engineering enhanced enzymes for suicide ......i am earnestly and eternally grateful to my parents...
TRANSCRIPT
ENGINEERING ENHANCED ENZYMES FOR SUICIDE
GENE THERAPY OF CANCER
By
CANDICE LYNN WILLMON
A dissertation submitted in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
WASHINGTON STATE UNIVERSITY Program in Pharmacology and Toxicology
MAY 2006
© Copyright by CANDICE LYNN WILLMON, 2006
All Rights Reserved
© Copyright by CANDICE LYNN WILLMON, 2006 All Rights Reserved
ii
To the Faculty of Washington State University: The members of the Committee appointed to examine the dissertation of CANDICE LYNN WILLMON find it satisfactory and recommend that it be accepted.
___________________________________ Chair
___________________________________
___________________________________ ___________________________________
iii
ACKNOWLEDGMENT
My graduate school experience has been a journey of unbelievable extremes: trying,
rewarding, shaming, at times lonely and, yet, exceptionally fulfilling! I am sincerely indebted to
many people who have influenced my decisions, shaped the beginning of my career and have
just been great friends. Firstly, thank you to my major advisor, Dr. Margaret Black. I know I
have been an amazing pain, I admit it, but she has patiently directed me in how to approach
science and answered my most idiotic questions. I would also like to thank Margaret for her
enthusiasm and encouraging me to succeed. She has not only been an advisor and role model,
but a friend, and hopefully will continue to be so once I leave.
Many other amazing people in my life have paved the road I have followed to this point.
I am earnestly and eternally grateful to my parents and family for always wanting me to be the
best that I could be and showering me in love. Many thanks go to Miryam Salter for being my
“big sister” and guiding me through some of the rough times in my life and not accepting my
self-doubts. Without Mr. David Johnson, my high school chemistry teacher who introduced me
to the world of stoichiometry, I probably wouldn’t have even entertained the idea of going into
science. For my education as an undergraduate, I thank my professors at LCSC. Dr. Christine
Pharr, my undergraduate advisor and the person who pushed me to go to graduate school and
reinforced my anal retentiveness. We relentlessly paid practical jokes on her (sorry about the
balloons and Styrofoam), and while all her plans of retribution usually backfired onto unintended
people, she always forgave us and willing joined us in making chemistry exciting by dressing up
with us as elements during National Chemistry Week, singing her redox songs in class, and
introducing me to liquid nitrogen ice cream. I would also like to thank Dr. Thomas Urquhart.
He is an amazing teacher, whom I hope one day to emulate in my career, and led me down the
iv
path to the biological sciences - the “other” science. Thanks to Dr. William D. Samuels at
Pacific Northwest National Laboratory for taking me into his laboratory and trusting me with all
of those hazardous and combustible chemicals, but mostly, for urging me to get my PhD.
I would like to acknowledge the members of the Black Lab, past and present, which have
helped or contributed to my research – Kathleen Bongiovanni, Chris Detzel, Dr. Elangovan
Gopal, Cathy Knoeber, Lyssa Krabbenhoft, Ling-Yu Kuan, Sheri Mahan, Mariana Sheldon,
Tiffany Stolworthy, and Karina Villa. In addition, Andressa Ardiani has been my joy and my
best labmate since she joined our laboratory. She has been so willing to help me when needed at
any time and I couldn’t thank her enough for this. What will I do without her smile and our
chats? Thanks to the amazing and funny Amanda Goyke - I thank her for befriending me,
showing me around the lab when I first began, having those looks at the most appropriate times
and most of all for putting up with me during my crises. My gratitude also goes to Michi
Fuchita, my braces twin, for her helpfulness and even though she was a quiet presence in our
laboratory, she added happiness. I would like to acknowledge Dr. Jean-Emmanuel Kurtz and
his preliminary work on implementing the dCK screening. Sincerest thanks also go to our
collaborators at the Fred Hutchinson Cancer Research Center, especially Aaron Korkegian, Dr.
Django Sussman and Dr. Barry Stoddard for their amazing science and helpfulness.
Additionally, I am grateful to Mark Kokoris.
Without the help of my committee members I certainly wouldn’t be writing this! I
appreciate the time and energy they have put into listening to me, reading my qualifying and
preliminary exams, writing recommendation letters and, of course, now having to thumb through
this monstrosity. I would especially like to thank Dr. Lisa Gloss for helping me with my
kinetics, appreciating my giggle and just giving me general guidance. Her seeming tough
v
demeanor scared me in the beginning; however, she is very sincere, kind and approachable and
has become one of my favorite professors. The same can be said for Dr. Joseph Harding and his
amazing knowledge in everything. My first year in Cellular Physiology made me seriously
rethink what I was doing in graduate school when I had no idea what a tyrosine kinase was and
yet he could recite every signaling pathway and their constituents by memory. Thanks to Dr.
Catherine Elstad for being in her usual, but not official, position as “counselor” and advising me
on the direction and approaches I should take for my future and just making me laugh.
I would like to thank the Pharmacology/Toxicology program: all of the professors who
have taught me, the staff for their never-ending help and the amazing students. I have made
some great friends while being at WSU! I would like to especially thank Manpreet Chahal who
always helped me with techniques, but more importantly, is a dear friend. I sincerely appreciate
my surrogate laboratory, Team Dioxin (The Lawrence Lab). Dr. Kristen Mitchell who was my
graduate student role model and has always had wise words of advice. Dr. Beth Vorderstrasse
and her sanity! She is a remarkable person who doesn’t have an unkind bone in her body and
will achieve great things in her life. I would also like to thank her family – Eric, Annika and
Katrina. They have always welcomed me into their home for parties and holidays. Additionally,
her beautiful girls have been rays of sunshine and I have enjoyed seeing them from in utero to
being born to becoming adorable, intelligent children. Thanks to Dr. Sabine Teske for our
serious studying, shopping and movie times. Jennifer Cundiff and her family have been
wonderful friends. And, thank you to Dr. B. Paige Lawrence for allowing me to be part of the
group!
I don’t know if I can succinctly express my gratitude to my best friend, Haley Neff-
LaFord – I could write pages about what a great friend she is. Haley is a rock and like a sister to
vi
me! We have been through so much together and it is amazing we have been able to share our
undergraduate years, our DOE internships, and graduate school years, and now, the day we
receive our PhDs! I believe it is through times of adversity people reveal their true selves.
Haley has definitely had her fair share of hardship over the past half decade and she has shone
brilliantly. She has forged forward academically, emotionally and spiritually while not
neglecting to care for the people around her. Additionally, she has unquestionably continued to
give more than needed to our friendship. Although I am deeply saddened that it is time for us to
pursue our careers (and lives) in different places, I know she can achieve whatever her heart is
set upon and we will always have each other to rely on and to support.
Lastly, I would sincerely like to thank those that have contributed monetarily or
materially to my project. Without donors like the Motsenbocker, Bracken or Sue Harriet Monroe
Mullen families I would not have been able to attend some of the conferences I have been very
fortunate to attend. Many thanks go to the individuals that have contributed cell lines and
antibodies. Thanks to Dr. Earnest Johnson II for his gift of the dCK gene. I am also grateful to
Drs. Charles Dumontet and Lars Petter Jordheim for the generous gift of the MESSA 10K cells.
I am appreciative to Drs. Jean-Emmanuel Kurtz and Richard Jund for the CD-containing
plasmid, as well as Drs. Oliver Press (Jurkat cells), Neal Davies (HCT116 cells), Steven Albelda
(WiDr cells) and Alnawaz Rehemtulla (polyclonal yCD antibody). And, of course, to the NIH
for the grant monies, CA097328 and CA085939.
vii
ENGINEERING ENHANCED ENZYMES FOR SUICIDE
GENE THERAPY OF CANCER
Abstract
by Candice Lynn Willmon, Ph.D. Washington State University
May 2006
Chair: Margaret E. Black
Suicide gene therapy of cancer offers a selective approach to eliminating tumor cells. A
drawback with suicide gene therapy is the low activity of the enzyme towards prodrug. In an
attempt to circumvent this, we sought to optimize three suicide enzymes for increased sensitivity
towards their respective prodrugs.
We have altered the sensitivity of herpes simplex virus-1 thymidine kinase (TK) for the
prodrugs acyclovir and ganciclovir in two site-directed mutants, A168F and A168Y. While both
mutants display increased prodrug specificity, the most noteworthy alteration from wild-type TK
is the 763-fold greater relative specificity that A168F displays towards acyclovir. In addition to
the A168 mutagenesis study, we have generated a fusion protein of TK and mouse guanylate
kinase that displays a greater than 175-fold decreased IC50 for ganciclovir in tumor cells.
Yeast cytosine deaminase (yCD) activates the anti-fungal agent, 5-fluorocytosine (5FC).
From a regio-specific random mutagenesis library, we identified three mutant enzymes (D92E,
M93L, and I98L) with enhanced 5FC activities. These three mutant were individually overlayed
onto a previously engineered thermostable mutant (triple yCD). Enhanced tumor cell killing was
demonstrated with the three 5FC-sensitizing mutants, along with two thermostable (double and
viii
triple) and two superimposed variants (triple/M93L and triple/I98L). Interestingly, one
superimposed mutant (triple/D92E) displays even greater thermostability than the
computationally designed triple yCD variant, but achieves less cell killing.
Human deoxycytidine kinase (dCK) phosphorylates anti-neoplastic agents, such as
gemcitabine. We have performed regio-specific random mutagenesis on an important region of
dCK and have identified eight mutant enzymes with the ability to sensitize bacteria to
gemcitabine at a level 35,000-fold greater than wild-type dCK. In vitro evaluation of the eight
dCK mutants shows no improved cancer cell killing compared to wild-type dCK. Current
attempts to elucidate the cause for the observed activities in the tumor cell models evaluation in a
dCK-deficient cell line.
In summary, because of their enhanced activities for their prodrugs, the mutant enzymes
we have created should provide the means to use otherwise unfeasible drugs, overcome
bottlenecks in pathways, and, overall, demonstrate increased tumor ablation and expanded
bystander effects in future suicide gene therapy applications.
ix
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS……………………………………………………...………………iii
ABSTRACT………………………………………………………………………...…………...vii
LIST OF TABLES………………………………………………………………………………..xi
LIST OF FIGURES……………………………………………………………………………...xii
CHAPTER
1. INTRODUCTION
Enzyme/prodrug combinations……………………………………………5
Enzyme improvements…………………………………………………..12
2. THE ROLE OF HERPES SIMPLEX VIRUS-1 THYMIDINE KINASE
ALANINE 168 IN SUBSTRATE SPECIFICITY
Abstract…………………………………………………………………..20
Introduction………………………………………………………………21
Materials and Methods…………………………………………………...23
Results…………………………………………………………………....26
Discussion………………………………………………………………..31
3. A GUANYLATE KINASE/HSV-1 THYMIDINE KINASE FUSION
PROTEIN ENHANCES PRODRUG MEDIATE CELL KILLING
Abstract…………………………………………………….…………….36
Results and Discussion…………………………………………………..37
x
4. IMPROVED 5-FC-MEDIATED CELL KILLING BY ENGINEERED
AND SELECTED YEAST CYTOSINE DEAMINASE MUTANTS
Abstract…………………………………………………………………..45
Introduction………………………………………………………………46
Materials and Methods…………………………………………………...49
Results…………………………………………………………………....57
Discussion………………………………………………………………..64
5. DIRECTED EVOLUTION AND SELECTION OF DEOXYCYTIDINE
KINASE MUTANTS THAT INDUCE HYPERSENSITIVITY TO
GEMCITABINE
Abstract…………………………………………………………………..69
Introduction………………………………………………………………70
Materials and Methods…………………………………………………...72
Results…………………………………………………………………....79
Discussion………………………………………………………………..87
6. SUMMARY……………………………………………………………………...92
APPENDICES
Appendix A – Construction of a positive selection for human
deoxycytidine kinase in E. coli…………………………………………………..97
Appendix B – Screening and characterization of an error-prone
Polymerase chain reaction deoxycytidine kinase library……………………….100
LIST OF REFERENCES……………………………………………………………………….103
xi
LIST OF TABLES
1.1 Enzyme/prodrug combinations for suicide gene therapy………………………………….5
1.2 Key differences between the E. coli (bCD) and S. cerevisiae (yCD)
cytosine deaminases……………………………………………………………………...10
2.1 Kinetic parameters of mutant and wild-type thymidine kinases…………………………27
3.1 Kinetic parameters of purified enzymes…………………………………………………43
4.1 Mutant yCD enzymes identified…………………………………………………………58
4.2 Kinetic parameters of wild-type and mutant yCD variants……………………………...63
5.1 Kinetic parameters of wild-type and mutant dCK enzymes……………………………..82
xii
LIST OF FIGURES
1.1 Schematic representation of suicide gene therapy………………………………………...1
1.2 Activation of ganciclovir (GCV) and acyclovir (ACV) by herpes simplex
virus thymidine kinase……………………………………..…………………………..….6
1.3 Cytosine deaminase activities……………………………………………………………..8
1.4 5FC metabolites and their fates in the pyrimidine metabolism pathway………………….9
1.5 Chemical structure of deoxycytidine (dC), gemcitabine (dFdC), and
cytosine arabinoside……………………………………………………………………...11
1.6 Salvage pathway of deoxycytidine or deoxycytidine analog metabolism
in mammalian cells………………………………………………………………...…….12
1.7 Regio-specific random mutagenesis methodology………………………………………14
1.8 Human dCK amino acid sequence (in single-letter code) spanning
residues 120-145…………………………………………………………………………19
2.1 Deducing amino acid sequence of the two HSV TK mutants, mutant 30 and SR39…….22
2.2 Mutant enzyme relative specificity differences compared to wild-type
enzyme for each prodrug………………………………………………………………...29
2.3 Superposition of HSV TK active site residues…………………………………………..30
3.1 Functional complementation assays in E. coli TS202A(DE3)…………………………..40
3.2 GCV sensitivity assays…………………………………………………………………..42
4.1 Yeast cytosine deaminase amino acid sequence (in single-letter code)
spanning residues 81-100………………………………………………………………...48
4.2 Sensitivity of yCD-expressing rat C6 constructs to 5FC………………………………...61
4.3 Thermal denaturation measurements of the wild-type and yCD variants………………..64
xiii
5.1 Human dCK amino acid sequence (in single-letter code) spanning
residues 120-145…………………………………………………………………………72
5.2 Comparison of catalytic efficiencies (kcat/Km) and the relative specificity differences of
the mutant enzymes compared to wild-type dCK ……………………………………….83
5.3 Gemcitabine sensitivity of deoxycytidine kinase (dCK)-transfected
C6 glioma and HCT116 cells…………………………………………………………….86
5.4 Human deoxycytidine kinase structure with bound deoxycytidine and ADP and
positions of residues mutated in a screen for increased sensitivity to
gemcitabine………………………………………………………………………………88
5.5 Gemcitabine sensitivity of various cancer cell lines……………………………………..90 7.1 Pathways in the pyrimidine metabolism of E. coli………………………………………98
xiv
DEDICATION
To my younger brothers, Dexter and Tim.
Life is a challenge – meet it! Life is a goal - achieve it!
1
CHAPTER ONE
Introduction
A major limitation to conventional therapies for cancer is the lack of specificity to target
tumor cells. As a treatment for cancer, chemotherapy is a systemic approach and can often cause
damage to normal tissue. In general these drugs affect all dividing cells, including normal cells
and especially those with a high proliferation rate, such as bone marrow and gastrointestinal tract
cells. In fact, treatment failure is largely a result of the lack of specificity of chemotherapeutic
agents since the concentration of drugs successfully administered to the patient without severe
repercussions on normal tissues may be insufficient for therapeutic efficacy. Additionally,
because of the high degree of acquired resistance to current chemotherapeutic and radiation
treatments, new treatment approaches must be sought.
To enhance the selectivity of tumor treatment and to prevent the emergence of drug
resistance one approach is gene therapy. One type of gene therapy used to treat cancer is
Figure 1.1. Schematic representation of suicide gene therapy.
Cells expressing suicide gene are killed
Suicide gene
Gene delivery to target
Enzyme production
Prodrug administration
Transfected cell Untransfected cell
Bystander effect
2
suicide gene therapy, often referred to as gene-directed enzyme prodrug therapy (GDEPT).
Suicide gene therapy involves the introduction of a gene encoding an enzyme into cancer cells
followed by prodrug administration (Figure 1.1). Expression of the enzyme allows for the
conversion of the non-toxic prodrug into a highly toxic antimetabolite, whereby it results in the
elimination of cancer cells. It is this increase in drug and tissue selectivity that makes suicide
gene therapy very appealing and over the past decade the public awareness about this new
therapeutic approach has increased.
It was not until 1991 that Huber et al. first described the use of a transgene for enzyme
prodrug activation (1). Since that time, 101 clinical trials have been registered worldwide that
contain various elements of GDEPT strategies, of which 69 were active as of January 2006.
Three of these trials have advanced to phase III multicenter programs (2). However, as expected
with any new and innovative treatment, there have been several setbacks that have come along
with the successes in the general gene therapy field. In 1999, a gene therapy clinical trial raised
questions regarding the safety of experimental gene therapy treatments. Jesse Gelsinger, an 18-
year old suffering from an ornithine transcarboxylase deficiency (OTCD), which prevented his
body from properly metabolizing nitrogen, died due to an intrahepatic infusion of a modified
adenoviral vector. The first cure from gene therapy occurred in April 2000 at Hôpital Necker-
Enfants Malades in Paris. The treatment was given to children afflicted with X-linked severe
combined immunodeficiency (SCID-X1), often known as the “baby in the bubble” or “bubble
boy” syndrome. SCID-X1 is caused by a mutation in interleukin-2 receptor gamma, which is
necessary for the production of the γ chain (γc) subunit, a common component to numerous
interleukin receptors. Children with this disorder have typical B cell function, but the lack of γc,
and consequently defective interleukin signaling that prevents the proper maturation of T
3
lymphocyte and natural killer (NK) cells. This in turn leads to susceptibility to numerous
infections because the immune system is unable to identify invading agents as well as activate
and regulate other immune cells. Whereas most of the children from the Paris study today are
able to live “normally,” have normal numbers of T cells, and have responded well to several
childhood immunizations, three of the children have developed leukemia. Two of these cases
were due to chromosomal integration of the retroviral vector and subsequent activation of the
oncogene LMO-2, and, unfortunately, led to the death of one child (3). The cause of the third
child’s leukemia has yet to be determined (4). Overall, whereas these incidences have been
tragic and unfortunate, they have led to enhancing the knowledge to prevent such events in future
gene delivery systems, improving preclinical testing and monitoring of gene therapy trials, and,
importantly, stress the need for more innovative and successful gene therapy strategies.
The effectiveness of tumor-specific gene therapy relies upon several key factors. The
first aspect is that the gene must be efficiently transfected and exclusively expressed in tumor
cells (or at least to a high ratio of tumor cells to normal cells). Many of the current gene delivery
mechanisms rely on viral vectors derived from viruses, such as retroviruses or adenoviruses,
because of their inherit ability to deliver genetic material to host cells. Despite their decent
infectivity of cancer cells the potential hazards of using viral vectors, such as those described in
the two cases above, have led to an insurgence of effort to advance non-viral means of delivery
because of their decreased pathogenicity and ability to integrate into the host genome. The
current non-viral methods include episomal plasmids, artificial chromosomes, liposomes,
nanoparticles and naked DNA delivered by injection. The second key aspect of effective gene
therapy is that the level and regulation of the gene must be sufficient for clinical benefit.
However, it is highly unlikely that delivery of the suicide gene to all target cells will be achieved.
4
Therefore, a third important factor should be taken into consideration: the therapeutic effect of
the enzyme and cytotoxic agent (prodrug) combination. In fact, for suicide gene therapy
efficient prodrug activation may be the essential factor.
In choosing an appropriate enzyme/prodrug combination it is necessary for the enzyme to
be non-cytotoxic when expressed, stable and highly active against the prodrug candidate. If
possible, the reaction pathway should be unique from endogenous enzymes, in order to avoid
cytotoxic activation in normal tissues. On the other hand, using an exogenous enzyme can be a
drawback because foreign enzymes will usually elicit an immune response. While the use of
enzymes from human origin could alleviate this problem, normal tissue damage may occur due
to activation of the anti-metabolites in non-transfected tissues. In addition, there should be
increased concentrations of the active drug at the tumor site when compared to the level of
classic chemotherapeutic agents typically available to the tumor. Furthermore, the prodrug and
its activated counterpart should be highly diffusible throughout the tumor or be efficiently
transported through gap junctions or by membrane-associated transport machinery, should be
chemically stable under physiological conditions and have suitable pharmacological and
pharmacokinetic properties. Additionally, in order to generate a local bystander effect (discussed
below) the toxic agent should have a half-life that allows for diffusion to surrounding
untransfected cells. And lastly, the drug would ideally be active in both proliferative and
quiescent cells to kill a wide range of tumor cell populations.
The bystander effect, as mentioned above, was first described by Moolten (5) and refers
to the extension of cell killing effects of the active drug to untransfected neighboring cells.
Without the bystander effect the success of gene therapy would not be possible because of
inefficient gene delivery and targeting. This effect is often facilitated by the transfer of the toxic
5
drug from transfected cells to neighboring untransfected cells through gap junctions (6-8),
apoptotic vesicles (9,10) or by passive diffusion (11-13). Whereas current gene therapy
strategies rely strictly upon the transfer of the cytotoxic drug for the bystander effect, recently it
has been observed that there is a cell-mediated (immune) bystander effect involved in not only
ablating the targeted tumor (14-16), but also antitumor immunity (17) and clearing metastatic
tumors (18).
Table 1.1. Enzyme/prodrug combinations for suicide gene therapy. Adapted from information in Dachs et al. (19). Asterisks denote enzymes currently being used in clinical trials.
Enzyme/prodrug combinations
Numerous enzyme/prodrug combinations have been studied over the past decade for use
in suicide gene therapy applications (Table 1.1). In this section, three prevalent combinations
will be discussed: herpes simplex virus (HSV) thymidine kinase (TK)/ganciclovir (GCV);
cytosine deaminase (CD)/5-fluorocytosine (5FC); and human deoxycytidine kinase
(dCK)/gemcitabine (dFdC).
Thymidine kinase/ganciclovir (TK/GCV)
The TK/GCV combination is the prototypic combination for suicide gene therapy and has
been the most extensively studied and the most widely used combination to date. In the clinical
trials recorded up to July 2004, 82 out of 95 clinical trials utilized TK/GCV and it is currently the
Enzyme Prodrug Predominant toxicity
Carboxypeptidase G2 Nitrogen mustards DNA cross-linking agent
Cytochrome P450 Cyclophosphamide DNA cross-linking agent
*Cytosine deaminase 5-fluorocytosine Inhibition of thymidylate synthase
Deoxycytidine kinase Gemcitabine; AraC Inhibition of DNA synthesis
Horseradish peroxidase Indole-3-acetic acid Forms DNA/protein adducts
Nitroreductase CB1954 DNA cross-linking agent
*Thymidine kinase Ganciclovir; acyclovir DNA chain termination
6
only enzyme and prodrug being studied in advanced phase III clinical trials. Thymidine kinase is
an important enzyme in the salvage pathway of nucleosides, and is responsible for
phosphorylating deoxythymidine (dT) to its monophosphate form (dTMP). Herpes simplex virus
TK has a broad substrate specificity and this is what dictates its success in gene therapy. Unlike
human thymidine kinase, HSV TK is able to phosphorylate the purine nucleoside analog
ganciclovir (GCV), a clinically used antiviral treatment (Figure 1.2). Once phosphorylated by
TK to its monophosphate form, GCV is further phosphorylated to its di- and triphosphate forms
by endogenous cellular kinases. In its triphosphate form, GCV is incorporated into DNA and
acts as a chain terminator to prevent further DNA synthesis, thereby eliciting cell death (20). In
addition to activating GCV, TK is capable of phosphorylating other clinically relevant
nucleoside analogs, such as acyclovir (ACV). Whereas ACV is less myelosuppresive than GCV,
the poor activity wild-type TK has towards this analog precludes its use in current gene therapy
applications.
GCV GCV-MP GCV-DP GCV-TP
ACV ACV-MP ACV-DP ACV-TP
Figure 1.2. Activation of ganciclovir (GCV) and acyclovir (ACV) by herpes simplex virus thymidine kinase (tk). GCV and ACV are initially phosphorylated by tk and subsequent phosphorylations to the di- and triphosphates are performed by the endogenous cellular enzymes guanylate kinase (gmk) and nucleoside diphosphokinase (ndk). The triphosphate forms of GCV and ACV are incorporated into nascent DNA by DNA polymerase (pol) during cell division, causing DNA synthesis termination, and therefore, cell death. The killing of non-transfected cells with the TK/GCV combination through the bystander
effect is largely mediated through gap junctions. Gap junctions are intracellular channels
between cells that physiologically allow for communication between cells via chemical signals
and allow neighboring cells to share nutrients. Once GCV or ACV is activated, the charged,
gmk pol tk ndk DNA
7
phosphorylated product is unable to diffuse from the cell and requires cell-to-cell contact for
transfer of the antimetabolites. This can be problematic because many tumor cells lack or
express low levels of gap junction proteins (21,22). Despite these issues, it was observed that a
mixed population of only 10% of TK expressing cells was killed at a concentration of GCV that
was not cytotoxic to non-TK expressing cells (10).
Cytosine deaminase/5-fluorocytosine (CD/5FC)
After TK, cytosine deaminase is the next most widely studied gene. This gene is found in many
fungi and bacteria, but not in higher eukaryotes. The CD enzyme encoded by this gene catalyzes
the deamination of cytosine to uracil (Figure 1.3). Uracil is then converted to uridine and
phosphorylated or further converted to thymidylate and ultimately utilized for DNA synthesis.
CD also selectively converts the antifungal agent, 5FC into its active metabolite 5-fluorouracil
(5FU) (Figure 1.3). 5FU is one of the most widely used chemotherapeutic agents and has been
clinically used for decades. It is active against many different tumor types, but is most
commonly used for colon, breast, stomach and pancreatic cancers (23). 5FU is metabolized by
both normal and tumor cells into 5-fluoro-2’-deoxyuridine 5’-monophosphate (5FdUMP) and 5-
fluorouridine triphosphate (FUTP) (Figure 1.4). These metabolites can cause cell death through
two different mechanisms of action (MOA). In the first MOA, 5FdUMP and N5-10-methylene-
tetrahydrofolate, a folate cofactor, bind to thymidylate synthase (TS) forming a covalently bound
ternary complex and inhibits the formation of thymidylate (Figure 1.4). Thymidylate is the
precursor for thymidine triphosphate, an essential building block for DNA synthesis. Inhibition
of TS by 5FdUMP is the primary mechanism responsible for cell death (24-27). The second
MOA is mistaken incorporation of FUTP, instead of UTP, during transcription and leads to
interference in RNA processing and protein synthesis (28).
8
Figure 1.3. A. The deamination of cytosine to uracil catalyzed by cytosine deaminase. B. The deamination of 5-fluorocytosine and 5-fluorouracil catalyzed by cytosine deaminase.
In contrast to the TK/GCV combination, the CD/5FC bystander effect is not dependent
upon gap junctions (12) and in comparison to other enzyme/prodrug combinations it has one of
the strongest bystander effects (12,29-32). This is due to 5FU being a small, uncharged molecule
capable of non-facilitated diffusion diffuse through cellular membranes. Despite this advantage,
bacteria in the intestinal flora can utilize 5FC leading to normal cell killing and unwanted side
effects. Therefore, the dose of 5FC administered must be low if gastrointestinal tract toxicity is
to be limited. Another advantage of this combination is that 5FU has radiosensitizing properties
(33,34). Since it is unlikely that treatment with gene therapy would be the only course of action
in patients, radiosensitizing effects are important. In fact, in vivo results show a significant
bystander effect (33,35) at clinically relevant 5FC doses and radiation regimens (33,36).
A.
B.
9
Figure 1.4. 5FC metabolites and their fates in the pyrimidine metabolism pathway. DPD: dihydropyrimidine dehydrogenase, TK: thymidine kinase, UPRT: uracil phosphoribosyltransferase, OMP: orotidine-5’-monophosphate, CH2FH4: N5,N10-methylenetetrahydrofolate, 5FUH2: 5,6-dihydro-5-fluorouracil, 5FUPA: 5-α-fluoro-β-ureidopropionic acid, 5-α-fluoro-β-alanine.
As mentioned above, cytosine deaminase is present in both yeast and bacteria. While
these two enzymes have the same primary function, they have evolved independently, which is
reflected in distinct characteristics at several levels: primary sequence, quaternary structure,
metal requirements, relative substrate specificities and affinities, and thermostability (Table 1.2).
Even though the yeast CD (yCD) displays better kinetic properties towards 5FC and improved
efficacy for treating tumors in mice (37), the bacterial cytosine deaminase (bCD) has greater
thermostability making it the enzyme of choice in current gene therapy applications.
5FU
OMP
5FUH2
5dUrd 5FUrd
5FUPA 5FUMP
CO2
+ NH3
+ 5FBAL 5FdUMP 5FdUDP 5FUDP
urea 5FdUTP 5FUTP
Further
utilization
Elimination
De novo pathway
Salvage pathway
Catabolism
Anabolism
5FC
CH2FH4
Thymidylate
synthase
No TMP
DNA synthesis
inhibition
DNA and RNA
incorporation
Cytosine
deaminase
(cod A)
OMP
decarboxylase
(pyr F)
DPD
UPRT
TK
10
Table 1.2. Key differences between the E. coli (bCD) and S. cerevisiae (yCD) cytosine deaminases.
The CD gene from Escherichia coli has been utilized in numerous in vitro studies and has
been found to enhance mammalian cell sensitivity (IC50) to 5FC up to 2000-fold (38). bCD has
also proven to be very effective in vivo (38-43), and as of January 2006 is being used in eight
clinical trials alone or in combination with TK (2).
Deoxycytidine kinase/gemcitabine (dCK/dFdC)
Human deoxycytidine kinase (dCK) is an important enzyme in the salvage pathway of
deoxyribonucleosides. This enzyme displays broad substrate specificity and is capable of
phosphorylating deoxycytidine analogs in addition to deoxycytidine. Deoxycytidine kinase is
responsible for the initial phosphorylation, the rate-limiting step, of anti-neoplastic agents such
as gemcitabine (dFdC or 2', 2'-difluorodeoxycytidine) and cytosine arabinoside (araC or 1-β-D-
arabinofuranosylcytidine), as shown in Figures 1.5 and 1.6. Further phosphorylation of these
drugs is carried out by other endogenous enzymes. The final phosphorylated products of both
drugs are incorporated into replicating DNA and terminate DNA chain synthesis (44,45). An
additional aspect of gemcitabine’s mechanism of action is through the inhibition of
ribonucleotide reductase (RNR) (46,47). RNR is the key enzyme responsible for the conversion
of ribonucleotide diphosphates to their deoxyribonucleotide diphosphate counterparts. In the
absence of dNTP substrates for DNA polymerase, DNA synthesis is stalled.
Escherichia coli Saccharomyces cerevisiae
Monomer Size 48 kD 17.2 kDFold TIM Barrel Amidohydrolase
Active Enzyme Hexamer HomodimerThermostability Stable Labile
kcat (Cyt, 5-FC) 185 s-1, 16 s-1 84 s-1, 68 s-1
Km (Cyt, 5-FC) 2.2 mM, 18 mM 3.9 mM, 0.8 mM
11
Figure 1.5. Chemical structure of deoxycytidine (dC), gemcitabine (dFdC) and cytosine arabinoside (araC).
Although dCK is present in different tissues, many tumors have low levels of dCK,
indicating that gene therapy with dCK may be an appropriate means to sensitize tumors to
cytotoxic compounds. Whereas the concept to use dCK in gene therapy for treatment of solid
tumors with dFdC or araC has been discussed for a number of years, supporting evidence for the
use of dCK as a suicide gene has only been reported recently (44,48-53). Gene therapy studies
using dCK resulted in 50-400 fold reduction in IC50 in vitro in dCK-deficient CHO cells with
dFdC or araC as compared to untransfected cells (51). Manome et al. (44) observed significant
anti-tumor effects in a 9L glioma model in rats and demonstrated a statistically significant
survival time compared to that of non-dCK transfected tumor-bearing animals.
Similar to TK, Manome et al. (44) demonstrated the dependence of the bystander effect
with dCK/araC on cell-cell contact, due to the transfer of antimetabolites through gap junctions.
Furthermore, Sanda et al. (50) found that the presence of 25% dCK-expressing cells caused a
significant bystander effect (~80-90%) in mixed culture. Unlike TK and cytosine deaminase,
dCK is of human origin. Whereas the application of TK/GCV or CD/5FC is limited due to rapid
O
R2OH
R1H
HH
HO
N
N
NH2
O
R1 R2
dC H H
dFdC F F
araC OH H
12
clearing by the immune system in sequential administrations, dCK provides an opportunity to
perform multiple administrations because of its human origin, which may be necessary for
complete tumor ablation. This is a substantial advantage of the dCK/dFdC system over the
TK/GCV or CD/5FC systems.
dck ucmk ndk pol dC dCMP dCDP dCTP DNA dFdC dFdC-MP dFdC-DP dFdC-TP (-) araC araC-MP araC-DP araC-TP
(-) dU
Figure 1.6. Salvage pathway of deoxycytidine or deoxycytidine analog metabolism in mammalian cells. The (-) symbol represents inhibition. dC, deoxycytidine; dFdC, gemcitabine; araC, cytosine arabinoside; dU, deoxyuridine; dck, deoxycytidine kinase; cdd, deoxycytidine deaminase; ucmk, uridine/cytidine monophosphate kinase; ndk, nucleoside diphosphokinase; pol, DNA polymerase; rnr, ribonucleotide reductase.
Enzyme improvements
Despite the success of several of the current suicide gene therapy applications in vitro and
in vivo, clinical trials have not yet proven completely fruitful. One limitation of suicide gene
therapy is the inefficient gene delivery systems, although it is unlikely that a perfect, tumor-
specific system will ever be realized. Another limitation is that the enzyme/prodrug
combinations used are not necessarily enzymatically efficient with respect to prodrug activation,
and it can be difficult to achieve clinically relevant doses without inducing toxicities. In order to
overcome these two limitations, several approaches have been reported including the use of
multiple gene copies to yield high expression (54), synthesizing novel prodrugs for gene therapy
(55), modulation of nucleoside metabolizing pathways using drugs (56), application of dual
cdd
NDP dNDP rnr
13
suicide gene/dual prodrug approaches (57) and the use of mutant suicide genes with improved
kinetic parameters towards clinically relevant prodrugs (58,59).
This next section is a brief summary of the approaches which have been taken for
improving the enzyme/prodrug combinations discussed above. In addition, the motives for the
work that has been performed for my dissertation project will be discussed. This work centers
on the concept of improving these enzymes’ ability to utilize their prodrugs for use in suicide
gene therapy applications. To begin the discussion of enzyme optimization, an overview of
random mutagenesis schemes will be given.
Mutagenesis schemes
Regio-specific random mutagenesis
Regio-specific random mutagenesis allows for the introduction of multiple, random
mutations within a target region. Often residues in the active site of the enzyme or involved in
important functions of catalysis are targeted. Regio-specific random mutagenesis in conjunction
with a selection system, such as genetic complementation, is a powerful method to identify
functional mutants. By performing this type of mutagenesis, it is possible to generate large
numbers of functional mutant enzymes, to evaluate the diversity of the individual residues that
can occur throughout the target region and to identify mutants that display desired
characteristics. Regio-specific random mutagenesis is especially applicable to structure/function
studies of enzymes, and can lead to the design of enzymes with altered or novel functions (60).
Regio-specific random mutagenesis introduces mutations into a short segment of DNA.
To perform regio-specific mutagenesis the first step is to design overlapping oligonucleotides
that contain either completely random nucleotides at specified positions to introduce all possible
residues simultaneously or a limited percent randomness to skew the introduction of mutations
14
R1
3’
RI
R1
R1
5’ NNNNNNN 3’
3’
R2
5’ NNNNNNN 3’
R2
PCR Amplification (Optional)
5’ NNNNNNN
3’
R2
NNNNNNN
R2
5’ NNNNNNN
3’ NNNNNNN
One or multiple oligonucleotides contain random sequence
Anneal overlapping oligonucleotides
Extend hybrid DNA with DNA polymerase
Cut product with restriction enzymes (R1 and R2) to create overhangs
Dummy Vector
R1
R2
Random mutants
Cut dummy vector with restriction enzymes to create overhangs
Ligate digested random fragments and dummy vector, transform into E. coli and screen clones
5’
5’
5’
3’
5’
3’
Figure 1.7. Regio-specific random mutagenesis methodology.
NNNNNNN
15
toward single changes (61-64). The oligonucleotides are randomized and contain restriction
endonuclease sites at each end. Using the oligonucleotides, a library of random sequences is
constructed by first annealing the oligonucleotides followed by extension of the hybrid DNA
with DNA polymerase (Figure 1.7). The extended DNA product is then cut at each end with the
corresponding restriction enzymes and cloned into a similarly restricted, dummy vector. The use
of a non-functional or dummy vector as the recipient of randomized fragments eliminates the
possibility of contamination by the wild-type enzyme and will prevent false-positives from
emerging in the selection system. The dummy vector is an expression vector carrying the gene
in which the section of the gene to be replaced contains a mutation that renders the gene non-
functional. After insertion of the randomized fragments, the recombinant molecules are
transformed into E. coli followed by genetic complementation or another selection system to
identify mutants with the desired properties.
Error-prone PCR and DNA shuffling
In contrast to regio-specific random mutagenesis, error-prone polymerase chain reaction
(epPCR) with DNA shuffling is a technique that can be utilized if little or nothing is known
about the active site of an enzyme. Unlike regio-specific random mutagenesis that targets
specific amino acid (a.a.) codons, epPCR is capable of introducing mutations into the entire
gene. Random mutations are introduced using a low fidelity polymerase chain reaction, which
can be achieved by increasing the concentration of MgCl2, adding MnCl2, skewing the levels of
the dNTP pool, increasing the concentration of polymerase or increasing the extension time (65).
To provide a degree of recombination among the mutations in the PCR pool, DNA shuffling is
used to randomly fragment the PCR products followed by reassembly of the gene (66). DNA
shuffling involves fragmenting the PCR product with DNAse I. The fragments are then
16
subjected to primerless PCR, where the fragments prime to each other and ultimately lead to the
generation of the recombined full-length gene. Several rounds of shuffling can be completed to
obtain a higher degree of diversity in the gene of interest. Finally, the pool of mutants is cloned
into the expression vector and used to transform E. coli. The desired phenotype is identified
using the selection system, as done with regio-specific random mutagenesis
Improving thymidine kinase
The sensitivity of wild-type HSV TK for GCV and ACV in cells is not optimal and this is
a limitation with this combination for use in gene therapy because high doses of these nucleoside
analogs, especially GCV, can be myelosuppressive. In order to increase the level of nucleoside
analog activation several investigators have engineered HSV TK mutants. Initial mutagenesis
for an improved HSV TK for gene therapy was first described in 1993 when Munir et al. (67)
altered the substrate preference for 3’-azido-3’deoxythymidine (AZT) and also enhanced this
enzyme’s thermostability. Soon after, the same technique was used to create a large library of
variants for increased sensitivity to either GCV or ACV (68). This particular library, named the
LIF-ALL library, targeted two conserved motifs that had previously been identified to be
involved in substrate binding (69). The codons targeted correspond to the residues leucine (L),
isoleucine (I), and phenylalanine (F) at positions 159-161 and residues alanine (A), leucine (L),
and leucine (L) at positions 168-170, hence the designation of LIF-ALL. From this library a
total of 26 mutants displayed enhanced sensitivity to GCV and 54 to ACV. Six of these mutants
demonstrated increased sensitivity to both GCV and ACV in E. coli (58). In order to further
evaluate the diversity of the LIF-ALL library, a second semi-random library was created using
the sequences of the best mutants from the first library. From this seven mutants were identified
to have the desired GCV or ACV sensitivity. Through extensive characterization of the two
17
libraries of TK mutants, two mutants containing multiple amino acid substitutions were of
particular interest due to their increased sensitivity to prodrugs. These mutants, mutant 30 and
SR39, were found to have amino acid substitutions at five of the six targeted codons. SR39,
when evaluated in a rat C6 glioma xenograft tumor model, was able to significantly prevent
tumor growth, compared to wild-type upon treatment of both ACV and GCV. Additionally,
mutant 30 and SR39 have shown improved prodrug-mediated killing by retroviral (70) and
adenoviral (71) transfection in various human tumor cells.
Molecular modeling of these multiple amino acid substituted variants led us to suggest
that substitutions at position 168, one of the sites mutated in both mutant 30 and SR39, may be
crucial for the observed substrate alterations found in mutant 30 and SR39 (59). Results from a
site-directed mutagenesis study to elucidate whether A168 is responsible for the increased
sensitivity seen in mutant 30 and SR39 is described in Chapter 2.
Even with these dramatic increases in the sensitivity of TK achieved from the
mutagenesis libraries, evidence suggests that once a prodrug monophosphate is formed, a
bottleneck occurs, leading to accumulation of the ineffective intermediate product (72-74). In
order to overcome this a fusion gene of HSV TK and mouse guanylate kinase (mgmk), the
second gene in the activation of GCV and ACV (Figure 1.2), was created. This novel fusion
enzyme displays favorable kinetics and nearly 175-fold increased sensitivity to GCV in tumor
cells and is discussed in Chapter 3.
Improving cytosine deaminase
As with TK, the activation of 5FC by cytosine deaminase is not optimal. While yCD
displays more favorable kinetics for this substrate and significantly enhanced tumor cell killing
in mice (37), bCD has greater thermostability, the reason it has been the preferred enzyme in
18
current gene therapy applications. Several mutagenesis studies to date have been completed by
our laboratory to increase the sensitivity of bCD for 5FC (75-77), whereas none have specifically
targeted the sensitivity of yCD. The sensitive bCD mutants created in these studies display 3 to
33-fold decreased IC50s in tumor cell lines (75,78). Several studies have fused the yCD gene or
bCD gene with TK (57,79-86) or yeast uracil phosphoribosyltransferase (87-89) and have
conferred increased tumor cell killing upon the treatment with GCV and/or 5FC. However,
much of the focus of current mutagenesis and fusion studies has been on bCD and little work has
been attempted toward improving yCD activity. Therefore, in order to create mutant yCD
enzymes with enhanced 5FC activity a regio-specific random mutagenesis library targeting the
nucleoside-binding site was created and the results from this work are presented in Chapter 4.
Recently, a computational design study aimed at improving the thermostability of yCD
for the use in gene therapy applications was performed (90). One resulting mutant enzyme,
coined the triple mutant, has A23L, V108I, and I140L substitutions and was found to have over a
30-fold increased enzymatic half-life at physiological temperatures with a preservation of kcat/Km
for cytosine. Furthermore, we have recently observed that this mutant has an increased
sensitivity to 5FC in bacteria and in tumor cell lines and is further described in Chapter 4 in
conjunction with the regio-specific random mutagenesis study of yCD.
Improvement of deoxycytidine kinase
Obviously any of the enzymes used in suicide gene therapy have not naturally evolved to
utilize their prodrugs as substrates, and just as with CD and TK, dCK displays poor kinetics
towards gemcitabine. However, unlike cytosine deaminase and thymidine kinase, relatively little
mutagenesis to improve dCK for gene therapy has been done. Specifically, only one mutant
enzyme (A100V/R104M/D133A) has been created based on molecular modeling studies to
19
improve the catalytic efficiency of dCK for any prodrug (53) and it displays only a 4-fold
increase in kcat/Km compared to wild-type dCK for gemcitabine. Moreover, if you look at the
relative specificity of this enzyme, a measurement of the preference for this enzyme to utilize the
prodrug in a milieu of nucleosides within the cell, there is no increase, but rather there is a
decrease of nearly 30%. This suggests the mutant enzyme would not favorably phosphorylate
gemcitabine in the cells and would preferentially use deoxycytidine, although in vitro studies in
tumor cells have not been performed to establish this.
Figure 1.8. Human dCK amino acid sequence (in single-letter code) spanning residues 120-145. The amino acid region targeted for regio-specific random mutagenesis is highlighted in blue. A loop structure found in this region has been boxed. Starred residues interact with substrate (53).
As previously performed in our laboratory with TK (58,69,91), regio-specific random
mutagenesis to improve gemcitabine activation was used to introduce substitutions in the 124-
141 a.a. region of dCK (Figure 1.8) and is discussed in Chapter 5. We believe that creation of
super-sensitive dCK enzymes by regio-specific mutagenesis could be as successful and
important for cancer research as the TK mutagenesis studies. Importantly, the creation of
improved dCK variants for gene therapeutic applications will (i) allow the use of low prodrug
doses, (ii) reduce toxic side effects, and (iii) enhance the bystander effect and tumor ablation.
Additionally, because dCK is of human origin, multiple administrations may be possible.
Finally, this research is significant because it will provide an exciting, novel alternative for gene
therapy applications for a wide variety of cancers.
E K P V L F F E R S V Y S D R Y I F A S N L Y E S E 124 141
20
CHAPTER TWO
The Role of Herpes Simplex Virus-1 Thymidine Kinase
Alanine 168 in Substrate Specificity1
ABSTRACT
Herpes simplex virus type 1 (HSV) thymidine kinase (TK) has been widely used in
suicide gene therapy for the treatment of cancer due to its broad substrate specificity and the
inability of the endogenous human TK to phosphorylate guanosine analogs, such as ganciclovir
(GCV) and acyclovir (ACV). The basis of suicide gene therapy is the introduction of a gene that
encodes a prodrug-activating enzyme into tumor cells. After administration, the prodrug is
selectively converted to a toxic drug by the suicide gene product thereby bringing about the
eradication of the cancer cells. A drawback with this therapy is the low activity of the enzyme
towards the prodrugs, requiring high doses of prodrug that result in adverse side effects. Earlier
studies revealed two HSV TK variants derived by random mutagenesis with enhanced activities
towards ACV and GCV in vivo. While these mutants contain multiple amino acid substitutions,
molecular modeling suggests that the alanine 168 (A168) substitutions to phenylalanine (SR39)
or tyrosine (mutant 30) may be responsible for the observed increase in prodrug sensitivity. To
evaluate this, site-directed mutagenesis was used to individually substitute A168 with
phenylalanine or tyrosine to reflect the mutations found in mutant SR39 and 30, respectively.
While both mutants display increased prodrug specificity, the most noteworthy alteration is the
763-fold greater relative specificity A168F displays towards ACV, compared to wild-type TK,
making the use of ACV feasible in gene therapy applications.
1 Data in this chapter were submitted to Protein Science in the following manuscript: Willmon, C.L., Sussman, D., and Black, M.E. (2006) Role of herpes simplex virus-1 thymidine kinase alanine 168 in prodrug activation.
21
INTRODUCTION
Thymidine kinase (TK), a key enzyme in nucleotide salvage metabolism, catalyzes the
phosphorylation of thymidine (dT) to thymidylate (dTMP). While human TK is a strict
thymidine kinase, the TK from herpes simplex virus 1 (HSV) TK can phosphorylate dT, dTMP,
deoxycytidine, and various pyrimidine and guanosine analogs (92,93). Guanosine analogs, such
as ganciclovir (GCV) and acyclovir (ACV) are selectively phosphorylated by TK. The initial
phosphorylation of the analogs is performed by TK while additional phosphorylations are carried
out by endogenous enzymes. In their triphosphate form, the guanosine analogs are incorporated
into DNA and act as chain terminators to prevent further DNA synthesis, thereby eliciting cell
death (20).
The ability of TK to phosphorylate nucleoside analogs is the basis of treatment for
herpetic infections and for its use in gene therapy. TK has been widely investigated for suicide
gene therapy. The principle of suicide gene therapy is to deliver a gene encoding a prodrug-
activating enzyme to tumor cells. When the cells are subsequently treated with the non-toxic
prodrug, the drug is converted into its toxic form only in the cells containing the suicide gene,
thereby ablating the tumor cells. The TK is being used in combination with GCV and has been
shown to result in complete tumor regression in a xenograft mouse model of glioma (92).
Although the TK/GCV combination is currently being used in many clinical trials for a wide
variety of cancers (94-99), the gene and its delivery to tumor cells have not been optimized.
With respect to the suicide enzyme, wild-type is very inefficient at binding GCV (Km = 45 µM)
(92), which limits its therapeutic efficacy. Additionally, although ACV is less toxic than GCV,
the poor affinity which TK has for ACV (Km = 471 µM) (59) precludes the use of ACV in
current gene therapy studies.
22
Figure 2.1. Deduced amino acid sequence of the two TK mutants, mutant 30 and SR39. The top line shows the amino acid sequence of the residues 159-174 of wild-type TK. Two highly conserved tripeptide motifs are boxed and denoted as Sites 3 and 4. The red-colored amino acids in bold in the top line denote codons that were targeted for the creation for the randomized libraries and subsequent mutations are shown in the following lines (58,60).
In order to improve the effectiveness of suicide gene therapy, protein engineering was
used to optimize TK for increased sensitivity to GCV and/or ACV. Our previous studies used
random sequence mutagenesis to target six amino acid residues that neighbor two highly
conserved tripeptide motifs shown to be involved in substrate binding (Figure 2.1) (58,60).
Several variants conferred enhanced prodrug sensitivity in rat C6 glioma transfected cells. In a
xenograft tumor model, the most promising two mutants, mutant 30 and
SR39, displayed impaired tumor growth at doses of ACV and GCV that did not impact wild-type
TK transfected tumors (59). Molecular modeling of these multiple amino acid substituted
variants led us to suggest that substitutions at position 168 may be crucial for the observed
substrate alterations found in mutant 30 and SR39 (59). To examine the role of A168 in substrate
specificity, we created individual substitutions to reflect mutations at A168 found in mutant 30
or SR39. These mutant and wild-type TK enzymes were expressed and purified to near
homogeneity from E. coli, and characterized for their kinetic properties. The substrate binding
and catalytic efficiency information generated from theses studies were used to correlate
functional enzyme data and modeled structures with three different substrates. Our results reveal
Mutant 30 I L A Y F SR39 I F L F M Residue number 159 160 161
A L T L I F D R H P I A A L L C Y P A 168 169 170
Site 3 Site 4
23
that the phenylalanine (F) and tyrosine (Y) substitutions at the A168 are mainly accommodated
by the side-chain rearrangements that maintain interactions between TK and the nucleoside
analogs. More importantly, the A168F mutant has a dramatically increased specificity towards
ACV compared to wild-type TK, now making the use of ACV relevant to suicide gene therapy
applications.
MATERIALS AND METHODS
Materials
The bacterial expression vector, pET23d, was purchased from Novagen (Madison, WI,
USA). Oligonucleotides used for site-directed mutagenesis and DNA sequencing were
purchased from Operon Technologies (San Pablo, CA). Restriction endonucleases used for
screening mutants were purchased from New England Biolabs (Beverly, MA). Activated CH-
Sepharose 4B and 3’-aminothymidine used to make the thymidine affinity column were
purchased from Sigma (St. Louis, MO). Tritiated thymidine ([methyl-3H] thymidine, specific
activity, 90 Ci/mmol) was purchased from Amersham (Arlington Heights, IL); [8-3H]-acyclovir
(specific activities, 15 and 15.3 Ci/mmol) and [8-3H]-ganciclovir (specific activities, 14.7 and
19.1 Ci/mmol) were purchased from Moravek Biochemicals (Brea, CA). All other reagents were
purchased from Sigma (St. Louis, MI) unless otherwise denoted.
Mutagenesis
The oligonucleotide used to construct the A168F mutant is: 5’
GCAAGGAGGAAGGCGATGGGGTGGCG 3’. Oligonucleotides used to construct the A168Y
mutant are: 5’ CTTCGACCGCCACCCCATCGCCTACCTCCTGTGC 3’ and 5’
GGAGGTAGGCGATGGGGTGGCGG 3’. The single-stranded DNA template for site-directed
24
mutagenesis was isolated from pET23d:HSVTK (CJ236) (58) as described in Black and Hruby
(100). The single-stranded DNA was then used as a template for Kunkel-based site-directed
mutagenesis (101,102) for the A168F mutant. The A168Y mutant was created using a Stratagene
QuikChange kit (La Jolla, CA) according to the manufacturer’s directions, with dsDNA of
pET23d:HSVTK as the backbone. Transformants were screened by restriction enzyme digestion
for the loss of a BstF51 site. Sequence analysis was performed on purified mutant plasmid DNA
to confirm the mutation at the A168 position.
Initial evaluation of thymidine and prodrug activity in E. coli
To evaluate the level of activity to dT, GCV and ACV, mutant and wild-type TK
plasmids were transformed into a thymidine kinase-deficient E. coli (BL21 (DE3) tdk-) and used
to inoculate 3 mL of TK selection broth (69). After an overnight incubation, cultures were
serially diluted in 0.9% NaCl and spread onto 2 x YT containing carbenicillin (50 µg/mL) or TK
selection plates containing different concentrations of prodrug as described in Black and Loeb
(69). Colony growth was scored after 16-24 hr at 37°C.
Protein expression and purification
Expression and lysate preparation of wild-type and mutant HSV thymidine kinases were
performed as described by Black et al. (103). Enzymes were purified by affinity
chromatography to near homogeneity using a 3’-aminothymidine sepharose column as described
in Kokoris and Black (104). Enzyme concentrations were determined using a Pierce BCA
Protein Assay Kit (Rockford, IL) according to the manufacturer’s protocol, using BSA as the
protein standard.
25
Enzyme assays and kinetics
Thymidine kinase activity assays were performed using varying substrate concentrations
and enzyme concentrations. The assay used to detect the phosphorylation of the nucleosides
[methyl-3H]-thymidine, [8-3H]-GCV and [8-3H]-acyclovir is a filter binding assay previously
described by Hruby and Ball (105), except all washes were done at room temperature and the
assays were performed at 37°C. The assays were conducted in triplicate three or more times.
Data were plotted as the double reciprocal of velocity versus substrate concentration for each
substrate. The Michaelis-Menten constant (Km) was determined using the double-reciprocal
plots for each substrate. A conversion factor was determined for each [3H]-nucleoside
monophosphate by measuring the cpms for a known number of moles of each tritiated
nucleoside. The kcat values were calculated, using this conversion factor, from the double
reciprocal plots assuming one active site per monomer.
Structure refinement and molecular modeling
Coordinates for the TK structures with thymidine, GCV, and ACV bound were obtained
from the Protein Data Bank files 1KIM, 1KI2, and 2KI5 respectively (106,107). Structures were
refined using the Crystallography and NMR System (CNS) software version 1.1 (108). Each of
these structures was used to generate a model with either a phenylalanine or a tyrosine in
position 168. All models (mutant and wild-type) were first subjected to 200 steps of gradient
minimization, and then 1000 steps of torsion angle dynamics with slow-cooled annealing were
performed from 1250 to 298 K with 25 K temperature drops per cycle. Molecular modeling and
model analysis was performed using the computer graphics program XtalView (109). Raster3D
was used to generate figures (110).
26
RESULTS
Mutagenesis and evaluation of thymidine and prodrug sensitivity in E. coli
To explore the role of the A168 substitutions seen in the TK mutants, mutant 30 (58) and
SR39 (60), site-directed mutagenesis of wild-type TK was used to create two mutant TK
enzymes, A168F and A168Y. The A168 residue was substituted by a phenylalanine, as seen in
SR39, or a tyrosine, as seen in mutant 30. An initial evaluation of the thymidine and prodrug
activity of these mutants was performed in a tk-deficient E. coli, BL21 (DE3) tdk-, as outlined in
Materials and Methods (data not shown). Both A168 mutants were able to complement the tk-
deficient E. coli under selective conditions. When GCV or ACV is used in a secondary
selection, A168Y displays similar sensitivity as compared to wild-type TK expressing cells.
A168F, however, revealed an increased sensitivity to both prodrugs as monitored by a reduced
colony size relative to wild-type expressing cells.
Enzyme Kinetics.
Further biochemical characterization of the mutant and wild-type enzymes was done
using dT, GCV and ACV as substrates. First, enzymes were purified by affinity chromatography
to near homogeneity using a 3’-aminothymidine sepharose column (104). Next, purified wild-
type and mutant TK enzymes were characterized using a filter binding assay previously
described by Hruby and Ball (105). The optimal substrate concentrations and enzyme amounts
were determined for each substrate. All three enzymes (wild-type, A168F and A168Y) display
Michaelis-Menten kinetics. The kinetic parameters were determined using double-reciprocal
plots for each substrate.
A168Y
As shown in Table 2.1, the A168Y mutant displays a similar Km value, or affinity, for dT
27
compared to wild-type with a value of 1.18 µM. However, the kcat, or turnover number, for this
mutant is about 2% of the wild-type value with dT. Therefore, the overall efficiency (kcat/Km =
0.70 s-1/µM) of this mutant is less than 2% wild-type (kcat/Km = 38.02 s-1/µM). The A168Y
mutant has a Km value for GCV of 45.8 µM – equivalent to the wild-type Km of 45.5 µM. Once
again, this mutant displays a lower kcat for GCV than wild-type, causing the efficiency value to
be less than 11% that of wild-type. Moreover, an important point to note is that endogenous
thymidine within the cell competes with GCV for the active site. Therefore, the relative
specificity constant for the prodrug and dT should be taken into consideration. For A168Y the
relative specificity constant [kcat/Km (prodrug)/(kcat/Km (prodrug) + kcat/Km (dT))] is over 5.3-fold higher
than that for the wild-type enzyme with GCV as the prodrug substrate.
Using ACV as substrate, the A168Y mutant has a Km of 447.23 µM, which is similar to
the value obtained for the wild-type enzyme (Km = 486.11 µM). The kcat value for this mutant is
Table 2.1. Kinetic parameters of mutant and wild-type thymidine kinases.
HSV TK A168Y A168F
Thymidine
Km (µM) 1.06a 1.18 1.70
kcat (s-1) 40.30 0.83 51.00
kcat/Km (s-1/µM) 38.02 0.70 30.00
Ganciclovir
Km (µM) 45.50 45.80 8.60
kcat (s-1) 55.30 6.00 91.10
kcat/Km (s-1/µM) 1.22 0.13 10.59
Relative Specificityb 0.03 0.16 0.26
Acyclovir
Km (µM) 486.11 447.23 16.04
kcat (s-1) 0.31 7.70 x 10-2 6.31
kcat/Km (s-1/µM) 6.44 x 10-4 1.72 x 10-4 0.39
Relative Specificity 1.69 x 10-5 2.45 x 10-4 1.3 x 10-2
a Values are the averages of at least three independent measurements. Standard error of the means are within a range of ± 20%.
b Relative Specificity = kcat/Km (prodrug)/(kcat/Km (prodrug) + kcat/Km (dT)).
28
25% of the wild-type value, but the relative specificity is over 14-fold higher for ACV compared
to wild-type TK. Overall, the A168Y mutation in TK positively impacts the activity of the
enzyme for both of the prodrug substrates, such that the enzyme prefers GCV and ACV to dT
more so than wild-type TK does. However, in comparison to kinetic values previously obtained
for mutant 30 (59), the A168Y single mutation does not lead to as great an increase in activity
towards the prodrugs (Figure 2.2). Mutant 30 has a 67- and 330-fold increase in relative
specificity for GCV and ACV, respectively, while the A168Y mutant has a 5.3- and 14-fold
increase in relative specificity for GCV and ACV, respectively.
A168F
Like the A168Y mutant, the A168F mutant displays a similar affinity (Km) for the natural
substrate, thymidine, compared to wild-type (Table 2.1). The kcat value for A168F (kcat = 51.0
s-1) with dT is similar to the wild-type value at 40.3 s-1, revealing a similar efficiency to that of
wild-type. With ganciclovir as the substrate, the A168F mutant has a Km of 8.6 µM and a kcat of
91.1 s-1. The efficiency (kcat/Km) of A168F for GCV is close to 9-fold higher when compared to
wild-type values. When taking into consideration the competition at the active site between dT
and GCV in the cell, this mutant clearly prefers GCV based upon the relative specificity
observed, which is close to 14-fold higher than that of wild-type. Furthermore, with ACV as a
substrate, the A168F mutant displays a Km of 16.04 µM, about 97% decreased from wild-type,
and a kcat of 6.31 s-1, over 20-fold higher than wild-type. When compared to wild-type the
relative specificity displayed by A168F for ACV was 763-fold higher, suggesting this mutant has
an immense advantage over wild-type for ACV. Additionally, as shown in Figure 2.2, A168F
has a relative specificity constant that is over triple what was observed with SR39 for ACV
(104). The kinetic values obtained for the A168F mutant clearly show that the mutation at the
29
A168 position plays a key role in the increased sensitivity previously seen with SR39 (60).
Figure 2.2. Mutant enzyme relative specificity differences compared to wild-type enzyme for each prodrug (A. Ganciclovir; B. Acyclovir). Using the equation, [kcat/Km
(prodrug)/(kcat/Km (prodrug) + kcat/Km (dT))], the relative specificities were calculated. This equation takes into account the presence of endogenous thymidine that could compete with prodrug for the active site. The data are plotted as the fold increases from wild-type enzyme. Mutant 30 and SR39 values have been taken from Kokoris et al. (59) and (104), respectively.
Molecular dynamics
Despite the replacement of a small hydrophobic side-chain at position 168 with either an
aromatic phenylalanine residue or bulky and polar tyrosine, many of the original contacts
between protein and ligand are conserved after molecular dynamics/simulated annealing. In fact,
these dramatic mutations are mainly accommodated by rearrangements in side-chain positions
while the backbone undergoes relatively minor adjustments (Figure 2.3). These rearrangements
allow the maintenance of hydrophobic and ionic interactions between the protein and both the
A. B.
0
200
400
600
800
Fo
ld I
ncre
ase
HS
V T
K
A1
68
Y
A168F
Mu
tan
t 3
0
SR
39
Acyclovir
0
25
50
75
100
Fo
ld I
ncre
ase
HS
V T
K
A16
8Y
A168F
Mu
tant 30
SR
39
Ganciclovir
30
Figure 2.3. Superposition of TK active site residues (residue numbers 58, 128-132,168-172, 222-225). All three panels have consistent9 coloring: wild-type: yellow; A168F: blue; and A168Y: green. Models of wild-type and mutant proteins in complex with: A. Thymidine (thy); B. Ganciclovir; and C. Acyclovir.
31
nucleoside base analog and the sugar constituents of all ligands. Most notably, the hydrophobic
Y172 and methionine 128 (M128) residues sandwich the ring and define its placement within the
binding site.
This conservation of the binding site is achieved largely because the substituted residues
point directly away from the ligand and fit nicely into a hydrophobic pocket that is present in the
wild-type structure, composed of residues leucine 169 (L169), F190, and L193. The filling of
the binding pocket results in a slight rearrangement, presenting a highly complementary surface
to the tested prodrugs.
It should be noted that the final orientation of the mutated residue is somewhat dependent
on its placement in the initial modeling. After molecular dynamics, one of two orientations is
acquired; the residue either points toward the ligand (into contact with charged residues and
disturbing the binding pocket) or away from it (as described above). These two structures
converged to a similar total energy and might represent two structures that are present in vivo.
However, because the binding pocket is greatly perturbed when the mutant residue points
towards the ligand, a finding that is inconsistent with the kinetic data presented, we think it is
highly unlikely the mutation results in this conformation.
DISCUSSION
Although the use of TK in suicide gene therapy for cancer has been widely investigated,
the efficacy of this enzyme is rather limited due to low activity towards GCV and ACV and
inefficient gene delivery to tumor cells. Additionally, GCV is myelosuppressive at doses that are
required for complete tumor regression. While ACV is relatively nontoxic at high doses, it is not
feasible for gene therapy because of the poor activity displayed towards it by TK. In order to
32
improve the efficacy of suicide gene therapy, protein engineering was used to optimize TK for
increased sensitivity to the prodrugs by Black et al. (58,60). Earlier studies of two mutant
enzymes show that mutant 30 (58) and SR39 (60) confer increased GCV and ACV sensitivity in
tumor cells. Molecular modeling of these multiple amino acid substituted mutants led us to
hypothesize that the tyrosine or phenylalanine substitutions at the A168 position may cause
neighboring side chains to move and thereby enlarge the active site allowing for greater prodrug
accessibility to the active site and, in particular, lead to the increased prodrug sensitivity seen
with these two variants (59). In order to examine the role of the A168 substitutions seen in
mutant 30 and SR39, site-directed mutagenesis was used to create individual A168 mutants and
kinetic data were obtained.
The single A168Y mutation, corresponding to the substitution in mutant 30, led to the
creation of an enzyme that has similar Km values compared to wild-type values for all three
substrates (dT, GCV, and ACV). A previous report by Mercer et al. (111) describing a site 4 TK
mutant Q30-3, which contains A168Y and L169F substitutions, concluded that the site 4 half site
mutations are detrimental to substrate binding. In contrast, we found that the A168Y single
substitution was active and, compared to wild-type TK, has similar substrate specificity for all
three substrates. Other studies by Larder et al. (112) and Munir et al. (63) found that the A168
position when singly mutated with polar, uncharged amino acids maintained similar activity for
dT compared to the wild-type TK enzyme. Additionally, Larder et al. (112) found the A168T
mutation displays not only a Km value for dT similar to wild-type enzyme, but also for ACV.
While affinity towards the substrates was maintained by the A168Y mutant, the kcat values are
substantially reduced relative to wild-type. More importantly, the relative specificity of this
mutant TK for GCV and ACV over dT is over 5- and 14-fold higher, respectively, compared to
33
wild-type TK. This indicates that the A168Y mutant has an advantage over wild-type enzyme to
catalyze the phosphorylation of GCV or ACV in an environment consisting of other nucleosides,
particularly thymidine.
Comparatively, mutant 30 has poorer kinetic parameters with dT than does the A168Y
mutant (59). The catalytic efficiency mutant 30 displays for dT is 0.03% wild-type, while
A168Y has a value 1.84% that of wild-type. Like A168Y, mutant 30 preferentially utilizes the
prodrugs in an environment consisting of other nucleosides, as evidenced by the relative
specificities and IC50 values that mutant 30 displays (59). This is due to the lower efficiency that
mutant 30 has for the natural substrate. Mutant 30 has a 67- and 330-fold increase in relative
specificity to GCV and ACV, respectively, compared to wild-type TK values. While mutant 30
clearly displays a much greater preference for GCV and ACV than the A168Y mutant, the A168
substitution in mutant 30 is important for its activity.
Unlike A168Y, the A168F mutant displays major changes in prodrug substrate binding
(ACV and GCV), but not dT binding. Furthermore, the turnover number of the A168F mutant
was favorably, albeit marginally, impacted; the observed kcat of the A168F mutant is nearly
doubled that of wild-type. Compared to wild-type TK, the relative specificity A168F displays
towards GCV is over 8-fold higher, while the relative specificity SR39 displays is 85-fold higher
for GCV. Like mutant 30, SR39 displays poor kinetics for dT, with a catalytic efficiency 0.53%
of wild-type for dT (104), while A168F has a catalytic efficiency about 79% of wild-type TK.
The low efficiency of SR39 towards dT presumably results in less competition at the active site
between dT and GCV than would be seen with the A168F mutant and is likely to be responsible
for the increased relative specificity that SR39 displays towards GCV compared to A168F.
With ACV as substrate the kcat value for the A168F mutant is over 20-fold higher than
34
that of wild-type enzyme. This reflects a 3.3-fold enhancement in the relative specificity SR39
displays toward ACV (104). More significantly, the relative specificity of A168F for ACV,
compared to wild-type, is over 763-fold higher.
Despite the lack of efficient gene transfer and the low affinity that TK displays towards
GCV and ACV, it has been shown that complete tumor regression can occur when only a small
percentage of tumor cells are transfected with the suicide gene (59,113). This phenomenon,
known as the bystander effect, is important for complete tumor ablation in suicide gene therapy.
The bystander effect, first described by Moolten (5), refers to the extension of cell killing effects
of the active drug to non-transfected neighboring tumor cells. This effect is often due to the
transfer of the toxic drug from transfected cells to neighboring non-transfected cells through gap
junctions; (6-8), apoptotic vesicles (9,10) or by passive diffusion (11-13).
Previously reported in vitro studies demonstrated a 180-fold decrease in IC50 value for
ACV for rat C6 glioma cells stably transfected with SR39 compared to wild-type TK transfected
cells (60). Furthermore, in a xenograft tumor model SR39 was capable of causing significant
growth restriction with low doses of ACV, while wild-type TK-expressing tumors were not
affected (60). While those studies highlight the ability of mutant TKs to enhance tumor cell
killing, both mutant 30 and SR39 failed to elicit a strong bystander effect with ACV (25 mg/kg)
in a mouse tumor model with 20% of tumor cells expressing the mutant TK (unpublished
results). This contrasts with the significant bystander effect observed with both mutants in
similar xenograft studies with GCV at 5 mg/kg (59). Because A168F displays increased
preference for ACV (763-fold over wild-type TK and 3.3-fold over SR39) a greater bystander
effect should be observed with A168F than with the wild-type enzyme, mutant 30 or SR39
because of the elevated amount of phosphorylated drug produced. Therefore, we expect that the
35
A168F TK mutant will result in complete tumor ablation at very low prodrug doses, even if only
a small percentage of the tumor cells are transfected.
In conclusion, the two single amino acid substitutions A168 show that the substitutions at
position 168 corresponding to ones found in mutant 30 and SR39 play a key role in the increased
sensitivity towards GCV and ACV seen in these mutants. With A168Y, the Km for all substrates
is unchanged from the wild-type enzyme, whereas the turnover number (kcat) is impaired most
dramatically for dT (2% wild-type TK). While the A168F Km value for dT is similar to wild-
type TK, significant improvements have occurred with the ability of A168F to bind the prodrugs.
Compared to wild-type TK, similar or improved kcat values for A168F were determined with the
greatest improvement found when ACV is the substrate (20.4-fold). Molecular modeling
suggests that an enlargement of the active site induced by the introduction of large aromatic
residues may be responsible for the improved affinity of the TK mutants for GCV and ACV.
The additional mutations in the previously characterized TK variants (mutant 30 and SR39)
might serve to change the shape of the binding pocket even further. While further in vitro cell
culture studies and in vivo mouse tumor models are ongoing, we have previously shown good
correlation between the altered substrate specificity of mutant TK enzymes and how they
perform in cell culture and tumor models (58,59,114). Accordingly, the A168F mutant displays
kinetic parameters that make it a strong candidate for use in gene therapy studies utilizing the
relatively non-toxic ACV, rather than the predominately used and myelosuppressive GCV.
36
CHAPTER THREE
A Guanylate Kinase/HSV-1 Thymidine Kinase Fusion Protein
Enhances Prodrug Mediated Cell Killing1
ABSTRACT
Herpes simplex virus thymidine kinase (HSV TK) with the guanosine analog ganciclovir
(GCV) is currently the most widely used suicide gene/prodrug system for gene therapy of cancer.
Despite the broad application of the HSV TK/GCV approach, phosphorylation of GCV to its
active state is inefficient such that high, myelosuppressive doses of GCV are needed to observe
an antitumor effect. One strategy used to overcome the poor substrate specificity of HSV TK
towards GCV (Km = 45 µM) has been to create novel forms of TK with altered substrate
preferences. Such mutant TKs have shown benefit and are currently in clinical use. We describe
here a second strategy to increase the amount of intracellular triphosphorylated GCV by involved
the second enzyme in the GCV activation pathway, guanylate kinase (GMK). As a means to
overcome the bottleneck of prodrug activation from the monophosphate to the disphosphate, we
sought to combine both the critical HSV TK and GMK activities together. In this report we
describe the construction of a fusion or chimeric protein of HSV TK and guanylate kinase, show
data that demonstrate it confers ~175-fold decrease in IC50 compared to HSV TK alone in
response to ganciclovir treatment in stably transfected C6 glioma cells and finally, we present
biochemical evidence of a kinetic basis for this improved cell killing.
1 Data in this chapter are in press as part of the following manuscript: Willmon, C.L., Krabbenhoft, E., and Black, M.E. (2006) A guanylate kinase/HSV-1 thymidine kinase fusion protein enhances prodrug mediated cell killing. Gene Therapy.
37
RESULTS AND DISCUSSION
Herpes simplex virus thymidine kinase is widely used as a suicide gene in combination
with ganciclovir (GCV) for the treatment of a variety of cancers. To date over 70 clinical gene
therapy trials using HSV TK have been approved (115). Moolten first described the potential
application of HSV TK for cancer treatment based on its ability to phosphorylate the antiviral
drug ganciclovir (116). Following delivery of the gene encoding HSV TK to cancer cells, the
prodrug is administered. Because HSV TK, but not the endogenous thymidine kinase, is able to
phosphorylate ganciclovir, cytotoxicity is limited to the site of transfection. Once
monophosphorylated, GCV is further phosphorylated to the triphosphate by the endogenously
expressed enzymes, guanylate kinase (GMK) and nucleoside diphosphokinases, respectively. In
the triphosphorylated state, GCV competes with dGTP for incorporation into the nascent DNA
strand by DNA polymerase and once incorporated prevents chain elongation that subsequently
results in apoptosis (117,118).
A major factor in tumor ablation using the HSV TK/GCV approach involves the transfer
of antimetabolites to neighboring untransfected cells through gap junctions or via apoptotic
vesicles (10,119). Implicit in the bystander effect is that sufficient phosphorylated GCV be
transferred to neighboring cells to elicit cell killing. As such, to achieve complete ablation
myelosuppressive doses of GCV are required. Several approaches have been reported with
varying success to overcome high dose treatments including the use of multiple gene copies to
yield high expression (54), modulation of nucleoside metabolizing pathways using drugs (56),
application of dual suicide gene/dual prodrug approaches (57) and the use of mutant suicide
genes with improved kinetic parameters towards the prodrug (58,59).
38
In this report we describe a novel approach taking advantage of the GCV phosphorylation
pathway to overcome the limitations in the intracellular conversion of GCV-monophosphate
generated by HSV TK to the toxic GCV-triphosphate. There is evidence that once the prodrug
monophosphate is formed, a bottleneck occurs, leading to accumulation of the ineffective
intermediate product (72-74). Responsible for the second phosphorylation step of GCV and its
close relative acyclovir (ACV), guanylate kinase is an essential enzyme whose expression level
and specificity is likely a key determinant in the overall efficacy of the HSV TK/GCV approach
(120,121). While thymidine kinase from herpes simplex virus type I displays broad substrate
specificity and can phosphorylate pyrimidines and both pyrimidine and purine analogs including
thymidine, deoxycytidine, azidothymidine, acyclovir and ganciclovir, guanylate kinase has a
restricted substrate specificity and phosphorylates only the mono-phosphorylated forms of
guanosine (GMP), deoxyguanosine (dGMP) and guanosine analogs such as GCV and ACV
(GCV-MP, ACV-MP) (120-123). From an enzyme kinetic perspective, the high Km of guanylate
kinase for GCV-MP (Km = 42-54 µM) compared to the Km for GMP (~25 µM) supports the
notion that low endogenous GMK activity may limit production of the pharmacologically active
form of GCV (120,124). As a means to overcome the bottleneck of prodrug activation at the
mono- to diphosphate step we combined both the critical HSV TK and GMK activities together
in a single fusion or chimeric protein and assessed the ability of pathway engineering to reduce
to the level of GCV required for effective cell killing.
To construct the guanylate kinase/HSV TK fusion gene, the mouse gmk (mgmk) gene
was isolated from pET23d:mgmk as an BglII/MscI fragment and ligated to BglII/MluI(blunt-
ended) digested pET23d:HSVTK (58,124). The resulting plasmid, designated
pET23d:mgmk/TK, was sequenced to confirm in-frame fusion of the two genes. As a result of
39
the cloning sites used, the first nine amino acids of HSV TK and the last two amino acids of
mGMK were deleted.
To rapidly assess the functionality of the fusion construct with respect to both HSV TK
and guanylate kinases activities genetic complementation for both activities was independently
evaluated. We previously described the establishment of an E. coli strain (TS202A(DE3)) that is
deficient in thymidine kinase activity (tdk-) and is conditionally guanylate kinase deficient (125).
Briefly, the mouse guanylate kinase gene was integrated into the arabinose operon of E. coli
KY895 (ilv- tdk-) to provide induction of GMK protein expression in the presence of arabinose.
The bacterial guanylate kinase was disrupted by insertion of a kanamycin resistance gene into the
bacterial guanylate kinase locus by recombination. Because guanylate kinase is an essential
enzyme, E. coli TS202A(DE3) requires the presence of arabinose. In the absence of arabinose,
no guanylate kinase is expressed and the cells are nonviable. TS202A(DE3) cells were
transformed with pET23d, pET23d:HSVTK, pET23d:mgmk or the fusion construct,
pET23d:mgmk/TK. Cultures were grown in M9 minimal medium in the presence of arabinose
to allow the expression of endogenous mouse guanylate kinase at 37ºC overnight prior to
streaking onto TK selection plates (58) containing arabinose, M9 plates plus arabinose or M9
plates without arabinose (125). As anticipated, cells harboring pET23d and pET23d:mgmk were
not able to complement the TK deficiency of TS202A(DE3), whereas cells harboring
pET23d:HSVTK and pET23d:mgmk/TK were viable (Figure 3.1A). Similarly, only
pET23d:mgmk and pET23d:mgmk/TK transformed TS202A(DE3) were viable on M9 plates in
the absence of arabinose (Figure 3.1B). These results indicate that the fusion protein is
expressed and maintain both thymidine kinase and guanylate kinase activities. All cultures grew
on the control M9 with arabinose plates.
40
Figure 3.1. Functional complementation assays in E. coli TS202A(DE3). E. coli TS202A(DE3) [LAM-, tdk-1, IN(rrnD-rrnE)1, ilv-276 kanR ara- ] was used for genetic complementation of guanylate kinase and thymidine kinase activities as well as for protein expression (125). For culturing purposes cells were grown on 2 x YT (16 g tryptone, 10 g yeast extract, 5 g NaCl per liter) containing 0.2% L-arabinose and 50 µg/ml kanamycin. Carbenicillin was added to 50 µg/ml to media for growth of vector-transformed cells. For complementation of the tdk deficiency, TK selection medium plates (58) supplemented with arabinose and kanamycin were used. For guanylate kinase complementation, M9 minimal medium plates containing kanamycin with or without arabinose were used (125). A) Diagram of the plating pattern used (left) and TS202A(DE3) cells harboring pET23d, pET23d:mgmk, pET23d:HSVTK and pET23d:mgmk/TK on TK selection plates after overnight incubation at 37°C (right). B) Same strains as in (A) after incubation on M9 plus arabinose (left) or M9 minus arabinose plates (right).
With confirmation of both thymidine kinase and guanylate kinase activities in the fusion
protein we next sought to evaluate the ability of the fusion gene to sensitize cancer cells to GCV.
Toward this end, the HSV TK, mgmk and mgmk/TK genes were subcloned into the pREP8D7
vector as described in Kokoris et al. (126). Rat C6 glioma cells were transfected using
electroporation with the vector control (pREP8D7:dual-GFP) and vectors containing HSV TK,
mgmk or mgmk/TK. Insertion of HSV TK, mgmk or the fusion gene into this vector allows
expression from an RSV LTR promoter. Following initial selection on histidinol, GFP expressed
from a constitutive metallothionein promoter enabled cell sorting based on GFP expression to
41
enrich transfectant pools. Immunoblots using polyclonal serum raised against HSV TK or
mGMK were performed on transfectant lysates and show the presence of HSV TK at around 45
kD when the anti-TK serum is used and mGMK at around 23 kDa when anti-mGMK serum is
used (data not shown). These results are in accord with the proteins’ respective predicted
molecular masses. The fusion protein cross-reacted with both anti-sera at the predicted
molecular mass of approximately 68 kDa. Relatively equivalent protein levels were also
observed.
To determine the cytotoxicity of GCV, transfected pools were plated in eight replicates
and treated with a wide range of GCV concentrations (0 to 100 µM). Cell survival was
assessed after 7 days using the redox indicator Alamar Blue and is expressed as a percentage of
control wells that did not receive GCV (0 µM). Figure 3.2 shows a representative GCV
sensitivity curve of the rat C6 glioma cells transfected with pREP8D7:dual-GFP (vector alone),
pREP8D7:TK-GFP, pREP8D7:mgmk-GFP and pREP8D7:mgmk/TK-GFP. Vector control and
mgmk transfected cells were unaffected by GCV treatment. Cells expressing the fusion protein
mGMK/TK were sensitive to GCV concentrations two orders of magnitude lower than those
cells expressing HSV TK (IC50 = 70 µM and 0.4 µM, respectively). As such, the approximately
175-fold lower IC50 value observed with the fusion construct may be highly relevant in clinical
settings where high doses of GCV are not well tolerated by patients. Erbs et al. (87) report
similar findings when cells were transduced with an adenoviral vector expressing a yeast
cytosine deaminase:uridine phosphoryltransferase fusion protein compared to cytosine
deaminase.
While earlier experiments co-expressing HSV TK and mGMK as separate proteins did
not demonstrate more than 2-fold alteration in GCV sensitivity (data not shown and Akyürek et
42
Figure 3.2. GCV sensitivity assays. Ten µg of each DNA (pREP8Δ7:dual-GFP (vector alone), pREP8Δ7:mgmk-GFP, pREP8Δ7:TK-GFP and pREP8Δ7:mgmk/TK-GFP ) was used to transfect 1 x 106 rat C6 glioma cells by electroporation. Cells were grown in Dulbecco’s modified Eagle’s medium plus supplements (5% fetal bovine serum, 1 mM sodium pyruvate, 10 mM HEPES, 100 µM nonessential amino acids, 100 units/ml penicillin G, 10 µg/ml streptomycin sulfate, 292 µg/ml L-glutamine, 100 µM sodium citrate and 0.0014% NaCl). After 2 days of incubation, the medium was replaced with DMEM minus histidine plus supplements containing 0.25 mM histidinol. After two weeks histidinol resistant clones (100-500) were collected and pooled. Approximately 200,000 pooled transfectants were bulk sorted using a MoFlo cell sorter (Cytomation) based on photoactivation and emission of green fluorescent protein (GFP). Data was analyzed using Cyclops software (Cytomation, Fort Collins, CO). To determine the cytotoxicity of GCV, pools of transfectants were transferred to 96 well microtiter plates at an initial density of 600 cells per well in DMEM plus supplements (59). After cell adherence overnight, GCV (0-100µM) was added in sets of eight wells for each concentration tested. The plates were incubated for 6 days at which time the redox indicator dye Alamar Blue was added. Cell survival was determined several hours later as according to the manufacturer’s instructions and is plotted with standard deviation bar. Legend: pREP8Δ7:dual-GFP (vector alone; ), pREP8Δ7:mgmk-GFP (), pREP8Δ7:TK-GFP () and pREP8Δ7:mgmk/TK-GFP ().
43
al. (127)), the substantial improvement observed with the fusion protein prompted us to
investigate whether functional changes towards GCV are a result of fusing the two proteins
together. Towards this end, mGMK, HSV TK and the fusion protein were purified to >95%
homogeneity using affinity chromatography and their respective kinetic parameters determined
(Table 3.1). For all substrates examined, the kcat is 1.8- to 2.9- fold higher (P <0.002) with the
fusion enzyme versus the respective single enzymes. This indicates that the fusion enzyme has
an overall improved turnover rate for all substrates. Interestingly, while mGMK/TK Km values
for the normal substrates (GMP and thymidine) are 2.4- to 3.8-fold higher (P <0.008 for both
substrates), respectively, the Km for GCV is over 3-fold lower (P =0.006) and is suggestive of an
increased relative preference for binding GCV. Perhaps most revealing is that the kcat/Km value
Table 3.1. Kinetic parameters of purified enzymes. Overexpression of TK, mgmk and mgmk/TK from pET vector constructs was as previously reported (59,124). Mgmk was expressed from pETHT which fuses a histidine tag to mGMK and the enzyme was purified using nickel affinity chromatography (124). Purification of TK and mGMK/TK was performed by affinity chromatography using CH-Sepharose 4B coupled p-aminophenylthymidine 3’-phosphate resin as previously described (59). Phosphorylation of radiolabeled substrates (thymidine and GCV) was detected by a filter binding assay (59) and guanylate kinase assays (GMP) were done as described in Brady et al. (124). Radioisotopes for kinetic assays: [methyl-3H] thymidine (specific activity, 85 Ci/mmol) was purchased from Amersham and [8-3H]-ganciclovir (specific activity, 17.6 Ci/mmol) was purchased from Moravek Biochemicals (Brea, CA). Data were plotted as the double reciprocal of velocity (min/µMole x 104) versus substrate concentration (µM-1) and the intercepts values used to determine Km and Vmax. These values were used to calculate the kcat from the double reciprocal plot data as per active site (one active site per monomer). Kinetic values were compared using Student’s unpaired t-test, using KaleidaGraph 4.01 software: a or b indicates a p value < 0.015 compared to mGMK or TK, respectively. *N.D., not determined.
GMP thymidine GCV
Km kcat kcat/Km Km kcat kcat/Km Km kcat kcat/Km
(µM) (s-1) (s-1/µM) (µM) (s-1) (s-1/µM) (µM) (s-1) (s-1/µM)
mGMK 25 426.6 17.1 N.D.* N.D. N.D. N.D.
mGMK/TK 95a 779.4a 8.3a 2.5b 92.2b 36.9 14.6b 162.6b 11.1b
HSV TK N.D. N.D. 1.1 40.3 38 45.5 55.3 1.2
44
of mGMK/TK for GCV is 9-fold better (P <0.002) than the catalytic efficiency displayed by
wild-type HSV TK towards GCV. These data provide, at least in part, a kinetic explanation for
the improved sensitivity of mGMK/TK transfected tumor cells to GCV. Another possibility is
that the fusion protein localizes in transfected cells in a way that provides a more efficient cell
killing.
Many nucleoside analogs require activation by several different enzymes before they can
exert their cytotoxic effects. Evidence suggests that not all enzymes involved in a pathway act
efficiently and thus can lead to an accumulation of potentially toxic intermediate products, e.g.,
AZT-MP. Several groups now recognize this as a rate-limiting step for antiviral, antineoplastic
and suicide gene therapeutic applications (72-74). We report here the first attempt to overcome
the bottleneck in the widely used HSV TK driven GCV activation pathway using a kinetically
favorable fusion protein. By providing effective cell killing at low GCV concentrations, this
novel fusion protein could benefit many patients by improving tumor ablation perhaps through
an expanded bystander effect and by reducing the side effects associated with the high doses of
GCV currently administered.
ACKNOWLEDGMENTS
We are grateful to Mark Kokoris for initial cell culture and kinetic studies. This project was
supported by NIH grant CA85939.
45
CHAPTER FOUR
Improved 5-FC-mediated cell killing by engineered and selected yeast cytosine
deaminase mutants1
ABSTRACT
Cytosine deaminase (CD) converts cytosine to uracil and is responsible for the
conversion of the antifungal agent, 5-fluorocytosine (5FC) to the potent chemotherapeutic drug,
5-fluorouracil (5FU). Since cytosine deaminase (CD) is found in yeast and bacteria but not
mammals, it is being evaluated for use in suicide gene therapy applications for cancer. Local
cytotoxicity is generated in transfected tumors upon administration of the prodrug, 5FC, due to
the actions of CD expressed after delivery of the gene to target cells. The therapeutic efficacy is
limited, in part, by low prodrug activation by CD. As a means to improve the enzyme activity or
preference towards 5FC, a library of 34,000 mutant yeast cytosine deaminases was generated
using regio-specific random mutagenesis, three of which conferred enhanced 5FC sensitivity to a
CD-deficient E. coli as compared to wild-type CD. Sequence analysis revealed each to contain
single unique amino acid substitutions (D92E, M93L, or I98L) at residues within or near the
active site pocket. Furthermore, a previously engineered thermostable mutant (triple yCD) (90)
was used as the backbone for overlaying 5FC-sensitive substitutions obtained from the regio-
specific random mutagenesis. Rat C6 glioma cells were transfected with vectors containing the
wild-type yCD, the three 5FC-sensitizing mutants, along with the two thermostable mutants
(double and triple) and three superimposed variants (triple/D92E, triple M93L, and triple/I98L).
Results from 5FC sensitivity assays reveal a 30% decrease in IC50 of the C6 cells transfected
1Data in this chapter will be submitted for publication in the following manuscript: Willmon, C.L., Stolworthy, T.S., Korkegian, A.M., Stoddard, B.L, and Black, M.E. (2006) Improved 5-FC-mediated cell killing by engineered and selected yeast cytosine deaminase mutants.
46
with the D92E and the thermostable double mutant (A23L/I140L) and a 50% decrease in IC50
with the best thermostable variant (triple), compared to wild-type yCD transfected cells.
Interestingly, one superimposed mutant displays even greater thermostability than the
computationally designed triple yCD variant. These yCD variants should prove useful in future
gene therapy applications.
INTRODUCTION
Suicide gene therapy involves the transduction of tumor cells with a gene encoding an
enzyme. Expression of the enzyme allows for the conversion of the non-toxic prodrug into a
highly toxic antimetabolite, whereby it results in the elimination of cancer cells. Cytosine
deaminase and the antifungal agent, 5-fluorouracil (5FC), is one enzyme/prodrug combination
that has been widely studied and is currently being evaluated in several clinical trials for cancer
(2). The cytosine deaminase (CD) gene is found in yeast (yCD) and bacteria (bCD), but not in
higher eukaryotes (128). CD selectively converts 5FC into its active metabolite 5-fluorouracil
(5FU), an extensively used antineoplastic agent. This selective utilization of 5FC and its absence
from mammalian cells makes CD of great interest for biomedical applications, such as gene
therapy. Once 5FC is deaminated to 5FU, it is further metabolized by endogenous enzymes to
toxic metabolites causing cell death through three mechanisms of action, including the inhibition
of thymidylate synthase leading to inhibition of DNA synthesis, and interference in RNA and
protein processing (129-131).
The CD/5FC combination has advantages over other enzyme/prodrug combinations.
Firstly, the CD/5FC bystander effect is not dependent upon cell-to-cell contact (12) and in
comparison to other enzyme/prodrug combinations it has one of the strongest bystander effects
47
(12,29-32). Without the bystander effect complete tumor ablation with gene therapy would not
be possible because of the inability to deliver a gene to every cell in a tumor. The strong
bystander effect of CD occurs because 5FU is a small, uncharged molecule capable of non-
facilitated diffusion and, therefore, it can pass through cellular membranes allowing it to kill not
only transduced tumor cells in which it is activated but also non-transduced cells in close
proximity. Another advantage of this combination is that 5FU has radiosensitizing properties
(33,34). Since it is unlikely that treatment with gene therapy would be the only course of action
in patients, radiosensitizing effects are important. In fact, in vivo results utilizing CD show a
significant bystander effect (33,35) at clinically relevant 5FC doses and radiation regimens
(33,36).
As mentioned above there are two cytosine deaminases currently being used in suicide
gene therapy applications, the E. coli (bCD) and S. cerevisiae (yCD) cytosine deaminases.
Although product release from the yeast cytosine deaminase is rate-limiting (132), the
homodimeric yCD has been observed to display superior kinetic properties (22-fold lower Km)
towards 5FC and slightly improved efficacy for treating tumors in mice than bCD (37).
However, wild-type yCD is quite thermolabile compared to bCD, limiting its production and
enzymatic efficiency in transfected cells at 37°C. Therefore, many studies to date have focused
on bCD due to the more favorable thermostability it displays. In order to address the
thermostability of yCD, Korkegian et al. (90) created two yCD variants by computational design
with increased thermal stability at physiological temperatures and, therefore, possible better
enzymes for use in gene therapy applications. The double mutant (A23L/I1140L) and the final
redesigned mutant (A23L/V108I/I140L or triple mutant) display over a 5- and 30-fold increased
enzymatic activity half-life at physiological temperatures, respectively, and preserved catalytic
48
efficiency for cytosine. Other than the Korkegian et al. study, no attempts have specifically
focused on improving yCD for gene therapy.
Although yCD displays better 5FC kinetics compared to bCD, the side effects associated
with the requisite therapeutic 5FC doses remains problematic. To address this we sought to shift
the preference of yCD further towards this prodrug for use in suicide gene therapy applications.
To enhance the specificity of yCD towards 5FC two approaches were taken. The first approach
was to re-engineer the active site by performing regio-specific random mutagenesis of the 83-98
amino acid (a.a.) region of yCD (Figure 4.1). This region was chosen based upon sequence
homology studies (133) and our previous crystal structure studies (134) that led us to believe this
site is involved in substrate binding and, thus, would be an excellent target for mutagenesis. In
this region, two cysteines (positions 91 and 94) coordinate the zinc ion and leucine 88 (L88) is
directly involved in substrate coordination within the active site of yCD. From the regio-specific
random library, we identified three mutants (D92E, M93L and I98L) that conferred superior
activity in an E. coli based screening system and were chosen for further evaluation. For the
second approach to improve yCD, we surmised that the increased half-life displayed by the two
thermostable mutants (double and triple) described above would correlate to an increased degree
of 5FC activation. In order to exploit the enhanced thermostability of the triple mutant and the
increased activity of the selected random mutants we engineered yCD variants to contain both
the thermostable and 5FC-sensitive substitutions identified from the regio-specific random
D T T L Y T T L S P C D M C T G A I I M 83 98
Figure 4.1. Yeast cytosine deaminase amino acid sequence (in single-letter code) spanning residues 81-100. The amino acid region targeted for regio-specific random mutagenesis is highlighted in blue. Highly conserved residues are underlined and were not targeted for mutagenesis.
49
library (A23L/V108I/I140L with D92E, M93L or I98L) and tested these yCD variants in a rat C6
glioma cell model for enhanced sensitivity to 5FC. While, the two thermostable mutants and one
regio-specific mutant display superior tumor cell killing in comparison to wild-type yCD, the
superimposed mutant did not. Perhaps most unexpectedly, we created a variant yCD
(triple/D92E) with even greater thermostability than the previously characterized triple mutant.
We believe these mutants could provide a significant advantage compared to wild-type yCD for
suicide gene therapy applications.
MATERIALS AND METHODS
Materials
Oligonucleotides used to mutate and sequence yCD were obtained from either Sigma-
Proligo (St. Louis, MO) or Integrated DNA Technologies (Coralville, IA). Restriction enzymes
used to construct and screen the yCD mutants were obtained from New England Biolabs
(Beverly, MA). The CD-containing plasmid, pRS306-FCY1 (a kind gift from Dr. Jean-
Emmanuel Kurtz and Dr. Richard Jund (Strasbourg, France)), was used as the source of the
FCY1 (yCD) gene used for the construction of the plasmid pETHT:yCD expressing His-tagged
yCD as described in Ireton et al. (134). Wizard PCR preps from Promega (Madison, WI),
HiSpeed Plasmid Mini Kit from Qiagen (Valencia, CA), and StrataPrep EF Plasmid Midikit
from Stratagene (LaJolla, CA) were used to purify plasmid DNA. Alamar Blue was purchased
from Serotec Limited (Oxford, UK). ). All cell culture reagents were purchased from Gibco
(Carlsbad, CA). All other reagents were purchased from Sigma (St. Louis, MO) unless
otherwise noted.
50
Bacterial strains
Escherichia coli GIA39, a strain deficient in cytosine deaminase and orotidine 5’-
phosphate decarboxylase, was obtained from the E. coli Genetic Stock Center (CGSC #5594:
thr- dadB3 fhuA21 codA1 lacY1 tsk-95 glnV44(AS) λ- pyrF101 his-108 argG6 ilvA634 thi-1
deoC1 glt-15). E. coli GIA39 was lysogenized with DE3 according to the manufacturer’s
instructions (Novagen, Madison, WI). The derived strain, GIA39(DE3), was used in the genetic
complementation assays for cytosine deaminase activity. E. coli strain CJ236 (F+LAM-, ung-1,
relA1, dut-1, spoT1, thi-1) was used to produce single-stranded DNA for Kunkel site-directed
mutagenesis procedures (101,102). The E. coli strain NM522 (F+ lacIq ∆(lacZ)-M15
proA+B+/supE thi∆(lac-proAB) ∆(hsdMS-mcrB)5(rk-mk
-McrBC-)) and E. coli strain XL1-Blue
(F'::Tn10 proA+B+ laclq D(lacZ) M15/recA1 endA1 gyrA96 (Nalr) thi hsdR17 (rK-mK
+) supE44
relA1 lac) were used as recipients for certain cloning procedures.
Construction of the yCD regio-specific random library
Preparation of the 83-98 a.a. insert: Random mutagenesis at the 83-98 a.a. region of yCD was
performed as described in Kurtz and Black (135). The following oligonucleotides were designed
to synthesize a 139 bp dsDNA fragment including the 12 codons (T83, L84, Y85, T86, L88, S89,
D92, M93, T95, G96 and I98) that were randomized at 21% as indicated in bold in the MB 224
oligonucleotide:
MB 223 (56mer), 5’ GTGAGATCTCCACTTTGGAAAACTGTGGGAGATTAGAGGGCAAA
GTGTACAAAGAT 3’
MB 224 (95mer), 5’ CCGACAA(CACAGCGTG)GAATACCATACATGATGATGGCACC
TGTACACATGTCGCATGGAGACAGCGTCGTATACAAAGTGGTATCTTTGTACAC 3’
51
MB 225 (18mer), 5’ GTG(AGATCT)CCACTTTGG 3’
MB 226 (17mer), 5’ CCGACAACACAGCGTGG 3’
DraIII and BglII restriction sites are noted in the parenthesis in MB 224 and MB 225,
respectively, and were used to insert the synthesized dsDNA fragments into the inactivated yCD
gene described below. Briefly, the 139 bp dsDNA fragment was synthesized by annealing 50
pmol each of MB 223 and MB 224 with 10 x annealing buffer (70 mM Tris-HCl at pH 7.5, 60
mM MgCl2, 200 mM NaCl) in a final volume of 50 µL at 95°C for 5 min, 65°C for 20 min, and
room temperature for 10 min. Next, the annealed product was extended with the Klenow
fragment of E. coli DNA polymerase in an 80 µL reaction mixture consisting of the 40 µL
annealed product, 4 µL of 10 x annealing buffer, 5.6 µL of 10 mM dNTPs, 1.6 µL of 0.1 M
DTT, and 4.8 µL of Klenow (5U/µL), at 37°C for 30 min, 65°C for 10 min, and room
temperature for 10 min.
Amplification of the random insert. A master mix was first prepared consisting of 110 µL of 10
x PCR buffer (200 mM Tris-HCl at pH 8.3, 250 mM KCl, 15 mM MgCl2, 0.5% Tween-20), 200
pmol of MB 225 and MB 226 each, 4.4 µL of 10 mg/mL BSA, 5.5 µL of 10 mM dNTPs, 4.4 µL
of Taq polymerase (5U/µL). The extended product (16 pmol or 5 µL) and 27.8 µL of the master
mix were mixed to a final volume of 200 µL and split into four tubes of 50 µL each. The 50 µL
mixtures were subjected to amplification using an Eppendorf Thermocycler by 30 cycles with 1
cycle at 94°C for 1 min, 34°C for 2 min, and followed with a 7 min extension at 72°C.
Amplification of the 139 bp insert was confirmed by 2% agarose gel electrophoresis.
Construction of recombinant yCD variants: To construct the vector carrying the inactive or
dummy gene the DraIII and BglII sites within the backbone of pETHT first needed to be
inactivated. To do this Kunkel site-directed mutagenesis (101,102) of pETHT:yCD was
52
performed. The following oligos were used for the mutagenesis: MB 277 (21mer), 5’ CGATGG
CCCAATACGTGAACC 3’ for the DraIII removal and MB 307 (22mer), 5’ CGGGATCGC
GATCGCGGGCAGC 3’ for removal of the BglII site. Next, the yCD gene was inactivated by
restriction with AccI followed by extension with the Klenow fragment and religation. The
resulting inactivated yCD was designated as “pETHT:yCD dummy.” To ligate the insert with
the dummy vector, all 200 µL of the amplified insert was digested with BglII and DraIII,
electrophoresed and excised from the agarose gel using the Promega Wizard SV Gel and PCR
Clean-Up System according to the manufacturer’s directions. Additionally, 9.5 µg of the
dummy vector was restricted with BglII and DraIII enzymes, followed by treatment with shrimp
alkaline phosphatase (SAP) (Promega, Madison, WI) at 37°C for 15 min, and incubation at 65°C
for 15 min. Next, 2 µL of the SAP-treated vector and 14 µL of the purified random insert were
ligated using 2 µL T4 DNA ligase (Promega, Madison, WI) in a total of 20 µL at 12°C
overnight.
Transformation of E. coli GIA39(DE3) and positive selection: Approximately 3.5 µL of the
ligated product was electroporated into 40 µL of electrocompetent E. coli GIA39(DE3) and then
shaken at 37°C for 1 hr in 1 mL of SOC medium (3 g Bactopeptone, 2.5 g yeast extract, 1 M
NaCl, 1 M KCl, 5 mM MgSO4, 5 mM MgCl2 and 1.8% glucose per liter). The transformation
mixture was concentrated by pelleting, resuspended in 100 µL of 0.9% NaCl and plated at
various volumes onto 2 x YT rich medium, uracil and cytosine minimal media plates
supplemented with 50 µg/mL carbenicillin. The use of the cytosine minimal medium determines
functionality of the yCD variants, while 2 x YT and uracil minimal media are used as positive
controls. Uracil minimal medium (500 mL) was prepared from 0.36 g yeast synthetic dropout
without leucine, 50 mL 10 x M9 salts (15 g KH2PO4, 33.9 g anhydrous NaHPO4, 2.5 g NaCl, 5.0
53
g NH4Cl), 1 mM MgSO4, 2.5 mL, 20% glucose, 0.1 mM CaCl2, 1 mL 2% leucine, and 7.5 g
Bactoagar. Cytosine minimal medium (500 mL) was prepared from 1 mM cytosine, 0.96 g yeast
synthetic dropout without uracil, 8.5 g Bactoagar, 50 mL 10 x M9 salts, 1 mM MgSO4, 2.5 mL
20% glucose, and 0.1 mM CaCl2. The 2 x YT plates were incubated at 37°C overnight, and the
uracil and cytosine plates were incubated at 37°C for approximately 36 hr. The number of
transformant colonies from the 2 x YT plate were counted to estimate the size of the library.
Transformants from the cytosine plates were picked and restreaked onto fresh cytosine plates to
confirm the phenotype.
Negative selection of 5FC sensitive yCD mutants: Genetic complementation of the yCD variants
was done as previously established with bCD (77). To determine the ability of the mutants to
confer 5FC sensitivity, the functional variants determined by the positive selection, described
above, were streaked onto cytosine plates supplemented with 5FC at 10 µg/mL, a sublethal dose
for wild-type yCD, and incubated for approximately 36 hr at 37°C. Colonies unable to grow on
the 5FC plates were selected from the control plates and
subjected to additional rounds of negative selection by decreasing the 5FC concentration to 5, 2,
1 and 0.5 µg/mL. Next, plasmid DNA of the yCD variants was purified by Wizard PCR preps,
and the randomized region was sequenced using the T7 terminator primer (5’TATGCTAGTTAT
TGCTCAG 3’) at the core sequencing laboratory at Washington State University (WSU).
Construction of the thermostable and overlay mutants
yCD genes from pET15b:yCD-A23L/V108L (double) and pET15b:yCD-
A23L/V108L/I140L (triple) (90) were sub-cloned into the mammalian expression vector,
pCDNA6/myc-His B (Invitrogen, Carlsbad, CA). Briefly, to isolate the yCD double and triple
54
genes, a 50 µL reaction of 5 µg of pET15b:yCD double and triple plasmid DNA was restriction
digested with NcoI, followed by fill-in with the Klenow fragment, and then by digestion with
XhoI. The digested DNA was electrophoresed and purified from the agarose gel using the
QIAquik Gel Extraction Kit (Qiagen, Valencia, CA) according to the manufacturer’s directions.
Next, in a 50 µL reaction, 5 µg of pCDNA6/myc-His B was digested with EcoRV and XhoI,
followed by SAP treatment. Then, 2 µL of the SAP-treated vector and 4 µL of the purified
random insert were ligated using 1 µL T4 DNA ligase (Promega, Madison, WI) in a total of 10
µL at 12°C overnight. Approximately 2 µL of the ligation reaction was used to transform E. coli
NM522. After restriction enzyme verification, DNA sequencing analysis was performed at the
core sequencing facility at WSU to confirm the presence of the yCD double and triple genes.
Next, site-directed mutagenesis was performed to introduce the mutations derived from
the regio-specific random mutagenesis to the triple mutant using the QuikChange Site-directed
Mutagenesis Kit from Stratagene (La Jolla, CA) according to the manufacturer’s protocol. Three
individual oligonucleotides containing the D92E, M93L, or I98L and a silent mutation to remove
a restriction site for screening purposes were synthesized by Integrated DNA Technologies
(Coralville, IA). In the mutagenic oligonucleotides the bolded nucleotide indicates the regio-
specific random substitution and the underlined nucleotide indicates the removal of the
restriction site.
D92E, loss of AflIII site
MB 374 (31mer), 5’ CGCTGTCTCCATGCGAAATGTGTACAGGTGC 3’
MB 375 (31mer), 5’ CGACCTGTACACATTTCGCATGGAGACAGCG 3’
M93L, loss of AflIII site
MB 376 (31mer), 5’ CGCTGTCTCCATGCGACCTGTGTACAGGTGC 3’
55
MB 377 (31mer), 5’ GCACCTGTACACAGGTCGCATGGAGACAGCG 3’
I98L, loss of BanI site
MB 378 (34 mer), 5’ GCGACATGTGTACAGGAGCCCTCATCATGTATGG 3’
MB 379 (34 mer), 5’ CCATACATGATGAGGGCTCCTGTACACATGTCGC 3’
After restriction enzyme verification, DNA sequencing analysis was performed at the core
sequencing facility at WSU to confirm the presence of the correct mutation.
5FC sensitivity assays
One µg of each DNA (pCDNA (vector alone), pCDNA:yCD, pCDNA:yCD D92E,
pCDNA:yCD M93L, pCDNA:yCD I98L, pCDNA:double, pCDNA:triple;, pCDNA:triple/D92E;
pCDNA:triple/M93L; pCDNA:triple/I98L) was used to transfect 1 x 105 rat C6 glioma cells by
lipofection using FuGENE 6 transfection reagent (Roche Diagnostics, Penzberg, Germany) at a
3:1 ratio according to the manufacturer’s directions. Immunoblots were performed to assess
protein levels. Briefly, pools of transfectants were harvested and resuspended at 100,000
cells/µL in lysis buffer (for 2 mL: 2 µL 1 M DTT, 20 µL 1 M HEPES, 40 µL Nonidet P40
(Roche Diagnostics, Pernzberg, Germany) 2 µL MgAc2, H2O to final volume). The resuspended
pellets were incubated on ice for 20 min and subjected to centrifugation at 4°C for 20 min to
pellet debris. Samples (10 µL per well) were run on a 15% SDS gel, transferred to a
nitrocellulose membrane and blocked with 3% gelatin in Tris-buffered saline. The membrane
was probed with rabbit polyclonal yCD antibody (gift from Dr. Alnawaz Rehemtulla, U.
Michigan, Ann Arbor, MI) followed by goat anti-rabbit AP-conjugated antibody. The blot was
developed using the AP Conjugate Substrate Kit (Bio-Rad, Hercules, CA). To determine the
cytotoxicity of 5FC, pools of transfectants were transferred to 96 well microtiter plates at an
56
initial density of 250 cells per well in DMEM plus supplements (59). After cell adherence
overnight, 5FC (0-10 mM) was added in sets of eight wells for each concentration tested. The
plates were incubated for 6 days at 37ºC in 5% CO2 at which time the redox-indicator dye
Alamar Blue was added. Cell survival was determined several hours later as according to the
manufacturer’s instructions and the data were plotted with a standard error of the mean bar. At
least three replicates were performed.
Protein expression and purification
Protein expression and purification of yCD was carried out as previously described (134),
with the exception that buffer exchange was carried out with Biorad pre-packed Econo-Pac
10DG buffer exchange columns rather than overnight dialysis. The protein was expressed from
a pET15b vector with a thrombin cleavable 6-His tag within BL21-RIL cells. All kinetics and
thermostability experiments were carried out with fresh, unfrozen protein stored at 4ºC for 2
weeks or less.
Activity assays
The conversion of 5FC to 5FU by yCD was measured by UV spectroscopy by monitoring
change in absorbance at 238 nM. The protein was diluted to 2 µM in 50 mM Tris-Cl (pH 7.5)
and mixed 1:1 with a range of nine cytosine concentrations from 0.2 – 1 mM in the same buffer.
Absorbance at 238 nm was measured every 5 seconds until baseline was reached with the first
reading taken 5 s after mixing. Measurements were taken in quadruplicate and averaged to
reduce error. Initial velocity was calculated as a function of the initial slope by curve-fitting the
resulting plot, taking the derivative and extrapolating back to time zero. Km and kcat values of
57
wild-type yCD and mutant constructs were determined from a double reciprocal (Lineweaver-
Burke) plot of the resulting data and the catalytic efficiency kcat/Km was calculated from these
values.
Circular dichroism measurement and thermal denaturation experiments
CD data were collected on an Aviv 62A DS spectrometer as described in Dantas et al.
(136). Wavelength scans were run from 200-260 nm to determine the folded state of the protein,
the ratio of concentration to signal strength and the wavelength where signal strength was at its
highest. Temperature melts were run with 8-12 µM protein in a 2 mm pathlength cuvette from
10 to 98ºC in 2º increments with temperature regulated by a Peltier device. Sample temperature
was allowed to equilibrate for 30 s before measurement and signal was collected and averaged
over 30 s. Denaturation was recorded as a change in elipticity over temperature. Apparent Tms
were determined by curve-fitting.
RESULTS
Screening and characterization of the 83-98 a.a. regio-specific random library by positive and
negative genetic complementation
Previously we established genetic complementation in E. coli GIA39(DE3) to identify
bCD variants with enhanced 5FC sensitivity (77). Using this same technique we evaluated the
yCD variants for 5FC sensitivity. Since E. coli GIA39(DE3) lacks the salvage pathway enzyme,
cytosine deaminase (codA-), and the de novo pathway enzyme, orotidine 5’-monophosphate
decarboxylase (pyrF-), pyrimidine synthesis in this strain is prevented. Because RNA and DNA
synthesis in E. coli GIA39(DE3) requires an external uracil source this strain is not viable on
58
medium containing cytosine as a sole pyrimidine source. When GIA39(DE3) harbors a plasmid
carrying a functional yCD, growth occurs on both uracil- and cytosine-containing media.
However, cells that do not express a functional yCD will only be viable on uracil-containing
plates. To eliminate the possibility of false positive due to contamination by the wild-type yCD,
we inactivated the parental yCD gene, as described in Materials and Methods, by introducing a
frame-shift in the yCD open reading frame. The reading frame should be restored in variants
containing an introduced 83-98 a.a. randomized fragment.
Using this strain, a two-step selection procedure was used to identify functional yCD
variants with enhanced 5FC activity. Firstly, functional yCD variants were selected based upon
their growth on cytosine-containing plates. Secondly, the functional variants were screened on
cytosine plates containing 5FC. It was determined that the 5FC dose at which wild-type yCD
survives, or the sub-lethal dose, is 10 µg/mL. Although wild-type pETHT:yCD will grow on
plates containing the sub-lethal dose of 5FC, any mutant with increased 5FC sensitivity will not
be viable. To identify the best variants within the library pool the 5FC concentration in these
plates was sequentially decreased from 10 µg/mL to the lowest concentration of 0.5 µg/mL.
To identify yCD variants with increased activity to 5FC, we subjected 11 codons within
the 83-98 a.a. region of yCD (T83, L84, Y85, T86, L88, S89, D92, M93, T95, G96 and I98) to
regio-specific random mutagenesis (Figure 4.1). In order to decrease the likelihood of creating a
great number of non-functional variants (137), we limited the randomness of the library to 21%.
Additionally, we did not target specific amino acids found to be highly conserved (T87, P90,
C91 and C94) in alignment studies (133). Using the positive selection system described above
and in Materials and Methods, an estimated total of 34,000 transformants were screened. Of
these 34,000, 50 colonies (~0.15%) were identified as functional yCDs. After 5 rounds of
59
negative selection on successively lower 5FC concentrations, we identified six mutants that
displayed death at 0.5 µg/mL of 5FC, the lowest level of sensitivity in the selection system.
In order to evaluate the library diversity, along with identifying which a.a. substitutions
are tolerated, DNA from colonies on the non-selective 2 x YT, cytosine- and 5FC-containing
plates were isolated and sequenced. Sequence analyses revealed a broad spectrum of mutations
observed in the regio-specific library (data not shown). This indicates that the oligos used to
generate the library contained random sequences, and the selection/screening worked as
evidenced by the limited number of nucleotide substitutions found at certain residues. The
amino acid sequences of the six best 5FC mutants were found to have similar substitutions.
Two of the six had a substitution at D92 to glutamic acid, another two had M93L
substitutions, and the last two had substitutions of I98 to leucine. Importantly, the I98L
mutant has different nucleotide-level changes, suggesting these mutants were derived
Table 4.1. Mutant yCD enzymes identified from the regio-specific random mutagenesis along with all mutants analyzed in vitro and their corresponding amino acid (a.a.) residue substitutions.
Mutant Name A.A. Substitutions
Regio-specific MutantsD92E
D92E
M93L
I98L
Thermostable Mutants
Double A23L/I140L
Triple A23L/V108I/I140L
Superimposed Mutants
Triple/D92E A23L/D92E/V108I/I140L
Triple/M93L A23L/M93L/V108I/I140L
Triple/I98L A23L/I98L/V108I/I140L
60
independently. Therefore, from a library of 34,000 transformants, three mutants were identified
that confer 5FC sensitivity (D92E, M93L, and I98L) at the lowest concentration in our selection
scheme (Table 4.1).
Construction of superimposed yCD mutants
Recently, we performed a study aimed at improving the stability of the thermolabile yCD
(90). From this study, two thermostable mutants (double (A23L/I140L) and triple
(A23L/V108I/I140L)) were generated that provide significantly longer enzymatic activities than
wild-type at physiological temperatures (Table 4.1). From that study, two hypotheses emerged:
(1) The two thermostable mutants created would display dramatic enhancements in 5FC
activation in vitro by virtue of an extended protein half-life, and (2) by overlaying the
substitutions found in the 5FC sensitive mutants onto the best thermostable mutant, novel yCD
variants (triple/D92E, triple/M93L, and triple/I98L) would be created that would result in even
greater 5FC activity in mammalian tumor cells (Table 4.1). Using site-directed mutagenesis, as
described in the Materials and Methods section, the superimposed yCD mutants were engineered
and tested in a mammalian tumor cell line.
In vitro 5FC sensitivity assays
To determine the 5FC activity of these mutants in vitro, mammalian expression vectors
encoding the yCD variants were constructed and used to transfect rat C6 glioma cells (see
Materials and Methods). Immunoblot analyses of lysates from the pools of transfectants show
similar expression levels for all of the mutants and wild-type yCD, with no detectable expression
in vector control pools (data not shown). Pools of stable transfectants were assayed for their
61
level of 5FC sensitivity over a drug range of 1-10 mM. Representative results of the 5FC
sensitivities displayed by the yCD mutants, wild-type yCD and a vector control are shown in
Figure 4.2. Little to no toxicity was observed with vector alone at the lower 5FC doses;
however, above 6 mM 5FC an inherent cytotoxicity is observable in the glioma cell line. No
increase in sensitivity was observed with the I98L substitution (Figure 4.2A). The M93L mutant
displays an IC50 of 9.5 mM, an estimated decrease of 15% from wild-type yCD. D92E displays
the greatest reduction in IC50 (~30%) for 5FC of the regio-specific random mutants, compared to
wild-type yCD.
Figure 4.2. Sensitivity of yCD-expressing rat C6 constructs to 5FC. Pools of stable transfectants containing vector only (pCDNA), wild-type yeast cytosine deaminase (yCD), A. regio-specific random mutants (D92E, M93L or I98L), B. thermostable mutants (double and triple) and the superimposed mutants (triple/D92E, triple/M93L or triple/I98L) were constructed and transfected in rat C6 glioma cells and evaluated for 5FC sensitivity as described in Materials and Methods. After six days of 5FC treatment, the growth inhibition was determined by staining with Alamar Blue and fluorescence recorded at 530/590 nm. Each data point (mean ± SEM, n=3 performed with at least fifteen replicates) is expressed as a percentage of the value for control wells with no 5FC treatment.
B.
A.
0
10
20
30
40
50
60
70
80
90
100
110
0 1 2 3 4 5 6 7 8 9 10 11
pCDNAyCDD92EM93LI98L
% S
urv
ival
[5FC] (mM)
0
10
20
30
40
50
60
70
80
90
100
110
0 1 2 3 4 5 6 7 8 9 10 11
pCDNAyCDDoubleTripleTriple/D92ETriple/M93LTriple/I98L
% S
urv
ival
[5FC] (mM)
62
The thermostable double and triple mutants display increased sensitivity towards 5FC in
glioma cells (Figure 4.2B). The double mutant displays a similar IC50 (8 mM) to D92E, an
approximate 30% reduction in IC50 compared to wild-type yCD-transfected cells. The greatest
enhancement in activity was observed in the triple mutant with an IC50 of approximately 6 mM,
an estimated 50% reduction from wild-type yCD. Unfortunately, none of the superimposed
mutants exhibit greatly enhanced activities towards 5FC (Figure 4.2B). The triple/I98L and
triple/M93L have an IC50 = 9 mM, while the triple/D92E has a similar IC50 to wild-type. These
results led us to hypothesize that the addition of the further substitutions to the triple mutant may
destabilize the superior triple mutant leading to a decrease in sensitivity towards 5FC, evidenced
in the superimposed enzymes activities in the glioma cells in comparison to the triple mutant.
Enzyme kinetics and thermal melts of purified proteins
The enhanced activity towards 5FC with the D92E, triple and triple/D92E mutants
observed in tumor cells led us to investigate the functional changes of these yCD variants with
regard to the their enzymatic activities and thermal stabilities. Using purified proteins, enzyme
assays with cytosine and 5FC as substrates were performed. The Km values for cytosine
observed with the triple and triple/D92E mutants are approximately 75% and 87% decreased
from the Km of wild-type yCD, respectively (Table 4.2). The cytosine kcat values for the triple
and triple/D92E variants are decreased by 75% and 79%, when compared to wild-type. Overall,
the relative catalytic efficiency (kcat/Km) with cytosine for the triple mutant was not appreciably
different, while the triple/D92E has an approximate 65% decrease in efficiency from the wild-
type yCD value.
63
Table 4.2. Kinetic parameters of the purified wild-type and mutant yCD enzymes. a N.D., not determined.
With 5FC as the substrate, the triple and triple/D92E enzymes were observed to have
32% and 16% reductions in Km values compared to wild-type (Table 4.2). The 5FC kcat values
are 16 s-1 and 10 s-1 for the triple and triple/D92E variants, respectively, which are 23% and 52%
decreased from the wild-type kcat value of 20.7 s-1. This results in the catalytic efficiencies of
these enzymes being reduced from wild-type yCD by 8% for the triple yCD and 29% for the
triple/D92E yCD. Although the kinetics of the triple mutant and the triple/D92E mutant are
similar for the natural substrate cytosine, the triple mutant has more favorable kinetics for 5FC.
However, when the relative specificities of these mutant enzymes, or their substrate preference
for 5FC, is compared (calculated as [kcat/Km (5FC)/(kcat/Km (5FC) + kcat/Km (cyt))]), there is no
considerable difference in substrate preference between the wild-type enzyme and triple mutant.
The triple/D92E mutant has only a modest 1.3-fold increase in preference for 5FC.
In light of the introduced thermostability of the triple mutant, we examined the stabilizing
effects of the individual D92E substitution and of the superimposed triple/D92E mutant.
Thermal denaturation experiments were performed on the wild-type, D92E, and triple/D92E
proteins using circular dichroism (CD) spectroscopy as outlined in Materials and Methods
(Figure 4.3). Wild-type yCD has an apparent melting temperature (Tm) of 52ºC. The
triple/D92E construct has a dramatic 16ºC increase in apparent Tm compared to wild-type yCD,
and 6ºC higher than the triple mutant.
Cytosine 5FC
Km kcat kcat/Km Km kcat kcat/Km
(mM) (s-1) (s-1/M) (mM) (s-1) (s-1/M)
yCD 5.35 433 8.10 x 104 0.195 20.7 1.06 x 105
D92E N.D.a N.D. N.D. N.D. N.D. N.D.
Triple 1.33 107 8.08 x 104 0.164 16.0 9.78 x 104
Triple/D92E 0.73 20.7 2.85 x 104 0.133 10.0 7.52 x 104
64
Figure 4.3. Thermal denaturation measurements of the wild-type and yCD variants. The temperature melt measures the change in signal at 220 nm over a range of temperatures. All constructs show a folded baseline followed by a sigmoidal two-state transition to an unfolded baseline. The baseline plateaus correspond to an assignment of 100% folded protein.
DISCUSSION
While the bacterial cytosine deaminase/5FC combination has been widely investigated
and is currently being used in several gene therapy clinical trials, the yeast cytosine deaminase
might prove to be a more efficient enzyme for use in suicide gene therapy because of its inherent
prodrug activity (37). The advantage of having a suicide enzyme with the ability to use the
prodrug to a greater extent would be highly advantageous since the active metabolite of 5FC,
5FU, has adverse side effects when given at high doses. However, because the yeast cytosine
deaminase is thermolabile the focus has been on utilizing the less active, but thermostable, bCD.
Using computational redesign we created two thermostable yCD mutants (90), the double and
triple mutant yCDs, that could advance the use of yCD in antitumor gene therapy applications.
In this paper we report two attempts to enhance the activity of yCD for the prodrug 5FC. First,
we targeted an important amino acid region (83-98 a.a.) of yCD for regio-specific random
0
20
40
60
80
100
5 15 25 35 45 55 65 75 85 95 105
yCDTripleTriple/D92EP
erc
en
t F
old
ed
Temperature (C)
65
mutagenesis. Secondly, combination mutant yCDs were engineered that contain the
substitutions seen in the thermostable as well as the regio-specific mutants.
Previously, sequence homology alignment studies of 16 sequences, which included yCD,
bCD, deoxycytidylate deaminase from bacteriophages T2, T4 and human sources, suggested that
the 83-98 a.a. region of yCD might be important in nucleobase binding. Of particular interest are
two cysteine residues (positions 91 and 94) within this region that were found to be directly
involved in coordinating the zinc ion in human deoxycytidylate deaminase (138) and E. coli
cytidine deaminase (139). The crystal structure of yCD (134) clearly demonstrates C91 and C94
along with H62 and a water molecule tetrahedrally coordinate the catalytic zinc atom within the
active site. In addition to the importance of the cysteine residues within the target region, the
structure suggests that L88 is involved in substrate coordination. Taken together, these studies
indicated this region to be an excellent target for re-engineering.
Of the 34,000 regio-specific random yCD transformants that complemented the CD-
deficient E. coli, six were identified to have a 20-fold increased activity towards 5FC. Sequence
analysis of these mutants revealed three sets of identical single a.a. substitutions (D29E, M93L
or I98L), with different nucleotide substitutions that were evaluated further in vitro. Mutants
D92E and M93L were found to display approximately 30% and 15% decreases, respectively, in
IC50 for 5FC in rat C6 glioma cells when compared to wild-type. The I98L showed no
appreciable difference in IC50 for 5FC from wild-type yCD. The D92E substitution is localized
near the active site of yCD at the homodimer interface.
The thermostable mutants initially described in Korkegian et al. (90) display superior
tumor cell killing in the rat C6 glioma model (Figure 4.2). The double mutant was observed to
have a 30% decrease in IC50 for 5FC and the triple mutant shows a 50% decrease in IC50,
66
compared to wild-type. Because these mutants display not only desirable thermostability but
also an appreciable increase in sensitivity towards 5FC, presumably because of the extended
half-life of the enzymes, we engineered superimposed mutants containing the thermostable and
regio-specific substitutions in an attempt to create a yCD variant that is catalytically efficient
towards 5FC, as well as thermostable. The resulting mutants, however, did not display
substantial enhancement in activities towards the prodrug. The triple/M93L mutant displays a
similar IC50 for 5FC compared to the M93L mutant (IC50 ≈ 9 mM), but the triple/D92E,
unexpectedly, exhibits no alteration in IC50 compared to wild-type yCD. Therefore, the addition
of the D92E substitution to the triple mutant results in an increased in IC50. Although the IC50 of
I98L by itself is not as greatly enhanced as it is for the triple mutant, the triple/I98L shows a
decreased IC50 towards 5FC compared to wild-type. Since the I98L alone displays no increase in
activity in tumor cell killing with 5FC in the rat glioma model compared to wild-type, we believe
the slight decrease in IC50 for the triple/I98L mutant is a result of the stabilizing effects of the
triple mutant substitutions providing an extended half-life at 37ºC. It is unclear why overlaying
one mutant (D92E) causes an increase in IC50 whereas, the I98L superimposition results in a
decrease.
The kinetic parameters of the triple and triple/D92E show no major improvements when
compared to wild-type yCD. Rather, the triple/D92E displays even poorer kinetics for 5FC, with
over a 50% decrease in kcat and a 32% decrease in Km. Moreover, the relative specificities of the
two mutants when compared to the wild-type enzyme were not improved, signifying an overall
preference for utilizing 5FC in a milieu of nucleosides within the cell is not responsible for the
triple mutant’s overall enhancement in tumor cell killing. We also observed a significant
increase in thermostability with the triple/D92E mutant when compared with wild-type yCD
67
(16ºC increase) and the triple mutant enzyme (6ºC increase) using CD spectroscopy. Therefore,
it appears that the D92E substitution stabilizes the protein, and that is what is responsible for
providing sensitization to 5FC in a manner similar to the triple mutant. Interestingly, the D92E
substitution would not have been predicted in the computation redesign study because any
residue involved in catalysis, located within 4Å of the active site, or involved in the dimer
interface was held fixed (90). Therefore, either method of stabilizing the protein, improved
packing of the hydrophobic core (as observed with the triple mutant) or stabilization of the dimer
interface, may put the enzyme at a threshold for the 5FC sensitivity phenotype, beyond which no
extra effect is observed.
Future experiments utilizing the triple or triple/D92E mutant as the scaffold for directed
evolution followed by screening for the desired alteration in 5FC activity may identify mutants
that could prove to be of great interest for gene therapy. Additionally, several studies have fused
the yCD gene with herpes simplex virus thymidine kinase (80,81,140) or yeast uracil
phosphoribosyltransferase (87,88) and have conferred increased tumor cell killing upon the
treatment with GCV and/or 5FC. By introducing the superior triple yCD mutant in these fusions
it is likely an even greater tumor cell killing will be achieved.
The bystander effect is an essential element for gene therapy and without it complete
tumor killing with the current inefficient gene delivery techniques would not be possible. The
CD/5FC enzyme/prodrug combination has been observed to elicit a strong bystander effect
because of the ability of 5FC to passively diffuse from transfected cells to neighboring non-
transfected cells (12,29-32).
In conclusion, the goal of this study was to create yCD variants with enhanced activity
towards the prodrug 5FC for use in biomedical applications. Using regio-specific random
68
mutagenesis we have identified three mutant yCDs with greatly altered sensitivities for 5FC in an
E. coli screening method. The best mutant (D92E) from this group displays an approximate 30%
decrease in IC50 for 5FC when compared to wild-type yCD. In addition, two previously
characterized thermostable yCD variants were demonstrated to increase tumor cell killing in cells
treated with 5FC up to 50% compared to wild-type yCD. Moreover, we have engineered a
mutant yCD enzyme by superimposing the D92E substitution identified in the regio-specific
random mutagenesis library with the substitutions in the thermostable triple mutant with a
substantial increase in thermal stability, approximately 6ºC and 16ºC higher than the Tm observed
with the triple and wild-type yCD enzymes, respectively. We believe the D92E, double and
triple mutants display tumor cell killing that makes them strong candidates for use in future gene
therapy studies.
69
CHAPTER FIVE
Directed evolution and selection of deoxycytidine kinase mutants that induce
hypersensitivity to gemcitabine1
ABSTRACT
There are several limitations to suicide gene therapy, specifically current gene delivery
systems as well as the specificity of targeting tumor cells. One approach to circumvent these
limitations is to create mutant suicide enzymes with enhanced activity towards the prodrugs.
Specifically, our goal was to create mutant human deoxycytidine kinase (dCK) enzymes with
enhanced activity, by regio-specific random mutagenesis, towards the anti-neoplastic nucleoside
analog gemcitabine (dFdC). Over 19,000 transformants were surveyed for enhanced prodrug
specificity in an E. coli screening system. Approximately 335 mutants display super-sensitivity
towards gemcitabine at the lowest concentration of prodrug examined. The sequences of
variants that display increased activity towards gemcitabine reveal particular residues that appear
to interact with substrate. The altered activity of the super-sensitive mutants was then verified by
a semi-quantitative enzyme assay using bacterial cell lysates. From these data eight super-
sensitive mutants with a 35,000-fold increased activity to dFdC compared to wild-type dCK were
chosen for further evaluation. Determination of the alterations in the kinetic parameters of the
best mutant enzymes was assessed with two substrates, deoxycytidine and dFdC. In vitro
evaluation of eight super-sensitive dCK mutants for improved cancer cell killing in two cell lines
shows no improved activity for the mutant dCK enzymes compared to wild-type. Current
attempts to elucidate the cause for the observed activities in the tumor cell models include using
1 Data in this chapter will be submitted for publication in the following manuscript: Willmon, C.L., Sussman, D., Stoddard, B.L., and Black, M.E. (2006) Directed evolution and selection of deoxycytidine kinase mutants that induce hypersensitivity to gemcitabine.
70
a dCK-deficient cell line to further evaluate the enzymes. Despite no observed enhanced activity
in the tumor cell lines, data obtained from this mutagenesis study will reveal important structural
and functional information regarding catalysis, structural integrity, and substrate specificity that
will provide insight for further enzyme improvements and novel anti-cancer drug design.
INTRODUCTION
Human deoxycytidine kinase (dCK) is an important enzyme in the salvage pathway of
deoxyribonucleosides. As one of only four human salvage pathway kinases, dCK can
phosphorylate the purine nucleosides deoxyadenosine (dA) and deoxyguanosine (dG), but has
the greatest activity for the pyrimidine nucleoside deoxycytidine (dC) (141). In addition, this
enzyme is responsible for the initial phosphorylation of several anti-neoplastic agents, such as
gemcitabine (dFdC or 2', 2'-difluorodeoxycytidine), cytosine arabinoside (araC or 1-β-D-
arabinofuranosylcytidine) and fludarabine (FaraA or 9-β-D-arabinofuranosyl-2-fluoroadenine-5'-
monophosphate). Further phosphorylation of these drugs is carried out by other cellular
enzymes. The final phosphorylated products of these drugs are incorporated into replicating
DNA and terminate DNA chain synthesis (44,45,142). An additional aspect of dFdC and
FaraA’s mechanisms of action is through the irreversible inhibition of ribonucleotide reductase
(RNR) (46,47,143) leading to the depletion of dNTP pools. In the absence of dNTP substrates
for DNA polymerase, DNA synthesis is stalled.
Suicide gene therapy, unlike chemotherapy, offers a localized and selective cancer
treatment. This is accomplished through the introduction of a gene encoding an enzyme into
cancer cells. Upon administration of the prodrug, the enzyme converts the prodrug selectively
into its toxic form, thereby resulting in the elimination of cancer cells. However, there are
71
several limitations of current gene delivery systems as well as a lack of specificity to target
tumor cells. One approach to circumvent these limitations is to create mutant enzymes with
enhanced activity towards the prodrugs.
The enzymes used in suicide gene therapy have not naturally evolved to utilize their
prodrugs as substrates. Deoxycytidine kinase displays poor kinetics, overall. This enzyme has a
slow turnover rate (< 1 s-1) (144) for dC and in comparison to other deoxynucleoside kinases
such as deoxyguanosine kinase (dGK) and D. melanogaster deoxyribonucleoside kinase (dNK)
its activity is markedly reduced (145,146). While dCK’s activity has the potential to be
enhanced for its substrates for gene therapy purposes, relatively little mutagenesis to improve the
activity of dCK has been done. Specifically, only one mutant enzyme (A100V/R104M/D133A)
has been created based on molecular modeling studies to improve the catalytic efficiency of dCK
for any prodrug (53) and it displays a 4-fold increase in kcat/Km compared to wild-type dCK for
gemcitabine. However, the relative specificity of this enzyme, a measurement of the preference
for this enzyme to utilize the prodrug in a milieu of nucleosides within the cell, there is no
increase, but rather a decrease of nearly 30%. This suggests the mutant enzyme would not
favorably phosphorylate gemcitabine in cells and would preferentially use deoxycytidine,
although in vitro studies in tumor cells have not been performed to establish this.
Our goal for this study was to utilize regio-specific random mutagenesis to improve
gemcitabine activation by introducing substitutions in the 124-141 a.a. region of dCK (Figure
5.1). Alignment studies with other pyrimidine kinases have provided some insight to the
location of a putative nucleoside binding site in dCK (147,148). An additional alignment of
human dCK with Herpesviridae thymidine kinases and thymidylate kinases from various sources
reveals additional conserved motifs. Two of these motifs (sites 3 and 4) in the herpes simplex
72
virus thymidine kinase (TK) (149) have been shown by mutagenesis and structural studies to be
involved in substrate binding (69,103,150). The 124-141 a.a. region of dCK targeted for
mutagenesis is in and around the region corresponding to sites 3 and 4 of TK. Furthermore, a
crystal structure study of dCK (53) confirms that several target residues lie within the active site
and form a loop structure or are involved in substrate binding (Figure 5.1).
Figure 5.1. Human dCK amino acid sequence (in single-letter code) spanning residues 120-145. The amino acid region targeted for regio-specific random mutagenesis is highlighted in blue. A loop structure found in this region has been boxed. Starred residues interact with substrate (53), while highly conserved residues are underlined and were not targeted for mutagenesis.
From the regio-specific random library eight mutant enzymes with 35,000-fold increased
activity in an E. coli based screening were chosen for further evaluation, including biochemical
characterization and improved tumor cell killing. Importantly, the creation of improved dCK
variants for gene therapeutic applications will (i) allow the use of low prodrug doses, (ii) reduce
toxic side effects, and (iii) enhance the bystander effect and tumor ablation. Additionally,
because dCK is of human origin multiple administrations may be possible. Furthermore, this
research is significant because it will provide an exciting, novel alternative for gene therapy
applications for a wide variety of cancers.
MATERIALS AND METHODS
Materials
Oligonucleotides used to mutate and sequence dCK were obtained from either Sigma-
Proligo (St. Louis, MO) or Integrated DNA Technologies (Coralville, IA). Restriction enzymes
E K P V L F F E R S V Y S D R Y I F A S N L Y E S E 124 141
73
used to construct and screen the dCK mutants were obtained from New England Biolabs
(Beverly, MA). The dCK gene was a kind gift from Dr. Earnest Johnson II (Chapel Hill, North
Carolina). Wizard PCR preps from Promega (Madison, WI), HiSpeed Plasmid Mini Kit from
Qiagen (Valencia, CA), and StrataPrep EF Plasmid Midikit from Stratagene (LaJolla, CA) were
used to purify plasmid DNA. Alamar Blue was purchased from Serotec Limited (Oxford, UK).
All cell culture reagents were purchased from Gibco (Carlsbad, CA). All other reagents were
purchased from Sigma (St. Louis, MO) unless otherwise noted.
Bacterial strains
E. coli strain CJ236 (F+LAM-, ung-1, relA1, dut-1, spoT1, thi-1) was used to produce
single-stranded DNA for site-directed mutagenesis procedures. The E. coli strain NM522 (F+
lacIq∆(lacZ)-M15proA+B+/supE thi∆(lac-proAB)∆(hsdMS-mcrB)5(rk-mk
-McrBC-)) was used as a
recipient for certain cloning procedures. E. coli BL21(DE3) (F-ompT[lon] hsdSb (rB- mB
-) gal
dcm met (DE3)) or BL21(DE3)tdk- were used for screening purpose and expression of wild-type
and mutant dCK enzymes.
Cell lines
All cell lines were maintained in a humidified incubator at 37ºC in 5% CO2. The human
uterine sarcoma cell line, MESSA 10K, was a kind gift from Drs. Lars Petter Jordheim and
Charles C. Dumontet (INSERM, Lyon, France) (151). The Jurkat T cell leukemia line, was a
kind gift from Dr. Oliver Press (Fred Hutchinson Cancer Research Center, Seattle, WA). The
MESSA 10K and Jurkat cells were grown in RPMI 1640 media containing 25mM HEPES, 100
U/ml penicillin G, 10 µg/ml streptomycin sulfate, 292 µg/ml L-glutamine and fetal bovine serum
74
(10%). Human colorectal carcinoma cell line, HCT116, was a considerate gift from Dr. Neal
Davies (Washington State University, Pullman, WA) and were maintained in McCoy’s 5A
medium (Sigma, St. Louis, MO) containing 25 mM HEPES, 100 U/ml penicillin G, 10 µg/ml
streptomycin sulfate, 292 µg/ml L-glutamine and 10% fetal bovine serum. The Rat C6 glioma
cells were purchased from ATCC (Manasass, VA) and were maintained in Dulbecco’s Modified
Essential Medium containing 5% fetal bovine serum, 1mM sodium pyruvate, 10 mM HEPES,
100 µM nonessential amino acids, 100 U/ml penicillin G, 10 µg/ml streptomycin sulfate, 292
µg/ml L-glutamine, 100 µM sodium citrate and 0.0014% NaCl. The WiDr cells, a human
colorectal adenocarcinoma cell lines, was a generous gift from Dr. Steven Albelda (University of
Pennsylvania, Philadelphia, PA), and maintained in Minimum Essential Medium containing
Earle's salts, 10% fetal bovine serum, 1mM sodium pyruvate, 10 mM HEPES, 100 µM
nonessential amino acids, 100 U/ml penicillin G, 10 µg/ml streptomycin sulfate, 292 µg/ml L-
glutamine, 100 µM sodium citrate and 0.0014% NaCl. Transfected cells were cultured in media
supplemented with blasticidin at 4 µg/mL (C6 glioma) and 6 µg/mL (HCT116).
Construction of the dCK regio-specific random library
Construction of the inactivated dCK vector: The use of a non-functional or dummy vector as the
recipient of randomized fragments eliminates the possibility of contamination by wild-type dCK
and prevent the emergency of false-positives in the selection system. To create the dummy
vector (pETHT:dCK dummy), pETHT:dCK (wild-type) was mutated by the Kunkel method of
site-directed mutagenesis (101,102) to simultaneously introduce a KpnI restriction site
(underlined nucleotides) for ease of mutant identification and a stop codon (bolded nucleotides)
within the targeted region of the dCK gene using the following oligonucleotide, MB 339: 29mer,
75
5’ CAAATTAGATGCTTAAAGGTACCTGTCAC 3’. After restriction enzyme verification,
DNA sequence analysis was performed to confirm the presence of the desired mutations within
pETHT:dCK dummy at the core sequencing laboratory at Washington State University.
Preparation of the 124-141 a.a. random DNA: Random mutagenesis at the 124-141 a.a. region
of dCK was performed by using a modified version of megaprimer polymerase chain reaction of
whole plasmid as described in Miyazaki and Takenouchi (152). The following oligonucleotides
were designed for amplification of the target region, including ten codons that were randomized
at 12% as indicated in bold in the MB 345 oligonucleotide:
MB 345 (84mer), 5’ GCAGAGAAACCTGTATTATTTTTTGAACGATCTGTGTATAGT
GACAGGTATATTTTTGCATCTAATTTGTATGAATCTGAATGC 3’
MB 371 (61 mer), 5’ CCAGTCATGCCCGTCTTGATAAATTGTCCACTCTGTCTCATTCAT
GCATTCAGATTCATAC 3’. Briefly, regio-specific random mutants were created by first
performing a PCR reaction carried out in 20 mM Tri-HCl (pH = 8.8), 10 mM KCl, 10 mM
(NH4)SO4, 2 mM MgSO4, 0.1 % Triton X-100, 0.1 mg/mL BSA, 0.2 mM each dNTP, 50 ng of
purified pETHT:dCK dummy DNA, 125 ng MB 345, 125 ng of MB 371, 5 µL of 10 x reaction
buffer and 2.5 U PfuTurbo DNA Polymerase (Stratagene, LaJolla, CA) in total volume of 50 µL.
The entire reaction was subjected to cycling in an Eppendorf Cycler at 95°C for 1 min, followed
by 30 cycles of 95°C for 1 min, 60°C for 1 min and 68°C for 8 min. After cycling, the mixture
was treated with DpnI (1 uL, 20 U) and incubated at 37°C for 1 hr.
Screening for gemcitabine sensitivity: Approximately 2 µL of the DpnI-treated reaction mixture
was electroporated into 40 µL of electrocompetent E. coli BL21(DE3) cells and then shaken at
37°C for 1 hr in 1 mL of SOC medium (3 g Bactopeptone, 2.5 g yeast extract, 0.5 g NaCl, 2.5
mM KCl, 5 mM MgSO4, 5 mM MgCl2 and 1.8% glucose per liter) and plated onto non-selective
76
medium. The transformants produced were surveyed for altered prodrug specificity in E. coli
BL21(DE3) on plates supplemented with gemcitabine at a sublethal (0.1 µM) and a lethal (35
µM) dose of gemcitabine. The medium (500 mL), with or without gemcitabine (i.e., non-
selective), was prepared from 0.36 g yeast synthetic dropout without leucine, 50 mL 10 x M9
salts (15 g KH2PO4, 33.9 g anhydrous NaHPO4, 2.5 g NaCl, 5.0 g NH4Cl), 1 mM MgSO4, 2.5
mL, 20% glucose, 0.1 mM CaCl2, and 7.5 g Bactoagar and 50 µg/mL carbenicillin. Using single
colonies from non-selective plates, 96 well microtiter plates containing 200 µL per well of non-
selective medium (minus Bactoagar) were inoculated and incubated at 37°C for 16 hr. Next, the
colonies were replica plated onto control, sublethal, and lethal gemcitabine-containing plates and
colony size and formation were scored. Colonies that grew at a slower rate or were smaller than
wild-type dCK on the plates containing gemcitabine were classified as “super-sensitive” and re-
tested on plates containing the lethal, sublethal or lower concentrations (5 and 1 nM) of
gemcitabine. Next, plasmid DNA of the dCK variants was purified by Wizard PCR preps, and
the randomized region was sequenced using the T7 terminator primer (5’ TAT GCT AGT TAT
TGC TCA G 3’) at the core sequencing laboratory at WSU.
Determination of dCK activities in bacterial lysates
In order to select the most sensitive mutants in the library, a preliminary filter-binding
assay was performed using crude bacterial lysates. Colonies identified from the lowest
gemcitabine-containing plate (1 nM) were used to inoculate 3 mL of non-selective medium and
grown at 37°C until similar optical densities (OD) at 600 nm were achieved. Verification that
the cultures expressed similar protein levels was performed using SDS-PAGE. Next, a 1.5-ml
aliquot from the cultures was removed and pelleted. The bacterial pellets were resuspended in
77
0.5 mL of buffer A (50 mM Tris-Cl (pH = 7.6), 1 mM EDTA, 50 mM NaCl, 0.5 mM PMSF, 2
mM DTT, 1 mg/mL lysozyme), incubated at room temperature for 30 min and subjected to
centrifugation (14 krpm for 20 min). Immediately following the supernatants (5 µL) were used
for enzyme assays using [5-3H]-gemcitabine (9.7 Ci/mmol) as the substrate (Moravek
Biochemicals, Brea, CA). The assay used to detect the phosphorylation of gemcitabine is a
filter-binding assay modified from the previously described assay by Usova and Eriksson (153).
Briefly, in triplicate, 3 µL of the reaction mixture (50 mM Tris-CL (pH = 7.4), 10 mM MgCl2, 5
mM ATP (pH = 7), 2 mM DTT, 0.5 mg/mL BSA, and 5 µL of 3H-gemcitabine (in a total of 75
µL) and 5 µL of bacterial cell lysates were incubated for 30 min at 37 °C. The reactions were
terminated by boiling for 1 min, adding 40 µL of ice-cold H2O and then spotting 40 µL of the
reaction mixture onto Whatman DE-81 filter discs. The filter discs were washed three times in 5
mM ammonium formate, twice in 95% Ethanol, allowed to air dry and then added to 5 mL of
scintillate and radioactivity was measured by scintillation counting. Using the recorded counts
per minute (CPM), the individual mutants were compared to the positive control (wild-type
dCK) and negative control (dummy dCK) CPMs to determine the mutants with the greatest
activities. This assay was performed at least three times.
Purification of wild-type and variants dCKs
The selected dCKs were expressed in E. coli BL21(DE3) and purified by Ni-NTA
chromatography as previously described (77). The purified mutants were dialyzed against
dialysis buffer (50 mM Tris-HCl (pH = 7.5), 50 mM NaCl) at 4°C and stored at 4°C. Enzyme
concentrations were determined using a Pierce BCA Protein Assay Kit (Rockford, IL) according
to the manufacturer’s protocol, using BSA as the protein standard.
78
Antiserum production
Purified deoxycytidine kinase was supplied to Harlan Bioproducts (Indianapolis, IN) for
the generation of rabbit polyclonal antiserum according to their standard protocol.
Enzyme kinetics of purified proteins
The deoxycytidine kinase activity assays were performed using varying substrate
concentrations and enzyme concentrations. The assay used to detect the phosphorylation of the
nucleosides (deoxy[3H]cytidine and [5-3H]-gemcitabine) is the filter binding assay described
above. The deoxycytidine radioisotope (deoxy[3H]cytidine, 23 Ci/mmol) was purchased from
Amersham (Buckinghamshire, UK). The resulting data were fit to the Michaelis-Menten
equation to determine the Km and Vmax values of wild-type dCK and mutant constructs. A
conversion factor was determined for each [3H]-nucleoside monophosphate by measuring the
cpms for a known number of moles of each tritiated nucleoside. The kcat values were calculated
using this conversion factor, from the determined Vmax values assuming one active site per
monomer. The catalytic efficiency kcat/Km is calculated from these values.
Gemcitabine sensitivity assays
One µg of each DNA (pCDNA (vector alone), pCDNA:dCK, pCDNA:F125L,
pCDNA:F126L, pCDNA:Y131H; pCDNA:S139A; pCDNA:S139A; pCDNA:S139F;
pCDNA:S139T; pCDNA:L141F; pCDNA:V130I/L141S) was used to transfect 1 x 105 the C6
glioma and HCT116 cells by lipofection using FuGENE 6 transfection reagent (Roche
Diagnostics, Penzberg, Germany) at a various ratios according to the manufacturer’s directions.
Immunoblots were performed to ensure protein levels were equally expressed. Briefly, pools of
79
transfectants were harvested and resuspended in lysis buffer (for 2 mL: 2 µL 1 M DTT, 20 µL 1
M HEPES, 40 µL Nonidet P40 (Roche Diagnostics, Pernzberg, Germany) 2 µL MgAc2, H2O to
final volume) at 100,000 cells/µL. The resuspended cells were incubated on ice for 20 min and
subjected to centrifugation at 4°C for 20 min to pellet debris. Samples (10 µL per well) were run
on a 12% SDS gel, transferred to a nitrocellulose membrane and blocked with 3% gelatin in Tris-
buffered saline. The membrane was probed with rabbit polyclonal dCK antibody followed by
goat anti-rabbit AP-conjugated antibody according to manufacturer’s specifications. The blot
was developed using the AP Conjugate Substrate Kit (Bio-Rad, Hercules, CA). To determine
the cytotoxicity of gemcitabine, pools of transfectants or the parental cell lines were transferred
to 96 well microtiter plates in the corresponding medium plus supplements at an initial density of
500 cells per well for C6 and HCT116, 1000 cells per well for Jurkat and MESSA 10K, and 3000
cells per well for the WiDr cell line (59). After cell adherence overnight, gemcitabine (0-1.25
µM) was added in sets of eight wells for each concentration tested. The plates were incubated
for 4 days at which time the redox indicator dye Alamar Blue was added. Cell survival was
determined by fluorescence recorded at 530/590 nm several hours later as according to the
manufacturer’s instructions and data were plotted with standard deviation bar. At least three
replicates were performed.
RESULTS
Screening and characterization of the 124-141 a.a. regio-specific random library.
E. coli does not express dCK so a simple genetic complementation screening method
could not be established to identify transformants with the ability to phosphorylate
deoxycytidine. However, E. coli harboring a dCK-containing plasmid are sensitive to dFdC and
80
viability is diminished or completely lost while E. coli without a dCK gene or expressing a non-
functional dCK gene are viable. Using this fact, we established a screening method for
surveying mutant dCK enzymes with altered activities towards dFdC. The plates containing
dFdC were at both a sublethal (0.1 µM) and lethal (35 µM) dose of dFdC for wild-type dCK.
The plates containing the lethal dose allow for the selection of any functional dCK because the
dCK-expressing bacteria will not be viable on these plates. The plates at a sublethal
concentration of gemcitabine allow bacteria with wild-type dCK-expressing plasmids to grow,
whereas bacteria that contain dCK mutants with increased activity towards dFdC die.
Furthermore, to prevent false positives from emerging in the screening system and eliminating
the possibility of contamination by the wild-type dCK, we inactivated the parental dCK gene by
introducing a stop codon in the target region of the gene, as described in Materials and Methods.
In order to create mutant dCK enzymes with enhanced activity for dFdC regio-specific
random mutagenesis was employed to target the 124-141 a.a. region (L124, F125, F126L, V129,
Y131, D133, I136, A138, S139, and L140) of dCK (Figure 5.1) using a modified version of
megaprimer polymerase chain reaction of whole plasmid initially described in Miyazaki and
Takenouchi (152). Based upon other studies which have demonstrated that a lower percentage
of randomization will result in a larger number of non-functional variants (137), 12%
randomization was chosen for this library. Furthermore, we eliminated several codons
(underlined in Figure 5.1) from the mutagenesis because they are highly conserved in alignment
studies with other nucleoside kinases (53) and therefore, likely to be essential. In total, 19,834
transformants were screened as described above. Of those transformants, 65% (or 13,109)
displayed sensitivity at the lethal concentration on gemcitabine-containing plates.
Approximately 335 transformants (1.7%) displayed super-sensitivity on plates containing 1 nM
81
gemcitabine, the lowest level of sensitivity in the screening system. In order to refine the
number of super-sensitive variants, the 335 transformants were surveyed for altered enzyme
activity using bacterial cell lysates, as described in Materials and Methods. From these data
eight mutants with the greatest altered activity were chosen to be further characterized by
enzyme kinetics and in tumor cell models.
We evaluated the library, along with which a.a. substitutions were tolerated, by
performing DNA sequence analysis of non-selected, gemcitabine-sensitive, and non-sensitive
transformants. The broad spectrum of mutations observed in the regio-specific library indicates
that the oligonucleotides were random at the designated codons. Furthermore, a more restricted
range of substitutions were observed in the 335 “super-sensitive” mutants (data not shown).
Sequence analysis revealed the introduction of single and multiple mutations and trends in types
of amino acid substitutions for the super-sensitive mutants. For example, leucine 124 appears to
tolerate only amino acid substitutions that also have nonpolar, hydrophobic side chains.
Phenylalanine 125 has substitutions that are restricted to very similar nonpolar, hydrophobic side
chains, as well. Aspartic acid 133 was found to only accept glutamic acid substitutions;
otherwise, any other D133 variants were insensitive to gemcitabine in the screening system. The
eight variants with the greatest increase in activities towards gemcitabine in the screening system
and the bacterial crude lysates assays were identified as: F125L, F126L, Y131H, S139A, S139F,
S139T, L141F and V130I/L141S.
Determination of the kinetic parameters of wild-type and mutant dCK proteins
To establish the effects of the substitutions found in the super-sensitive mutants on
substrate specificity, Michaelis-Menten kinetic constants for dC and dFdC were determined and
82
are summarized in Table 5.1. First, enzymes were purified by nickel affinity chromatography to
near homogeneity. Next, purified wild-type and mutant dCK enzymes were characterized using
a filter binding assay previously described by Usova and Eriksson (153). The optimal substrate
concentrations and enzyme amounts were determined for each substrate. The kinetic values for
wild-type enzyme were found to be similar to previously reported values (Table 5.1)
(46,153,154).
Table 5.1. Kinetic parameters of wild-type and mutant dCK enzymes.
Mutants with impaired deoxycytidine activity (F125L, S139A and S139F)
With dC as the substrate, F125L and S139F display 2.3 and 1.4-fold increases,
respectively, in Km compared to wild-type yCD (Table 5.1). However, the turnover number
(kcat) for both of these mutant enzymes, compared to wild-type dCK, was decreased by 30% and
dCK F125L F126L Y131H S139A
Deoxycytidine
Km (µM) 1.71 ± 0.25 3.86 ± 0.17 2.79 ± 0.17 1.49 ± 0.11 1.42 ± 0.11
kcat (s-1) 0.010 ± 0.0008 0.0047 ± 0.0004 0.013 ± 0.0007 0.0102 ± 0.0002 0.0048 ± 0.0002
kcat (s-1)/Km (M) 5847.95 1217.62 4659.50 6845.64 3380.28
Gemcitabine
Km (µM) 5.83 ± 0.89 2.13 ± 0.23 5.43 ± 1.19 6.1 ± 1.5 5.34 ± 2.23
kcat (s-1) 0.10 ± 0.03 0.028 ± 0.001 0.065 ± 0.006 0.082 ± 0.015 0.067 ± 0.023
kcat (s-1)/Km (M) 17152.66 13145.54 11970.53 13442.62 12546.82
Relative Specificitya 0.75 0.92 0.72 0.66 0.79
S139F S139T L141F V130I/L141S
Deoxycytidine
Km (µM) 2.36 ± 0.26 1.55 ± 0.12 0.39 ± 0.07 1.81 ± 0.47
kcat (s-1) 0.0044 ± 0.0004 0.017 ± 0.003 0.00601 ± 0.00146 0.0268 ± 0.0149
kcat (s-1)/Km (M) 1864.41 10967.74 15410.26 14806.63
Gemcitabine
Km (µM) 1.44 ± 0.06 3.00 ± 0.34 4.39 ± 1.52 0.77 ± 0.04
kcat (s-1) 0.0114 ± 0.0009 0.054 ± 0.008 0.215 ± 0.062 0.016 ± 0.003
kcat (s-1)/Km (M) 7916.67 18000.00 48974.94 20779.22
Relative Specificity 0.81 0.62 0.76 0.58
a Relative specificity = kcat/Km (dFdC)/(kcat/Km (dFdC) + kcat/Km (dC)).
83
56%, respectively. Therefore, the overall catalytic efficiency (kcat/Km) of these enzymes was
greatly reduced from the wild-type by 81% for F125L and 68% for S139F (Figure 5.2A).
Compared to wild-type dCK, the Km for S139A is slightly decreased (17%) for deoxycytidine,
and its kcat is markedly reduced by 52%, decreasing the kcat/Km value by 43%.
Figure 5.2. Comparison of (A.) catalytic efficiencies (kcat/Km) and (B.) the relative specificity differences of the mutant enzymes compared to wild-type dCK. Using the equation, [kcat/Km (dFdC)/(kcat/Km (dFdC) + kcat/Km (dC))], the relative specificities were calculated. The data (B.) are plotted as the fold differences from wild-type enzyme.
Mutants with similar kinetics for deoxycytidine (F126L and Y131H)
The F126L and Y131H display relatively similar kinetics for deoxycytidine compared to
wild-type enzyme. Despite a slightly improved affinity (Km) and turnover number of F126L for
dC with a 1.6-fold increase in Km and a 1.3-fold increase in kcat than wild-type enzyme, the
overall efficiency of this enzyme is about 20% reduced from dCK (Figure 5.2A). Y131H has no
change in its kcat value, compared to wild-type enzyme, but has a slight decrease (13%) in Km.
Mutants with improved deoxycytidine activity (S139T, L141F and V130I/L141S)
With dC, S139T (Km = 1.55 µM) and V130I/L141S (Km = 1.81 µM) display very similar
0
5
10
15
20
25
45
50
dCK F125L F126L Y131H S139A S139F S139T L141FV130I/L141S
kcat/Km (dC)kcat/Km (dFdC)
s-1
/M
0
0.2
0.4
0.6
0.8
1
1.2
1.4
dCK F125L F126L Y131H S139A S139F S139T L141FV130I/L141S
Fo
ld D
iffe
ren
ce
A. B.
84
Km values to wild-type dCK (Km = 1.71 µM). However, these mutants display kcat values which
are 1.7- and 2.7-fold greater than wild-type, respectively. Therefore, the overall catalytic
efficiencies of these enzymes were increased 2-fold for S139T and 2.7-fold for V130I/L141S, in
comparison to wild-type dCK (Figure 5.2A). L141F has a considerably reduced Km for dC than
wild-type, although its turnover number was reduced by 40% from the wild-type value. Overall,
the enzyme displays an enhanced catalytic efficiency (2.6-fold increase) compared to wild-type
enzyme (Figure 5.2A).
Mutant enzymes with impaired or similar gemcitabine activity (F125L, F126L, Y131H, S139A
S139F and S139T)
The majority of the dCK mutant enzymes identified were determined to have very similar
Km values for dFdC compared to wild-type (Table 5.1). The one exception was the Km for
S139T was found to be 49% decreased from wild-type enzyme. While the Km values of these
enzymes were relatively unaffected, the kcat values were significantly impacted. For example, the
kcat for F125L and S139F are reduced by 72% and 89%, respectively, compared to wild-type
dCK. Overall, the F125L, F126L, Y131H, S139A and S139F were found to be less catalytically
efficient enzymes for dFdC than wild-type for this substrate. The greatest decrease in efficiency
was observed with the S139F mutant enzyme which has a 54% decrease in kcat/Km when
compared to wild-type enzyme (Figure 5.2A). However, S139T has a slightly improved
efficiency for dFdC compared to wild-type enzyme because of its improved Km.
Mutants with improved gemcitabine activity (L141F and V130I/L141S)
The V130I/L141S mutant enzyme was found to have the best ability to bind dFdC (Km =
0.77 µM), an 87% decrease from wild-type dCK (Table 5.1). However, in comparison to wild-
type, the turnover number of this enzyme was decreased by 84%. Yet, because of the reduced
85
Km, the overall catalytic efficiency was only modestly increased by 1.2-fold, compared to wild-
type dCK (Figure 5.2A). The L141F has a similar Km value for dFdC as wild-type, but the kcat
was increased about 2-fold. Therefore, this enzyme displays the greatest increase from wild-type
in relative specificity for all of the mutants, with an approximate 3-fold increase (Figure 5.2B).
In vitro gemcitabine sensitivity assays
To determine the ability of the mutants to sensitize cells to dFdC in vitro, rat C6 glioma
and HCT116 human colorectal carcinoma cells were transfected with mammalian expression
vectors encoding the wild-type and eight mutant dCKs. Immunoblot analyses of lysates from the
pools of transfectants show similar expression levels for all of the mutants and wild-type dCK,
but an increased expression in comparison to vector control (data not shown). Using drug
concentrations between 5-1250 nM, the cells were assayed for their level of gemcitabine
sensitivity. Overall, no considerable enhancement in sensitivity was observed between the
mutant enzymes and wild-type dCK in either cell line (Figure 5.3) and several mutant enzymes
were found to have decreased sensitivities. However, because these cells express endogenous
dCK, the cells transfected with vector alone have an inherent cytotoxicity upon gemcitabine
treatment. In comparison to the C6 glioma cells, where an approximate 2-fold improvement of
IC50 for dFdC was achieved in dCK-transfected cells over control cells (vector alone) (Figure
5.3A), there is no detectable difference in IC50 for dFdC between vector or dCK-expressing
HCT116 cells (Figure 5.3B).
86
Figure 5.3. Gemcitabine sensitivity of deoxycytidine kinase (dCK)-transfected (A.) C6 glioma and (B.) HCT116 cells. Pools of stable transfectants containing vector alone, wild-type dCK and the eight regio-specific random mutants were constructed in rat C6 glioma or HCT116 cells and evaluated for gemcitabine sensitivity as described in Materials and Methods. After four days of gemcitabine treatment, the growth inhibition was determined by staining with Alamar Blue and fluorescence recorded at 530/590 nm. Each data point (mean ± SEM, n=3 performed with at least fifteen replicates) is expressed as a percentage of the value for control wells with no gemcitabine treatment.
0
10
20
30
40
50
60
70
80
90
100
110
120
-50 0 50 100 150 200 250 300 350 600 700 800 900 1000 1100 1200 1300
pCDNAdCKF125LF126LY131HS139A
% S
urv
ival
[dFdC] (nM)
0
10
20
30
40
50
60
70
80
90
100
110
120
-50 0 50 100 150 200 250 300 350 600 700 800 900 1000 1100 1200 1300
pCDNAdCKS139FS139TL141FV13OI/L141S
% S
urv
ival
[dFdC] (nM)
0
10
20
30
40
50
60
70
80
90
100
110
120
-50 0 50 100 150 200 250 300 350
pCDNAdCKF125LF126LY131HS139A
% S
urv
ival
[dFdC] (nM)
0
10
20
30
40
50
60
70
80
90
100
110
120
-50 0 50 100 150 200 250 300 350
pCDNAdCKS139FS139TL141FV13OI/L141S
% S
urv
ival
[dFdC] (nM)
A.
B.
87
DISCUSSION
Although dCK is present in different tissues (155), many tumors have low levels of dCK,
indicating that gene therapy with dCK may be an appropriate means to sensitize tumors to
cytotoxic compounds (156). Whereas the concept to use dCK in gene therapy for treatment of
solid tumors with gemcitabine or araC has been discussed for a number of years, supporting
evidence for the use of dCK as a suicide gene has only been reported recently (44,48-53). Gene
therapy studies using dCK resulted in 50-400 fold reduction in IC50 in vitro in dCK-deficient
CHO cells with dFdC or araC as compared to untransfected cells (51). Manome et al. (44)
observed significant anti-tumor effects in a 9L glioma model in rats and demonstrated a
statistically significant survival time compared to that of non-dCK transfected tumor-bearing
animals.
Furthermore, Sanda et al. (50) found that the presence of 25% dCK-expressing cells
caused a significant bystander effect (~80-90%) in mixed culture. Unlike TK and cytosine
deaminase, two extensively studied suicide enzymes, dCK is of human origin. Whereas the
sequential administration of herpes TK/ganciclovir (GCV) or cytosine deaminase (CD)/5FC is
presumably limited due to rapid clearing by the immune system, dCK provides an opportunity to
perform multiple administrations because of its human origin, which may be important for
complete tumor ablation. This is a substantial advantage of the dCK/dFdC system over the
TK/GCV or CD/5FC systems.
While the aforementioned studies have demonstrated that dCK can be utilized for gene
therapy applications, these attempts have only been partially successful. One particular problem
is that dCK is a poor catalyst for its substrates. Additionally, because most normal cells have a
low level of endogenous dCK expression, the dose of dFdC must be administered at a level that
88
is not toxic to normal cells. Therefore, we believe by improving binding and, especially, the
turnover of this enzyme through mutagenesis a mutant enzyme might demonstrate enhanced
tumor cell killing.
Figure 5.4. Human deoxycytidine kinase structure with bound deoxycytidine and ADP (blue) and positions of residues mutated in a screen for increased sensitivity to gemcitabine (magenta side chains). The majority of mutated positions (F125, F126, V130, Y131, and S139) involve positions that are involved in core protein fold packing around the active site cleft, and display only small effects on the kinetic behavior of the enzyme towards deoxycytidine or gemcitabine nucleosides. The single position that is in contact with substrate (L141) enhances the affinity of the enzyme for both substrates, with a marked shift towards the gemcitabine moiety.
In order to create mutant dCK enzymes we utilized regio-specific random mutagenesis to
introduce substitutions in the 124-141 a.a. region of dCK. Seven singly substituted and one
doubly substituted enzymes were identified to have a 35,000-fold increased activity for dFdC
killing in E. coli compared to wild-type. Overall, three of the mutant dCKs (S139T, L141F and
V130I/L141S) display considerable increases from wild-type in their efficiency for using dC as a
substrate (Table 5.1). Mutants L141F and V130I/L141S also demonstrate improved catalytic
efficiency with dFdC compared to wild-type, with nearly a 3-fold increase by L141F.
89
In addition, we modeled the substitutions onto the dCK crystal structure (Figure 5.4).
Interestingly, the substitutions of the mutant enzymes are largely located within hydrophobic
core pockets surrounding the enzyme active site, rather than at positions that interact directly
with bound substrates. Only one mutated position (L141) is within contact distance to the
deoxycytidine substrate. Interestingly, the mutation at this residue (L141F) is the only
substitution from the screen that demonstrates an increase in kcat for dFdC. Given that
stabilization of yCD (90) has produced a robust improvement in its activity in similar assays with
5FC, we hypothesize that this screen has identified a series of mutations within the dCK scaffold
that affect either enzyme stability or its global dynamic flexibility, with one independent
mutation (L141F) acting directly on substrate turnover.
In the tumor cell killing models, we unfortunately did not see large dFdC sensitivity
differences between the mutant enzymes and wild-type dCK-expressing cells. Moreover, with
the HCT116 colorectal carcinoma cell line we were unable to demonstrate a difference in
sensitivity to dFdC between the vector control and dCK-transfected cells. We speculate there is
a threshold of sensitization to dFdC based upon dCK activity in certain types of cells and that by
increasing the dCK activity no further difference will be observed. Furthermore, the results in
the HCT116 are not an unusual observation. Other groups have reported that upon prodrug
treatment no increase in growth inhibition was observed in cell lines with increased dCK activity
(29,157). In those studies little to no increase in cytotoxicity was achieved when cells were
transduced with dCK and treated with dFdC.
Additionally, unlike several of the other dCK gene therapy studies which have used cells
lines with little or no expression of dCK (44,49,50), we attempted to evaluate the mutant
enzymes in cell lines with endogenous dCK activity. Similar to our findings of an approximately
90
Figure 5.5. Gemcitabine sensitivity of various cancer cell lines. Several different cancer cell lines were evaluated for dFdC sensitivity as described in Materials and Methods. After four days of gemcitabine treatment, the growth inhibition was determined by staining with Alamar Blue and fluorescence recorded at 530/590 nm. Each data point (mean ± SEM, performed with at least eight replicates) is expressed as a percentage of the value for control wells with no gemcitabine treatment.
50% decrease in IC50 for dFdC in dCK-transfected cells from vector-transfected cells in the C6
glioma model, Luminczky et al. (157) observed that increased dCK expression could chemo-
sensitize C6 glioma cells to dFdC. In order to elucidate if we are indeed achieving an upper limit
of toxicity in these cell lines and, therefore, are not able to demonstrate improved cellular killing
with the dCK mutants, they will be evaluated in a cell line with little to no endogenous dCK
expression. Preliminary evaluation of several other cancer cell lines (Jurkat, MESSA 10K, and
WiDr) for their intrinsic cytotoxicity upon dFdC treatment has revealed these cell lines exhibit
similar sensitivity to dFdC as the C6 glioma and HCT116 cells, except for the dCK-deficient
MESSA 10K cells, which display little sensitivity with dFdC (Figure 5.5). We believe future
experiments, which include transfecting MESSA 10K cells with the super-sensitive mutants and
0
10
20
30
40
50
60
70
80
90
100
110
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300
C6HCT 116JurkatMESSA 10KWiDr
% S
urv
ival
[dFdC] (nM)
91
testing for dFdC sensitivity, would be valuable to understand more fully the impact of the
substitutions in the mutant enzymes and their activities.
In summary, this is the first report of a directed evolution strategy aimed at identifying
mutant dCK enzymes with enhanced sensitivity to any prodrug for gene therapy purposes. For
this study, we have designed and implemented an E. coli based screening method for identifying
mutant dCK enzymes with enhanced sensitivity to dFdC. Using this screening method, a regio-
specific random library of over 19,000 transformants was surveyed for increased activity and
eight mutant enzymes with 35,000-fold increased activity were isolated. The biochemical
characterization of these mutants revealed one particular mutant (L141F) with a 3-fold enhanced
catalytic efficiency for dFdC. Further characterization to fully establish the achievable level of
chemo-sensitivity upon expression of the mutant enzymes in tumor cell lines needs to be
performed. We believe the creation of mutant dCK enzymes with enhanced sensitivity for dFdC
will (i) allow the use of low prodrug doses, (ii) reduce toxic side effects, and (iii) enhance the
bystander effect and tumor ablation in future gene therapy applications. Moreover, the use of
dCK in gene therapy applications should allow for multiple administrations because of its human
origin, which may be necessary for complete tumor ablation.
92
CHAPTER SIX
Summary
Although tremendous hurdles in regards to the feasibility and therapeutic response of
gene therapy have been overcome in the past two decades, further research to improve suicide
gene therapy must be performed before gene therapy becomes a mainstream therapeutic
approach for treating cancer. One difficult problem in gene therapy has been achieving a
delivery method that is specific and efficient, but not immunogenic. Obviously the task of
perfectly targeting every cancerous cell in a complex system is quite an undertaking, and more
than likely will never be fully achieved. However, one approach to bypassing the current
inefficient delivery methods is to improve the suicide enzyme itself; and, this is the focus of the
work in the preceding chapters. Suicide enzymes can be engineered to be better catalysts for
their prodrugs, and consequently in response, several other limiting factors in suicide gene
therapy could then be overcome. The first is administration of lower prodrug doses could result
in fewer unwanted side effects for the patients. Second, if more of the prodrug is subsequently
activated because of enhanced catalytic efficiency of the enzyme, more tumor cells should be
eradicated. Additionally, with a greater extent of drug being activated, we should see expanded
bystander effects, leading to an increased degree of tumor cell elimination.
In addition to overcoming some of the current limitations with suicide gene therapy, it is
also necessary to realize that having a repertoire of only a few of enzyme/prodrug combinations
may be impractical for the treatment of cancer. It is well demonstrated patients easily develop
resistance or tolerance to chemotherapeutic treatments. Additionally, because many of the
currently studied enzymes are exogenous, multiple administrations are not feasible due to
activation of the immune response. In order to address this, new enzyme/prodrug combinations
93
are being developed. In particular, one promising approach would be to use endogenous
enzymes, such as human deoxycytidine kinase (dCK) discussed in Chapter 5.
Overall, several exceptional mutant enzymes have been generated and are presented in
this work. As discussed in Chapter 2, herpes simplex virus thymidine kinase (TK) is the most
extensively studied suicide enzyme. This enzyme is capable of phosphorylating several
clinically relevant prodrugs; however, one particular prodrug, acyclovir (ACV), has not been
readily used because of inefficient phoshporylation by TK. Even with high prodrug doses, the
enzyme so poorly utilizes the prodrug that no therapeutic effect is achieved. Using the
information obtained from two extensively characterized libraries (58,60,104), we created a
mutant enzyme (A168F) with 763-fold increased substrate preference towards ACV compared to
wild-type TK, and a 3.3-fold increase in comparison to the best mutant TK engineered to date
(SR39) (104).
An additional aspect of engineering enzymes is to ensure activation of the product does
not lead to an accumulation of ineffective intermediate compounds because of the low activity of
other enzymes responsible for complete activation of the drug. For example, with TK, ultimate
activation of ACV or GCV is dependent upon the second and third phosphorylations by
endogenous enzymes. In particular, it has been demonstrated that guanylate kinase (gmk) does
not efficiency phosphorylate GCV-MP to GCV-DP (120,124). Using pathway engineering, we
combined the TK and GMK activities in a fusion protein and were able to achieve significantly
improved reductions (~175-fold) in the IC50 for GCV in tumor cell killing (Chapter 3). Future
work, aimed at further improving this fusion protein, particularly the GMK portion of the
protein, might provide even more dramatic results.
94
Discovering or enhancing enzymes from different sources but with similar functions
towards already clinically approved prodrugs could also lead to increasing the repertoire of
suicide enzyme/prodrug combinations. For example, both bacterial (bCD) and yeast cytosine
deaminases (bCD) are able to activate the anti-fungal agent, 5-fluorocytosine (5FC) to 5-
fluorouracil (5FU), a widely studied and used chemotherapeutic drug. Historically bCD has
been used in gene therapy applications because of its thermostability and reasonable activity
towards 5FC. However, yCD displays better kinetics with 5FC, but has been demonstrated to be
less thermostable than bCD (37). We have previously created two thermostable mutants of yCD
(double and triple yCD) (90) and also demonstrated these mutant enzymes display improved
tumor cell killing (Chapter 4). In addition to these yCD variants, directed evolution of yCD was
performed and three resulting mutants were identified to confer increased 5FC sensitivity in E.
coli, as well as to tumor cells, compared to wild-type yCD (Chapter 4). Furthermore, we
superimposed the substitutions of the 5FC-sensitizing mutants from the regio-specific library
onto the thermostable triple mutant and have created a variant with even greater thermostability,
with 6ºC and 16ºC increases in apparent melting temperature compared to the triple and wild-
type yCD, respectively. As with the mGMK/TK fusion protein, pathway engineering to combine
these exceptional yCD mutants with an enhanced uracil phosphoribosyl transferase (UPRT)
enzyme may provide even greater cytotoxicity and broader bystander effects (80,89). Or, as has
been previously demonstrated with the wild-type enzymes (81,140), the fusion of a yCD variant
with a TK variant (SR39) could lead to a double mutant enzyme capable of using two drugs
synergistically to achieve a dramatically superior enzyme/prodrug combination. This mutant is
currently in phase I clinical trials for prostate cancer.
95
Lastly, as mentioned above, the use of endogenous enzymes, such as dCK, for gene
therapy could provide a substantial improvement to the use of foreign enzymes because
sequential administrations of endogenous enzymes may limit activation of the immune response.
However, as with all of the suicide enzymes, dCK has not evolved to use prodrugs. We have
attempted to improve the activity of dCK towards gemcitabine (dFdC), a clinically used anti-
cancer agent, by regio-specific random mutagenesis (Chapter 6). In this study, we identified
eight dCK variants capable of extensively sensitizing E. coli to gemcitabine, up to 35,000-fold.
Unexpectedly though, these mutants have not proven to necessarily be more catalytically
efficient for dFdC or to confer improved cytotoxicity in tumor cells. However, we believe future
work to elucidate why this discrepancy between the bacterial and mammalian cell models is
occurring should be pursued.
In summary, the identification of enzymes with enhanced or novel functions for suicide
gene therapy of cancer is significant and has already accelerated the current effectiveness of gene
therapy. We believe the mutant enzymes created in these studies will provide additional insight
and resources for future suicide gene therapy applications and may be applied to treating
numerous types of cancer.
96
APPENDICES
In the following appendices data associated with the creation of the dCK random library
will be discussed.
Appendix A: Construction of a positive selection system for the human deoxycytidine kinase in
E. coli
Appendix B: Screening and characterization of an error-prone polymerase chain reaction
deoxycytidine kinase library
97
APPENDIX A
Construction of a positive selection for human deoxycytidine kinase in E. coli
The deoxyribonucleoside savage pathway in mammalian cells is dependent upon four
enzymes, including deoxycytidine kinase (dCK), thymidine kinase 1 and 2 (TK), and
deoxyguanosine kinase (dGK). However, the salvage pathways of deoxynucleosides in
prokaryotes and eukaryotes, especially deoxyadenosine (dA), deoxyguanosine (dG) and
deoxycytidine (dG), show large variations. In fact, E. coli does not express dCK activity.
Because of this, simple genetic complementation in E. coli, by creating deoxycytidine
auxotrophs and then compensating for the deficiency by introducing a plasmid harboring a dCK
gene, is not possible for the selection of dCK mutant enzymes.
As can be imagined, the ability to select only the functional mutants from a large library
would greatly decrease the manual work of having to screen thousands of individual. Therefore,
we attempted to create a positive selection system for the dCK mutagenesis using several
different E. coli strains.
A previous study aimed at identifying mutant E. coli strains which could be used to
screen pharmacological nucleoside analogs characterized the cytidine deaminase (cdd) deficient-
strain, SØ441 (cdd-5, relA1, rpsL254(strR), metB1, upp-1) (158) (Figure 7.1). This strain was
found to be sensitive to gemcitabine upon introduction of the dCK gene indicating that it may
potentially be useful in screening for deoxycytidine kinase mutants. In addition, the E. coli
strain, χ2465 (∆(proB-lac)41, glnV42(AS), λ-, pyrF111, cdd-7, pyrG77, thyA702, cycA1,
deoC34, T 3
R), was identified as a possible strain for the selection of functional dCK mutants
(159). This strain contains mutations in several genes: cdd, deoC (deoxyriboaldolase), pyrF
(orotidine 5’-monophosphate pyrophosphorylase), pyrG (CTP synthetase), thyA (thymidylate
98
synthase), and upp (UMP pyrophosphorylase). Therefore, because of these gene deficiencies,
χ2465 is not able to synthesize CTP, and requires cytidine for growth in minimal media.
Figure 7.1. Pathways in the pyrimidine metabolism of E. coli. (Adapted from Maturin and Curtiss (159).) The enzymes are: cdd, cytidine deaminase; cmk, deoxycytidylate kinase; cod, cytosine deaminase; deoC, deoxyriboaldolase; pyrF, orotidine 5’-phosphate decarboxylase; pyrG, CTP synthetase; thyA, thymidylate synthase; udk, uridine/cytidine kinase; udp, uridine phosphorylase; upp, uracil phosphoribosyltransferase. Enzymes in blue are involved salvage pathways. Enzymes deficient in SØ441 are underlined, while enzymes deficient in χ2465 are starred. Metabolic reactions in red are not found in E. coli.
E. coli SØ441 and χ2465 were both obtained from the E. coli Genetic Stock Center (Yale
University) and were lysogenized with DE3 to permit gene expression from pET-based vectors
according to the manufacturer’s instructions (Novagen, Madison, WI) and designated
SØ441(DE3) and χ2465(DE3), respectively. Competent cells from both strains were
deoC
upp
pyrG
cod
cdd
pyrF thyA
dCTP
dCDP
UdR
dUMP
dUDP
dUTP
CDP
CMP
CR
C
CTP UTP
UDP
UMP
U
UR
T
TdR
dTDP
dTMP
dTTP
OMP
d-5-P
d-I-P
G-3-P + A
udp
udk udk
*
*
*
*
*
*
*
dC
dCMPMPOP dck
99
transformed with pETHT, pETHT:dCK or pETHT:dCK dummy. Single colonies were then
plated onto M9 minimal medium supplemented with various nucleosides:
1. uracil (dU) (80 µg/mL),
2. thymidine (dT) (1 or 2 µg/mL),
3. dC (25, 100 or 200 µg/mL),
4. dT (1 µg/mL) and dC (25 µg/mL, 100 µg/mL, 200 µg/mL, 1 mg/mL, or 2 mg/mL),
5. dC (1 mg/mL) and uracil (5 µg/mL), and
6. dT (1 µg/mL), dC (1 or 2 µg/mL), uracil (5 µg/mL).
On plates containing all of the above there was no observable difference in the viability
of E. coli from either strain harboring vector alone (no dCK gene), a non-functional dCK gene or
with a functional dCK gene. CTP synthetase (pyrF), SØ441(DE3) is able to metabolize CTP
from UTP, ATP and glutamine and is not affected by the addition or lack of dC to the medium.
The χ2465(DE3) strain did not grow well on any of the above plates or on rich medium and we
believe the sheer amount of genetic deficiencies in both de novo and salvage nucleotide
pathways has impaired its viability. Furthermore, looking at the pyrimidine synthesis pathways
it appears that χ2465(DE3) would need to have a deficiency in uridine kinase (udk) to inhibit the
possibility of cytidine being metabolized to CMP. However, we believe creating even further
deficiencies in the χ2465(DE3) strain would not be advantageous because its hardiness is quite
affected already. One possibility for future work would be to create a strain deficient in cdd,
pyrG and udk, forcing the strain to metabolize dCTP from the supplemented dC by dCK, while
still allowing for metabolism of the other nucleotides through intact de novo or salvage
pathways.
100
APPENDIX B
Screening and Characterization of an Error-Prone Polymerase Chain Reaction
Deoxycytidine Kinase Library
In the first attempt at the regio-specific random mutagenesis of dCK eight overlapping
oligonucleotides were used. Although several approaches were taken to produce the random
PCR fragment of the correct size including: annealing only two of the primers at a time,
adding/decreasing the primer(s) concentration, optimization of the annnealing temperature, and
other protocol modifications, full-length product was never obtained.
As an alternate strategy to the regio-specific random mutagenesis library, error-prone
PCR (epPCR) was performed to create mutant dCK enzymes. Unlike regio-specific random
mutagenesis that targets specific amino acid (a.a.) codons, epPCR is capable of introducing
mutations at random sites throughout the entire gene. Random mutations are introduced using a
low fidelity polymerase chain reaction. They can be achieved by increasing the concentration of
MgCl2, adding MnCl2, skewing the levels of the dNTP pool, increasing the concentration of
polymerase or increasing the extension time (65). Using epPCR, a random PCR fragment was
generated and isolated. The fragment was then digested and sub-cloned into the similarly cut
pETHT vector. After ligation, the E. coli strain BL21(DE3)tdk- was transformed with the newly
created pETHT:dCK epPCR mutant DNA. Transformants were then tested on gemcitabine-
containing plates (as described in Chapter 5) and DNA from each colony was isolated and
digested with restriction enzymes to evaluate the presence of the dCK gene. Several different
ligations and transformations revealed no transformants with an introduced dCK gene. In order
to improve the cloning efficiency the vector pETHT was first cut with the two restriction
enzymes and then treated with shrimp alkaline phosphatase to remove the 5’ phosphate groups
101
from the DNA to prevent self-ligation. Once again, from several transformations colonies were
chosen and tested to determine if they contained the introduced dCK gene. From 358 colonies,
17 were dCK positive, indicating only a 4.7% cloning efficiency.
In order to improve the cloning efficiency, megaprimer PCR of whole plasmid
(MEGAWHOP) was utilized. This method does not involve traditional cloning steps, such as
restriction digestions or ligations, but instead utilizes the premise that some bacterial strains
methylate their DNA during replication. Additionally, MEGAWHOP helps to eliminate the
unwanted plasmids, such as those with no insert/gene or plasmids with inserts. Using
MEGAWHOP both the epPCR and regio-specific random libraries (Chapter 5) were created
side-by-side. For the epPCR product, T7 promoter and terminator primers that hybridize to
vector sequences flanking the dCK gene were used to amplify the entire gene. Once the gene
was synthesized, it was purified by gel isolation and used to prime for the next round of PCR.
For the second PCR reaction, the dCK dummy vector, isolated from a DNA adenine
methyltransferase positive (dam+) E. coli, was used as the template. The product from this PCR
reaction was then treated with DpnI, a restriction enzyme that will digest methylated DNA.
Thus, template DNA isolated from the dam+ bacteria will be digested while newly formed,
mutated DNA will not be digested. Finally, the digested product was used to transform E. coli
and resulting colonies were screened for increased activity towards gemcitabine as outlined in
Chapter 5.
From the epPCR library 5,007 transformants were screened. Of those, 3,775 (or 75%)
displayed gemcitabine sensitivity on the plates containing the lethal dose of gemcitabine.
However, only three variants (0.06%) displayed super-sensitivity on the 5 nM dFdC-containing
plates (mutants 1545, 1546 and 1552). As done with the regio-specific random mutagenesis
102
library, DNA from the non-selected and gemcitabine-sensitive mutants was isolated and
sequenced. Mutant 12, a gemcitabine-sensitive variant, was found to be a truncated protein,
having a stop codon in place of glutamine 230. This is particularly interesting because even
though mutant 12 lacks the last 30 amino acids it still retains activity similar to wild-type
enzyme, suggesting that these last amino acids are not particularly important for the overall
catalytic properties of dCK. Super-sensitive mutant 1545, which has a V38L substitution, was
found to be more sensitive than wild-type to gemcitabine. This residue is located near the P-loop
on alpha helix 1 of the dCK structure (53). Interestingly, in an alignment study D. melanogaster
deoxyribonucleoside kinase and human thymidine kinase 2 have a leucine rather than valine
naturally (53). Super-sensitive mutant 1546 has an uncharged, polar side chain (glutamine165)
to a charged, polar side chain (lysine) substitution. Residue 165 is located on the alpha helix 7,
which with alpha helix 4 makes up the four-helix bundle of the dimer interface. Lastly, mutant
1552, which was also found to be super-sensitive, has a valine to aspartic acid change at residue
250. The 250 residue is found on alpha helix 10 of the dCK structure.
The super-sensitive mutants from the epPCR library were next evaluated using bacterial
cell lysates assays for gemcitabine (see Materials and Methods, Chapter 5), however, the mutant
enzymes from the regio-specific random library were ultimately found to be more active. While
the epPCR library would have been informative in regards to the structure/function relationship
of this enzyme, this library was eventually abandoned in favor of the regio-specific library
because the latter library was producing a greater number of mutants that displayed sensitivity to
gemcitabine at the lowest concentration in the screening system.
103
LIST OF REFERENCES 1. Huber, B. E., Richards, C. A. & Krenitsky, T. A. (1991) Retroviral-mediated gene therapy for the treatment of hepatocellular carcinoma: an innovative approach for cancer therapy. Proc Natl Acad Sci U S A 88: 8039-8043. 2. Gene therapy clinical trials world wide (www.wiley.co.uk/genetherapy/clinical). Journal of Gene Medicine. 3. Bester, A. C., Schwartz, M., Schmidt, M., Garrigue, A., Hacein-Bey-Abina, S., Cavazzana-Calvo, M., Ben-Porat, N., Von Kalle, C., Fischer, A. & Kerem, B. (2006) Fragile sites are preferential targets for integrations of MLV vectors in gene therapy. Gene Ther. 4. Ginn, S. L., Curtin, J. A., Kramer, B., Smyth, C. M., Wong, M., Kakakios, A., McCowage, G. B., Watson, D., Alexander, S. I. et al. (2005) Treatment of an infant with X-linked severe combined immunodeficiency (SCID-X1) by gene therapy in Australia. Med J Aust 182: 458-463. 5. Moolten, F. L. (1986) Tumor chemosensitivity conferred by inserted herpes thymidine kinase genes: paradigm for a prospective cancer control strategy. Cancer Res 46: 5276-5281. 6. Dilber, M. S., Abedi, M. R., Christensson, B., Bjorkstrand, B., Kidder, G. M., Naus, C. C., Gahrton, G. & Smith, C. I. (1997) Gap junctions promote the bystander effect of herpes simplex virus thymidine kinase in vivo. Cancer Res 57: 1523-1528. 7. Mesnil, M., Piccoli, C., Tiraby, G., Willecke, K. & Yamasaki, H. (1996) Bystander killing of cancer cells by herpes simplex virus thymidine kinase gene is mediated by connexins. Proc Natl Acad Sci U S A 93: 1831-1835. 8. Touraine, R. L., Vahanian, N., Ramsey, W. J. & Blaese, R. M. (1998) Enhancement of the herpes simplex virus thymidine kinase/ganciclovir bystander effect and its antitumor efficacy in vivo by pharmacologic manipulation of gap junctions. Hum Gene Ther 9: 2385-2391. 9. Colombo, B. M., Benedetti, S., Ottolenghi, S., Mora, M., Pollo, B., Poli, G. & Finocchiaro, G. (1995) The "bystander effect": association of U-87 cell death with ganciclovir-mediated apoptosis of nearby cells and lack of effect in athymic mice. Hum Gene Ther 6: 763-772. 10. Freeman, S. M., Abboud, C. N., Whartenby, K. A., Packman, C. H., Koeplin, D. S., Moolten, F. L. & Abraham, G. N. (1993) The "bystander effect": tumor regression when a fraction of the tumor mass is genetically modified. Cancer Res 53: 5274-5283. 11. Greco, O., Folkes, L. K., Wardman, P., Tozer, G. M. & Dachs, G. U. (2000) Development of a novel enzyme/prodrug combination for gene therapy of cancer: horseradish peroxidase/indole-3-acetic acid. Cancer Gene Ther 7: 1414-1420.
104
12. Lawrence, T. S., Rehemtulla, A., Ng, E. Y., Wilson, M., Trosko, J. E. & Stetson, P. L. (1998) Preferential cytotoxicity of cells transduced with cytosine deaminase compared to bystander cells after treatment with 5-flucytosine. Cancer Res 58: 2588-2593. 13. Stribbling, S. M., Friedlos, F., Martin, J., Davies, L., Spooner, R. A., Marais, R. & Springer, C. J. (2000) Regressions of established breast carcinoma xenografts by carboxypeptidase G2 suicide gene therapy and the prodrug CMDA are due to a bystander effect. Hum Gene Ther 11: 285-292. 14. Caruso, M., Panis, Y., Gagandeep, S., Houssin, D., Salzmann, J. L. & Klatzmann, D. (1993) Regression of established macroscopic liver metastases after in situ transduction of a suicide gene. Proc Natl Acad Sci U S A 90: 7024-7028. 15. Kuriyama, S., Kikukawa, M., Masui, K., Okuda, H., Nakatani, T., Akahane, T., Mitoro, A., Tominaga, K., Tsujinoue, H. et al. (1999) Cancer gene therapy with HSV-tk/GCV system depends on T-cell-mediated immune responses and causes apoptotic death of tumor cells in vivo. Int J Cancer 83: 374-380. 16. Kuriyama, S., Sakamoto, T., Masui, K., Nakatani, T., Tominaga, K., Kikukawa, M., Yoshikawa, M., Ikenaka, K., Fukui, H. & Tsujii, T. (1997) Tissue-specific expression of HSV-tk gene can induce efficient antitumor effect and protective immunity to wild-type hepatocellular carcinoma. Int J Cancer 71: 470-475. 17. Jones, R. K., Pope, I. M., Kinsella, A. R., Watson, A. J. & Christmas, S. E. (2000) Combined suicide and granulocyte-macrophage colony-stimulating factor gene therapy induces complete tumor regression and generates antitumor immunity. Cancer Gene Ther 7: 1519-1528. 18. Agard, C., Ligeza, C., Dupas, B., Izembart, A., El Kouri, C., Moullier, P. & Ferry, N. (2001) Immune-dependent distant bystander effect after adenovirus-mediated suicide gene transfer in a rat model of liver colorectal metastasis. Cancer Gene Ther 8: 128-136. 19. Dachs, G. U., Tupper, J. & Tozer, G. M. (2005) From bench to bedside for gene-directed enzyme prodrug therapy of cancer. Anticancer Drugs 16: 349-359. 20. Elion, G. B. (1980) The chemotherapeutic exploitation of virus-specified enzymes. Adv Enzyme Regul 18: 53-66. 21. Holder, J. W., Elmore, E. & Barrett, J. C. (1993) Gap junction function and cancer. Cancer Res 53: 3475-3485. 22. Hotz-Wagenblatt, A. & Shalloway, D. (1993) Gap junctional communication and neoplastic transformation. Crit Rev Oncog 4: 541-558. 23. Longley, D. B., Harkin, D. P. & Johnston, P. G. (2003) 5-fluorouracil: mechanisms of action and clinical strategies. Nat Rev Cancer 3: 330-338.
105
24. Hyatt, D. C., Maley, F. & Montfort, W. R. (1997) Use of strain in a stereospecific catalytic mechanism: crystal structures of Escherichia coli thymidylate synthase bound to FdUMP and methylenetetrahydrofolate. Biochemistry 36: 4585-4594. 25. Kalman, T. I. (1975) Molecular aspects of the mechanism of action of 5-fluorodeoxyuridine. Ann N Y Acad Sci 255: 326-331. 26. Matthews, D. A., Villafranca, J. E., Janson, C. A., Smith, W. W., Welsh, K. & Freer, S. (1990) Stereochemical mechanism of action for thymidylate synthase based on the X-ray structure of the covalent inhibitory ternary complex with 5-fluoro-2'-deoxyuridylate and 5,10-methylenetetrahydrofolate. J Mol Biol 214: 937-948. 27. Santi, D. V., McHenry, C. S. & Sommer, H. (1974) Mechanism of interaction of thymidylate synthetase with 5-fluorodeoxyuridylate. Biochemistry 13: 471-481. 28. Ghoshal, K. & Jacob, S. T. (1994) Specific inhibition of pre-ribosomal RNA processing in extracts from the lymphosarcoma cells treated with 5-fluorouracil. Cancer Res 54: 632-636. 29. Hoganson, D. K., Batra, R. K., Olsen, J. C. & Boucher, R. C. (1996) Comparison of the effects of three different toxin genes and their levels of expression on cell growth and bystander effect in lung adenocarcinoma. Cancer Res 56: 1315-1323. 30. Kuriyama, S., Masui, K., Sakamoto, T., Nakatani, T., Kikukawa, M., Tsujinoue, H., Mitoro, A., Yamazaki, M., Yoshiji, H. et al. (1998) Bystander effect caused by cytosine deaminase gene and 5-fluorocytosine in vitro is substantially mediated by generated 5-fluorouracil. Anticancer Res 18: 3399-3406. 31. Nishihara, E., Nagayama, Y., Narimatsu, M., Namba, H., Watanabe, M., Niwa, M. & Yamashita, S. (1998) Treatment of thyroid carcinoma cells with four different suicide gene/prodrug combinations in vitro. Anticancer Res 18: 1521-1525. 32. Trinh, Q. T., Austin, E. A., Murray, D. M., Knick, V. C. & Huber, B. E. (1995) Enzyme/prodrug gene therapy: comparison of cytosine deaminase/5-fluorocytosine versus thymidine kinase/ganciclovir enzyme/prodrug systems in a human colorectal carcinoma cell line. Cancer Res 55: 4808-4812. 33. Kievit, E., Nyati, M. K., Ng, E., Stegman, L. D., Parsels, J., Ross, B. D., Rehemtulla, A. & Lawrence, T. S. (2000) Yeast cytosine deaminase improves radiosensitization and bystander effect by 5-fluorocytosine of human colorectal cancer xenografts. Cancer Res 60: 6649-6655. 34. Lawrence, T. S., Davis, M. A. & Maybaum, J. (1994) Dependence of 5-fluorouracil-mediated radiosensitization on DNA-directed effects. Int J Radiat Oncol Biol Phys 29: 519-523. 35. Hamstra, D. A., Rice, D. J., Pu, A., Oyedijo, D., Ross, B. D. & Rehemtulla, A. (1999) Combined radiation and enzyme/prodrug treatment for head and neck cancer in an orthotopic animal model. Radiat Res 152: 499-507.
106
36. Stackhouse, M. A., Pederson, L. C., Grizzle, W. E., Curiel, D. T., Gebert, J., Haack, K., Vickers, S. M., Mayo, M. S. & Buchsbaum, D. J. (2000) Fractionated radiation therapy in combination with adenoviral delivery of the cytosine deaminase gene and 5-fluorocytosine enhances cytotoxic and antitumor effects in human colorectal and cholangiocarcinoma models. Gene Ther 7: 1019-1026. 37. Kievit, E., Bershad, E., Ng, E., Sethna, P., Dev, I., Lawrence, T. S. & Rehemtulla, A. (1999) Superiority of yeast over bacterial cytosine deaminase for enzyme/prodrug gene therapy in colon cancer xenografts. Cancer Res 59: 1417-1421. 38. Ge, K., Xu, L., Zheng, Z., Xu, D., Sun, L. & Liu, X. (1997) Transduction of cytosine deaminase gene makes rat glioma cells highly sensitive to 5-fluorocytosine. Int J Cancer 71: 675-679. 39. Bentires-Alj, M., Hellin, A. C., Lechanteur, C., Princen, F., Lopez, M., Fillet, G., Gielen, J., Merville, M. P. & Bours, V. (2000) Cytosine deaminase suicide gene therapy for peritoneal carcinomatosis. Cancer Gene Ther 7: 20-26. 40. Consalvo, M., Mullen, C. A., Modesti, A., Musiani, P., Allione, A., Cavallo, F., Giovarelli, M. & Forni, G. (1995) 5-Fluorocytosine-induced eradication of murine adenocarcinomas engineered to express the cytosine deaminase suicide gene requires host immune competence and leaves an efficient memory. J Immunol 154: 5302-5312. 41. Ichikawa, T., Tamiya, T., Adachi, Y., Ono, Y., Matsumoto, K., Furuta, T., Yoshida, Y., Hamada, H. & Ohmoto, T. (2000) In vivo efficacy and toxicity of 5-fluorocytosine/cytosine deaminase gene therapy for malignant gliomas mediated by adenovirus. Cancer Gene Ther 7: 74-82. 42. Kanai, F., Lan, K. H., Shiratori, Y., Tanaka, T., Ohashi, M., Okudaira, T., Yoshida, Y., Wakimoto, H., Hamada, H. et al. (1997) In vivo gene therapy for alpha-fetoprotein-producing hepatocellular carcinoma by adenovirus-mediated transfer of cytosine deaminase gene. Cancer Res 57: 461-465. 43. Topf, N., Worgall, S., Hackett, N. R. & Crystal, R. G. (1998) Regional 'pro-drug' gene therapy: intravenous administration of an adenoviral vector expressing the E. coli cytosine deaminase gene and systemic administration of 5-fluorocytosine suppresses growth of hepatic metastasis of colon carcinoma. Gene Ther 5: 507-513. 44. Manome, Y., Wen, P. Y., Dong, Y., Tanaka, T., Mitchell, B. S., Kufe, D. W. & Fine, H. A. (1996) Viral vector transduction of the human deoxycytidine kinase cDNA sensitizes glioma cells to the cytotoxic effects of cytosine arabinoside in vitro and in vivo. Nat Med 2: 567-573. 45. Peters, G. J., Ruiz van Haperen, V. W., Bergman, A. M., Veerman, G., Smitskamp-Wilms, E., van Moorsel, C. J., Kuiper, C. M. & Braakhuis, B. J. (1996) Preclinical combination therapy with gemcitabine and mechanisms of resistance. Semin Oncol 23: 16-24.
107
46. Bouffard, D. Y., Laliberte, J. & Momparler, R. L. (1993) Kinetic studies on 2',2'-difluorodeoxycytidine (Gemcitabine) with purified human deoxycytidine kinase and cytidine deaminase. Biochem Pharmacol 45: 1857-1861. 47. Goan, Y. G., Zhou, B., Hu, E., Mi, S. & Yen, Y. (1999) Overexpression of ribonucleotide reductase as a mechanism of resistance to 2,2-difluorodeoxycytidine in the human KB cancer cell line. Cancer Res 59: 4204-4207. 48. Blackstock, A. W., Lightfoot, H., Case, L. D., Tepper, J. E., Mukherji, S. K., Mitchell, B. S., Swarts, S. G. & Hess, S. M. (2001) Tumor uptake and elimination of 2',2'-difluoro-2'-deoxycytidine (gemcitabine) after deoxycytidine kinase gene transfer: correlation with in vivo tumor response. Clin. Cancer Res. 7: 3263-3268. 49. Kojima, H., Iida, M., Miyazaki, H., Koga, T., Moriyama, H. & Manome, Y. (2002) Enhancement of cytarabine sensitivity in squamous cell carcinoma cell line transfected with deoxycytidine kinase. Arch Otolaryngol Head Neck Surg 128: 708-713. 50. Sanda, A., Zhu, C., Johansson, M. & Karlsson, A. (2001) Bystander effects of nucleoside analogs phosphorylated in the cytosol or mitochondria. Biochem Biophys Res Commun 287: 1163-1166. 51. Zhu, C., Johansson, M. & Karlsson, A. (2000) Incorporation of nucleoside analogs into nuclear or mitochondrial DNA is determined by the intracellular phosphorylation site. J Biol Chem 275: 26727-26731. 52. Bergman, A. M., Pinedo, H. M. & Peters, G. J. (2002) Determinants of resistance to 2',2'-difluorodeoxycytidine (gemcitabine). Drug Resist Updat 5: 19-33. 53. Sabini, E., Ort, S., Monnerjahn, C., Konrad, M. & Lavie, A. (2003) Structure of human dCK suggests strategies to improve anticancer and antiviral therapy. Nat. Struct. Biol. 10: 513-519. 54. Kim, Y. G., Bi, W., Feliciano, E. S., Drake, R. R. & Stambrook, P. J. (2000) Ganciclovir-mediated cell killing and bystander effect is enhanced in cells with two copies of the herpes simplex virus thymidine kinase gene. Cancer Gene Ther 7: 240-246. 55. Davies, L. C., Friedlos, F., Hedley, D., Martin, J., Ogilvie, L. M., Scanlon, I. J. & Springer, C. J. (2005) Novel fluorinated prodrugs for activation by carboxypeptidase G2 showing good in vivo antitumor activity in gene-directed enzyme prodrug therapy. J Med Chem 48: 5321-5328. 56. Boucher, P. D., Ostruszka, L. J. & Shewach, D. S. (2000) Synergistic enhancement of herpes simplex virus thymidine kinase/ganciclovir-mediated cytoxicity by hydroxyurea. Cancer Res 60: 1631-1636.
108
57. Rogulski, K. R., Kim, J. H., Kim, S. H. & Freytag, S. O. (1997) Glioma cells transduced with an Escherichia coli CD/HSV-1 TK fusion gene exhibit enhanced metabolic suicide and radiosensitivity. Hum Gene Ther 8: 73-85. 58. Black, M. E., Newcomb, T. G., Wilson, H. M. & Loeb, L. A. (1996) Creation of drug-specific herpes simplex virus type 1 thymidine kinase mutants for gene therapy. Proc Natl Acad Sci U S A 93: 3525-3529. 59. Kokoris, M. S., Sabo, P., Adman, E. T. & Black, M. E. (1999) Enhancement of tumor ablation by a selected HSV-1 thymidine kinase mutant. Gene Ther. 6: 1415-1426. 60. Black, M. E., Kokoris, M.S., and Sabo, P. (2001) Herpes simplex virus-1 thymidine kinase mutants created by semi-random sequence mutagenesis improve prodrug-mediated tumor cell killing. Cancer Research 61: 3022-3026. 61. Dube, D. K. & Loeb, L. A. (1989) Mutants generated by the insertion of random oligonucleotides into the active site of the beta-lactamase gene. Biochemistry 28: 5703-5707. 62. Horwitz, M. S., Dube, D. K. & Loeb, L. A. (1989) Selection of new biological activities from random nucleotide sequences: evolutionary and practical considerations. Genome 31: 112-117. 63. Munir, K. M., French, D. C., Dube, D. K. & Loeb, L. A. (1992) Permissible amino acid substitutions within the putative nucleoside binding site of herpes simplex virus type 1 encoded thymidine kinase established by random sequence mutagenesis [corrected]. J Biol Chem 267: 6584-6589. 64. Reidhaar-Olson, J. F. & Sauer, R. T. (1988) Combinatorial cassette mutagenesis as a probe of the informational content of protein sequences. Science 241: 53-57. 65. Cadwell, R. C. & Joyce, G. F. (1992) Randomization of genes by PCR mutagenesis. PCR Methods Appl 2: 28-33. 66. Stemmer, W. P. C. (1994) Rapid evolution of a protein in vitro by DNA shuffling. Nature 370: 389-391. 67. Munir, K. M., French, D. C. & Loeb, L. A. (1993) Thymidine kinase mutants obtained by random sequence selection. Proc Natl Acad Sci U S A 90: 4012-4016. 68. Black, M. E., Rechtin, T. M. & Drake, R. R. (1996) Effect on substrate binding of an alteration at the conserved aspartic acid-162 in herpes simplex virus type 1 thymidine kinase. J. Gen. Virol. 77 ( Pt 7): 1521-1527. 69. Black, M. E. & Loeb, L. A. (1993) Identification of important residues within the putative nucleoside binding site of HSV-1 thymidine kinase by random sequence selection: analysis of selected mutants in vitro. Biochemistry 32: 11618-11626.
109
70. Qiao, J., Black, M. E. & Caruso, M. (2000) Enhanced ganciclovir killing and bystander effect of human tumor cells transduced with a retroviral vector carrying a herpes simplex virus thymidine kinase gene mutant. Hum Gene Ther 11: 1569-1576. 71. Wiewrodt, R., Amin, K., Kiefer, M., Jovanovic, V. P., Kapoor, V., Force, S., Chang, M., Lanuti, M., Black, M. E. et al. (2003) Adenovirus-mediated gene transfer of enhanced Herpes simplex virus thymidine kinase mutants improves prodrug-mediated tumor cell killing. Cancer Gene Ther 10: 353-364. 72. Fridland, A., Connelly, M. C. & Ashmun, R. (1990) Relationship of deoxynucleotide changes to inhibition of DNA synthesis induced by the antiretroviral agent 3'-azido-3'-deoxythymidine and release of its monophosphate by human lymphoid cells (CCRF-CEM). Mol Pharmacol 37: 665-670. 73. Johansson, M., Amiri, M. & Karlsson, A. (2005) Phosphorylation of 9-beta-D-arabinofuranosylguanine monophosphate by Drosophila melanogaster guanylate kinase. Biochem Pharmacol 70: 987-992. 74. Ostermann, N., Lavie, A., Padiyar, S., Brundiers, R., Veit, T., Reinstein, J., Goody, R. S., Konrad, M. & Schlichting, I. (2000) Potentiating AZT activation: structures of wild-type and mutant human thymidylate kinase suggest reasons for the mutants' improved kinetics with the HIV prodrug metabolite AZTMP. J Mol Biol 304: 43-53. 75. Fuchita, M., Ardiani, A., Zhao, L., Stoddard, B. L. & Black, M. E. Engineering bacterial cytosine deaminase for efficient 5-FU production for cancer gene therapy. Manuscript in preparation. 76. Mahan, S. D., Ireton, G. C., Knoeber, C., Stoddard, B. L. & Black, M. E. (2004) Random mutagenesis and selection of Escherichia coli cytosine deaminase for cancer gene therapy. Protein Eng. Des. Sel. 17: 625-633. 77. Mahan, S. D., Ireton, G. C., Stoddard, B. L. & Black, M. E. (2004) Alanine-scanning mutagenesis reveals a cytosine deaminase mutant with altered substrate preference. Biochemistry 43: 8957-8964. 78. Kaliberov, S. A., Markert, J. M., Krendelchtchikova, V., Manna, D. D., Sellers, J. C., Kaliberova, L. N., Black, M. E. & Buchsbaum, D. J. (2006) Gene directed/enzyme prodrug therapy of human glioma using mutant Escherichia coli cytosine deaminase. Molecular Therapy Submitted. 79. Blackburn, R. V., Galoforo, S. S., Corry, P. M. & Lee, Y. J. (1999) Adenoviral transduction of a cytosine deaminase/thymidine kinase fusion gene into prostate carcinoma cells enhances prodrug and radiation sensitivity. Int J Cancer 82: 293-297.
110
80. Bourbeau, D., Lavoie, G., Nalbantoglu, J. & Massie, B. (2004) Suicide gene therapy with an adenovirus expressing the fusion gene CD::UPRT in human glioblastomas: different sensitivities correlate with p53 status. J Gene Med 6: 1320-1332. 81. Freytag, S. O., Khil, M., Stricker, H., Peabody, J., Menon, M., DePeralta-Venturina, M., Nafziger, D., Pegg, J., Paielli, D. et al. (2002) Phase I study of replication-competent adenovirus-mediated double suicide gene therapy for the treatment of locally recurrent prostate cancer. Cancer Res 62: 4968-4976. 82. Freytag, S. O., Rogulski, K. R., Paielli, D. L., Gilbert, J. D. & Kim, J. H. (1998) A novel three-pronged approach to kill cancer cells selectively: concomitant viral, double suicide gene, and radiotherapy. Hum Gene Ther 9: 1323-1333. 83. Hanna, N. N., Mauceri, H. J., Wayne, J. D., Hallahan, D. E., Kufe, D. W. & Weichselbaum, R. R. (1997) Virally directed cytosine deaminase/5-fluorocytosine gene therapy enhances radiation response in human cancer xenografts. Cancer Res 57: 4205-4209. 84. Rogulski, K. R., Wing, M. S., Paielli, D. L., Gilbert, J. D., Kim, J. H. & Freytag, S. O. (2000) Double suicide gene therapy augments the antitumor activity of a replication-competent lytic adenovirus through enhanced cytotoxicity and radiosensitization. Hum Gene Ther 11: 67-76. 85. Rogulski, K. R., Zhang, K., Kolozsvary, A., Kim, J. H. & Freytag, S. O. (1997) Pronounced antitumor effects and tumor radiosensitization of double suicide gene therapy. Clin Cancer Res 3: 2081-2088. 86. Wu, D. H., Liu, L. & Chen, L. H. (2005) Antitumor effects and radiosensitization of cytosine deaminase and thymidine kinase fusion suicide gene on colorectal carcinoma cells. World J Gastroenterol 11: 3051-3055. 87. Erbs, P., Regulier, E., Kintz, J., Leroy, P., Poitevin, Y., Exinger, F., Jund, R. & Mehtali, M. (2000) In vivo cancer gene therapy by adenovirus-mediated transfer of a bifunctional yeast cytosine deaminase/uracil phosphoribosyltransferase fusion gene. Cancer Res 60: 3813-3822. 88. Graepler, F., Lemken, M. L., Wybranietz, W. A., Schmidt, U., Smirnow, I., Gross, C. D., Spiegel, M., Schenk, A., Graf, H. et al. (2005) Bifunctional chimeric SuperCD suicide gene -YCD: YUPRT fusion is highly effective in a rat hepatoma model. World J Gastroenterol 11: 6910-6919. 89. Tiraby, M., Cazaux, C., Baron, M., Drocourt, D., Reynes, J. P. & Tiraby, G. (1998) Concomitant expression of E. coli cytosine deaminase and uracil phosphoribosyltransferase improves the cytotoxicity of 5-fluorocytosine. FEMS Microbiol Lett 167: 41-49. 90. Korkegian, A., Black, M. E., Baker, D. & Stoddard, B. L. (2005) Computational thermostabilization of an enzyme. Science 308: 857-860.
111
91. Black, M. E., Munir, K. M. & Loeb, L. A. (1993) Random sequence selection for the generation of active recombinant molecules. Meth Mol Gen 2: 26-43. 92. Balzarini, J., Bohman, C. & De Clercq, E. (1993) Differential mechanism of cytostatic effect of (E)-5-(2-bromovinyl)-2'-deoxyuridine, 9-(1,3-dihydroxy-2-propoxymethyl)guanine, and other antiherpetic drugs on tumor cells transfected by the thymidine kinase gene of herpes simplex virus type 1 or type 2. J Biol Chem 268: 6332-6337. 93. Fyfe, J. A. (1982) Differential phosphorylation of (E)-5-(2-bromovinyl)-2'-deoxyuridine monophosphate by thymidylate kinases from herpes simplex viruses types 1 and 2 and varicella zoster virus. Mol Pharmacol 21: 432-437. 94. Alvarez, R. D., Gomez-Navarro, J., Wang, M., Barnes, M. N., Strong, T. V., Arani, R. B., Arafat, W., Hughes, J. V., Siegal, G. P. & Curiel, D. T. (2000) Adenoviral-mediated suicide gene therapy for ovarian cancer. Mol Ther 2: 524-530. 95. Germano, I. M., Fable, J., Gultekin, S. H. & Silvers, A. (2003) Adenovirus/herpes simplex-thymidine kinase/ganciclovir complex: preliminary results of a phase I trial in patients with recurrent malignant gliomas. J Neurooncol 65: 279-289. 96. Kubo, H., Gardner, T. A., Wada, Y., Koeneman, K. S., Gotoh, A., Yang, L., Kao, C., Lim, S. D., Amin, M. B. et al. (2003) Phase I dose escalation clinical trial of adenovirus vector carrying osteocalcin promoter-driven herpes simplex virus thymidine kinase in localized and metastatic hormone-refractory prostate cancer. Hum. Gene Ther. 14: 227-241. 97. Morris, J. C., Ramsey, W. J., Wildner, O., Muslow, H. A., Aguilar-Cordova, E. & Blaese, R. M. (2000) A phase I study of intralesional administration of an adenovirus vector expressing the HSV-1 thymidine kinase gene (AdV.RSV-TK) in combination with escalating doses of ganciclovir in patients with cutaneous metastatic malignant melanoma. Hum Gene Ther 11: 487-503. 98. Singh, S., Cunningham, C., Buchanan, A., Jolly, D. J. & Nemunaitis, J. (2001) Toxicity assessment of intratumoral injection of the herpes simplex type I thymidine kinase gene delivered by retrovirus in patients with refractory cancer. Mol Ther 4: 157-160. 99. Sung, M. W., Yeh, H. C., Thung, S. N., Schwartz, M. E., Mandeli, J. P., Chen, S. H. & Woo, S. L. (2001) Intratumoral adenovirus-mediated suicide gene transfer for hepatic metastases from colorectal adenocarcinoma: results of a phase I clinical trial. Mol Ther 4: 182-191. 100. Black, M. E. & Hruby, D. E. (1990) Identification of the ATP-binding domain of vaccinia virus thymidine kinase. J Biol Chem 265: 17584-17592. 101. Kunkel, T. A. (1985) Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc Natl Acad Sci U S A 82: 488-492.
112
102. Kunkel, T. A., Roberts, J. D. & Zakour, R. A. (1987) Rapid and efficient site-specific mutagenesis without phenotypic selection. Methods Enzymol 154: 367-382. 103. Black, M. E. & Loeb, L. A. (1996) Random sequence mutagenesis for the generation of active enzymes. Methods Mol Biol 57: 335-349. 104. Kokoris, M. S. & Black, M. E. (2002) Characterization of herpes simplex virus type 1 thymidine kinase mutants engineered for improved ganciclovir or acyclovir activity. Protein Sci. 11: 2267-2272. 105. Hruby, D. E. & Ball, L. A. (1981) Cell-free synthesis of enzymatically active vaccinia virus thymidine kinase. Virology 113: 594-601. 106. Bennett, M. S., Wien, F., Champness, J. N., Batuwangala, T., Rutherford, T., Summers, W. C., Sun, H., Wright, G. & Sanderson, M. R. (1999) Structure to 1.9 Å resolution of a complex with herpes simplex virus type-1 thymidine kinase of a novel, non-substrate inhibitor: X-ray crystallographic comparison with binding of aciclovir. FEBS Lett 443: 121-125. 107. Champness, J. N., Bennett, M. S., Wien, F., Visse, R., Summers, W. C., Herdewijn, P., de Clerq, E., Ostrowski, T., Jarvest, R. L. & Sanderson, M. R. (1998) Exploring the active site of herpes simplex virus type-1 thymidine kinase by X-ray crystallography of complexes with aciclovir and other ligands. Proteins 32: 350-361. 108. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M. et al. (1998) Crystallography & NMR system: A new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crystallogr 54 ( Pt 5): 905-921. 109. McRee, D. E. (1999) XtalView/Xfit--A versatile program for manipulating atomic coordinates and electron density. J Struct Biol 125: 156-165. 110. Merritt, E. A. & Bacon, D. J. (1997) Raster3D: Photorealistic molecular graphics. Methods Enzymol 277: 505-524. 111. Mercer, K. E., Ahn, C. E., Coke, A., Compadre, C. M. & Drake, R. R. (2002) Mutation of herpesvirus thymidine kinase to generate ganciclovir-specific kinases for use in cancer gene therapies. Protein Eng 15: 903-911. 112. Larder, B. A., Derse, D., Cheng, Y. C. & Darby, G. (1983) Properties of purified enzymes induced by pathogenic drug-resistant mutants of herpes simplex virus. Evidence for virus variants expressing normal DNA polymerase and altered thymidine kinase. J Biol Chem 258: 2027-2033. 113. Culver, K. W., Ram, Z., Wallbridge, S., Ishii, H., Oldfield, E. H. & Blaese, R. M. (1992) In vivo gene transfer with retroviral vector-producer cells for treatment of experimental brain tumors. Science 256: 1550-1552.
113
114. Black, M. E., Kokoris, M. S. & Sabo, P. (2001) Herpes simplex virus-1 thymidine kinase mutants created by semi-random sequence mutagenesis improve prodrug-mediated tumor cell killing. Cancer Res 61: 3022-3026. 115. Edelstein, M. L., Abedi, M. R., Wixon, J. & Edelstein, R. M. (2004) Gene therapy clinical trials worldwide 1989-2004-an overview. J Gene Med 6: 597-602. 116. Moolten, F. L. & Wells, J. M. (1990) Curability of tumors bearing herpes thymidine kinase genes transferred by retroviral vectors. J Natl Cancer Inst 82: 297-300. 117. Field, A. K., Davies, M. E., DeWitt, C., Perry, H. C., Liou, R., Germershausen, J., Karkas, J. D., Ashton, W. T., Johnston, D. B. & Tolman, R. L. (1983) 9-([2-hydroxy-1-(hydroxymethyl)ethoxy]methyl)guanine: a selective inhibitor of herpes group virus replication. Proc Natl Acad Sci U S A 80: 4139-4143. 118. Reardon, J. E. (1989) Herpes simplex virus type 1 and human DNA polymerase interactions with 2'-deoxyguanosine 5'-triphosphate analogues. Kinetics of incorporation into DNA and induction of inhibition. J Biol Chem 264: 19039-19044. 119. Bi, W. L., Parysek, L. M., Warnick, R. & Stambrook, P. J. (1993) In vitro evidence that metabolic cooperation is responsible for the bystander effect observed with HSV tk retroviral gene therapy. Hum Gene Ther 4: 725-731. 120. Boehme, R. E. (1984) Phosphorylation of the antiviral precursor 9-(1,3-dihydroxy-2-propoxymethyl)guanine monophosphate by guanylate kinase isozymes. J Biol Chem 259: 12346-12349. 121. Miller, W. H. & Miller, R. L. (1980) Phosphorylation of acyclovir (acycloguanosine) monophosphate by GMP kinase. J Biol Chem 255: 7204-7207. 122. Agarwal, K. C., Miech, R. P. & Parks, R. E., Jr. (1978) Guanylate kinases from human erythrocytes, hog brain, and rat liver. Methods Enzymol 51: 483-490. 123. Gaidarov, I. O., Suslov, O. N. & Abdulaev, N. G. (1993) Enzymes of the cyclic GMP metabolism in bovine retina. I. Cloning and expression of the gene for guanylate kinase. FEBS Lett 335: 81-84. 124. Brady, W. A., Kokoris, M. S., Fitzgibbon, M. & Black, M. E. (1996) Cloning, characterization, and modeling of mouse and human guanylate kinases. J Biol Chem 271: 16734-16740. 125. Stolworthy, T. S., Krabbenhoft, E. & Black, M. E. (2003) A novel Escherichia coli strain allows functional analysis of guanylate kinase drug resistance and sensitivity. Anal. Biochem. 322: 40-47.
114
126. Kokoris, M. S., Sabo, P. & Black, M. E. (2000) In vitro evaluation of mutant HSV-1 thymidine kinases for suicide gene therapy. Anticancer Res. 20: 959-963. 127. Akyurek, L. M., Nallamshetty, S., Aoki, K., San, H., Yang, Z. Y., Nabel, G. J. & Nabel, E. G. (2001) Coexpression of guanylate kinase with thymidine kinase enhances prodrug cell killing in vitro and suppresses vascular smooth muscle cell proliferation in vivo. Mol Ther 3: 779-786. 128. Nishiyama, T., Kawamura, Y., Kawamoto, K., Matsumura, H., Yamamoto, N., Ito, T., Ohyama, A., Katsuragi, T. & Sakai, T. (1985) Antineoplastic effects in rats of 5-fluorocytosine in combination with cytosine deaminase capsules. Cancer Res 45: 1753-1761. 129. Kanamaru, R., Kakuta, H., Sato, T., Ishioka, C. & Wakui, A. (1986) The inhibitory effects of 5-fluorouracil on the metabolism of preribosomal and ribosomal RNA in L-1210 cells in vitro. Cancer Chemother Pharmacol 17: 43-46. 130. Pinedo, H. M. & Peters, G. F. (1988) Fluorouracil: biochemistry and pharmacology. J Clin Oncol 6: 1653-1664. 131. Thomas, D. M. & Zalcberg, J. R. (1998) 5-fluorouracil: a pharmacological paradigm in the use of cytotoxics. Clin Exp Pharmacol Physiol 25: 887-895. 132. Yao, L., Li, Y., Wu, Y., Liu, A. & Yan, H. (2005) Product release is rate-limiting in the activation of the prodrug 5-fluorocytosine by yeast cytosine deaminase. Biochemistry 44: 5940-5947. 133. Hayden, M. S., Linsley, P. S., Wallace, A. R., Marquardt, H. & Kerr, D. E. (1998) Cloning, overexpression, and purification of cytosine deaminase from Saccharomyces cerevisiae. Protein Expr Purif 12: 173-184. 134. Ireton, G. C., Black, M. E. & Stoddard, B. L. (2003) The 1.14 Å crystal structure of yeast cytosine deaminase: evolution of nucleotide salvage enzymes and implications for genetic chemotherapy. Structure (Camb) 11: 961-972. 135. Kurtz, J. E. & Black, M. E. (2004) Enhancement of suicide gene prodrug activation by random mutagenesis. Methods Mol Med 90: 331-344. 136. Dantas, G., Kuhlman, B., Callender, D., Wong, M. & Baker, D. (2003) A large scale test of computational protein design: folding and stability of nine completely redesigned globular proteins. J Mol Biol 332: 449-460. 137. Suzuki, M., Christians, F. C., Kim, B., Skandalis, A., Black, M. E. & Loeb, L. A. (1996) Tolerance of different proteins for amino acid diversity. Mol Divers 2: 111-118. 138. Weiner, K. X., Weiner, R. S., Maley, F. & Maley, G. F. (1993) Primary structure of human deoxycytidylate deaminase and overexpression of its functional protein in Escherichia coli. J Biol Chem 268: 12983-12989.
115
139. Yang, C., Carlow, D., Wolfenden, R. & Short, S. A. (1992) Cloning and nucleotide sequence of the Escherichia coli cytidine deaminase (ccd) gene. Biochemistry 31: 4168-4714. 140. Barton, K. N., Paielli, D., Zhang, Y., Koul, S., Brown, S. L., Lu, M., Seely, J., Kim, J. H. & Freytag, S. O. (2006) Second-generation replication-competent oncolytic adenovirus armed with improved suicide genes and ADP gene demonstrates greater efficacy without increased toxicity. Mol Ther 13: 347-356. 141. Bohman, C. & Eriksson, S. (1988) Deoxycytidine kinase from human leukemic spleen: preparation and characteristics of homogeneous enzyme. Biochemistry 27: 4258-4265. 142. Iwasaki, H., Huang, P., Keating, M. J. & Plunkett, W. (1997) Differential incorporation of ara-C, gemcitabine, and fludarabine into replicating and repairing DNA in proliferating human leukemia cells. Blood 90: 270-278. 143. Baker, C. H., Banzon, J., Bollinger, J. M., Stubbe, J., Samano, V., Robins, M. J., Lippert, B., Jarvi, E. & Resvick, R. (1991) 2'-Deoxy-2'-methylenecytidine and 2'-deoxy-2',2'-difluorocytidine 5'-diphosphates: potent mechanism-based inhibitors of ribonucleotide reductase. J Med Chem 34: 1879-1884. 144. Hughes, T. L., Hahn, T. M., Reynolds, K. K. & Shewach, D. S. (1997) Kinetic analysis of human deoxycytidine kinase with the true phosphate donor uridine triphosphate. Biochemistry 36: 7540-7547. 145. Johansson, M., van Rompay, A. R., Degreve, B., Balzarini, J. & Karlsson, A. (1999) Cloning and characterization of the multisubstrate deoxyribonucleoside kinase of Drosophila melanogaster. J Biol Chem 274: 23814-23819. 146. Yamada, Y., Goto, H. & Ogasawara, N. (1982) Deoxyguanosine kinase from human placenta. Biochim Biophys Acta 709: 265-272. 147. Harrison, P. T., Thompson, R. & Davison, A. J. (1991) Evolution of herpesvirus thymidine kinases from cellular deoxycytidine kinase. J Gen Virol 72 (Pt 10): 2583-2586. 148. Koonin, E. V. & Senkevich, T. G. (1993) Fowlpox virus encodes a protein related to human deoxycytidine kinase: further evidence for independent acquisition of genes for enzymes of nucleotide metabolism by different viruses. Virus Genes 7: 289-295. 149. Balasubramaniam, N. K., Veerisetty, V. & Gentry, G. A. (1990) Herpesviral deoxythymidine kinases contain a site analogous to the phosphoryl-binding arginine-rich region of porcine adenylate kinase; comparison of secondary structure predictions and conservation. J Gen Virol 71 ( Pt 12): 2979-2987.
116
150. Brown, D. G., Visse, R., Sandhu, G., Davies, A., Rizkallah, P. J., Melitz, C., Summers, W. C. & Sanderson, M. R. (1995) Crystal structures of the thymidine kinase from herpes simplex virus type-1 in complex with deoxythymidine and ganciclovir. Nat Struct Biol 2: 876-881. 151. Jordheim, L. P., Galmarini, C. M. & Dumontet, C. (2006) Gemcitabine resistance due to deoxycytidine kinase deficiency can be reverted by fruitfly deoxynucleoside kinase, DmdNK, in human uterine sarcoma cells. Cancer Chemother Pharmacol: 1-8. 152. Miyazaki, K. & Takenouchi, M. (2002) Creating random mutagenesis libraries using megaprimer PCR of whole plasmid. Biotechniques 33: 1033-1034, 1036-1038. 153. Usova, E. V. & Eriksson, S. (1997) The effects of high salt concentrations on the regulation of the substrate specificity of human recombinant deoxycytidine kinase. Eur. J. Biochem. 248: 762-766. 154. Usova, E. V. & Eriksson, S. (2002) Identification of residues involved in the substrate specificity of human and murine dCK. Biochem. Pharmacol. 64: 1559-1567. 155. Karlsson, A., Johansson, M. & Eriksson, S. (1994) 2 cloning and expression of mouse deoxycytidine kinase. Pure recombinant mouse and human enzymes show differences in substrate specificity. J. Biol. Chem. 269: 24374-24378. 156. Hapke, D. M., Stegmann, A. P. & Mitchell, B. S. (1996) Retroviral transfer of deoxycytidine kinase into tumor cell lines enhances nucleoside toxicity. Cancer Res 56: 2343-2347. 157. Lumniczky, K., Szatmari, T., Huszty, G., Desaknai, S., Sasvari-Szekely, M., Staub, M. & Safrany, G. (2005) Combined treatment of experimental gliomas with radiotherapy, radiosensitizing and chemosensitizing gene therapy. 13th European Cancer Conference, Paris, France. 158. Wang, J., Chattopadhyaya, J. & Eriksson, S. (1999) The enantioselectivity of the cellular deoxynucleoside kinases. Nucleosides & Nucleotides 18: 807-810. 159. Maturin, L. J., Sr. & Curtiss, R., 3rd (1981) Role of ribonucleic acid synthesis in conjugational transfer of chromosomal and plasmid deoxyribonucleic acids. J Bacteriol 146: 552-563.