Understanding and Improving the Anticancer Activity of
Cisplatin
by
Qing He
B.A., Chemistry
Franklin and Marshall College, 1995
SUBMITTED TO THE DEPARTMENT OF CHEMISTRY IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY IN BIOLOGICAL CHEMISTRY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2001
© Massachusetts Institute of Technology, 2001
All rights reserved
Signature of Author:
._
Depawnent of Chemistry
November 9, 2000
Certified by:
Ste en J. Lippard
Arthur Amos Noyes Professor of Chemistry
Thesis Supervisor
Accepted by:
Robert W. Field
Chairman, Departmental Committee on Graduate Studies
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
MAR 0 1 2001
LIBRARIES
2
This doctoral thesis has been examined by a Committee of the Department of
Chemistry as follows:
.h-•1
E. Essigmarn
Professor of Chemistry and Toxicology
Committee Chairman
Ste hen. Lippard
Arthur Amos Noyes Chair and Professor of Chemistry
Thesis Supervisor
Carl O. Pabo
Professor of Biology
UNDERSTANDING AND IMPROVING THE
ANTICANCER ACTIVITY OF CISPLATIN
By
Qing He
Submitted to the Department of Chemistry on Nov 2000
In Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy in Chemistry
ABSTRACT
The purpose of this thesis is to further our understanding of the mechanism of
action of cis-diamminedichloroplatinum(II)(cisplatin), one of the most effective
anticancer drugs. Since its serendipitous discovery in 1970, cisplatin has served to help
cure testicular cancer and treat a variety of human malignancies. It is widely accepted
that DNA is the cellular target for cisplatin. Prior to this work, several structures of
duplex DNA modified by cisplatin revealed the distinctive distortions caused by
cisplatin-DNA adducts. High mobility group (HMG) domain proteins are DNAbinding proteins that bind to cisplatin-modified DNA in vitro with high specificity and
affinity. HMG-domain proteins block nucleotide excision repair of cisplatin-DNA
adducts in vitro, suggesting that such proteins may mediate cisplatin cytotoxicity in
cells.
The structure of HMG1 domain A bound to site-specifically cisplatin modified
DNA reveals an unprecedented protein-DNA binding mode and a key phenylalanine
side-chain intercalation. Factors contributing to the affinity of HMG-domain proteins
for cisplatin-modified DNA are not well understood. In Chapter 2 is described a
biochemical approach to evaluate the contribution of intercalating residues to the
affinity of HMG-domain proteins for platinated DNA. Site-directed mutagenesis,
bandshifts and footprinting methods show that the position of the side-chain
intercalator determines the protein binding mode. This study provides a new paradigm
to understand why and how HMG domains interact with platinated DNA.
In addition to understanding the molecular basis of protein platinated-DNA
interaction, the role of HMG-domain proteins in the cisplatin mechanism was
investigated on the cellular level. Overexpression of HMG1 had been predicted to
enhance the sensitivity of mammalian cells to cisplatin. Previous attempts from our
laboratory and others failed to overexpress HMG1 stably in cells. When it was reported
that HMG1 mRNA is upregulated in mammalian breast cancer MCF-7 cells after
estrogen treatment, the effects of steroid hormone treatment on HMG1 protein
expression and cisplatin sensitivity in mammalian cell lines from breast and ovarian
tumors were studied. The ability to modulate cisplatin sensitivity in cells has useful
clinical implications such as enhancing the efficacy of cisplatin chemotherapy. The
results of this study led to a phase I clinical trial to investigate the efficacy of hormonecarboplatin combination therapy for treatment of ovarian cancer patients.
It can concluded from Chapters 2 and 3 respectively, that the affinity of HMG
domains for cisplatin-modified DNA can be improved by protein modifications and
that the cytotoxicity of cisplatin can be enhanced by HMG1 overexpression. Because
cisplatin lesions are not natural targets for HMG domain proteins, the protein-DNA
binding affinity may not be optimal. It is of interest to design novel proteins to be used
in gene therapy for further improvement of the therapeutic effects of cisplatin in
patients. In order to achieve this goal, the phage display method was employed to
search for novel HMG domain proteins with higher affinity for cisplatin-modified DNA
than those naturally occurring. It was successfully demonstrated that HMG-domains
can be expressed on the phage surface, and protocols were established to enable
selection for cisplatin-damaged DNA targets based on DNA structure rather than
sequence. Chapter 4 sets the foundation for future phage display protocols to design
proteins of high affinity for cisplatin-modified DNA.
Thesis Supervisor: Stephen J. Lippard
Title: Arthur Amos Noyes Professor of Chemistry
This thesis is dedicated to my family.
Acknowledgments
There are many people who have helped shape my scientific career prior to MIT
as well as many who supported, encouraged, and inspired me to be my best during my
graduate years. I am very grateful to each of them. First of all, I must thank my
research advisor, Stephen J. Lippard. He is truly an amazing scientist, with an
incredible amount of enthusiasm for science, a constant push for excellence, and an
unparalleled capacity and efficiency for work. He has assembled a fantastic lab
environment in which I am fortunate to work. Steve has been supportive throughout
my graduate career and has helped me become a better scientist, for which I will be
always grateful.
I have received a great deal of helpful advice and suggestions from other labs. I
thank my thesis committee members, Professor John M. Essigmann and Professor Carl
O. Pabo, for sharing their expertise in the steroid hormone, phage display and related
biological fields with me, discussing my research progress and offering tremendous
insights and helpful suggestions. I thank Professor Richard Hynes and Jane Trevithick
for their help with the immunofluorescence experiments.
I am grateful for the opportunity to work closely with MIT undergraduate
Cynthia H. Liang. She is one of the brightest, most dedicated, and efficient people I
have met, while keeping balance in life with so many interests: violin, piano, dancing,
tennis, cooking...... As her "UROP mentor" for 2 years, I hope she has learned as
much from me as I have learned from her! She has become such an amazing scientist
and I am sure she will succeed in graduate school and whatever she does in the future.
The Lippard lab is like an extended family. I have been fortunate to work with
many exceptional people and I have enjoyed both the scientific interactions and lasting
friendships, past, present, and future. When I first joined the lab, many people taught
me techniques and also became good friends. Deborah Zamble and Karen Sandman are
both excellent scientists. They taught me how to run a gel and how to label DNA as
well as many other things! I am sure Deborah will become a very successful professor
soon. Karen and Betsy are good friends and their daily cat stories and "pilgrimage"
definitely made the afternoons more fun! Stephanie Kane, Andy Gelasco, and Shari
Dunham and many others made the platinum bay feel like a family. Uta-Maria
Ohndorf is not only a great labmate with whom to work and collaborate, but also my
tennis partner, and we took many tennis classes together in the summers - lots of fun!
Sudharkar Marla has been my baymate for the last 3 years. His calm and
extremely careful way of doing things definitely taught me a lot! I thank him and his
wife Chandrika for the friendship and Chandrika in particular for being my "fashion
consultant"! Krista Kneip is a postdoc from Germany and I really enjoyed the many
late-night conversations in the reading room and her constant stock of cookies and
snacks! Amanda Yarnell is so good at everything she puts her mind to, and she leads a
very balanced life with lots of volunteer works. She also has a keen sense of justice. She
has helped me tremendously by carefully reading and editing my thesis chapters. She
will be successful at whatever she may choose to do in the future. Seth Cohen and Yuji
Mikata are two other great platinum baymates who have been great collaborators and
friends.
Min Wei is an excellent postdoc who recently arrived in the lab, just in time to
lend her expertise in Molscript to help me make an important figure in a paper. We
have become good friends. Dong Wang and Joey Bautista are two very smart second
year students, and they make the biobay more lively - especially Joey after coffee! I
wish Dong, Min and Joey great success! There are many other people in lab who I have
enjoyed interaction with, including Natasha, Jessica, Amy, Chuan, and Dirk and many
others. Bryan Wang from Professor Pabo's lab had many discussions with me about
phage display problems and showed me some techniques in phage display.
I want to thank my many friends, some students at MIT or other schools, others
friends outside of the science world. They are my tennis or skating partners and friends
who I can share a good hike/meal/laugh with. Their diverse experience remind me to
maintain perspective and balance in life.
I thank my mother for everything she gave, taught and sacrificed for me. She
showed me how to be a good person, how to live optimistically, how to love and help
those around her. She had touched so many people's lives and she forever will be my
role model. I am glad that I fulfilled her wish that I get a Ph.D. I miss you and you
always will be in my heart.
I thank my dad and my stepmother for their unconditional love and support
over the years. They are always so proud of me and I hope I will live up to their
expectations. I also want to thank my in-laws and my other relatives for their care and
love.
To my husband, Zhenqing: your constant love, support, understanding, and
optimism have brightened my days throughout the years. You always encourage me
and believe in me and I thank you for who you are and everything you have done for
me. You are my better half and my best friend, you complete me. Having you as my
life companion is the best thing that has happened to me.
8
Table of Contents
3
Ab stract ............................................................................................................................
5
Dedication .................................................................................................................
6
A cknow ledgm ents ....................................................................................................
8
Table of C ontents ......................................................................................................
............................. 12
List of T ables ...............................................................................
............................... 13
List of Figures............................................................................
G lossary of Term s .................................................... ................................................ 15
Chapter 1.
............................. 17
In trodu ction ................................................................................
18
DNA is the Biological Target of Cisplatin ...................................................
Cellular Proteins that Bind Cisplatin-Modified DNA................................... 20
Potential Roles of HMG-Domain Proteins in
Mediating Cisplatin Cytotoxicity..........................................22
............................... 25
R eferen ces ..................................................................................
Chapter 2.
Intercalating Residues Determine the Mode of HMG1 Domains A
and B Binding to Cisplatin-Modified DNA..........................................................39
.................................. 40
...................................................
Introduction ...................
M aterials and M ethods..................................................................................................41
Site-Directed Mutagenesis ..............................................................................
41
Circular Dichroism..........................................................42
Gel Mobility Shift Assays ............................................................................ 42
D ata Analysis.................................................. .............................................. 42
43
Hydroxyl Radical Footprinting Assays .........................................................
..... 43
Photocross-linking with HMG1 Domains.....................
Results ....... ................................................................................................ . ......... 44
44
The Domain A and Domain B Mutants .........................................................
Contribution of Helix I-II Spacing and Loop Length to
Platinated DNA Protein Binding Affinity .............................................. 44
Contribution of Protein-Base Hydrogen Bonding to
Platinated DNA Protein Binding Affinity.....................................................45
Contribution of Intermolecular Stacking at Position 37 to
.... 45
Platinated DNA Protein Binding Affinity ....................................
An Intercalating Residue at Position 16 Influences DNA Affinity ............. 46
Position of the Intercalator Determines the Binding Orientation ............... 47
Comparison of HMG Domain Intercalating Sites ..................................... 48
Sequence Selectivity of Mutant HMG1 Domains A and B .......................... 49
.... 49
Photocross-linking with Mutant Proteins .....................................
D iscu ssion .................................................................................................................... 51
Factors Affecting Protein Affinity for Cisplatin-Modified DNA ................ 51
Sequence Selectivity of Mutant HMG-Domain Proteins ............................. 52
Factors that Govern the Photocross-linking Efficiency ................................... 52
53
The DNA Binding Mode.....................................
....... 54
......
The Recognition Mechanism .....................................
Implications for Full Length HMG1-DNA Binding .................................... 56
A cknow ledgem ents ...................................................... ........................................... 57
R eferen ces ............................................................... .................................................. 58
Chapter Three.
The Effect of Steroid Hormones on HMG1 Protein Levels and Platinum Sensitivity in
78
M am m alian Cell Lines ...........................................................................................
................................................. 79
Introduction .............................................................
M aterials and M ethods................................................... ......................................... 82
Cell C ultures ...................................................... ........................................... 82
.................... 82
Clonogenic Assays .........................................................
Western Analysis for HMG1 Protein Levels ..................................... ... 83
Immunofluorescence to Detect HMG1 Protein Levels.................................83
Double Labeling and Immunofluorescence of
84
M ammalian Cancer Cells ...................................................................................
DA PI A ssay ....................................................... ............................................ 85
................... 85
Apoptosis TUNEL Assay................................................
Mammalian Genomic DNA Rapid Preparation .............................................. 86
32
P Post-Labeling Methods to Detect Platinum-DNA Adducts ...................... 86
Results and Discussion ........................................................................ .................... 87
HMG1 Overexpression in Breast Cancer MCF-7 and Evsa-T Cells ............ 87
Sensitization of Cells to Cisplatin by Estrogen and Progesterone .............. 88
Effects of Hormone Concentration on the Degree of Sensitization ............ 89
Timing of the Hormone Treatment and Sensitivity
10
Toward Cisplatin and Carboplatin........................................90
Tamoxifen and Estrogen..................................................91
Cell Proliferation and Sensitivity to Other Cytotoxic Agents ..................... 92
Platinum -DNA Adduct Levels.......................................................................93
94
Apoptosis Assays .............................................................
Clinical Implications of the Hormone/Carboplatin
Combination Therapy ......................................................... 95
Detection of CA125 Expression on the Surface of
........ 97
Ovarian Cancer Cells in Vitro ...........................................
99
Conclusions and Future Directions ....................................................
Acknow ledgm ents .................................................. ............................................... 101
.............................. 102
R eferen ces .................................................................................
Chapter 4.
The Search for Novel HMG-Domain Proteins with High Affinity toward CisplatinM odified D NA by Phage D isplay .................................................................... ..... 129
.................................. 130
Introduction ..........................................................................
Materials and Methods...........................................................133
134
M aterials .. ......................................................................................................
................................ 134
M ethods ..................................................................................
Cloning of the pHDB Phagemid ................................................................... 134
Site-Directed Mutagenesis to Mutate the Amber
Stop Codon TAG in pHDB ..................................................... 135
...... 135
VCS-M13 Helper Phage Preparation ......................................
Titering the Phage .......................................................... 135
136
.......................
pHDB Phage Preparation ..................
Electron Microscopy of the Phage Particles ........................................ 137
Gel Mobility Shift Assays of pHDB Phage with DNA Probes ................... 137
pHDB Phage Panning ....................................................... 138
....... 139
Cloning of the pHDA Phagemid ........................................
....... 139
DNA Targets for pHDA Phage Panning ...............................
......... 140
Phage Panning with pHDA...........................
Multivalent and Monovalent Phage ................................................................ 141
Site-Directed Mutagenesis to Incorporate
Tw o BbsI Sites in pH DA2 ............................................................................. 141
........ 142
Constructing Phagemid Library ........................................
11
Cloning of HMG1 DomA into pcDNA3.1 ................................................... 144
Cloning of HMG1 DomB into pcDNA6/V5His ...................................... 144
Overexpression of HMG1 Domains in HeLa Cells ..................................... 145
Results and D iscussion ....................................................................... ................... 145
.... 145
pH DB Phage ....................................................................................
Gel Mobility Shift Assays of pHDB Phage with DNA...............................145
Panning of the pHDB Phage ........................................................................ 146
Cloning of the pHDA Phagemid ................................................................ 146
Multivalent and Monovalency ................................................ 146
Bacterial Strain ................................................... ......................................... 147
Size of the DN A Target ........................................................................... ...... 148
150
pHDA2 Control Phage Panning..............................
pH DA Phage Library ............................................ ..................................... 151
155
Overexpression of HMG1 DomA and DomB .....................................
156
Conclusions and Future Directions ..................................... ............
HMG1 Domain A Phage Bind Cisplatin-Modified
.................... 156
DNA Preferentially .......................................................
Intrinsic Problems with HMG Domain Phage Libraries......................... 156
..... 157
Potential of Alternative Phage Libraries ....................................
A cknow ledgm ents ...................................................... ........................................... 157
................................................. 158
R eferen ces..............................................................
Biographical N ote ............................................................................
...................... 176
List of Tables
Table 1.1. Selected structural features of cisplatin-DNA adducts.........................30
Table 1.2. Structural parameters for X-ray and NMR solution structures of DNA
duplexes containing the 1,2-intrastrand d(GpG) cisplatin adduct ......... 31
Table 1.3. Proteins that bind cisplatin-modified DNA...........................................32
Table 2.1. Affinities of HMG1 domain A and domain B mutants toward
63
cisplatin-m odified DNA ..........................................................................
Table 2.2. Codes and sequences for 15 bp probes containing 7-deaza-dA,
...... 64
7-deaza-dG and dG-rich sequences .....................................
Table 3.1. List of estrogen-regulated genes or proteins
107
involved in cell growth.............................. ..................
Table 3.2. Cell cycle profiles of MCF-7 cells after hormone treatments .............. 108
Table 4.1. Retention efficiency of the pHDB phage after the panning...................160
Table 4.2. Retention efficiency of the pHDA2 phage after the panning................161
...... 162
Table 4.3. pHDA phagemid library sequence.....................
Table 4.4 pHDA phage library sequence before and after panning ................... 163
List of Figures
Figure 1.1.
Figure 1.2.
Figure 1.3.
Figure 1.4.
Figure 1.5.
Figure 1.6.
Figure 2.1.
Figure 2.2.
Figure 2.3.
Figure 2.4.
Figure 2.5.
Figure 2.6.
Figure 2.7.
Figure 2.8.
Figure 2.9.
Figure 2.10.
Figure 2.11.
Figure 3.1.
Figure 3.2.
Figure 3.3.
Figure 3.4.
Figure 3.5.
Figure 3.6.
Structures of some platinum complexes investigated
33
for biological activity ...............................................................................
N7 positions of the purine bases where platinum binds .................. 34
The structures of cisplatin 1,2-intrastrand DNA adducts.....................35
Schemes of nucleotide excision repair.................................................36
Structures of HMG domains, free or complexed with DNA ............37
Models of how HMG-domain proteins may mediate
38
cisplatin cytotoxicity ..................................... ...............
Structures of HMG1 domain A bound to cisplatin-modified DNA ....65
Sequence of HMG domain proteins and cisplatin-modified DNA ...... 66
Circular dichroism spectra of HMG1 domain A and
................ 67
domain B mutants ..................................... .......
Gel mobility shift assays of domain A and domAF37A .................... 68
The potential steric crowding between domA F37W and DNA..........79
... 70
Hydroxyl radical footprint analysis .....................................
Structural diagram of deoxyadenosine monophosphate (AMP),
deoxyguanosine monophosphate (GMP), 7-deaza-dA
monophosphate (AMP), and 7-deaza-dG monophosphate (GMP) ..... 73
Bandshift experiments showing the flanking sequence
preferences of mutant HMG-domain proteins ...................................... 74
Denaturing polyacylamide gel demonstrating the photocross-linking
reaction of platinated 15-bp DNA with HMG1 domain B protein ...... 75
Photocross-linking experiment with mutant HMG-domain proteins.76
Comparison of two binding modes of HMG domains bound to
....... 77
cisplatin-modified DNA .................................... ....
Chemical structures of the steroid hormones ................................... 109
Hormone receptor mechanism, free and bound and HMG1
............................ 110
involvement ....................................
HMG1 mRNA levels in estrogen treated MCF-7 and Evsa-T cells ....111
Western blots analysis of MCF-7 cells treated with estrogen .......... 112
Immunofluorescence of MCF-7 cells treated with estrogen and
progesterone ........................................... ......................................... 113
Immunofluorescence of Evsa-T cells treated with estrogen and
114
progesterone ........................................................................................
Figure 3.7.
Immunofluorescence of BG-1 cells treated with estrogen and
progesterone .......................................................... 115
Figure 3.8. Colony counting assays of MCF-7, Evsa-T, BG-1 and HeLa cells......116
Figure 3.9. Hormone concentration affects the cisplatin sensitivity..................117
Figure 3.10 Hormone treatment schedule affects the cisplatin sensitivity............118
Figure 3.11. Colony counting assays of MCF-7 cells to carboplatin ..................... 119
Figure 3.12. Colony counting assays of MCF-7 cells with estrogen and tamoxifen
............. 120
......
treatm ent ..................................... ...........
Figure 3.13. Immunofluorescence of MCF-7 cells with estrogen and tamoxifen
............. 121
...........
treatment ........................................................
Figure 3.14 Proliferation of MCF-7 cells with hormone treatment ..................... 122
Figure 3.15. Colony counting assays of MCF-7 cells to trans-DDPand
123
calicheamicin .................................... .....................
Figure 3.16. 32 P post-labeling to detect Pt-DNA adducts.....................124
Figure 3.17. DAPI staining for apoptosis in HeLa and MCF-7 cells .................... 125
Figure 3.18. TUNEL based FACS to detect apoptosis in MCF-7 cells .................. 126
Figure 3.19. Immunofluorescence of CA125 staining in BG-1 cells ...................... 127
128
Figure 3.20. Immunofluorescence of CA125 staining in OVCAR-3 cells ........
Figure 4.1. Cartoon of filamentous phage ............................................................ 165
Figure 4.2.
Figure 4.3.
Figure 4.4.
Figure 4.5.
Figure 4.6.
Figure 4.7.
Figure 4.8.
Figure 4.9.
Figure 4.10.
Scheme of phage panning...................................................................166
Maps of pZif and pHDA and pHDB .................................................. 167
Visualization of phage under electron microscopy...................... 168
Gel mobility shift assay of pHDB phage and cisplatinm odified DNA ........................................................................................ 169
Genomic DNA digestion ................................................. 170
Size of the DNA targets affects phage panning...................................171
172
pHDA2 control phage panning results ........................................
Design of pH DA library ..................................................................... 173
Phage retention efficiency vs. selection cycles...................................... 174
Figure 4.11. Western blot analysis of HeLa cells transfected with domains A and B
overexpression vector ......................................................................... 175
1,2-d(GpG)
1,2-d(ApG)
4WJ
bp
BSA
cis-DDP
cisplatin
domA
domB
CD
DTT
EDTA
E. coli
ER
ERE
FPLC
HEPES
Histone H1
HMG
HMG1
HR
HRE
tsHMG
HPLC
hsp
IPTG
Ixrl
Kd
LEF-1
P-ME
n. a.
n. d.
n. t.
PAGE
PBS
Abbreviations
cis-[Pt(NH 3)2{d(GpG)-N7(1)-N7(2)} ]
cis-[Pt(NH3)2{d(ApG)-N7(1)-N7(2)}]
4-way junction DNA
base pair
bovine serum albumin
cis-diamminedichloroplatinum(II)
cis-diamminedichloroplatinum(II)
HMG1 domain A
HMG1 domain B
circular dichroism
dithiothreitol
ethylenediaminetetraacetic acid
Escherichiacoli
estrogen receptor
estrogen responsive element
fast protein liquid chromatography
N-(2-hydroxyethyl)piperazine-N'-(2-ethanesulfonic acid)
A linker histone that serves as a chromatin architectural protein
high-mobility group
high-mobility group protein 1
hormone receptor
hormone responsive element
testis-specific high-mobility group protein
high performance liquid chromatography
heat shock protein
isopropyl-p-D-thiogalactopyranoside
intrastrand cross-link recognition protein
apparent dissociation constant
lymphoid enhancer binding factor 1
P-mercaptoethanol
not applicable
not determined
nucleotide
polyacrylamide gel electrophoresis
phosphate buffered saline
PEG
PR
XPA
YT
polyethyleneglycol
progesterone receptor
progesterone responsive element
root-mean-square deviation
replication protein A
sodium dodecyl sulfate
sex determining region Y protein
TATA binding protein
trans-diamminedichloroplatinum(II)
tris(hydroxymethyl)aminomethane
Xeroderma pigmentosum
Xeroderma pigmentosum group A protein
cell growth media containing 16 g trypton, 10 g yeast extract and 5
YTG
g NaC1 per liter
YT cell growth media containing 10 g / liter of D-glucose
PRE
rmsd
RPA
SDS
SRY
TBP
trans-DDP
Tris
XP
Chapter One
Introduction
cis-Diammindichloroplatinum(cisplatin), known also as Peyrone's chloride, was
first synthesized in 1845 (1). Its cell growth inhibitory properties were discovered
serendipitously by Barnett Rosenberg in 1965 during a study of the effect of an electric
field on the growth of bacteria (2). Rosenberg's discovery led to subsequent clinical
trials, demonstrating the antitumor activity of cisplatin. Cisplatin first was used to treat
genitourinary tumors in 1979 and has made testicular cancer curable for over 90% of the
cases (3).
Cisplatin has since been used to treat other types of tumors, including
ovarian, cervical, head and neck, and nonsmall cell lung cancer (4-7).
Despite the great success of cisplatin in the clinic, the mechanism of its anticancer
activity is incompletely understood. Both toxic side effects and intrinsic or acquired
tumor resistance limit the usage of the drug (8). New platinum compounds have been
investigated in order to overcome these clinical limitations. Of the > 3000 compounds
tested, only cisplatin and carboplatin have been approved by the FDA, and JM216 is
still in clinical trials (Figure 1.1) (8). The structure of the inactive compound trans-DDP,
an isomer of cisplatin, is also shown in Figure 1.1. Greater understanding of detailed
mechanism of cisplatin will help to design novel and more effective platinum drugs as
well as to improve current therapy.
DNA is the Biological Target of Cisplatin. Cisplatin is administered by intravenous
injection. After the neutral compound diffuses through the cell membrane, the lower
chloride concentration inside the cell (- 2 - 30 mM) leads to the formation of mono- or
diaquadiammineplatinum(II) complexes. These species readily react with DNA, RNA
and proteins. DNA is the cytotoxic target of cisplatin in cells (9, 10). DNA synthesis is
selectively inhibited at a cisplatin dosage lower than that affecting protein and RNA
synthesis (11, 12). In a separate study, the number of platinum atoms bound to DNA,
RNA, and proteins in HeLa cells treated with l95mPt-radiolabeled cisplatin at its mean
lethal concentration were determined (13). The results indicated that nine platinum
atoms are bound per DNA molecule (about 2x105 bp), but only one platinum atom per 3
x 10' to 3 x 10' protein molecules and 1 platinum atoms per 10 to 1000 RNA molecules.
Other evidence, including that DNA repair-deficient E. coli and human cells are more
sensitive to cisplatin treatment compared to wild-type cells, further supports the
conclusion that DNA is the primary target for cisplatin (14-20).
Platinum covalently cross-links the N7 atoms of the DNA purine bases (Figure
1.2). Platinum-DNA adducts include 1,2-intrastrand cross-links, 1,3-intrastrand crosslinks, interstrand cross-links, monofunctional adducts, and protein-DNA cross-links.
The major products, demonstrated by enzymatic digestion of cisplatin-treated DNA, are
the 1,2-intrastrand cross-links. The species cis-[Pt(NH3)2{d(GpG-N7(1),N7(2))}] and cis[Pt(NH 3) 2{d(ApG-N7(1),N7(2))}] account for = 65% and = 20% of the total, respectively
(21).
Such adducts can not be formed by trans-DDP,the clinically inactive isomer,
because of steric constraints. Therefore, 1,2-intrastrand cross-links have been our focus
for understanding the mechanism of cisplatin.
The square-planar coordination geometry of platinum distorts cisplatin-modified
DNA. Structural features of the cisplatin-DNA adducts have been revealed by many
biochemical and biophysical techniques, including gel electrophoresis, NMR and X-ray
crystallography methods. Table 1.1 lists the structural features of different types of
cisplatin-DNA adducts (8). Differences between the adducts include DNA bending and
unwinding angles and the location of the platinum atom. The crystal structure of a
cisplatin-cross-linked dinucleotide cis-[Pt(NH3)2{d(GpG)}] reveals that the two guanine
bases are almost perpendicular to each other so as to accommodate the platinum planar
coordination (Figure 1.3 A) (22). Later, structures of duplex DNA containing a sitespecific 1,2-intrastrand cisplatin adduct were solved by NMR and crystallography
(Figure 1.3 B and C) (23, 24). The DNA is distinctively bent towards the major groove,
whereas the minor groove is wide and shallow. It should be noted, however, that there
are some differences between the two structures, possibly due to the influence of crystal
packing. Table 1.2 summarizes the features of several X-ray and NMR structures of the
major cisplatin adduct on duplex DNA (8).
Cellular Proteins that Bind Cisplatin-Modified DNA. DNA distortions caused by
platinum adducts interfere with normal cellular functions, including replication and
transcription (8, 25, 26). Such damage on the DNA is recognized by cellular repair
complexes.
Both nucleotide excision repair (NER) and mismatch repair (MMR)
pathways have been implicated in the processing of cisplatin-DNA adducts.
In the mammalian NER pathway, the cisplatin-DNA adduct is excised as a 24 to
32-mer single-stranded DNA fragment and DNA is synthesized to fill the resulting gap
(27-29). Many proteins are involved in this repair process (Figure 1.4). Xeroderma
pigmentosum (XP) patients have defects in NER and are hypersensitive to DNA
damage, such as that caused by UV light. Cell lines from XP patients are deficient in
certain components of the excision repair complex (28, 29) and are highly sensitive to
cisplatin treatment, implicating NER in cisplatin damage processing (Figure 1.4) (8, 3032). In vitro repair assays demonstrate that platinum-DNA adducts can be repaired by
NER (33). Furthermore, 1,2-intrastrand cross-links are repaired less efficiently than 1,3intrastrand adducts (34).
Mismatch repair deficient cell lines are resistant to cisplatin (35, 36), suggesting
that cisplatin adducts may also be processed by MMR. During mismatch repair, the
strand to be corrected is nicked, and the mismatch excised followed by new DNA
synthesis.
MutSca, a protein involved in mismatch recognition, and one of its
components MSH2, both selectively bind to cis-GG adducts (37, 38).
Besides repair proteins, many other proteins bind cisplatin-modified DNA.
Table 1.3 lists some of these proteins and their natural functions where known (8).
Among these proteins is a category called high mobility group (HMG)-domain proteins.
SSRP1 was the first HMG-domain protein found to bind specifically to cisplatinmodified DNA (39).
The HMG domain is a DNA binding motif of about 80 amino acids. HMGdomain proteins commonly are associated with chromatin and recognize DNA by
either sequence or structure (40, 41). Sequence-specific HMG domains are generally
transcription factors or cellular differentiation regulators, contain a single domain, and
are cell-type specific. Examples include the testis-determining factor SRY, lymphoid
enhancer binding factor LEF-1, and the Sox proteins. Sequence-neutral HMG-domain
proteins generally have no known recognition sequence, contain more than one
domain, and maintain chromatin structure or serve as architectural proteins. Examples
include HMG1 and HMG2. HMG-domain proteins from both categories bind the major
cisplatin-DNA adducts, supercoiled DNA and four way junctions, but not DNA
modified with trans-DDPand other inactive compounds (42-44).
Sequence homology between HMG domains is quite low (25%), but tertiary
structure is highly conserved (45). HMG domains share a common structure of a
twisted L-shape with three helices, as demonstrated by several NMR determinations
(Figure 1.5) (46-48). The structure of the HMG domain complexed with its recognition
target has been solved for SRY and LEF-1 (49, 50). These sequence-specific HMG
domains bind to minor groove of the DNA through its concave surface. Both proteins
use a hydrophobic residue to intercalate between two DNA base steps, forming a
distinctive bend of 1170 for LEF1 DNA and 70-80' for SRY DNA (49, 50). In both cases,
the DNA backbone is severely bent and unwound, resembling that of DNA modified by
cisplatin. Structures of the sequence-neutral HMG-domain proteins drosophila HMG-D
and yeast NHP6A bound to linear DNA are available (51, 52). These two proteins bend
linear DNA by way of multiple intercalating residues.
The structure of HMG1 domain A bound to a site-specific 1,2-intrastrand
cisplatin-DNA cross-links reveals the molecular basis for HMG domains interaction
with cisplatin-modified DNA (53). The DNA is bent by about 610 and a Phe37 side
chain intercalates into the hydrophobic notch in the minor groove created by cisplatin
coordination in the major groove (Figure 1.5) (53). The protein is positioned
asymmetrically relative to the major bend locus, representing an unprecedented
binding mode. The contribution of intercalating residues and other factors that affect
the binding affinity and positioning of the protein with respect to its DNA target is the
focus for Chapter 2.
Potential Roles of HMG-Domain Proteins in Mediating Cisplatin Cytotoxicity. HMG
domains bend specifically to the major cisplatin-DNA adducts, forming a stable
platinum-DNA-protein ternary complex. There are several hypotheses for how HMGdomain proteins can subsequently mediate cisplatin cytotoxicity.
One hypothesis the hijacking model (Figure 1.6). Cisplatin-DNA adduct sites
may titrate HMG domains away from their natural binding sites. hUBF binds with
similar affinity to the major cisplatin-DNA adducts as to its cognate target sequences
(54). Such titration could disrupt gene regulation and may cause cell death. In addition,
cisplatin-modified DNA is recognized by the 3-methyladenine DNA glycosylase (AAG)
family of mammalian repair proteins. Binding of AAG to cisplatin damage inhibits the
excision of 1,N6-ethenoadenine, a natural target of AAG (56), presumably by way of a
titration mechanism.
An alternative, but not exclusive, hypothesis is repair-shielding. Platinum-DNAprotein complexes can block nucleotide excision repair of the DNA damage, leading to
cell death or arrest (Figure 1.6). HMG domains can block NER of site-specific cisplatinmodified DNA in vitro (57, 58). In addition, depleting HMG1 and HMG2 from cell
extracts by immunoprecipitation enhances nucleotide excision repair of cisplatinmodified DNA (59). Knockout of the yeast HMG-domain protein Ixrl led to a 2-6 fold
decrease in cisplatin sensitivity, suggesting that Ixrl can sensitize cells by blocking
repair of cisplatin-modified DNA (60, 61). The decrease in cisplatin sensitivity is
attributed to damage recognition and the formation of the excision repair complex (60).
A recent kinetic study shows that HMG1 domains binds to cisplatin-damaged DNA
with rates near the diffusion limit, consistent with no other nuclear proteins binding
faster (62). Therefore, HMG domains may compete for and block access to the 1,2intrastrand cross-links recognized by repair proteins. A competition binding study
with both HMG1 and the repair protein RPA demonstrates that HMG1 binds to
cisplatin-modified DNA and can block formation of the RPA complex (63).
The repair shielding hypothesis predicts that higher levels of HMG-domain
proteins should block NER more efficiently and lead to higher cisplatin sensitivity.
Previous attempts to upregulate HMG1 using overexpression vectors have been
unsuccessful (Zamble, D.B., unpublished results). Chapter 3 describes a novel way to
upregulate HMG1 in cancer cells via a natural signal transduction pathway.
Mammalian cell lines with appropriate steroid hormone receptors upregulate HMG1
following hormone treatment. The effect of HMG1 overexpression on the cisplatin
sensitivity of cells is also investigated. If cisplatin sensitization in cell lines can be
achieved, the efficacy of platinum anti-cancer drugs in clinical settings may be
enhanced.
Our search for improved cisplatin therapy continues in Chapter 4. We propose
that novel proteins may be engineered to exhibit higher affinity and specificity for
cisplatin-modified DNA, because HMG-domain proteins have not evolved to bind
cisplatin-modified DNA. Such novel HMG domains, if transfected into cells, may be
able to block the repair of platinum-DNA adducts more efficiently than the natural
proteins, thereby potentiating cisplatin sensitivity. Proteins sensitizing cancer cells to
cisplatin will have clinical application by way of gene therapy. To screen millions of
HMG domain mutants for platinated-DNA binding properties, a combinatorial method
is more desirable than the conventional site-directed mutageneisis. In Chapter 4, a
phage display protocol is designed to search HMG-domain libraries for peptides with
high affinities and specificities for cisplatin-modified DNA.
References:
1. Peyrone, M. (1845) Ann. Chem. Pharm. 51, 1-29.
2. Rosenberg, B., VanCamp, L., and Krigas, T. (1965) Nature 205, 698-699.
3. Bosl, G. J., and Motzer, R. J. (1997) New Eng. J. Med. 337, 242-253.
4. Loehrer, P. J., and Einhorn, L. H. (1984) Ann. Intern. Med. 100, 704-713.
5. Morris, M., Eifel, P. J., Lu, J., Grigsby, P. W., Levenback, C., Stevens, R. E., Rotman,
M., Gershenson, D. M., and Mutche, D. G. (1999) The New England Journal of Medicine
340, 1137-1143.
6. Rose, P. G., Bundy, B. N., Watkin, E. B., Thigpen, J. T., Deppe, G., Maiman, M. A.,
Clarke-Pearson, D. L., and Insalaco, S. (1999) The New England Journalof Medicine 340,
1144-1153.
7. Keys, H. M., Bundy, B. N., Stehman, F. B., Muderspach, L. I., Chafe, W. E., Suggs, C.
L., Walker, J. L., and Gersell, D. (1999) The New England Journal of Medicine 340, 11541161.
8. Jamieson, E. R., and Lippard, S. J. (1999) Chem. Rev. 99, 2467-2498.
9. Bruhn, S. L., Toney, J. H., and Lippard, S. J. (1990) Prog.Inorg. Chem. 38, 477 - 516.
10. Brown, S. J., Chow, C. S., and Lippard, S. J. (1994) in Encyclopedia of Inorganic
Chemistry (King, R. B., Ed.) pp 3305-3315, John Wiley $ Sons Ltd, West Sussex, England.
11. Harder, H. C., and Rosenberg, B. (1970) Int. J. Cancer 6, 207-216.
12. Howle, J. A., and Gale, G. R. (1970) Biochemical Pharmacology19, 2757-2762.
13. Akaboshi, M., Kawai, K., Maki, H., Akuta, K., Ujeno, Y., and Miyahara, T. (1992)
Jpn. J. CancerRes. 83, 522-526.
14. Beck, D. J., and Brubaker, R. R. (1973) J. Bacteriology 116, 1247-1252.
15. Drobnik, J., Urbinkova, M., and Krekulova, A. (1973) Mutation Research 17, 13-20.
16. Markham, B., and Brubaker, R. R. (1980) J. Bacteriol. 143, 455-462.
17. Brouwer, J., van de Putte, P., Fichtinger-Schepman, A. M. J., and Reedijk, J. (1981)
Proc.Natl. Acad. Sci. USA 78, 7010-7014.
18. Beck, D. J., Popoff, S., Sancar, A., and Rupp, W. D. (1985) Nucleic Acids Res. 13, 73957412.
19. Fram, R. J., Cusick, P. S., Wilson, J. M., and Marinus, M. G. (1985) Mol. Pharmacol.28,
51-55.
20. Popoff, S. C., Beck, D. J., and Rupp, W. D. (1987) Mutation Research 183, 129-137.
21. Fichtinger-Schepman, A. M. J., van der Veer, J. L., den hartog, J. H. J., Lohman, P. H.
M., and Reedijk, J. (1985) Biochemistry 24, 707-713.
22. Sherman, S. E., Gibson, D., Wang, A. H.-J., and Lippard, S. J. (1985) Science 230, 412417.
23. Takahara, P. M., Rosenzweig, A. C., Frederick, C. A., and Lippard, S. J. (1995) Nature
377, 649 - 652.
24. Gelasco, A., and Lippard, S. J. (1998) Biochemistry 37, 9230-9239.
25. Corda, Y., Job, C., Anin, M.-F., Leng, m., and Job, D. (1991) Biochemistry 30, 222-230.
26. Harder, H. C., Smith, R. G., and Leroy, A. F. (1976) Cancer Res. 36, 3821-3829.
27. Zamble, D. B., and Lippard, S. J. (1999) in Cisplatin-Chemistryand Biochemistry of a
Leading Anticancer Drug (Lippert, B., Ed.) pp 73-110, Verlag Helvetica Chimica Acta,
Zurich.
28. Wood, R. D. (1997) J. Biol. Chem. 272, 23465.
29. Sancar, A. (1996) Annu. Rev. Biochem. 65, 43.
30. Fraval, H. N. A., Rawlings, C. J., and Roberts, J. J. (1978) Mutat. Res. 51, 121-132.
31. Plooy, A. C. M., van Dijk, M., Berends, F., and Lohman, P. H. M. (1985) CancerRes.
45, 4178-4184.
32. Dijt, F. J., Fichtinger-Schepman, A. M. J., Berends, F., and Reedijk, J. (1988) Cancer
Res. 48, 6058-6062.
33. Hansson, J., and Wood, R. D. (1989) Nucleic Acids Res. 17, 8073-8091.
34. Szymkowski, D. E., Yarema, K., Essigmann, J. M., Lippard, S. J., and Wood, R. D.
(1992) Proc. Natl. Acad. Sci. U.S.A. 89.
35. Fink, D., Nebel, S., Aebi, S., Zheng, H., Cenni, B., Nehme, A., Christen, R. D., and
Howell, S. B. (1996) CancerRes. 56, 4881.
36. Aebi, S., Fink, D., Gordon, R., Kim, H. K., Zheng, H., Fink, J. L., and Howell, S. B.
(1997) Cancer Res. 3, 1763.
37. Duckett, D. R., Drummond, J. T., Murchie, A. I. H., Resardon, J. T., Sancar, A.,
Lilley, D. M. J., and Modrich, P. (1996) Proc.Natl. Acad. Sci. U.S.A. 93, 6443.
38. Mello, J. A., Acharya, S., Fishel, R., and Essigmann, J. M. (1996) Chem. & Biol. 3, 579589.
39. Bruhn, S. L., Pil, P. M., Essigmann, J. M., Housman, D. E., and Lippard, S. J. (1992)
Proc. Natl. Acad. Sci. 89, 2307-2311.
40. Bustin, M., and Reeves, R. (1996) Prog.Nucleic Acid Res. Mol. Biol. 54, 35-100.
41. Read, C. M., Cary, P. D., Crane-Robinson, C., Driscoll, P. C., Carrillo, M. O. M., and
Norman, D. G. (1995) in Nucleic Acids and Molecular Biology (Eckstein, F., and Lilley, D.
M. J., Eds.) pp 222-250, Springer-Verlag, Berlin.
42. Whitehead, J. P., and Lippard, S. J. (1996) in Metal Ions in Biological Systems (Sigel, A.,
and Sigel, H., Eds.) pp 687-725, Marcel Dekker, Inc., New York.
43. Lilley, D. M. J. (1992) Nature 357, 282-283.
44. Grasser, K. D., Teo, S. H., Lee, K. B., Broadhurst, R. W., Rees, C., Hardman, C. H.,
and Thomas, J. 0. (1998) Eur. J. Biochem. 253, 787-795.
45. Baxevanis, A. D., and Landsman, D. (1995) Nucleic Acids Res. 23, 1604-1613.
46. Hardman, C. H., Broadhurst, R. W., Raine, A. R. C., Grasser, K. D., Thomas, J. O.,
and Laue, E. D. (1995) Biochemistry 34, 16596-16607.
47. Jones, D. N. M., Searles, M. A., Shaw, G. L., Churchill, M. E. A., Ner, S. S., Keeler, J.,
Travers, A., and Neuhaus, D. (1994) Structure 2, 609-627.
48. Weir, H. M., Kraulis, P. J., Hill, C. S., Raine, A. R. C., Laue, E. D., and Thomas, J. O.
(1993) EMBO J. 12, 1311-1319.
49. Love, J. J., Li, X., Case, D. A., Giese, K., Grosschedl, R., and Wright, P. E. (1995)
Nature 376, 791 - 795.
50. Werner, M. H., Huth, J. R., Gronenborn, A. M., and Clore, G. M. (1995) Cell 81, 705714.
51. Allain, F. H. T., Yen, Y. M., Masse, J. E., Schultze, P., Dieckmann, T., Johnson, R. C.,
and Feigon, J. (1999) EMBO J. 18, 2563-2579.
52. Murphy IV, F. V., and Churchill, M. E. A. (2000) Structure 8, R83-R89.
53. Ohndorf, U. M., Rould, M. A., He, Q., Pabo, C. 0., and Lippard, S. J. (1999) Nature
399, 708-712.
54. Treiber, D. K., Zhai, X., Jantzen, H.-M., and Essigman, J. M. (1994) Proc. Nat. Acad.
Sci. USA 91, 5672 - 5676.
55. Trimmer, E. E., Zamble, D. B., Lippard, S. J., and Essigmann, J. M. (1997)
Biochemistry 37, 352-362.
56. Kartalou, M., Samson, L. D., and Essigmann, J. M. (2000) Biochemistry 39, 8032-8038.
57. Zamble, D. B., Mu, D., Reardon, J. T., Sancar, A., and Lippard, S. J. (1996)
Biochemistry 35, 10004-10013.
58. Huang, J., Zamble, D. B., Reardon, J. T., Lippard, S. J., and Sancar, A. (1994) Proc.
Nat. Acad. Sci. USA 91, 10394 - 10398.
29
59. Li, L., Liu, X., Glassman, A. B., Keating, M. J., Stros, M., Plunkett, W., and Yang, L.
(1997) CancerResearch 57, 1487-1494.
60. McA'Nulty, M. M., and Lippard, S. J. (1996) Mutat. Res. 362, 75-86.
61. McA'Nulty, M. M., Whitehead, J. P., and Lippard, S. J. (1996) Biochemistry 35, 60896099.
62. Jamieson, E. R., and Lippard, S. J. (2000) Biochemistry 39, 8426-8438.
63. Patrick, S. M., and Turchi, J. J. (1998) Biochemistry 37, 8808-8815.
Table 1.1. Selected Structural Features of Cisplatin-DNA Adducts. Reprinted with
permission from Jamieson, E.R. and Lippard, S.J., Chem. Rev., (1999), 99 2467-2498.
Unwinding
Pt
Average
Bend
Method
DNA Sequence
Site
Helical
Angle
Twista
:
d (CCTCTG*G'TCTCC)*
X-ray
39-550
320
Major
n.d.
Groove
d (GGAGAC C AGAGG)
d (CCTCTG*G*TCTCC) d (GGAGAC C AGAGG)
NMR
780
270
Major
Groove
n.d.
d (CCTG*G*TCC).
d (GGAC C AGG)
NMR
580
2 5 ob
Major
Groove
210
d (GACCATATG*G*TC) d (GACCATATG*G*TC)
NMR
400
n.d.
Major
Groove
n.d.
Major
Groove
n.d.
d (CTCTCG'G'TCTC).
d (GAGACC G AGAG)
Paramagnetic
NMR
NMR and
Molecular
Modeling
550
n.d.
Major
Groove
n.d.
d (CTCTAG*TG*CTCAC)d (GTGAGC AC TAGAG)
NMR
20-240
n.d.
Major
Groove
190
d (CCTCG*CTCTC)d (GAGAG*CGAGG)
X-ray
470
n.d.
Minor
Groove
700
d (CATAG*CTATG)
d (CATAG*CTATG)
NMR
200
n.d.
Minor
Groove
400
NMR
d (CCTCG*CTCTC)TC
d (GAGAGTCGAGG)
aValues not determined are denoted as n.d.
n.d.
Minor
Groove
d (CTCA*G*CCTC).
d (GAGG CT GAG)
760
Table 1.2. StructuralParameters for X-ray and NMR Solution Structures of DNA
Duplexes Containing the 1,2-Intrastrand d(GpG) Cisplatin Adduct. Reprint with
permission from Jamieson, E.R. and Lippard, S.J., Chem. Rev., (1999), 99: 2467-2498.
Parameter
X-ray
X-ray
NMR
NMR
NMR
DNA Length
DNA Form
Minor Groove
12-bp
16-bp
12-bp
8-bp
11-bp
A/B Junction
B
B
B
B
9.5-11.0 A
5.5-12.0 A
4.5-7.8 A
9.0-12A
3.0 A
n.a.
1.4 A
3.2 A
2.1 A
A
n.a.
6.8 A
6.8 A
6.8 A
9.4-12.5
A
Width
Minor Groove
Depth
Average P-P
5.5
Distance
Dihedral Angle
Between
Platinated Bases
300
750
470
590
580
Average Helical
Twist
320
330
270
250
260
810
580
780
610
390 and 550
DNA Bendb
not available are denoted as n.a. bValues determined with the program Curves.
aValues
Table 1.3. Proteins that Bind to Cisplatin-Modified DNA. Adapted from Jamieson, E.R.
and Lippard, S.J., Chem. Rev., (1999), 99: 2467-2498..
Protein
Function
Repair Proteins
XPA (Xeroderma pigmentosum
A complementing protein)
Damage recognition protein in nucleotide excision
repair
RPA (Replication protein A,
single stranded binding protein)
Damage recognition protein in nucleotide excision
repair
XPC-HR23B
Damage recognition proteins in nucleotide excision
repair
MutScL, MSH2
Recognition component in mismatch repair
Ku Autoantigen (DNA-PK)
Takes part in V(D)J recombination and
double-strand break repair
HMG-Domain Proteins
Human SSRP1
Involved in transcription elongation
Drosophila SSRP1
Unknown
HMG1
Unknown
HMG2
Unknown
Ixrl
Transcription factor that regulates Cox5b promoter
HMG-D
Drosophilahomolog of HMG1
hUBF
Ribosomal RNA transcription factor
tsHMG
Unknown
SRY
Sex determining factor
Other Proteins
TATA binding protein
Part of the basal transcription factor TFIID
YB-1
Transcription factor
Histone H1
Linker histone
aValues not determined are denoted as n.d.
H3 N
/CI
Pt
H3 N
tCI
cisplatin
carboplatin
H3 N\ /CI
Pt
Cl /
JM216
NH3
trans-DDP
Figure 1.1. Structures of some platinum compounds.
O
I
Ii
minor groove
Figure 1.2. Cytosine-guanine and thymine-adenine Watson-Crick base pairs. Arrows
indicate the purine N7 position where platinum forms covalent adducts.
NH3
NH
3
Figure 1.3. Structures of cisplatin-modified DNA. (A) X-ray crystal structure of the cisplatin d(pGpG) dinucleotide cross-link, from Sherman et Science, (1985), 23(Q41241,7. Carbon atoms are shaded in light gray and heteroatoms are shaded in dark gray.
The drug is bound to the dinucleotide by the N7 nitrogen atoms of the guanine bases.
The bases are bound in a head-to-head oreitnation. (B and C) Diagrams of the NMR
solution structure (B) and X-ray structure (C) of cisplatin-modified DNA duplex, taken
from Gelasco and Lippard, Biochemistry, 1998, 37, 9230-9239 and Takahara et al, Nature,
(1995), 377:649-652. The overall structures are similar.
H3N \pt
H3N
.G
G
H3 N
Damage recogntion
by XPA, RPA, and
XPC-HR23B (C).
H3N
p
H3N
H3 N
TFIIH binds
and forms preincision
complex.
XPG binds and
makes 3'
incision.
H3N
4
H3 N
excised.
damage
DNA
Gap filled in and closed
by polymerases and ligases.
Process is PCNA dependent.
XPF(F)-ERCC1(1)
bind and make
5' incision.
Figure 1.4. Schematic diagram of nucleotide excision repair of a cisplatin-DNA
cross-link. DNA-damage is recognized by XPA, RPA, and XPC-HR23B (C). TFIIH
binds forming a preincision complex. XPG makes the 3' incision, and the 5' incision
is made by XPF-ERCC1 ((F) and (1), respectively). Once the damage is excised, the
DNA is filled in by polymerases and ligases in a PCNA dependent process. This
figure is reprinted from Jamieson, E.R. and Lippard, S.J., Chem. Rev., (1999), 99: 24672498.
A
B
He
Figure 1.5. Structures of free and DNA bound HMG domains. (A) NMR structure of HMG1
domain B. This figure is made from Weir et al, EMBO 1.,(1993), 12:1311-1319. (B) NMR structure
of LEF-1 complexed with its natural binding site. This figure is adapted from Love, et al, Nature
(1995) 376: 791-795 . (C) X-ray crystal structure of HMG1 domain A bound to cisplatin-modified
DNA. This figure is adapted from Ohndorf, et al. Nature(1999) 399:708-712.
DNA Damage Recognition
P
>
H3 N
Pto
H3N '
G
G
H3 N
H3 N
HMG-Domain
Protein Binds
Blocking Repair
Repair Co
Factors
H3N
H3 N
Hijacking of HMGDomain Protein
on of DNA
Figure 1.6. Schematic diagram of the hijacking and repair shielding hypothesis
proposed for mediation of cisplatin cytotoxicity by HMG-domain proteins. Reprinted
from Jamieson, E.R. and Lippard, S.J., Chem. Rev., 1999, 99:(2467-2498).
39
Chapter 2
Intercalating Residues Determine the Mode of HMG1 Domains A
and B Binding to Cisplatin-Modified DNA
Introduction (1-4)
cis-Diamminedichloroplatinum(II)(cisplatin) is a widely used anticancer drug that
forms covalent adducts on DNA (5-7). The major cisplatin-DNA adducts, 1,2intrastrand cross-links, form distinctive bends and distortions in the DNA, which are
recognized by a variety of cellular proteins including high-mobility group (HMG) domain proteins (8, 9). HMG-domain proteins bind specifically to the major cisplatinDNA adducts and form stable platinum-DNA-protein ternary complexes in vitro. Considerable evidence implicates the involvement of such complexes in cisplatin cytotoxicity (10-14). To delineate further the cisplatin mechanism of action and design better cancer therapies, it is important to understand thoroughly the interactions between HMGdomain proteins and cisplatin-modified DNA.
HMG-domain proteins are usually associated with chromatin (15, 16) and can be
categorized as sequence-specific or sequence-neutral (17). Sequence-specific HMG domains include transcription factors such as LEF-1 and SRY that are cell-type specific.
Sequence-neutral HMG-domain proteins, such as HMG1, HMG2, HMG-D, and
NHP6A, are present in many cell types and recognize distorted DNA independent of
sequence (17). Both subfamilies share a minor groove DNA-binding motif and bind to
distorted structures such as cisplatin-modified DNA (18), supercoiled DNA (19), and
four-way junctions (20).
HMG1 is an architectural protein with two tandem HMG domains, A and B. Each
domain alone specifically recognizes the major cisplatin-DNA adducts. The X-ray crystal structure of a 16-base-pair double-stranded deoxyoligonucleotide containing a single
cis-[Pt(NH 3)2{d(GpG)-N7(Gs)-N7(Gg)}] intrastrand cross-link in complex with HMG1
domain A has been described (Figure 2.1) (21). A striking feature is the intercalation of
the Phe 37 side chain into a hydrophobic notch in the minor groove formed by the drug-
DNA adduct. The opening of the two platinum cross-linked GC base pairs involving
guanine rings Gs and G9 allows for intermolecular stacking interactions between the Phe
37 phenyl ring and the G9 purine heterocycle, interlocking the two components like
pieces in a molecular jigsaw puzzle (Figure 2.1). This X-ray structure determination offered the first insight that cisplatin-DNA cross-links might provide a pre-formed intercalation site for minor groove DNA-binding proteins, a feature that is likely to contribute to the genotoxic effects of the drug. Recent work from our laboratory demonstrated
that TATA-binding protein interacts with cisplatin-modified TATA sequences in a
similar manner (22).
In the present article, we describe the site-directed mutagenesis that reveals the
contributions of specific protein-DNA contacts to the composition and stability of the
complexes. The binding orientations and affinities of HMG1 domain A and domain B
mutants for cisplatin-modified DNA were determined. Gel-mobility shift assays and
hydroxyl radical footprint analysis support a model in which sequence-neutral HMGdomain proteins utilize a two-pronged intercalation mode to bind to pre-bent DNA. We
demonstrate that the intercalating residues are key to how the protein is positioned on
the DNA relative to the platinum adduct. The binding mode can be dramatically
changed with a single side-chain mutation. The results are compared with those provided by the crystal and NMR structures of other HMG domain-DNA complexes (2326).
MATERIALS AND METHODS
Site-DirectedMutagenesis. Mutagenesis of HMG1 domains A and B was carried out
according to the Stratagene QuikChange site-directed mutagenesis kit protocol. Successful mutations were confirmed by DNA sequencing at the MIT Biopolymers Lab.
HMG1 domain A and B mutants were expressed in E. coli BL21(DE3) and purified as
described (27), with the addition of a FPLC size-exclusion purification column (highload Superdex 75, Pharmacia, 1 ml/min, 11.8 mM PBS, pH 7.4). The ability of protein
mutants to fold correctly was confirmed by circular dichroism spectroscopy.
CircularDichroism. CD spectra were recorded on an AVIV 62DS spectrophotometer. Proteins were diluted to 20 pM in 0.05 X PBS buffer. The measurements were made
at 4 'C with a 0.7 nm bandwidth and a 0.5 nm step size. Each spectrum is the average
of four recordings. The raw data were smoothed by using AVIV software. Ellipticities
are based on molar peptide rather than amino acid residue concentrations.
Gel Mobility Shift Assays. The oligonucleotide d(GGTTGGTCCAGAGAGG) was 5'
end-labeled and annealed to its complementary strand d(CCTCTCTG*G*ACCTTCC),
where the asterisks represent a cis-diammineplatinum(II)cross-link. The duplex (0.4
nM) was incubated with increasing concentrations of proteins in 20 pl sample volumes
containing 10 mM HEPES, pH 7.5, 10 mM MgCl2 , 50 mM LiC1, 100 mM NaC1, 1 mM
spermidine, 0.2 mg/ml BSA and 0.05% Nonidet P40. Samples were incubated on ice for
30 min, then made 7% in sucrose and 0.017% in xylene cyanol prior to loading on prerun, pre-cooled (4°C) 10% native polyacrylamide gels (29:1 acrylamide:bis). Gels were
electrophoresed for 3 h and vacuum dried onto Whatman 3 MM chromatography paper. Gels were visualized by using a BioRad phosphorimager and the bands were
quantitated with the BioRad Multi-Analysts software.
Data Analysis.
Apparent dissociation constants, Kd, were estimated from non-
linear least-squares fits of binding data to the Langmuir isotherm, eq 1 (28), where 0 is
the fraction of bound oligonucleotide probe and P is the total protein concentration.
Each Kd is the average of at least two replicates.
0 = P / (P + Kd)
(1)
Hydroxyl Radical FootprintingAssays. Labeled DNA samples with or without protein (20-fold excess protein over DNA) were incubated at 40C for 30 min in the same
buffer described for the gel mobility shift assays. To a 20 pl aliquot was added a fresh
mixture of sodium ascorbate (10 mM), Fe(NH 4)2(SO4) 2-6H 20 (2.5 mM), sodium EDTA (5
mM), and H 20 2 (0.3%) to initiate the reaction. After 5 min at room temperature, 10 FPl of
thiourea (1 M) was added to stop the reaction. DNA was extracted with phenol/chloroform and then precipitated with ethanol. The DNA samples were electrophoresed on 20% denaturing polyacrylamide gels. Gels were dried and quantitated on a
BioRad phosphorimager. The total counts in each lane were used to normalize the
bands. The peak heights corresponding to each band were compared and, when the difference between the control and protein-added lanes was more than 10%, then the corresponding base was scored as being protected. Each area of protection was reproduced
in at least two experiments.
Photocross-Linkingwith HMG1 Domains. The DNA used in the photocross-link ex-
periment is 5'-(CCTCTCTG*G*TTCTTC)-3' with 3'(GGAGAGACCAAGAAG)5' annealed. Reaction solutions for photcross-linking experiments were prepared in Thermowell TubesT (Costar) under exactly the same conditions as gel-mobility shift assays.
The reaction mixtures were incubated on ice for at least 30 min prior to irradiation. A
helium cadmium laser 32301 (Liconix) at 325 nm (2.5 mW) or an argon ion laser Innova
300 (Coherent) at 350 nm (900 mW, unless otherwise described) was employed as the
source. Irradiation was carried out at 40C in a quartz container maintained at ice-water
temperature. After irradiation, 5 pl1 of gel loading buffer (10 M urea, 1.5 mM EDTA,
005% (w/v) bromophenol blue and xylene cyanol) was added and the mixture was
heated to 90 0 C for 2 min, followed by quick cooling at ice temperature. The product
was separated by denaturing 8% PAGE (1.0 mm thickness) in 1X TBE (90 mM Tris, 90
mM boric acid, and 2.5 mM EDTA, pH 8.3) for 1.5 h at 300 V with cooling. The gels
were dried, visualized, and analyzed as described above. Data points in all figures represent the result of at least two (typically three) independent experiments.
RESULTS
The Domain A and Domain B Mutants. Figure 2.2A presents the sequence alignment
of HMG1 domains A and B together with several other HMG domains. The numbering
scheme is based on that of HMG1 domain A unless specified otherwise. The sequence
of the site-specifically platinated DNA used in this study is shown in Figure 2.2B.
A variety of HMG1 domain A and domain B mutant proteins were expressed and
purified. Circular dichroism spectroscopy showed that these mutants fold similarly to
the native protein, with high c-helical content (Figure 2.3). The affinities of the HMG1
domain A and B mutants for site-specifically cisplatin-modified DNA were investigated
by gel mobility shift assays. The Kd values are listed in Table 1.
Contribution of Helix I-II Spacing and Loop Length to Platinated-DNAProtein Binding
Affinity. Sequence alignment (Figure 2.2A) shows that the HMG1 domain A has two
amino acids, V35 and N36, inserted in the loop connecting helices I and II that are not
present in HMG1 domain B and most other HMG-domain proteins (29). In order to determine whether these two amino acids are critical for protein binding, a deletion mutant of HMG1 domain A lacking V35 and N36, domA AVN, was designed. In a complementary experiment these two amino acids were inserted in HMG1 domain B, domB
iVN. DomA AVN has a 3.3-fold lower affinity for platinated DNA compared to domA
(Table 1). The affinity of domB iVN, however, is comparable to that of the wild-type
domain B. These results suggest that the length of the intervening loop between helix I
and II is not a determining factor of the platinated DNA binding affinity in these HMG
domains.
Contribution of Protein-Base Hydrogen Bonding to Platinated-DNA Binding Affinity.
The crystal structure reveals only a single direct protein-base contact. Ser41 forms a hydrogen bond with the N3 atom of Ao
1 , the base directly adjacent to the cisplatin-DNA
cross-link. It had been suggested prior to the X-ray structure determinations that the
lower affinity of HMG1 domain B for cisplatin-modified DNA could be attributed to the
lack of such an interaction (30). Domain B has an alanine residue at the equivalent position. Indeed, the S41A mutant of HMG1 domain A exhibits a 5-fold decrease in affinity
for the cisplatin-modified substrate (Table 1).
Contributionof IntermolecularStacking at Position 37 to Platinated-DNA Binding Affinity. The crystal structure shows that Phe37 intercalates at the site of the cisplatin crosslink. It interacts through a
t-i
stacking interaction with the cisplatin-modified G 9 purine
ring and in an edge-to-face ring manner with the G8 base. This intercalation stabilizes
the unprecedented, large inter-base dihedral angle of 750 at the site of cisplatinmodification by compensating for the missing base stacking interaction of the 5' side of
the G9-C 24 base pair. Furthermore, it reduces the area of solvent accessibility of the
opened purine and pyrimidine ring faces of those base pairs by -100 A2. The ability of
HMG1 domain A to bind to the cisplatin-modified oligonucleotide is significantly diminished in a domF37A mutant, as revealed by gel mobility-shift analysis (Figure 2.4).
Since not all sequence-neutral HMG-domain proteins utilize an aromatic residue
for intercalation at position 37, mutagenesis experiments were designed to assess the
extent to which favorable -ntstacking interactions at this position contribute to binding
affinity. Replacing Ile37 in domain B with Phe increased the protein affinity toward cis-
platin-modified DNA by two-fold (Table 2.1), consistent with the conclusion that the
intercalating Phe37 residue contributes to the higher affinity of HMG1 domain A for
platinated DNA as compared to domain B. Trp has a larger aromatic ring and may
form more favorable n-t stacking interactions than Phe. The domA F37W mutant, however, displays diminished binding affinity (Table 2.1), presumably because of steric
crowding between the exocyclic oxo atom of the Cs base and the indole ring of Trp
(Figure 2.5).
An IntercalatingResidue at Position 16 Influences DNA Affinity. Since the N-termini
of both helices I and II lie within the minor groove, a potential intercalating residue
could be positioned within either helix. The sequence-specific HMG-domain proteins
LEF-1 and SRY intercalate into DNA by inserting Met or Ile, respectively, from helix I.
These residues occupy positions equivalent to that of 16 in domain A (31). Moreover, in
other sequence-neutral HMG domain-DNA complexes (23-26) there are multiple intercalators situated at the N-termini of helices I and II. At position 16, HMG1 domain A
has an alanine that is ill-suited to intercalate; instead, it forms a hydrophobic contact
with the ribose ring of T2. Loss of the Phe37 intercalation, as in the DomA F37A mutant,
drastically diminishes DNA binding affinity (21). The DomA A16F F37A double mutant
having a single potential intercalating residue at position 16, partially restores binding
affinity (Table 2.1). The domA A16F mutant, which has two potential intercalators,
however, displays decreased binding affinity possibly due to steric crowding of Phel6
with the adjacent Phe residues.
HMG1 domain B has two potential intercalating residues, Phel6 and Ile37. Mutation of either one to alanine decreases the affinity for cisplatin-modified DNA by about
2 to 2.5-fold (Table 2.1). The domain B double mutant, F16A I37A, has severely diminished platinated-DNA affinity, similar to that of the domain A F37A mutant. Despite
the presence of two intercalating residues, HMG1 domain B binds less well to cisplatinmodified DNA than domain A. These results confirm that favorable aromatic interactions provided by Phe37 and the overall complementarity of the protein and DNA
binding surfaces are both important determinants of binding affinity.
The Position of the IntercalatorDetermines the Binding Orientation. With a choice of
two different positions for an intercalator, two different binding modes of the HMG
domain with respect to the platination site are conceivable. These modes were probed
by hydroxyl radical footprinting experiments. Figure 2.6 A shows detailed footprinting
analysis of various mutants in which the bottom strand of the probe was labeled. Corresponding footprints for labeled, platinated-top strand gave similar results (Figure 2.6 B).
Two different protection patterns are observed: a symmetric footprint, in which the
protein-binding site extends to both sides of the platination site; or an asymmetric footprint, in which the binding site extends towards the 3' side of the platinum adduct. Regions of protection from the hydroxyl radical cleavage reactions are shown schematically in Figure 2.6 C. Both the general area of the footprint and the protection pattern
are clear and reproducible. Figure 2.6 C shows the more inclusive footprint pattern in
those cases where the areas protected is one base different between top or bottom
strand footprint. In proteins having a single intercalator at position 37, as in wild type
HMG1 domain A and domB F16A, an asymmetric footprint covering 5-6 bases is observed. On the other hand, the presence of a single intercalator at position 16, as occurs
for domA A16F F37A and domB I37A, converts the footprint to a symmetric pattern
extending over 7-8 bases. The footprint of domain B, which has two potential intercalators, covers about 8 base pairs and is centered around the platinum-adduct (Figure 2.6).
By contrast, the footprint of the domA A16F mutant, which has two potential intercalators, is asymmetric and similar to that of domain A.
Comparison of HMG Domain IntercalationSites. Minor groove binding of sequencespecific and sequence-neutral HMG-domain proteins is governed by intercalation of
one or more hydrophobic residues between base pairs of the DNA duplex. The sequence-specific HMG-domain proteins LEF-1 and SRY employ an aliphatic Met or Ile
residue at position 16, resulting in a DNA bend with its center two base pairs removed
from the major kink site in the HMG1 domain A complex with cisplatin-modified DNA
(21). Despite the disparate positioning of the protein with respect to the bend, comparison of the DNA structures in the proximity of these intercalation sites indicates a close
conformational similarity with an rmsd value of 2.4 A. The minor groove width in both
complexes reaches a value of - 11.5 A.
By contrast, the sequence-neutral HMG-D protein induces DNA deformations
through intercalation of multiple side chains. As in the sequence-specific LEF-1-DNA
complex, a Met residue at position 16 is employed. An additional double intercalation
of residues Va132 and Thr33 occurs at the site equivalent to position 37 in the sequenceneutral HMG1 domain A complex. Superposition of the latter sites reveals a dramatic
difference in DNA structure manifest by an rmsd value of 4.5
A. Despite the
difference
in protein positioning, the DNA structures at the intercalation sites of Met16 in HMG-D
and Phe37 in HMG1 domain A more closely resemble each other with a rmsd value of
2.7
A.
From this similarity, it is predicted that HMG-D binding to cisplatin-modified
DNA would occur mostly through Metl6 intercalation similar to Phe37 in domain A,
and that the Val32/Thr33 residues are not as dominant as Met16. Recently, an NMR
structural study of HMG-D bound to disulfide cross-linked DNA appeared. The structure resembled that of an HMG domain bound to DNA that was not pre-bent (32). This
result is probably due to the fact that cisplatin-modified DNA has a highly localized
bend whereas the disulfide-linked DNA does not (32).
Sequence Selectivity of Mutant HMG1 Domains A and B. The affinity of HMG1 domain A and domain B for cisplatin-DNA adducts is influenced by the sequences flanking the platinum-DNA adducts. In a sequence containing NG*G*N2, where G*G* represents a 1,2-intrastrand cross-link, the affinity of HMG1 domains is modulated for several orders of magnitude by the bases in the N, and N 2 positions (3, 30). Initially, only
nine out of sixteen possible flanking sequences, where N 1, N 2 = A, C, or T were examined due to synthetic difficulties with preparation of site-specifically modified oligonucleotides with more than two guanosines in a row. Recent efforts using unnatural 7deaza-guanosine or 7-deaza-adenine (Figure 2.7) (3), an isostere that lacks the reactive
N7 nitrogen atom in the purine base, successfully completed the study with the remaining seven sequences (3). The sequences used are shown in Table 2.2. The flanking
sequence selectivity of HMG1 domain A is much more dramatic than that of HMG1
domain B. In addition, HMG1 domain A strongly prefers A/T base pairs over C/G
base pairs at the N1 and N 2 position, with N 2 being the dominant position. This preference may be due to the higher flexibility of DNA caused by A/T base pairs (33).
HMG1 mutant proteins provide a means by which to probe the importance of
base-specific contacts in dictating sequence selectivity. The molecular structure of
HMG1 domain A complexed with cisplatin-modified DNA reveals a single base-specific
hydrogen bond contact between S41 and N3 of the neighboring adenine base located to
the 3' side of the G*G* adduct (21). Surprisingly, elimination of this interaction did not
alter the flanking sequence selectivity (Figure 2.8) (3). Mutation of the intercalating
residue from phenylalanine to tryptophan also caused no change in sequence preference.
Photocross-Linkingwith Mutant Proteins. Radiolabeled, platinated DNA duplex and
HMG1 domains form a noncovalent complex when incubated on ice, as evidenced by
gel-electrophoresis (4). Upon laser-irradiation of the complex at 325 nm for 30 min, a
fraction was detected on the denaturing gel with mobility slower than that of the free
probe (Figure 2.9) (4). This band disappeared after treatment with sodium cyanide
(Lane 4), to be replaced by a new band with faster mobility after additional treatment
with proteinase K (lane 5). This species is therefore assigned to a protein-DNA crosslink tethered by platinum-guanine and platinum-protein covalent bonds (34). The absence of such a band in the reaction probe alone (lane 2) further confirms that the protein-DNA photocross-linking reaction occurs only for the specific noncovalent complex
formed between platinated DNA and protein. The efficiency of the photocross-link is
defined as the percent of complex band relative to the total counts of the lane.
HMG1 domain B photocross-links to cisplatin-modified DNA via Lys6 residue
(34). A comparison of the N-terminal regions of HMG1 domain A and domain B reveals that domain A lacks a potential intercalating residue in position 16 in helix I,
where domain B has a phenylalanine (Figure 2.2). Intercalation by this residue alters
the positioning of the HMG1 domain and may help amino acid residues at the Nterminal region to move into the major groove and promote photocross-linking. Intercalation by Phe37 causes the domain to be offset toward the 3'-side of the platinated
strand. To assess the role of amino acid residues that intercalate into the hydrophobic
notch of the platinated DNA base pair, the domain mutants were subjected to laser
photocross-linking. The results (Figure 2.10) (4) show that mutation of Ala16 to Met in
domain A (domA A16M) increases the photocross-linking efficiency by about 4-fold
relative to that of native HMG1 domain A. Substitution of phenylalanine in this position (HMG1 domA A16F) has a small effect, possibly due to steric crowding as suggested from EMSA experiments and molecular modeling. Surprisingly, deletion of
phenylalanine at position 37 does not alter the photocoss-linking efficiency, because the
domA F37A still forms cross-links. In contrast, little interaction is observed in gel mobility shift experiments for this mutant (Figure 2.4). The domain B mutants exhibit negligible sensitivity to the identity of the amino acid residues in both intercalating positions for photocross-linking.
DISCUSSION
Factors Affecting ProteinAffinity for Cisplatin-ModifiedDNA. The affinity of HMG1
domain A for cisplatin-modified DNA is generally higher than that of domain B (30).
Although the helix I-II loop of domain A is longer than that of domain B (Figure 2.2),
the length of this loop is not critical for binding. Moreover, the specific base contacted
by Ser41 of domain A contributes only modestly to the strength of the interaction. The
intercalation mode of Phe37 observed in the crystal structure is important for DNA affinity (21), suggesting that if Ile37 in HMG1 domain B was replaced by a phenylalanine
residue, the enhanced n-interactions would enhance the binding. Indeed, domB I37F
has a higher affinity towards platinated DNA than wild type domain B.
The domA F37A mutant protein has low binding affinity for the platinated DNA
probe (21), owing to the loss of the Phe37 intercalator. The affinity is partially restored
in the A16F F37A mutant, indicating the side chain at position 16 to be a potential intercalator. Instead of abolishing binding affinity, as occurs with domA F37A, mutating either Phel6 or Ile37 in domain B lowers the affinity by only 2-2.5 fold. When both residues are mutated, in domB F16A I37A, the affinity for the platinated probe is very low,
as is the case for domA F37A. These results suggest that either Phel6 or Ile37 can intercalate in the domain B platinated-DNA complex, in support of previous reports on the
contribution of intercalating residues to the affinity of HMG-domain proteins for DNA.
A model of HMG-D binding to pre-bent bulged DNA indicates intercalations by both
Metl3 and Va132, equivalent to positions 16 and 37 in domain A (25). The HMG-D
V32A mutant, which is similar to HMG1 domB I37A, binds to bulged DNA with 5-6
fold lower affinity (25). The affinity of several NHP6A mutants for linear DNA has been
reported, and deleting Met29 or Phe48, equivalent to position 16 or 37 in domain A,
causes a 2- or 4-fold reduction in affinity, respectively (23).
A recent study of HMG1 domain B mutants revealed that domB F16A and domB
I37A bind more tightly to four-way-junction (4WJ) DNA (35). The authors propose that
these mutations increase the flexibility of the angle between helices I and II, enhancing
protein-4WJ DNA interactions. This result seems contrary to our findings. The binding
mode to 4WJ of HMG domains, however, may be very different from that of cisplatinmodified DNA, and intercalating residues may not contribute to the strength of the interaction.
Sequence Selectivity of Mutant HMG-Domain Proteins. Mutation of the base-specific
contact or the intercalating residue in HMG1 domain A did not alter the flanking sequence selectivity (Figure 2.8). These findings support the hypothesis that sequence selectivity for HMG1 domain A is dominated by the presence of A/T base pairs that result in a more flexible DNA structure. The preference for flexibility is similar to that observed for the TATA-Binding protein, which shows an increased affinity for consensus
sequences having this property (33, 36).
Factors that Govern the Photocross-linkingEfficiency. Figure 2.10 indicates that photocross-linking can capture even weak or transient interactions between the platinated
DNA duplex and protein. The presence of an amino acid side chain, such as that of an
intercalating phenylalnine, that tightly fixes the protein to the DNA is not necessary for
protein-DNA covalent bond formation as mediated by the platinum atom upon photolysis. Because the reaction site is located in the major groove of the duplex and pho-
tocross-linking is afforded by Lys6 in HMG1 domain B, wrapping of N-terminus region
around the duplex would appear to be a prerequisite for photocross-linking. Tight
binding of the protein in the minor groove might prevent the movement necessary for
HMG1 domain A to wrap its N-terminus into the major groove and bond to the photoactivated platinum atom.
Although the data presented here are consistent with the notion that loosely
bound protein is most efficiently photocross-linked to platinated DNA, the difference in
photocross-linking efficiency between HMG1 domain A and HMG1 domain B may also
be attributed to the lack of a suitable amino acid at the proper location in HMG1 domain A (Lys6). It is interesting that this residue plays a crucial role for DNA-bending
by HMG1 domain B, as revealed in a ligase-mediated circularization assay (37). The
poor photocross-linking activity of HMG1 domain A may also be correlated to its much
weaker DNA-bending ability compared to HMG1 domain B, which has flanking amino
acids in its N- and C-terminal regions that mediate this activity (37-39).
The DNA Binding Mode. The mutagenesis and footprinting experiments described
here confirm that HMG domains can bind in more than one manner to duplex DNA, as
postulated previously (30, 38, 40, 41). Both symmetric and asymmetric binding modes
are possible, as determined by the intercalator position in the protein. Furthermore, our
results reveal that it is possible to alter rationally the protein binding orientation by
changing the position of the intercalator within the protein scaffold. An HMG domain
can be engineered to cover a specific region on the DNA with respect to the cisplatin
lesion. If the intercalating residue is situated at position 37, as is the case in domA and
domB F16A, the hydroxyl radical cleavage pattern is asymmetric with respect to the
platination site. Coverage occurs exclusively to the 3' side of the lesion. The footprints
of the mutants domA A16F F37A and domB 137A, in which the potential intercalator is
positioned in helix I, are symmetrically centered at the site of the major bend locus (Figure 2.6). These results indicate that Phel6 redirects the protein-DNA interactions and
functions as the intercalating residue.
The domA A16F mutant has two potential intercalating residues. The binding
mode remains the same as that of wild-type domain A, however, consistent with Phe37
being the dominating residue. The favorable hydrogen bond between Ser41 and Ao
1 , in
conjunction with other side chain-DNA interactions observed in the crystal structure,
could explain the preference of the asymmetric over the symmetric orientation. In contrast, domain B, which like domA A16F has two possible intercalating residues, binds in
a symmetric manner. The footprint region is slightly larger than that of domain A and
extends over 8 bases. This observation suggests that domain B may bind in more than
one orientation. The following scenarios are conceivable: either Phel6 or Ile37 intercalate alternatively into the hydrophobic notch, possibly in a dynamic manner; domain B
can pivot around the intercalating residue; or both might occur. Figure 2.11 portrays a
model of how domain B might bind cisplatin-modified DNA symmetrically with Phel6
as the intercalating residue as compared with the structure of domain A binding to cisplatin-modified DNA. The ability of a protein to bind DNA in several different orientations has been reported previously. The engrailed homeodomain binds to certain DNA
sequences in two different complexes as demonstrated by hydroxyl radical footprinting,
although the crystal structure captured only one binding mode (42).
The Recognition Mechanism. Compared to the major groove, the DNA minor
groove offers few distinctive hydrogen bonding features. Recognition of specific sequences by DNA minor groove binding proteins is postulated to depend largely on the
inherent flexibility of a sequence (43, 44) and is often dominated by hydrophobic protein-DNA interactions (33, 36). In addition to hydrophobic minor groove complemen-
tarity, sequence-specific HMG-domain proteins use an intercalating residue for binding
site recognition. The protein-DNA complex structures of hSRY and LEF-1 reveal that
residue 16 serves this function, whereas residue 37 is polar and makes a base-specific
contact. Minor groove DNA-binding proteins often organize elements in chromatin
structure by offering a DNA folding surface. Hydrophobic residues are employed as an
intercalative wedge to pry open one or several base pair steps, thereby inducing a positive roll and bending the DNA away from the protein towards the major groove. In addition to the sequence-specific and sequence-neutral HMG-domain proteins, the TATA
binding protein TBP (45, 46), the hyperthermophile chromosomal proteins Sac7d (47)
and Sso7d (48), the E. coli purine repressor protein PurR (49), and the prokaryotic architectural protein integration host factor IHF (50) belong to this group.
Many of these proteins also bind to DNA containing the cisplatin-1,2-intrastrand
cross-link (48, 51-53). It had been suggested previously that pre-bending of the DNA
would be the recognition signal for protein binding (43, 44). The structure of HMG1
domain A with cisplatin-modified DNA and the mutagenesis results presented here
suggest, however, that pre-formation of a hydrophobic notch through cisplatin modification is at least of equal importance. Favorable intercalation interactions direct protein
binding and orientation. The general mechanism of action of this drug is therefore likely
to be based on its ability to provide an intercalation site for minor groove binding proteins at a much lower energetic cost, the energy penalty for base pair destacking being
paid upon cis-DDP modification. In further support of this hypothesis, the HMGdomain protein Rox1, which has polar residues at both positions 16 and 37, is unable to
serve an intercalating function and is the only HMG-domain protein found to date that
does not bind to the 1,2-intrastrand cisplatin-DNA cross-link (12).
Implicationsfor Full-Length HMG1-DNA Binding. The mutagenesis results presented
here show for the first time that the HMG domains in HMG1 possess two different
DNA binding modes with respect to a defined intercalation site. It is striking that, in
both wild-type domains, a Phe residue is known or presumed to interact with the hydrophobic notch created by the cisplatin-DNA adduct. As a result of the different positioning of these residues within the protein scaffold, the binding region of the two
HMG domains differs by two base-pair steps with respect to the cisplatin-damaged site.
Moreover, there is evidence that the two domains can be differentiated functionally.
The isolated domain A of pig HMG2, which has 90% sequence identity to rat HMG1
employed in our study, is more effective in DNA unwinding than domain B (54). Most
sequence-neutral HMG-domain proteins comprise several HMG DNA-binding domains, with hUBF having as many as six. It is thus likely that fine tuning of HMG domain protein specificity during evolution has been achieved through sequential arrangement of several domains with different DNA recognition properties.
The mechanism of full-length HMG domain proteins binding to cisplatinmodified DNA remains unresolved. In a recent study, it was found that domain A
dominates the AB didomain in binding to 4WJ DNA (55) despite each individual domain binding to the 4WJ in a similar fashion. In addition, a stopped-flow fluorescence
competition study of HMG1 domain A and domain B binding to cisplatin-modified
DNA provides evidence that HMG1 domain A controls binding when both domain A
and B are present (56). It is quite conceivable that domain A is the dominating domain
in HMG1 that binds to the platinum site, while domain B facilitates binding by providing additional protein-DNA interactions. A structure of full-length HMG1 complexed
with cisplatin-modified DNA will provide more insights into how HMG-domain proteins can mediate cisplatin cytotoxicity.
57
Acknowledgements - The crystal structure and analysis were performed by Dr. U.M. Ohndorf with the help of Drs. C. O. Pabo and M. A. Rould, the sequence-preference
experiment was done by Dr. S. M. Cohen and the photocross-linking was done by Dr. Y.
Mikata. I thank Mr. R. J. Kennedy for helping with the CD measurements and Prof. D.
Kemp for access to instrumentation. I thank Drs. S. M. Cohen, S. S. Marla, C. Kneip and
Ms. A. T. Yarnell for helpful discussions. I thank Johnson Matthey for a gift of cisplatin.
REFERENCES
1. This chapter is published in references 2-4.
2. He, Q., Ohndorf, U.-M., and Lippard, S. J. (2000) Biochemistry in press.
3. Cohen, S. M., Mikata, Y., He, Q., and Lippard, S. J. (2000) Biochemistry 39, 1177111776.
4. Mikata, Y., He, Q., and Lippard, S. J. (2000) Nucleic Acids Research submittedfor publication.
5. Gelasco, A., and Lippard, S. J. (1998) Biochemistry 37, 9230-9239.
6. Takahara, P. M., Rosenzweig, A. C., Frederick, C. A., and Lippard, S. J. (1995) Nature
377, 649 - 652.
7. Yang, D., van Boom, S. S. G. E., Reedijk, J., van Boom, J. H., and Wang, A. H.-J. (1995)
Biochemistry 34, 12912-12920.
8. Zamble, D. B., and Lippard, S. J. (1999) in Cisplatin-Chemistryand Biochemistry of a
Leading Anticancer Drug (Lippert, B., Ed.) pp 73-110, Verlag Helvetica Chimica Acta, Zurich.
9. Jamieson, E. R., and Lippard, S. J. (1999) Chem. Rev. 99, 2467-2498.
10. He, Q., Liang, C., and Lippard, S. J. (2000) Proc. Nat. Acad. Sci. USA 97, 5768-5772.
11. Brown, S. J., Kellett, P. J., and Lippard, S. J. (1993) Science 261, 603-605.
12. McA'Nulty, M. M., Whitehead, J. P., and Lippard, S. J. (1996) Biochemistry 35, 60896099.
13. Treiber, D. K., Zhai, X., Jantzen, H.-M., and Essigman, J. M. (1994) Proc. Nat. Acad.
Sci. USA 91, 5672 - 5676.
14. Zamble, D. B., Mu, D., Reardon, J. T., Sancar, A., and Lippard, S. J. (1996) Biochemistry 35, 10004-10013.
15. Bustin, M., and Reeves, R. (1996) Prog.Nucleic Acid Res. Mol. Biol. 54, 35-100.
16. Read, C. M., Cary, P. D., Crane-Robinson, C., Driscoll, P. C., Carrillo, M. O. M., and
Norman, D. G. (1995) in Nucleic Acids and Molecular Biology (Eckstein, F., and Lilley, D.
M. J., Eds.) pp 222-250, Springer-Verlag, Berlin.
17. Grosschedl, R., Giese, K., and Pagel, J. (1994) Trends Gen. 10, 94 - 100.
18. Whitehead, J. P., and Lippard, S. J. (1996) in Metal Ions in Biological Systems (Sigel, A.,
and Sigel, H., Eds.) pp 687-725, Marcel Dekker, Inc., New York.
19. Grasser, K. D., Teo, S. H., Lee, K. B., Broadhurst, R. W., Rees, C., Hardman, C. H.,
and Thomas, J. 0. (1998) Eur. J. Biochem. 253, 787-795.
20. Lilley, D. M. J. (1992) Nature 357, 282-283.
21. Ohndorf, U. M., Rould, M. A., He, Q., Pabo, C. 0., and Lippard, S. J. (1999) Nature
399, 708-712.
22. Cohen, S. M., Jamieson, E. J., and Lippard, S. J. (2000) Biochemistry 39, 8259-8265.
23. Allain, F. H. T., Yen, Y. M., Masse, J. E., Schultze, P., Dieckmann, T., Johnson, R. C.,
and Feigon, J. (1999) EMBO J. 18, 2563-2579.
24. Balaeff, A., Churchill, M. E. A., and Schulten, K. (1998) Proteins: Structure, Function
and Genetics 30, 113-135.
25. Payet, D., Hillisch, A., Lowe, N., Diekmann, S., and Travers, A. (1999) J. Mol. Biol.
294, 79-91.
26. Murphy IV, F. V., Sweet, R. M., and Churchill, M. E. A. (1999) EMBO J. 18, 66106618.
27. Falciola, L., Murchie, A. I. H., Lilley, D. M. J., and Bianchi, M. E. (1994) Nucleic Acids
Res. 22, 285-292.
28. Lohman, T. M., and Mascotti, D. P. (1992) Methods Enzymol. 212, 400-424.
29. Baxevanis, A. D., and Landsman, D. (1995) Nucleic Acids Res. 23, 1604-1613.
30. Dunham, S. U., and Lippard, S. J. (1997) Biochemistry 36, 11428-11436.
31. Love, J. J., Li, X., Case, D. A., Giese, K., Grosschedl, R., and Wright, P. E. (1995) Nature 376, 791 - 795.
32. Dow, L. K., Jones, D. N. M., Wolfe, S. A., Verdine, G. L., and Churchill, M. A. (2000)
Biochemistry 39, 9725-9736.
33. Juo, Z. S., Chiu, T. K., Leiberman, P. M., Baikalov, I., Berk, A. J., and Dickerson, R. E.
(1996) J. Mol. Bol. 261, 239-254.
34. Kane, S. A., and Lippard, S. J. (1996) Biochemistry 35, 2180-2188.
35. Taudte, S., Xin, H., and Kallenbach, N. R. (2000) Biochem. J. 347, 807-814.
36. Grove, A., Galeone, A., Yu, E., Mayol, L., and Geiduschek, E. P. (1998) J. Mol. Biol.
282, 731-739.
37. Stros, M. (1998) JBC 273, 10355-10361.
38. Teo, S. H., Grasser, K. D., and Thomas, J. 0. (1995) Eur.J. Biochem. 230, 943-950.
39. Suda, T., Mishima, Y., Takayanagi, K., Asakura, H., Odani, S., and Kominami, R.
(1996) Nucleic Acids Res. 24, 4733-4740.
40. Pontiggia, A., Rimini, R., Harley, V. R., Goodfellow, P. N., Lovell-Badge, R., and Bianchi, M. E. (1994) EMBO J. 13, 6115-6124.
41. Berners-Price, S. J., Corazza, A., Guo, Z. J., Barnham, K. J., Sadler, P. J., Ohyama, Y.,
Leng, M., and Locker, D. (1997) Eur.J. Biochem. 243, 782-791.
42. Draganescu, A., and Tullius, T. D. (1998) J. Mol. Biol. 276, 529-536.
43. Grove, A., Galeone, A., Mayol, L., and Geiduschek, E. P. (1996) J. Mol. Biol. 260, 120125.
44. Lebrun, A., and Lavery, R. (1999) Biopolymers 49, 341-353.
45. Kim, J. L., Nikolov, D. B., and Burley, S. K. (1993) Nature365, 520-527.
46. Kim, Y. C., Geiger, J. H., Hahn, S., and Sigler, P. B. (1993) Nature 365, 512-520.
47. Robinson, H., Gao, Y.-G., McCrary, B. S., Edmondson, S. P., Shriver, J. W., and
Wang, A. H.-J. (1998) Nature 392, 202-205.
48.
Gao, Y.-G., Su, S.-Y., Robinson, H., Padmanabhan, S., Lim, L., McCrary, B. S.,
Edmondson, S. P., Shriver, J. W., and Wang, A. H.-J. (1998) Nature Struct. Biol. 5, 782-
786.
49. Schumacher, M. A., Choi, K. Y., Zalkin, H., and Brennan, R. G. (1994) Science 266,
763-770.
50. Rice, P. A., Yang, S., Mizuuchi, K., and Nash, H. A. (1996) Cell 87, 1295-1306.
51. Chow, C. S., Whitehead, J. P., and Lippard, S. J. (1994) Biochemistry 33, 15124-15130.
52. Trimmer, E. E., Zamble, D. B., Lippard, S. J., and Essigmann, J. M. (1997) Biochemistry 37, 352-362.
53. Coin, F., Frit, P., Viollet, B., Salles, B., and Egly, J.-M. (1998) Mol. Cell. Biol. 18, 39073914.
54. Yoshioka, K., Saito, K., Tanabe, T., Yamamoto, A., Ando, Y., Nakamura, Y., Shirakawa, H., and Yoshida, M. (1999) Biochemistry 38, 589-595.
55. Webb, M., and Thomas, J. 0. (1999) J. Mol. Biol. 294, 373-387.
56. Jamieson, E. R., and Lippard, S. J. (2000) Biochemistry 39, 8426-8438.
Table 2.1. Affinities of I-HMG1 Domain A and Domain B
Mutants Towards Cisplatin-Modified DNA
Protein
Kd (nM)
Kdrelativea
DomA
1.5 ± 0.5
1
DomA AVN
5.0 ± 0.9
3.3
DomA F37W
9.9 ± 2.1
6.6
DomA A16F
17 + 5.7
11
DomA A16F F37A 167 ± 32
111
DomA F37A
> 1000
>667
DomA S41A
5.8 + 0.7
3.9
DomB
39 + 9.9
1
DomB iVN
39 + 13
1
DomB I37F
19.5 ± 2.6
0.5
DomB F16A
119 ± 39
3
DomB I37A
85 ± 17
2.2
DomB F16A I37A
> 1000
>25
aKd/ Kd (DomA) or Kd/ Kd (DomB)
64
Table 2.2. Codes and Sequences for 15 bp Probes Containing 7-deaza-dA, 7-deaza-dG,
and dG-rich Sequences.
Codea
Sequence"
CGGA
5'-CCTCTCCGGATCTTC-3'
3 '-GGAGAGGCCTAGAAG-5'
AGGC
5'-CCTCTCAGGCTCTTC-3'
3 '-GGAGAGGCCGAGAAG-5'
AGGA
5'-CCTCTCAGGATCTTC-3'
3 '-GGAGAGTCCTAGAAG-5'
GGGA
5'-CCTCTCGGGATCTTC-3'
3 '-GGAGAGCCCTAGAAG-5'
GGGC
5'-CCTCTCGGGCTCTTC-3'
3 '-GGAGAGCCCGAGAAG-5'
GGGT
5'-CCTCTCGGGTTCTTC-3'
3'-GGAGAGCCCAAGAAG-5'
GGGG
AGGG
CGGG
TGGG
5'-CCTCTCGGGGTCTTC-3'
3 '-GGAGAGCCCCAGAAG-5'
5' -CCTCTCAGGGTCTTC-3'
3 '-GGAGAGTCCCAGAAG-5'
5' -CCTCTCCGGGTCTTC-3'
3'-GGAGAGGCCCAGAAG-5'
5'-CCTCTCTGGGTCTTC-3'
3'-GGAGAGACCCAGAAG-5'
AGGG
5'-CCTCTCAGGGTCTTC-3'
3' -GGAGAGTCCCAGAAG-5'
CGGG
5'-CCTCTCCGGGTCTTC-3'
3'-GGAGAGGCCCAGAAG-5'
TGGG
5'-CCTCTCTGGGTCTTC-3'
3 '-GGAGAGACCCAGAAG-5'
aA indicates 7-deaza-dA and G indicates 7-deaza-dG.
Figure 2.1. Structure of the HMG1 domain A complex with cisplatin-modified DNA. The
protein backbone is shown in yellow, the intercalating Phe 37 residue as van der Waals
spheres, and the DNA in red and blue with the cis-[Pt(NH3 )2 {d(GpG)-N7(G8),-N7(G9)}]
intrastrand adduct in green. Numbers indicate the first (N terminus) and last (C terminus)
ordered residues in the crystal structure.
S
Helix I
I Helix I. .
Helix III
10
20
30
40
50
60
70
80
MGKGDPKKPRGKMsSS
F1FVQTCREEHKKKHPDASV SE FSKKCSERWKTMSAKEKGK FEDMAKADKARYEREMKTYIPPKGETKKK
FKDPNAPKRPPSAL FFCSEYRPKIKGEHPGLS-- GDVAKKLGEMWNNTAADDKQPYEKKAAKLKEKYEKDIAAYRAK
MVTPREPKKRTTRKKKDPNAPKRALSA) FFANENRDIVRSENPDIT--GQVGKKLGEKWKALTPEEKQPYEAKAQADKKRYESEKELYNATLA
MS DKPKRPLSALWLNSARES IKRENPGIK-- TEVAKRGGELWRAM- -KDKSEWEAKAAKAKDDYDRAVKEFEANGGSSAANGGGAKKR
I
HMGI domA
HMG1 domB
NHP6A
HMG-D
LEF1
SRY
MHIKKPLNAfL YMKEMRANVVAECTLKE-AAINQ ILGRRWHALSREEQAKYYELARKERQLHMQLYPGWSARDNYGKKKKRKREK
QDRVKRPMNA IVWSRDQR-RKMAENPRMR-- SE I SKQLGYQWKMLTEAEKWPFFQEAQKLQAMHREKYPNYK
B
5'-C01C02T03C04T05C06 o7Go
08 9A10C 11C12T1T
3 14c15C16 -3.
3'-G 3 2G31 A 30 G2 9 A2 8 G2 7A2 6 C2 5 C2 4 T 2 3G2 2 G2 A2 0 A1 9 G1 8G1 7 -5'
Figure 2.2. Sequences of HMG domain proteins and cisplatin-modified DNA. A.
Sequence alignment of HMG domain proteins. B. Sequence of DNA used in this
study; asterisks denote nucleotides cross-linked by cisplatin.
CAIV\~~n
Mn
JLRJUUU.UUU
4000000.000
3000000.000
2000000.000
1000000.000
A.2
0.000
T -1000000000
-2000000000
-3000000000
0
Wavelength (nm)
5UUUUUU.UU
4000000.00C
,
3000000.00C
5 2000000.00O
B.
g 1000000.00C
0.00C
'
-1000000000
-2000000000
-3000000000
Wavelength (nm)
Figure 2.3. Circular dichroism spectra of HMG1 domain A and domain B
mutants. A. CD spectra of domA, domA F37W, domA A16F F37A and
domA A16F. B. CD spectra of domB, domB F16A and domB I37F.
HMG1 domain A
wild-type
c
1 2
3
4
5
6
7
HMG1 domain A
Phe37Ala mutant
8
9 10 11 12 13 14 15
Figure 2.4. Gel-mobility shift assays comparing the DNA-binding
ability of wild-type HMG1 domain A (lanes 2-8) with that of a
domAF37A mutant (lanes 8-15). Lane 1 contain no protein. The
cisplatin-modified probe (5 nM) was incubated with 10, 20, 40, 60,
80, 100 and 200 nM protein. The band at high mobility corresponds
to unbound DNA, where as that of lower mobility represents the
specific protein-DNA complex.
Figure 2.5. The potential steric crowding between domA F37W mutant
and DNA.
-+
Domain A
_
I
I
I
04400·-I-
-
+
T-%--
A
A 1 I-~
1-n
DomB F16A
+
-
A
G
A
G
A
G
A
]r
MIJAW
4"Now
ý
40MMOMW
-
*avo
-
DomA A16F
+
-
+
DomB 137A
A
C
~·
4"
ot
460*"**
rr/·
rigure L.6A
-
T
+
l
I l
'l' ' a ,
+
-
Domain A
, ,l
.
Domain B
l
G
T
C
'
T
-+
DomA A16F F37A
+
-
DomB F16A
...
'U
I I
T
LCA
""
-
G
G
T
T
-
DomA A16F
-+
.
T
.
.
.
.
,
.
.
.
+
DomB 137A
.
44O
G
T
Figure 2.6B
Domain A
5' -CCTCTCTGGACCTTCC-3'
Domain B
3 '-GGAGAGACCTGGAAGG-5'
5' -CCTCTCTGGACCTTCC-3'
3 '-GGAGAGACCTGGAAGG-5'
DomA A16F
DomB F16A
5'-CCTCTCTGGACCTTCC-3 '
3'-GGAGAGACCTGGAAGG-5 '
5' -CCTCTCTGGACCTTCC-3 '
DomA A16F F37A
DomB I37A
5'-CCTCTCTGGACCTTCC-3'
3 '-GGAGAGACCTGGAAGG-5'
3 '-GGAGAGACCTGGAAGG-5'
5 ' -CCTCTCTGGACCTTCC-3'
3 '-GGAGAGACCTGGAAGG-5'
Figure 2.6. Hydroxyl radical footprint analysis. (A) Footprint analysis of the
bottom strand DNA. Dotted lines are the traces for control samples without protein
added. Solid lines are the traces for the footprint with protein added. Boxed and
shaded letters indicate the protection areas. (B) Footprint analysis of the top strand
DNA. (C) Summary of footprint results. Bold letters indicate the protected areas
and half circles above the GG bases indicate the site of platinum cross-linking.
N NH
H2
AMP
oN
0
0-
0OH
OH
S
0
GMP
0
GMP
NH
NH2
H
<
-0-ro0-
NH2
AMP
O
OH
ýNH2
0
-"-o0OH
Figure 2.7. Structural diagram of deoxyadenosine monophosphate (AMP),
deoxyguanosine monophosphate (GMP), 7-deaza-dA monophosphate (AMP), and 7deaza-dG monophosphate (GMP).
n rn
U.2U
0.40
0.30
0.20
0.10
0.0
GG*G*A
GG*G*C
GG*G*T
-
-
-
GG*G*G
AG*G*C
CG*G*A
AG*G*A
PROBE -
Figure 2.8. Bandshift experiments showing the flanking sequence preferences of
mutant HMG-domain proteins. The graph shows the fraction of bound probe DNA
(0) and the sequence selectivity for HMG1 domAS41A (white bars) and HMG1
domAF37W (striped bars). Both mutant proteins display the same sequence
selectivity as wild-type HMG1 domain A. Error bars indicate one standard deviation
derived from at least three independent experiments. [HMG1 domAS41A] 148 nM;
[HMG1 domAF37W] 203 nM, and [probe] 5.0 nM.
lane
1
2
3
4
_.rotein-DNA
cross-link
.-.
igested protein-DNA
cross-link
._efree
probe
Figure 2.9. Denaturing polyacylamide gel demonstrating the photocrosslinking reac-tion of platinated 15-bp DNA with HMG1 domain B protein
induced by laser (325 nm, 30 min) irradiation. Lane 1: with protein, no
irradiation. Lane 2: no protein, irradiated. Lane 3: with protein, irradiated.
Lane 4: same as lane 3, followed by NaCN treatment (0.2 M, pH 8.5). Lane
5: same as lane 3, followed by proteinase K treatment (0.1 mg/mL).
native A16M A16F F37A A16M A16F native F16A I37A F16A
F37A F37A
I37A
HMG1domA
HMGldomB
protein
Figure 2.10. Photocross-linking experiment with mutant HMG-domain
proteins using 325 nm light (30 min).
A
B
Figure 2.11. Comparison of two binding modes of HMG domains bind to cisplatinmodified DNA. A. Domain A binding to cisplatin-modified DNA (21). B. A model of
domain B binding to Pt-DNA with Phel6 as the intercalating residue. The LEF-1DNA complex binding orientation is used as the model, and platinated-DNA was
superimposed with LEF-1 DNA (see Materials and Methods), where as domain B
was superimposed with LEF-1.
Chapter 3
The Effect of Steroid Hormones on HMG1 Protein Levels and
Platinum Sensitivity in Mammalian Cell Lines
Introduction (1,2)
Cisplatin (cis-diamminedichloroplatinum(II) or cis-DDP) is a widely used
antitumor drug for the treatment of testicular, breast, ovarian, lung, and head and neck
tumors (3-7). DNA is the primary cytotoxic target of cisplatin in vivo (8). The major
cisplatin-DNA adducts formed in vivo are 1,2-intrastrand d(GpG) and d(ApG) crosslinks.
The adducts significantly bend and distort normal B-DNA (8, 9), thereby
affecting DNA replication and transcription (10-14). In addition, structural distortions
imposed upon DNA by cisplatin coordination are recognized by a variety of structurespecific DNA-binding proteins, such as DNA repair and high-mobility group (HMG)
domain proteins (14).
HMG-domain proteins are architectural proteins that facilitate cellular functions
requiring chromosomal DNA (15). HMG-domain proteins bind specifically to the major
cisplatin-DNA adducts, forming stable platinum-DNA-protein ternary complexes (16,
17). There is evidence implicating the involvement of such platinum-DNA-protein
complexes in mediating cisplatin cytotoxicity by blocking nucleotide excision repair
(NER) of the DNA damage, a process termed repair shielding. In vitro experiments
revealed that a variety of HMG-domain proteins, including HMG1, tsHMG, and SRY,
blocked removal of cisplatin intrastrand d(GpG) adducts when added in a NER assay
(18-20). Depleting HMG1 and HMG2 from cell extracts by immunoprecipitation
enhanced excision repair of cisplatin-modified DNA (21). In yeast, interruption of the
HMG-domain protein Ixrl caused a 2- to 6-fold desensitization to cisplatin compared to
wild type cells (22, 23). Furthermore, introducing HMG2 by transfection enhanced the
cisplatin sensitivity of a lung adenocarcinoma cell line (24). Until the present work,
however, there was no evidence that overexpression of an HMG-domain protein via a
natural signal transduction pathway in human cancer cells could increase their
sensitivity to cisplatin.
HMG1 is a structure-specific HMG-domain protein with little or no sequence
specificity. It is abundant in all tissues and species (25). HMG1 binds preferentially to
cisplatin-modified DNA, cruciform DNA, and other distorted structures (15). It has
numerous functions including association with chromosomes and interaction with Oct,
Hox, p53, and some components of the basal transcriptional machinery (26-30). HMG1
also functions as an architectural protein to facilitate the binding of steroid hormone
receptors, such as estrogen, progesterone, androgens, and glucocorticoid receptors (ER,
PR, AR and GR) to their cognate DNA binding sites (31).
Humans use steroid hormones to transmit information between different cells
and tissues (32). Steroid hormones are mainly synthesized from cholesterol precursors
in the mitochondria of adrenal glands, gonads, and placenta. Estrogen, progesterone,
and androgen belong to the family of steroid hormones. The structures of the common
steroid hormones are shown in Figure 3.1.
Steroid hormones influence many processes, including development, cell growth
and cancer, metabolism, immune functions, cardiovascular actions, bone functions, and
reproductive functions. On the cellular level, steroid hormones act in concert with
hormone receptors, thereby initiating a signaling cascade that activates chromatin in the
nucleus. Steroid hormones enter targeted cells by diffusion and bind to the respective
hormone receptors (HR). Progesterone receptor, androgen receptor, and glucocorticoid
receptor exist in the cytoplasm, but the estrogen receptor is confined to the nuclear
compartment. The hormone receptors are usually complexed with heat shock proteins
(hsp) in the absence of hormone. Upon ligand binding, the HR undergoes a series of
conformational changes, including dimerization, liberation of heat shock proteins,
exposure of zinc-finger DNA-binding domains, and interactions with accessory
proteins. The receptor-hormone complexes bind to the hormone responsive element
(HRE), a DNA sequence recognized by the zinc-finger domains of the hormone
receptors. Figure 3.2 illustrates the mechanism postulated for the estrogen receptor.
Transcription regulation can subsequently occur for many controlled genes (32). For
example, there are about 30 known proteins that are under the control of ER (Table 3.1)
(33).
Binding of ER to estrogen responsive elements (ERE) and PR to progesterone
responsive elements (PRE) induces DNA bending (34) and, accordingly, attracts HMG1
and increases HMG1 affinity at the site. The binding of HMG1 further alters the
structure of the target DNA and facilitates formation of a more stable receptor/DNA
complex (Figure 3.2) (35-37). The transcriptional activity of these steroid hormones is
enhanced in mammalian cells that are transiently expressing HMG1 (31, 35). Mice
deficient in HMG1 die of hypoglycemia, implicating a role for HMG1 in glucocorticoiddependent gene regulatory pathways (38).
The rationale for the discovery described here was derived from previous work
reporting that the mRNA level of HMG1 is up-regulated when human MCF-7 breast
cancer cells are treated with estrogen (39). We corroborate such overexpression at the
protein level. Previous efforts to overexpress HMG1 in mammalian cell lines with
either an inducible or a stable vector system has been unsuccessful (Zamble, D.B.,
unpublished results).
Therefore, such hormone-induced
"natural" HMG1
overexpression, such as that described (39), is highly desirable. The inverse relationship
demonstrated between the levels of HMG-domain proteins and the ability of cells to
repair cisplatin-DNA adducts both in vivo and in vitro predicts that steroid hormone
treatment would sensitize these cells to cisplatin. The present results confirm this
prediction and suggest that a combination of cisplatin or carboplatin with estrogen
and/or progesterone, all FDA-approved drugs, may be of clinical significance for the
treatment of certain kinds of cancer.
Materials and Methods
Cell Culture. MCF-7 cells derived from breast adenocarcinoma were purchased
from the American Type Culture Collection (Rockvilla, MD). BG-1 cells were a gift
from J. Barrett (NIH, MD) (40). Evsa-T cells were a gift from G. Leclercq (Brussels,
Belgium) (41). Cells were grown in DMEM (GIBCO/BRL) containing 10% heatinactivated fetal bovine serum (GIBCO/BRL), 2 mM glutamine, 100 units/ml penicillin
and 100 pg/ml streptomycin at 370 C under a 5% CO2(g) atmosphere.
Clonogenic Assays. Cells were seeded on 6-well plates (Corning) at a density of
400-800 cells per well. After 24 h, a fresh stock of estrogen (17P-estradiol) or
progesterone was prepared in N,N-dimethyl formamide (DMF) and added to the plates
at a final hormone concentration described in each experiment. Control plates were
treated with the same volume of DMF without hormone. For co-treatment, the hormone
was added at the same time as cisplatin. For pre-treatment, the hormone was added 2 h,
4 h or 24 h prior to cisplatin. After 4 h of cisplatin treatment, the cells were washed with
PBS and fresh media was added. After 10 days, the cell colonies were stained with a 1%
methylene blue (Fluka), 50% ethanol solution and counted. Each point is an average of
three independent determinations +/- one standard deviation (eq 1) of the cell count.
S
[1]
N-i
For carboplatin cytotoxicity assays, cells were incubated with the drug (0 - 160
pM) for 16 or 24 h, and then treated with either 10' M estrogen or 10 -6M progesterone
for 4 h. Cells were washed, and fresh media was added. After ten days, cells were
counted and plotted as described above.
Western Analysis for HMG1 Protein Levels. Cells grown to 80% confluence were
treated with P-estradiol or progesterone in DMF to a final concentration of 10-'M and
106 M respectively for 0 h, 2 h, 4 h, and 24 h. After hormone treatment, DMEM was
removed and cells were washed with PBS and lysed (2% SDS, 15 pg/ml pepstatin,
leupeptin, and aprotinin protease inhibitors). Lysates were passed through a 25 G5 /,
syringe needle and boiled for 5 min. The bicinchoninic acid protein kit (Sigma) was
used to determine the concentration of protein in each cell extract. Protein samples (200
pg) were separated on a 12% SDS-polyacrylamide gel. The proteins were transferred to
a Protran pure nitrocellulose membrane (Schleicher & Schuell) by electroblotting. The
membrane was blocked with 5% nonfat dry milk in TPBS (0.05% Tween in PBS) for 1 h,
then washed with TPBS. The blot was incubated with primary antibody (1:200 dilution)
for 1.5 h then washed with TPBS followed by incubation with ' 25I-protein A for 1 h. The
blot was washed with TPBS, and then exposed to a phosphorimager screen and protein
bands were quantitated on a Molecular Imager Multianalyst system.
Immunofluorescence to Detect HMG1 Protein Levels. The cells were grown to 70%
confluence on 12 mm glass cover slips in 6-well plates and subsequently treated with
estrogen or progesterone for 0 to 24 h. After hormone treatment, the cells were
permeablized with 25% acetic acid in methanol for 10 min at room temperature and
washed with PBS. The permeablized cells were incubated for 30 min at 370 C with 1:100
dilution (for MCF-7 cells) or 1:500 dilution (for Evsa-T cells) of anti-HMG1 polyclonal
antibody (PharMingen), washed, incubated subsequently with 1:200 dilution of goat-
anti-rabbit IgG conjugated to fluorescein (Biosource International) for 30 min at 370 C.
Finally, the cells were visualized with fluorescent light microscope (Zeiss Axiophot).
Double Labeling and Immunofluorescence of Mammalian Cancer Cells. Ovarian cancer
BG-1 or NIH:OVCAR-3 cells were grown to 80% confluence on 12 mm coverslips
(Fisher Scientific). Cells were treated with 10 7 M estrogen or progesterone. Cells were
washed with PBS containing 10 pg/ml CaCl, and MgC12 then incubated with mouse
anti-CA125 antibody (1:150 for BG-1 and 1:300 for NIH-OVCAR-3) (Chemicon
International) or normal mouse IgG control (1:100 dilution) (Santa Cruz Biotechnology)
in 10% goat serum in PBS for 30 min at 370 C. Cells were washed with PBS, then
incubated with TRITC conjugated goat anti-mouse antibody (1:100 dilution) (Jackson
ImmunoResearch Laboratories) in 10% goat serum in PBS for 30 min at 37 0 C. After
washing the cells with PBS, different protocols of fixation-permeabilization were tested:
(1) fixation in 1%, 2%, or 4% paraformaldehyde in PBS for 10 min and permeabilization
in 0.5% NP-40 for 10 min at room temperature; (2) fixation/permeabilization in 70%
ethanol for 10 min at room temperature; (3) fixation in 70% ethanol for 10 min and
permeabilization
in 0.5%
NP-40 for 10
min at room temperature;
(4)
fixation/permeabilization in 3:1 methanol: acetic acid for 10 min at room temperature.
Cells were washed with PBS, then incubated with rabbit anti-HMG1 antibody
(PharMingen) in 10% goat serum in PBS for 30 min at 370 C. After the cells were washed
with PBS, they were incubated with FITC conjugated goat anti-rabbit antibody (1:500
dilution) (Biosource International) in 10% goat serum in PBS for 30 min at 370 C.
Following one wash with water, the coverslips were mounted on slides with Gelvatol.
Cells were visualized using a fluorescence microscope (Zeiss Axiophot) and pictures
were taken with a CCD camera.
DAPI Assay. MCF-7 cells from 24-well plates were treated with varying amount
of cisplatin (0 to 20 pM) and estrogen (10- 7 M) for 4 h. After 24 h, both attached and
unattached cells were collected, pelleted and resuspended in 25 pl of PBS. A 5 pl
aliquot of Trypan blue dye was added to 5 pl of cells and the number of dead cells
counted. Dead cells appeared blue under the microscope due to permeable membranes.
The remaining cells were fixed with 0.5 ml of 4 'C MeOH, and incubated on ice for 10
min, followed incubation with 1 pg/ml of DAPI (4,6-diamidino-2-phenylindole) for 5
min at room temperature. Cells were then resuspended in Mowiol (10 pl), mounted
onto slides and visualized under a fluorescent light microscope.
Apoptosis TUNEL Assay. Apo-BrdU kit (PharMingen) is a staining method for
labeling DNA breaks and total cellular DNA to detect apoptotic cells by flow cytometry.
Terminal deoxynucleotidyl transferase is used to add bromodeoxyuridine triphosphate
(Br-dUTP) to the 3'-OH DNA ends, and the DNA termini are labeled with fluorescein
conjugated anti-BrdU antibody. Propodium iodide (PI) is used to stain total DNA.
A total of 106 cells were resuspended in 0.5 ml of PBS. The cell suspension was
added to 5 ml of 1% (w/v) paraformaldehyde in PBS and placed on ice for 15 min.
Paraformaldehyde chemically cross-links the smaller fragments of DNA to avoid loss of
DNA in the subsequent wash steps. Cells were washed twice with PBS, fixed in 5 ml of
ice-cold 70% (v/v) ethanol, and stored at -20 OC overnight. After several washes,
thawed cells were incubated with Br-dUTP and TdT enzymes for 1 h at 30 oC. The cells
were rinsed and incubated with fluorescein labeled anti-BrdU antibody in a total
volume of 100 pl1 for 30 min at room temperature. Propodium iodide (PI) solution (0.5
ml) was added to the cells and incubated at room temperature for 30 min. The cells
were analyzed by flow cytometry equipped with a 488 nm argon laser as the light
source. PI fluoresces at about 623 nm and fluorescein at 520 nm when excited at 488
nm.
Mammalian Genomic DNA Rapid Preparation. Plates of cells were harvested with
PBS and lysed in 2 ml of 100 mM Tris-HC1, pH 8.5, 5 mM EDTA, 0.2% SDS, 200 mM
NaCI, 100 pg/ml proteinase K. The samples were agitated overnight at 55 OC. Genomic
DNA was extracted twice with 1X volume of phenol/chloroform/isoamyl alcohol, and
followed by two extractions with chloroform/isoamyl alcohol.
The DNA was
precipitated with 1X volume of isopropanol, and the string-like DNA was removed and
dissolved in 10 mM Tris-HC1 pH 7.5 buffer. The DNA was treated with RNase and
quantitated by UV-vis spectroscopy.
32 P
Post-LabelingMethods to Detect Platinum-DNA Adducts. Genomic DNA (10-50
pg) was first dissolved in 50 pl of 50 mM ammonium acetate (pH 5.0) and 1 mM ZnC12
and 6 U of Nuclease P1 and incubated at 55 'C for 2 h. To the mixture was added 10
mM of Tris-HC1 (pH 8.2) and 50 U of DNase I, followed by incubation at 37 'C for 2 h.
The mixture was adjusted to pH 9.0, 20 U of alkaline phosphatase added, and incubated
at 37 'C overnight.
The pH of the samples was adjusted to 3.0 by addition of 0.8 vol 0.05 M HC1. The
samples were applied to pre-washed, pre-equilibrated SCX columns containing 100 mg
benzene sulfonic acid resin in 1 cc syringes (Varian). This step isolates Pt-adducts on
basis of positive charge. Unmodified nucleotides and nucleosides were eluted by eight
washes with 10 mM ammonium formate buffer (pH 6.0). Platinated products were
recovered by addition of 1 ml of 0.25 M NH 4OH. DNA samples were dried in a
Speedvac, and then dissolved in 12 p1 of 0.2 M NaCN, pH 10, and incubated overnight
at 65 oC.
The DNA samples were neutralized to pH 8 and labeled with T4 DNA kinase
and y-32P-ATP. Pure ApG and GpG dinucleotide were also labeled as standards. DNA
samples were separated on TLC sheets (20x40 cm, Polygram gel 300 PEI TLC) with 1.5
M ammonium formate buffer (pH 4.0). The spots were integrated by phosphorimage
analysis.
RESULTS and DISCUSSION
HMG1 Overexpression in Breast Cancer MCF-7 and Evsa-T Cells. MCF-7 breast
cancer cells express both ER and PR. Estrogen induces an increase in HMG1 mRNA
levels in MCF-7 cells (Figure 3.3) (39). To determine whether elevation in transcription
level of HMG1 is accompanied by increased protein levels, western blots were
performed on cell extracts from MCF-7 cells. Cells treated with estrogen showed
increased HMG1 level compared to the untreated cells (Figure 3.4). HMG1 protein
levels appear to increase within 2 h of estrogen treatment and gradually decrease over
the course of 24 h. The relatively small differences in the increase makes the
quantitative western analysis difficult and vary between experiments, thus an
alternative method is sought to validate the protein upregulation.
Immunofluorescence techniques use primary and fluorescently labeled
secondary antibodies to detect the protein of interest. Immunofluorescence is more
specific and allows for visualization of protein localization within individual cells. A
protocol was optimized to detect HMG1 as a nuclear protein. A 3:1 methanol/acetic
acid fixation and permeabilization is the best method for nuclear staining of HMG1.
MCF-7 cells treated with estrogen or progesterone have a higher level of HMG1 than
that of the untreated cells (Figure 3.5). Evsa-T cells are ER negative, PR positive breast
cancer cells. Evsa-T cells overexpress HMG1 with progesterone treatment but not with
estrogen (Figure 3.6). Therefore, we conclude that the HMG1 upregulation by steroid
hormones requires the appropriate hormone receptors.
The time-course of HMG1 upregulation in these cell lines is different. Detailed
immunofluorescence analysis shows that MCF-7 cells upregulate HMG1 within 30 min
of hormone treatment. On the other hand, HMG1 levels in Evsa-T cells respond after 4
h of progesterone treatment. Bowman Gray-1 (BG-1) cells are ER and PR positive
ovarian cancer cells derived from a solid tumor. BG-1 cells respond differently to the
two hormones. Estrogen treatment induced HMG1 after 30 min, achieved a maximum
after 2 h, and decreased to basal levels after 24 h (Figure 3.7). In contrast, progesterone
caused maximum HMG1 overexpression after 30 min, remained at a maximum up to 2
h, and decreased to slightly above basal levels after 24 h (Figure 3.7). The HMG1
overexpression is obvious under the fluorescence microscope. The exact amount of
increase can not be quantitated with the microscope because the cell density and
distribution are different on each cover slip, and a FACS based protocol needs to be
established to quantitate the intensity of fluorescence.
HMG1 facilitates the binding of both ER and PR to their response elements (31).
HMG1 induces a structural change in DNA that helps promote the binding of
progesterone receptor to its DNA response element (36), which may be why the cell
naturally upregulates HMG1 upon hormone treatment. Because no HRE was found in
the HMG1 promoter region, the exact mechanism of HMG1 upregulation is not known
(39).
Sensitization of Cells to Cisplatin by Estrogen or Progesterone. To assess further the
involvement of HMG-domain proteins in mediating cisplatin cytotoxicity, we
investigated the effects of elevated HMG1 levels on the sensitivity of cells to the drug.
As predicted by the repair shielding hypothesis, we found that elevated HMG1
expression levels were indeed paralleled by increased sensitivity to cisplatin. In MCF-7
cells, estrogen or progesterone treatment increased cisplatin sensitivity about 2-fold,
according to the LC 0values, the cisplatin concentrations where only 50% of the cells are
viable (Figure 3.8A). A combination of estrogen and progesterone sensitized the cells to
cisplatin by a factor of 4 (Figure 3.8A); this additive effect suggests that the two
hormones may independently up-regulate HMG1. In Evsa-T cells, progesterone
treatment induced 1.5-fold sensitization towards cisplatin, whereas estrogen treatment
had no effect due to the absence of ER (Figure 3.8B).
For BG-1 cells, estrogen
cotreatment caused a 2.2-fold increase in cisplatin sensitivity and progesterone
treatment caused a 2.5-fold increase (Figure 3.8C). In addition, the cisplatin sensitivity
of HeLa cells, an ER negative cervical cancer cell line, did not change following estrogen
treatment (Figure 3.8D).
Effects of Hormone Concentration on the Degree of Sensitization. MCF-7 cells were
treated with 10-7 M, 10-6 M, 10-5 M, 10-' M, or 10-3 M estrogen to investigate the
concentration dependence of cisplatin sensitization by hormone treatment. The degree
of sensitization increased slightly with increasing concentrations of estrogen, with a 2.1fold sensitization after 10-6 M estrogen treatment and 2.5-fold increase under 10-s M
estrogen (Figure 3.9). MCF-7 cells failed to survive under the conditions of 10-4 M or 103
M estrogen treatment. An increase in estrogen concentration induced higher levels of
HMG1, leading to greater HMG1 overexpression in the cells.
If the estrogen
concentration is too high, however, the cells do not survive. We conclude that the
concentration of the hormone affects the degree of sensitization. The concentration
used for hormone replacement therapy in post-menopausal women for hormone
replacement therapy is around 10- 7 M in plasma and thus this concentration is clinically
relevant.
Timing of the Hormone Treatment and Sensitivity Toward Cisplatin and Carboplatin.
The timing of hormone treatment plays an important role in the sensitizing the cells to
cisplatin. Co-treatment of estrogen and cisplatin caused 2-fold sensitization in MCF-7
cells; pre-treatment with estrogen for 24 h did not cause any sensitization (Figure 3.10).
In Evsa-T cells, the effect of sensitization by progesterone was not observed with cotreatment (data not shown) but was 2-fold with a 2-h pre-treatment of progesterone
(Figure 3.8B).
Carboplatin is an oral analog of cisplatin.
Carboplatin forms the same
bifunctional DNA adducts as cisplatin because the aquation products of both are the
same (42). Carboplatin is used widely in the clinical treatment of cancer since, unlike
cisplatin, it does not produce nephrotoxicity (43).
In comparison to cisplatin,
carboplatin is not as potent; 10 times the concentration and 7.5 times longer incubation
are necessary to induce the same degree of DNA damage (44). Also, the uptake of
carboplatin is 1.5 to 13 times lower than the uptake of cisplatin into human ovarian
cancer cell lines (43). Estrogen had no effect on the carboplatin sensitivity of cells
treated simultaneously with the two reagents, but a 24-h pretreatment of carboplatin
followed by a 4-h estrogen treatment increased the sensitivity by 2-fold (Figure 3.11).
Hormone treatments must be termed to reflect both the kinetics of HMG1 upregulation and cisplatin aquation chemistry. HMG1 protein levels increase within 0.5-2
h after estrogen treatment in MCF-7 cells, whereas cisplatin may take a few hours to
enter the cell, undergo aquation, and bind to DNA (45). By the time platinum-DNA
damage occurs, HMG1 levels are high and can block NER efficiently. When cells are
pre-treated with estrogen for 24 h, the amount of HMG1 has leveled off before
platinum-DNA adduct formation. Consequently, the sensitization effect is less. On the
other hand, pre-treatment of estrogen for 24 h sensitizes MCF-7 cells to carboplatin
more than co-treatment. This observation is readily explained by the difference in the
rate of DNA binding for the two drugs. A 7.5-fold longer incubation time for
carboplatin than cisplatin is necessary to obtain the same degree of DNA damage (43,
44, 46), owing to the slow rate of aquation of carboplatin. In contrast to MCF-7 cells,
HMG1 levels increase in Evsa-T cells only after 4 h of progesterone treatment. The
effect of sensitization by progesterone is not observed with co-treatment but is 2-fold
with 2-h pretreatment of progesterone.
We conclude that the degree of drug
sensitization is affected by the timing of hormone treatment, which reflects the kinetics
of HMG1 upregulation and of platinum aquation.
Tamoxifen and Estrogen. Tamoxifen is an nonsteroidal antiestrogen that binds to
estrogen receptor with low affinity (47). Tamoxifen is both a partial agonist and a
partial antagonist for estrogen. In breast cells, tamoxifen behaves as an estrogen
antagonist (47) and is commonly used for breast cancer treatment or prevention.
Synergy between tamoxifen and cisplatin has been reported in human melanoma and
ovarian tumor cells (48-50), which may be due to enhancement of DNA platination
levels in tamoxifen treated cells (51).
MCF-7 cells were treated with equimolar
concentrations of estrogen and tamoxifen (10-7 M) with cisplatin to investigate the
effects that cotreatment has on cisplatin sensitivity. MCF-7 cells treated with estrogen
alone were 2-fold more sensitive to cisplatin treatment whereas those treated with
tamoxifen alone were 1.3-fold more sensitive (Figure 3.12).
The combination of
tamoxifen and estrogen treated MCF-7 cells produced no significant effect on cisplatin
sensitivity. Since tamoxifen can bind to estrogen receptor, tamoxifen could be
competing with estrogen for receptor binding without causing HMG1 overexpression,
thus diminishing the overall level of HMG1 overexpression in the cells. The reduction
in HMG1 levels could reduce the sensitization of tamoxifen and estrogen cotreated cells
to cisplatin by diminishing the extent of repair shielding in the cells.
Immunofluorescence was used to measure the effects of tamoxifen on HMG1
level in MCF-7 cells. Over the course of 24 h, there was no significant difference in
HMG1 protein expression in MCF-7 cells treated with 10 7 M tamoxifen than the
untreated cells (Figure 3.13A). When cells were treated with a combination of tamoxifen
and estrogen, no obvious HMG1 overexpression was observed (Figure 3.13B). The
immunofluorescence data confirm our hypothesis that tamoxifen blocks estrogen
binding to the ER, preventing HMG1 upregulation and subsequent cisplatin
sensitization in cells.
Cell Proliferation and Sensitivity to Other Cytotoxic Agents. Estrogen induces
general cell proliferation (52) and regulates human mammary epithelial cell
morphogenesis (Table 3.1) (33, 53). We have considered the possibility that the
sensitization of cells toward cisplatin may be a consequence of hormone-induced cell
proliferation. Accordingly, we investigated whether hormone treatment conditions
used in the cell survival assays affected cell growth. Fluorescence assisted cell sorting
(FACS) analysis revealed no change in cell cycle profile following hormone treatment
(Table 3.2). Cell proliferation rate, determined in cell counting assays, was similarly
unaffected by the transient hormone treatments (Figure 3.14).
trans-DDPis a clinically inactive isomer of cisplatin that forms DNA adducts not
recognized by HMG1 (16). The trans-DDPsensitivity of MCF-7 cells was unaffected by
estrogen treatment (Figure 3.15).
Calicheamicin is another cytotoxic agent, which
causes double-stranded DNA cleavage (54) in a manner that does not involve HMG1.
The sensitivity of MCF-7 cells towards calicheamicin was also unaffected by estrogen
treatment (Figure 3.15). The above evidence argues against higher transcriptional
activity being the main reason for the enhanced platinum-induced toxicity of treated
cells. We conclude that the hormone treatment did not sensitize the cells to all cytotoxic
agents, but only to cisplatin or carboplatin via a pathway that involves HMG1,
presumably by a repair-shielding mechanism.
Platinum-DNA Adduct Levels. Another hypothesis is that DNA platination levels
could be elevated as a consequence of hormone treatment, based on reports that active
promoter sites are preferentially platinated by the drug (55, 56). Such sites should be
equally accessible to cisplatin and trans-DDP,however, and the trans-DDPsensitivity
was not influenced by hormone treatment. We have investigated the bound-platinum
levels on genomic DNA in cells treated with steroid hormones by platinum atomic
absorption spectroscopy. MCF-7 cells were treated with a range of cisplatin
concentrations (0 to 100 pM) for 4 h and then immediately harvested and genomic DNA
was extracted to evaluate the initial platination levels. The platinum signals were not
detected with the atomic absorption spectrometer in samples treated with 10 pM, 20 pM
or 50 pM cisplatin. For samples treated with 100 pM of cisplatin, the rb values, defined
as the number of platinum atom per nucleotide, were 3.48 x 10 ' Pt/n.t. for control cells
and 3.33 x
10 '
Pt/n.t. for estrogen treated cells. The results indicate that the platinum
adduct levels on genomic DNA are comparable in control cells and cells treated with
estrogen.
Similar experiments were also performed with BG-1 cells. BG-1 cells were
treated with cisplatin (0 to 100 pM) for 4 h with or without estrogen. Platinum atomic
absorption spectroscopy was performed on the genomic DNA extracted from the
samples. For samples treated with 50 pM of cisplatin, the rb values were 4.6 x 104 for
control cells and 3.2 x 10-4 for estrogen treated cells. A [32P] post-labeling method was
also used to estimate the amount of platinum-DNA adducts. As shown in Figure 3.16,
the amount of GG and AG adducts were comparable in either the control cells or
estrogen-treated cells. Synthetic, purified GpG and ApG were used as standards. The
bottom band corresponded to free ATP, and the top band was present in all lanes
including untreated DNA samples and not identified (57). We conclude that the
transient hormone treatment did not cause a higher level of platinum-DNA adducts in
cells.
Apoptosis Assays. Cytotoxicity assays assess total survival rate but do not
differentiate between the mechanisms of cell death. It also takes 10 days to form cell
colonies. We first tried DAPI staining to detect apoptosis. DAPI is a blue fluorescent
dye that binds to the DNA in the major groove. When cells are undergoing apoptosis, a
series of morphological changes occur, including chromatin condensation. Using DAPI
staining, this condensation can be visualized, and apoptotic cells can be identified and
counted to determine the percentage of such cells within a population.
Under 24 h continuous cisplatin treatment, HeLa cells treated with 1 pM cisplatin
did not exhibit high levels of apoptosis (Figure 3.17A). With 2 pM of cisplatin, there
were more apoptotic cells as seen by an condensed chromatin. The number of apoptotic
cells increased as the concentration of cisplatin treatment was increased from 5 pM to 20
-pM. In contrast, MCF-7 cells treated under these conditions from 1 pM to 20 pM
cisplatin did not show the same DAPI staining pattern (Figure 3.17B). A slightly
different pattern was seen, with highly condensed DAPI staining along a condensed
nuclear envelope. This staining pattern was observed in cells treated with 10 pM and 20
pM cisplatin. Condensation of the nucleus is a morphological change that occurs in
95
cells undergoing apoptosis, so the DAPI staining may indicate that this process occurs
in these MCF-7 cells. Since the phenotype is not very obvious, it was not possible to
quantitative for the percent of apoptotic cells.
A TUNEL-based FACS assay was then performed on MCF-7 cells. No apoptotic
population could be detected in the FACS results, however, although many cisplatin
conditions were tried (Figure 3.18). To obtain good staining on the fragmented DNA,
the cell and nuclear membranes must be well permeablized to permit antibody entry
and binding. It is possible that the inability to detect TUNEL signal was caused by
insufficient fixation and permeabilization. This assay has been known to give false
negative results (Glenn Paradis, MIT Cancer Center, personal communications).
Clinical Implications of the Hormone/CarboplatinCombination Therapy. In summary,
all available evidence supports the hypothesis that hormone receptors are essential for
HMG1 up-regulation and subsequent cisplatin/carboplatin sensitization in the cells.
Cisplatin produces many side effects including nephrotoxicity, neurotoxicity, and
emesis (58). The present study shows that the potency of cisplatin can be increased
through hormone treatment. From a clinical perspective, estrogen and/or progesterone
treatment should allow the currently applied cisplatin regimen to increase its
cytotoxicity towards cancer cells. Even a factor of two increases in sensitibity could be
quite important in medical applications. In addition, the enhanced potency of the drug
may be sufficient to overcome some types of acquired and intrinsic resistance.
Responses to hormone can vary between tissues and individuals due to
differences in the hormone receptor distribution. In normal mammary glands, estrogen
receptor levels are estimated to be between 5000 and 20,000 ER molecules per cell (32).
Between 50-60% of human breast cancers overexpress the ER, perhaps because of
transcription from a promoter that is inactive in normal breast cells (59-61). Between
40-50% of ovarian cancers have high ER and PR levels (60, 61).
The status of the
receptor is used to diagnose breast cancer and determine clinical therapy. Compounds
that inhibit estrogen and androgen action, such as tamoxifen, are commonly used as
endocrine therapy for breast and prostate tumors.
Cancers with functional ER and PR would be good candidates for
cisplatin/carboplatin treatment in conjunction with hormone therapy, because they can
overexpress HMG1. Since estrogen has been implicated in the etiology of breast cancer
owing to its proliferative properties, progesterone may be preferable for combination
treatment of breast cancer patients. For ovarian and cervical cancer patients, carboplatin
is the standard chemotherapeutic agent because of its diminished nephrotoxicity
compared to that of cisplatin (42). Although pathways other than repair shielding by
HMG1 may be responsible for the observed sensitization, this work establishes the
potential to treat ovarian cancer patients with estrogen/progesterone in combination
with carboplatin in clinical trials.
A clinical protocol was developed in conjunction with Dana Farber Cancer
Institute researchers to combine estrogen and progesterone with carboplatin to treat
ovarian cancer. Women with ovarian cancer often manifest a buildup of ascites fluid in
their abdomen into which malignant cancer cells are shed. In the clinical study with the
Dana Farber Cancer Institute and Massachusetts General Hospital, the ascites fluid will
be tapped before hormone treatment, as well as 4 h and 24 h after hormone treatment.
The levels of HMG1 in the malignant cells will be detected by immunofluorescence, and
the fluorescence intensity before any treatment will be compared to that at 4 h or 24 h
after treatment to detect any change in the HMG1 expression in ovarian cancer cells.
The application of immunofluorescence to analyze changes in HMG1 levels in
clinical samples is complicated by the presence of many different cell types in the
ascites fluid. Ascites fluid may contain leukocytes, blood cells, and mesothelial cells in
addition to the malignant ovarian cancer cells of interest. The percentage of ovarian
cancer cells in the fluid varies greatly between patients but is generally low, from 1%10%, making comparison of HMG1 levels in malignant cells quite difficult.
To
distinguish ovarian cancer cells from all the other types of cells, cells first will be stained
for the protein CA125. CA125 is a receptor expressed on the surface of ovarian cancer
cells but not on the other cells in the ascites fluid. CA125 is a high molecular weight
glycoprotein, the detection of which in the blood often is used as a diagnostic screening
assay for ovarian cancer (62).
Detection of CA125 Expression on the Surface of Ovarian Cancer Cells in Vitro. To
develop a protocol to analyze clinical samples, fixation/permeabilization conditions
and antibody dilutions were optimized not only to achieve good CA125 staining, but
also to reflect accurately relative HMG1 levels in BG-1 cells. The CA125 status of BG-1
ovarian cancer cells has not been reported in the literature. Double-labeling for CA125
(detected with a TRITC conjugated secondary antibody) and HMG1 (detected with a
FITC
conjugated
secondary
antibody)
was
used
with
different
fixation/permeabilization conditions to detect whether BG-1 cells express CA125 on
their surface. Under all staining conditions, BG-1 cells showed specific expression of
CA125 on some of the cell surface. Staining for CA125 clearly followed the shape of the
cell membrane, showing a speckled appearance consistent with the published results
for such CA125 staining of other ovarian cancer cell lines (62). Only 10% of BG-1 cells
express CA125 (Figure 3.19). Expression of CA125 is suspended when ovarian cancer
cells from patients are immortalized (63). It is possible that the percentage of BG1
CA125+ cells is low because this immortalized BG-1 cell line was passaged many times.
Since the clinical samples should be neither immortalized nor passaged, loss of CA125
expression should not be an issue. A more important factor to consider is that only 5060% of ovarian tumors are CA125 positive (64). Thus, some of the clinical ovarian
carcinomas will not express CA125. Cell morphology could reveal the different cell
types, so consulting a pathologist can help identify the cancer cells.
Fixation/permeabilization methods significantly affected the HMG1 staining of
BG-1 cells. Fixation and permeabilization with 70% ethanol yielded non-specific
labeling on the surface with no nuclear HMG1 labeling (Figure 3.19). Cells fixed and
permeablized with 2% paraformaldehyde/0.5% NP-40 showed very strong HMG1
labeling in the nucleus (Figure 3.19), but the FITC intensity was very intense that it was
difficult to distinguish changes in fluorescence in cells with decreasing dilutions of
HMG1 antibody. With 3:1 methanol:acetic acid, there was specific labeling of HMG1 in
the nucleus (Figure 3.19), allowing for detection of changes in fluorescence intensity
reflecting changes in HMG1 levels.
Double-labeling also was performed with another ovarian cancer cell line,
NIH:OVCAR-3 (OVCAR-3) cells, to test whether the same protocol could be used
successfully with different ovarian cell lines. Unlike BG-1 cells, which were established
from a solid tumor, OVCAR-3 cells were established from ovarian-cancer cells from
ascites fluid, making it a better model to optimize staining conditions for clinical
samples. Under all fixation/permeabilization conditions investigated, OVCAR-3 cells
showed very strong CA125 expression on the cell surface (Figure 3.20). In contrast to
the heterogeneous CA125 expression observed in BG-1 cells, all OVCAR-3 cells were
positive for CA125. OVCAR-3 cells were stained after only a few passages, so it is
possible that after many more passages, these cells also could lose their CA125
expression.
As with the BG-1 cells, not all fixation/permeabilization methods produced good
HMG1 staining in the nucleus. With the 3:1 methanol:acetic acid fixative, antibodies
appeared to be unable to enter the nucleus and only non-specific staining on the surface
was seen (Figure 3.20). This result was unexpected since 3:1 methanol:acetic acid has
been used successfully to obtain excellent staining for HMG1 in the nucleus in HeLa,
Evsa-T, MCF-7, and BG-1 cells. Non-specific HMG1 labeling on the cell surface also
occurred with the 70% ethanol/0.5% NP-40 fixative (Figure 3.20). Only with the 1% and
2% paraformaldehyde/0.5% NP-40 fixatives was specific HMG1 staining in the nucleus
obtained (Figure 3.20).
In conclusion, CA125 expression was detected on the surface of two ovarian
cancer cells lines, BG-1 and OVCAR-3 by using immunofluorescence under several
fixation/permeabilization conditions. Detection of HMG1 protein in the nucleus of the
cells was possible under at least one of the fixation/permeabilization conditions tested
for each of the two cell lines. The application of immunofluorescence to detect CA125
and HMG1 in clinical ovarian cancer samples appears promising.
Conclusions and Future Directions
It has been demonstrated that HMG1 can be transiently overexpressed in cells
with the appropriate hormone receptor after steroid hormone treatment. HMG1
already exists at a high level, between 10,000 to 100,000 copies per cell (25), and an
excess amount of the protein is toxic (65), which may explain why only a moderate level
of overexpression of HMG1 was observed in the present study. Why then does the
moderate up-regulation observed here increase drug sensitivity? One possibility is that
endogenous HMG1 proteins are already involved in complexes with chromatin and
transcription factors. HMG1 transiently expressed as a scaffold to facilitate estrogen or
100
progesterone receptor mediated transcription may be more readily available to bind to
cisplatin-DNA intrastrand cross-links. The two-fold difference in cisplatin sensitivity is
in good accord with the approximately two-fold increase in HMG1 protein levels.
The correlation between up-regulation of HMG1 protein and increase in cisplatin
sensitivity of cancer cells adds to growing evidence that a repair shielding mechanism
plays a role in the HMG1 enhancement of cisplatin cytotoxicity. However, these studies
fail to prove that HMG1 overexpression is the absolute cause for the increase in
cisplatin sensitivity. Hormone treatment causes many physiological changes in cells
that could contribute to the increased sensitivity to cisplatin. Control experiments have
ruled out several alternative hypotheses for the sensitization. Transient hormone
treatment did not cause cell growth proliferation, nor did it make the cells generally
sensitive to cytotoxic agents or induce higher platinum-DNA adducts in hormone
treated cells. There may be many other genes to be discovered that are under the
hormone control and can potentially cause cisplatin sensitization, our HMG1 repairshielding model is still just a hypothesis.
The hypothesis can be further tested by an antisense RNA strategy to inhibit the
overexpression of HMG1 in hormone treated cells. If the cisplatin sensitivity of
hormone treated cells did not persist in the presence of the HMG1 antisense vector, the
increase in cisplatin cytotoxicity in cells could be attributed to HMG1 overexpression.
Recently, our laboratory has obtained both a HMG1 -1-mice fibroblast cell line and the
corresponding HMG1 /' ÷ cell line. These cells are from HMG1 knock-out mouse (38). It
/ cells will be more resistant towards cisplatin than the control
is expected that HMG1-
cells. HMG1 will be reintroduced into HMG1V cells with a vector to test for restored
cisplatin sensitivity.
101
Although pathways other than repair shielding by HMG1 may be responsible for
the sensitization, the fundamentals of this work have led to a clinical protocol for
ovarian cancer patients to receive combination therapy of hormone and carboplatin at
Dana Farber Cancer Institute and Massachusetts General Hospital. Ascites samples will
be collected from patients and subjected to protocols for measuring relative HMG1
levels developed in ovarian cancer cells. CA125 can be detected clearly on the surface
of ovarian cancer cells and will be a useful marker for identifying malignant cells in a
heterogeneous cell population.
The limits for the usage of these platinum drugs include severe side effects and
propensity for acquired resistance. By increasing the cytotoxicity of these platinum
drugs, a decrease in tumor resistance may be achieved. We hope for exciting results
from our clinical study based on in vitro studies showing the effectiveness of steroid
hormones in sensitizing ovarian cancer cells with the appropriate hormone receptors to
cisplatin and carboplatin treatment.
Acknowledgment:
Cynthia H. Liang was instrumental in performing many of the experiments described. I
thank Dr. J.M. Essigmann for access to tissue culture facilities and his helpful input in
the work. Dr. S. S. Marla synthesized and purified GpG and ApG standards used for
32P
post-labeling. I thank MIT Center for Cancer Research for providing the FACS
facilities. I appreciate help from R. Hynes and J. Trevithick in conjunction with the
immunofluorescence study.
102
References:
1. Part of this chapter is published in reference 2.
2. He, Q., Liang, C., and Lippard, S. J. (2000) Proc. Nat. Acad. Sci. USA 97, 5768-5772.
3. Loehrer, P. J., and Einhorn, L. H. (1984) Ann. Intern. Med. 100, 704-713.
4. Morris, M., Eifel, P. J., Lu, J., Grigsby, P. W., Levenback, C., Stevens, R. E., Rotman,
M., Gershenson, D. M., and Mutche, D. G. (1999) The New England Journal of Medicine
340, 1137-1143.
5. Rose, P. G., Bundy, B. N., Watkin, E. B., Thigpen, J. T., Deppe, G., Maiman, M. A.,
Clarke-Pearson, D. L., and Insalaco, S. (1999) The New England Journal of Medicine 340,
1144-1153.
6. Keys, H. M., Bundy, B. N., Stehman, F. B., Muderspach, L. I., Chafe, W. E., Suggs, C.
L., Walker, J. L., and Gersell, D. (1999) The New England Journal of Medicine 340, 11541161.
7. Thomas, G. M. (1999) The New England Journal of Medicine 340, 1198-1200.
8. Yang, D., and Wang, A. J. (1996) Prog.Biophys. Molec. Biol. 66, 81-111.
9. Gelasco, A., and Lippard, S. J. (1998) in Topics in Biol. Inorg. Chem. (Clarke, M., Ed.)
pp 1-43, Springer-Verlag, Berlin.
10. Pinto, A. L., and Lippard, S. J. (1985) Proc. Nat. Acad. Sci. USA 82, 4616 - 4619.
11.
Ciccarelli, R. B., Solomon, M. J., Varshavsky, A., and Lippard, S. J. (1985)
Biochemistry 24, 7533-7540.
12.
Comess, K. M., Burstyn, J. N., Essigmann, J. M., and Lippard, S. J. (1992)
Biochemistry 31, 3975 - 3990.
13. Mello, J.A., Lippard, S. J., and Essigmann, J. M. (1995) Biochemistry 34, 14783-14791.
14. Jamieson, E. R., and Lippard, S. J. (1999) Chem. Rev. 99, 2467-2498.
15. Bustin, M., and Reeves, R. (1996) Prog.Nucleic Acid Res. Mol. Biol. 54, 35-100.
103
16. McA'Nulty, M. M., and Lippard, S. J. (1995) in Nucleic Acids and Molecular Biology
(Eckstein, F., and Lilley, D. M. J., Eds.) pp 264-284, Springer-Verlag, Berlin.
17. Whitehead, J. P., and Lippard, S. J. (1996) in Metal Ions in Biological Systems (Sigel, A.,
and Sigel, H., Eds.) pp 687-725, Marcel Dekker, Inc., New York.
18. Huang, J., Zamble, D. B., Reardon, J. T., Lippard, S. J., and Sancar, A. (1994) Proc.
Nat. Acad. Sci. USA 91, 10394 - 10398.
19.
Zamble, D. B., Mu, D., Reardon, J. T., Sancar, A., and Lippard, S. J. (1996)
Biochemistry 35, 10004-10013.
20. Trimmer, E. E., Zamble, D. B., Lippard, S. J., and Essigmann, J. M. (1997)
Biochemistry 37, 352-362.
21. Li, L., Liu, X., Glassman, A. B., Keating, M. J., Stros, M., Plunkett, W., and Yang, L.
(1997) Cancer Research 57, 1487-1494.
22. Brown, S. J., Kellett, P. J., and Lippard, S. J. (1993) Science 261, 603-605.
23. McA'Nulty, M. M., and Lippard, S. J. (1996) Mutat. Res. 362, 75-86.
24.
Arioka, H., Nishio, K., Ishida, T., Fukumoto, H., Fukuoka, K., Nomoto, T.,
Kurokawa, H., Yokote, H., Abe, S., and Saijo, N. (1999) Jpn. J. Cancer Res. 90, 108-115.
25. Zlatanova, J., and van Holde, K. (1998) BioEssays 20, 584-588.
26. Landsman, D., and Bustin, M. (1993) BioEssays 15, 539 - 546.
27. Bianchi, M. E., and Beltrame, M. (1998) Am. J. Hum. Genet. 63, 1573-1577.
28. Falciola, L., Spada, F., Calogero, S., Langst, G., Voit, R., Grummt, I., and Bianchi, M.
E. (1997) J. Cell Biol. 137, 19-26.
29. Grosschedl, R., Giese, K., and Pagel, J. (1994) Trends Gen. 10, 94 - 100.
30. Grosschedl, R. (1995) Current Opinion in Cell Biology 7, 362-370.
104
31.
Boonyaratanakomkit, V., Melvin, V., Prendergast, P., Altmann, M., Ronfani, L.,
Bianchi, M. E., Taraseviciene, L., Nordeen, S. K., Allegretto, E. A., and Edwards, D. P.
(1998) Molecular and CellularBiology 18, 4471-4487.
32. Greenspan, F. S., and Strewler, G. J. (1999), Appleton & Lange, Stamford.
33. Ciocca, D. R., and Fanelli, M. A. (1997) Trends Endocrinol.Metab. 8, 313-321.
34. Nardulli, A. M., Greene, G. L., and Shapiro, D. J. (1993) Molecular Endocrinology 2,
331-340.
35. Zhang, C. C., Krieg, S., and Shapiro, D. J. (1999) Molecular Endocrinology 13, 632-643.
36. Onate, S. A., Prendergast, P., Wagner, J. P., Nissen, M., Reeves, R., Pettijohn, D. E.,
and Edwards, D. P. (1994) Molecularand CellularBiology 14, 3376-3391.
37. Melvin, V. S., and Edwards, D. P. (1999) Steroids 64, 576-586.
38. Calogero, S., Grassi, F., Aguzzi, A., Voigtlander, T., Ferrier, P., Ferrari, S., and
Bianchi, M. E. (1999) Nature Genetics 22, 276-280.
39. Chau, K. Y., Lam, H. Y. P., and Lee, K. L. D. (1998) Experimental Cell Research 241,
269-272.
40. Baldwin, W. S., Curtis, S. W., Cauthen, C. A., Risinger, J. I., Korach, K. S., and
Barrett, J. C. (1998) In Vitro Cell. Dev. Biol. 34, 649-654.
41. Borras, M., Lacroix, M., Legros, N., and Leclercq, G. (1997) Cancer Letters 120, 23-30.
42. Fichtinger-Schepman, A. M. J., van Dijk-Knijnenburg, H. C. M., van der VeldeVisser, S. D., Berends, F., and Baan, R. A. (1995) Carcinogenesis 16, 2447-2453.
43. Los, G., Verdegaal, E., Noteborn, H. P. J. M., Ruevekamp, M., Graeff, A. D.,
Meesters, E. W., Huinink, D. T. B., and McVie, J. G. (1991) Biochemical Pharmacology42,
357-363.
44. Hongo, A., Seki, S., Akiyama, K., and Kudo, T. (1994) Int. J. Biochem. 26, 1009-1016.
105
45. Bancroft, D., Lepre, C. A., and Lippard, S. J. (1990) Journal of the American Chemical
Society 112, 6860-6871.
46. Knox, R. J., Friedlos, F., Lydall, D. A., and Roberts, J. J. (1986) Cancer Research 1986,
1972-1979.
47. Hossain, M. B., Vanderhem, d., Pordesimo, E. O., Magarian, R. A., Meyers, K. L.,
Overacre, L. B., and Day, B. W. (1994) Journal of Medicinal Chemistry 37, 1670-1683.
48. McClay, E. F., Albright, K. D., Jones, J. A., Christen, R. D., and Howell, S. B. (1993)
CancerResearch 53, 1571-1576.
49. Jones, J. A., Albright, K. D., Christen, R. D., Howell, S. B., and McClay, K. F. (1997)
CancerResearch 57, 2657-2660.
50. Scambia, G., Ranelletti, F. O., Panici, P. B., Piantelli, M., Vincenzo, R. D., Bonanno,
G., Ferrandina, G., Isola, G., and Mancuso, S. (1992) European Journalof Cancer 28A, 18851889.
51. Ercoli, A., Scambia, G., De Vincenzo, R., Alimonti, A., Petrucci, F., Fattorossi, A.,
Isola, G., Panici, P. B., Caroli, S., and Mancuso, S. (1996) Cancer Letters 108, 7-14.
52. Lippman, M., Bolan, G., and Huff, K. (1976) Cancer Research 36, 4595-4601.
53. Diaz-Perez, M. J., Zannis-Hadjopoulos, M., Price, G. B., and Wainer, I. W. (1998)
Journalof CellularBiochemistry 70, 323-329.
54. Kuduvalli, P. N., Townsend, C. A., and Tullius, T. D. (1995) Biochemistry 34, 38993906.
55. Haghighi, A., Lebedeva, S., and Gherset, R. A. (1999) Biochemistry 38, 12432-12438.
56. Caliaro, M. J., Vitaux, P., Lochon, I., Nehme, A., Valette, A., Canal, P., Bugat, R., and
Jozan, S. (1997) British Journalof Cancer 75, 333-340.
57. Blommaert, F. A., and Saris, C. P. (1995) Nucleic Acids Research 23, 1300-1306.
106
58. Van Hoff, D. D., Schilsky, R., Reichert, C. M., Reddick, R. L., Rozenscweig, M.,
Young, R. C., and Muggia, F. M. (1979) CancerTreat. Rep. 63, 1527-1531.
59. Weigel, R. J., Crooks, D. L., Iglehart, J. D., and deConinck, E. C. (1995) Cell Growth
Diff. 6, 707-711.
60. Femrno, M., Johansson, B. U., Olsson, N. H., Ryden, S., and Sellberg, G. (1990) Acta
Oncol. 29, 129-135.
61. Slotman, B.J., and Rao, B. R. (1988) Anticancer Research 8, 417-434.
62. Chang, K., Pastan, I., and Willingham, M. C. (1992) InternationalJournalof Cancer 50,
373-381.
63. Auersperg, N., Mainesbandiera, S., Booth, J. H., Lynch, H. T., Godwin, A. K., and
Hamilton, T. C. (1995) American Journalof Obstetricsand Gynecology 173, 558-565.
64.
Leake, J., Woolas, R. P., Daniel, J., Oram, D. H., and Brown, C. L. (1994)
Histopathology24, 57-64.
65. Wang, H., Bloom, O., Zhang, M., Vishnubhakat, J. M., Ombrellino, M., Che, J.,
Frazier, A., Yang, H., Ivanova, S., Borovikova, L., Manogue, K. R., Faist, E., Abraham, E.,
Andersson, J., Andersson, U., Molina, P. E., Abumrad, N. N., Sama, A., and Tracey, K. J.
(1999) Science 285, 248-251.
107
Table 3.1. Estrogen-regulated genes or proteins involved in cell growth. Adapted from
Ciocca, D.R. and Fanelli, M.A., Trends Endocrinol.Metab. (1997), 8: 313-321
Gene/Protein
Function
c-jun, jun-B/jun-D
Protooncogene, transcription regulator
N-myc, c-myc
Protooncogene, transcription regulator
c-myb
Protooncogene, transcription regulator
c-fos
Protooncogene, transcription regulator
ras
Oncogene, transducer, GTP binding
c-src3 and c-yes
Tyrosine kinases
erk-1 and erk-2
Serine/threonine kinases
TGF-a
Transforming growth factor alpha
Amphiregulin
Epidermal growth factor-related peptide
EGF
Epidermal growth factor
IGF-I/II
Insuline-like growth factors
IGFBPs
Insuline-like growth factor binding proteins
CSF-1
Colony stimulating factor
bFGF
Basic fibroblast growth factor
GH
Growth hormone
PRL
Prolactin
EGF R
Epidermal growth factor receptor
c-erB-2
Protooncogene, protein involved in cell growth
IGF-I, IGFBP-2 R
Insuline-like growth factor/binding proteins receptors
PRLR
Prolactin receptor
GH R
Growth hormone receptor
PR
Progesterone receptor
TK/DNA polymerase/ODC
Thymidine kinase, DNA polymerase
Cyclins D1, B1, E
Cell cycle regulators
TFG P
Transforming growth factor 3
RAR a
Retinoic acid receptor ca
Hsp27
Heat shock protein 27,000
p53
Antioncogene
p21 WAF1/CP1
Cell cycle inhibitor
108
Table 3.2. Cell cycle profiles of MCF-7 cells after hormone treatments.
Treatment Conditions
GO/G1
S
G2/M
DMF, t = 4 h
57%
27%
13%
10-7M estrogen, t = 4 h
57%
27%
17%
10'6 M progesterone, t = 4 h
61%
27%
12%
DMF, t = 4 h, wait 24 h
67%
24%
6%
10-7 M estrogen, t = 4 h, wait 24 h
61%
32%
7%
106 M progesterone, t = 4 h, wait 24 h
67%
24%
9%
109
17-3 estradiol
Progesterone
-CH 2OH
Hydrocortisone
Figure 3.1. The structures of steroid hormones.
Testosterone
110
Estrogen
Diffuses through membrane
?ý?ý?
55941
ggg
I
Activated estrogenreceptor complex
Estrogen receptor
complexed with heat shock
proteins
Low affinity binding
\1
\&QV2Estrogen
%4r/ responsive
element
HMG-1
^
Nuclear membrane
High level of
Transcription
Figure 3.2. The mechanism of estrogen receptor action and HMG1
involvement. Adapted from Onate et al, Mol. and Cell. Biol., (1994), 14: 33763391
111
MCF-7 cells
Evsa-T cells
0.00
0.25
0.50
1.00
1.50
3.00
5.00
24.00
Time of Estrogen Treatment (h)
Figure 3.3. A bar chart showing the relative HMG1 mRNA level change in MCF-7 and Evsa-T
cells under estrogen treatment. Reprinted from Chau et al, ExperimentalCell Research, (1998),
241:269-272.
112
r,,h'u
LUU -
150-
100-
50-
0. I.
0-
0
2
4
24
Treatment time (h)
200U -
150.
100-
500-
iI
0
2
4
24
Treatment time (h)
Figure 3.4. Western blot quantitation of HMG1 protein levels in MCF-7
cells after 10-7 M of estrogen treatment.
Levels were normalized
according to the sample without estrogen treatment. A and B are two
separate experiments
113
A
2h
1h
0.5 h
I
B
C
I
I
24h
4h
I
I
I
I
Figure 3.5. HMG-1 protein levels in hormone-trea ted MCF-7 cells detected by
immunofluorescence. Fluorescence images of MCF-7 cells (A) untreated, (B) treated
with 10-7 M estrogen, and (C) treated with 106 M progesterone for 0.5 24 h.
114
0.5 h
2h
1h
A
B
I
24 h
4h
!i
I
C
Figure 3.6. HMG-1 protein levels in hormone treated Evsa-T cells detected by
immunofluorescence. Fluorescence images of Evsa-T cells (A) untreated, (B)treated
with 10-7 M estrogen, and (C) treated with 10-7 M progesterone for 0.5 - 24 h.
115
0.5 h
2h
24 h
I
I
Figure 3.7. HMG-1 protein levels in hormone treated BG-1 cells detected by
immunofluorescence. Fluorescence images of BG-1 cells (A) untreated (B) treated
with 10-7 M estrogen, and (C) treated with 10-7 M progesterone for 0.5 - 24 h.
116
IIHI
IINI
A
C"4
0a
0
1
2
3
4
5
6
7
R
[cisplatin] (pM)
[cisplatin] (pM)
100
^^
IUU00
D
C
10
1
1
0.1
[cisplatin] (pM)
[cisplatin] (pM)
Figure 3.8. Cell survival assays. (A) The effects of estrogen and progesterone co-treatment on
-7
cisplatin sensitivity of MCF-7 cells. MCF-7 cells were co-treated with 2 x 10-7 M estrogen, 2 x 10 M
progesterone, or both hormones with cisplatin for 4 h. (B) The effects of hormones on cisplatin
sensitivity of Evsa-T cells. Evsa-T cells were pre-treated with 10-7 M estrogen or 10-7 M progesterone
for 2 h prior to a 4 h cisplatin treatment. (C) The effects of hormones on cisplatin sensitivity of BG-1
cells. BG-1 cells were co-treated with 10-7 M of estrogen or progesterone with cisplatin for 4 h. (D)
The effects of 10-7 M estrogen on cisplatin sensitivity of HeLa cells.
117
W
P-
[cisplatin] pM
Figure 3.9. The effects of hormone concentration on cisplatin sensitivity of
MCF-7 cells. MCF-7 cells were cotreated with 10 -7 M, 10-6 M, and 10-5 M estrogen
and cisplatin.
118
~T\A
AA
IUU.UU
10.00
1.00
0.0
2.0
4.0
6.0
8.0
10.0
[cisplatin] (pM)
Figure 3.10. The timing of estrogen treatment affects cisplatin sensitivity
of MCF-7 cells. MCF-7 cells were either cotreated with estrogen, or treated
with 10- 7 M estrogen in DMF for 2 h or 24 h prior to cisplatin treatment. The
cells were treated with cisplatin for 4 h.
119
100
p 10
0-0
1
0.1
[carboplatin] (pM)
Figure 3.11. Cell survival assays of MCF-7 cells toward carboplatin with or without
estrogen treatment. Cells were pretreated with carboplatin for 24 h, and then treated
with estrogen for 4 h.
120
100-
.
h
10-
-
-m- -lU
M tamoxiten
------ 10-7 Mestrogen and
10-7 M tamoxifen
1
1-
0
I
1
,
I
2
r
I
3
r
I
4
1
I
5
I
I
I
6
I
7
1
I
8
,
I
9
10
[cisplatin] pM
Figure 3.12. The effects of estrogen and tamoxifen cotreatment on cisplatin
sensitivity of MCF-7 cells. MCF-7 cells were cotreated with estrogen or tamoxifen
or both with cisplatin for 4 h.
121
A
B
Figure 3.13. HMG-1 protein levels in tamoxifen treated MCF-7 cells detected by
immunofluorescence. Fluorescence images of MCF-7 cells treated with
10-7 M tamoxifen (A) or 10 7 M tamoxifen plus 10-7 M estrogen (B) for 0 h, 0.5
h, 1h,2h, or 4h.
122
A1
40
"* 35
*
7
30
30
25
20
-
20
15
10
C5
0
10
20
30
40
50
60
70
80
time (h)
Figure 3.14. Cell proliferation profile of MCF-7 cells. Shown are control,
with 10-7 M of estrogen treatment for 4 h, and with 10-6 M of progesterone
treatment for 4 h.
123
IUU
A
10
50
0
150
100
(pM)
[trans-DDP]
Iuu
B
10
1
0
0.05
0.1
0.15
0.2
0.25
[calicheamicin] (pM)
Figure 3.15. The effect of estrogen on the sensitivity of MCF-7 cells towards other
cytotoxic agents. Cell survival assays of MCF-7 cells towards trans-DDP(A) or
calicheamicin (B) with or without 10-7 M estrogen.
124
gr~
o-ounidentified
4 -AG
GG
1
2
3
4
SI
5
6
7
40
8
free ATP
9
Figure 3.16. 32p post-labeling methods to detect platinum-DNA adducts, Pt-GG and PtAG. Lane 5 is the GG control, and lane 6 is the AG control. Lane 1-4 are samples
without hormone, and lane 7-9 are samples treated with 10-7 M estrogen for 4 h. Lane
1, no platinum treatment. Lane 2 and 7, [Pt] = 5 pM, lane 3 and 8, [Pt] = 10 pM, lane 4
and 9, [Pt] = 50 pM.
125
|
A
,,
--~------------
B
Figure 3.17. DAPI staining of (A) HeLa cells or (B) MCF-7 cells treated with
(i) 0 1pM, (ii) 1 pM, (iii) 2 pM, (iv) 5 pM, (v) 10 pM, or (vi) 20 pM cisplatin for 24 h.
126
Non
Apoptotic Apoptotic
0
OD
A
0
C
0
s
0
01
0
o
i,
cJ
D
0
C
C
z
0
0
Q
o
FL1-H
Cr
-L1-H
Figure 3.18. TUNEL based FACS analysis to detect apotosis. (A) Positive control cells
containing apoptotic cells. (B-D) MCF-7 cells treated with 0 (B), 5 pM (C) and 10 pM (D)
of cisplatin for 24 h.
127
70% ethanol
2% PFA/0.5% NP-40
3:1 methanol:acetic acid
Figure 3.19. BG-1 cells stained for HMG-1 and CA125 under different fixation/
permeabilization conditions. Under each condition the same cells were visualized by
(A) phase contrast, (B)FITC fluorescence for HMG-1 staining, or (C) TRITC fluorescence
for CA125 staining.
128
70% ethanol/0.5% NP-40
3:1 methanol:acetic acid
1%PFA/0.5% NP-40
2% PFA/0.5% NP-40
Figure 3.20. OVCAR-3 cells stained for HMG-1 and CA125 under different fixation/
permeabilization conditions. Under each condition the same cells were visualized by
(A) phase contrast, (B) FITC fluorescence for HMG-1 staining, or (C) TRITC fluorescence
for CA125 staining.
129
Chapter 4
The Search for Novel HMG-Domain Proteins with High Affinity toward
Cisplatin-Modified DNA by Phage Display
130
Introduction
High-mobility group (HMG) domain proteins bind 1,2-intrastrand
cisplatin-DNA adducts to form stable platinum-DNA-protein ternary complexes.
The affinity and stability of such complexes are important for cisplatin activity in
repair shielding and hijacking pathways (1-5) and, if increased, may enhance the
cisplatin cytotoxicity. As shown in Chapter 2, the affinity of HMG1 domains for
platinated DNA can be improved by side-chain modifications.
Chapter 3
demonstrated that higher levels of endogenous HMG1 correlate with enhanced
cisplatin sensitivity in mammalian cells. In this chapter, we use a combinatorial
system to search for novel HMG-domain proteins with high affinity towards
cisplatin-modified DNA. Such proteins may potentially augment cisplatin
chemotherapy.
The phage display protocol was first developed by George P. Smith and
others (6-9). Phage is a virus that infects bacteria, using the bacterial machinery
to produce progeny. A typical M13 filamentous phage is a flexible rod about 1
pm long and 6 nm in diameter. A phage particle is composed of a singlestranded viral DNA (6407 nucleotides) encapsulated by 2700 copies of pVIII, the
major coat protein, and five copies of each minor proteins, pIII, pVI, pVII and
pIX (Figure 4.1) (10). The major pVII coat proteins are on the sides of the phage
and minor coat proteins, such as pIll, are at the ends (Figure 4.1). Infection of
bacteria is initiated by the attachment of the N-terminal domain of p1I to the tip
of the thread-like appendage called the F' pilus of the male bacteria, after which
the genome is released into the bacterium (11)(12). Inside the bacterium, the
phage genome can replicate itself and use the bacterial machinery to produce
coat proteins. For a lytic phage system, the assembly of the single stranded
131
genome into coat proteins occurs inside the cell, and the bacterium is lysed to
release the phage progenies. For non-lytic filamentous phage, the coat proteins
and the genome are transported through the cell envelope and assembled on the
bacterial surface, a process that couples assembly with export (10). The phage
progeny are continuously extruded from the bacteria.
Foreign DNA can be cloned into the phage genome and subsequently
expressed by the phage-infected E. coli machinery. In most cases, the foreign
amino acid sequence is fused to either pVIII or pIII (11). The resulting fusion
protein is displayed on the outer surface, thus the name phage "display". A
phage-display library is a mixture of such phage clones, each with a different
foreign DNA insert and different peptide on the surface. Each phage can infect
bacteria and produce progeny encoding for the engineered fusion protein. The
fusion proteins on the phage surface exposed to solvent can behave
independently from the virus. Affinity-based selection can be applied to the
phage library according to the ability of fusion proteins to bind the desired
ligand targets, with the targets usually immobilized to a solid surface (Figure 4.2)
(11). Phage peptides with high affinity towards a ligand receptor are captured
on the solid surfaces, and can be amplified in bacteria to produce a large crop of
progeny phage, which can be subjected to another round of selection. Enormous
libraries with millions or even billions of different peptides can be processed
simultaneously, and peptide phenotype and genotype are effectively linked.
Phage display technology has found many applications. It is extensively
used for antibody mapping and receptor mimicking. DNA-binding proteins
with novel sequence specificities can be selected through phage display. Phage
display has been very successful in designing zinc finger proteins with high
132
DNA binding affinity and sequence specificity (13-16). It has also been used to
probe enzymatic catalysis and protein-protein interactions (17).
In addition,
novel DNA binding peptide motifs were identified from random phage libraries
(18, 19). In summary, phage display has been extensively used as a tool for
protein engineering.
Since HMG-domain proteins are not designed by nature to bind to
platinated-DNA, we propose that novel HMG-domain proteins with high
affinities towards cisplatin-DNA adducts can be identified through phage
display. HMG1 domain A has a good affinity (nM) for cisplatin-modified DNA
sequences (20).
The crystal structure of domain A bound to a 16-bp site-
specifically platinated DNA duplex offers structural information about proteinDNA interactions (21). The loop region between helix II and helix III, which also
include the intercalator Phe 37, is in close proximity to the platination site and
can be important in binding to the platinated DNA. Changes in the loop (amino
acids 32 to 37) may allow more favorable interactions with cisplatin-modified
DNA. To investigate this hypothesis, we designed a phage display library of
HMG1 domain A mutants with amino acids 32-37 randomized.
Here, the randomized HMG1 domain A peptide is fused to the coat
protein pIII of the filamentous phage M13 and displayed on the phage surface.
In our non-lytic phage system, the domain A phagemid cannot produce progeny
after transfection into the bacterium. It requires the "helper phage" VCS-M13 to
"super-infect" the bacterium.
The helper phage can facilitate single strand
replication from the double stranded phagemids, and produce other necessary
coat proteins. New phage progeny have coat proteins from both the helper
phage and the domain A phagemid, but it has the genome of domain A
133
phagemid because the helper phage genome is genetically engineered not to be
packaged. Cisplatin-modified DNA targets are immobilized in 96-well plates
using the biotin-streptavidin linking method. Phage particles that have domain
A fusion proteins with high affinity for the platinated DNA target are selected
from the phage library through biopanning (Figure 4.2). If novel proteins with
unusually high affinity for the platinated DNA are identified, they will be
introduced into cells to examine their effects on cisplatin cytotoxicity.
A
polypeptide that enhances the cytotoxic effect of cisplatin in cells may have
clinical applications through gene therapy or the use of peptidomimetic
compounds.
Materials and Methods
Materials
cis-DDP was a gift from Johnson-Matthey.
The pHMG1 and pHMG1
domA plasmids were obtained from Dr. Marco Bianchi. The pZifl2 phagemid
was obtained from C. Pabo and coworkers. The HMG1 domain B antibody was
affinity purified with HMG1 domB protein by Deborah Zamble. The ARI 229
bacterial strain was donated by the Affymax Research Institute. The QuickChange site-directed mutagenesis kit and VCS-M13 helper phage were
purchased from Strategene. T4 polynucleotide kinase, Taq polymerase and all
restriction enzymes were obtained from New England Biolabs. Streptavidincoated 96-well plates were purchased from Pierce. BiotinTEG CPG 500 columns
were obtained from Glen Research. Human placental DNA was purchased from
Sigma. Streptavidin-coated iron beads (Dyna beads M288) and a magnetic
particle concentrator were obtained from Dynal.
134
Methods
Platinum atomic absorption spectroscopy was performed on a Varian AA
1475 instrument equipped with a GTA 95 graphite furnace. Oligonucleotides
were synthesized by the phosphoramidite method on an Applied Biosystems 392
DNA synthesizer. DNA was purified by either HPLC or denaturing
polyacrylamide gel electrophoresis. UV/visible spectrophotometry was carried
out on a Varian instrument (Cary). DNA sequencing was performed by the MIT
Biopolymers Lab on an automated ABI Prism system.
Cloning of the pHDB Phagemid. The phagemid pZifl2 is a plasmid
encoding a zinc finger protein fused to M13 coat protein III (Figure 4.3 A). The
zinc finger protein was excised by digestion with XhoI and XbaI, and the
digested phagemid fragment was purified on an agarose gel. The HMG-1
domain B sequence, from K85 to K163, was PCR amplified with Taq polymerase
from the pHMG1 with primers containing XhoI and XbaI sites.
PCR top primer with XhoI site underlined:
5'- ACG GAG CCT CGA GTC AAA AAG AAG TTC AAG GAC -3'
PCR bottom primer with XbaI site underlined:
5'- AG GCA AAG TCT AGA TIT AGC TCT GTA GGC AGC AAT ATC -3'
The amplified domain B sequence was digested with XhoI and XbaI, gelpurified and ligated into the digested phagemid with T4 DNA ligase. The
ligation product was transformed into competent XL1-Blue cells and grown on
LB ampicillin plates. The partial map of the cloning product, HMG1 domain B
phagemid (pHDB1), is shown in Figure 4.3 B. Selected colonies were grown to
saturation in LB media. Plasmid DNA was isolated by using the boiling lysis
miniprep method (22). The plasmid DNA was digested with StuI and XbaI, and
135
the digested plasmid with the domain B sequence had the anticipated 200-bp
fragment on agarose gels. The plasmid was sequenced by the MIT Biopolymers
lab.
Site-DirectedMutagenesis to Mutate the Amber Stop Codon TAG in pHDB. To
mutate the amber stop codon TAG to glutamine, the following primers were
synthesized:
5' - AGAGCTAAATCTAGAGACCAGAAAAAGGCCGACAAGTCC -3'
3 '- TCTCGATTTAGATCTCTGGTCTTTTTCCGGCTCTTCAGG -5 '
where the point mutations are in bold. Standard mutagenic procedures were
followed as described above. The new phagemid, pHDB2, is shown in Figure
4.3C.
VCS-M 13 Helper Phage Preparation. VCS-M13 helper phage was streaked
onto an LB plate with a sterile loop. Each plate was then covered with 3 mL of
liquid top LB agar and 0.5 ml of freshly-grown XL1-Blue cells. After incubation
at 37 'C over-night, plaques appeared as clear spots on the lawn of bacteria as a
result of the slower growth of phage infected bacteria compared to uninfected
ones. A single plaque was picked and grown in 50 ml of LB kanamycin (70
pg/ml) overnight. The cells were pelleted twice by centrifugation and discarded.
The VCS-M13 phage in the supernatant can be stored at 4 oC for 2-3 months.
Titering the Phage. To estimate the phage concentration, a titering
procedure described by Rebar (14) was followed. Phages infect strains of E. coli
that display a thread-like appendage called F' pilus (11). Male XL1-blue cells
have the F' pilus. A fresh culture of XL1-Blue cells that were grown overnight at
37 'C in LB containing 12.5 pg/ml of tetracycline was placed into the wells of 96
well plate (50 pl/well). Serial dilutions of phage samples were prepared in 96-
136
well plates.
A 10 pl aliquot of each phage dilution was added to a well
containing cells; the mixture was incubated at room temperature for 10 min.
After 190 pl of 2x YT was added to each well, the plate was incubated at 37 'C for
30 min.
A 5 pl sample of each mix was then spotted onto LB agar plates
containing the appropriate antibiotics. The plates were incubated at 37 'C until
visible colonies formed.
pHDB Phage Preparation. A single colony containing the phagemid pHDB
was grown in 25 mL of 2xYT containing 100 pg/ml ampicillin and 12.5 pg/ml
tetracycline for XL1-Blue cells or 70 pg/ml kanamycin for ARI cells. The culture
was grown at 37 'C and agitated on a rotary shaker at 300 rpm. After the OD
reading of the culture at 600 nm reached 0.3, the culture was titered and
subsequently superinfected with VCS-M13 helper phage (1*1012 kanamycin
transducing unit, or KTU) and grown at 37 'C, 125 rpm for 1.5 h. The growth
was titered again to make sure that over 90% of the cells were superinfected with
the helper phage, and 10 ml of the culture was added to 400 ml of 2xYT
containing ampicillin and kanamycin. After 24 h of growth at room temperature
with no agitation, the turbid culture reached saturation. The cells were pelleted
twice at 7.5 krpm for 15 min and 300 ml of supernatant were mixed with 70 ml of
5xPEG/NaCl to precipitate the phage. After 1.5 h of incubation on ice, the phage
was pelleted at 9 krpm and then resuspended in 32 ml of 2xYT and 8 ml
5xPEG/NaCl, and incubated on ice for 0.5 h. The final spin at 12 krpm yielded a
dark phage pellet. The pellet was dissolved in 0.5 ml of binding buffer (10 mM
Hepes-NaOH, pH 7.4, 10 mM MgCl2, 50 mM KC1, 1mM EDTA, 4% (v/v)
glycerol, 0.05% (v/v) Nonidet P-40). The phage was stored at -80 oC.
137
Electron Microscopy of the Phage Particles. The negative stain method was
employed to visualize the phage particles. Copper grids (300 mesh) were freshly
coated with carbon film. A phage sample (1010 particles/ml) was applied to the
grid for 10-30 sec, and then washed with 40 Fl of H 20. Uranyl acetate (20 pl of 23%, pH 4.5) was dropped onto the grid surface, and washed with water after 30
sec. The grid was blotted with filter paper briefly. The phage were visualized
under a JEOL 1200 CX electron microscope, at 80 KV, with 60,000x magnification.
Gel Mobility Shift Assays of pHDB Phage with DNA Probes. Two probes were
used for the gel mobility shift assays. 123-bp DNA was obtained by digestion of
the 123-bp DNA ladder with Ava I followed by polyacrylamide gel purification
(23). The 123-bp DNA was globally platinated at an rb of 0.02 (Pt/nucleotide). A
20-bp oligonucleotide with the following sequence was synthesized:
5'-TCTCCTTCAG*G*TCTCTTCTC-3'
3' -AGAGGAAGTC C AGAGAAGAG-5'
where as the asterisks indicate the platination sites.
The HPLC purified DNA was labeled with [y-32P]-ATP and T4 DNA
kinase. Reaction mixtures (10 pl total volume) contained 20,000 cpm of probe
([DNA] = 10 nM), 10 mM Hepes-NaOH, pH 7.4, 0.2 mg/ml BSA, 10 mM MgCl 2,
50 mM KC1, 1 mM EDTA, 4% (v/v) glycerol and 0.05% (v/v) Nonidet P-40.
Competitor chicken erythrocyte DNA was added to the mixture to reduce
nonspecific DNA-protein binding. The number of phage particles used was
estimated from the titering results. Reaction mixtures were incubated at room
temperature for 70 min. To each reaction mixture was added 0.5 pl of gel loading
buffer (30% (v/v) glycerol, 0.25% (w/v) bromophenol blue and 0.25% (w/v)
138
xylene cyanol) and the final mixture was loaded on a pre-run, pre-cooled (4 'C)
10% native polyacrylamide gel (29:1 acrylamide:bis). The gel was run in 0.5X
TBE buffer at 300 V for 3 h and vacuum dried onto Whatman 3 MM
chromatography paper.
The gel was quantitated on a Molecular Dynamics
Phosphorimager using ImageQuant software.
pHDB Phage Panning. Several oligonucleotide sequences were employed
as DNA targets in the panning process.
20-mer with biotin at the 3' end:
5'- TCT CCT TCA G*G*T CTC TTC TC biotin -3'
3'- AGA GGA AGT C C A GAG AAG AG -5'
This particular sequence, having A and T flanking the GG platination site, has
the most favorable interaction with HMG domB protein (Kd = 48 nM) compared
to other sequences with different flanking bases (20).
30-mer with multiple platination sites with biotin at the 3' end:
5'- CAGGGTCGGAACATGGCCCCTTCCACGGTA Biotin -3'
3 ' - GTCCCAGCCTTGTACCGGGGAAGGTGCCAT -5'
pHDB phage (~1010 ATU) was mixed with varying amounts of either
unmodified or cisplatin modified DNA (1 pmole to 1 nmole) in 20 pl of binding
buffer (10 mM Hepes, pH 7.4, 0.2 mg/ml BSA, 10 mM MgCl2, 50 mM KC1, 1mM
EDTA, 4% (v/v) glycerol, 0.05% (v/v) Nonidet P-40).
After 1 h at room
temperature, the mixture was applied to wells of streptavidin-coated plates
containing 10 pl of binding buffer. The plate was agitated at 200 rpm for 1 h.
The samples were removed from the wells and each well was washed three times
with 50 pl of the binding buffer. During each wash the plate was agitated for 5
139
min at 200 rpm. Elution buffer, 4M NaCl in binding buffer, was added to each
well (50 pl). After 1 h, the eluted material was removed and titered.
Cloning of the pHDA Phagemid. The HMG-1 domain A sequence, from M1
to F89, was PCR amplified from the HMG-1 expression plasmid with Taq
polymerase and primers containing XhoI and XbaI sites.
PCR top primer with XhoI site underlined:
5'-CTTTAAGAAGGACCTCGAGTCATGGGCAAAGGAGATCCTAAG-3'
PCR bottom primer with XbaI site underlined:
5'-GGGGGCATTGGGTCTAGAGAACTTCTTITTGGTCTCCTCCCC-3'
The amplified domain A sequence was digested with XhoI and XbaI, gel purified
and ligated into the digested pZif phagemid with T4 DNA ligase (Figure 4.3 C).
The ligation product was transformed into XL1-blue cells, which were grown on
LB ampicillin plates. Selected colonies were picked and grown to saturation in
LB media. Plasmid DNA was isolated by using the boiling lysis miniprep
method (Sambrook et al., 1989). The plasmid DNA having the correct insert from
restriction digestion mapping, pHDA, was confirmed by sequencing.
DNA Targets for pHDA Phage Panning. Several oligonucleotide sequences
were used as the DNA targets for the panning process.
One of the DNA sequences was a 20-bp DNA with a single platination site
-AG*G*T-:
5' -TCTCCTTCAG*G*TCTCTTCTC biotin-3'
3'
-AGAGGAAGTC C AGAGAAGAG -5'
Another DNA target was a 30-bp DNA with a single -TG*G*A- platination site:
5' -
TCTCTCTCTCTCTCTTG*G*ACTCTCTCTCT -3'
3 '-Biotin AGAGAGAGAGAGAGAAC C TGAGAGAGAGA -5'
140
Yet another DNA sequence was a 30-bp DNA with multiple platination sites:
5' -CAGGGTCGGAACATGGCCCCTTCCACGGTA biotin-3 '
3 ' -GTCCCAGCCTTGTACCGGGGAAGGTGCCAT -5'
Phage Panning with pHDA. pHDA phage (=10" ATU) were mixed with
varying amounts of either unmodified or cisplatin-modified DNA (50 nM to 250
nM) in 20 pl of binding buffer (10 mM Hepes-NaOH, pH 7.4, 0.2 mg/ml BSA, 10
mM MgC12, 50 mM KC1, 1 mM EDTA, 4% (v/v) glycerol, 0.05% (v/v)Nonidet P40). After 1 h of binding at room temperature, the mixture was applied to wells
of streptavidin-coated plates containing 10 pl of binding buffer. The plate was
agitated at 600 rpm for 1 hr. The samples were removed from the wells and each
well was washed 5 times with 190 pl of the above binding buffer containing 100
mM NaC1. During each wash the plate was agitated for 5 min at 600 rpm and
then the supernatant was removed thoroughly. The elution buffer, 4 M NaCl in
binding buffer, was added to each well (30 pl). After 1 h, the eluted phage was
removed and titered.
Streptavidin-coated iron beads also can be used for biopanning. A 50 pl
portion of the beads were resuspended in 200 pl of binding buffer as described
above and a magnetic particle concentrator was used to pull the beads down in
the Eppendorf tube. The binding reaction described earlier was mixed with the
washed beads and more binding buffer was added to make the total volume 200
pl. After 1 h, the beads were pulled down with the magnet, and the supernatant
was removed. The beads were washed five times with 500 pl of 100 mM NaCl in
the binding buffer. To elute the bound phage, 50 pl of 4 M NaCl in binding
buffer was used.
141
Multivalent and Monovalent Phage. There are five copies of pIII coat protein
for every phage particle. If there is only one copy of HMG-domain-pIII fusion
protein per phage, then the phage is monovalent; if there are two or more copies
of the fusion protein per phage, then the phage is multivalent. Panning of
monovalent phage is based on the affinity of a single fusion protein with the
target, whereas panning of multivalent phage may be affected by how many
copies of fusion protein the phage has. Therefore, the monovalent phage system
is more desirable for selecting protein with high affinity for target.
To produce monovalent phage, an amber stop codon TAG was designed
in between domain A and pIII gene (Figure 4.3 C). XL1-blue cells are amber
suppressant, which means that 86% of the time, TAG is recognized as the stop
codon and the fusion coat protein is not made; but 15% of the time, a tRNA will
translate TAG as Gln and the fusion coat protein is made. Helper phage VCSM13 produces endogenous pIII coat proteins which, along with the domA-pIII
fusion protein, will be packaged in the new phage particles. Statistically, this
procedure ensures that there is no more than one copy of fusion protein per
phage.
To produce multivalent phage, the amber stop codon was removed in
between domain A and pHI gene (Figure 4.3 D). The phagemid only expresses
domA-pIm fusion protein. Together with the pHI protein provided by the VCSM13 helper phage, newly packaged phage have two or more copies of domA-pmI
fusion proteins on the surface.
Site-DirectedMutagenesis to IncorporateTwo BbsI Sites in pHDA2. Plasmid
digestion with non-palindromic restriction enzymes generates non-compatible
sticky ends that can decrease the inserts self-ligation probability in cloning.
142
Thus, it is desirable to design such enzyme sites for library DNA cloning. The
following primers will delete the original BbsI site in pHDA2 by using a silent
mutation:
5' -AAGTGCTCAGAGAGGTGGAAAACCATGTCTGCTAAAGAA -3'
3 ' -TTCTTTAGCAGACATGGTTTTCCACCTCTCTGAGCACTT -3'
The following primers will incorporate a BbsI site N-terminus to amino acid 28:
5 ' -TTCTTTGTGCAAACCTGCCGGTCTTCGCACAAGAAGAAGCACCCGGAT -3'
3 ' -AAGAAACACGTTTGGACGGCCAGAAGCGTGTTCTTCTTCGAGGGCCTA -5'
The following primers will incorporate a second BbsI site in the C-terminus
region relative to amino acid 37, and also incorporate an one base pair deletion,
thus forming a frame-shift in the domA-pIII fusion gene:
5 ' -GTCAACTTCTCAGAGTTCT GAAGACAGTGCTCAGAGAGGTGGAAG -3'
3 ' -CAGTTGAAGAGTCTCAAGA CTTCTGTCACGAGTCTCTCCACCTCT -5'
If the library DNA fails to be inserted in the vector, the self-ligated phagemid can
not express pIII coat proteins in frame. The phage produced by the self-ligated
vector do not have any fusion proteins on the surface, and are thus unable to
bind to DNA target and cannot be selected in the panning process. Only when
the library DNA is correctly inserted can the phagemid restore the frame and
produce phage that have domA-pIII fusion proteins on the surface to bind DNA.
Constructing the Phagemid Library. Library DNA with amino acid 32-37
randomized and primers were synthesized and purified:
5' -CTGCCGGGAGGAGCACAAGAAGAAGCACCCG (NNB) 6TCAGAGTTCTCCAAGAAG-3'
3'-
GCCCTCCTCGTGTTCTTCTTCGTGGGC
AGTCTCAAGAGGTTCTTCACGA-5'
The purified library DNA containing the randomized 6 amino acids and the
DNA primers were mixed at a range of mole ratios of library : primer 1 : primer 2
143
of 1 : 2: 2 to 1 : 50 : 50. The library mixture was treated with T4 DNA kinase and
ATP and then ligated into the modified pHDA2 phagemid. The ligation product
was precipitated by adding 1/10 volume of NaOAc and 2 volumes of cold
ethanol. The ligated DNA was pelleted, air-dried, and dissolved in 5 p• of dd
H 20. A 40 pl1portion of ARI 229 cells was added to each ligation mixture and
electroporation was performed.
A total of 30 pg of phagemid DNA was
transformed into ARI 229 cells to yield a total of 108 transformants.
After a short growth period in 10 ml SOB (20 g/liter trypton, 5 g/liter
yeast extract, 0.5 g/liter NaC1) supplemented with 1% glucose for 1 h at 37 'C
with aeration, the transformants from each electroporation were pooled into five
samples (40 ml each). Each sample was diluted into 340 ml of 2X YTG/AK and
amplified (2 hr, 37 'C with aeration). To ensure continued repression of the lac
promoter of the fusion gene, glucose [10 ml of 40% (w/v)] was added after 1 h
and again after 2 h. Cells from each sample were pelleted (15 min, 6000 g, 4 °C),
resuspended in 10 ml of 2x YTG containing 15% glycerol (v/v), frozen in liquid
nitrogen and stored at -80 oC.
To produce the phage library, 5 ml of each sample were thawed and
diluted into 400 ml of 2x YTG/AK. The resulting cultures were grown for 1 h (37
oC, with aeration) and then more glucose [10 ml of 40% (w/v)] was added to
each. Cells were pelleted (15 min, 6000 g, 4 °C), resuspended in 40 ml of 2x
YT/AK, and incubated with 101 kanamycin-transducing units (KTU) of VCSM13 helper phage (2.5 h, room temperature, no agitation). Each of the five
cultures was diluted into 2x YT/AK (360 ml) and grown to saturation (room
temperature, 36 hr, no agitation).
144
Phage were harvested from the large volume of culture supernatant by
using two PEG precipitations.
The cells were first pelleted by two
centrifugations (6000 g, 4 'C, 15 min), and 345 ml of the final supernatant was
mixed with 87 ml of PEG solution in fresh centrifuge tubes. After incubation on
ice for 1.5 h, the phage were precipitated by centrifugation at 13,700 g, 4 'C for 20
min. Phage pellets were resuspended in 32 ml of 2x YT and 8 ml of PEG, then
incubated on ice for 30 min before the final centrifugation at 26,900 g, 4 "C for 20
min. The supernatant was thoroughly removed and the phage pellets were
resuspended in 0.5 ml of binding buffer. Aliquots of phage were stored at -80
OC.
Cloningof HMG1 DomA into pcDNA 3.1. pcDNA3.1 is an overexpression
vector from Invitrogen with a CMV promoter. The HMG-1 domain A sequence,
from M1 to F89, was PCR amplified with Taq polymerase from HMG1domA
expression plasmid with primers containing XhoI and EcoRI sites.
PCR top primer with XhoI site underlined:
5' - TTAGACTCGAGTATACATATGGCAAAGGAGAT -3'
PCR bottom primer with EcoRI site underlined:
5' - GGGGGCAGAATTCTCTTAGAACTTCTTITTGGTCTCCCCTTT -3'
The amplified domain A sequence was digested with XhoI and EcoRI, gelpurified and ligated into the digested pcDNA 3.1(-) plasmid with T4 DNA ligase.
The ligation product was transformed into competent XL1-Blue cells and grown
on LB ampicillin plates.
Plasmid with the correct domain A insert was
sequenced.
Cloning of HMG1 DomB into pcDNA6/V5His.
pcDNA6/V5His
is
a
mammalian cell overexpression vector from Invitrogen with a CMV promoter.
145
The HMG-1 domain B sequence, from K85 to K163, was PCR amplified with Pfu
polymerase from HMG1 domB expression plasmid with primers containing
EcoRI and XhoI sites.
PCR top primer with EcoRI site underlined:
5'-ATCTGGAATTCTGGGAAGATGAAAAAGAAGTTCAAGGACCCC -3'
PCR bottom primer with XhoI site underlined:
5'- CGCTCTCGAGGGTTTTTIIATTTAGCTCTGTAGGCAGCAATATC -3'
Plasmid with the correct domain B insert was sequenced.
Overexpression of HMG1 Domains in HeLa Cells. SuperFect protocol from
Qiagen was followed to transfect HeLa cells with HMG1 domains in pcDNA
overexpression vectors. For transient transfection, cells were harvested after 24
and 48 h of transfection. For stable transfection, cell colonies were picked and
grown into individual plates before harvesting.
Results and Discussion
pHDB Phage. HMG1 domain B was initially designed as the basis for
phage display prior to the knowledge that domain A has higher affinity for
platinated-DNA than domain A.
The crystal structure of the domain A
complexed with cisplatin-modified DNA was also not available at the time. The
HMG-1 domain B gene was successfully cloned into the phagemid to form
pHDB (HMG-1 domain B phagemid) (Figure 4.3 B). Results from the MIT
Biopolymers Lab confirmed the correct sequence. Electron microscopy of the
phage particles confirmed the rod-like appearance of the phage (Figure 4.4).
Gel Mobility Shift Assays of pHDB Phage with DNA. To test whether the
HMG domB-pIII fusion protein on the pHDB1 phage surface is able to bind
146
platinated DNA sequences, gel mobility shift assays were performed with phage
and different DNA sequences (Figure 4.5). Because of the large size of the phage,
a band of very slow mobility would indicate the binding of intact phage to the
DNA. Only a band corresponding to the HMG domB-DNA complex appeared
in the gel; however, indicating the existence of free HMG domB protein in the
phage preparation. The occurrence of free protein is normal. The absence of the
expected slow mobility bands presumably was caused by the low affinity of the
phage for the platinated DNA.
Panning of the pHDB Phage. The process of pHDB phage panning with
different DNA targets immobilized on the surface of microtiter plates resulted in
mostly background level retention (0.01%) and no difference in the retention
percentage between the platinated and unplatinated DNA targets (Table 1). This
results may be due to the low affinity of the phage for the DNA targets, and is in
agreement with the gel mobility shift assays of these phage with DNA.
Cloning of the pHDA Phagemid. The unsuccessful attempts to use pHDB
phage as a control in the panning process indicated that a higher affinity between
phage and platinated DNA targets is necessary. HMG1 domA has a higher
affinity than HMG domB towards cisplatin-modified DNA (20).
The Kd for
HMG1 domA with a 15mer AG*G*A platinated duplex is 1.6 nM, two orders of
magnitude smaller than the dissociation constant of HMG1 domB for cisplatinmodified DNA (20). The possibility of using pHDA phage in the panning
process to search for novel HMG1 domA proteins was therefore investigated.
Multivalency and Monovalency. There are five copies of the pIII coat
proteins on every phage surface. pHDA phagemid encoding the domA-pIII
fusion protein lacks the complete phage genome, and thus is unable to produce
147
phage without superinfection with VCS-M13 helper phage. After superinfection,
both the pHDA phagemid and the helper phage express coat proteins that will
be packaged into the new phage particles. The phagemid gene is preferentially
packaged into new phage due to the defect engineered in the helper phage gene.
In pHDA1 phagemid, the HMG1 domain A gene was fused to the N-terminus of
the coat protein gene III with an amber stop codon between the two genes
(Figure 4.3 C). The amber codon TAG in pHDA1 permits the expression of the
HMG1 domain A polypeptide either as a fusion protein or as an independent
protein, depending on the bacterial strain. In amber suppresser strains of
bacteria, e.g., XL1-blue cells, TAG is suppressed 14% of the time to be expressed
as glutamine. Statistically, the phage particles produced have no more than one
copy of domA-pIII fusion protein on the surface, therefore affording a
monovalent phage system. This system has low affinity towards DNA target
because only one copy of fusion protein is available for binding.
When the amber stop TAG is mutated to TAC encoding for glutamine
(Figure 4.3 D), the pHDA2 phagemid only expresses the domA-pIII fusion
protein. Together with the pIII coat protein provided by the VCS-M13 helper
phage, each new phage particle should have two or more copies of domA-pIII
fusion proteins on the surface, as in the multivalent phage system. Multiple
DNA-binding fusion proteins enable the phage to have a higher affinity towards
DNA targets. However, the selection may not depend on the "real" affinity of
each individual domA-pmII protein to the DNA targets, but on how many copies
of the fusion protein the phage has. Ideally, a monovalent phage system with a
high affinity for the DNA target is more desirable for the phage panning system.
148
Bacterial Strain. The pIII coat protein is critical for phage infection (11).
This coat protein is synthesized with an 18 amino acid N-terminal signal peptide
that is cleaved by signal peptidase during the insertion of mature pIII into the
inner membrane of the cell (24). HMG1 domA was fused to the N-terminus of
the pIII coat protein and positioned adjacent to the signal cleavage site.
Positively charged amino acids adjacent to the signal peptide affect the proper
insertion of the pHI into the inner membrane, and block the assembly of phage
particles (24). HMG1 domA is positively charged and has lysines in the first ten
amino acids that might inhibit pIII insertion, and consequently, phage
production and infection ability.
The ARI 229 bacterial strain has suppresser mutations in the prlA (secY)
component of the protein export apparatus, thus compensating for phage growth
defects caused by the positively charged amino acids (24).
This strain is
kanamycin resistant, but does not suppress amber stop codon, and thus cannot
produce monovalent phage. Multivalent pHDA2 phages produced by ARI 229
cells demonstrated specificity to platinated DNA targets and were used in this
study.
Size of the DNA Target.
The length and sequence of platinated DNA
targets used in phage panning are important for the selection. The structurespecific HMG domain A has different affinity towards different platinated DNA
sequences (20). DNA of different length, 20-mers or 30-mers, with either a single
platination site or multiple platination sites, were tested as targets for phage
panning.
These targets could bring sequence bias into the procedure.
Alternatively, commercially available human genomic DNA was fragmented
into 100-bp lengths. Since neither manual shearing with 25G,/s fine syringe
149
needle nor sonication broke up the long DNA, we employed triple enzymatic
digestion, using enzymes that recognize short DNA sequences, to fragment the
DNA to the desired length. The recognition sequence for Sau3AI is 5'-GATC-3',
for RsaI is 5'-GTAC -3', and for MseI is 5'- TTAA-3', all of these 4-bp sequences
are frequently encountered in the genome. As shown in Figure 4.6, the digested
genomic DNA was less than 500-bp. The digested genomic DNA was run out on
an agarose gel and extracted.
Fragmented genomic DNA contains millions of different sequences and
more closely reflects the platinated genome encountered in vivo than specific
synthetic DNA. However, when phage bind long pieces of DNA (>50 bp), nonspecific binding of the domain A protein to unplatinated segments of the DNA
will dominate the specific binding to the platinated region (Figure 4.7 A). The
panning selection then may be biased by non-specific binding (P. Sharp and
coworkers, unpublished results). Thus, short oligonucleotide targets (no more
than 30 bp) were used for panning (Figure 4.7 B).
When a specific DNA sequence is used as the panning target, the selection
may be biased by the sequence context. If several sequences were mixed
together as DNA targets, the panning would still be biased, because each
selected phage may prefer one specific DNA among the mixture. If a different
DNA sequence. were used for each round of panning selection, the selection
conditions would be very harsh. In a certain panning step, a particular sequence
for which the domain A fusion proteins have very low affinity may offset the
previous selections and result in loss of the library. After comparing these
scenarios, it was decided that a specific DNA sequence should be used for
panning in all rounds.
150
pHDA2 Control Phage Panning. pHDA2 phage having domA-pmI fusion
proteins on the phage surface were used as a control for the panning conditions.
Binding of the phage to the DNA targets may be influenced by several variables
including the salt concentration in the binding buffer, the DNA target
concentration, the amount of non-specific DNA competitor (chicken erythrocyte
DNA) and the amount of specific competitor. The specific DNA competitor has
the same DNA sequence as the DNA target, except that it is unplatinated and
unbiotinylated. Specific competitor is intended to minimize sequence bias and
ensure that the selection is based on structure recognition. Furthermore, in the
washing step, the optimal salt concentration of wash buffer, the time of each
wash, and the total number of washes all need to be determined.
After testing many combinations of the above variables, the optimal
panning conditions for the system was determined. In the binding reaction, the
DNA target concentration was 250 nM, the non-specific CE DNA competitor was
0.005 pg/pl, and the specific competitor was 1.25 pM. The binding buffer
contained 50 mM KC1. After binding to the streptavidin-coated plate wells, five
washes each of 190 p~ wash buffer with 100 mM of NaC1 were performed. Under
such conditions, the selectivity of pHDA2 phage binding to platinated DNA
targets over unplatinated DNA was 10-20 fold (Table 4.2) (Figure 4.8).
Streptavidin-coated iron beads also were tested for biopanning. The
larger surface area of the beads compared to the flat surface of the wells may
facilitate more efficient DNA binding and the subsequent washing. Titering
results from streptavidin-coated iron beads were comparable with those from
streptavidin-coated plates, with about 10 to 20-fold selectivity for platinatedDNA targets over unplatinated-DNA.
151
pHDA Phage Library. A phagemid library with amino acids 32-37
randomized was constructed as described in the methods section (Figure 4.9).
The optimal ligation condition for constructing the library was phagemid :
library DNA : primer 1 : primer 2 = 1 : 10 : 250: 250. A total of 108 transformants
were obtained when 30 pg of phagemid were electroporated into ARI 229 cells.
Several colonies were picked and grown in 2 ml cultures followed by extraction
and sequencing of the phagemid DNA. Among the colonies picked, 5% were
self-ligated vectors without the library insert. The designed frame-shift within
the vector should prevent expression of any fusion coat proteins to bind to DNA
targets, and thus should not survive the panning selection. The remaining
phagemids contained randomized sequences (Table 4.3).
A total of 1010 ATU of phage were produced from this phagemid library as
described. The resulting phage library was subjected to panning with 30 mer
DNA target with a single site-specific platination site -TG*G*A- situated 20 bp
away from the 5' biotinylated end. This flanking sequence was chosen because it
forms a tight complex with domain A (20), and the 20-bp distance from the 5'
end to the platination site should allow domA to access the binding site.
Following each panning selection, the eluted, pooled phage were mixed with a
freshly saturated culture of ARI 229 cells (4.8 ml). The fusion gene is under the
control of lac promoter. If the lac promoter is not repressed before superinfection
by VCS-M13 helper phage, the expressed fusion protein may be excreted, bind to
the pilus of the bacteria, and block superinfection. The addition of glucose
represses the lac promoter of the fusion gene. Cultures were pelleted and
resuspended in 2 ml of 2x YTG containing 15% glycerol (v/v) and stored at -80
OC.
152
The infected cells were thawed, diluted into 40 ml of 2x YTG/AK, and
grown for 1 hr (37 'C, with aeration). Cells were pelleted and resuspended in 5
ml of 2x YT and incubated with VCS-M13 helper phage. At this stage, the
phagemid expresses the domA-pIII fusion protein and the helper phage
expresses the pIII protein and other coat proteins. The helper phage also initiates
replication of the single-stranded DNA from the phagemid encoding the fusion
gene to be packaged into the progeny phage. A 45 ml portion of 2x YT/AK was
added and the cultures were grown to saturation (room temperature, 38 h, no
agitation). Cell were cleared by two centrifugation steps and the phage were
pelleted following a PEG precipitation and stored at -80 oC.
Theoretically, each panning cycle should enrich the sample in phage
expressing domA peptide mutants with good affinity for platinated DNA
targets. Phage from the first cycle are enriched and subjected to a second cycle of
selection (Figure 4.2). The retention efficiency is the percentage of eluted phage
from the total input. Retention efficiency should increase with each round of
selection and eventually plateau (Figure 4.10 A).
However, the retention
efficiency observed did not increase steadily, but showed a random profile
(Figure 4.10 B).
To investigate this puzzling result, the original phage library produced
from the phagemid library was re-introduced into bacteria cells. Resulting
colonies were picked and grown in 2 ml cultures. Extracted phagemid DNA was
sequenced. Sequencing results showed these phagemids to be predominately
self-ligated vectors (>90%) lacking library DNA inserts. These phage should fail
to produce functional coat proteins because of the frameshift engineered in the
gene, but are able to survive and infect in the presence of coat proteins provided
153
by VCS-M13 helper phage. Over 95% of the initial phagemid library contained
library DNA inserts. When phage were produced from this phagemid library,
however, self-ligated vectors were selected over the phagemids with the insert.
Several possibilities may explain the observed results. First, growth
conditions may favor the background phage. Instead of growing the phage at
room temperature for 38 h (proteolytic activity is lower at room temperature),
some other published conditions, such as growing the phage at 37 'C with
aeration for 5 h or 12 h with aeration, were also tested. These conditions still
selected the background phage growth. Second, if glucose repression of lac
promoter of the fusion gene were insufficient, phagemids with library DNA
inserts may produce fusion proteins, thereby blocking superinfection at the
bacterial pilus and preventing subsequent phage production. Undesirable
background phagemid, however, fails to produce coat protein capable of
blocking superinfection, and phage are produced efficiently. XL1-Blue cells have
a lacI Q phenotype, a super-repressor for lac promoter, and glucose can repress
the production of the fusion protein very efficiently prior to superinfection. The
ARI 229 cells are variants of the K91 cell line, which has lacI repressor but not the
super-repressor.
When the glucose concentration was increased to 4%,
background phage still dominated, which excludes this hypothesis. Finally, The
HMG1 domain A protein may be toxic to the cells or affect phage production.
pHDA2 control phage were successfully produced in ARI 229 cells, so the
existence of domain A peptide does not completely inhibit the phage production
but may have slowed down the process.
When pHDA phagemids producing
fusion proteins are mixed with background phagemids unable to make any
fusion protein, the background phagemid may overtake the culture. pHDA
154
phagemids lack the complete phage genome and require helper phage to
produce progeny. There are other filamentous phage systems, such as T7 phage,
that retain the complete phage genome, and do not require the helper phage.
Without the helper phage, the phage must produce coat protein themselves;
however, self-ligated frame-shifted vectors are unable to provide pIII coat
proteins and fail to propagate. To use the "self-sufficient" phage system, domain
A phage must be cloned into a new phage vector and all the experiments and
panning conditions need to be re-designed.
Instead of changing the phage display system, however, an alternative
approach was attempted. Domain A and domain A mutants should have similar
growth rates because all retain the N-terminal positive charges that slow down
phage production. When the frame-shift is deleted and the reading frame
restored, the self-ligated phagemid pHDA2 expresses domain A-pIII fusion and
propagates at the same rate as the domA mutant phage. Since we seek domain A
mutant phage with ten-fold or higher affinity for the platinated DNA targets
than that of domain A, the background (less than 5%) of self-ligated pHDA2
phage should be acceptable in this system.
The frameshift in the pHDA2 vector was deleted and a new domain A
library without any frameshifts was made, yielding a total of 6 x 108 different
protein variants. The phage library produced 1.5 x 1010 ATU/pl titer. Four
rounds of panning with Pt-30mer TGGA were carried out.
However, the
retention efficiency over four rounds of selection remained between 0.001% and
0.01%.
Several phage from each round of selection was sequenced and the results
shown in Table 4.4.
After each round of panning, the number of phage
155
containing stop codons or frameshifts resulting in "nonsense" phage lacking
domA protein on the surface increased. The first possible explanation is that the
library does not harbor a novel peptide with higher affinity for platinated-DNA
than domain A; alternatively, the selection conditions may be too harsh. The
absence of domain A sequence after each round of selection argues against this
hypothesis. Another possibility is that domain A is toxic to the bacterial host,
and phage producing domain A grow poorly. "Nonsense" phage may grow
better, be more infective, package better or making cells less sick, than the domA
phage.
Natural selection for growth robustness benefits nonsense phage
production.
Overexpression of HMG1 DomA and DomB. Before putting a domain A
based library into another phage system, efforts must be undertaken to validate
the potential of novel HMG1 based proteins for gene therapy. First, the ability to
overexpress HMG-domains in mammalian cells was investigated. Second, the
correlation between protein platinated-DNA affinity and cisplatin sensitivity of
cell lines overexpressing such protein should be tested.
HMG1 domains A and B were cloned into pcDNA overexpression vectors.
SuperFect was used to transfect HMG1 domain A and B transiently into HeLa
cells and cells were harvested after 24 and 48 h. Western blots were performed
on cell extracts from the HeLa cells, but no signal corresponding to HMG1
domain overexpression was observed on the gel (Figure 4.11). Stable transfection
yielded the same results with no overexpression of HMG1 domains. Other
laboratories also have had similar troubles overexpressing HMG1 domains
(Bianchi, M. personal communication).
156
Conclusions and Future Directions
HMG1 Domain A Phage Bind to Cisplatin-Modified DNA Preferentially. It has
been successfully established that domain A protein can be expressed on
filamentous phage as a pIII fusion. The VCS-M13 phage display system is nonlytic, and thus the HMG1 domain A-pIII fusion proteins must be secreted
through the bacterial membrane to be assembled into phage (24).
Highly
positively charged proteins, such as domain A, cause secretion problems and
phage growth defects (24).
Many growth conditions, including different
bacterial strains and growth temperature, were tested for the ability to support
proper HMG1 domain A phage growth. ARI 229 cells were able to produce
pHDA phage. pHDA phage are used as a control to determine the optimal
panning condition for subsequent experiments. pHDA phage bind to platinated
DNA targets about 20-fold better than the unplatinated DNA, thereby
demonstrating specific selection towards platinum-DNA adducts.
IntrinsicProblems with HMG Domain Phage Libraries. During the library phage
production, nonsense phagemids that have stop codons or frameshifts
dominated those retaining the library inserts. The presence of domain A mutants
fused to the coat protein may inhibit fusion-protein secretion and phage
assembly. The nonsense phagemid vectors are unable to produce domain A-pIII
coat proteins, but instead obtain all coat proteins from helper phage. These
phage, lacking domain A mutants displayed on the phage surface, propagate
faster and win the growth competition.
The secretion of foreign proteins
determines the growth rates of phage. Such problem is specific for non-lytic
phage system such as M13.
157
Potentialof Alternative Phage Libraries. Alternative phage system could be used
to discover novel HMG domains, such as the T7 phage system (25), which has
rapid growth and is lytic. Phage burst the cell membrane upon maturation and
there is no selective advantage according to protein characteristics (26).
However, because HMG1 domains cannot be overexpressed in mammalian cell
lines, a new phage library should be based on proteins other than HMG1.
Short peptides from random peptide libraries can also recognize DNA
specifically (18), although structure-specific DNA-binding peptides have not
been reported to do so. Alternative phage libraries may be able to select new
proteins with high affinity for cisplatin-modified DNA. For the long term goal of
gene therapy, it remains to be established that (1) such novel protein can be
overexpressed in mammalian cells and (2) high affinity of a protein for
platinated-DNA correlates with better cisplatin sensitivity.
It is also possible that the inability of the current domain A library to be
selected is due to the low affinity between the protein and the DNA in the
panning process. Zinc fingers can be used to enhance the binding affinity
between protein and target. Protein of interest can be randomized and fused to
two zinc fingers, and the DNA target will consist two zinc finger binding site
adjacent to the platinum-binding site, although the exact spacing needs to be
optimized. This work has established the basis for future library selections for
cisplatin-modified DNA.
Acknowledgments:
I thank Professor Carl O. Pabo and Dr. Bryan Wang for the many helpful
discussions.
158
References:
1.
Brown, S. J., Kellett, P. J., and Lippard, S. J.(1993) Science 261, 603-605.
2.
McA'Nulty, M. M., and Lippard, S. J. (1996) Mutat. Res. 362, 75-86.
3.
Zamble, D. B., Mu, D., Reardon, J.T., Sancar, A., and Lippard, S. J. (1996)
Biochemistry 35, 10004-10013.
4.
He, Q., Liang, C., and Lippard, S. J. (2000) Proc. Nat. Acad. Sci. USA 97,
5768-5772.
5.
Treiber, D. K., Zhai, X., Jantzen, H.-M., and Essigman, J. M. (1994) Proc.
Nat. Acad. Sci. USA 91, 5672 - 5676.
6.
Smith, G. P. (1985) Science 228, 1315-1317.
7.
Clarkson, T., and Wells, J. A. (1994) Trends Biotechnol. 12, 173-184.
8.
Kay, B. K., Winter, J., and McCafferty, J. (1996) Phage display ofpeptides and
proteins,Academic Press, San Diego.
9.
Scott, J.K., and G.P.Smith. (1990) Science 249, 386-390.
10.
Wilson, D. R., and Finlay, B. B. (1998) Can. J. Microbiol. 44, 313-329.
11.
Smith, G. P., and Petrenko, V. A. (1997) Chem. Rev. 97, 391-410.
12.
Russel, M., Wirhlow, H., Sun, T.-P., and Webster, R. E. (1988) J. Bacteriol.
170, 5312-5316.
13.
Rebar, E. J., and Pabo, C. 0. (1994) Science 263, 671-673.
14.
Rebar, E. J., Greisman, H. A., and Pabo, C. 0. (1996) Methods Enzymol. 267,
129-149.
15.
Choo, Y., and Klug, A. (1994) Proc.Natl. Acad. Sci. USA 91, 11163-11167.
16.
Choo, Y., Sanchez-Garcia, I., and Klug, A. (1994) Nature 372, 642-645.
17.
O'Neil, K. T., and Hoess, R. H. (1995) Curr.Opin. Struct. Biol. 5, 443-449.
18.
Cheng, X., Kay, B. K., and Juliano, R. L. (1996) Gene 171, 1-8.
159
19.
Wang, B., Dickinson, L. A., Koivunen, E., Ruoslahti, E., and Kohwi-
Shigematsu, T. (1995) J. Biol. Chem. 270, 23239-23242.
20.
Dunham, S. U., and Lippard, S. J. (1997) Biochemistry 36, 11428-11436.
21.
Ohndorf, U. M., Rould, M. A., He, Q., Pabo, C. 0., and Lippard, S. J. (1999)
Nature 399, 708-712.
22.
Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A
Laboratory Manual,Cold Spring Harb. Lab, Cold Spring Harbor.
23.
Hartley, J. L., and Gregori, T. J. (1981) Gene 13, 347-353.
24.
Peters, E. A., Schatz, P. J., Johnson, S. S., and Dower, W. J. (1994) J.
Bacteriol. 176, 4296-4305.
25.
Rosenberg, A., and Mierendorf, R. (1996) Novations 6, 1-6.
26.
Sche, P. P., McKenzie, K. M., White, J. D., and Austin, D. J. (1999) Chem.
Biol. 6, 707-716.
160
Table 4.1. Panning table of pHDB control phage with different DNA targets.
DNA sequences are described in the Methods section. 30-bp DNA has multiple
platination sites.
Samples
DNA targets
CE DNA (pg)
% of phage
selected
control
1
2
3
4
5
6
none
unPt 30bp 0.1 pM
Pt 30bp 0.1 pM
unPt 30bp 1 pM
Pt 30bp 1 pM
unPt 30bp 10 pM
Pt 30bp 10lpM
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.05
0.03
0.09
0.09
0.04
0.01
0.04
Selectivity of
platinated/unplatinated
3
0.5
4
161
Table 4.2. Panning table of pHDA2 control phage with different DNA targets.
DNA sequences are described in the Methods section. 30-bp DNA has multiple
platination site. 20-bp DNA has a single -TG*G*T- platination site.
CE
DNA
(pg)
control none
0.1
1
unPt 30bp 250nM 0.1
2
0.1
Pt 30bp 250 nM
unPt 20bp 250 nM 0.1
3
0.1
4
Pt 20bp 250 nM
unPt 20bp 250 nM 0.1
5
0.1
6
Pt 20bp 250 nM
Sample DNA targets
s
Specific competitor % of
phage
(5 times the
selected
targets)
0.011
unPt 30bp 1.25 pM 0.015
unPt 30bp 1.25 pM 0.29
0.037
0
0.35
0
unPt 20bp 1.25 pM 0.017
0.33
Pt 20bp 1.25 pM
Selectivity of
platinated/unplatinated
20
9.3
18
162
Table 4.3: Representative Phagemid Sequences from the Domain A Library.
CGT GTG ATT CTG ACG GAG
Arg Val Ile Leu Thr Glu
AGG GCG CCT GGG GCG TGT
Arg Ala Pro Gly Ala Cys
ACG GCT TGG CTG TGG GTG
Thr Ala Trp Leu Trp Val
ACG TAG CTG GTG AAT GAG
stop codon
GGG TGT CAG TTT CGT CGG
Gly Cys Gin Phe Arg Arg
TAG GCT CCG TGG AAT GTG
stop codon
ATG TCG CCG TGG GGT GGG
Met Ser Pro Trp Gly Gly
ATT GCG GTG GAT ACG TAG
stop codon
GAG GAG GCG CCG TGG GGT
Glu Giu Ala Pro Trp Gly
GTG GTG CGG GTT GGG CAG
Val Val Arg Val Gly Gin
AGG TGG GCT CAG GGG CTG
Arg Trp Ala Gin Gly Leu
GGG CGT CAG AGT ATG AGG
Gly Arg Gin Ser Met Arg
Position
G
A
T
C
1
41%
21%
19%
19%
2
32%
22%
22%
24%
3
66%
33%
163
Table 4.4A. Representative Sequences of Phage Produced by the Domain A
Phagemid Library:
GGT AGT TAG ATT TGT GCT
stop codon
CCG CAG GGG CGG GAT TGG
Pro Gln Gly Arg Asp Trp
AGG GCG GAT CCG TCT GGG
Arg Ala Asp Pro Ser Gly
GAG GCT TGT GTT TTC GTG
Glu Ala Cys Val Phe Val
AGG AAG GGG GCT GGG GTT
Arg Lys Gly Ala Gly Val
CCG CAG GGG CGG GAT TGG
Pro Gln Gly Arg Asp Trp
CGG CAT ACG CTG TAG GCG
stop codon
GTG ACG GCG ACT AGA TTC
Val Thr Ala Thr Arg Phe
ATG ATT GCG GGT AGT ATG
Met Ile Ala Gly Ser Met
Table 4.4B. Representative Sequences of Phage after First Panning
CAG CGG GGG AAG TAT GAG
Gin Arg Gly Lys Tyr Glu
TCA CTG CTA AGG GGG GG
frameshift
GAT GCT TCT GTC AAC TTC
Asp Ala Ser Val Asn Phe
TAG CAG AGT -GAT TGG GGG
stop codon
CGG ATC CTG CGT TGA GT
frameshift
GTG AGG ACG GGG GCG CGT
Val Arg Thr Gly Ala Arg
AGG GCG CAT ATT GAG TTG
Arg Ala His Ile Glu Leu
164
GAG ACG GAG GAG AAG CGG
Glu Thr Glu Glu Lys Arg
TCT GCT AAG CAG CGT TGG
Ser Ala Lys Gln Arg Trp
Table 4.4C. Representative Sequences of Phage after Fourth Panning
TAG AGT TCT CCA AGA AGT GCCT
frameshift
GAT GGG CGG GTG TAG TCT
stop codon
AGG AAG CAT CCT TGG GCT
Arg Lys His Pro Trp Ala
AGG ATG AGT GGC TTA TG
frameshift
GCG TAG CCT TGG TAT CGG
Stop codon
TTT TGG AGT GGT GAG ATG
Phe Trp Ser Gly Glu Met
TTG TAG CAT CAG GAG TCG
stop codon
GGG GGT TTT CGA GTG GT
frameshift
CAT TTG CCT GCG AAG GGT
His Leu Pro Ala Lys Gly
CAT TTG CCT GCG AAG GGT
His Leu Pro Ala Lys Gly
GTG CAG TAT CCT CAG CAT
frameshift
TCG GGG ACT TTG TAG CAT
stop codon
165
pVII, pIX:
4 or 5 copies each
Single-stranded
genome
pVIII:
2700 copies
(number varies
with genome length)
pVI:
4 or 5 copies
pIII:
4 or 5 copies
Figure 4.1. Cartoon illustrating filamentous phage structure. A single-stranded circular genome
is surrounded by about 2700 copies of the major coat protein pVIII and 4 or 5 copies of each of
four species of minor coat proteins, including, pIII, which binds to a host cell F-pilus. This figure
is adapted from Wilson, D.R. and Finlay, B.B., Can. J. Microbiol. 1998 44 (313-329).
166
_____r
Bind
S0@
M
4
Wash and Select
L
A
--
1f
r
kAmpiry ana Repeat
H3 N
4
=target ligand e.g. H3 N
Streptavidin
coated petri
dish
Figure 4.2. The phage display protocol to select for proteins with high affinity for
target ligands. For example, the target ligand can be cisplatin-modified DNA.
167
lacZpromoter
NcoI XhoI
I
XbaI
TAG - M13 coat protein
amber gene III
codon
lacZpromoter
NcoI XhoI
XbaI
codon
lacZvromoter
NcoI XhoI
XbaI
codon
lacZtoromoter
NcoI XhoI
XbaI
Figure 4.3. Partial map of the plasmid used for HMG domain B and domain A library.
A. Map of the phagemid pZifl2. pelB encodes the signal peptide necessary for
membrane export. The hybrid gene of zinc fingers and coat protein III is under control
of the lac promoter. B. Map of pHDB that encodes the fusion protein of HMG domB and
the coat protein III. C. Map of pHDA that encodes the fusion protein of HMG domA and
the coat protein III. Note the amber stop codon before the pIII gene in A, B and C. D.
Map of pHDA2 that encodes the fusion protein of HMG domB and the coat protein III.
168
r
'~
s
·aal
~·'
;Ir
· · c%
~i;
'·31
>~r ·
Figure 4.4. pHDB phage particles visualized under the electron microscope at 60 K magnification.
The phage appear as long flexible rods.
169
free DNA-*
I
I
0.33
0
0
pHDB1 phage (pM)
Salmon sperm DNA (pg/pl)
unPt 20 mer DNA competitor (pg/pl)
Pt 20 mer DNA
2.5
0.33
0
1
0
0
0.33
0
1
2.5
0
1
domB bound DNA
free DNA
I I
1
I
Pt 123 bp
pHDB1 phage (pM)
0
Salmon sperm DNA(pg/pl) 0
0.33
0
2.5
0
unPt 123 bp
0.33
1
2.5
1
0.33
0
2.5
0
0.33
1
Figure 4.5. Gel mobility shift assays of pHDB1 phage with DNA targets. A. Each lane contained 10
nM platinated 20 mer DNA probe. B. Bandshift of pHDB1 phage solutions with both platinated
and unplatined 123 bp DNA. For the platinated 123 bp DNA, rb = 0.02 and the concentration is 20 nM.
170
10000 bp
-P
2000 bp
1000 bp
500 bp
200 bp
-4
100 bp
-
1
2
3
Figure 4.6. Human placenta DNA electrophoresed on a 1%agarose gel.
Lane 1: molecular weight markers; Lane 2: undigested human placenta
DNA, Lane 3: human placenta DNA digested with RsaI, MseI and Sau3AI
restriction enzymes.
171
denotes pII coat protein while
denotes domA-pIII fusion protein
denotes platinum moiety
\'>
denotes oligonucleotide modified by cisplatin
4--'
denotes the interaction between fusion protein and DNA
Figure 4.7. Schematic diagram of how phage binds to DNA targets of different length.
A. When a long piece of DNA (>50 bp) is used as DNA target, the non-specific interaction
between the fusion protein and unplatinated region of the DNA will dominate the
specific interaction between the fusion protein and the cisplatin-modified DNA site.
B. When a short piece of DNA is used as target, the specific binding between the fusion
protein and the cisplatin-modified DNA dominates.
172
O unplatinated
0 Platinated
0.400
0.350
0.300
0.250
0.200
0.150
0.100
0.050
0
0.00
20bp DNA 250nM
30bp DNA 250nM
DNA targets (250 nM)
Figure 4.8. pHDA phage has higher specific affinity for platinated DNA than unplatinated
DNA. Binding reactions contain specific competitor DNA (1.25 pM), that has the same
sequence as the DNA target, but is unplatinated and unbiotinylated.
173
5'
CTGCCGGGAGGAGCACAAGAAGAAGCACCCG (NNK) 6TCAGAGTTCTCCAAGAAG
3
Random library of amino acids 32-37 of domA (the loop between helix I and helix II)
N= A/T/G/C
K= G/T
Anneal with complementary
bottom strand primers
5'
3'
CTGCCGGGAGGAGCACAAGAAGAAGCACCCG (NNK) 6TCAGAGTTCTCCAAGAAG
AGTCTCAAGAGGTTCTTCACGA
GCCCTCCTCGTGTTCTTCTTCGTGGGC
3'
5'
+
AAGAAGACCCGGATGCTCTGTCAACTTCTCACGAGTTCTGAAGACAG TGCTCAGA
AAAC CTGCCGGTCTTCGCAC
TTTGGACG GCGAACGTGTTCTTCTTCGTGGGCCTACGAAGACAGTTGAAGAGTCTCAAGCCTTCTGTCACGA GTCT
Bbs I site
Bbs I site
Modified pHDA phagemid was digested with Bbs I and purified
[
Figure 4.9. The scheme for constructing domain A phagemid library.
174
I
S
(U
1
............................
I
'
EI
-
I
/
'
1
0
r .
II
I
I
I
I
Ii
i
·i
···...........
.....
S-......
.......
i!.......
.......
0.01
I,
I
I-...I
...............
_
.
0.001
*
I
.......
0.1
0-
I
SI
I.
3
I,
,I
4
,,
6
,
7
Selection cycle
...
I
...
:
I1 ....
11
I
... 1
1
11
1...
.-
C)
C)
W
C)
/\
*
0.01
i
...........
. . . . I. . . . . . .ilI\i. . . . . . . \is
. . . . i.. . . . . . . . . . .
I
0D
0,
...... .
0.
%
0.001
Selection Cycle
Figure 4.10. Plot of phage retention efficiency versus selection cycles. (A) pZifl2 library
selection from Pabo and coworkers. (B) pHDA library selections.
175
HMG1
Do
1
2
3
4
5
6
7
8
9
Figure 4.11. Western blot of HeLa cells transiently transfected with overexpression vector for
HMG1 domains A and B. The primary antibody used was 1:200 dilution of HMG1 antibody
and 1:2000 dilution of HMG1 domB antibody mixture. chemiluminescent method was used to
visualize the bands. Lane 1. HMG1 protein standard. Lane 2. HMG1I domain A protein
standard. Lane 3. HeLa cell control. Lanes 4 and 5, HeLa cells transiently tranfected with
domA vector for 24 and 48 h. Lanes 6 and 7, HeLa cells transiently tranfected with domB
vector for 24 and 48 h. Lanes 8 and 9, HeLa cells stably transfected with domA vector.
176
Biographical Note
The author was born Qing He on December 25, 1973 in Tianjin, People's Republic of
China. She grew up in Tianjin untill the age of 17, when she came to the United
States for the senior year of high school in Seattle in 1991. She attended Franklin and
Marshall College in Lancaster, PA, completing her B.A. summa cum laude in
chemistry in 1995. For four years at F&M College, she performed research in the
laboratory of Professor J. Spencer, investigating the mechanism of inclusion between
alcohols and a-cyclodextrin and analyzing acid-base chemical reactions and hostguest complexation by calorimetric techniques. She has also conducted biochemistry
and molecular biology research internships in the laboratory of Professor E. Davie at
the University of Washington and Professor M. Poncz at the University of
Pennsylvania during the summers. The scholarships and awards she received
include the Merck Scholar, the Dana Scholarship, the American Chemical Society
POLYED Award for Organic Chemistry, the Spalding Leadership Award, the
American Chemical Society Analytical Chemistry Award and election to Phi Beta
Kappa. In 1995, she began graduate studies in biological chemistry at Massachusetts
Institute of Technology. As a Howard Hughes Medical Institute Predoctoral Fellow,
the author joined the laboratory of Professor S. J. Lippard, where she studied the
anticancer activity of cisplatin by focusing on high-mobility group proteins and
improving the efficacy of the drug. She also mentored the laboratory studies of an
undergraduate student, Cynthia H. Liang. She married Zhenqing Wu in 1998.
Following graduation, she plans to join industry in the New Jersey area.
5,~~r23G