality control and purification of oligonucleotides on teh WAVE(R) nucleic acid fragment analysis system. In: Transgenomics, Application note 103., 2000.

184. Geary RS, Yu RZ, Levin AA. Pharmacokinetics of phosphorothioate antisense oligodeoxynucleotides Curr Opin Investig Drugs 2001; 2: 562-573.

185. Grieg MJ, Gaus H, Cummins L, Sasmor H, Griffey RH. Measurement of macromolecular binding using electrospray mass spectrometry. Determination of disassociation constants for oligonucleotide-serum albumin complexes. J Am Chem Soc 1995; 10765-10766.

186. Obach RS. Nonspecific binding to microsomes: impact on scale-up of in vitro intrinsic clearance to hepatic clearance as assessed through examination of warfarin, imipramine, and propranolol Drug Metab Dispos 1997; 25: 1359-1369.

187. Tang C, Lin Y, Rodrigues AD, Lin JH. Effect of albumin on phenytoin and tolbutamide metabolism in human liver microsomes: an impact more than protein binding Drug Metab Dispos 2002; 30: 648-654.

188. McLure JA, Miners JO, Birkett DJ. Nonspecific binding of drugs to human liver microsomes Br J Clin Pharmacol 2000; 49: 453-461.

306 189. Kuhnz W, Gieschen H. Predicting the oral bioavailability of 19-nortestosterone progestins in vivo from their metabolic stability in human liver microsomal preparations in vitro Drug Metab Dispos 1998; 26: 1120-1127.

190. Houston JB, Kenworthy KE. In vitro-in vivo scaling of CYP kinetic data not consistent with the classical Michaelis-Menten model Drug Metab Dispos 2000; 28: 246-254.

191. Houston JB, Galetin A. Modelling atypical CYP3A4 kinetics: principles and pragmatism Arch Biochem Biophys 2005; 433: 351-360.

192. Witherow LE, Houston JB. Sigmoidal kinetics of CYP3A substrates: an approach for scaling dextromethorphan metabolism in hepatic microsomes and isolated hepatocytes to predict in vivo clearance in rat J Pharmacol Exp Ther 1999; 290: 58-65.

193. Cornish-Bowden Athel. Introduction to enzyme kinetics. In: Fundamental of enzyme kinetics, London : Portland, 1995: 16-37.

194. Oda Y, Kharasch ED. Metabolism of levo-alpha-Acetylmethadol (LAAM) by human liver cytochrome P450: involvement of CYP3A4 characterized by atypical kinetics with two binding sites J Pharmacol Exp Ther 2001; 297: 410-422.

195. Kenworthy KE, Clarke SE, Andrews J, Houston JB. Multisite kinetic models for CYP3A4: simultaneous activation and inhibition of diazepam and testosterone metabolism Drug Metab Dispos 2001; 29: 1644-1651.

196. Harlow GR, Halpert JR. Analysis of human cytochrome P450 3A4 cooperativity: construction and characterization of a site-directed mutant that displays hyperbolic steroid hydroxylation kinetics Proc Natl Acad Sci U S A 1998; 95: 6636-6641.

197. Emoto C, Yamazaki H, Iketaki H, Yamasaki S, Satoh T, Shimizu R, Suzuki S, Shimada N, Nakajima M, Yokoi T. Cooperativity of alpha-naphthoflavone in cytochrome P450 3A-dependent drug oxidation activities in hepatic and intestinal microsomes from mouse and human Xenobiotica 2001; 31: 265-275.

198. Venkatakrishnan K, von Moltke LL, Obach RS, Greenblatt DJ. Microsomal binding of amitriptyline: effect on estimation of enzyme kinetic parameters in vitro J Pharmacol Exp Ther 2000; 293: 343-350.

199. Isoherranen N, Kunze KL, Allen KE, Nelson WL, Thummel KE. Role of itraconazole metabolites in CYP3A4 inhibition Drug Metab Dispos 2004; 32: 1121-1131.

307 200. Yu RZ, Geary RS, Monteith DK, Matson J, Truong L, Fitchett J, Levin AA. Tissue disposition of 2'-O-(2-methoxy) ethyl modified antisense oligonucleotides in monkeys J Pharm Sci 2004; 93: 48-59.

201. Houston JB. Utility of in vitro drug metabolism data in predicting in vivo metabolic clearance Biochem Pharmacol 1994; 47: 1469-1479.

202. Houston JB, Carlile DJ. Prediction of hepatic clearance from microsomes, hepatocytes, and liver slices Drug Metab Rev 1997; 29: 891-922.

203. Branden LJ, Christensson B, Smith CI. In vivo nuclear delivery of oligonucleotides via hybridizing bifunctional peptides Gene Ther 2001; 8: 84-87.

204. Shiraishi T, Hamzavi R, Nielsen PE. Targeted delivery of plasmid DNA into the nucleus of cells via nuclear localization signal peptide conjugated to DNA intercalating bis- and trisacridines Bioconjug Chem 2005; 16: 1112-1116.

205. van Pelt K, de Haan G, Vellenga E, Daenen SM. Administration of low-dose cytarabine results in immediate S-phase arrest and subsequent activation of cell cycling in murine stem cells Exp Hematol 2005; 33: 226-231.

308