NON-NATURAL NUCLEOTIDES AS
MODULATORS OF ATPASES
BY
KEVIN ENG
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy
Dissertation Advisor: Dr. Anthony J. Berdis
Department of Pharmacology
CASE WESTERN RESERVE UNIVERSITY
May, 2010
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of
______
candidate for the ______degree *.
(signed)______(chair of the committee)
______
______
______
______
______
(date) ______
*We also certify that written approval has been obtained for any proprietary material contained therein.
This work is dedicated to my previous mentors, Silvana Gaetani, Danielle Piomelli, and
Steve Barnett. You are the ones responsible for bringing me this far. Table of Contents
Signature Sheet i
Dedication ii
Table of Contents 1
List of Figures 4
List of Tables 7
List of Abbreviations 8
Acknowledgements 9
Abstract 11
Chapter 1: Introduction
1.1 Overview of ATP and ATPases 13
The chemical properties of ATP and its role in biological systems
Structural analysis of ATP binding pocket
Existing drugs that target the ATP binding site
1.2 Non-natural Nucleotides as Therapeutic Agents 19
Non-natural nucleotides and translesion DNA synthesis
Clinical importance of nucleoside analogs
1.3 Model Systems for Evaluating the Influence of
1 Non-Natural Nucleotides on ATP-dependent
Processes 22
Sliding clamp loader ATPase
P-glycoprotein
1.4 Statement of Purpose 31
Figures 32
Tables 68
Chapter 2: Selective inhibition of DNA replicase assembly by
a non-natural nucleotide: Exploiting the structural
diversity of ATP-binding sites
2.1 Abstract 69
2.2 Introduction 71
2.3 Materials and Methods 72
2.4 Results and Discussion 76
Figures 89
Tables 112
Chapter 3: A Novel Non-Natural Nucleoside that Influences P-
glycoprotein Activity and Mediates Drug Resistance
3.1 Abstract 115
2 3.2 Introduction 116
3.3 Materials and Methods 118
3.4 Results 121
3.5 Discussion 128
Figures 134
Tables 150
Chapter 4: Conclusions and future directions
4.1 Overview 154
4.2 Chapter 2 Conclusions 155
4.3 Chapter 2 Future directions 156
4.4 Chapter 3 Conclusions 163
4.5 Chapter 3 Future directions 164
Figures 170
References 179
3 List of Figures
Figure 1.1 Structures of ATP and dATP 33
Figure 1.2 Schematic of a molecular motor and actin assembly 35
Figure 1.3 Representation of a typical ATPase P-loop 37
Figure 1.4 Chemical mechanism of ATP hydrolysis. 39
Figure 1.5 Structures of ATP competitive inhibitors 41
Figure 1.6 Imatinib and cAbl kinase 43
Figure 1.7 Structures and electron density maps of non-natural nucleobases. 45
Figure 1.8 Schematic of translesion DNA synthesis 47
Figure 1.9 Structures of nucleoside analogs used in cancer chemotherapy 49
Figure 1.10 Structures of nucleoside analogs used in anti-viral chemotherapy 51
Figure 1.11 Schematic of replicase formation 53
Figure 1.12 Subunit compositions of clamp loaders from different species 55
Figure 1.13 Sample of the structural diversity of P-gp substrates 57
Figure 1.14 Model of P-glycoprotein transport 59
Figure 1.15 Structure of P-glycoprotein 61
Figure 1.16 ATP switch model 63
Figure 1.17 The occluded nucleotide model 65
Figure 1.18 Structures of the most recent P-glycoprotein inhibitors 67
Figure 2.1 Natural and non-natural nucleobases used in this study 90
Figure 2.2 Hydrolysis of nucleotide substrates by gp44/62 and the effect of
r5-NITP and d5-NITP on gp44/62 ATPase activity 92
Figure 2.3 Effects of d5-NITP on T4 replicase formation and gp45 opening 94
4 Figure 2.4 Plot of K i values for various non-natural nucleotides as a function of
their respective nucleobase size 96
Figure 2.5 Molecular modeling of the active sites of gp44/62 and γ-complex. 98
Figure 2.6 Alignment of the ATP-binding region of the bacteriophage
T4 gp44/62, Saccharomyces cerevisiae RFC, and Escherichia coli γ-complex 100
Figure 2.7 Inhibition of T4 plaque formation by d5-NI 102
Figure 2.8 d5-NI is not bactericidal or bacteriostatic against E. coli . 104
Figure 2.9 Kinetic model for the effects of a competitive inhibitor on the
concurrent activity of two independent enzymes that possess different K m
values for a common substrate and different K i values for a common
competitive inhibitor 106
Figure 2.10 Comparison of the DNA replication mechanisms of E.coli and T4
bacteriophage 111
Figure 3.1 Structures and electrostatic potential models of non-natural
nucleosides used in this study 135
Figure 3.2 Stimulation of P-gp ATPase activity by non-natural nucleosides 137
Figure 3.3 Stimulation of P-gp ATPase activity by increasing concentrations of
calcein-AM in the presence and absence of d5-CHI or CsA 139
Figure 3.4 Modulation of drug resistance by d5-CHI in KB-V1 cells 141
Figure 3.5 Effects of d5-CHI, d5-CEI, and d5-PhI on the cytotoxicity of
paclitaxel and vinblastine in KB-V1 cells 143
Figure 3.6 Effects of d5-CHI on cell viability. 145
Figure 3.7 Comparison of the structures of P-gp interacting compounds 147
Figure 3.8 Trend in ATPase V max /K m ratios and MDR modulation 149
5 Figure 4.1 Schematic of chapter 2 conclusions 171
Figure 4.2 Kinase profiling of non-natural nucleo(s)tides 173
Figure 4.3 Chemical structures of ligands for the 5HT2, histamine H2,
adenosine, and melatonin receptors. 175
Figure 4.4 Schematic of model correlating changes in p-glycoprotein ATPase
catalytic efficiency with changes in p-glycoprotein mediated drug resistance 177
6 List of Tables
Table 1.1 Subunit compositions and functions of clamp loaders from various
species 68
Table 2.1 Summary of inhibition constants for natural and non-natural
nucleotides against the ATP-dependent clamp loaders from bacteriophage
T4 (gp44/62) and Escherichia coli (γ-complex) 112
Table 2.2 Summary of inhibition constants for various non-natural nucleotides
against wild-type, R175K, and R175L mutants of gp44/62 114
Table 3.1 Kinetic parameters for the stimulation in ATPase activity by drug
substrates 150
Table 3.2 Permeability coefficients of substrates across MDCK and MDCK-MDR
monolayers 151
Table 3.3 Effects of d5-CHI on the catalytic efficiency of drug-stimulated P-gp
ATPase activity 152
Table 3.4 Effects of d5-CHI on the cytotoxicity of vinblastine (VBL),
doxorubicin (DOX), colchicine (COLC), and paclitaxel (TAX) in KB-V1
cells and parental KB-3-1 cells 153
Table 4.1 ABC transporters involved in multidrug resistance and their known
substrates 178
7 List of Abbreviations
d4-NITP, 4-nitro-indolyl-2’-deoxyriboside triphosphate; d5-NITP, 5-nitro-indolyl-2’- deoxyriboside triphosphate; r5-NITP, 5-nitro-indolyl-2’-ribose triphosphate; d5-NI, 5-nitro- indolyl-2’-deoxyriboside; d6-NITP, 6-nitro-indolyl-2’-deoxyriboside triphosphate; d5-EtITP,
5-ethyl-indolyl-2’-deoxyriboside triphosphate; d5-EyITP, 5-ethylene-indolyl-2’-deoxyriboside triphosphate ; d5-EyInd, 5-ethylene-indolyl-2’-deoxyriboside; d5-FITP, 5-fluoro-indolyl-2’- deoxyriboside triphosphate; d5-FI, 5-fluoro-indolyl-2’-deoxyriboside; dITP, indolyl-2’- deoxyriboside triphosphate; Ind, indolyl-2’-deoxyriboside; d5-CHITP, 5-cyclohexyl-indolyl-
2’-deoxyriboside triphosphate; d5-AITP, 5-amino-indolyl-2’-deoxyriboside triphosphate; d5-
CEITP, 5-cyclohexene-indolyl-2’-deoxyriboside triphosphate; d5-CITP, 5-carboxylate- indolyl-2’-deoxyriboside triphosphate; d5-PhITP, 5-phenyl-indolyl-2’-deoxyriboside triphosphate; AMP-PNP, 5'-adenylyl-beta,gamma-imidodiphosphate; gp44/62, bacteriophage T4 sliding clamp loader; gp45, bacteriophage T4 sliding clamp; gp43 exo -, exonuclease-deficient mutant of bacteriophage T4 DNA polymerase; γ complex, E. coli sliding clamp loader; β clamp, E. coli sliding clamp. d5-CEI, 5-cyclohexene indolyl 2’- deoxyribose; d5-CHI, 5-cyclohexyl indolyl 2’-deoxyribose; d5-PhI, 5-phenyl indolyl 2’- deoxyribose; AP→BL, apical to basolateral; BL→AP, basolateral to apical; CsA, cyclosporine A; MDR, multi drug resistance; MDCK, Madin Darby Canine Kidney; MTT, 3-
(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; NIMH, National Institutes of
Mental Health; PDSP, Psychoactive Drug Screening Program; Peff , permeability coefficient;
P-gp, P-glycoprotein; SAR, structure activity relationship.
8 Acknowledgements
This work would not be possible without the contributions of several individuals.
First and foremost, my advisor Dr. Anthony Berdis for providing support and guidance throughout my training. I would also like to thank my committee members Dr. Chris
Dealwis, Dr. Shigemi Matsuyama, and Dr. Robert Bonomo whose insights were invaluable to my professional development and Dr. John Mieyal for his guidance. Special thanks to
Edward Motea, Xuemei Zhang, and Sandra Craig for their time and effort in providing non- natural nucleotides of the highest quality used throughout all my studies. I am also grateful to Dr. Irene Lee for teaching me the principles of protein purification. I would also like to thank Dr. Babho Devadoss for providing me with the invaluable guidance of a senior graduate. I am also indebted to my undergraduate student, Raphael Bendriem, for his fresh insights to the P-glycoprotein project which helped shape the project as it is today. Special thanks to members of the Berdis lab (past and present); Jackelyn Golden, Andrea Ramos,
Asim Sherrif, Robert Bowers, Dave McCutcheon, and Kevin Costanzo for making every
workday more enjoyable. Several researchers have provided me with their advice and
expertise throughout my studies: Philip Kiser, David Lodowski, Jim Fairman, Jay
Prendergast, Seunghwan Lim, Vivian Gama, Jose Gomez, Reema Wahdan, Andrea
Moomaw, and Monica Montano.
During my time here at Case, I have been truly blessed with many great friends: Eric
Lam, Vaibhav Pathak, Chohee Yun, Charlotte Chung, Alejandro Colozo, Paul Park,
Debarshi Mustafi, Rod Nibbe, Dasha Hajkova-Leary, Ndiya Ogba, the Marksz family, Stasha
Razack, Jared Klaus, and Joe Dea.
9 To my family, thank you for always believing in me and supporting me with your constant encouragement.
10 Non-Natural Nucleotides as Modulators of ATPases
Abstract
By
KEVIN ENG
ATP is the chemical medium of energy transfer for all living organisms. A common assumption is that the active site of all ATP-binding proteins share similar architecture. This study challenges this assumption by using non-natural nucleotides as chemical probes for exploring the diverse architecture of the ATP binding site. Two systems were used to explore this principle; first was the sliding clamp loader ATPase which is responsible for processive DNA replication and second was the ATP binding cassette transporter P- glycoprotein (P-gp) that is involved in multidrug resistance (MDR) phenotype. A library of non-natural nucleotides was used to probe both the T4 bacteriophage and E.coli clamp loaders. One compound, 5-nitroindolyl 2’-deoxyribose triphosphate, was found to selectively inhibit the T4 bacteriophage clamp loader (gp44/62). Mutagenesis studies strongly suggest that specific electrostatic interactions between the nitro moiety and Arg175 of gp44/62 are responsible for the selectivity. Functionally, this inhibition was shown to inhibit processive
DNA replication and effectively reduce T4 plaque formation over a lawn of E.coli . Several non-natural nucleosides were submitted to a screening service which identified 5-cyclohexly
2’-deoxyriboside (d5-CHI) as a functional inhibitor of P-gp. From this screen we hypothesized that d5-CHI acts as an ATP competitive inhibitor, however, this compound
was shown be a non-transportable substrate of P-glycoprotein that is able to stimulate its
ATPase activity and influence the drug stimulated ATPase catalytic efficiency (V max /K m) of
11 other important chemotherapeutic drugs. The effects of d5-CHI on drug stimulated P-gp
ATPase catalytic efficiency was shown to be in direct correlation with changes in P-gp mediated drug resistance in a P-gp overexpressing MDR cell line. Although the results of this study were contrary to the original hypothesis, d5-CHI proved to be a useful tool for developing a diagnostic method for determining P-gp mediated drug interactions using
ATPase activity. This study also demonstrated a unique selectivity of P-gp for its substrates as structurally similar compounds such as 5-cyclohexeneindolyl 2’-deoxyriboside and 5- phenylindolyl 2’-deoxyriboside failed to illicit any interactions with P-gp. Overall, these studies demonstrate the diverse architecture of ATPases and the use of non-natural nucleotides as chemical probes for understanding protein function.
12 Chapter 1: Introduction
1.1 Overview of ATP and ATPases
ATP binding proteins comprise a large portion of cellular proteins in any given species. It has been predicted that the human genome encodes for as many as 2000 ATP binding proteins (1). Of these proteins, 518 kinases have been identified and are responsible for controlling as many as 20,000 protein phosphorylation states to form a vast network of intracellular signal transduction (2). Mechanical ATPases are estimated to be as many as
~1500 and carry out functions such as intracellular transport, DNA replication and repair,
protein folding, and ion transport (3). Although ATPases have vast therapeutic potential,
these proteins share a common substrate (Figure 1.1) and therapeutically modulating an
ATPase would not be an ideal approach. However, this assumption has been challenged by
the emergence of selective kinase inhibitors used in cancer chemotherapy (4). How is it
possible that one agent can selectively target a kinase in a binding pocket that is common to
all ATPases? Two important lines of evidence address this question. First, the kinetics of
ATP utilization are not the same across all ATPases (5) indicating a divergence in substrate
binding mechanism and/or chemistry. Second, variability exists in the adenine binding
region shown directly by several X-ray structures (6).
The chemical properties of ATP and its role in biological systems
ATP is comprised of an adenine ring, ribose sugar, and three phosphate groups designated as α, β , and γ based on their proximity to the ribose sugar. Cleavage of either the α−β or
β−γ phosphodiester bonds is a net exothermic process (-∆G) yielding a ∆G of −30.5 kJ/mol
13 for the breaking of a β−γ bond and −45.6 kJ/mol for the breaking of a α−β bond (7). At
neutral pH, ATP is negatively charged and able to chelate metals such as magnesium,
sodium, and calcium. Consequently, the association of ATP with metals provides an
important co-factor in hydrolysis reactions. These metals are able to act as Lewis acids to
stabilize either the α−β or β−γ phosphodiester bond and render the corresponding non- stabilized bond susceptible to cleavage. The β−γ bond is typically cleaved in energy
dependent and signaling processes and the α−β bond is cleaved for incorporation of
nucleotides into RNA.
ATP is the energy currency of the cell. Biosynthetic pathways invest energy to
generate ATP and energy dependent cellular processes hydrolyze ATP to harness the release
of free energy. The human body stores less than 10 grams of ATP, however, it is ceaselessly
synthesized from fat, carbohydrate, and protein stores and degraded resulting in an average
turnover of ~50 kg of ATP for a human adult in a 24 hour period (8). The proteins that use this massive amount of ATP fall broadly into two categories, mechanical ATPases and signaling ATPases.
Mechanical ATPases: These proteins utilize the energy of ATP hydrolysis to perform
mechanical work like the kinetic movement of a protein (Figure 1.2A) and cytoskeleton
filament assembly/disassembly (Figure 1.2B) . For example, actin and tubulin utilize ATP
(or GTP) for the maintenance of cytoskeleton structure, function, and scaffolding protein-
protein interactions (3). These filaments are also important for intracellular transport as they
provide a medium for mechanical ATPases that act as molecular motors in cargo movement
(Figure 1.2A) (3). Another mechanical ATPase, myosin, together with actin utilizes ATP to control muscle contraction. In addition to the structural roles, AAA+ family ATPases
14 (ATPases associated with a variety of cellular functions) play many important roles in DNA replication (9). This family includes clamp loader subunits, such as replication factor C
(RFC) and gp44/62, and other replication proteins such as cdc6 and MCM. ATP-binding
cassette (ABC) transporters are another important class of mechanical ATPases which utilize
ATP for the active transport of a wide range of substrates. In humans, there are 48 known
ABC transporters present responsible for the import or export of substrates that range from
lipids to drugs (10 ).
Signaling ATPaes: ATP is critically involved in intracellular signal transduction because it is
utilized as a substrate by kinases to perform phosphoryl-transfer reactions. These reactions
can result in changes in enzymatic activity, protein localization, and/or
association/dissociation of protein complexes (3). Cellular growth, proliferation, and gene
expression are regulated by the ATP-dependent signal transduction cascades. Activation of
protein kinase A by cyclic adenosine monophosphate (cAMP) and the mitogen activated
protein kinase (MAPK) pathway are two instances of these ATP-dependent signaling
pathways (3).
Structural analysis of ATP binding pocket
The most highly conserved structural motif in all ATP binding proteins is the Walker motif originally identified by John E. Walker in ATP synthase subunits (11 ). Homology was later observed in ATP binding proteins such as myosin, phosphofructokinase and adenylate kinase. The Walker motif contains the sequence: G-X-X-X-X-G-K(T)X-X-X-X-X-X-I/V and X-ray structures have shown this motif forming a loop-like structure around the nucleotide, therefore stabilizing interactions of the lysine and threonine with the phosphate
15 groups of the nucleotide triphosphate. The Walker motif is also referred to as the P-loop
(Figure 1.3) because the exact sequence is found in many other proteins but only the
Walker sequences of ATP binding proteins adopt a distinctive loop conformation (12 ).
Another feature of ATPases required for catalytic function is a catalytic base side chain utilized in activation of an active site nucleophile. A water molecule serves as the nucleophile in most mechanical ATPases. The α−β anhydrous phosphate bond is stabilized by the metal co-factor, usually magnesium, allowing the activated water molecule to hydrolyze the β−γ
bond (Figure 1.4A) . Kinases share a similar chemistry but the main difference is the
nucleophile is a serine, threonine, or tyrosine residue on the acceptor protein of the
phosphoryl transfer reaction (Figure 1.4B) .
The highly conserved Walker motif interacts exclusively with the triphosphate moiety of the nucleotide whereas structural motifs associated with the binding of adenine are not as clear. Most ATPases do not utilize the hydrogen bonding capabilities of adenine for binding (13 ). Instead the primary driving factor appears to be the hydrophobic nature of the adenine binding cavity (14 ). In contrast, ATPase inhibitors utilize many different molecular interactions to maximize binding to the adenine cavity. The structures of select ATPase inhibitors (Figure 1.5) show a variety of groups such as aromatic, hydrophobic, hydrogen bonding, and electron withdrawing elements (15 ). Size of the ATP binding cavity is also relevant as larger pockets are easier targets for drug treatment as more protein-drug contacts are present in a large cavity. Despite all these factors, it can be argued that the entropic contributions from the nucleobase binding region are negligible in comparison to the enthalpic contributions of the triphosphate. While this is a valid argument, almost all
ATPases do not use other nucleotides such as GTP, CTP, or TTP suggesting that the adenine binding region does plays a selective role in ATP binding. Thus rationally designing
16 ATP competitive nucleoside analogs makes sense because these compounds could maximize interactions with the nucleobase binding region as well as the Walker domain. It should be noted that interactions of nucleosides with the Walker domain would depend upon the metabolism of these compounds to their nucleotide forms. However, this is a reasonable assumption since many existing nucleoside-based drugs have been shown to be metabolized by kinases in the nucleoside salvage pathway (16 ).
Existing drugs that target the ATP binding site
The most prevalent class of FDA-approved ATPase inhibitors are tyrosine kinase inhibitors used in the treatment of various forms of cancer. Perhaps the most well known ATPase inhibitor is the BCR-ABL inhibitor, Imatinib (Figure 1.6A) (17). The BCR-ABL fusion protein is the result of a translocation between chromosomes 9 and 22 and was found to be a biomarker that is strongly associated with chronic myelogenous leukemia (CML) (18 ).
From this genetic biomarker, Imatinib was developed to specifically target the fusion protein
(Figure 1.6B) . Its subsequent success in treating CML marked the first therapeutic kinase inhibitor and validated targeted therapies as an alternative to classical DNA damaging agents.
Since then, many more ATP competitive tyrosine kinase inhibitors have been developed such as erlotinib, gefitinib, and lapatinib (19-20 ). Although the binding affinities and mechanisms of these kinase inhibitors vary with different mutant forms of their target receptors, they generally rely on a combination of hydrogen bonding, hydrophobic and steric factors. For example, gefitinib competitively inhibits EGFR by forming many key contacts
within the ATP binding site. The quinazoline ring forms a hydrogen bond with main chain amide of Met793 while the 3-chloro-4-fluoro aniline occupies a hydrophobic pocket in the back of the ATP binding site (20 ).
17 While the recent coverage given to the kinase inhibitor field has made it synonymous
with ATPase inhibitors, targeting mechanical ATPases has received much less attention.
Currently, mechanical ATPases are the targets of several drugs such as podophyllotoxins
(topoisomerase II) (21 ), digoxin (Na +, K + ATPase) (22 ), and monastrol (kinesin Eg5) (23 ).
However, these drugs inhibit their targets allosterically or by other modes of non-
competitive inhibition rather than ATP competition. Despite this, the ATP binding cavity of
mechanical ATPases remains an attractive target for drug development. Indeed, efforts are
currently underway to develop ATP competitive inhibitors of flavivirus NS3 helicase (24 )
and kinesin (25 ).
18 1.2 Non-natural Nucleotides as Therapeutic Agents
Non-natural nucleotides are defined as any nucleotide not based on a purine or pyrimidine scaffold. In this dissertation, I describe a unique set of non-natural nucleotides that are 5- subtituted indolyl-2’-deoxyriboside triphosphates which mimic the core structure of dATP.
The 5’ substituent groups represent a diverse selection of chemistry ranging from an amine group to a napthyl (Figure 1.7) (26-30 ). These compounds have been used as probes to study the dynamics of translesion DNA synthesis and as chain terminating DNA replication inhibitors (28 ). Overall, the core structure of these compounds suggests that they will be
biologically active and may be potentially useful as a scaffold for drug design.
Non-natural nucleotides and translesion DNA synthesis
Translesion DNA synthesis is the ability of a DNA polymerase to replicate past damaged templating templating DNA (31 ). The underlying consequence of translesion DNA synthesis is the potential propagation of misreplicated base pairs resulting in mutations and eventually cancer (Figure 1.8) (). Accordingly, lesions such as an abasic site, thymine dimer, and O 6- methylguanine are strongly associated with cancer (32 ). In the study of translesion synthesis of abasic sites, a phenomenon called the “A-rule” was observed. This rule states that in the absence of templating DNA, a high fidelity DNA polymerase preferentially incorporates adenine opposite the abasic site (33 ). Based on this observation, Berdis and co-workers synthesized the library of non-natural nucleotides to investigate biophysical parameters that govern translesion synthesis such as size, shape, hydrophobicity, π-electron density, and hydrogen bonding capability (26-30 ). This library was screened for insertion opposite an
abasic site and identified 5-nitroindolyl 2’-deoxyribose triphosphate (d5-NITP) as an analog
capable of being incorporated opposite an abasic site with ~1000-fold greater efficiency
19 compared to adenine (34 ). The importance of this observation to this dissertation is that
since d5-NITP appears to act as a superior adenine mimetic, it may have profound effects
on other adenine utilizing proteins.
Clinical importance of nucleobase and nucleoside analogs
Nucleotides are among the most biologically active classes of chemicals. They are intimately involved in many major processes in the cell such as DNA replication, transcription, nucleotide metabolism, and energy dependent NTPase processes. Therefore they have long been used for the treatment of diseases such as cancer and viral infections. Examples of nucleoside analogs used in cancer include 5-fluorouracil, clofarabine, cytarabine, and gemcitabine (Figure 1.9). All inhibit the process of DNA replication at different points. For
example, the nucleobase 5-fluorouracil inhibits DNA replication by blocking the
biosynthesis of thymidine resulting in depletion of nucleotide pools (35 ). Clofarabine and
gemcitabine also deplete nucleotide pools by inhibiting ribonucleotide reductase (36-37 ). The
skewing of nucleotide pools by these compounds then enables their insertion into DNA
resulting in chain termination. Cytarabine is an example of a direct inhibitor of DNA
polymerase (38 ). Examples of anti-viral nucleoside analogs are acyclovir, ribavirin, and
ganciclovir (Figure 1.10). These compounds exploit fundamental differences between virus
and host replication proteins to selectively inhibit virus replication. The mechanism of action
of acyclovir illustrates the ability of a nucleoside analog to selectively influence several viral
NTP dependent processes. First, the acyclovir is rapidly metabolized to its monophosphate
form by viral thymidine kinase with an efficiency ~3000-fold greater than human thymidine
kinase (39 ) and is then metabolized further to its triphosphate form by cellular kinases.
Second, acyclovir triphosphate acts as a selective chain terminator by being incorporated
20 opposite templating RNA by viral DNA polymerase with ~100-fold greater affinity compared to human DNA polymerase (40 ). The mechanisms of action of all these drugs highlight the complex role of nucleotides in biological systems and suggest that many potential points of therapeutic intervention exist. This suggests that nucleoside analogs will continue to be a useful class of drugs and novel compounds such as non-natural nucleotides may have great therapeutic potential.
21 1.3 Model Systems for Evaluating the Influence of Non-Natural
Nucleotides on ATP-dependent Processes
The observation of d5-NITP as a superior adenine mimetic in translesion DNA synthesis raises the question of whether it is a superior adenine mimetic in the context of ATPases.
However, this postulate is not limited to d5-NITP as other non-natural nucleotides also displayed superior incorporation efficiency compared to adenine. In a broad sense, the overall goal of this dissertation is to explore the effects of non-natural nucleosides/tides on
ATP utilizing enzymes. As previously mentioned, ATP utilizing enzymes fall into two major categories: mechanical ATPases and kinases. Since a large berth of literature describing the interactions between the ATP binding site and a kinase inhibitor exists (4), this dissertation
will focus on exploring the ATP binding site of mechanical ATPases through use of novel inhibitors. Throughout this study, two model systems were used. The first is the DNA replicase complex whose formation is catalyzed by the ATP-dependent sliding clamp loader.
The second is the ATP binding cassette (ABC) transporter P-glycoprotein (P-gp) which confers multidrug resistance in many forms of cancer.
Sliding clamp loader ATPase
The first aim of this dissertation is to use non-natural nucleotides as inhibitors of clamp loader ATPases to demonstrate an alternative strategy to inhibit DNA replication. Currently, damaging agents are widely used to inhibit DNA replication. These agents select for cancer cells by damaging all rapidly dividing cells. However, several normal cell types such as hair follicles, the lining of the gastrointestinal tract and blood cells are rapidly dividing and will be affected by DNA damaging agents (41 ). These off-target affects result in nausea, alopecia,
22 and anemia (38 ). Inhibiting replicase formation could have fewer drawbacks because this
strategy would not depend on damaging DNA, only halting the process.
The genome of an organism is replicated by a replicase complex consisting of DNA
polymerase in complex with the “sliding clamp” processivity factor (42 ). While DNA
polymerase alone is capable of synthesizing DNA, the replicase complex is necessary to
increase the efficiency of polymerization to levels necessary for sustaining the organism (43 ).
In the absence of the clamp, DNA polymerase will replicate ~800 base pairs before dissociating whereas in complex with the clamp, DNA polymerase is capable of processively replicating thousands of base pairs (44 ). Typically, sliding clamps are circular oligomers that
encircle DNA through actions of their respective clamp loader proteins. Clamp loaders are
also protein complexes generally consisting of static subunits and ATP-binding subunits that
belong to the ATPases associated with a variety of cellular activities (AAA+) family of
ATPases. In general, clamp loaders bind and hydrolyze ATP to catalyze the association of
the sliding clamp with DNA. This is followed by the association of the DNA-sliding clamp
complex with DNA polymerase (Figure 1.11).
Based on previous studies with d5-NITP (vide supra), we hypothesize that d5-NITP
could act as an ATP competitive inhibitor of the bacteriophage T4 clamp loader. Selectivity
of d5-NITP was also assessed by comparing the inhibition constants for d5-NITP and other
non-natural nucleotides against the T4 clamp loader and E.coli clamp loader. These two
replicase systems were chosen as model systems for several reasons. Both organisms have
long been used as model systems for studying the process of DNA replication (45-46 ).
Additionally, purification protocols for each system are readily available allowing for rapid
generation of starting materials necessary for in vitro inhibition screening and characterization
(47-48 ). Functional inhibition of replicase formation can be easily assessed since T4 replicase
23 catalyzed DNA replication can be easily reconstituted in vitro (49-50 ). Finally, since T4 phage
is the obligate parasite of E.coli , in vitro selective inhibition can be validated in vivo by testing for inhibition of T4 plaque formation over a lawn of E.coli .
Comparison of clamp loader complexes across species: While clamp loaders serve the same function
for replicases of each species, key differences exist in the catalytic mechanisms of each
ortholog of clamp loader that could provide a platform for developing selective clamp loader
inhibitors. The most well studied systems are from yeast, E.coli , and Bacteriophage T4
(Figure 1.12, Table 1.1) (9). The T4 clamp loader (gp44/62) consists of four ATP binding
domains (gp44) and one wrench domain (gp62). The mechanism of gp44/62 ATP utilization
starts with ATP binding at all four sites that drive the association of gp44/62 with its clamp,
gp45, followed by hydrolysis of two ATP equivalents for the opening of gp45. The
remaining two ATP molecules of gp44/62 are hydrolyzed to encircle gp45 over DNA. The
proper association of gp43 with the gp45: DNA complex is then catalyzed by the
matchmaker functions of gp44/62 (50 ) followed by ATP-independent dissociation (51 ).
In contrast, the E.coli clamp loader γ-complex contains three unequal ATP binding domains
(γ), one distinct stator domain ( δ’), and one wrench domain ( δ). It should be noted that ATP hydrolysis is not necessary for the opening of its clamp, β-clamp, as previous studies have demonstrated that ATP binding alone is able to open the β-clamp (52 ). Binding of ATP to γ- complex is also necessary for the association of γ-complex with β-clamp and DNA.
Hydrolysis appears to only be required for closing the β-clamp over DNA and the dissociation of γ-complex (53 ). Although sequential ATP binding and hydrolysis is crucial
for γ-complex function, neither binding nor hydrolysis is cooperative.
24 The mechanism of the Saccharomyces cerevisae RFC has been extensively characterized and shares similarities with both gp44/62 and γ-complex. Its mechanism of ATP hydrolysis
and clamp loading resembles γ-complex. ATP binding drives the association of RFC with its respective clamp, proliferating cell nuclear antigen (PCNA), and hydrolysis facilitates the closing of PCNA and release of RFC from the complex (54 ). However, unlike γ-complex
there is no sequential binding and hydrolysis of ATP. The subunit composition of RFC
mirrors that of gp44/62 consisting of four ATPase active domains (RFC1-4) and one
ATPase inactive domain (RFC5). Interestingly, RFC5 is able to bind ATP but is hydrolysis
incompetent (55 ) and the complex of RFC2&5 is able to unload PCNA from DNA in an
ATP independent manner (56 ). However, RFC2&5 must be in complex to unload PCNA.
Although the effects of RFC2&5 on nucleotide binding, PCNA unloading, and overall RFC
function has not yet been determined; it is possible that these sites could modulate PCNA
opening and influence RFC. Interestingly, sequence-structure homology tools (57 ) show that
all clamp loaders adopt a similar shape but the primary amino acid sequence of the RFC
ATP-binding region more closely resembles gp44/62 with a 68% similarity compared to
48% similarity shared with the γ-complex.
P-glycoprotein
P-glycoprotein (P-gp) was the first mammalian ABC transporter to be identified (58 ). The
170 kDa membrane protein was isolated from multi drug resistant (MDR) cells that were selected by chronic exposure to increasing concentrations of cytotoxic agents (59 ). P-gp is expressed in all tissue types but at higher levels within epithelial cell types, primarily in the gastrointestinal tract and the blood brain barrier. In these tissues, P-gp primarily serves
25 cellular defense functions such as the export of xenobiotics and metabolites (60). Substrates for P-gp tend to be hydrophobic weak bases (61 ). Therefore, many drug-like molecules are transported by P-gp. It is also highly expressed in the adrenal glands where it is involved in the secretion of steroids (62 ). Knocking out P-gp is not an embryonic lethal mutation as P- gp knock-out mice have normal life spans. However, when challenged with xenobiotics, the mice are particularly susceptible to neurotoxicity as P-gp is not present to prevent to accumulation of drugs in the brain (63 ).
P-gp is a major target in cancer chemotherapy because it confers the MDR phenotype to a tumor by actively transporting cytotoxic chemotherapeutics out of the cell in an ATP-dependent manner. P-gp amplification can be intrinsic in tumors where P-gp is normally expressed, or acquired as an adaptation after multiple rounds of chemotherapy in a tumor where P-gp is normally not expressed (64-68 ). A correlation between P-gp expression and poor prognosis has been observed tumors such as ovarian cancer that typically do not express P-gp. A notable study (69 ) demonstrates that P-gp expression and prognosis are inversely correlated when treating an ovarian carcinoma with paclitaxel. In hematological malignancies, expression of P-gp increased with chemotherapeutic treatment rounds, leading to a poor prognosis (70 ). Given the clinical importance of P-gp in mediating drug resistance, a significant effort has been placed on identifying potential P-gp inhibitors. The second aim of this dissertation will explore the effects of a non-natural nucleoside inhibitor of P-gp activity identified by one such screen. Of several non-natural nucleosides screened for P-gp inhibition, only one tested positive suggesting a unique mechanism of action. In addition, these studies offer mechanistic insights to basic P-gp function.
26 Proposed transport mechanisms of P-glycoprotein: P-gp displays the unique ability to transport a wide
variety of structurally diverse substrates (Figure 1.13). This phenomenon has been extensively studied, resulting in two non-mutually exclusive models. The first model is the
“hydrophobic vacuum cleaner” (71 ) model that proposes that P-gp extracts substrates directly from the plasma membrane into its transmembrane pore for subsequent transport.
The basis for this model was the observed requirement of plasma membrane localization for transport and the fact that most substrates are in fact hydrophobic and predicted to partition into the lipid bilayer (72-73 ). The x-ray structure of P-gp was consistent with this model as it shows entry portals of P-gp directly accessible to the plasma membrane (74 ).
Another model, the flippase model, suggests that P-gp transports substrates from the inner leaflet of the plasma membrane directly to the outer leaflet rather than the extracellular space
(71 ). This model is energetically favorable because the transport of a substrate to the lipid
outer leaflet would not require hydration, which is a large energy barrier to overcome.
Experimental evidence exists for this model, for example, monolayer efflux experiments
show no detectable transport of doxorubicin although it is a classical P-gp substrate (75 ). By action of a flippase, doxorubicin is likely transported to the outer leaflet of the plasma membrane rather than the aqueous phase making it undetectable by the monolayer efflux assay. Although doxorubicin provides a clear example of a flippase transport, many other substrates such as vinblastine and digoxin are readily detectable in the aqueous phase (75 ).
This strongly suggests that transport to the extracellular aqueous phase may be a two step
process consisting of flippase transport to the outer leaflet followed by dissociation to the
aqueous phase (Figure 1.14). Extremely hydrophobic molecules such as doxorubicin may
have a longer residence time in the outer leaflet phase rendering the dissociation to aqueous
phase step virtually undetectable.
27
The role of ATP in the catalytic cycle of P-glycoprotein: P-gp has the structure of a prototypical ABC transporter, a monomer consisting of two transmembrane domains and two nucleotide binding domains (Figure 1.15). The transmembrane domains consist of six α-helices that
form the transmembrane pore for transport of substrates (74 ). Each nucleotide binding
domain is in the cytoplasm and contains a Walker A and Walker B motif. Regardless of the
transport mechanism, dimerization of the nucleotide binding domains is required for
substrate transport and is controlled by ATP binding and hydrolysis (76 ). However, the role
of ATP in the catalytic cycle of P-gp is a subject of debate in the field. Currently, two major
models attempt to describe the role of ATP in substrate transport. The first model is the
ATP switch model (77 ) which proposes that ATP binding drives dimerization of the
nucleotide binding domains followed by a change in the transmembrane domain
conformation from an “open” cytoplasmic facing conformation to a “closed” extracellular
facing conformation (Figure 1.16). Hydrolysis of the bound nucleotides resets the
transmembrane domains back to the open conformation. The second model is referred to as
the “occluded nucleotide” model (78 ). In this model, ATP also drives dimerization of the
nucleotide binding domains but in this model dimerization does not result in a transition of
the transmembrane domains from open to closed. The model describes two distinct ATP
bound states; a high affinity “occluded” state and low affinity “non-occluded” state. In the
beginning of the catalytic cycle, ATP binding drives the dimerization of the nucleotide
binding domains in a non-occluded state whiel the transmembrane domains remain in an
open conformation. The switch of one bound ATP to an occluded state promotes the
change from open to closed transmembrane domain conformation followed by substrate
transport (Figure 1.17). The switch from non-occluded to occluded is proposed to cause the
28 nucleotide bound at that site to hydrolysis and subsequent dissociation. At that point, the other non-occluded nucleotide can enter the same cycle to stimulate another round of transport while the empty nucleotide binding site can be occupied by another nucleotide.
Modulating P-glycoprotein: Over the years, several compounds have been shown to influence the activity of P-gp. The first inhibitor identified was verapamil (79 ). Shortly after, the immunosuppressant drug cyclosporine A was identified as an inhibitor (80 ). However, as
more inhibitors were discovered it became apparent that inhibitors had different effects on
the efflux of different substrates. Therefore, inhibitors were more appropriately termed
“modulators.” Modulators are able to reverse MDR against a subset of drugs and restore
sensitivity evidenced through lower LD 50 for those drugs. Several compounds have been shown to modulate P-gp and the structures of these compounds share the same degree of diversity as P-gp substrates. This suggests that modulators are most likely substrates and modulation of P-gp activity is actually the result of a drug-drug interaction. A possible explanation for observed inhibition is that a P-gp substrate-like inhibitor readily re-crosses the plasma membrane after being exported. This results in a futile cycling of export that effectively prevents the transport of other intracellular substrates. To date, all known modulators are substrate-like and inhibiting P-gp by binding to the nucleotide binding domains remains unexplored. Due to the structural similarities between ATP and non- natural nucleotides, it is possible that one or more analogs can bind and inhibit P-gp through the nucleotide binding site.
Utilizing P-gp modulators to treat MDR cancers has proven to be difficult because of the important role of P-gp in protecting non-cancer cells from potentially harmful xenobiotics. For example, a modulator would sensitize an MDR tumor to a cytotoxic agent
29 but may expose the brain to the same cytotoxic agent, putting the patient at risk for neurotoxicity. In addition, inhibition of P-gp would alter the pharmacokinetics of a drug as
P-gp is responsible for the excretion of xenobiotics. Drugs would have a longer half-life
which may result in additional undesired side-effects. Despite these risks, cancer chemotherapy is not a chronic treatment. Patients are often treated under close supervision allowing for the strategic administration of P-gp modulators for a better clinical outcome.
Several modulators are now in clinical trials, most notably zosuquidar (81 ) and tariquidar
(Figure 1.18) (82 ) which are highly selective small molecule P-gp inhibitors.
30 1.4 Statement of Purpose
Non-natural nucleotides have been shown to be useful tools in exploring the dynamics of translesion DNA replication and the overall process of DNA replication. Consequently, several analogs have been shown to behave as superior adenine mimetics. The overall aim of this dissertation is to explore the biological role of mechanical ATPases by using non-natural nucleotides as chemical entities for modulating their function. Chapter 2 focuses on using non-natural nucleotides as inhibitors and structural probes of clamp loader proteins. This chapter demonstrates structural variability in the ATP binding sites of clamp loader proteins from two different species and also demonstrates a new therapeutic strategy for inhibiting
DNA replication. Chapter 3 addresses the effects of non-natural nucleosides on the ABC transporter P-gp. From a screen of nine non-natural nucleosides, one compound was identified as a potent P-gp inhibitor. This chapter explores the mechanism of inhibition and
while focusing on the ATPase activity of P-gp, leads to the development of an in vitro
screening method for determining P-gp drug-drug interactions. Chapter 4 will discuss the
broader implications of this work and future directions.
31 Figures
Figure 1.1 Structure of (A) adenosine triphosphate (ATP) and (B) deoxyadenosine triphosphate (dATP). Phosphate groups are designated α, β, and γ . The β−γ phosphodiester
bond of ATP is cleaved by ATPases for energy dependent processes; ADP and inorganic
phosphate are released. ATP also serves as a substrate for phosphoryl transfer reactions
catalyzed by kinases. dATP is used in DNA replication for the incorporation of adenine. In
this reaction, the α−β phosphodiester bind is cleaved, releasing pyrophosphate.
32 Figure 1.1
(A)
γ β α
(B)
γ β α
33 Figure 1.2 Schematic of a molecular motor and actin assembly (A) Molecular motor moving across an actin filament. Head units hydrolyze ATP to ‘walk’ along the actin filament while associated carrier proteins bind cargo. (B) Assembly of actin. Actin monomers are activated
by binding ATP. Nucleation of activated actin monomers initiates the polymerization of a
filament. Hydrolysis of ATP and P i release stabilize the filament.
34 Figure 1.2
(A) CargCargo o
Walking
ADP Head 1 + Head 1 ATP Head 2 Pi
(B) Actin ATP Activated Actin
Assembly
ADP bound Pi
35 Figure 1.3 Representation of a typical ATPase P-loop (taken from T4 bacteriophage
gp44/62). The triphosphate of ATP forms electrostatic interactions with the Walker A
domain (glycine, lysine, threonine) of an ATPase.
36 Figure 1.3
Threonine
Glycine
Lysine
37 Figure 1.4 Chemical mechanism of ATP hydrolysis. (A) Hydrolysis of ATP by a mechanical
ATPase. Magnesium acts as a Lewis acid and stabilizes the α−β phosphodiester bond. A catalytic base, typically a histidine or lysine, then activates a water molecule which then cleaves the more labile β−γ bond. (B) Phosphoryl transfer by a kinase. Similar reaction mechanism as the mechanical ATPase; however, the catalytic base activates a serine,
threonine, or tyrosine to accept the γ -Pi.
38 Figure 1.4
(A)
NH2
N N NH2
N O O N N N :B HOH :OH HO P O P O O OO O N N -O O- H H HO P O P O P O H H O OH H Mg++ O- -O O- H H H H + OH OH Mg++ O
-O P O-
O-
(B)
NH2
N N :B HO :O
OO O N N O
HO P O P O P O O -O P O O- -O O- H H H H OH OH O- Mg++
39 Figure 1.5 Structures of ATP competitive inhibitors. (A) Lapatinib, (B) sorafenib, (C)
staurosporine, (D) erlotinib. Compounds are all generally hydrophobic with aromatic groups
highly prevalent on all structures.
40 Figure 1.5
(A)
N
N
HN
O NH Cl F O S O
(B)
F FF O
Cl O O N H N N N H H
(C) (D)
H O N
O N O N N O O N O
HN O
NH
41 Figure 1.6 Imatinib and cAbl kinase. (A) Structure of Imatinib. (B) Imatinib bound to cAbl
kinase domain (pdbID: 1IEP). The translocation between chromosome 9 and 22 results in
the fusion of the Break Point Cluster (BCR) and abelson kinase (Abl) genes. This fusion
gene expresses a constitutively active Abl kinase which is strongly associated with chronic
myelogenous leukemia (CML). Imatinib competes with ATP for the kinase domain of ABL
kinase and stabilizes cAbl in its inactive conformation.
42 Figure 1.6
(A)
N
HN NN
CH3
HN
CH3 O N
N
(B)
43 Figure 1.7 Structures and electron density surface potentials of non-natural nucleobases. All
models were constructed using Spartan ’04 software (www.wavefun.com). The electron
density surface potentials of adenine and non-natural nucleobases were then generated. The
most electronegative regions are in red, neutral charges are in green, and the most
electropositive regions are in blue.
44 Figure 1.7
H F N H
N N N H H H Indole 5-aminoindole 5-fluoroindole N N H O N O- O H N 5-cyclohexylindole 5-phenylindole O- O N N N N H H H -O 4-nitroindole 5-nitroindole 6-nitroindole
O CH2 CH3 C CH CH 2 O- N N H H N N N H H H 5-cyclohexeneindole 5-napthylindole 5-ethyleneindole 5-ethylindole 5-carboxylindole
45 Figure 1.8 Schematic of translesion DNA synthesis. ‘X’ represents an abasic lesion. On the
left side, a DNA polymerase will bypass this lesion by incorporating dAMP. However, if the
abasic site previously contained a guanine or cytosine, the incorporation of adenine will
result in a point mutation which will be propagated in the next round of replication. On the
right, a non-natural nucleotide such as 5-NITP is incorporated 1000-fold more efficiently
than dATP opposite an abasic site by a high fidelity DNA polymerase. 5-NITP acts as a
chain terminator, effectively halting the propagation of mutations.
46 Figure 1.8
O +N -O N O O O -O P O P O P O O O- O- O- H H H H OH H
Propagation of Growth arrest, apoptosis miscoding information DNA repair
47 Figure 1.9 Structures of nucleobase and nucleoside analogs used in cancer chemotherapy.
Nucleobase (A) 5-fluorouracil and nucleosides (B) clofarabine, (C) cytarabine, (D) gemcitabine.
48 Figure 1.9
(A) (B) NH2
N N H O NO N N Cl
HO O HN H H H H F OH F
(C) (D) NH2 NH2
N N
N O N O
HO HO O O H OH H F H H H H OH OH F
49 Figure 1.10 Structures of nucleoside analogs used in anti-viral chemotherapy. (A) acyclovir,
(B) ribavirin, (C) ganciclovir
50 Figure 1.10
(A) (B) O O NH2 H N HN HO N N H O H2N H HO O OH OH
(C) O
N NH
HO N N NH O 2
OH
51 Figure 1.11 Schematic of replicase formation. DNA replication in vivo is carried out by a confederation of proteins including DNA polymerase, and sliding clamp and loader processivity factors. Clamp loaders utilize the energy of ATP hydrolysis to load sliding clamps on to DNA. Sliding clamps then associate with DNA polymerases to form replicase complexes.
52 Figure 1.11
Clamp loader Sliding clamp
ATP
DNA polymerase ADP + P i
5’ 3’ 5’ 3’ 3’ 5’ 3’ 5’
53 Figure 1.12 Subunit compositions of clamp loaders from different species. ‘Motor’ refers to
ATP hydrolyzing units, ‘stator’ refers to structural subunits, and ‘wrench’ refers to pivot subunits that physically bind to sliding clamps.
54 Figure 1.12
E.coli γγγ-complex Eukarya Motor Motor γ 3 γ γ 2 4 δ’ δ 5 1
Stator Wrench Stator Wrench
Archaea T4 Bacteriophage Motor Motor
S3 44 S2 L2 44 44
S1 L1 44 62
Stator Wrench Stator Wrench
55 Figure 1.13 Sample of the structural diversity of P-gp substrates. (A) Paclitaxel, (B)
Cyclosporine A, (C) vinblastine, (D) doxorubicin. Cyclosporine A inhibits P-gp by acting as an alternative substrate. P-gp expression confers resistance to paclitaxel, vinblastine, and doxorubicin.
56 Figure 1.13
(A)
O O OH O
O NH O H O O O OH O OH O O
(B)
HO O O H N N N N N O O O O N O O H H N N N N N H O O O
(C) (D)
OH N OOH O OH H OH N H N O H O O OO OH O O N H
O H OH O OH O O NH2
57 Figure 1.14 Model of P-glycoprotein transport. Blue ovals represent transmembrane domains and red circle is the substrate. (Left side) The hydrophobic vacuum cleaner model shows that substrates partition into the lipid bilayer and then are extracted from the bilayer into the drug binding cavity of P-gp. Transmembrane domains are in the “closed” conformation. (Right side) Transport cycle: Step 1 is the flippase activity of P-gp that
translocates the substrate to the membrane outer leaflet. Transmembrane domains are now
in the “open” conformation Step 2 is the dissociation of the substrate from the membrane
outer leaflet to the extracellular aqueous phase.
58 Figure 1.14
1
2
59 Figure 1.15 Structure of P-glycoprotein (pdbID: 3G5U) (74 ). P-gp has a typical ABC
transporter structure consisting of transmembrane domains (TMD) and nucleotide binding
domains (NBD).
60 Figure 1.15
TMD TMD
NBD NBD
61 Figure 1.16 ATP switch model. Ovals represent transmembrane domains and pentagons represent nucleotide binding domains. Black triangle is the substrate. The open transmembrane conformation allows for ligand binding from the plasma membrane. ATP binding drives dimerization of nucleotide binding domains leading to the closed transmembrane conformation and ligand transport. Subsequent hydrolysis and release of
ADP and P i resets the open conformation.
62 Figure 1.16
ATP ADP ADP P i P ATP Hydrolysis i release Ligand P binding ATP ADP i ATP
ADP ATP Pi
Closed dimer Open dimer
63 Figure 1.17 The occluded nucleotide model. Blue rectangles represent transmembrane domains and gray crescents are nucleotide binding domains. ATP binding promotes the open transmembrane conformation and ligand binding. One ATP site enters the ‘occluded’ state and drives the switch from the open transmembrane conformation to the closed conformation and promotes ligand transport. The switch to occluded state commits the nucleotide to subsequent hydrolysis and release of ADP regenerates the open transmembrane conformation. The second ATP site is now available to enter the occluded state and power another round of transport.
64 Figure 1.17
High-affinity → Low-affinity Switch
ATP Hydrolysis
ATP ADP ADP ATP 1 1 1 1
ATP ATP ATP ATP 2 2 2 2 Dissociation of ATP-site 1
65 Figure 1.18 Structures of the most recent P-glycoprotein inhibitors. (A) zosuquidar
(LY335979) and (B) tariquidar (XR9676). Both compounds are currently in clinical trials
(searchable on clinicaltrials.gov).
66 Figure 1.18
(A) H O N
H O N
O N
N O
O
O
(B)
F F
N OH
N
ON
67 Tables
Table 1.1 Subunit compositions and functions of clamp loaders from various species.
Organism Subunits Functions
Escherichia coli γ motor (γ-complex) δ wrench δ’ stator
Eukaryotic Yeast/human (RFC) Rfc 2-3-4/p37-p36- motor p40 wrench Rfc1/P140 stator Rfc5/p38
Archael RFC S motor/stator (RFC) RFC L motor/wrench Bacteriophage T4 gp44 motor/stator (gp44/62) gp62 wrench
68 Chapter 2: Selective inhibition of DNA replicase assembly by a non- natural nucleotide: Exploiting the structural diversity of ATP-binding sites
Kevin Eng a, Sarah K. Scouten-Ponticelli b, Mark Sutton b, and Anthony Berdis a,#
aDepartment of Pharmacology, Case Western Reserve University, Cleveland OH 44106 bDepartment of Biochemistry, University of Buffalo, State University of New York, Buffalo
NY, 14214
Reproduced in part with permission from [Eng K, Scouten-Ponticelli SK, Sutton M, Berdis
A. Selective inhibition of DNA replicase assembly by a non-natural nucleotide: exploiting the structural diversity of ATP-binding sites. ACS Chem Biol. 2010 Feb 19;5(2):183-94.]
Copyright [2010] American Chemical Society.
2.1 Abstract
DNA synthesis is catalyzed by an ensemble of proteins designated the replicase. The efficient assembly of this multi-protein complex is essential for the continuity of DNA replication and is mediated by clamp-loading accessory proteins that use the binding and hydrolysis of ATP to coordinate these events. As a consequence, the ability to selectively inhibit the activity of these accessory proteins provides a rational approach to regulate DNA synthesis. Toward this goal, we tested the ability of several non-natural nucleotides to inhibit
ATP-dependent enzymes associated with DNA replicase assembly. Kinetic and biophysical
69 studies identified 5-nitro-indolyl-2'-deoxyribose-5'-triphosphate as a unique non-natural nucleotide capable of selectively inhibiting the bacteriophage T4 clamp loader versus the homologous enzyme from Escherichia coli . Modeling studies highlight the structural diversity
between the ATP-binding site of each enzyme and provide a mechanism accounting for the
differences in potencies for various substituted indolyl-2'-deoxyribose-5'-triphosphates. An
in vivo assay measuring plaque formation demonstrates the efficacy and selectivity of 5-nitro- indolyl-2'-deoxyribose as a cytostatic agent against T4 bacteriophage while leaving viability of the E. coli host unaffected. This strategy provides a novel approach to develop agents that selectively inhibit ATP-dependent enzymes that are required for efficient DNA replication.
70 2.2 Introduction
DNA replication is essential for the proliferation and survival of all forms of life ranging from simple viruses and bacteria to more complex organisms including humans. However, the importance of DNA replication is often highlighted by various pathological states caused by dysfunctional and/or unregulated replication activity. Diseases such as cancer, autoimmune disorders, and viral infections require abnormally high levels of DNA synthesis.
As a consequence, these pathological states are treated with compounds that inhibit DNA synthesis. For example, DNA damaging agents and chain-terminating nucleosides inhibit
DNA synthesis by chemically transforming nucleic acid into an ineffective substrate for elongation (83 ). Unfortunately, these agents often cause severe side effects induced by the
non-selective killing of pathogenic and healthy cells (84 ). In addition, these agents can
accelerate disease development by altering the integrity and stability of genomic material
(85 ). For example, anti-viral nucleosides can cause symptoms mimicking diabetes mellitus
(86 ) while DNA damaging agents are linked with the development of therapy-related cancers
(87 ).
To avoid these complications, other molecular targets involved in DNA synthesis
must be evaluated as potential points for therapeutic intervention. Indeed, efficient DNA
replication is dependent upon a confederation of proteins (46 ) that must function in a
collaborative effort. Several of these proteins, including DNA helicases and "clamp-loading"
accessory proteins, require ATP binding and hydrolysis to function properly (42 ). Clamp
loaders are an attractive target as they function to increase the efficiency of DNA synthesis
by placing accessory proteins, referred to as "sliding clamps", onto nucleic acid in an ATP-
dependent manner. Sliding clamps increase the processivity of DNA polymerases involved
in chromosomal replication, and disrupting the interactions between a DNA polymerase and
71 its processivity factor dramatically reduces the overall efficiency of DNA synthesis (88 ).
Since replicative accessory proteins are species-specific, inhibiting the function of a pathogenic protein without affecting the activity of the host protein could be developed into a selective therapeutic agent to inhibit cellular proliferation.
However, this is not an easy task since clamp loaders from viruses, bacteria, and
eukaryotes all rely on the binding and hydrolysis of ATP for their biological function. As
such, the commonality in primary amino acid sequence for the ATP binding site suggests a
low probability of identifying a unique molecule that inhibits a clamp loader from one
species while leaving another unaffected. In spite these obstacles; this report describes the
ability of various non-natural nucleotides (Figure 2.1) to disrupt processive DNA synthesis
by inhibiting the activity of ATP-dependent accessory proteins. Kinetic, biophysical, and in
vivo data reveal a specific non-natural nucleotide, 5-nitro-indolyl-2'-deoxyribose-5'-
triphosphate (d5-NITP), that selectively inhibits the bacteriophage T4 clamp loader versus
the functionally homologous clamp loader from Escherichia coli . This represents a novel
strategy to develop therapeutic agents against hyperproliferative diseases such as viral
infections and cancer by selectively inhibiting ATP-dependent enzymes involved in DNA
replication and recombination.
2.3 Methods and Materials Materials. Non-natural nucleosides and nucleotides were synthesized as previously described (27-30 ). Purification of wild-type gp44/62 and gp45 from overproducing strains obtained from William Konigsberg (Yale University) was performed as previously described
(47 ). The R175K and R175L mutants of gp44/62 were constructed and purified as described
in Supplemental Table 1. Purification of exonuclease-deficient T4 DNA polymerase
72 (D129A) was performed as previously described (89 ). The pET26b vector harboring the
double mutant gp45 W199F V162C (90 ) was a generous gift from Dr. Stephen Benkovic
(Pennsylvania State University). Expression, purification, and labeling of the mutant gp45
with 7-diethylamino-3-(4'-maleimidylphenyl)-4-methylcoumarin (CPM) (Molecular Probes)
was done as described (90 ). Purification of γ-complex and β-clamp were done as previously described (48 ).
Initial velocity studies in the presence and absence of inhibitor. All experiments with
gp44/62 used the following buffer system (designated T4 buffer): 50 mM Tris pH 7.5 , 10
mM DTT, 150 mM potassium acetate, 10% glycerol. A typical ATPase assay contained 10
mM Mg 2+ , 1 µM 13/20 DNA, 500 µM (r)NTP or (d)NTP, 500 nM gp45, and 500 nM
gp44/62 in T4 buffer. ATPase activity was measured by monitoring the release of P i using a malachite green assay (91 ) or via the hydrolysis of [ γ-32 P]-ATP. Reaction inhibition studies
used identical conditions except for the inclusion of variable concentrations of inhibitor (0-
400 µM) and a fixed concentration of ATP (32.5 nM [ γ-32 P]-ATP and 100 µM unlabeled
ATP). Reactions were quenched at variable times (0, 30, 60, 90, and 120 seconds) by the
addition of an equal volume of 1 N formic acid and analyzed by TLC on PEI-cellulose plates
(EM Science) using 0.6 M KH 2PO 4 pH 3.5. The ATPase activity of the γ-complex was measured as above using the following modifications: all reactions were performed at 37°C,
γ-complex buffer (20 mM Tris pH 7.5, 5 mM DTT, 10% glycerol) was used, the concentration of ATP was fixed at 20 µM, and 100 nM of β-subunit and γ-complex was
used. Reactions were quenched using 0.5 M EDTA. In both cases, product formation was
32 detected using a Packard Cyclone PhosphorImager. The ratio of free Pi to non-hydrolyzed
[γ-32 P]-ATP was multiplied by the final concentration of ATP to obtain total product
73 concentration. Product formation in the absence of enzyme was measured and subtracted from all measurements. Initial velocities were obtained by fitting the data to equation (1).
y = mx + b (1)
Y is product concentration, x is time, m is the slope, and b is the y-intercept. IC 50 values
were obtained by fitting initial velocities to equation (2).
n y = 100/(1+( IC 50 /I) ) (2)
y = fractional activity, I is the concentration of inhibitor, IC 50 is the concentration of inhibitor that yields 50% enzyme activity, and n is the Hill coefficient for the inhibitor. True
Inhibition constants designated as K i values were obtained from equation (3):
Ki = IC 50 /1+[ATP]/ Km (3)
Km is the Michaelis-Menten constant for ATP, [ATP] is the concentration of ATP, and IC 50
the concentration of inhibitor that yields 50% enzyme activity.
Replicase Formation Assay: 34/Bio-62 (50SP)/36-mer was prepared and labeled as
previously described (49 ). T4 buffer was mixed with 34/Bio-62 (50SP)/36-mer (500 nM),
ATP (100 µM), Mg 2+ (10 mM), streptavidin (1 µM), dCTP (500 µM), gp44/62 (500 nM), gp45 (500 nM), and gp43 exo - (150 nM). Reaction was initiated with gp43 exo - and incubated
for 10 seconds. Elongation was initiated by addition of (d)NTPs (100 µM) with salmon
sperm DNA trap (3 mg/ml) and quenched in an equal volume of 1 M HCl at variable time
(0-15 seconds). DNA was extracted using phenol: chloroform: isoamyl alcohol (25:24:1) and
neutralized with 1 M Tris/NaOH. Strands were separated on a 16% denaturing acrylamide
74 gel and detected by phosphorimaging. The presence of full-length product (50-mer) is indicative of replicase formation.
Fluorescence measurements: Tryptophan fluorescence was measured with a Kintek SF-
2004 stopped-flow. Excitation wavelength was 290 nm and emission cutoff filter was 310 nm. Syringe A contained T4 buffer, 2 µM gp44/62 and 20 mM Mg 2+ . Syringe B contained
T4 buffer, 2 µM 45 W199F V162C-CPM, and 1 mM ATP or 1 mM d5-NITP. Single mixing reactions were monitored over 3 seconds and data were fit to the equation for a single exponential (Equation (4)).
Y=A(1-e-kt ) (4)
A is the amplitude, k is the first-order rate constant, and t is time.
In vivo Phage assay and toxicity screen: Initial screening of potential toxicity against E. coli was performed growing JA300 E. coli cells (ATCC) in the absence and presence of
various non-natural nucleosides. The culture media was LB supplemented with 0.8 g glucose/liter. A 1:250 dilution of an overnight culture was grown to mid-log phase and treated with non-natural nucleoside for one hour. Growth curves were obtained by
measuring OD 600 and toxicity was assessed by comparing growth curves of cells treated with d5-XIs and DMSO. Long-term toxicity was assessed by plating dilutions of suspension cultures on LB/agar plates and counting colonies after 24 hours. The inhibitory effects of
various non-natural nucleosides against phage proliferation was assesses using a plaque forming assay. An overnight culture of JA300 E. coli cells was diluted 1:150 in LB and pre- incubated with varying concentrations of non-natural nucleosides for 60 minutes at 37°C.
Cultures were then infected with ~300 pfu of T4 bacteriophage (ATCC). The suspension
was decanted over an LB/agar plate. After the suspension absorbed into the LB/agar, plates
75 were incubated at 37°C for 24 hours and plaques were manually counted. ANOVA analysis and student’s t-test were done using Graphpad Prism v4.0. To test the dependence of non- natural nucleoside metabolism on phage inhibition, a deoxythymidine kinase knock out E.coli strain (KY895)(92-93 ) was used in place of JA300 in the plaque forming assay.
Molecular modeling: An energy minimized coordinates of d5-NITP was generated from the Dundee PRODRG2 Server ( http://davapc1.bioch.dundee.ac.uk/programs/prodrg/ )
(94 ). The structure of γ-complex used for modeling was obtained from the RCSB Protein
Data Bank (PDBID: 1NJF). A model of gp44/62 was obtained by threading the primary
sequence of gp44/62 was into the structure of P. furiosus RFC (24% sequence identity
PDBID: 1IPQ) using the Imperial College Protein Homology/analogy Recognition Engine
(PHYRE) ( http://www.sbg.bio.ic.ac.uk/phyre/ ). d5-NITP was docked into the active site of each structure using Crystallographic Object-Oriented Toolkit (95 ). Docked models were then subject to energy minimization using CNSolve v. 1.1.
2.4 Results and Discussion
5-NITP is a Potent Inhibitor of the Bacteriophage T4 Clamp Loader. d5-NITP is a non-natural
nucleotide that mimics the size and shape of (d)ATP (Figure 2.1A) and is used as an effective surrogate for dATP by various DNA polymerases during the misreplication of damaged DNA (27 ). These features prompted us to evaluate if d5-NITP could also act as a
substrate for the ATP-dependent bacteriophage T4 clamp loader, gp44/62. Despite
structural and functional similarities to dATP, d5-NITP is not hydrolyzed by gp44/62
(Figure 2.2A) . The inability to hydrolyze d5-NITP is not caused by the absence of the 2'-
hydroxyl moiety since gp44/62 is also incapable of hydrolyzing r5-NITP, the ribose form of
76 the non-natural nucleotide (Figure 2.2A) . This contrasts data obtained with natural nucleotides in which gp44/62 hydrolyzes dATP just as efficiently as ATP (Figure 2.2A) .
The lack of hydrolysis could simply reflect an inability to bind the non-natural nucleotide. This possibility was tested by monitoring the dose-dependency of d5-NITP toward inhibiting the ATPase activity of gp44/62. Figure 2.2B shows that d5-NITP inhibits
gp44/62 with a K i value of 4.8 ± 0.5 µM. A similar K i of 10.8 ± 0.7 µM is obtained using r5-
NITP (Figure 2.2B) , indicating that the bacteriophage clamp loader is promiscuous in its ability to bind either ribose or deoxyribose nucleotides (44 ). The calculated Hill coefficient for both non-natural nucleotides is ~1, indicating a lack of positive or negative cooperativity between the four ATP binding sites of gp44/62.
The mode of inhibition by d5-NITP was determined by measuring ATP hydrolysis at several different fixed concentrations of d5-NITP while varying the concentration of ATP.
The double reciprocal plot yields a series of lines intersecting on the y-axis and are diagnostic
for reversible, competitive inhibition (Figure 2.2C) (96 ). The measured K i of 5.7 +/- 1.1
µM is identical, within error, to the value of 4.8 µM measured using Dixon plot analysis. It is
striking that the K i for d5-NITP is ~20-fold lower than the K m of 110 µM for ATP (51 ). In
addition, the K i for d5-NITP is ~5-fold lower than the inhibition constant of 29 µM for
ATP γS (Table 2.1) and~300 fold lower that the value of 1,200 µM measured for 5'-
adenylyl-beta,gamma-imidodiphosphate (AMP-PNP) (97 ). The lower inhibition constant for
d5-NITP compared to other competitive inhibitors indicates superior binding affinity that is
influenced by the unique chemical features present on the 5-nitro-indolyl moiety. However,
the triphosphate group is essential for binding as 5-nitro-indoly 2'-deoxynucleoside (d5-NI)
does not inhibit gp44/62 at concentrations greater than 200 µM.
77 d5-NITP inhibits DNA synthesis by blocking replicase assembly. gp44/62 catalyzes formation of the replicase, a multi-protein complex that performs highly processive DNA synthesis. During this process, gp44/62 binds and hydrolyzes ATP to first load the processivity factor, gp45, onto DNA and then coordinates proper interactions of gp45 with the DNA polymerase
(gp43) in an ATP-independent manner (50 ). Although gp44/62 does not hydrolyze d5-
NITP, we tested if replicase formation can occur solely through the binding of the non- natural nucleotide using the strand displacement polymerization assay (98 ) (Figure 2.3A) .
This assay distinguishes between processive DNA synthesis catalyzed by replicase complex
(synthesis beyond a forked strand) from the activity of DNA polymerase that does not perform strand displacement synthesis. As illustrated in Figure 2.3B, longer replication products are generated by the replicase compared to DNA polymerase alone (compare lane
4 and lane 2, respectively). The inclusion of 100 µM d5-NITP inhibits formation of the
replicase complex as shorter replication products are produced (Figure 2.3B, lane 5)
compared to when d5-NITP is omitted (Figure 2.3B, lane 4) . The reduction in DNA
synthesis does not reflect direct inhibition of polymerase activity by d5-NITP since identical
amounts of products are generated by the polymerase in the absence or presence of d5-
NITP (compare lanes 2 and 3, respectively). The lack of inhibition is consistent with reports
indicating that d5-NITP is poorly incorporated opposite any of the four natural templating
nucleobases (34, 99 ).
The inhibitory effect by d5-NITP was further investigated using a FRET quenching
assay developed by Benkovic and co-workers (100 ) that monitors the ability of gp44/62 to
open the closed ring of the homotrimeric gp45 labeled with fluorescent probes. When CPM-
labeled gp45 is mixed with gp44/62 and 1 mM ATP, a rapid change in fluorescence with an
amplitude of 0.2199 ± 0.0004 units is obtained (Figure 2.3D) and confirms that clamp
78 opening occurs upon ATP binding and hydrolysis. However, a significantly smaller change in fluorescence (amplitude = 0.044 ± 0.002) is detected when ATP is replaced with 1 mM d5-
NITP (Figure 2.3D) , indicating that clamp opening does not occur upon binding of the non-natural nucleotide. These results collectively indicate that d5-NITP inhibits replicase assembly and subsequent processive DNA synthesis by preventing gp44/62 from opening the closed gp45 trimer.
Structure-activity relationships for nucleotide binding. d5-NITP represents the most potent inhibitor of gp44/62 identified to date as it binds 5- and 300-fold more tightly than other competitive inhibitors such as ATP γS and AMP-PNP, respectively. A structure-activity relationship explaining the unprecedented potency of d5-NITP was developed by testing the ability of the other non-natural nucleotides (Figure 2.1B) to inhibit gp44/62. The data summarized in
Table 1 indicate that binding affinity is not dependent on the shape and size of the non-
natural nucleotide (Figure 2.4) . This is evident as d5-NITP binds with a significantly higher
affinity compared to analogs such as ATP γS and AMP-PNP that are similar in shape and size. In addition, the majority of small non-natural nucleotides such as dITP, d5-AITP, and
d5-FITP bind poorly as their K i values are greater than 200 µM. One notable exception is
d5-EtITP as it inhibits gp44/62 with a K i value of 80 µM. Surprisingly, the closely related
analog, d5-EyITP, binds far worse with a K i greater than 200 µM. The unexpected difference in potency between the two analogs may be caused by entropic factors as the ethyl moiety is more flexible than the ethylene moiety. Indeed, entropic effects appear to play important roles in binding as bulky, hydrophobic analogs such as d5-CHITP and d5-PhITP
bind to gp44/62 with low K i values of ~40 µM. The K i values of these analogs are lower than analogs such as dITP and d5-AITP that more closely resemble ATP and again re-iterate
79 that nucleobase size or shape does not influence binding affinity. In addition, a strong correlation between binding affinity and overall π-electron surface area is not apparent since
the K i value for the electron rich d5-PhITP (K i = 42 µM) is identical to that for d5-CHITP
(K i = 42 µM) which lacks significant π-electron density at the 5-position. It is quite
surprising that the hydrophilic analog, d5-CITP, binds to gp44/62 with a relatively low K i of
37 µM. This provides an interesting paradox as the small hydrophilic nucleotide, d5-CITP,
binds with nearly the same affinity as large, hydrophobic analogs such as d5-CHITP and d5-
PhITP. This dichotomy becomes even more intriguing when one considers that d5-NITP,
an analog possessing both hydrophobic and hydrophilic character, binds 5-fold more tightly
than any of these analogs.
Exploring the active site of gp44/62 . Predictive in silico models of the active site of gp44/62 bound with ATP (Figure 2.5A) or with d5-NITP (Figure 2.5B) were generated to provide more insight into the mechanism of nucleotide binding. In the absence of a structure for gp44/62, the PHYRE (101 ) server was used to create a threaded model based on the structure of P. furiosus RFC (1IQP) with a high degree of confidence ( E value of 2.5 x 10 -27 ).
In this model, gp44/62 binds ATP through interactions with each individual component of the nucleoside triphosphate. The triphosphate moiety interacts with a positively charged arginine (Arg205) as well as G55, K56, and T57 that compose part of the Walker A motif.
The hydroxyl group on the ribose moiety interacts with Arg16 through hydrogen bonding interactions while contacts with the adenine ring are stabilized through π-π stacking interactions with Phe204 and hydrogen-bonding interactions with amide bonds on the adjacent helix.
80 The model of gp44/62 bound with d5-NITP shows many of the same interactions.
Two noticeable differences, however, include more favorable stacking interactions between the indolyl ring and Phe204 as well as potential electrostatic interactions between the nitro group and Arg175. This model shows strong alignment between the two oxygen atoms on the nitro moiety with the guanidinium nitrogen atom of Arg175 that is not present in the model of gp44/62 bound with ATP. As such, the orientation and close proximity (<4Å) between these complementary functional groups could account for the higher affinity of d5-
NITP compared to ATP.
To investigate this mechanism, inhibition constants for d5-NITP were measured using two active site mutants, R175L and R175K (Table 2.2) . The binding affinity for d5-
NITP decreases ~3-fold upon the conservative substitution of arginine with lysine. This small decrease is consistent with a minimal loss of electrostatic interactions between the
oxygen atom of the nitro group and the guanidinium nitrogen. Surprisingly, the K i of 16 µM
for the R175L mutant is identical to14 µM measured with the R175K mutant. This result is unexpected as replacement of the positively charged arginine with the hydrophobic leucine is predicted to lower binding affinity by at least 10-fold. However, hydrophobic interactions between the nitro and leucine may compensate for the loss of the electrostatic interaction.
Another possibility is a compensatory conformational change in the protein that allows additional interactions between the nitro moiety and other amino acids or a water molecule.
However, this possible compensatory mechanism is not evident with hydrophobic analogs such as d5-EtITP and d5-EyITP since their binding affinity to the R175L mutant was observed to increase (vide infra). This shows that a R175L is capable of participating in hydrophobic interactions with non-natural nucleotides; however, the chameleon-like properties of d5-NITP would more readily accommodate an interaction with a water
81 molecule as compared to d5-EtITP or d5-EyITP and a compensatory mechanism cannot be completely ruled out.
The identity in K i values for d5-NITP with the R175K and R175L mutants could reflect its bipolar character as d5-NITP possesses both hydrophilic and hydrophobic properties. This possibility was investigated measuring the inhibition constant for d5-CITP
with these mutants. The K i of 41 µM with the R175K mutant is very similar to the K i of 37
µM with wild-type gp44/62. This minimal change is expected since the carboxyl moiety can form favorable electrostatic contacts with either arginine or lysine. However, a dramatic
effect is observed with the R175L mutant as indicated by the large K i of >200 µM, and indicates that replacing a positively charged amino acid with the hydrophobic leucine abolishes the binding of the negatively charged d5-CITP.
Additional evidence for the role of Arg175 in binding non-natural nucleotides comes
from comparing the K i values for hydrophobic analogs such as d5-EtITP and d5-EyITP
(Table 2.2) . These analogs bind poorly to wild-type and R175K mutant. However, their K i
values are ~10-fold lower in the R175L mutant compared to wild type enzyme. This increase in binding affinity is consistent with a model invoking entropic stabilization of the hydrophobic nucleobase within a hydrophobic active site upon replacement of arginine with leucine. By inference, these data suggest that the high binding affinity of d5-NITP is caused by the diverse chemical nature of the nitro moiety which can interact with positively charged amino acids through enthalpic effects or with hydrophobic amino acids through entropic/desolvation effects. These data suggests that the nitro group is a promiscuous pharmacophore that can blend into different protein environments via equal interactions
with diverse functional groups such as positively charged and hydrophobic amino acids. The
electron withdrawing potential and zwitterionic character of -NO 2 allows it to participate in
82 non-covalent, electrostatic interactions while its double-bond character provides potential interactions though π-cation and π-π stacking arrangements. Finally, the surprisingly low solvation energy of the nitro moiety allows it to interact with molecular targets through entropic effects. These collective properties provide d5-NITP with its superior inhibitory effects against ATP-dependent clamp loaders, especially compared with other competitive inhibitors such as ATP γS and AMP-PCP that accurately mimic ATP. Since the triphosphate moiety of d5-NITP is unmodified, the inhibition by this non-natural nucleotide likely occurs
via non-productive binding that prevents conformational changes in gp44/62 required for proper interactions with gp45. We argue that interactions between d5-NITP and specific amino acids such as Arg175 and Phe204 within the ATP-binding site are responsible for non-productive binding as these interactions do not exist when ATP is bound to gp44/62.
To investigate this further, the K i values for d4-NITP and d6-NITP were measured to evaluate if the position of the nitro pharmacophore impacts nucleotide binding. With wild
type gp44/62, the K i of 34 µM for d4-NITP is ~6-fold higher than that for d5-NITP (5.7
µM) and suggests that placement of the nitro group at the 4-position prohibits favorable
contacts with Arg175 (Table 2.1) . Consistent with this argument, the K i value for d4-NITP
is unaltered when Arg175 is replaced with either lysine or leucine (Table 2.2) . In contrast,
the K i for d6-NITP (5.1 µM) is identical to that for d5-NITP (5.7 µM) and suggests that the binding mode for d6-NITP and d5-NITP are identical (Table 2.1) . However, the model in
Figure 4B argues otherwise as Arg175 does not directly interact with the pharmacophore at the 6-position. Instead, the high binding affinity for d6-NITP likely results from favorable electrostatic interactions with Arg16, another positively charged amino acid in close
83 proximity. Collectively, these data indicate that the position of the nitro pharmacophore influences binding affinity through discrete molecular contacts with active site amino acids. d5-NITP is a selective inhibitor of gp44/62. Clamp loaders across all species serve identical functions by using ATP to assemble their respective DNA replicases (9). In fact, the clamp loaders from bacteriophage T4, E. coli , and eukaryotes all display significant similarity
(~56%) and identity (~33%) in regions that interact with ATP (Figure 2.6) . The similarities in function and active site composition predict that all clamp loaders should display an identical structure-activity relationship for the non-natural nucleotides used in this study.
This hypothesis is inaccurate as the K i values for non-natural nucleotides differ significantly between gp44/62 and the related E. coli γ-complex (Table 2.1) . In fact, gp44/62 binds these analogs with affinities ranging from 5 µM to greater than 200 µM while the γ-complex binds the same analogs with an average affinity of ~20 µM. Compared to gp44/62, the γ-complex binds the majority of analogs with higher affinity. In fact, d5-NITP is the only nucleotide analog that binds more tightly to gp44/62 than to the γ-complex. The difference in binding affinity provides a selectivity factor of 4.5 for the phage clamp loader whereas most other analogs display selectivity factors of <1 (Table 2.1) .
The selectivity of d5-NITP was investigated comparing in silico models of the active site of the γ-complex (1NJF) bound with ATP (Figure 2.5C) or d5-NITP (Figure 2.5D)
with corresponding models of gp44/62 (Figure 2.5A and 2.5B) . Visual inspection reveals
some obvious similarities between the two clamp loaders. The triphosphate moieties of ATP
and d5-NITP interact with amino acids within the Walker A motif (G57, K58, and T59) as
well as Arg63 in the E. coli γ-complex. A major difference between the two clamp loaders,
however, is the absence of an aromatic amino acid in the active site of the γ-complex that
84 can interact with adenine or the indole of d5-NITP. In addition, the γ-complex lacks a positively charged amino acid analogous to Arg175 in gp44/62 that could interact with the nitro pharmacophore. In fact, the active site of the γ-complex (Figure 2.5C) resembles a simple hydrophobic pocket lined with small, aliphatic amino acids including P12, V18, and
V19. This hydrophobic environment could explain why most hydrophobic non-natural nucleotides bind with similar affinities (~20 µM) that appear independent of shape/size, π-
electron surface area, and dipole moment.
Testing the in vivo selectivity of non-natural nucleosides. We hypothesize that the differences in K i
values for d5-NITP between clamp loaders could be exploited to inhibit phage replication
while leaving DNA synthesis in the E. coli host unperturbed. This system provides a simple and convenient model to test non-natural nucleos(t)ides as potential agents that selectively inhibit pathogenic DNA synthesis in a host. This was investigated using a plaque-forming assay to quantify the ability of non-natural nucleosides to inhibit phage infection in an E. coli host (102 ). Data provided in Figure 2.7A shows that E. coli preincubated with 100 µg/mL d5-NI have ~40% fewer plaques compared to untreated E. coli . The protective effect is dose-dependent as treatment with 50 µg/mL d5-NI provides ~2.5-fold more protection
than 25 µg/mL (Figure 2.7B) . The protective effects of the non-natural nucleosides against
phage infection correlate with the inhibitory effects against the bacteriophage T4 clamp
loader. It should be noted that the protective effects of d5-NI appear to be dependent upon
its conversion to d5-NITP. Figure 2.7E shows that the effects of d5-NI are significantly
reduced in a deoxythymidine kinase deficient ( tdk-1) E.coli strain (KY895). This also offers an
explanation for the maximal effect of 40% plaque reduction. At 100 µg/mL, full conversion
of d5-NI to d5-NITP would yield a theoretical maximum intracellular concentration of d5-
85 NITP of 200 µM. While this concentration would be sufficient to completely inhibit all
gp44/62 function, it is highly unlikely that 100% of d5-NI is converted which would limit its
efficacy.
It should be emphasized that the nitro group acts as the primary pharmacophore
since other non-natural nucleosides generate significantly less protective effects. This is
evident as treatment with 150 µg/mL 5-ethylene-indolyl-2'-deoxyribose (d5-Ey) only inhibits
15% of plaque formation while other analogs such as 5-fluoro-indolyl-2'-deoxyribose (d5-
FI) do not inhibit plaque formation at concentrations of 100 µg/mL (Figure 2.7D) . The inability of these analogs to inhibit plaque formation correlates well with the poor potency of the corresponding nucleoside triphosphate to inhibit gp44/62. In addition, none of the non- natural nucleosides tested here produce a significant inhibitory effect on E. coli growth in
suspension (Figure 2.7C) or plated on LB/agar (Figure 2.8) .
The collective in vitro and in vivo data demonstrate that replicative accessory proteins
are bona fide targets for therapeutic intervention against pathological disorders caused by
uncontrollable DNA synthesis. Although inhibiting DNA polymerase activity is the most
logical target for therapeutic intervention, agents that target DNA polymerase activity can
cause toxic side effects by non-selectively inhibiting DNA synthesis in diseased and healthy
cells. Using the simple bacteriophage T4 replication system as a tool, this report illustrates a
way to circumvent this complication by selectively inhibiting DNA synthesis by targeting the
activity of an essential replicative accessory protein. In this respect, the ability of the non-
natural nucleoside, d5-NI, to differentially inhibit bacteriophage DNA synthesis without
affecting E. coli proliferation likely reflects the ability of the corresponding nucleotide to
inhibit ATP-dependent processes involved in assembly of protein complexes at the DNA
replication fork of the bacteriophage. From a pharmacological perspective, this preferential
86 inhibition can be rationalized by simply evaluating the selectivity factor for d5-NITP, defined
as the ratio of its K i value for γ-complex versus gp44/62 (K i host /K i pathogen ). In general, high
values of greater than 100 predict exclusive inhibition of the target enzyme without an influence of the activity of host enzymes. As such, it is quite surprising that d5-NI displays any in vivo selectivity since the calculated in vitro selectivity factor for d5-NITP is only 4.5.
This low value suggests that d5-NITP should elicit an appreciable cytostatic effect against E. coli by inhibiting bacterial DNA synthesis. This dichotomy can be rectified by taking into
account a limitation of "selectivity" which assumes that the Km value for the substrate will be
identical amongst all potential targets. Indeed, selectivity factors can greatly underestimate
the therapeutic potential of a compound if K m values for the substrate differ by only 5-fold amongst various enzyme targets. As such, the therapeutic potential of an inhibitor must take
into account the relationship between the K i for the inhibitor with respect to the K m for the substrate. This relationship, calculated as the ratio of [(K m/K i) pathogen ]/[(K m/K i) host ], defines the s ensitivity factor for an inhibitor. Using the parameters listed in Table 1, the sensitivity factor for d5-NITP against gp44/62 is 20, and this higher value could explain the cytostatic effects of d5-NI against phage proliferation in vivo . Kinetic simulations of this model
(Figure 2.9) indicate that due to differences in K m for ATP, a competitive inhibitor such as d5-NITP has a more pronounced inhibitory effect on the ATPase activity of the bacteriophage clamp loader compared to the γ-complex.
Another possibility is that fundamental differences in the biology of DNA replication between E. coli and the phage invader contribute to the inhibitory effects of d5-
NITP (Figure 2.10) . E. coli replicates its circular genome in a bidirectional manner after initiation from a single origin of replication. This simple mode of replication requires minimal clamp loading events to achieve continuous and uninterrupted leading and lagging
87 strand DNA synthesis. In contrast, bacteriophage T4 DNA synthesis is more complicated as replication of its linear genome occurs in two distinct phase. After bidirectional DNA synthesis commences from fixed locations in the phage genome, there is a switch to recombination-dependent replication (RDR) that produces long concatemers of the phage genome generated via homologous recombination. RDR requires the activity of two additional bacteriophage enzymes, UvsX and UvsY, that utilize ATP to catalyze strand invasion of single stranded 3’ ends of DNA into homologous regions of duplex DNA (D).
After replication of the resulting ‘D-loop’ structures, the concatemers are processed into smaller genomic segments by terminase prior to packaging into new phage particles. Since efficient concatemeric replication depends upon a high frequency of clamp loading events, the bacteriophage is predicted to be more sensitive to the inhibitory effects of d5-NITP. In addition, d5-NITP could inhibit multiple targets associated with bacteriophage replication and/or infection including other ATP-dependent enzymes such as UvsX, UvsY, gp59 (DNA helicase), and gp 61 (helicase loader) that are required for RDR. Thus, d5-NITP could function as a "magic-shotgun" by inhibiting multiple proteins associated with bacteriophage
DNA replication. These data support a new approach to disrupt the activity of a specific replicative accessory protein in various DNA- and RNA-dependent viruses .
88 Figures
Figure 2.1 Natural and non-natural nucleobases used in this study. (A) Comparison of the structures of adenine and 5-nitroindole (B) Library of all non-natural nucleobases. All models were constructed using Spartan ’04 software (www.wavefun.com). The electron density surface potentials of adenine and non-natural nucleobases were then generated. The most electronegative regions are in red, neutral charges are in green, and the most electropositive regions are in blue.
89 Figure 2.1
90 Figure 2.2 Hydrolysis of nucleotide substrates by gp44/62 and the effect of r5-NITP and
d5-NITP on gp44/62 ATPase activity. (A) Hydrolysis of various nucleotide substrates by
gp44/62 quantified by colorimetric ATPase assay. Assay conditions are listed in methods
section and concentration of all substrates was 500 µM (B) Dose-dependent inhibition of
gp44/62 ATPase activity by d5-NITP (▲) and r5-NITP (■) K i values are 4.8 ± 0.5 µM and
10.8 ± 0.7 µM, respectively. (C) Double reciprocal analysis of d5-NITP versus gp44/62.
Competitive inhibition was determined by fitting data to the following equation: rate =
Vmax [S]/K m(1+[I]/K i) + [S]. No inhibitor (■), 5 µM d5-NITP (▲), 25 µM d5-NITP (▼),
and 50 µM d5-NITP (♦). Inset: K i value of d5-NITP determined from double reciprocal data. Slope of each line ([ATP]/rate) was plotted against [d5-NITP]. The x-intercept defines
the K i value (5.7 µM).
91 Figure 2.2
92 Figure 2.3 Effects of d5-NITP on T4 replicase formation and gp45 opening. (a) Diagram of
strand displacement assay. DNA polymerase alone (I) is unable to extend the primer beyond
the forked strand (longest product: 44-mer). DNA polymerase in the presence of accessory
proteins (II) is able to extend the primer beyond the forked strand up to the abasic site (SP)
present on the 51 st nucleotide of the template (longest product: 50-mer). When an ATPase
inhibitor is present (III), the accessory proteins are unable to facilitate replicase formation
and extension beyond the forked strand is not observed (longest product: 44-mer). (b) Gel
images of DNA substrate alone (lane 1), polymerase alone (lane 2), polymerase with
accessory proteins (lane 4), polymerase alone with 100 µM d5-NITP (lane 3), and
polymerase with accessory proteins and 100 µM d5-NITP (lane 5) (c) Quantification of
products in strand displacement assay (d) FRET quenching of W91 by CPM in the presence
of either 1 mM ATP or d5-NITP.
93 Figure 2.3
94 Figure 2.4 Plot of K i values for various non-natural nucleotides as a function of their respective nucleobase size. The low r 2 value of 0.4591 indicates a poor correlation between binding affinity and nucleobase size.
95 Figure 2.4
210 Ind 5-FITP 5-EyITP 200 190 5-AITP 180 170 160 150 140 130 120 M) 110 µ µ µ µ
( 100 i
K 90 5-EtITP 80 70 60 5-CHITP 50 5-PhITP 40 5-CITP 30 20 5-CEITP 10 5-NITP 0 120 130 140 150 160 170 180 190 200 210 220 230 240 250 Volume (Å 3)
96 Figure 2.5 Molecular modeling of the active sites of gp44/62 and γ-complex.. (A) ATP or
(B) d5-NITP bound to T4 bacteriophage gp44/62. (C) ATP or (D) d5-NITP bound to E.coli
γ-complex.
97 Figure 2.5
98 Figure 2.6 Alignment of the ATP-binding region of the bacteriophage T4 gp44/62,
Saccharomyces cerevisiae RFC, and Escherichia coli γ-complex. The sequences compared were
obtained from the NCBI protein database, bacteriophage T4 gp44/62 (NCBI Reference
Sequence: NP_049665.1), S. cerevisiae RFC, (PDB: 1SXJD), and E. coli γ-complex (PDB:
1NJFD). The NCBI protein Basic Local Alignment Search Tool (BLAST) was used to align
the sequence of RFC and γ-complex to gp44/62. The ATP binding region compared was defined by the 25 amino acid segment containing the Walker A motif (GKT). Green indicates identity and blue indicates similarity in amino acid composition. The three sequences display an overall similarity of 28% and an overall identity of 24%. gp44/62 and γ- complex share 36% identity and 48% similarity. Comparison between gp44/62 and RFC reveals 44% identity and 68% similarity. Comparison between the γ-complex and RFC
reveals 48% similarity and 48% identity.
99 Figure 2.6
100 Figure 2.7 Inhibition of T4 plaque formation by d5-NI. (A) Effects of 100 µg/ml d5-NI on
T4 Bacteriophage plaque formation, arrows indicate plaques. (B) Graphical quantification of plaque reduction by d5-NI, n=4, **p<0.01 vs others. (C) Effects of 100 µg/ml d5-NI (♦),
d5-EyI (▲) and ampicillin (●) compared to a DMSO control treated normal growth curve
of E.coli (■). (D) Comparison of the plaque-reducing effects of d5-NI with d5-EyI, d5-FI, and chloramphenicol n=4 **p<0.01 vs others. (E) Effects of 100 µg/ml d5-NI on plaque
formation in E.coli JA300 and KY895 ( tdk-1) n=3 *p<0.05.
101 Figure 2.7
102 Figure 2.8 d5-NI is not bactericidal or bacteriostatic against E. coli . Cytotoxicity of 50 µg/ml chloramphenicol (bottom row) and 100 µg/ml d5-NI (middle row) compared to DMSO
control (top row). A stationary phase culture was diluted 1:250 and allowed to grow for 2.5
hours. Cultures were then treated and allowed to grow for an additional hour. Cultures were
diluted 1:10 2, 1:10 4, 1:10 6, and 1:10 8 (left to right) and plated on LB/agar for 24 hours.
Comparison of the cells displayed in the far right lane, it is clear that chloramphenicol is a bactericidal agent since there are fewer cells compared to cells treated with DMSO (vehicle).
Likewise, d5-NI does not generate a bactericidal effect as the cell density and viability remains invariant compared to cells treated with DMSO (vehicle).
103 Figure 2.8
104 Figure 2.9 Kinetic model for the effects of a competitive inhibitor on the concurrent
activity of two independent enzymes that possess different K m values for a common substrate and different K i values for a common competitive inhibitor. KinTekSim v2.03 was used to generate the all kinetic models using the following minimal kinetic mechanism:
k+1 k+2 k+3 E + S ES EP E + P k-1 k-2 k-3
k+4 k+5 k+6 F + S FS FQ F + Q k-4 k-5 k-6
k+7 E + I EI k-7
k+8 F + I FI k-8
In this mechanism, E represents the γ-complex, F represents gp44/62, S represents the substrate (ATP), I represents the inhibitor (d5-NITP), P represents products generated by the γ-complex, and Q represents products generated by gp44/62. Since clamp loading and
ATP hydrolysis are directly coupled, the amount of product formed (P or Q) is equivalent to
the frequency of clamp loading in the system. Inhibitor concentrations for all simulations
were varied at 0, 100, 200, and 500 µM. The substrate binding step was set equal to the
published K d values of ATP for γ -complex (K d = 0.9 µM) (103 ) and gp44/62 (K d = 20 µM)
(104 ). Product release steps, k 3 and k 6, were set equal to the k cat for γ-complex (k cat = 1.4 s-1)
(53 ) and gp44/62 (k cat = 0.4 s-1) (51 ), respectively. The K i value for d5-NITP against the γ- complex and gp44/62 were 22 µM and 5 µM, respectively.
105 Model 1. The effects of 5-NITP on clamp loaders possessing different k cat and K m values.
650 5-NITP 0 - 500 µM
600
550
500
450 M) µ µ µ µ 400
350
300
250
ATP Hydrolyzed ( ATP Hydrolyzed 200 5-NITP 0 - 500 µM 150 100 50 0 0 10 20 30 40 50 60 70 Time (sec)
This simulation represents the effects of d5-NITP on γ-complex (red) and gp44/62 (blue)
ATPase activity using the equilibrium and kinetic parameters listed in the adjacent table. All
concentrations are in micromolar (µM). In this simulation, the k cat values for gp44/62 and γ- complex (k+3 and k+6) are set equal to their reported values. Although the catalytic activity of gp44/62 is lower than that for the γ-complex, the inhibitory effect of d5- NITP on gp44/62 is still more prevalent compared to the γ-complex. d5-NITP displays a minimal effect on the ATPase activity of the γ-complex ATPase activity. In contrast, a dose- dependent decrease in gp44/62 ATPase activity is observed when the concentration of d5-
NITP is increased.
106 Model 2. The effects of d5-NITP on the bacteriophage and E. coli clamp loaders with
identical k cat values but with different K m values.
550 5-NITP 0 - 500 µM 500
450 5-NITP 0 - 500 µM 400 M) µ µ µ µ 350
300
250
200 ATP Hydrolyzed ( ATP Hydrolyzed 150 100
50
0 0 10 20 30 40 50 60 Time (sec)
The above simulation represents the effects of d5-NITP on γ-complex (red) and gp44/62
(blue) ATPase activity using the equilibrium and kinetic rate constants listed in the adjacent
table. All concentrations are provided in micromolar (µM). In this simulation, the k cat values for gp44/62 and γ-complex (k+3 and k+6) are set equal to each other to illustrate the degree
of inhibition by d5-NITP on each respective enzyme. The inhibitory effects of d5-NITP on
gp44/62 is more prevalent compared to the γ-complex as a result of the 5-fold higher K m
value for ATP.
107 Model 3. The effects of variable concentrations of d5-NITP on clamp loaders at different protein concentrations.
750 5-NITP 0 - 500 µM 700 650 600 550 500 M) µ µ µ µ 450 400 350 300 5-NITP 0 - 500 µM 250 ( ATP Hydrolyzed 200 150 100 50 0 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 Time (sec)
The above simulation shows that the sensitivity of gp44/62 is independent of enzyme concentration. Conditions are identical to Model #2 and differ only by a 10-fold greater concentration of gp44/62 (species F) compared to γ-complex (species E). In this case, the
catalytic activity of gp44/62 is higher than the γ-complex due to the higher enzyme
concentration. However, the inhibitory effect of 5-NITP on gp44/62 is still more prevalent
compared to the γ-complex. d5-NITP displays a minimal effect on the ATPase activity of the γ-complex ATPase activity. In contrast, a dose-dependent decrease in gp44/62 ATPase activity is observed when the concentration of d5-NITP is increased.
108 hydrolysis are directly coupled, the amount of product formed (P or Q) is equivalent to the frequency of clamp loading in the system. Inhibitor concentrations for all simulations were
varied at 0, 100, 200, and 500 µM. The substrate binding step was set equal to the published
Kd values of ATP for γ-complex (K d = 0.9 µM) (103 ) and gp44/62 (K d = 20 µM) (104 ).
Product release steps, k 3 and k 6, were set equal to the k cat for γ-complex (k cat = 1.4 s-1) (53 )
and gp44/62 (k cat = 0.4 s-1) (51 ), respectively. The K i value for 5-NITP against the γ- complex and gp44/62 were 22 µM and 5 µM, respectively.
109 Figure 2.10 Comparison of the DNA replication mechanisms of E.coli and T4
bacteriophage. (A) Inset of E.coli genome. Circular genome is replicated in a single event
from one origin of replication using a bidirectional replication fork. Bidirectional fork shown
in blue and red arrows, solid arrows represent leading strands and dashed arrows are lagging
strands, DNA polymerase is shown in brown, γ-complex in red, and β-clamp in blue. (B)
Scheme of T4 Bacteriophage replication. Origin dependent replication is followed by “join- cut-copy” recombination dependent replication (RDR). (I) T4 phage initiates origin dependent replication of its linear genome; (II) products have RDR competent single stranded 3’ ends coated in single stranded binding protein (green) and UvsY (dark red). (III)
UvsX (bright red) displaces single stranded binding protein and 3’ ends invade homologous regions of double stranded DNA (join), 3’ ends act as leading strand primers. Arrow points to endonuclease VII cut site (cut). (IV) Replication proceeds (copy), dashed lines represent newly synthesized DNA, and solid lines are templating DNA. DNA polymerase shown in blue, gp45 in light blue, and gp44/62 in grey. One round of RDR yields duplex DNA exceeding the length of the original template, and RDR competent intermediates. (V)
Multiple rounds of RDR results in a long concatemeric structure. (VI) Cleavage of concatemer and packaging of phage heads is carried out by the terminase complex.
110 Figure 2.10
111 Tables
Table 2.1 . Summary of inhibition constants for natural and non-natural nucleotides against the ATP-dependent clamp loaders from bacteriophage T4 (gp44/62) and Escherichia coli (γ-
complex).
Nucleotide gp44/62 γ-complex Selectivity a b c Analog K i ( µM) K i ( µM) Factor ______ATP γS 28.9 ± 11.6 13.0 ± 4.5 0.44 dITP <200 d 60.9 ± 19.5 0.30 d5-AITP <200 45.0 ± 18.5 0.23 d5-FITP <200 34 ± 4 0.17 d5-EtITP 81.5 ± 17.0 9.0 ± 3.2 0.11 d5-EyITP <200 30.0 ± 15.7 0.15 d5-CITP 37 ± 8 40 ± 13 0.93 d5-NITP 4.8 ± 0.5 21.7 ± 2.1 4.52 d5-CHITP 47.5 ± 10.0 24.0 ± 10.5 0.51 d5-CEITP 10.0 ± 1.6 11.8 ± 1.5 1.18 d5-PhITP 42 ± 10 7.5 ± 1.1 0.18 d4-NITP 34.2 ± 6.7 11.1 ± 2.4 0.32 d6-NITP 5.1 ± 1.4 8.1 ± 2.3 1.59 ______a Reactions were performed using 500 nM gp44/62 and gp45, 10 mM Mg 2+ , 100 µM ATP, 1 µM DNA. The concentration of nucleotide was varied from 0.5-400 µM. Assays were performed at 25°C. Initial rates in ATP consumption were obtained from the time courses that were linear over the time frame measured (120 seconds). IC 50 values were converted to dissociation constants (K i) using equation (3). b Reactions were performed using 100 nM γ-complex and β-clamp, 10 mM Mg 2+ , 20 µ M ATP, 1 µM DNA. The concentration of nucleotide was varied from 0.5-400 µM. Assay was performed at 37°C. Initial rates in ATP consumption were obtained from the time courses
112 that were linear over the time frame measured (120 seconds). IC 50 values were converted to dissociation constants (K i) using equation (3). c Selectivity Factor = K i ( γ-complex) / K i (gp44/62). d No inhibition was observed at nucleotide concentrations greater than 200 µM. e Not determined
113 Table 2.2 Summary of inhibition constants for various non-natural nucleotides against wild- type, R175K, and R175L mutants of gp44/62.
Nucleotide Wild-type R175K R175L a Analog Ki ( µM) K i ( µM) K i ( µM) ______ATP γS 28.9 ± 11.6 14.0 ± 3.5 33.2 ± 8.0 d5-NITP 4.8 ± 0.5 13.8 ± 4.2 16.4 ± 2.4 d5-CITP 37 ± 8 40.6 ± 17.6 <200 b d5-EtITP 81.5 ± 17.0 25.2 ± 4.2 11.0 ± 3.5 d5-EyITP <200 26.2 ± 9.0 34 ± 14 d4-NITP 34.2 ± 6.7 30.2 ± 7.4 41.6 ± 10.9 d6-NITP 5.1 ± 1.4 9.1 ± 2.1 5.0. ± 1.8 ______a Reactions were performed using 500 nM gp44/62 and gp45, 10 mM Mg 2+ , 100 µM ATP, 1 µM DNA. The concentration of nucleotide was varied from 0.5-400 µM. Assays were performed at 25°C. Initial rates in ATP consumption were obtained from the time courses that were linear over the time frame measured (120 seconds). IC 50 values were converted to dissociation constants (K i) using equation (3). b No inhibition was observed at nucleotide concentrations greater than 200 µM.
114 Chapter 3: A Novel Non-Natural Nucleoside that Influences P- glycoprotein Activity and Mediates Drug Resistance
Kevin T. Eng and Anthony J. Berdis #
Department of Pharmacology, Case Western Reserve University, 10900 Euclid Ave
Cleveland OH 44106
Reproduced in part with permission from [Eng KT, Berdis AJ. A novel non-natural
nucleoside that influences P-glycoprotein activity and mediates drug resistance. ACS
Biochemistry. 2010 Mar 2;49(8):1640-8.]. Copyright [2010] American Chemical Society.
3.1 Abstract
Multi drug resistance during cancer chemotherapy is commonly acquired by overexpression of the ATP binding cassette transporter, P-glycoprotein (P-gp). As such, inhibitors that target P-gp activity represent potential therapeutic agents against this form of drug resistance. This study evaluated the ability of various non-natural nucleosides that mimic the core structure of adenosine to modulate drug resistance by inhibiting the ATPase activity to
P-gp. Of several analogs tested, only one novel non-natural nucleoside, 5-cyclohexyl-indolyl
2’-deoxyribose (d5-CHI), behaves as a P-gp inhibitor. Although d5-CHI is an adenosine analog, it surprisingly stimulates the ATPase activity of P-gp in vitro . However, d5-CHI is not
a true substrate for P-gp as it is not transported across an MDCK-MDR1 monolayer. In
addition, d5-CHI differentially modulates MDR by decreasing or increasing the cytotoxicity
of several chemotherapeutic agents. This modulation directly correlates with changes
115 observed in the drug-stimulated ATPase catalytic efficiency induced by d5-CHI. The collective data set are used to develop a rapid and accurate in vitro method for predicting drug-drug interactions with P-gp.
3.2 Introduction
P-glycoprotein (P-gp) is a member of the ATP binding cassette (ABC) transporter family that is encoded by the MDR1 gene (105-106 ). P-gp utilizes the energy derived from ATP
binding and hydrolysis to export xenobiotics from the intracellular space. Although the
normal physiological role of P-gp is to defend the cell against potentially toxic chemical
entities, amplification of the MDR1 gene leads to development of the multi-drug resistance
(MDR) phenotype in many types of cancers (107 ). This occurs since P-gp efficiently exports
a number of structurally diverse chemotherapeutic agents (Figure 1.13) (61 ). Indeed,
amplification of MDR1 strongly correlates with a poor prognosis in many forms of cancer
(69 ). As a consequence, P-gp is a major clinical target in cancer chemotherapy (108 ).
At the cellular level, functional P-gp exists as a dimer containing two transmembrane
domains and two nucleotide-binding domains (109 ). Substrate transport is an ATP-
dependent process that follows a sequential mechanism. As mentioned in chapter 1, two
major models attempt to describe the role of ATP in substrate transport. The first model is
the ATP switch model (Figure 1.16) (77 ), and the second model is the occluded nucleotide
model (Figure 1.17) .
Perhaps the most recognized feature of P-gp is its polyspecificity during drug
transport. In this regard, P-gp facilitates the export of structurally diverse chemical entities
such as vinblastine, doxorubicin, and digoxin. While these compounds do not share a
common structural feature, they are all considered to be relatively hydrophobic. Indeed, this
116 common feature is often used to justify why a drug-like compound behaves as a substrate for P-gp. The most widely accepted model accounting for the polyspecificity of P-gp is the
“hydrophobic vacuum cleaner” model originally proposed by Higgins and Gottesman (71 ).
In this model, the hydrophobic nature of a potential P-gp substrate facilitates partitioning
into the lipid membrane. At this point, the substrate enters the transmembrane domains of
P-gp and is transported out of the cell. Indeed, the recently described X-ray structure of P-
gp is consistent with this model as there are four α-helices in the transmembrane domains of
P-gp that could function as an entry portal for a substrate since they are directly accessible to the plasma membrane (74 ).
Attempts to generate inhibitors of P-gp have primarily focused on developing compounds that prevent entry of a substrate into the transmembrane domains of P-gp. This approach has met with some success as compounds such as VX-710 and S9788 can block the export of chemotherapeutic agents including vinsblastine and doxorubicin (110-111 ).
However, significant limitations in this approach exist as most agents that inhibit the export
of a P-gp substrate often serve as alternative substrates for P-gp themselves and can cause
serious alterations in the pharmacokinetics of co-administered drugs (112 ). Thus, developing
P-gp inhibitors represents an important yet daunting challenge in drug design.
Since substrate transport depends upon ATP binding and hydrolysis, an alternative
approach is to develop small molecule inhibitors that target the ATP binding site of P-gp.
Targeting the ATP binding site provides a new set of challenges that include potency and
selectivity. However, there are several examples in which a small molecule has been designed
to selectively inhibit various ATP-dependent enzymes. A classical example of this is Imatinib
(Gleevac). Imatinib is used to treat chronic myelogenous leukemia by selectively targeting the
BCR-ABL fusion protein in its kinase domain (17 ). Another less selective kinase inhibitor is
117 Seliciclib (Roscovitine) which is known to target several cell cycle dependent kinases such as
CDK2 (113 ), CDK7, and CDK9 (114 ). Further supporting this approach is the most recent
P-gp inhibitor tariquidar (82 ) which has been shown to have superior potency and a unique mechanism of being able to bind to multiple sites on P-gp.
This report outlines our efforts to achieve similar pharmacological effects against P-
gp by using various non-natural nucleosides (Figure 3.1) that mimic the core structure of
adenosine. This small library of structurally related nucleoside analogs was screened as
inhibitors of P-gp activity using a cell-based assay. Of nine structural related compounds
tested, only 5-cyclohexyl indolyl 2’-deoxyribose (d5-CHI) produced any ex vivo inhibitory
effects against the transport activity of P-gp. Although the non-natural nucleoside influences
the export of several known substrates of P-gp, the inhibitory effect is not caused by
perturbing the ATPase activity of P-gp. In fact, d5-CHI potently stimulates P-gp ATPase
activity, displaying defined Michaelis-Menten kinetic behavior. The non-natural nucleoside
does not behave as a typical P-gp substrate as it is not transported across a cellular
membrane under ex vivo conditions. In addition, d5-CHI differentially influences the catalytic
efficiency of drug-stimulated ATPase activity for several important chemotherapeutic agents.
In this regard, an accurate correlation is observed between the effects of d5-CHI on the
ATPase activity of P-gp with its ability to affect the efflux capability of the drug transporter.
Collectively, these data are used to develop a model directly correlating changes in the
catalytic efficiency of drug-stimulated ATP hydrolysis activity with the modulation of drug
resistance in MDR cells which can be used to predict multiple drug interactions with P-gp.
3.3 Materials and Methods
118 Materials. All non-natural nucleosides and nucleotides were synthesized as previously described (27-30 ). Doxorubicin, paclitaxel, and cyclosporine A were purchased from Tocris
Bioscience (Ellisville, MO). Verapamil, vinbastine, and colchicine were purchased from
Sigma-Aldrich (St. Louis, MO). Calcein-AM was purchased from Calbiochem (San Diego,
CA). All other chemicals were purchased from high quality vendors. PEI cellulose TLC plates were purchased from EM Science (Gibbstown, NJ). P-gp membranes were purchased from BD Biosciences (San Jose, CA). Transwell inserts (polycarbonate, 12-well, 11 mm diameter, 0.4-µm pores) were purchased from Corning Costar (Cambridge, MA). KB-V1,
KB-3-1, MDCK-MDR1, and MDCK cells were a generous gift from Dr. Michael
Gottesman (National Cancer Institute, NIH). All cell culture media and supplements were
purchased from Invitrogen (Carlsbad, CA).
ATPase Assays and Measurement of Kinetic Parameters. ATPase activity was monitored by hydrolysis of [ γ-32 P]-ATP. Reaction buffer consisted of 1 mM ATP, 50 mM Tris-MES, 2
mM EGTA, 2 mM DTT, 50 mM KCl, and 5 mM sodium azide. All reactions were carried
out at 37°C in triplicate. Reactions were initiated by addition of 5 µg of P-gp membranes and quenched at 5 and 20 minutes in an equal volume of 10% SDS. Quenched samples were
analyzed by thin layer chromatography on PEI-cellulose plates using 0.6 M KH 2PO 4 pH 3.5.
Imaging of thin layer chromatography was done using a Packard Cyclone PhosphorImager.
32 32 The ratio of free Pi to non-hydrolyzed [ γ- P]-ATP was multiplied by the final concentration of ATP to obtain total product concentration. Product formation in the absence of enzyme was measured and subtracted from all measurements. Initial velocities
were obtained by fitting the data to equation 1.
y = mx + b (1)
119
y is product concentration, x is time, m is the slope, and b is the y-intercept. K m and V max
values were obtained by fitting initial velocities to equation 2.
v =V max *[S]/(K m+[S]) (2)
v is velocity, V max is the maximal velocity, S is substrate concentration, and K m is the
Michaelis-Menten constant.
Cell culture techniques. All cell lines were generously provided by Dr. Michael Gottesman
(NIH). All cells were cultured at 37°C in humidified air with 5% CO 2. KB-3-1 cells were grown and maintained in Dulbecco’s Modified Eagle Media with 10% FBS, 2 mM L- glutamine, 4500 mg/L glucose, 110 mg/L sodium pyruvate, 100 µg/ml penicillin/streptomycin. Doubling time was approximately 24 hours. Cells were split weekly at a ratio of 1:4. KB-V1 cells were grown and maintained in the same conditions with the addition of 100 nM vinblastine. MDCK cells were grown and maintained in Minimum
Essential Media with 10% FBS, 2 mM L-glutamine, and 100 µg/ml penicillin/streptomycin.
Cells were split twice weekly at a ratio of 1:10. MDCK-MDR cells were grown and
maintained in the same conditions with the addition of 200 nM colchicine.
Monolayer Efflux Experiments. MDCK-MDR and MDCK cells were seeded in polycarbonate transwell membranes at a density of 300,000 cells/cm 2. Media was changed every other day and monolayers were ready for experimentation 7 days after plating. For a typical transport experiment, monolayers were washed and preincubated with transport buffer (Hanks
120 Balanced Salt Solution) for 30 minutes. Transport was initiated by the addition of drug to the donor well and assays were conducted for 60 minutes. Drug transport was measured in the
BL→AP direction and final drug concentrations in the donor and receiver wells were
quantified via spectroscopic measurements using the following parameters (d5-CHI λmax =
-1 -1 -1 -1 266 nm, ε266 = 6,840 M cm ; calcein-AM: λmax = 496 nm, ε496 = 6,000 M cm ; rhodamine
-1 -1 123: λmax = 500 nm, ε500 = 75,000 M cm ). Permeability coefficients were obtained from equation 3:
Peff = 1/AC o (dQ/dt) (3)
Peff is permeability coefficient, A is membrane surface area, C o is initial drug concentration in
donor well, and (dQ/dt) is the amount of drug transported over a given period of time.
Measure of in vivo Cytotoxicity. The cytotoxicity of d5-CHI and other chemotherapeutic agents
was quantified using a standard MTT assay. Cells were seeded at 20,000 cells/well in a 96-
well plate. Cells were incubated for 24 hours after plating to ensure proper adhesion. At the
time of drug treatment, 10% FBS media was replaced with 2.5% FBS media, and DMSO
content did not exceed 0.4%. Raw data was normalized to 100% viability (no treatment) and
0% viability (lethal dose) and LD 50 values were obtained through a fit of the data to equation
4.
n y = 100/(1+(LD 50 /I) ) (4)
121 y = fractional viability, I is the concentration of drug, and LD 50 is the concentration of drug that causes 50% cell death compared to cells treated with vehicle, and n is the Hill coefficient.
3.4 Results
Identification of d5-CHI as a modulator of P-gp activity. Conventional P-gp inhibitors are generally
hydrophobic compounds that target the drug export site of the transporter. Our attempts to
develop an alternative class of inhibitor focused on targeting the ATP binding site of P-gp.
We used the non-natural nucleosides depicted in Figure 3.1 since these analogs mimic the
core structure of adenosine. In addition, we have previously demonstrated that the
triphosphate form of certain non-natural nucleosides can act as either surrogates of dATP
(34 ) or as inhibitors of ATP-dependent enzymes (data unpublished). To test the ability of these various non-natural nucleosides to inhibit P-gp, we submitted them to the NIMH
PDSP screening facility (http://pdsp.med.unc.edu/indexR.html). This facility screens compounds using a cell-based assay that measures the accumulation of calcein-AM, a highly selective substrate for P-gp, in Caco-2 cells. In cells that overexpress P-gp, the non- fluorescent calcein-AM is actively exported and thus generates a low fluorescence signal.
However, inhibition of P-gp causes calcein-AM to accumulate within the cell. Hydrolysis by intracellular esterases converts calcein-AM into calcein which is highly fluorescent, and the production of a fluorescence signal indicates a functional inhibition of drug efflux. From a screen of these structurally related non-natural nucleosides, d5-CHI was identified as the only compound that inhibited the efflux of calcein-AM with an efficacy equal to cyclosporine A (CsA) (data not shown).
122 Effects of d5-CHI on P-gp ATPase activity. Although d5-CHI inhibited calcein-AM efflux, this effect could be caused by several mutually exclusive mechanisms. These mechanisms include the ability of the non-natural nucleoside to act as a competitive inhibitor for ATP binding, as an alternative substrate for efflux, or as a non-transported substrate. To differentiate amongst these possibilities, we measured the effect of d5-CHI on the ATPase activity of P- gp. If d5-CHI acts as an ATP competitive inhibitor, then the low basal ATPase activity of P- gp should be reduced even further by the addition of increasing concentrations of d5-CHI.
Conversely, the ATPase activity of P-gp should be stimulated if d5-CHI behaves as either an alternative or non-transported substrate.
The ATPase activity of P-gp was measured in the presence of increasing concentrations of d5-CHI using a single, fixed concentration of 1 mM ATP. Surprisingly, the rate of ATP hydrolysis increased as a function of increasing concentrations of d5-CHI.
Initial velocities were plotted as a function of d5-CHI concentration (Figure 3.2A) and the
data were fit to the Michaelis-Menten equation to define K m and V max values. The data
illustrated in Figure 2A shows that d5-CHI potently stimulates P-gp ATPase activity with a
Km of 60 +/- 10 nM. The selectivity of d5-CHI was interrogated by testing the ability of
other structurally related analogs such as d5-CEI and d5-PhI to influence the ATPase
activity of P-gp. As illustrated in Figure 3.2B , neither compound displays any stimulatory or
inhibitory activity against the ATPase activity of P-gp. Other non-natural nucleosides also do
not stimulate the ATPase activity of P-gp (data not shown). These results are consistent with
data obtained using the calcein-AM screening assay (vide supra) indicating that these analogs
do not interact with P-gp.
To compare the potency of d5-CHI with other classical P-gp substrates, K m and V max
values were measured for compounds such as paclitaxel, vinblastine, doxorubucin,
123 colchicine, verapamil, and cyclosporine A (Table 3.1) . While all of these agents stimulate P- gp ATPase activity with Michaelis-Menten kinetics, none are as potent as d5-CHI. In fact, the potency for these substrates varies by several orders of magnitude (Table 3.1) with d5-
CHI being the most potent substrate (K m = 60 nM) and colchicine being the least potent
(K m = 37 µM). Despite vast differences in these K m values, the V max for all substrates showed
less variation with an average value of ~5 nmols P i/min/mg protein. Collectively, these data
strongly suggest that d5-CHI does not behave as a competitive inhibitor against ATP but
rather as an alternative or a non-transported substrate.
d5-CHI is a non-transported substrate of P-gp. To differentiate between these mechanisms, the
ability of P-gp to transport d5-CHI was directly measured using the monolayer efflux assay
with the MDCK and MDCK-MDR1 cell lines. Substrate transport was measured by
calculating the Peff in the basolateral (BL) to apical (AP) direction and vice versa. In general,
the ratio of Peff values in the BL→AP to AP→BL defines the degree of P-gp mediated transport (75 ). Since P-gp is localized to the surface of these cells, positive ratios indicate
polarized P-gp mediated transport. To define the degree of transport, these ratios are
compared against values obtained with known substrates such as calcein-AM and rhodamine
123. Using the MDCK-MDR1 cell line, calcein-AM and rhodamine 123 display Peff ratios of
2940 and 72.7, respectively. These high values validate that they are indeed substrates of P-
gp. In contrast, the P eff ratio for d5-CHI in the MDCK-MDR1 cell line is significantly lower at 11.1 (Table 3.2) and suggests that the non-natural nucleoside is not exported by P-gp. In
addition, a nearly identical Peff ratio is obtained using the MDCK cell line that does not
overexpress MDR1 (Table 3.2) . To further validate that d5-CHI is a non-transported substrate, we tested the ability of d5-CHI to block the transport of calcein-AM across
MDCK-MDR1 monolayers. As predicted, d5-CHI completely blocks the transport of
124 calcein-AM as its Peff ratio is reduced from ~3,000 in the absence of d5-CHI to ~1 in the
presence 10 µ M of the non-natural nucleoside (Table 3.2) .
d5-CHI is imported by the nucleoside transporter. The low Peff ratio of d5-CHI could reflect passive diffusion through the monolayer. Alternatively, d5-CHI could be imported into the cell by the action of a nucleoside transporter. Measuring the effects of adenosine on the permeability of d5-CHI assessed this possibility since it is a high affinity substrate for the nucleoside transporter (115 ) that should block the transport of other potential substrates.
The data provided in Table 3.2 validate this mechanism as the Peff ratio for d5-CHI decreases from 11.1 in the absence of adenosine to that of 1.56 in its presence. A similar result is observed using the MDCK cell line that does not overexpress P-gp (Table 3.2) .
Collectively, these results indicate that d5-CHI is imported via a nucleoside transporter rather than via passive diffusion as expected for a typical hydrophobic compound.
Correlating ATPase activity with modulation of drug export. The data suggest that d5-CHI is a non- transported substrate that stimulates the ATPase activity of P-gp. If correct, then the binding of d5-CHI should block the binding of another substrate such as calcein-AM. The resulting
inhibition would be reflected in a decrease in the V max /K m value in the presence of d5-CHI.
For a competitive inhibitor (116 ), the decrease in the overall catalytic efficiency would be
caused by an increase in the K m for calcein-AM while the V max for the reaction remains unchanged. To validate this approach, the kinetic parameters for stimulated ATPase activity
by calcein-AM efflux were measured in the absence and presence of K m concentrations of
d5-CHI. The double reciprocal plot shown in Figure 3.3 reveals that the K m for calcein-AM
increases ~7-fold in the presence of d5-CHI (compare K m values of 3.1 ± 0.4 µM versus
125 22.5 ± 6.5 µM, respectively). Similar results are obtained using the known P-gp inhibitor,
CsA, in which the K m for calcein-AM increases significantly in the presence of 400 nM CsA
(compare K m values of 3.1 ± 0.4 µM versus 20.9 ± 8.7 µM, respectively). In addition, nearly
identical V max values are obtained with calcien-AM in the absence (9.2 ± 0.2 µM/min) or
presence of d5-CHI (9.4 ± 1.3 µM/min) or CsA (8.0 ± 1.6 µM/min). Collectively, the ability
of d5-CHI to influence the K m of another substrate without affecting its V max is consistent
with a competitive inhibition mechanism and validate the results from the PDSP screening efforts (vide supra).
Vmax /K m values for other P-gp substrates were measured in the absence and presence of d5-CHI. These values, summarized in Table 3.3, are useful in defining the potential interactions of d5-CHI with the second substrate. This analysis is based on comparing the
ratio of V max /K m value measured with drug and d5-CHI (V max /K m(drug+d5-CHI) compared to the
Vmax /K m of drug alone (V max /K m(drug) ). In this analysis, a ratio less than 1 represents a decrease in activity caused by d5-CHI inhibition while a ratio greater than 1 indicates an increase in activity that could reflect cooperative efflux. As expected, the ratio using d5-CHI against calcein-AM as the substrate is less than 1. Nearly identical results are obtained using CsA as the inhibitor to influence the ATPase activity by calcien-AM. In fact, identical ratios of
~0.14 indicate that both d5-CHI and CsA inhibit the efflux of calcein-AM.
The effect of d5-CHI on the efflux of vinblastine, doxorubicin, colchicine, paclitaxel, and verapamil was next investigated by determining these ratios in the catalytic efficiencies in
ATPase activity. As reported in Table 3.3, the ratio using paclitaxel is significantly less than
1 and indicates that d5-CHI also inhibits its efflux. However, d5-CHI does not behave as a universal inhibitor since the ratios for vinblastine and doxorubicin are greater than 1. These data suggest that the non-natural nucleoside enhances their export by binding to at least two
126 mutually exclusive sites that functionally interact with each other. Finally, the inclusion of
d5-CHI has no effect on the V max /K m values for colchicine or verapamil. These data suggest a lack of interaction between the binding of d5-CHI and colchicine or verapamil. d5-CHI modulates drug resistance in MDR positive cells. The differential effects of d5-CHI on the
ATPase activity of P-gp could be used to influence resistance to the effects of certain drugs in cell lines that overexpress P-gp. To test this hypothesis, we used an MDR positive cell line
(KB-V1 MDR1 +/+) to evaluate the ability of d5-CHI to alter the cytotoxic effects of
various anti-cancer agents. Cytotoxicity was assessed using an MTT assay that measures the
loss of cell viability as a function of drug concentration. Since d5-CHI is cytotoxic at
concentrations greater than 50 µM (vide infra) 2, changes in drug resistance were evaluated by
comparing the LD 50 value for a chemotherapeutic agent in the absence and presence of a
sub-lethal dose of d5-CHI. As shown in Figure 4A, the LD 50 for paclitaxel decreases in the presence of d5-CHI, thereby sensitizing the MDR-positive cells to the cytotoxic effects of the anti-cancer agent. The ability of d5-CHI to potentiate the effects of paclitaxel correlates
with the inhibition in ATPase activity. An opposite effect is observed when vinblastine or doxorubicin is combined with d5-CHI (Figure 3.4A) as the cells become more resistant
their cytotoxic effects when d5-CHI is present. These data suggest that d5-CHI enhances
their efflux to further enhance resistance. In this case, the ability of d5-CHI to protect the
cell against the effects of these chemotherapeutic agents correlates with ratios greater than 1
measured in the catalytic efficiencies in ATPase activity. Finally, the presence of d5-CHI did
not influence the LD 50 values for colchicine. The lack of cellular effect is expected since no change in the catalytic efficiency ratio for ATP hydrolysis was detected in the presence of d5-CHI. Collectively, these data reveal an excellent correlation between the ability of d5-CHI
127 to modulate drug resistance in cell lines overexpressing P-gp with the in vitro derived catalytic activity ratios for the ATPase activity of P-gp with these agents.
To further characterize the MDR reversing effects of d5-CHI, KB-V1 cells were treated with increasing concentrations of paclitaxel and a fixed concentration of d5-CHI (50
µM) for 48 hours. Under these conditions, the cytotoxicity of paclitaxel is increased ~10- fold by a sub-lethal dose of d5-CHI (Figure 3.4B) . This potentiating effect is not observed using the isogenic cell line, KB3-1, that does not overexpress P-gp (Table 3.4) . This result
confirms that potentiation by d5-CHI is caused by functional inhibition of P-gp.
Furthermore, the related structural analogs, d5-CEI and d5-PhI, had no significant effects on
MDR phenotype (Figure 3.5) and again highlight the unique selectivity of d5-CHI as a modulator of P-gp activity.
Cytotoxicity of d5-CHI. As indicated above, d5-CHI displays cytotoxic effects in a dose- and
time-dependent manner. The LD 50 value of d5-CHI was measured using the KB3-1 and KB-
V1 cell lines in which cells were treated with concentrations of d5-CHI that range over
several log units (0.001 µM to 200 µM). At 72 hours post-treatment, cell viability was
assessed using the MTT assay. The LD 50 value for d5-CHI in the MDR positive cell line is
44+/- 5 µM. An identical LD 50 value of 45 +/- 3 µM was measured in the isogenic cell line
not overexpressing P-gp (KB3-1 MDR -/-) (Figure 3.6). The identity in LD 50 values
indicates that the cytotoxic effect of d5-CHI is independent of P-gp and again validates that
the non-natural nucleoside is not a substrate for the drug transporter.
3.5 Discussion
A goal of molecular medicine is to develop small molecule inhibitors against therapeutically important targets. P-gp provides an excellent example of an important therapeutic target that
128 functions under certain conditions to generate an MDR phenotype by actively transporting
various chemotherapeutic agents out of a cell. However, developing an efficient and selective inhibitor against P-gp proves to be a difficult challenge due to the complexities in the mechanism by which P-gp catalyzes the drug efflux. In this report, we investigate the potential use of non-natural nucleosides as inhibitors of P-gp. The key findings of this paper include the following: 1.) the identification of a novel non-natural nucleoside that modulates
P-gp activity by stimulating its ATPase activity without being a true transportable substrate,
2.) the ability of the non-natural nucleoside to influence the export of other drugs either by competitive inhibition or through cooperative efflux; and 3.) the development of a model correlating changes in the catalytic efficiency of drug-stimulated ATP hydrolysis activity with the modulation of drug resistance in MDR cells. Each point is discussed to highlight their importance.
The initial goal of this work was to identify inhibitors of P-gp by targeting its ATP- binding site. In this regard, we used a series of substituted indolyl 2’-deoxynucleosides as a novel class of purine nucleoside analogs that structurally mimic the basic structure of adenosine and that function as mimetics of ATP and dATP (27-30 ). Initial cell-based screening identified d5-CHI was the only inhibitor of P-gp export activity amongst this group of structurally related non-natural nucleosides. In spite of its inhibitory effects on drug efflux, d5-CHI does not behave as a competitive inhibitor targeting the ATP binding site of
P-gp as initially predicted. Instead, d5-CHI potently stimulates the in vitro ATPase activity of
P-gp with high potency as exhibited by a very low Km value of 60 nM. Monolayer efflux experiments validate that d5-CHI is not exported from cells overexpressing P-gp. This phenomenon is not limited to d5-CHI since Polli and coworkers reported similar results using drugs such as verapamil, ketoconazole, and nifedapine (75 ). In addition, d5-CHI
129 displays identical LD 50 values in both MDR-positive (KB-V1) and MDR-negative (KB3-1)
cell lines. Although d5-CHI is not exported by P-gp, it does inhibit the efflux of known P-gp
substrates such as calcien-AM and paclitaxel. These data collectively support a model in
which d5-CHI behaves as a non-transportable substrate for P-gp.
From a molecular perspective, it is surprising that closely related analogs of d5-CHI
such as d5-CEI and d5-PhI fail to elicit any effect on P-gp activity. This result is unexpected
since all three compounds are extremely hydrophobic (log Pd5-CHI = 3.63, log Pd5-CEI = 3.15, and log Pd5-PhI = 3.31) and should, according to the well-established hydrophobic vacuum cleaner model, behave as substrates for P-gp. In addition, the recently published X-ray structure of
P-gp describes the drug binding cavity as a large chamber possessing a potential volume of
6,000 Å 3 (74 ). The large size of the cavity can easily accommodate the binding of all three compounds. A possible mechanism to explain this dichotomy is that the larger volume of the cyclohexyl ring of d5-CHI (96.1 Å 3) makes more favorable contacts with hydrophobic residues that reside within this large binding cavity. The net effect of these favorable contacts is a slow dissociation rate constant that translates into high binding affinity. In contrast, the smaller volumes of the cyclohexene (91.2 Å 3) and phenyl (82.8 Å 3) groups and/or differences in their overall shape/configuration could provide less favorable interactions which would be reflected in poor affinity. Another possibility is with respect to differences in solvation energies between d5-CHI (-4.23 kJ/mol) compared to d5-CEI (-5.11 kJ/mol) and d5-PhI (-6.59 kJ/mol). Although hydrophobicity and desolvation are generally used interchangeably, each term represents a different biophysical feature that can influence the binding and transport of certain drugs. Hydrophobicity defines the tendency of a molecule to repel water whereas desolvation energy defines the quantity of energy required to remove water from a molecule. The lower solvation energy of d5-CHI compared to d5-
130 CEI and d5-PhI suggests that P-gp could more easily displace water molecules surrounding d5-CHI to accommodate binding. In this mechanism, the greater π-electron surface area and induced dipole moments associated with 5-CEInd and 5-PhInd could form a more defined solvation center that presents a larger barrier for desolvation. This would provide a kinetic that hinders their binding. Indeed, this feature may also explain why other analogs such as d5-NI and d5-NapI that contain large π-electron surface areas do not interact with P-gp.
Since d5-CHI lacks significant π-electron density or dipole moment at the 5-position of the
indolyl ring, it circumvents this desolvation step. This mechanism is being further explored
by developing other non-natural nucleosides containing differing degrees of π-electron density and solvation energies.
Regardless of these possibilities, it is clear that d5-CHI binds to and modulates the activity of drug efflux. In fact, the ability of d5-CHI to block the transport of compounds such as calcein-AM and paclitaxel suggests that all three compounds can occupy the same binding site. This provides another interesting conundrum as there are few, if any, structural similarities between d5-CHI and either calcein-AM or paclitaxel (Figure 3.7) . This feature
exemplifies the polyspecificity displayed by P-gp which has hindered the development of
“rules” that accurately define the interactions of P-gp with a potential substrate. In fact, the
classification of most P-gp substrates is based upon the ability of one compound to
influence the export of another. For example, Ling and colleagues were amongst the first to
report that rhodamine 123 and Hoechst 33342, two well known P-gp substrates, display a
synergistic increase in their rates of efflux when co-administered (117 ). Based upon these
observations, they assigned a potential binding site corresponding to each probe ("H" for
Hoechst 33342 and "R" for rhodamine 123). Other substrates were then classified as "H-
site" or "R-site" based upon their ability to influence the efflux rate of either probe.
131 At face value, the data presented here using d5-CHI is consistent with this model as the non-natural nucleoside behaves like an "H-site" drug by stimulating the ATP hydrolysis activity and transport capabilities of drugs such as vinblastine and doxorubicin. In addition, d5-CHI and paclitaxel appear to compete for the same binding site since d5-CHI squelches any increase in ATPase activity caused by paclitaxel. Finally, d5-CHI inhibits the transport of the taxane to potentiate its cytotoxic effects.
Unfortunately, d5-CHI does not always obey this model. In particular, d5-CHI has no effect on the ex vivo cytotoxicity of colchicine or its ability to stimulate ATP hydrolysis.
These data contradict the R and H site model since colchicine and d5-CHI are both predicted to be "H-site" drugs. These data collectively suggest that the R and H site model cannot be used to unambiguously predict the dynamics of multiple drug interactions.
The data provided in this report are used to develop a different method that can predict the functional interactions of two drug-like molecules with P-gp. This method is based on correlating the influence of d5-CHI on the ATPase activity of P-gp with the effect
of d5-CHI on P-gp transport activity. Specifically, the ratio of V max /K m values in ATP hydrolysis for a substrate measured in the absence or presence of d5-CHI provides an accurate indication of drug-drug interactions. For example, the ratio of less than 1 measured using paclitaxel indicates a decrease in ATPase activity, and this correlates with an inhibitory effect on its efflux as confirmed by the ability of d5-CHI to potentiate its cytotoxic effects in a cancer cell overexpressing P-gp. Likewise, drugs such as vinblastine and doxorubicin display ratios greater than 1 with d5-CHI which correlate with an enhancement in their export as demonstrated by the ability of d5-CHI to increase cellular resistance to vinblastine in
cells overexpressing P-gp. The data summarized in Figure 3.8 show a clear correlation
between the effects of d5-CHI on drug stimulated ATPase activity and the MDR phenotype
132 in cell culture for drugs such as paclitaxel, colchicine, and vinblastine. This correlation provides a valuable methodology to predict either positive or negative endpoints of MDR modulation by the ability of d5-CHI to influence drug efflux of various compounds.
This type of analysis has several advantages over conventional assays that are used to probe drug-drug interactions with P-gp. For example, the calcein-AM efflux assay is rapid and amenable to high throughput screening. However, as demonstrated here, it provides only qualitative information regarding drug efficacy and potency. In addition, it cannot distinguish between an inhibitor versus an alternative substrate. The monolayer efflux assay provides a more quantitative approach to measure substrate transport across a monolayer of cells. However, this cell-based assay is costly, technically challenging to perform, and time consuming. Also, this assay cannot easily measure changes in ATPase activity as a function
of drug concentration. We have shown here that measuring ATPase V max /K m presents a
simpler way to screen for novel compounds that interfere with the efflux of various drugs.
While this provides a powerful approach to screen for drug-drug interactions, it may
not be universally applied to all potential P-gp substrates. For example, stimulation in
ATPase activity by certain P-gp substrates such as etoposide and daunorubicin do not
display Michaelis-Menten kinetics (data not shown). The inability to accurately define their
Vmax /K m values in these cases precludes a careful evaluation by this method. Additionally, we
acknowledge that the magnitude of the changes in the catalytic efficiency in ATPase
stimulation by doxorubicin, for example, provides only a qualitative correlation with change
in doxorubicin resistance. Regardless, the speed of collecting data as well as its accuracy in
correlating in vitro and ex vivo cellular effects makes this approach an important tool in drug
discovery.
133 Figures
Figure 3.1 Structures and electrostatic potential models of non-natural nucleosides used in this study. Electrostatic potential models were constructed using Spartan ’04 software
(www.wavefun.com). Briefly, structures were built and geometry-optimized using the AM1 semi-empirical method of calculation. The optimized structures were then taken to perform single point energy minimization using the B3LYP 6-31G** basis set and the Density
Functional Theory (DFT) level of calculation. The electron density surface potentials were then generated. The most electronegative regions are in red, neutral charges are in green, and the most electropositive regions are in blue.
134 Figure 3.1
135 Figure 3.2 Stimulation of P-gp ATPase activity by non-natural nucleosides. (A) Inside-out
P-gp enriched membranes were used for ATPase measurements and the Michaelis-Menten
kinetic parameters, K m and V max, were obtained by stimulating P-gp ATPase activity with
increasing concentrations of d5-CHI. ATP concentration was fixed at 1 mM and initial
-1 velocities were measured over 20 minutes. K m = 60 ± 18 nM, V max = 3.25 ± 0.37 µM min .
(B) Stimulation of P-gp ATPase activity in inverted P-gp enriched membranes by increasing concentrations of d5-CHI (▼) d5-CEI (▲) and d5-PhI (■).
136 Figure 3.2
A
15.0
12.5
10.0
/min/mg protein) /min/mg 7.5 i
5.0
2.5
rate (nmols P rate(nmols 0.0 0.00 0.05 0.10 0.15 0.20 0.25 0.30 B [d5-CHI] ( µµµM)
17.5
15.0
12.5
10.0
7.5 /min/mg protein) i 5.0
2.5
0.0 -8 -7 -6 -5 -4 -2.5 rate (nmols P log [d5-XI] -5.0
137 Figure 3.3 Stimulation of P-gp ATPase activity by increasing concentrations of calcein-AM in the presence and absence of d5-CHI or CsA. Inside-out P-gp enriched membranes were used, ATP concentration was fixed at 1 mM and initial velocities were measured over 20 minutes. Double reciprocal analysis of d5-CHI (60 nM) and CsA (400 nM) versus P-
glycoprotein. Competitive inhibition was determined by fitting data to the following
equation: rate = V max [S]/K m(1+[I]/K i) + [S]. Calcein-AM alone (▲): K m = 3.1 ± 0.4 µM,
-1 Vmax = 9.2 ± 0.2 µM min , 60 nM d5-CHI (■): K m = 22.5 ± 6.5 µM, V max = 9.4 ± 1.3 µM
-1 -1 min , or 400 nM CsA (▼): K m = 20.9 ± 8.7 µM, V max = 8.0 ± 1.6 µM min .
138 Figure 3.3
139 Figure 3.4 Modulation of drug resistance by d5-CHI in KB-V1 cells. (A) Effects of 10 µM
d5-CHI over 72 hours on the log (LD 50 ) values of vinblastine (VBL), doxorubicin (DOX), colchicine (COLC), and paclitaxel (TAX). (B) Potentiation effects of 50 µM d5-CHI over 48
hours on the LD 50 of paclitaxel (TAX). No treatment (■) LD 50 = 4.0 ± 0.24 µM, 50 µM d5-
CHI (▲) LD 50 = 0.49 ± 0.03 µM.
140 Figure 3.4
A
0.3 ) 50 0.2
0.1
-0.0
-0.1 d5-CHI)-log (LD d5-CHI)-log 50 -0.2
-0.3 log(LD -0.4 L X C X B O L A V D O T C
B
100
75
50 %viable 25
0 -9 -8 -7 -6 -5 -4 -3 log[TAX] 0 50 uM 5-CHInd Sigmoidal dose-response Best-fit values BOTTOM -7.824 -1.501 141 TOP 95.51 101.9 LOGEC50 -5.390 -6.301 EC50 4.069e-006 4.998e-007 142 Figure 3.5 Effects of d5-CHI, d5-CEI, and d5-PhI on the cytotoxicity of paclitaxel and
vinblastine in KB-V1 cells. Concentrations of all nucleosides was 10 µM. *p<0.05 vs. TAX
+ d5-CEI and TAX + d5-PhI. #p<0.05 vs VBL + d5-CEI.
143 Figure 3.5
0.3 # 0.2 0.1 -0.0 -0.1 -0.2 -0.3 *
log (LD50 d5-XI)-log (LD50) d5-XI)-log log(LD50 -0.4 -0.5 I I HI EI 5-C 5-C d d5-CE d d5-Ph + + X X + L + d5-CHIL L + A B B TA T TAX + d5-PhIV VB V
144 Figure 3.6 Effects of d5-CHI on cell viability in (A) KB-V1 cells and (B) KB-3-1 cells over
72 hours. KB-V1 LD 50 = 44.1 ± 5.5 µM, KB3-1 LD 50 = 45 ± 3 µM.
145 Figure 3.6
A
120 110 100 90 80 70 60 50 40 % viable 30 20 10 0 -10 -8 -7 -6 -5 -4 -3 log[d5-CHI]
B
110 100 90 80 70 60 50 40 % viable 30 20 10 0 -10 -10 -9 -8 -7 -6 -5 -4 -3 log[d5-CHI]
146 Figure 3.7 Comparison of the structures of P-gp interacting compounds (A) 5-CHInd, (B) calcein-AM, (C) Cyclosporine A, and (D) paclitaxel.
147 Figure 3.7
(A) (B)
O O
H3C O O CH3
O O
CH3 O O O O CH3 N N N O O O OOO HO O O O O O O
CH3 CH3
O
HO O
(C) (D)
O O OH HO O O O H N N N N N O NH O O O O H O O O O N O O O OH H H OH N N N O O N N H O O O
148 Figure 3.8 Trend in ATPase V max /Km ratios and MDR modulation. The cytotoxic drugs used were vinblastine (VBL), doxorubicin (DOX), colchicine (COLC), and paclitaxel (TAX).
Left side: Changes in drug stimulated ATPase V max /K m ratios in the presence of 60 nM d5-
CHI. Right side: Changes in log (LD 50 ) for cytotoxic drugs in KB-V1 cells in the presence of
10 µM d5-CHI over 72 hours.
149 Figure 3.8
7.5 0.3 Vmax /K m ratios 7.0 ∆ LD 50 6.5 0.2 log (LD
m(drug) 6.0
/K 5.5
5.0 0.1 50 max 4.5 d5-CHI)-log (LD / V / 4.0 -0.0 3.5 3.0 2.5 -0.1 2.0 m(drug+d5-CHI) 50
1.5 ) /K -0.2 1.0
max 0.5 V 0.0 -0.3 VBL DOX COLC TAX
150 Tables
Table 3.1 Kinetic parameters for the stimulation in ATPase activity by drug substrates.
a Vmax (nmol P i/min/mg Drug Substrate K ( µM) a V /K m protein) max m
0.06 ± d5-CHI 16 ± 0.2 266.7 ± 39.0 0.01 Calcein-AM 3.1 ± 0.5 46 ± 2 14.8 ± 1.5 0.20 ± Vinblastine 17.5 ± 0.5 87.5 ± 2.0 0.03 Colchicine 37 ± 8 4.0 ± 0.2 0.1 ± 0.015 0.85 ± Doxorubicin 3.5 ± 0.2 4.1 ± 0.5 0.17 Paclitaxel 0.9 ± 0.2 9.5 ± 1.0 10.6 ± 1.5
Verapamil 4.0 ± 1.7 16.5 ± 2.5 4.1 ± 1.0 0.40 ± Cyclosporine A 7.0 ± 1.0 17.5 ± 1.5 0.16
a Km and V max values are mean ± SEM calculated from at least three independent experiments. Initial velocities were measured in the presence of fixed ATP concentration (1 mM) and varying drug concentrations. K m and V max values were obtained by fitting the initial velocities to the Michaelis-Menten equation.
151 Table 3.2 Permeability coefficients of substrates across MDCK and MDCK-MDR monolayers.
a MDCK-MDR1 BL →AP/ AP →BL
b d5-CHI 10 µM 11.1 ± 2.5 d5-CHI 10 µM e 1.56 ± 0.21 and Adenosine 10 µM
c Rhodamine 123 3 µM 72.7 ± 15.3
d Calcein-AM 4 µM 2940 ± 120
Calcein-AM 4 µM 0.98 ± 0.18 and d5-CHI 10 µM
Lucifer Yellow 25 µM 0.34 ± 0.05
MDCK BL →AP/ AP →BL d5-CHI 10 µM 8.7 ± 1.3 d5-CHI 10 µM e 2.3 ± 0.6 and Adenosine 10 µM
Rhodamine 123 3 µM 1.2 ± 0.2
Calcein-AM 4 µM 2.6 ± 0.3
Lucifer Yellow 25 µM 0.8 ± 0.1
aValues are mean ± SEM calculated from three independent experiments. bd5-CHI spectroscopic detection parameters: λ = 260 nm, ε = 6,843 M -1cm -1 cCalcein-AM spectroscopic detection parameters: λ = 496 nm, ε = 6,000 M -1cm -1 dRhodamine 123 spectroscopic detection parameters: λ = 500 nm, ε = 75,000 M -1cm -1 ed5-CHI and adenosine were measured at λ = 260 nm, ε = 6,843 M-1cm -1 (d5-CHI) and ε = 15,200 M-1cm -1 (adenosine)
152 Table 3.3 Effects of d5-CHI on the catalytic efficiency of drug-stimulated P-gp ATPase activity.
a Vmax /K m(drug+d5-CHI) / Vmax /K m(drug) Vmax/K m(drug + d5-CHI) Vmax /K m(drug) Calcein-AM 2.93 0.42 0.143 Calcein-AM + b 2.93 0.38 0.131 CsA Vinblastine 16.7 64.7 3.9
Colchicine 0.021 0.023 1.09
Doxorubicin 0.86 5.5 6.4
Paclitaxel 2.1 >0.02 >0.1
Verapamil 0.80 0.84 1.0 ad5-CHI concentration for all experiments was 60 nM bCyclosporin A (CsA) at 400 nM was used in place of d5-CHI
153 Table 3.4 Effects of d5-CHI on the cytotoxicity of vinblastine (VBL), doxorubicin (DOX), colchicine (COLC), and paclitaxel (TAX) in KB-V1 cells and parental KB-3-1 cells.
Relative KB-V1 LD ( µM) KB-3-1 LD (nM) a 50 50 Resistance VBL 0.445 ± 0.140 0.034 ± 0.005 13088
VBL + d5-CHI 10 µM 0.908 ± 0.252 0.038 ± 0.003 23894
DOX 5.6 ± 0.9 62.8 ± 6.2 89
DOX + d5-CHI 10 µM 6.7 ± 1.3 68.6 ± 9.5 97
COLC 2.3 ± 0.6 14.1 ± 3.1 163
COLC + d5-CHI 10 2.2 ± 0.5 15.3 ± 4.9 143 µM TAX 3.8 ± 0.2 5.1 ± 1.2 490
TAX + d5-CHI 10 µM 2.3 ± 0.05 5.5 ± 1.8 290
a Relative Resistance = LD 50 (KB-V1) / LD 50 (KB-3-1)
154 Chapter 4: Conclusions and future directions
4.1 Overview
This dissertation explored the use of non-natural nucleotides as inhibitors of mechanical
ATPases. Although all mechanical ATPases bind ATP, there is diversity in the architecture of their active sites and this can be exploited for the development of selective inhibitors.
While kinases have been the focus of ATP-competitive inhibitors, mechanical ATPases represent a vast array of potentially useful therapeutic targets (6). For this reason, this dissertation focuses solely on mechanical ATPases.
The non-natural nucleotides described in this work were originally developed to explore the dynamics of translesion DNA synthesis. These studies determined that several of these analogs act as effective dATP surrogates that display superior binding and insertion kinetics for DNA polymerase during insertion opposite a non-templating abasic site (26-30 ).
Based upon these similarities, I hypothesized that non-natural nucleotides could act as ATP competitive inhibitors for mechanical ATPases. Two model systems were chosen to evaluate this hypothesis. The first system was the bacteriophage T4 replicase complex which consists of DNA polymerase and a sliding clamp processivity factor. The formation of this complex is essential for DNA replication and is catalyzed by the clamp loader ATPase. By inhibiting the clamp loader, replicase formation will be prevented resulting in inhibition of processive
DNA replication. The second system is the ABC transporter P-glycoprotein. This transporter is commonly expressed in drug resistant cancers where it functions to export a
wide array of drugs using the power of ATP hydrolysis. By inhibiting ATP binding and hydrolysis, the export of cytotoxic drugs will be inhibited and drug resistance can be reversed.
155 4.2 Chapter 2 Conclusions
In chapter 2, I explored the effects of non-natural nucleotides on the T4 replicase system to demonstrate three important principles. First, processive DNA replication can be inhibited by preventing the formation of the replicase complex with an ATP competitor. Second, non- natural nucleotides can be used as chemical probes to explore the active site architecture of an ATPase. Third, non-natural nucleotides can act as selective inhibitors of clamp loader proteins from two different species (Figure 4.1) .
To demonstrate the first principle, non-natural nucleotides were screened for inhibition of T4 gp44/62 ATPase activity and d5-NITP was identified as the most potent inhibitor. Inhibition of replicase formation was then confirmed by strand displacement assay and FRET quenching. The second principle was used to determine the binding mechanism of d5-NITP that is responsible for its superior potency. Since the structure activity relationship for the library of non-natural nucleotides clearly showed d5-NITP as an outlier
(Figure 2.4) , I hypothesized that it alone participates in an electrostatic interaction with a key amino acid within the active site of gp44/62. Using molecular modeling, Arg175 was identified as an electrostatic contributor for d5-NITP binding and this interaction was confirmed by mutagenesis. To demonstrate the third principle I evaluated the selectivity of d5-NITP by testing its ability to inhibit E.coli γ-complex ATPase activity. d5-NITP was
determined to be a selective inhibitor of gp44/62 by binding with ~5-fold higher affinity
compared the γ-complex. A selectivity model was generated by molecular modeling and
showed that E.coli γ-complex lacked an electrostatic contributor corresponding to gp44/62
Arg175, thus binding d5-NITP with lower affinity. Finally, selectivity was demonstrated in vivo with a plaque forming assay that showed a decrease in T4 plaque formation upon treatment with d5-NI.
156 In this study, structure activity relationships were difficult to describe. However, the
collective data suggests that the chameleon-like properties of the nitro moiety allow it to
bind with superior affinity by maximizing interactions within the ATP binding pocket.
Specifically, its electron withdrawing potential and zwitterionic character allows it to
participate in non-covalent, electrostatic interactions. The double-bond character of -NO 2 provides potential interactions though π-cation and π-π stacking arrangements and the nitro group possesses a low solvation energy that allows it to interact with molecular targets through entropic effects. These collective properties provide d5-NITP with its superior inhibitory effects against ATP-dependent clamp loaders, especially compared with other competitive inhibitors such as ATP γS and AMP-PCP that are true mimics of ATP. Although these properties suggest that nitro-containing nucleos(t)ides will make excellent therapeutic agents, drawbacks warrant careful consideration. The most prominent complication is that several nitro-containing drugs are associated with unpredictable drug reactions that limit their use (118 ). For example, the catechol-O-methyltransferase inhibitor, tolcapone, was discontinued as a therapeutic agent against Parkinson's disease since it displayed a high incidence of liver toxicity (119 ). Despite this, there are numerous examples of nitro- containing compounds such as nifedipine (anti-arrhythmia agent), clonazepam
(anticonvulsant/muscle relaxer), and chloramphenicol (antibiotic) that are safe and effective therapeutic agents. Thus, the potential safety and efficacy of nitro-containing nucleosides needs to be carefully investigated before establishing clinical utility for this class of analog.
4.3 Chapter 2 Future directions
From a basic science perspective, this study demonstrates a utility for non-natural nucleotides as chemical probes for ATP binding sites. Localization of non-natural
157 nucleotides to the ATP binding pocket, by virtue of the P-loop and triphosphate interaction, allows for the specific evaluation of nucleobase binding contributions. Thus, hypotheses of the interactions between non-natural nucleobases and amino acids in the adenine binding region can be easily generated by screening non-natural nucleotides for inhibition of ATPase activity. However, the current library of non-natural nucleotides is limited to primarily 5’ modifications and expanding on the position of substituent groups could increase the probing abilities of this library and offer more precise determinations of the architechture of the ATP binding site. This principle was explored in chapter 2 where d4-NITP and d6-NITP
were used to probe gp44/62 and determined that the position of the nitro group was
important for ionic and/or hydrophobic interactions with Arg175. Briefly, the K i value for these two analogs was shown to be unchanged in both R175L and R175K mutants suggesting that the nitro groups of these compounds do not participate in any interactions
with Arg175. Since Arg175 was shown to participate in ionic interactions with d5-CITP, the
importance of the position of the carboxylic acid group can be analogously evaluated using
d4-CITP and d6-CITP. Based on the results with d4-NITP and d6-NITP, the K i value of
d4-CITP and d6-CITP are predicted to remain unchanged in either mutant due to the
orientation away from residue 175. Alternatively if the K i values for d4-CITP or d6-CITP change in either mutant, a compensatory conformational change may be occurring. Since this effect was not seen with the nitro compounds, it would suggest that the hydrophilic nature of the carboxylic acid may be driving this compensation. To validate this hypothesis, other hydrophilic substituents such as a 4 or 6 hydroxyl or amino should be tested against
both mutants. If K i values for the hydrophilic analogs are found to differ across all forms of gp44/62, it would be consistent with the alternative hypothesis.
158 Though the overall goal of this dissertation was to explore mechanical ATPases, chapter 2 demonstrated a broader implication of non-natural nucleotides as inhibitors of
ATPases and the extension of this implication to kinases cannot be ignored. The approach to this application would be to use in vitro screening services in combination with in silico predictions (docking scores) ( Figure 4.2 ). In addition to determining exploitable elements
within the ATP binding domains of kinase proteins, this proposed study may aide in the discovery and development of novel selective inhibitors. For this study, in vitro kinase data
will be provided by the PDSP which screens d5-NI, d4-NI, d6-NI, dInd, d5-PhI, d5-NapI, and d5-CHI for inhibition of an array of kinases targets. A commercial screening service such as Invitrogen’s kinase profiling service will be used to test d5-NITP, d4-NITP, d6-
NITP, dITP, d5-PhITP, d5-NapITP, d5-CHITP for inhibition of kinase targets. In collaboration with the Dealwis laboratory, the eight non-natural nucleo(s)tides will be docked into the available x-ray structures of the kinases that participated in the in vitro screening. From these screens, kinase profiles for each nucleo(s)tide will be generated and compared. Compounds that show a strong positive correlation between in vitro and in silico
screens will progress to full in vitro characterization (K i determination for all targets and
mode of inhibition). While this proposed study would be exploratory in nature, it has the
potential to generate many testable hypotheses. For example, a non-natural nucleo(s)tide
may display activity against an important kinase in a growth/proliferation pathway. This
compound would be tested in cell culture models and its target validated by siRNA and
overexpression vectors. In this scenario, knocking down the target in cells would have
similar effects as treatment with the compound. Conversely, the effects of the compound
can be mitigated by addition of an overexpression vector for the target in question.
However, the above example is a greatly simplified scenario of a testable hypothesis
159 generated from kinase profiling. Most likely, a compound will display activity against a number of kinase targets and combinations of knockdowns and overexpressions would be necessary to validate the targets. However, the broader implication of this study is to identify a broad spectrum agent similar to Sorafenib (120 ). Sorafenib was the first multikinase
inhibitor originally developed as a single agent inhibitor of Raf-1 kinase. It was later found to
inhibit multiple kinases such as VEGFR, p38, c-kit, and B-raf. This broad spectrum activity
of Sorafenib made it an effective treatment for advanced solid tumors such as renal cell
carcinoma (120 ). Kinase profiling can be used to match the spectrum of a potential inhibitor
with kinases that are known to be upregulated in certain cancers (datasets available on Gene
Expression Omnibus) resulting in a single agent that can hit multiple relevant targets.
A potential pitfall of this study would be the disparity between in vitro and in silico
data sets and/or unavailability of in silico data set due to unsolved x-ray structures. Some
degree of disparity is expected in a large data set comparing multiple compounds with
multiple targets; however, large disparities must be determined to represent either unique
binding mechanisms or experimental artifacts. This can be determined biochemically by
examining the mode of inhibition. In the event that X-ray structures are not available for
select targets, solving these structures is feasible given the large resources for structural
biology available at CWRU.
The PDSP tests the non-natural nucleosides against a variety of receptors that may
provide an indication to additional physiological and psychological effects. Most notable,
PDSP identified d5-PhI, d5-NapI, and d5-CHI as hits for the 5HT2A and 5HT2B receptors
with 50-80% efficacy compared to the respective controls ketaserin and LSD. The same
screen also identified d5-NI, dInd, d5-PhI, and d5-NapI as hits for the histamine H2
receptor with 50-70% efficacy compared to the control tiotidine (assay protocols available at
160 http://pdsp.med.unc.edu/pdspw/binding.php). However, it is not yet known whether they
act as agonists or antagonists. Since both 5HT2 and histamine H2 receptors are Gq/11
coupled receptors, intracellular calcium mobilization assays can be used to assess the
functionality of binding (121 ). Activity on 5HT2 receptors may result in CNS effects such as
hallucination and neuronal excitation/depression and cardiovascular effects such as
pulmonary vasoconstriction/dilation (41 ). The reactivity of non-natural nucleosides with
5HT2 receptors can be explained by their structural similarity with serotonin ( Figure 4.3 ).
Both are very similar in the sense that they share a common indolyl scaffold. Serotonin lacks a deoxyribose moiety; however, the reactivity of non-natural nucleosides against serotonin receptors suggest that the indolyl scaffold is the primary pharmacophore. The structural relationship between histamine and non-natural nucleosides is less obvious. The imidazole of histamine is similar to the 5-membered ring of the indolyl scaffold; however the enthanamine moiety does not fill the same space as the 6-membered portion of the indolyl.
This suggests that a 5-membered heterocycle may be an important pharmacophore for histamine receptor binding. This structural feature is also shared with the H2 antagonist cimetidine ( Figure 4.3 ). Nonetheless, If an agonist of histamine H2 receptor, non-natural nucleosides may cause inflammation and/or acid reflux (41 ). If an antagonist of histamine
H2 receptor, these compounds may have similar effects as cimetidine and be useful for the treatment of peptic ulcer, acid reflux, and dyspepsia. Although not yet identified by PDSP, based on structural similarities it is reasonable to hypothesize that non-natural nucleosides
would bind to adenosine and melatonin receptors ( Figure 4.3 ). The adenosine receptor is included in the PDSP thus the results for non-natural nucleosides against this receptor should be available in the near future. Pharmacological activity on the adenosine receptor
would manifest in changes in alertness, vascular resistance, and heart rate. However, PDSP
161 does not include the melatonin receptor, so in order to test for interactions with non natural nucleosides a melatonin receptor binding assay will be performed (122 ). This assay uses 2-
[125 I] iodomelatonin as the radioligand and chick brain membrane preparations as the source of melatonin receptor (123 ). If reactivity is observed, possible effects of this pharmacological action would be changes circadian rhythm.
One of the major implications of chapter 2 was the demonstration an alternative strategy for inhibiting DNA replication by inhibiting the processivity of DNA replicase.
Selectivity studies suggest that this strategy can be used to combat foreign pathogens such as bacteria or viruses. Since diversity in the ATP binding site was shown between T4 and E.coli clamp loaders, using non-natural nucleotides to probe the active site of a given ATPase can rapidly yield exploitable elements within the ATP binding site as well as identify selective non-natural nucleo(s)tide inhibitors. A possible target for the anti-viral application of non- natural nucleo(s)tides is the Hepatitis C virus (HCV). Acute infection of HCV (within 6 months) is typically asymptomatic and an estimated 3.2 million people in the United States are currently infected (Statistics available at http://www.cdc.gov/hepatitis/HCV/HCVfaq.htm#section1 ). Chronic HCV infection is a major health concern as it results in persistent liver disease that can progress to cirrhosis and liver cancer. There are no treatments for acute infection and chronic infections are treated
with interferon and ribavirin (124 ). However, these treatments are not beneficial to all patients and can cause serious side effects (125 ). Clearly, there is a need for more hepatitis C treatment options. The HCV helicase (NS3) has been implicated as a drug target to combat this disease. Several studies have shown that it has potential drug binding cavities (126 ), and crystal structures are available providing for structure-based approaches to drug design. NS3 is a multi-functional enzyme with NTPase activity, helicase activity and protease activity that
162 are all essential for productive viral infection (127 ). The NTPase activity fuels the helicase activity (128 ), and uniquely NS3 helicase has a catalytic site and an allosteric NTP binding site (128-129 ). The presence of two NTPase sites would predict the possibility of identifying non-competitive inhibitors. As previously mentioned, ribavirin is one of the only treatments available for hepatitis C. The mechanism of Ribivirin’s anti-viral effects is incorporation into
viral RNA leading to lethal hypermutation. However, in addition to this mechanism
Ribavirin triphosphate has been shown to be an inhibitor of both NS3 NTPase and helicase activity (130 ) and analogs of ribavirin have also been explored as inhibitors (131 ). Bretner and coworkers explored the inhibitory effects of inosine derivatives on NS3
NTPase/helicase activity and found that 5'-O-(4-fluorosulphonylbenzoyl)-esters of inosine displayed weak inhibitory effects. Since non-natural nucleotides are purine analogs and the triazole and carboxamide of ribavirin fill similar space to a purine, it is possible that non- natural nucleotides may have inhibitory effects on NS3 NTPase and/or helicase activity.
The overall hypothesis is that non-natural nucleo(s)tides are capable of influencing
NS3 NTPase/helicase activity. To test this hypothesis, a library of non-natural nucleo(s)tides
will be screened against NS3 to identify any inhibitors of NTPase activity . Based on structural features of ribavirin, I hypothesize that analogs with hydrophilic and hydrogen bonding groups such as d5-AITP and d5-CITP would be able to inhibit NTPase activity. It is also possible that the nitro containing compounds such as d5-NITP, d6-NITP, and d4-
NITP, may bind and inhibit NTPase activity by virtue of multiple chemical properties
(explored in chapter 2). Purified NS3 will be generated from well established protocols (132 )
and K i measurements will be obtained for NTPase activity for a library of non-natural nucleo(s)tides (133 ). NTPase inhibitors will then be tested for functional inhibition of NS3 by evaluating their ability to inhibit unwinding activity (134 ). Functional inhibitors will be
163 further characterized as competitive or non-competitive inhibitors by determining the mode of inhibition from double reciprocal analysis. The most potent inhibitors will be given the highest priority for docking studies to identify the binding mechanism using the oligonucleotide-bound structure of NS3 (pdbID: 1A1V). These docking studies will be carried out using the same method described in chapter 2 for determining the binding mechanism of d5-NITP to gp44/62. Of particular interest are non-competitive inhibitors because this may reflect an interaction with the allosteric NTPase site. Though a non- competitive inhibitor may reflect allosteric inhibition, exploring the allosteric NTP binding site would be difficult since it has not been structurally defined. Thus characterization of a non-competitive inhibitor would be limited to enzymology approaches. Finally, lead compounds will be evaluated for inhibition of HCV replication in a human hepatocyte model (135). If successful, lead compounds will be tested for further efficacy in a human
liver graft mouse model (136 ).
Potential pitfalls for the application of non-natural nucleotides in HCV
chemotherapy are the failure to identify any non-competitive inhibitors from the library
compounds. In this scenario, the results can still be useful for the probing of non-allosteric
NTPase site leading to rationale approaches for synthesizing more potent compounds or the
optimization of weaker binding inhibitors.
4.4 Chapter 3 Conclusions
The data presented in chapter 3 demonstrates the use of d5-CHI as a modulator of P-gp activity and provides interesting insight into the selectivity of the P-gp drug binding cavity using non-natural nucleosides as chemical probes. d5-CHI was originally identified as a P-gp interacting compound by its ability to block the efflux of calcein-AM. Based on its structural
164 similarity to ATP, I hypothesized that it inhibits P-gp activity by competing with ATP. To test this hypothesis, I evaluated the ability of d5-CHI to inhibit the basal ATPase activity of
P-gp enriched membranes in vitro . Unexpectedly, d5-CHI stimulated ATPase activity with
Michaelis-Menten kinetics, suggesting that it acts as an alternative substrate. The K m and V max
values for ATPase stimulation in P-gp enriched membranes was measured for various chemotherapeutic drugs and the results were expressed as ATPase catalytic efficiency
(V max /K m). The effect of d5-CHI on catalytic efficiency was determined by a ratio of
Vmax/K m(drug + d5-CHI) / V max /K m(drug) . Values greater than 1 indicate an enhancement of activity and predicted increase in resistance and values below 1 indicate an inhibition of activity and reversal of resistance. This prediction was confirmed in an MDR cell culture model (KB-V1)
validating our ATPase model as a useful strategy for determining P-gp drug interactions
(Figure 4.4). Another important aspect of this study was the effects of other non-natural nucleosides on P-gp activity. Despite their structural similarity and the well-known ability of
P-gp to transport an array of hydrophobic compounds, only one non-natural nucleoside out of nine displayed any interactions with P-gp. Structural analogs of d5-CHI such as d5-PhI and d5-CEI failed to elicit any effects on P-gp activity.
4.5 Chapter 3 Future directions
The overall model correlating ATPase kinetics and modulation of drug resistance makes the assumption that changes in drug cytotoxicity are inversely related with drug accumulation.
Although no direct evidence has been provided to validate this assumption, the fact that d5-
CHI does not change the LD 50 value for any drugs in non P-gp overexpressing cells ( Table
3.4 ) indicates that the effects of d5-CHI are P-gp mediated rather than multiple off-target
effects. However, the effects of d5-CHI on permeability coefficient ratios ( Peff ratio) of
165 vinblasine, doxorubicin, paclitaxel and colchicine must be measured in order to validate this model. This can be accomplished by using the monolayer efflux assay with commercially available radiolabeled drug substrates (American Radiolabeled Chemicals) in the absence and
presence of d5-CHI. The predicted results are that d5-CHI would increase the Peff ratio of
vinblastine and doxorubicin. Conversely, the Peff ratio of paclitaxel would be inhibited and no
effect in the Peff ratio of colchicine would be observed.
If this model is validated by direct transport experiments, it would lead to the hypothesis that there are no distinct drug binding sites; rather combinations of drugs interact
with P-gp according their own unique mechanism to either enhance of inhibit efflux. The R and H site model previously proposed that Rhodamine 123 and Hoechst 33342 were specific probes for two distinct drug binding sites and that unknown substrates can be classified based on their interactions with these probes (117 ). However, the data in chapter 3 as well as other studies using QB102 (137-138 ) have shown that not all modulators accurately adhere to this model. Brielfy, QB102 was found to increase the cytotoxicity of vinblastine but decrease the cytotocity of doxorubicin in P-gp overexpressing cells. This result is contrary to the R and H site model which posits that vinca alkyloids and anthracyclines both bind the R site and thus should behave similarly in response to a modulator. The recently solved X-ray structure offers important insights into substrate interactions with P-gp by examining the orientation of substrates within the binding cavity of P-gp (74 ). Aller and coworkers used two stereo specific hexipeptide inhibitors, QZ59-RRR and QZ59-SSS, to show distinct binding orientations within the P-gp substrate binding cavity. Co-crystallizsation of d5-CHI
with paclitaxel, vinblastine, colchicine, or doxorubicin within the P-gp binding cavity could provide a structural explanation for our biochemical observations (139 ). Based on these
studies, I would predict that a similar binding orientation would be observed between d5-
166 CHI and either vinblastine and doxorubicin. Specifically, the combination of d5-CHI and either vinblastine or doxorubicin within the P-gp drug binding cavity would form similar amino acid contacts. From a technical standpoint, obtaining electron density maps for large complex molecules such as vinblastine or paclitaxel in the P-gp crystal structure would be exceptionally difficult due to a lack of ordered binding. However, the presence of combinations of substrates may act to stabilize the structure and allow observation of these structurally complex molecules within the binding cavity of P-gp. Though technically challenging, if this structural work could be accomplished the findings would be monumental in the P-gp field. It would suggest that ordered binding results in cooperativity in transport and would provide a structural explanation for the long standing question of P- gp cooperative transport. It is possible that no static structural model can explain the cooperativity of transport. If this is the case, all combinations of substrates may result in different orientations or no ordered binding. In this scenario, an explanation may lie in the conformational changes in the P-gp catalytic cycle. Addressing this question would require the use of time-resolved crystallography techniques to visualize P-gp in various conformations in response to combinations of substrates. Of particular interest would be the transmembrane domains in the closed solvent exposed conformation. This model would provide insight to the interactions between substrates in the drug binding cavity and the membrane outer leaflet. If the closed conformation allows substrates to directly access the membrane outer leaflet in an analogous fashion to the open transmembrane conformation
(74 ) and the membrane inner leaflet, a flippase mode of action is likely and the outer leaflet
lipid composition may play an important role in the specificity of substrate transport by
dictating substrate dissociation.
167 Although d5-CHI did not behave as an ATP-competitive inhibitor, the correlation between ATPase kinetics and modulation of resistance in MDR cells provides a rapid in vitro method for determining P-gp drug-drug interactions that can be applied to all newly discovered modulators. Standard reports typically test the ability of a candidate modulator to inhibit verapamil stimulated ATPase activity and inhibit the efflux of rhodamine 123 in an
MDR cell model (140 ). However, these are two substrates out of a vast number of clinically relevant substrates whose interactions remain unexplored. This model allows for the determination of P-gp interactions between a modulator and several substrates or multiple drug combinations. This becomes important since P-gp has implications in pharmacokinetics and this method can be used to determine drug-drug interactions in all areas of chemotherapy (141 ). Following this conclusion, widely known P-gp drug-drug
interactions can be used to validate this model. For example, Lagas and coworkers observed
an increase in dasatinib accumulation in the brain after treatment with the P-gp inhibitor
elacridar (142 ). In this scenario, it would be predicted that elacridar would increase the K m
for dasatinib stimulated P-gp ATPase activity leading to a V max /K m ratio of less than 1.
Another example is the P-gp mediated increase in ivermectin plasma concentration with multiple treatments of ketoconazole (143 ). Ketoconazole was suggested to interfere with ivermectin clearance by interaction with P-gp. In the ATPase kinetic analysis, this would be
reflected in a V max /K m value of less than 1 for ivermectin stimulated ATPase activity in the presence of ketoconazole.
Multiple mechanisms of resistance have been implicated in the failure of P-gp inhibitors in clinical trials (144-145 ). Over the years, several ABC transporters have been
discovered and found to be associated with MDR cancers (146-147 ). These transporters
include multi-drug resistance proteins (MRPs) (148 ) and breast cancer resistance protein
168 (BCRP) (149 ). The simultaneous expression of such transporters is referred to as multi-
factorial drug resistance. Although other ABC transporters have not been studied as
extensively as P-gp, they share many similarities such as ATP-dependence and overall
structure (150 ) ( Table 4.1 ). Thus, similar approaches to studying drug interactions with these
transporters can be used.
The initial screen of non-natural nucleosides was done in using calcein-AM in Caco-
2 cells to assess P-gp activity. However, in addition to expressing P-gp Caco-2 cells are
known to express MRP1, MRP2, and BCRP. It should be noted that calcein-AM is also an
effective substrate for MRP1 (151 ) thus it is predicted that d5-CHI inhibits both MRP1 and
P-gp. To test this prediction, COR-L23 cells (MRP1 overexpressing) will be used in the
calcein-AM accumulation assay. If d5-CHI inhibits MRP1, it should result in an
accumulation of calcein within the COR-L23 cells. To evaluate the contributions of other
co-expressed ABC transporters in Caco-2 cells, specific probes such as pheophorbide A
(152 ) and diacetate ester of 5(6)-carboxy-2',7'-dichlorofluorescein (CDFDA) (153 ) can be
used to evaluate the contributions of BCRP and MRP2, respectively. ATPase kinetics can
then be measured to predict any potential interactions between d5-CHI and substrates of
interest for BCRP and MRPs. These predictions can then be tested for reversal of MDR in
the appropriate cell culture models. Cell lines that express single transporters (COR-L23/R
MRP1 overexpressing, MCF7-FLV1000 BCRP overexpressing, HepG2 MRP2
overexpressing) will be used to validate predictions and a cell line expressing multiple
transporters (Caco-2, MCF7-ADR) will be used to evaluate efficacy against multi-factorial
drug resistance.
The ultimate goal of this proposed study is to generate interaction profiles
interaction profiles for novel and existing drugs to predict the effects of combinations on
169 pharmacokinetics. This database of profiles will be similar to the PDSP database but will solely focus on MDR related ABC transporters. This information will be valuable in predicting both the safety and efficacy of drug cocktails.
170 Figure 4.1 Schematic of chapter 2 conclusions. d5-NITP was identified as a selective inhibitor of gp44/62 resulting in inhibition of T4 plaque formation while leaving the E.coli
host unperturbed. Arg175 of gp44/62 was shown to be an electrostatic contributor for drug
binding. This contribution was confirmed using site directed mutagenesis. The ability of d5-
NITP to inhibit T4 processive DNA replication was shown in a strand displacement. This
was followed by a plaque forming assay to evaluate the ability of d5-NI to inhibit T4 plaque
formation in vivo . The in vivo results show an inhibition of plaque formation upon treatment
with d5-NI suggesting that the d5-NI is converted to d5-NITP which then inhibits
processive DNA replication and lessens productive T4 infection.
171 Figure 4.1.
γγγ-complex βββ-clamp
ATP
ADP Bacteriophage Pi 5 3 3 5 E.col
gp44/62
gp45
ATP
ADP Pi
5 3 3 5 ’
172 Figure 4.2 Kinase profiling of non-natural nucleo(s)tides. Left side: Schematic of the human
kinome (Cell signaling). Upper right side: Virtual screening and commercial in vitro screening
will generate kinase profiles that can be used to predict the signaling pathways that may be
affected by non-natural nucleo(s)tides. Lower right side: Schematic of a representative data
set from kinase screening. Invitrogen’s kinase profiling service screens 315 kinases and
obtains IC 50 values for each target. ATP concentration will be set at the K m level for each kinase and the scale of inhibition will range from pink (higest IC 50 values) to black (lowest
IC 50 values). For in silico screening, the color scale will represent the docking scores with pink being the lowest and black being the highest.
173 Figure 4.2
174 Figure 4.3 Chemical structures of ligands for the 5HT2, histamine H2, adenosine, and
melatonin receptors. (A) histamine (B) serotonin, (C), Indolyl scaffold of non-natural
nucleoside (D) adenosine, (E) melatonin (F) cimetidine.
175 Figure 4.3
(A) (B)
HO NH2 N
N HN NH2 H
(C) (D)
NH2 X N N
N HO N N O HO O
OH HO HO
(E) (F)
S N O NH HN N O N HN N H HN
176
Figure 4.4 Schematic of model correlating changes in p-glycoprotein ATPase catalytic efficiency with changes in p-glycoprotein mediated drug resistance. d5-CHI was shown to inhibit p-glycoprotein mediated efflux of calcein-AM by acting as an alternative substrate. In addition, d5-CHI and other chemotherapeutic substrates were shown to activate p- glycoprotein ATPase activity with Michaelis-Menten kinetics Uniquely, d5-CHI acts as a non-transportable substrate which influences the transport activity of common chemotherapeutics in a positive or negative manner. Upon examination of the ATPase
catalytic efficiency (V max /K m) of p-glycoprotein substrates in the absence and presence of d5-
CHI, it was found that the Vmax /K m(drug + d5-CHI) / V max /K m(drug) was in direct correlation with changes in drug resistance.
177 Figure 4.4
178 Table 4.1 ABC transporters involved in multidrug resistance and their known substrates and tissues
179 References
1. Haystead, T. A. (2006) The purinome, a complex mix of drug and toxicity targets, Curr Top Med Chem 6, 1117-1127. 2. Johnson, S. A., and Hunter, T. (2005) Kinomics: methods for deciphering the kinome, Nat Methods 2, 17-25. 3. Alberts, B., Wilson, J. H., and Hunt, T. (2008) Molecular biology of the cell , 5th ed., Garland Science, New York. 4. Akritopoulou-Zanze, I., and Hajduk, P. J. (2009) Kinase-targeted libraries: the design and synthesis of novel, potent, and selective kinase inhibitors, Drug Discov Today 14 , 291-297. 5. Keith, C. T., Borisy, A. A., and Stockwell, B. R. (2005) Multicomponent therapeutics for networked systems, Nat Rev Drug Discov 4, 71-78. 6. Chene, P. (2002) ATPases as drug targets: learning from their structure, Nat Rev Drug Discov 1, 665-673. 7. Berg, J. M., Tymoczko, J. L., Stryer, L., and Stryer, L. (2007) Biochemistry , 6th ed., W.H. Freeman, New York. 8. Scott, C. B. (2008) A primer for the exercise and nutrition sciences : thermodynamics, bioenergetics, metabolism , Springer, Totowa, NJ. 9. Davey, M. J., Jeruzalmi, D., Kuriyan, J., and O'Donnell, M. (2002) Motors and switches: AAA+ machines within the replisome, Nat Rev Mol Cell Biol 3, 826-835. 10. Dean, M. (2005) The genetics of ATP-binding cassette transporters, Methods Enzymol 400 , 409-429. 11. Walker, J. E., Saraste, M., Runswick, M. J., and Gay, N. J. (1982) Distantly related sequences in the alpha- and beta-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold, Embo J 1, 945- 951. 12. Saraste, M., Sibbald, P. R., and Wittinghofer, A. (1990) The P-loop--a common motif in ATP- and GTP-binding proteins, Trends Biochem Sci 15 , 430-434. 13. Moodie, S. L., Mitchell, J. B., and Thornton, J. M. (1996) Protein recognition of adenylate: an example of a fuzzy recognition template, J Mol Biol 263 , 486-500. 14. Roe, S. M., Prodromou, C., O'Brien, R., Ladbury, J. E., Piper, P. W., and Pearl, L. H. (1999) Structural basis for inhibition of the Hsp90 molecular chaperone by the antitumor antibiotics radicicol and geldanamycin, J Med Chem 42 , 260-266. 15. Zhang, C., and Bollag, G. (2009) Scaffold-based design of kinase inhibitors for cancer therapy, Curr Opin Genet Dev . 16. Johnson, S. A. (2000) Clinical pharmacokinetics of nucleoside analogues: focus on haematological malignancies, Clin Pharmacokinet 39 , 5-26. 17. Schindler, T., Bornmann, W., Pellicena, P., Miller, W. T., Clarkson, B., and Kuriyan, J. (2000) Structural mechanism for STI-571 inhibition of abelson tyrosine kinase, Science 289 , 1938-1942. 18. Druker, B. J., Tamura, S., Buchdunger, E., Ohno, S., Segal, G. M., Fanning, S., Zimmermann, J., and Lydon, N. B. (1996) Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells, Nat Med 2, 561-566.
180 19. Carter, C. A., Kelly, R. J., and Giaccone, G. (2009) Small-molecule inhibitors of the human epidermal receptor family, Expert Opin Investig Drugs 18 , 1829-1842. 20. Yu, Z., Boggon, T. J., Kobayashi, S., Jin, C., Ma, P. C., Dowlati, A., Kern, J. A., Tenen, D. G., and Halmos, B. (2007) Resistance to an irreversible epidermal growth factor receptor (EGFR) inhibitor in EGFR-mutant lung cancer reveals novel treatment strategies, Cancer Res 67 , 10417-10427. 21. Hu, T., Sage, H., and Hsieh, T. S. (2002) ATPase domain of eukaryotic DNA topoisomerase II. Inhibition of ATPase activity by the anti-cancer drug bisdioxopiperazine and ATP/ADP-induced dimerization, J Biol Chem 277 , 5944- 5951. 22. Eichhorn, E. J., and Gheorghiade, M. (2002) Digoxin, Prog Cardiovasc Dis 44 , 251- 266. 23. Mayer, T. U., Kapoor, T. M., Haggarty, S. J., King, R. W., Schreiber, S. L., and Mitchison, T. J. (1999) Small molecule inhibitor of mitotic spindle bipolarity identified in a phenotype-based screen, Science 286 , 971-974. 24. Borowski, P., Niebuhr, A., Schmitz, H., Hosmane, R. S., Bretner, M., Siwecka, M. A., and Kulikowski, T. (2002) NTPase/helicase of Flaviviridae: inhibitors and inhibition of the enzyme, Acta Biochim Pol 49 , 597-614. 25. Groen, A. C., Needleman, D., Brangwynne, C., Gradinaru, C., Fowler, B., Mazitschek, R., and Mitchison, T. J. (2008) A novel small-molecule inhibitor reveals a possible role of kinesin-5 in anastral spindle-pole assembly, J Cell Sci 121 , 2293- 2300. 26. Zhang, X., Donnelly, A., Lee, I., and Berdis, A. J. (2006) Rational attempts to optimize non-natural nucleotides for selective incorporation opposite an abasic site, Biochemistry 45 , 13293-13303. 27. Zhang, X., Lee, I., and Berdis, A. J. (2004) Evaluating the contributions of desolvation and base-stacking during translesion DNA synthesis, Org Biomol Chem 2, 1703-1711. 28. Zhang, X., Lee, I., and Berdis, A. J. (2005) A potential chemotherapeutic strategy for the selective inhibition of promutagenic DNA synthesis by nonnatural nucleotides, Biochemistry 44 , 13111-13121. 29. Zhang, X., Lee, I., and Berdis, A. J. (2005) The use of nonnatural nucleotides to probe the contributions of shape complementarity and pi-electron surface area during DNA polymerization, Biochemistry 44 , 13101-13110. 30. Zhang, X., Lee, I., Zhou, X., and Berdis, A. J. (2006) Hydrophobicity, shape, and pi- electron contributions during translesion DNA synthesis, J Am Chem Soc 128 , 143- 149. 31. Berdis, A. J. (2008) DNA polymerases as therapeutic targets, Biochemistry 47 , 8253- 8260. 32. Batista, L. F., Kaina, B., Meneghini, R., and Menck, C. F. (2009) How DNA lesions are turned into powerful killing structures: insights from UV-induced apoptosis, Mutat Res 681 , 197-208. 33. Shibutani, S., Takeshita, M., and Grollman, A. P. (1997) Translesional synthesis on DNA templates containing a single abasic site. A mechanistic study of the "A rule", J Biol Chem 272 , 13916-13922. 34. Reineks, E. Z., and Berdis, A. J. (2004) Evaluating the contribution of base stacking during translesion DNA replication, Biochemistry 43 , 393-404.
181 35. Longley, D. B., Harkin, D. P., and Johnston, P. G. (2003) 5-fluorouracil: mechanisms of action and clinical strategies, Nat Rev Cancer 3, 330-338. 36. Bonate, P. L., Arthaud, L., Cantrell, W. R., Jr., Stephenson, K., Secrist, J. A., 3rd, and Weitman, S. (2006) Discovery and development of clofarabine: a nucleoside analogue for treating cancer, Nat Rev Drug Discov 5, 855-863. 37. Heinemann, V., Schulz, L., Issels, R. D., and Plunkett, W. (1995) Gemcitabine: a modulator of intracellular nucleotide and deoxynucleotide metabolism, Semin Oncol 22 , 11-18. 38. Perry, M. C. (2008) The chemotherapy source book , 4th ed., Wolters Kluwer Health/Lippincott Williams & Wilkins, Philadelphia. 39. Gallois-Montbrun, S., Schneider, B., Chen, Y., Giacomoni-Fernandes, V., Mulard, L., Morera, S., Janin, J., Deville-Bonne, D., and Veron, M. (2002) Improving nucleoside diphosphate kinase for antiviral nucleotide analogs activation, J Biol Chem 277 , 39953- 39959. 40. De Clercq, E., and Field, H. J. (2006) Antiviral prodrugs - the development of successful prodrug strategies for antiviral chemotherapy, Br J Pharmacol 147 , 1-11. 41. Goodman, L. S., Gilman, A., Brunton, L. L., Lazo, J. S., and Parker, K. L. (2006) Goodman & Gilman's the pharmacological basis of therapeutics , 11th ed., McGraw-Hill, New York. 42. Mace, D. C., and Alberts, B. M. (1984) The complex of T4 bacteriophage gene 44 and 62 replication proteins forms an ATPase that is stimulated by DNA and by T4 gene 45 protein, J Mol Biol 177 , 279-293. 43. Venkatesan, M., and Nossal, N. G. (1982) Bacteriophage T4 gene 44/62 and gene 45 polymerase accessory proteins stimulate hydrolysis of duplex DNA by T4 DNA polymerase, J Biol Chem 257 , 12435-12443. 44. Mace, D. C., and Alberts, B. M. (1984) Characterization of the stimulatory effect of T4 gene 45 protein and the gene 44/62 protein complex on DNA synthesis by T4 DNA polymerase, J Mol Biol 177 , 313-327. 45. Nossal, N. G. (1992) Protein-protein interactions at a DNA replication fork: bacteriophage T4 as a model, Faseb J 6, 871-878. 46. Johnson, A., and O'Donnell, M. (2005) Cellular DNA replicases: components and dynamics at the replication fork, Annu Rev Biochem 74 , 283-315. 47. Morris, C. F., Hama-Inaba, H., Mace, D., Sinha, N. K., and Alberts, B. (1979) Purification of the gene 43, 44, 45, and 62 proteins of the bacteriophage T4 DNA replication apparatus, J Biol Chem 254 , 6787-6796. 48. Pritchard, A. E., Dallmann, H. G., Glover, B. P., and McHenry, C. S. (2000) A novel assembly mechanism for the DNA polymerase III holoenzyme DnaX complex: association of deltadelta' with DnaX(4) forms DnaX(3)deltadelta', Embo J 19 , 6536- 6545. 49. Kaboord, B. F., and Benkovic, S. J. (1995) Accessory proteins function as matchmakers in the assembly of the T4 DNA polymerase holoenzyme, Curr Biol 5, 149-157. 50. Kaboord, B. F., and Benkovic, S. J. (1996) Dual role of the 44/62 protein as a matchmaker protein and DNA polymerase chaperone during assembly of the bacteriophage T4 holoenzyme complex, Biochemistry 35 , 1084-1092. 51. Berdis, A. J., and Benkovic, S. J. (1996) Role of adenosine 5'-triphosphate hydrolysis in the assembly of the bacteriophage T4 DNA replication holoenzyme complex, Biochemistry 35 , 9253-9265.
182 52. Hingorani, M. M., and O'Donnell, M. (1998) ATP binding to the Escherichia coli clamp loader powers opening of the ring-shaped clamp of DNA polymerase III holoenzyme, J Biol Chem 273 , 24550-24563. 53. Hingorani, M. M., Bloom, L. B., Goodman, M. F., and O'Donnell, M. (1999) Division of labor--sequential ATP hydrolysis drives assembly of a DNA polymerase sliding clamp around DNA, Embo J 18 , 5131-5144. 54. Chen, S., Levin, M. K., Sakato, M., Zhou, Y., and Hingorani, M. M. (2009) Mechanism of ATP-driven PCNA clamp loading by S. cerevisiae RFC, J Mol Biol 388 , 431-442. 55. Schmidt, S. L., Gomes, X. V., and Burgers, P. M. (2001) ATP utilization by yeast replication factor C. III. The ATP-binding domains of Rfc2, Rfc3, and Rfc4 are essential for DNA recognition and clamp loading, J Biol Chem 276 , 34784-34791. 56. Johnson, A., Yao, N. Y., Bowman, G. D., Kuriyan, J., and O'Donnell, M. (2006) The replication factor C clamp loader requires arginine finger sensors to drive DNA binding and proliferating cell nuclear antigen loading, J Biol Chem 281 , 35531-35543. 57. Shi, J., Blundell, T. L., and Mizuguchi, K. (2001) FUGUE: sequence-structure homology recognition using environment-specific substitution tables and structure- dependent gap penalties, J Mol Biol 310 , 243-257. 58. Ling, V., and Thompson, L. H. (1974) Reduced permeability in CHO cells as a mechanism of resistance to colchicine, J Cell Physiol 83 , 103-116. 59. Ling, V. (1975) Drug resistance and membrane alteration in mutants of mammalian cells, Can J Genet Cytol 17 , 503-515. 60. Huls, M., Russel, F. G., and Masereeuw, R. (2009) The role of ATP binding cassette transporters in tissue defense and organ regeneration, J Pharmacol Exp Ther 328 , 3-9. 61. Eckford, P. D., and Sharom, F. J. (2009) ABC efflux pump-based resistance to chemotherapy drugs, Chem Rev 109 , 2989-3011. 62. Ichikawa-Haraguchi, M., Sumizawa, T., Yoshimura, A., Furukawa, T., Hiramoto, S., Sugita, M., and Akiyama, S. (1993) Progesterone and its metabolites: the potent inhibitors of the transporting activity of P-glycoprotein in the adrenal gland, Biochim Biophys Acta 1158 , 201-208. 63. Schinkel, A. H., Smit, J. J., van Tellingen, O., Beijnen, J. H., Wagenaar, E., van Deemter, L., Mol, C. A., van der Valk, M. A., Robanus-Maandag, E. C., te Riele, H. P., and et al. (1994) Disruption of the mouse mdr1a P-glycoprotein gene leads to a deficiency in the blood-brain barrier and to increased sensitivity to drugs, Cell 77 , 491-502. 64. Goldstein, L. J., Galski, H., Fojo, A., Willingham, M., Lai, S. L., Gazdar, A., Pirker, R., Green, A., Crist, W., Brodeur, G. M., and et al. (1989) Expression of a multidrug resistance gene in human cancers, J Natl Cancer Inst 81 , 116-124. 65. Fojo, A. T., Ueda, K., Slamon, D. J., Poplack, D. G., Gottesman, M. M., and Pastan, I. (1987) Expression of a multidrug-resistance gene in human tumors and tissues, Proc Natl Acad Sci U S A 84 , 265-269. 66. Fojo, A., Cornwell, M., Cardarelli, C., Clark, D. P., Richert, N., Shen, D. W., Ueda, K., Willingham, M., Gottesman, M. M., and Pastan, I. (1987) Molecular biology of drug resistance, Breast Cancer Res Treat 9, 5-16. 67. Noonan, K. E., Beck, C., Holzmayer, T. A., Chin, J. E., Wunder, J. S., Andrulis, I. L., Gazdar, A. F., Willman, C. L., Griffith, B., Von Hoff, D. D., and et al. (1990) Quantitative analysis of MDR1 (multidrug resistance) gene expression in human tumors by polymerase chain reaction, Proc Natl Acad Sci U S A 87 , 7160-7164.
183 68. Bourhis, J., Goldstein, L. J., Riou, G., Pastan, I., Gottesman, M. M., and Benard, J. (1989) Expression of a human multidrug resistance gene in ovarian carcinomas, Cancer Res 49 , 5062-5065. 69. Penson, R. T., Oliva, E., Skates, S. J., Glyptis, T., Fuller, A. F., Jr., Goodman, A., and Seiden, M. V. (2004) Expression of multidrug resistance-1 protein inversely correlates with paclitaxel response and survival in ovarian cancer patients: a study in serial samples, Gynecol Oncol 93 , 98-106. 70. Han, K., Kahng, J., Kim, M., Lim, J., Kim, Y., Cho, B., Kim, H. K., Min, W. S., Kim, C. C., Lee, K. Y., Kim, B. K., and Kang, C. S. (2000) Expression of functional markers in acute nonlymphoblastic leukemia, Acta Haematol 104 , 174-180. 71. Higgins, C. F., and Gottesman, M. M. (1992) Is the multidrug transporter a flippase?, Trends Biochem Sci 17 , 18-21. 72. Romsicki, Y., and Sharom, F. J. (1999) The membrane lipid environment modulates drug interactions with the P-glycoprotein multidrug transporter, Biochemistry 38 , 6887- 6896. 73. Eytan, G. D., Regev, R., Oren, G., Hurwitz, C. D., and Assaraf, Y. G. (1997) Efficiency of P-glycoprotein-mediated exclusion of rhodamine dyes from multidrug- resistant cells is determined by their passive transmembrane movement rate, Eur J Biochem 248 , 104-112. 74. Aller, S. G., Yu, J., Ward, A., Weng, Y., Chittaboina, S., Zhuo, R., Harrell, P. M., Trinh, Y. T., Zhang, Q., Urbatsch, I. L., and Chang, G. (2009) Structure of P- glycoprotein reveals a molecular basis for poly-specific drug binding, Science 323 , 1718-1722. 75. Polli, J. W., Wring, S. A., Humphreys, J. E., Huang, L., Morgan, J. B., Webster, L. O., and Serabjit-Singh, C. S. (2001) Rational use of in vitro P-glycoprotein assays in drug discovery, J Pharmacol Exp Ther 299 , 620-628. 76. Liu, R., and Sharom, F. J. (1996) Site-directed fluorescence labeling of P-glycoprotein on cysteine residues in the nucleotide binding domains, Biochemistry 35 , 11865-11873. 77. Higgins, C. F., and Linton, K. J. (2004) The ATP switch model for ABC transporters, Nat Struct Mol Biol 11 , 918-926. 78. Sauna, Z. E., and Ambudkar, S. V. (2007) About a switch: how P-glycoprotein (ABCB1) harnesses the energy of ATP binding and hydrolysis to do mechanical work, Mol Cancer Ther 6, 13-23. 79. Cornwell, M. M., Pastan, I., and Gottesman, M. M. (1987) Certain calcium channel blockers bind specifically to multidrug-resistant human KB carcinoma membrane vesicles and inhibit drug binding to P-glycoprotein, J Biol Chem 262 , 2166-2170. 80. Goldberg, H., Ling, V., Wong, P. Y., and Skorecki, K. (1988) Reduced cyclosporin accumulation in multidrug-resistant cells, Biochem Biophys Res Commun 152 , 552-558. 81. Starling, J. J., Shepard, R. L., Cao, J., Law, K. L., Norman, B. H., Kroin, J. S., Ehlhardt, W. J., Baughman, T. M., Winter, M. A., Bell, M. G., Shih, C., Gruber, J., Elmquist, W. F., and Dantzig, A. H. (1997) Pharmacological characterization of LY335979: a potent cyclopropyldibenzosuberane modulator of P-glycoprotein, Adv Enzyme Regul 37 , 335-347. 82. Martin, C., Berridge, G., Mistry, P., Higgins, C., Charlton, P., and Callaghan, R. (1999) The molecular interaction of the high affinity reversal agent XR9576 with P- glycoprotein, Br J Pharmacol 128 , 403-411. 83. Gerson, S. L. (2002) Clinical relevance of MGMT in the treatment of cancer, J Clin Oncol 20 , 2388-2399.
184 84. David Golan, A. T. J., Ehrin Armstrong, Joshua Galanter, April Armstrong, Ramy Arnaout, Harris Rose (2005) Principles of Pharmacology: The Pathophysiologic Basis of Drug Therapy , Lippincott Williams & Wilkens, Philidephia. 85. Laghi, L., Bianchi, P., and Malesci, A. (2008) Differences and evolution of the methods for the assessment of microsatellite instability, Oncogene 27 , 6313-6321. 86. Vigouroux, C., Gharakhanian, S., Salhi, Y., Nguyen, T. H., Adda, N., Rozenbaum, W., and Capeau, J. (1999) Adverse metabolic disorders during highly active antiretroviral treatments (HAART) of HIV disease, Diabetes Metab 25 , 383-392. 87. Allan, J. M., and Travis, L. B. (2005) Mechanisms of therapy-related carcinogenesis, Nat Rev Cancer 5, 943-955. 88. Fradkin, L. G., and Kornberg, A. (1992) Prereplicative complexes of components of DNA polymerase III holoenzyme of Escherichia coli, J Biol Chem 267 , 10318-10322. 89. Frey, M. W., Nossal, N. G., Capson, T. L., and Benkovic, S. J. (1993) Construction and characterization of a bacteriophage T4 DNA polymerase deficient in 3'-->5' exonuclease activity, Proc Natl Acad Sci U S A 90 , 2579-2583. 90. Alley, S. C., Shier, V. K., Abel-Santos, E., Sexton, D. J., Soumillion, P., and Benkovic, S. J. (1999) Sliding clamp of the bacteriophage T4 polymerase has open and closed subunit interfaces in solution, Biochemistry 38 , 7696-7709. 91. Shirasaka, Y., Onishi, Y., Sakurai, A., Nakagawa, H., Ishikawa, T., and Yamashita, S. (2006) Evaluation of human P-glycoprotein (MDR1/ABCB1) ATPase activity assay method by comparing with in vitro transport measurements: Michaelis-Menten kinetic analysis to estimate the affinity of P-glycoprotein to drugs, Biol Pharm Bull 29 , 2465-2471. 92. Hiraga, S., Igarashi, K., and Yura, T. (1967) A deoxythymidine kinase-deficient mutant of Escherichia coli. I. Isolation and some properties, Biochim Biophys Acta 145 , 41-51. 93. Igarashi, K., Hiraga, S., and Yura, T. (1967) A deoxythymidine kinase deficient mutant of Escherichia coli. II. Mapping and transduction studies with phage phi 80, Genetics 57 , 643-654. 94. Schuttelkopf, A. W., and van Aalten, D. M. (2004) PRODRG: a tool for high- throughput crystallography of protein-ligand complexes, Acta Crystallogr D Biol Crystallogr 60 , 1355-1363. 95. Emsley, P., and Cowtan, K. (2004) Coot: model-building tools for molecular graphics, Acta Crystallogr D Biol Crystallogr 60 , 2126-2132. 96. Copeland, R. (2005) Evaluation of Enzyme Inhibitors in Drug Discovery: A Guide for Medicinal Chemists and Pharmacologists , 1st ed., Wiley-Interscience, New York. 97. Berdis, A. J., and Benkovic, S. J. (1997) Mechanism of bacteriophage T4 DNA holoenzyme assembly: the 44/62 protein acts as a molecular motor, Biochemistry 36 , 2733-2743. 98. Reineks, E. Z., and Berdis, A. J. (2003) Evaluating the effects of enhanced processivity and metal ions on translesion DNA replication catalyzed by the bacteriophage T4 DNA polymerase, J Mol Biol 328 , 1027-1045. 99. Kincaid, K., Beckman, J., Zivkovic, A., Halcomb, R. L., Engels, J. W., and Kuchta, R. D. (2005) Exploration of factors driving incorporation of unnatural dNTPS into DNA by Klenow fragment (DNA polymerase I) and DNA polymerase alpha, Nucleic Acids Res 33 , 2620-2628.
185 100. Alley, S. C., Abel-Santos, E., and Benkovic, S. J. (2000) Tracking sliding clamp opening and closing during bacteriophage T4 DNA polymerase holoenzyme assembly, Biochemistry 39 , 3076-3090. 101. Kelley, L. A., and Sternberg, M. J. (2009) Protein structure prediction on the Web: a case study using the Phyre server, Nat Protoc 4, 363-371. 102. Jim D. Karam, J. W. D. (1994) Molecular biology of bacteriophage T4 , illustrated ed., American Society for Microbiology, Washington DC. 103. Johnson, A., and O'Donnell, M. (2003) Ordered ATP hydrolysis in the gamma complex clamp loader AAA+ machine, J Biol Chem 278 , 14406-14413. 104. Young, M. C., Weitzel, S. E., and von Hippel, P. H. (1996) The kinetic mechanism of formation of the bacteriophage T4 DNA polymerase sliding clamp, J Mol Biol 264 , 440-452. 105. Gottesman, M. M., and Ling, V. (2006) The molecular basis of multidrug resistance in cancer: the early years of P-glycoprotein research, FEBS Lett 580 , 998-1009. 106. Juliano, R. L., and Ling, V. (1976) A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants, Biochim Biophys Acta 455 , 152- 162. 107. Ling, V., Kartner, N., Sudo, T., Siminovitch, L., and Riordan, J. R. (1983) Multidrug- resistance phenotype in Chinese hamster ovary cells, Cancer Treat Rep 67 , 869-874. 108. Szakacs, G., Paterson, J. K., Ludwig, J. A., Booth-Genthe, C., and Gottesman, M. M. (2006) Targeting multidrug resistance in cancer, Nat Rev Drug Discov 5, 219-234. 109. Higgins, C. F., Hiles, I. D., Salmond, G. P., Gill, D. R., Downie, J. A., Evans, I. J., Holland, I. B., Gray, L., Buckel, S. D., Bell, A. W., and et al. (1986) A family of related ATP-binding subunits coupled to many distinct biological processes in bacteria, Nature 323 , 448-450. 110. Germann, U. A., Shlyakhter, D., Mason, V. S., Zelle, R. E., Duffy, J. P., Galullo, V., Armistead, D. M., Saunders, J. O., Boger, J., and Harding, M. W. (1997) Cellular and biochemical characterization of VX-710 as a chemosensitizer: reversal of P- glycoprotein-mediated multidrug resistance in vitro, Anticancer Drugs 8, 125-140. 111. Dhainaut, A., Regnier, G., Atassi, G., Pierre, A., Leonce, S., Kraus-Berthier, L., and Prost, J. F. (1992) New triazine derivatives as potent modulators of multidrug resistance, J Med Chem 35 , 2481-2496. 112. McDevitt, C. A., and Callaghan, R. (2007) How can we best use structural information on P-glycoprotein to design inhibitors?, Pharmacol Ther 113 , 429-441. 113. De Azevedo, W. F., Leclerc, S., Meijer, L., Havlicek, L., Strnad, M., and Kim, S. H. (1997) Inhibition of cyclin-dependent kinases by purine analogues: crystal structure of human cdk2 complexed with roscovitine, Eur J Biochem 243 , 518-526. 114. Whittaker, S. R., Walton, M. I., Garrett, M. D., and Workman, P. (2004) The Cyclin- dependent kinase inhibitor CYC202 (R-roscovitine) inhibits retinoblastoma protein phosphorylation, causes loss of Cyclin D1, and activates the mitogen-activated protein kinase pathway, Cancer Res 64 , 262-272. 115. King, K. M., Damaraju, V. L., Vickers, M. F., Yao, S. Y., Lang, T., Tackaberry, T. E., Mowles, D. A., Ng, A. M., Young, J. D., and Cass, C. E. (2006) A comparison of the transportability, and its role in cytotoxicity, of clofarabine, cladribine, and fludarabine by recombinant human nucleoside transporters produced in three model expression systems, Mol Pharmacol 69 , 346-353.
186 116. Garrigues, A., Loiseau, N., Delaforge, M., Ferte, J., Garrigos, M., Andre, F., and Orlowski, S. (2002) Characterization of two pharmacophores on the multidrug transporter P-glycoprotein, Mol Pharmacol 62 , 1288-1298. 117. Shapiro, A. B., and Ling, V. (1997) Positively cooperative sites for drug transport by P-glycoprotein with distinct drug specificities, Eur J Biochem 250 , 130-137. 118. Boelsterli, U. A., Ho, H. K., Zhou, S., and Leow, K. Y. (2006) Bioactivation and hepatotoxicity of nitroaromatic drugs, Curr Drug Metab 7, 715-727. 119. Borges, N. (2005) Tolcapone in Parkinson's disease: liver toxicity and clinical efficacy, Expert Opin Drug Saf 4, 69-73. 120. Wilhelm, S., Carter, C., Lynch, M., Lowinger, T., Dumas, J., Smith, R. A., Schwartz, B., Simantov, R., and Kelley, S. (2006) Discovery and development of sorafenib: a multikinase inhibitor for treating cancer, Nat Rev Drug Discov 5, 835-844. 121. Davies, M. A., Compton-Toth, B. A., Hufeisen, S. J., Meltzer, H. Y., and Roth, B. L. (2005) The highly efficacious actions of N-desmethylclozapine at muscarinic receptors are unique and not a common property of either typical or atypical antipsychotic drugs: is M1 agonism a pre-requisite for mimicking clozapine's actions?, Psychopharmacology (Berl) 178 , 451-460. 122. Ozeki, Y., Yamada, N., Aoki, H., Yokoyama, T., Shirono, H., Koga, J., and Kato, N. (1999) A newly developed assay for melatonin using cells expressing human mel-1a receptor, Psychiatry Clin Neurosci 53 , 247-248. 123. Teh, M. T., and Sugden, D. (1998) Comparison of the structure-activity relationships of melatonin receptor agonists and antagonists: lengthening the N-acyl side-chain has differing effects on potency on Xenopus melanophores, Naunyn Schmiedebergs Arch Pharmacol 358 , 522-528. 124. Shiffman, M. L., Suter, F., Bacon, B. R., Nelson, D., Harley, H., Sola, R., Shafran, S. D., Barange, K., Lin, A., Soman, A., and Zeuzem, S. (2007) Peginterferon alfa-2a and ribavirin for 16 or 24 weeks in HCV genotype 2 or 3, N Engl J Med 357 , 124-134. 125. Mistry, N., Shapero, J., and Crawford, R. I. (2009) A review of adverse cutaneous drug reactions resulting from the use of interferon and ribavirin, Can J Gastroenterol 23 , 677-683. 126. Venkatraman, S., and Njoroge, F. G. (2007) Macrocyclic inhibitors of HCV NS3-4A protease: design and structure activity relationship, Curr Top Med Chem 7, 1290-1301. 127. Kim, J. W., Seo, M. Y., Shelat, A., Kim, C. S., Kwon, T. W., Lu, H. H., Moustakas, D. T., Sun, J., and Han, J. H. (2003) Structurally conserved amino Acid w501 is required for RNA helicase activity but is not essential for DNA helicase activity of hepatitis C virus NS3 protein, J Virol 77 , 571-582. 128. Locatelli, G. A., Spadari, S., and Maga, G. (2002) Hepatitis C virus NS3 ATPase/helicase: an ATP switch regulates the cooperativity among the different substrate binding sites, Biochemistry 41 , 10332-10342. 129. Cho, H. S., Ha, N. C., Kang, L. W., Chung, K. M., Back, S. H., Jang, S. K., and Oh, B. H. (1998) Crystal structure of RNA helicase from genotype 1b hepatitis C virus. A feasible mechanism of unwinding duplex RNA, J Biol Chem 273 , 15045-15052. 130. Borowski, P., Schalinski, S., and Schmitz, H. (2002) Nucleotide triphosphatase/helicase of hepatitis C virus as a target for antiviral therapy, Antiviral Res 55 , 397-412. 131. Bretner, M., Schalinski, S., Haag, A., Lang, M., Schmitz, H., Baier, A., Behrens, S. E., Kulikowski, T., and Borowski, P. (2004) Synthesis and evaluation of ATP-binding site directed potential inhibitors of nucleoside triphosphatases/helicases and
187 polymerases of hepatitis C and other selected Flaviviridae viruses, Antivir Chem Chemother 15 , 35-42. 132. Jin, L., and Peterson, D. L. (1995) Expression, isolation, and characterization of the hepatitis C virus ATPase/RNA helicase, Arch Biochem Biophys 323 , 47-53. 133. Borowski, P., Heising, M. V., Miranda, I. B., Liao, C. L., Choe, J., and Baier, A. (2008) Viral NS3 helicase activity is inhibited by peptides reproducing the Arg-rich conserved motif of the enzyme (motif VI), Biochem Pharmacol 76 , 28-38. 134. Dumont, S., Cheng, W., Serebrov, V., Beran, R. K., Tinoco, I., Jr., Pyle, A. M., and Bustamante, C. (2006) RNA translocation and unwinding mechanism of HCV NS3 helicase and its coordination by ATP, Nature 439 , 105-108. 135. Lindenbach, B. D., Meuleman, P., Ploss, A., Vanwolleghem, T., Syder, A. J., McKeating, J. A., Lanford, R. E., Feinstone, S. M., Major, M. E., Leroux-Roels, G., and Rice, C. M. (2006) Cell culture-grown hepatitis C virus is infectious in vivo and can be recultured in vitro, Proc Natl Acad Sci U S A 103 , 3805-3809. 136. Barth, H., Robinet, E., Liang, T. J., and Baumert, T. F. (2008) Mouse models for the study of HCV infection and virus-host interactions, J Hepatol 49 , 134-142. 137. Kondratov, R. V., Komarov, P. G., Becker, Y., Ewenson, A., and Gudkov, A. V. (2001) Small molecules that dramatically alter multidrug resistance phenotype by modulating the substrate specificity of P-glycoprotein, Proc Natl Acad Sci U S A 98 , 14078-14083. 138. Sterz, K., Mollmann, L., Jacobs, A., Baumert, D., and Wiese, M. (2009) Activators of P-glycoprotein: Structure-Activity Relationships and Investigation of their Mode of Action, ChemMedChem . 139. Eng, K. T., and Berdis, A. J. (2010) A novel non-natural nucleoside that influences P-glycoprotein activity and mediates drug resistance, Biochemistry 49 , 1640-1648. 140. Szakacs, G., Jakab, K., Antal, F., and Sarkadi, B. (1998) Diagnostics of multidrug resistance in cancer, Pathol Oncol Res 4, 251-257. 141. Rautio, J., Humphreys, J. E., Webster, L. O., Balakrishnan, A., Keogh, J. P., Kunta, J. R., Serabjit-Singh, C. J., and Polli, J. W. (2006) In vitro p-glycoprotein inhibition assays for assessment of clinical drug interaction potential of new drug candidates: a recommendation for probe substrates, Drug Metab Dispos 34 , 786-792. 142. Lagas, J. S., van Waterschoot, R. A., van Tilburg, V. A., Hillebrand, M. J., Lankheet, N., Rosing, H., Beijnen, J. H., and Schinkel, A. H. (2009) Brain accumulation of dasatinib is restricted by P-glycoprotein (ABCB1) and breast cancer resistance protein (ABCG2) and can be enhanced by elacridar treatment, Clin Cancer Res 15 , 2344-2351. 143. Hugnet, C., Lespine, A., and Alvinerie, M. (2007) Multiple oral dosing of ketoconazole increases dog exposure to ivermectin, J Pharm Pharm Sci 10 , 311-318. 144. Baer, M. R., George, S. L., Dodge, R. K., O'Loughlin, K. L., Minderman, H., Caligiuri, M. A., Anastasi, J., Powell, B. L., Kolitz, J. E., Schiffer, C. A., Bloomfield, C. D., and Larson, R. A. (2002) Phase 3 study of the multidrug resistance modulator PSC-833 in previously untreated patients 60 years of age and older with acute myeloid leukemia: Cancer and Leukemia Group B Study 9720, Blood 100 , 1224-1232. 145. Lhomme, C., Joly, F., Walker, J. L., Lissoni, A. A., Nicoletto, M. O., Manikhas, G. M., Baekelandt, M. M., Gordon, A. N., Fracasso, P. M., Mietlowski, W. L., Jones, G. J., and Dugan, M. H. (2008) Phase III study of valspodar (PSC 833) combined with paclitaxel and carboplatin compared with paclitaxel and carboplatin alone in patients
188 with stage IV or suboptimally debulked stage III epithelial ovarian cancer or primary peritoneal cancer, J Clin Oncol 26 , 2674-2682. 146. Gradhand, U., and Kim, R. B. (2008) Pharmacogenomics of MRP transporters (ABCC1-5) and BCRP (ABCG2), Drug Metab Rev 40 , 317-354. 147. Zhou, S. F., Wang, L. L., Di, Y. M., Xue, C. C., Duan, W., Li, C. G., and Li, Y. (2008) Substrates and inhibitors of human multidrug resistance associated proteins and the implications in drug development, Curr Med Chem 15 , 1981-2039. 148. Prime-Chapman, H. M., Fearn, R. A., Cooper, A. E., Moore, V., and Hirst, B. H. (2004) Differential multidrug resistance-associated protein 1 through 6 isoform expression and function in human intestinal epithelial Caco-2 cells, J Pharmacol Exp Ther 311 , 476-484. 149. Ishikawa, T., and Nakagawa, H. (2009) Human ABC transporter ABCG2 in cancer chemotherapy and pharmacogenomics, J Exp Ther Oncol 8, 5-24. 150. Rees, D. C., Johnson, E., and Lewinson, O. (2009) ABC transporters: the power to change, Nat Rev Mol Cell Biol 10 , 218-227. 151. Gutmann, H., Fricker, G., Torok, M., Michael, S., Beglinger, C., and Drewe, J. (1999) Evidence for different ABC-transporters in Caco-2 cells modulating drug uptake, Pharm Res 16 , 402-407. 152. Robey, R. W., Steadman, K., Polgar, O., Morisaki, K., Blayney, M., Mistry, P., and Bates, S. E. (2004) Pheophorbide a is a specific probe for ABCG2 function and inhibition, Cancer Res 64 , 1242-1246. 153. Zamek-Gliszczynski, M. J., Xiong, H., Patel, N. J., Turncliff, R. Z., Pollack, G. M., and Brouwer, K. L. (2003) Pharmacokinetics of 5 (and 6)-carboxy-2',7'- dichlorofluorescein and its diacetate promoiety in the liver, J Pharmacol Exp Ther 304 , 801-809.
189