Understanding the Functions of HMGB Proteins
in the Mechanism of Action of Cisplatin
by
Semi Park
B.S. Chemistry
Seoul National University, 2007
SUBMITTED TO THE DEPARTMENT OF CHEMISTRY IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY IN INORGANIC CHEMISTRY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
gggHyV
SITUTE
September 2012
© Massachusetts Institute of Technology, 2012
All rights reserved
Signature of Author:
Department of Chemistry
July 31, 2012
Certified by:
J St~phenJ Lippard
Arthur Amos Noyes Professor of Chemistry
Thesis Supervisor
I
~
Accepted by:
Robert W. Field
Chairman, Departmental Committee on Graduate Studies
This doctoral thesis has been examined by a committee of the Department of Chemistry as
follows:
Ifaniel G. Nocera
Henry Dreyfus Professor of Chemistry
Committee Chairman
0 Stfien J. Lippard
Arthur Amos Noyes Professor of Chemistry
Thesis Supervisor
(Ice Y. Ting
Ellen Swallow Richards Associate Professor
mistry
2
Understanding the Functions of HMGB Proteins
in the Mechanism of Action of Cisplatin
By
Semi Park
Submitted to the Department of Chemistry on August XX, 2012
In Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy in Chemistry
ABSTRACT
High mobility group box (HMGB) proteins are DNA-binding proteins that regulate many
important DNA-related processes. They are known to recognize the major lesion present in
cisplatin-modified DNA, and have been assumed to increase cisplatin cytotoxicity by "shielding"
the damaged site from the cellular repair apparatus. This thesis will describe the work in the area
of molecular biology and biochemistry to improve our understanding of the functions of HMGB
proteins in the mechanism of action of cisplatin.
In Chapter 1, the molecular biology of cisplatin and HMGB proteins is described based
on previously reported in vivo and in vitro experiments. The interaction of HMGB proteins and
platinated DNA is reviewed based on several structural studies of platinum-DNA adducts of
cisplatin, HMG box motif, and the binding complex of the two. Several cell-based studies
supporting a repair-shielding hypothesis, suggesting an inhibitory function of HMGB proteins in
the repair of cisplatin damage will be introduced with a focus on HMGB1, the most vigorously
investigated HMGB protein. Additionally, other HMGB 1-mediated processes that may be related
to cisplatin-triggered cell death will be discussed. Lastly, the correlation between testis-specific
HMGB proteins and the cisplatin hypersensitivity of testicular germ cell tumors will be
discussed.
Although some HMGB proteins including HMGB 1 inhibit the repair of platinated
damage on DNA in vitro, experiments conducted in live cells reveal conflicting correlation
between the expression level of HMGB1 and cisplatin cytotoxicity. Chapter 2 describes studies
in cultured human cancer cells aimed at examining the intracellular repair of platinum-modified
DNA and the influence of HMGB1 on the repair processes. The expression of HMGB1 is
artificially down-regulated by RNAi. HeLa and A549 cell lines present different trends in
cisplatin cytotoxicity upon HMGB1 knockdown. Intracellular repair of cisplatin is lower in
knockdown cells regardless of the parental cell line. This result stands in opposition to what is
expected from the repair-shielding hypothesis. In addition, the repair of different cisplatin
adducts was investigated in fibroblast cells deficient in one of the nucleotide excision repair
proteins in order to understand the repair mechanisms of each adduct.
Chapter 3 presents an in vitro study investigating the redox-dependence of the binding
interaction of HMGB1 to the cisplatin-1,2-d(GpG) intrastrand cross-link. Two cysteine residues
in HMGB1 domain A form a reversible disulfide bond under mildly oxidizing conditions. Both
3
HMGB 1 domain A and full-length HMGB 1 presented significantly weaker binding to a DNA
probe containing a 1,2-d(GpG) intrastrand cross-link when in their oxidized state compared to
their fully-reduced form. Mutagenesis studies on the cysteine residues revealed that this redoxdependence originates from disulfide bond formation. Footprinting analysis of a platinated DNA
probe bound to oxidized or reduced domain A showed that the asymmetric binding mode of
domain A to platinated DNA is not significantly altered by oxidation. These results suggest that
the cellular redox environment can influence the interaction of HMGB 1 with the platinated DNA.
In Chapter 4, the binding properties and repair inhibitory function of HMGB4 to
cisplatin-modified DNA are described. Based on its testis-restricted expression profile and
sequence similarity with HMGB 1, we propose that HMGB4 functions as a cisplatin cytotoxicity
enhancer in TGCT. To verify this hypothesis, HMGB4 and its binding domains were generated
recombinantly and interactions with cisplatin-modified DNA were investigated in vitro. The
binding properties of HMGB4 are quite similar to those of HMGB 1 except for a few differences:
i) full-length HMGB4 has stronger binding affinity than full-length HMGB1, because of a lack
of a C-terminal acidic tail. ii) binding of HMGB4 domain A is much more symmetric with
respect to the platinated lesion than that of HMGB 1 domain A. Furthermore, HMGB4 presented
stronger repair inhibition capacity than HMGB1 at an equimolar concentration. These results
support the hypothesis that HMGB4 enhances cisplatin cytotoxicity in TGCT.
Chapter 5 will be the conclusion chapter of this thesis. Works in this thesis will be
summarized with what we can learn from those results. In additions, future directions in the
study of HMGB proteins will be suggested.
Appendices A and B describe the incomplete work on HMGB4 in live cells. Appendix
A delineates the expression profile of human HMGB4 established by western blot analysis.
Human HMGB4 presented a testis-preferred expression. In Appendix B, attempts to establish a
HMGB4 knockdown testicular cancer cell line are described.
Thesis Supervisor: Stephen J. Lippard
Title: Arthur Amos Noyes Professor of Chemistry
4
Acknowledgments
I remembered how excited I was on my first day at MIT, with the expectation of the
wonderful days I would have at this place. The last five years at MIT have been filled with
exciting experiences that have far exceeded my expectations on that first day. There are many
individuals who have made my MIT life much more fruitful. At the end of my journey towards
my PhD, I am grateful for what I have achieved to the awesome people I have interacted with
over the last five years.
First, I would like to thank Prof. Stephen Lippard for giving me an opportunity to work
on this interesting project. It was a great pleasure to work in the Lippard Lab, where I could learn
about the different kinds of chemistry research. Prof Lippard is an amazing advisor who knows
how to guide each of his students in the best way, and he is a great role model as a scientist
whose commitment to science will never run out. I also admire Prof. Daniel Nocera and Prof.
Alice Ting for their commitment as committee members and the valuable suggestions they have
provided for my research.
I am very lucky to have had not only such a great advisor and committee members but
also amazing people to work with. Dr. Katherine Lovejoy was a talented mentor who helped me
when I was literally a novice in bioinorganic chemistry. I also received much help from Dr. Ryan
Todd when I was working on various experiments involving DNA and PAGE gel works. Dr.
Guangyu Zhu has been a reliable colleague during my days at MIT, and I would like to thank
him for giving me a chance to consider consulting as my next step. Thank you, Dr. Shanta Dhar
and Dr. Wee Han Ang, my good friends with whom I spent most of my early days in the Lippard
Lab. I owe a lot to Dr. Nora Graf, the most warm-hearted person I met in MIT, who encouraged
me when I going through a hard time. Dr. Yongwon Jung gave me valuable advice on EMSA
and footprinting experiments. I am grateful to Dr. Yang Li for all the valuable discussions on
science and history and the lots of coffee. Thank you to Dr. Woon Ju Song and Dr. Rachel
Wollacott for the training and advice on protein chemistry.
I am appreciative of all valuable discussion and suggestions on my research that the
Platinum Subgroup provided. Specifically, I would like to express my gratitude to Justin Wilson
and Tim Johnstone for their enormous help in improving my writing. Dr. Ying Song and Dr.
Patricia Marques-Gallego are really good friends and social events organizers who made my
5
days in the Lippard Lab quite exciting. Thank you to Dr. Ga Young Park and Dr. Yaorong Zheng
for helping me manage tissue culture facilities and to Meiyi Li for her help on the repair study.
I thank Julia, now Dr. Kozhukh, for being an awesome class/lab-mate for the last five
years. Two have been always better than one. All the steps I took in pursuing my PhD would
have been much more difficult if I had not had a good lab-mate like Julia. Thank you to Dr.
Shawn Lu, who was not only a nice lab-mate but also a good coffee-mate, for all the valuable
discussion on biochemistry. Thank you to Dr. Seung Jae Lee, for your help in protein chemistry.
I thank Ali Liang and Eric Victor, who always reminded me of how fun life could be when I was
having a gloomy day. Thank you to Dr. Ulf Apfel and Dr. Amit Majumdar for showing me how
good friends make lab life so happy. Thank you to Dr. Robert Radford, for his help on my thesis.
I also thank all Lippard Lab members for being awesome colleges to work with.
I thank Rich Girardi, who saved me countless times from troubles on all different kinds
of paperwork. Thank you to Susan Brighton, former chemistry education officer, for her help in
fulfilling the PhD requirements.
Sometimes, I have faced problems in research because of lack of my knowledge and
experience in an unfamiliar field. I would like to thank my friends in the biology and chemistry
society in Cambridge and Boston for being a big help on many of the troubleshooting processes I
that encountered during my thesis studies. Specifically, I received a lot of help from BAWI, the
alumni association of Seoul Science High School.
I thank my parents, Myungsun Park and HyeJung Kim, and my brother, Jihwan Park.
They have always respected my decisions and have been my biggest supporters at every single
moment. I appreciate their unconditional love. In addition to my PhD work, none of the things I
have achieved in my life would have been possible without them.
6
Table of Contents
Abstract
3
Acknowledgments
5
Table of Contents
7
List of Tables
12
List of Figures
13
Glossary of Terms
16
Chapter 1. HMGB Proteins and Cisplatin
18
1.1. Introduction
19
1.2. Molecular Mechanism of Cisplatin
21
1.2.1. DNA, the Major Target of Cisplatin
21
1.2.2. Structure of Cisplatin-Modified DNA
22
1.2.3. Nucleotide Excision Repair of DNA Damage
23
1.3. Repair-Shielding Model of HMGB Proteins
26
1.3.1. Damage Recognition Proteins of Cisplatin-Modified DNA
26
1.3.2. High Mobility Group Box Proteins
26
1.3.3. HMGB Proteins and Cisplatin-Modified DNA
28
1.4. Nature of HMGB 1 and its HMG Boxes Binding to Cisplatin-Platinated DNA
30
1.4.1. Structure of HMGB 1 and Homologous HMGB Proteins
30
1.4.2. Two HMG Box Domains in HMGB 1: Different Nature of Binding to DNA
31
1.4.3. Structure of Complexes of HMG Boxes with Cisplatin-Modified DNA
32
1.4.4. Function of the C-terminal Acidic Tail
35
1.5. Cellular Studies
36
1.6. Function of HMGB1 in Processes Other than Repair
37
1.6.1. Interaction of HMGB 1 with Intracellular Proteins
39
1.6.2. HMGB 1, an Extracellular Damage Associated Molecular Pattern (DAMP)
41
1.7. Testicular Cancer, Cisplatin, and HMGB Proteins
42
1.7.1. Cisplatin Hypersensitivity of Testicular Germ Cell Tumors (TGCTs)
42
1.7.2. HMGB Proteins in TGCT
43
1.8. Summary
45
7
1.9. References
46
Chapter 2. Effect of HMGB1 on NER of Platinated DNA in Cells
51
2.1. Introduction
52
2.2. Experimental
55
2.2.1. Cell Culture
55
2.2.2. Knockdown of HMGB 1 in HeLa and A549 Cells by Transfection of ShRNA-Expression
Vector
55
2.2.3. Western Blot
57
2.2.4. Immunofluorescence
58
2.2.5. Cytotoxicity Assay
58
2.2.6. Preparation of Platinated pGLuc Plasmids
59
2.2.7. Transcription Assay
60
2.3. Results
61
2.3.1. Establishment of HMGB 1 knockdown HeLa and A549
61
2.3.2. Cisplatin Sensitivity of HMGB1 Knockdown Cell Lines
63
2.3.3. Restoration of the Transcription Level of Platinated Plasmid in Knockdown Cells
65
2.3.4. Cisplatin and Phenanthriplatin Cytotoxicity in NER Proficient/Deficient Fibroblast Cell
Lines
68
2.3.5. Repair Profiling of Different Platinum-DNA Adducts in NER Proficient/Deficient Cell
Lines
69
2.4. Discussion
71
2.4.1. Cellular Response to Cisplatin in HMGB1 Knockdown HeLa and A549 Cells
71
2.4.2. Damage Repair Processes and HMGB 1
72
2.4.3. NER of Different Types of Platinum Adducts
74
2.5. Summary
77
2.6. References
79
Chapter 3. Redox State-Dependent Interaction of HMGB1 and Cisplatin-Modified DNA 81
3.1. Introduction
82
3.2. Experimental
85
8
3.2.1. Expression and Purification of HMGB 1 Proteins
85
3.2.2. Platination, Purification, and Characterization of 25-bp DNA Probes
86
3.2.3. Electrophoretic Mobility Shift Assays
89
3.2.4. Hydroxyl Radical Footprinting Assay
90
3.3. Results
91
3.3.1. Binding Affinity of Wild Type HMGBla to Platinated DNA Under Different Redox
Conditions
91
3.3.2. Site-Directed Mutagenesis of Cysteine Residues
93
3.3.3. Hydroxyl Radical Footprinting
96
3.3.4. Studies of Full-Length HMGB 1
98
3.3.6. Redox-Dependence of the Binding of Full-length HMGB1 to Cisplatin-Modified DNA 99
3.4. Discussion
102
3.4.1. Interaction of Cisplatin-Modified DNA with HMGB 1a in Different Redox States
102
3.4.2. Change of HMGBla Conformation Induced by Disulfide Bond Formation and Protein
Interaction with DNA
103
3.4.3. Influence of Cysteines on Conformation of HMGBla
104
3.4.4. Binding Properties of Reduced and Oxidized Full-Length HMGB1 and its Variants
105
3.4.5. In Vivo Redox Chemistry of HMGB1
106
3.5. Conclusion
108
3.6. References
109
Chapter 4. HMGB4, a Potential Enhancer of Cisplatin in Testicular Germ Cell Tumors 111
4.1. Introduction
112
4.2. Experimental
114
4.2.1. Cloning and Expression of HMGB4 Proteins
114
4.2.2. Preparation of 25-bp Site-Specifically Platinated DNA Probes
115
4.2.3. Preparation of a Repair Assay Substrate
115
4.2.4. Nucleotide Excision Repair Assay
117
4.2.5. Electrophoretic Mobility Shift Assay
117
4.2.6. Hydroxyl Radical Footprinting
118
4.3. Results
118
9
4.3.1. Expression and Solubility of HMGB4 Proteins
118
4.3.2. Binding of HMGB4 Proteins to Platinated DNA
119
4.3.3. Binding Affinities of Wild Type HMGB4 and its F37A Variant
122
4.3.5. Weak Asymmetry of HMGB4 Binding to a 1,2-Intrastrand Cross-Link
123
4.3.6. Repair Inhibitory Capacity of HMGB4 Proteins
125
4.4. Discussion
129
4.4.1. Natural Functions of Testis-Specific HMGB Proteins and Their Interaction with Cisplatin129
Modified DNA
4.4.2. Information about the Sequence of HMGB4 Relevant to its Binding Interaction with
130
Cisplatin-Modified DNA
131
4.4.3. Binding Affinity and Binding Specificity of HMGB4 to Platinated DNA
4.4.4. Function of Intercalating Residues and Their Influence on the Structure of the Platinated
DNA-Protein Complex
133
4.4.5. Repair Inhibition by HMGB Proteins
135
4.4.6. HMGB4, Cisplatin, and Testicular Cancer
135
4.5. Conclusion
136
4.6. References
138
Chapter 5. Summary and Future Directions
140
5.1. Summary and Future Directions
141
5.2. References
143
Appendix A. Expression Profiling of Human HMGB4
144
A. 1. Purpose
144
A.2. Experimental
144
A.2.1. Cell Culture
144
A.2.2. Western Blot Analysis
144
A.3. Results and Discussion
145
A.3.1. Expression Profile of Human HMGB4 in Cancer Cell Lines
145
A.3.2. Future Directions
146
A.4. References
146
10
Appendix B. Attempts to Knockdown of HMGB4 in NTera2
147
B.1. Purpose
147
B.2. Experimental
147
B.2. 1. HMGB4 Knockdown in NTera2 by Transfection of shRNA Expression Vector
147
B.2.2. HMGB4 Knockdown in NTera2 by Viral Transduction
147
B.3. Results and Discussion
148
B.3.1. Expression Levels of HMGB4 in Stably Transfected Cell Lines
148
B.3.2. Viral Transduction for HMGB4 Knockdown
149
B.3. Future Directions
150
B.4. References
151
Biography
152
Curriculum Vitae
153
11
List of Tables
Table 2.1 .Fibroblast Cell Lines Used for Repair Studies
55
Table 2.2. List of Stably Established HeLa and A549 Cell Lines and Their RNAi Efficiency
62
Table 2.3. IC50 Values of HMGB1 Knockdown HeLa and A549 Cell Lines
65
Table 2.4.Cisplatin and Phenanthriplatin IC50 Values of Each Fibroblast Cell Line
69
Table 2.5. The Change of Cellular Response to Cisplatin in HMGB1 Knockdown Cell Lines and
Expected Results Based on the Repair-Shielding Model
71
Table 3.1. Sequence of 25mer Oligonucleotides with Different Types of Cisplatin Binding
Sequences
87
-
Table 3.2. Dissociation Constants of Wild-Type and Variants of Domain A Binding to a 25-bp
Double-Stranded DNA Probe Containing a Site-Specific 1,2-d(GpG) Cross-Link as Determined
95
by Electrophoretic Mobility Shift Assays
Table 3.3. Dissociation Constants of Wild-Type and Variants of Full-Length HMGB1 Binding to
a 25-bp Double-Stranded DNA Probe Containing a Site-Specific 1,2-d(GpG) Cross-Link as
Determined by Electrophoretic Mobility Shift Assays
100
Table 4.1. Dissociation Constants of HMGB Proteins and Their Variants for DNA Probes
Containing a Central Cisplatin 1,2-d(GpG) Intrastrand Cross-Link
122
Table 4.2. Fraction of Repair Product of a Cisplatin-1,2-Intrastrand d(GpG) Cross-Linked DNA
in the Presence of Different Concentrations of HMGB Proteins
128
Table B. 1. List of pLKO l-Transfected NTera2 Cell Lines
147
Table B.2. List of Viral-Transduced NTera2 Cell Lines
148
12
List of Figures
Figure 1.1. Platinum-based compounds showing antineoplastic activity
20
Figure 1.2. Structure of different types of platinated DNA
23
Figure 1.3. The general mechanism of nucleotide excision repair
25
Figure 1.4. The solution NMR structures of HMGB 1 domain A and domain B
28
Figure 1.5. Sequences of human HMGB1-4
31
Figure 1.6. The crystal structure of the complex of HMGB1 domain A with cisplatin1,2-d(GpG)
intrastrand cross-linked DNA
34
Figure 1.7. Cellular functions of HMGBl presumably related to cisplatin-triggered cell death 38
Figure 2.1. Diagram of GLuc transcription assay using platinated substrates
54
Figure 2.2. HMGB1 knockdown efficiency analyzed by western blotting
62
Figure 2.3.
Immunofluorescence
results of HMGB1
knockdown HeLa in homogeneous
populations
63
Figure 2.4. Cisplatin cytotoxicity assay of HMGB 1 knockdown cell lines
64
Figure 2.5. Transcription profile of globally platinated pGLuc probe in heterogeneous
knockdown HeLa
66
Figure 2.6. Transcription profile of globally platinated pGLuc probe in heterogeneous
knockdown A549
67
Figure 2.7. Transcription profile of globally platinated probes in homogeneous HMGB1
knockdown A549 Cl1 and parental A549
68
Figure 2.8. Cisplatin and phenanthriplatin cytotoxicity of NER proficient/deficient fibroblast cell
lines and control fibroblast cell line
69
Figure 2.9. Transcription profile of site-specifically platinated probes in NER proficient/deficient
cell lines
70
Figure 3.1. X-ray crystal structure of the complex between HMGB1 domain A and a 16-bp
cisplatin-modified DNA under reducing conditions
84
Figure 3.2. Ion exchange HPLC purification of platinated 25mer oligonucleotides
88
Figure 3.3. S1 nuclease digestion of platinated 25mer oligonucleotides
89
Figure 3.4. EMSA analysis of wild type HMGB1a binding to the different DNA probes
92
Figure 3.5. EMSA analysis of wild type HMGBla bound to the cisplatin-modified DNA probe
under different redox conditions
93
13
Figure 3.6. The influence of reducing reagent on the binding of HMGB 1a to cisplatin-modified
94
DNA
Figure 3.7. Footprinting analysis of HMGB1 domain A bound to the cisplatin-modified DNA
97
probe
Figure 3.8. EMSA analysis of wild type domain A binding to cisplatin-modified DNA probe,
incubated in the presence of the reagents used to generate hydroxyl radicals
98
Figure 3.9. SDS-PAGE analysis of full-length HMGB1 in different redox states
99
Figure 3.10. EMSA analysis of full-length HMGB1 and its variants bound to the cisplatinmodified DNA probe
101
Figure 4.1. Sequences and structures of DNA binding motifs in HMGB 1 and HMGB4
113
Figure 4.2. Preparation of in vitro NER substrates
116
Figure 4.3. Purified full-length HMGB4 and its DNA binding domains resolved in 4-20%
gradient SDS gels
119
Figure 4.4. EMSA analysis of HMGB4 and HMGB4a bound to different probes
120
Figure 4.5. Binding preference of full-length HMGB4 and its DNA-binding domains to the
121
cisplatin-modified DNA probe
Figure 4.6. EMSA analysis of HMGB4 proteins and their F37A variants binding to platinated
123
DNA
Figure 4.7. Footprinting analysis of DNA-binding domains of HMGB 1 and HMGB4 bound to
124
the cisplatin-modified DNA probe
Figure 4.8. Footprinting analysis of HMGB 1 domain A and full-length HMGB4 proteins bound
to a cisplatin-modified DNA probe
125
Figure 4.9. Nucleotide excision repair products resolved on a 8%urea-PAGE gel
126
Figure 4.10. Excision of a cisplatin- 1,2-intrastrand d(GpG) cross-link by active repair extracts in
127
the presence of HMGB 1 or HMGB4
Figure 4.11. Effect of HMGB4 domain A on excision of a cisplatin 1,2-intrastrand d(GpG) cross128
link
Figure 4.12. X-ray crystal structure of the complex of the HMGB1 domain A and a 16-bp DNA
probe containing a cisplatin 1,2-intrastrand cross-link
131
Figure A. 1. Western blot analysis of human HMGB4
145
14
Figure B. 1. Western blot analysis of HMGB4 in NTera2 cells transfected by pLKO 1 shRNA
expression vectors
145
Figure B.2. Flow cytometry analysis of GFP expression in viral transduced NTera2 cells
150
Figure B.3. Western blot analysis of NTera2 cells transduced by viral particle with HMGB4
150
shRNA gene
15
Abbreviations
BER
bp
BSA
carboplatin.
CFE
cisplatin
CS
DAMP
DEAE
DNA-PK
DRP
DSBR
DTT
EC
EDTA
EMSA
ER
ERCC1
ERE
ESI-MS
FBS
GAPDH
GGR
Gluc
HBP1
HDR
HMG
HMGA
HMGB
HMGN
HPLC
IAP
ICL
IxrI
LEF-1
MAPK
MEF
MMR
MTT
mtTFA
MutSa
NER
NF-idB
NHEJ
base excision repair
base pair
bovine serum albumin
cis-diammine(1, 1-cyclobutanedicarboxylato)platinum(II)
cell free extract
cis-diamminedichloroplatinum(II)
Cockayne syndrome
damage associated molecular pattern
diethylaminoethyl cellulose
DNA-dependent protein kinase
damage recognition protein
double strand break repair
dithiothreitol
embryonal carcinomas
ethylenediaminetetraacetic acid
electrophoretic mobility shift assay
estrogen receptor
excision repair cross complementation group 1
estrogen responsive element
electrospray ionization mass spectrometry
fetal bovine serum
glyceraldehyde 3-phosphate dehydrogenase
global genome repair
Gaussian luciferase
HMG-box transcription factor 1
homology-directed repair
high mobility group
high mobility group AT-hook
high mobility group box
high mobility group nucleosome binding domain
high performance liquid chromatography
inhibitors of apoptosis protein
interstrand cross-link
intrastrand cross-link recognition I
lymphoid enhancer-binding factor 1
mitogen-activated proteins kinase
mouse embryonic fibroblast
mismatch repair
[3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide]
mitochondrial transcription factor A
mutator Sa
nucleotide excision repair
nuclear factor KB
non-homologous end joining
16
NHP6A
Ni-NTA
oxaliplatin
PBS
PAGE
PCNA
PDIA3
phenanthriplatin
RNA Pol II
PMSF
PNK
PVDF
pyriplatin
RAG
RAGE
RB
RNAi
RPA
SDS
shRNA
SRY
SSRP1
TBE
TBP
TBS
TCR
TFIIH
TGCT
TLR
trans-DDP
tsHMG
UBF
XP
non-histone chromosomal protein 6A
nickel-nitrilotriacetate
(1R,2R-diaminocyclohexane)oxalatoplatinum(II)
phosphate buffered saline
polyacrylamide gel electrophoresis
proliferating cell nuclear antigen
protein disulfide isomerase family A, member 3
cis-diammine(phenanthridine)chloroplatinum(II)
RNA polymerase II
phenylmethylsulfonylfluoride
polynucleotide kinase
polyvinylidene fluoride
cis-diammine(pyridine)chloroplatinum(II)
recombination activating gene
receptor for advanced glycation end product
retinoblastoma
RNA interference
Replication protein A
sodium dodecyl sulfate
short-hairpin RNA
sex-determining region Y
structure-specific recognition proteins 1
Tris/Borate/EDTA
TATA binding protein
Tris-buffered saline
transcription coupled repair
transcription factor II H
testicular germ cell tumor
toll-like receptor
trans-diamminedichloroplatinum(II)
testis specific HMG
upstream transcription factor
xeroderma pigmentosum
17
Chapter 1. HMGB Proteins and Cisplatin
18
1.1. Introduction
Cisplatin is one of the most widely prescribed anticancer drugs. Even though it was
reported more than one and half centuries ago as "Peyrone's salt", the biological activity of
cisplatin was not realized until Barnett Rosenberg serendipitously observed inhibition of E.coli
cell division at platinum electrodes. 1 This effect was later revealed to be due to cisplatin
produced by electrolysis. Following FDA approval in 1978, cisplatin has come to be prescribed
as the standard therapy for diverse types of cancers, such as testicular, ovarian, head and neck,
cervical, bladder, and non-small cell lung, usually in combination with other anticancer
reagents. 2
The success of cisplatin created a new category of drugs based on metal ions.3 Although
cisplatin is one of the most effective antitumor drugs, it has significant problems that restrict its
clinical use or cause negative outcomes following treatment. Since it was first introduced into the
clinic, side effects such as nausea, vomiting, ototoxicity, nephrotoxicity, and neurotoxicity have
5,6
been continuously reported. ' Another problem is the resistance to cisplatin that cancer cells can
develop.
'
As well as intrinsic resistance of specific types of cancer, tumors initially presenting
an acute response to cisplatin can develop drug resistance over the course of treatment.
Attempts to overcome these drawbacks have focused mainly on two goals: i) developing new
platinum-based anticancer drugs with better cytotoxicity and/or fewer side effects and ii)
understanding the molecular mechanism of cisplatin to improve the efficiency of its
chemotherapy.
As a consequence of studies in the first category, numerous platinum-based anticancer
drug candidates have been prepared and evaluated, including two FDA-approved compounds,
carboplatin and oxaliplatin (Figure 1.1).9 Carboplatin presents a cytotoxicity profile similar to
19
that of cisplatin with fewer side effects.'10 11 In particular, the dose limiting toxicity of carboplatin
no longer arises from nephrotoxicity as with cisplatin.12'13 Oxaliplatin is active in colorectal
,
cancers having intrinsic resistance to cisplatin.14 15
Pt 1
H3N
O 0
H3N
CI
H3N
H2
1Pt
H3N
Cl
.Pt
O
ON
H2
00
Oxaliplatin
Carboplatin
Cisplatin
0
0
0
H3NI
H3N
CI
CHC12
H3 N
+1
PtPt
CI
H3N
N-
O
C 2HC
1
0
Mitaplatin
Pyriplatin
Figure 1.1. Platinum-based compounds showing antineoplastic activity. Cisplatin, carboplatin, and oxaliplatin are
traditional bifunctional platinum(II) compounds approved by the FDA for clinical use. Platinum(IV) prodrugs like
17
16
mitaplatin or monofunctional platinum(II) compounds like pyriplatin are new types of drug candidates that present
promising properties for improved chemotherapy.
In addition to bifunctional platinum(II) compounds like cisplatin, recent studies have
focused on non-traditional platinum compounds, such as platinum(IV) compounds or
monofunctional platinum(II) compounds. Platinum(IV) compounds are kinetically inert prodrugs with two added axial ligands. Such compounds can remain in the +4 oxidation state under
extracellular conditions and be reduced to the "active" Pt(II) form only upon entering cells. It is
also possible to introduce the tumor-selectivity into the compounds by adding targeting moieties
at the axial positions.18 ' 19 Cytotoxicity can be increased by incorporating units that potentiate
cisplatin activity at the axial positions. 16 Monofunctional platinum compounds, like pyriplatin,
20
form monoadducts on DNA as opposed to DNA cross-links characteristic of traditional
bifunctional compounds. 17 ,2 0 This class of compounds exhibits a cytotoxicity profile quite
distinct from that of cisplatin.2 1
Molecular biology and biochemistry studies have helped to construct a picture of the
antitumor mechanism of cisplatin from uptake to cell death.22 Although much work has been
done, there remain many questions about the different steps of the mechanism. Compared to the
activation and DNA binding processes that occur early in the action of cisplatin, downstream
events triggered by platinum damage are less well understood, due to the complexity of the
cellular system. Several damage recognition proteins (DRPs) recognize DNA distorted by
platination.23,24 They are considered to affect the repair of cisplatin damage or to initiate damagetriggered cell death.
The present thesis focuses on high mobility group box (HMGB) proteins, which
recognize cisplatin-modified DNA. In this chapter, the general molecular mechanism of cisplatin,
the properties of HMGB proteins, and the relation of cisplatin to HMGB proteins will be
discussed. The involvement of HMGB proteins in "repair-shielding hypothesis" will be
described in detail. Various possible roles of HMGB proteins in the mechanism of action of
cisplatin will be presented. The possibility that some HMGB proteins act as enhancers of
cisplatin cytotoxicity in testicular germ cell tumors (TGCTs), which are cisplatin hypersensitive,
will be discussed.
1.2. Molecular Mechanism of Cisplatin
1.2.1. DNA, the Major Target of Cisplatin
21
Many studies using various methods have focused on one very fundamental question:
What are the intracellular interaction partners/reaction products that result in apoptotic/necrotic
cell death following treatment with cisplatin? These studies revealed that the major target of
platinum-based compounds is DNA,
and they identified several proteins that might be involved
in intracellular pathways triggered by platinated DNA.2 6-28 After the uptake of cisplatin into cells
from the extracellular matrix, the relatively low concentration of chloride ions in the intracellular
environment promotes replacement of chloride ligands with water molecules. 29 The resulting
aquated forms of the platinum compound are labile enough to undergo further ligand
replacement by nucleophilic cellular components. According to a 195Pt NMR study of the
cisplatin binding kinetics, the formation of cross-links occurs by two successive processes.30 The
half life of the first process, the formation of a cisplatin monofunctional adduct bound to the N7
position of a purine base, is about 2 h at 37 'C and pH 6.5. This value is in agreement with the
rate of replacement of the first chloride ligand with a water molecule, suggesting that this
aquation step is the rate-determining step of the monofunctional adducts formation. Subsequent
formation of bifunctional adducts occurs on a similar time scale (tu/2 = 2 h). Kinetic analysis
revealed that, despite the proximity of the other bases, the closure of mono- to bifunctional
adducts under in vitro conditions occurs via the aquation of monofunctional adducts rather than
the direct replacement of remaining chloride ligand with an adjacent DNA base.30 Cisplatin
mainly forms 1,2-intrastrand d(GpG) or d(ApG) cross-links, accounting for 65% and 25% of
total platinum-DNA adducts, respectively. It also produces 5-10% 1,3-intrastrand d(GpNpG)
3 132
cross-links and residual amounts of interstrand cross-links and monofunctional adducts. ,
1.2.2. Structure of Cisplatin-Modified DNA
22
The formation of a platinum cross-link causes a significant distortion in the DNA
structure. Both 1,2- and 1,3-intrastrand cross-links induce significant unwinding of DNA around
,
the platinum binding site and bending of the DNA toward the major groove (Figure 1.2.A,B).33 34
The X-ray crystal structure of a 1,2-intrastrand cross-linked DNA revealed a unique hybrid of A-
type DNA-like structure at the 5'-side and B-type DNA-like structure at the 3'-side of the
platinated guanines. 35 Solution phase NMR studies showed the presence of only B-type DNA
and a larger degree of DNA bending than in crystal structure.33,36 The interstrand cross-link of
cisplatin causes the DNA to bend toward the minor groove (Figure 1.2.C).37 Monofunctional
platinum adducts do not significantly alter the DNA structure (Figure 1.2.D). 38 It is generally
accepted that cyototoxic pathways triggered by cisplatin-DNA adducts are initiated by
recognition of the distorted DNA structure by damage recognition proteins.
(A)
(B)
(C)
(D)
Figure 1.2. Structure of different types of platinated DNA. Solution structure of a 1,2-intrastrand cross-linked 12mer (A),
a 1,3-intrastrand cross-linked 13mer (B), and an interstrand cross-linked 10mer (C) DNA duplex of cisplatin from NMR
data and (D) X-ray crystal structure of 16mer DNA duplex bearing a pyriplatin monoadduct.
1.2.3. Nucleotide Excision Repair of DNA Damage
Nucleotide excision repair (NER) is one of the main DNA damage repair mechanisms of
cellular systems. NER is involved in the repair of cisplatin 1,2-intrastrand, 1,3-intrastrand, and
23
interstrand cross-links.39 4 2 In addition, recent in vitro repair and intracellular transcription assays
showed that NER can repair DNA bearing a monofunctional pyriplatin adduct.38 The high
cisplatin sensitivity of cells deficient in NER components such as xeroderma pigmentosum (XP)
and Cockayne syndrome (CS) cells reflects the importance of NER in repair of platinated
DNA.
4 3,44
The NER process comprises recognition, DNA unwinding, damaged-strand excision, and
restoration. 4 5 There are two sub-pathways of NER, transcription coupled repair (TCR) and global
genomic repair (GGR), which differ in the way that the damage is recognized (Figure 1.3). In the
GGR mechanism, the xeroderma pigmentosum group C (XPC)- human homologue of RAD23B
(hHR23B) complex senses the damage.46 This complex has specific affinity for various types of
DNA damage and induces considerable alteration in DNA structure. This alteration is probably
necessary to recruit other proteins for subsequent steps. On the other hand, the stalling of RNA
polymerase II (RNA pol II) at the damaged site functions as an initiation signal for TCR.4 7
Cockayne syndrome proteins A (CSA) and B (CSB) interact with stalled RNA pol II on DNA
and play indispensable roles in TCR. 4 8 Once the damage is recognized in either sub-pathway,
several NER proteins are recruited to form the repair apparatus around the lesion. Transcription
factor II H (TFIIH) is a multi-protein complex containing xeroderma pigmentosum group B
(XPB) and D (XPD) helicases that partially unwind the DNA duplex at the damaged site.
Cleavage of 5'- and 3'-ends of the damaged strand is carried out respectively by the excision
repair cross-complementing 1 (ERCC 1)/xeroderma pigmentosum group F (XPF) complex and by
xeroderma
pigmentosum
group
G
(XPG),
producing
a
25-32-bp
single-stranded
oligonucleotide. 49 After dual incision, DNA pol &and pol 6 synthesize new DNA to fill the gap
from the 3'-side assisted by proliferating cell nuclear antigen (PCNA) and replication factor C
24
(RFC). Finally, the DNA is ligated on the 5'-end by DNA ligase I.50 The lack of any of these
NER proteins results in a significant decrease in the efficiency of cisplatin damage repair.
INA Damage
-~IA
HO
A
HR2
~w1=
20,
13
XPRERCCI
f
-j4
-0
6
Figure 1.3. The general mechanism of nucleotide excision repair. The damage is recognized either by stalled RNA pol H
(TCR, 1A.1) and subsequent interaction with CSB and CSA (1A.2) or the XPC/hHR23B complex (GGR, 1B). After
recognition, several proteins are recruited on the damaged site (2), and helicases in the TFH complex form an open
"bubble" around the damage (3). Then, exonucleases are recruited (4) and the single strand oligonucleotide bearing the
damage is removed by dual incision (5). The gap is filled by polymerases (6) and ligated at the 5' end (7).
NER efficiency varies with the type of DNA damage. Repair active cell free extract (CFE)
or reconstituted systems containing recombinant NER proteins have contributed significantly to
our understanding of the repair of different types of DNA damage. 51
25
52
Despite their utility as in
vitro models, these artificial repair systems have some shortcomings that limit their ability to
imitate in vivo systems. Quantitative measurement of repair using the artificial systems is
difficult owing to undesired background from the DNA degradation. Another important
complication arises from the use of 120- to 150-bp linear duplexes as a substrate for repair assays
in CFE systems. These repair substrates are typically used because it is easier to prepare than
closed circular substrates. They are, however, not proper substrates for TCR because they are not
long enough to involve essential transcription elements, such as the promoter sequence.5 1 Even
though this problem can be partially solved by changing the repair substrate to closed circular
DNA with appropriate transcription elements, the low transcription efficiency of in vitro
reconstituted or CFE systems still raises the question of the relevance of these model systems to
the true intracellular system.
1.3. Repair-Shielding Model of HMGB Proteins
1.3.1. Damage Recognition Proteins of Cisplatin-Modified DNA
trans-diamminedichloroplatinum(II) (trans-DDP) is the geometric isomer of cisplatin. The
two compounds exhibit comparable DNA-binding kinetics and similar Pt:DNA binding ratios in
vivo.
Moreover, they are both able to effectively inhibit DNA replications.
Despite these
similarities, the trans isomer is ineffective as a therapeutic agent. 54 The difference between these
geometric isomers gave rise to the hypothesis that the antitumor activity of platinum-based
compounds might be related to a specific structural alteration in DNA induced by the cis isomer.
If so, proteins recognizing platinated DNA in a structure-specific manner may play critical roles
in the mechanism of action of cisplatin. Many damage recognition proteins have been linked to
26
the molecular mechanism of cisplatin, including XPC,55 TATA-box binding proteins (TBP),56
replication protein A (RPA)," hMSH2,5 and high mobility group box (HMGB) proteins. 28 59
1.3.2. High Mobility Group Box Proteins
The first high mobility group box proteins were isolated in the early 1970s from calf
thymus and were designated as high mobility group 1 (HMG1) and 2 (HMG2) because of their
high mobility in an electrophoretic field.60 High mobility group proteins are categorized into
three sub-families depending on their DNA binding motifs: 1) the HMG AT hook (HMGA), 2)
HMG box (HMGB), and 3) HMG nucleosome binding (HMGN). Each sub-family has its unique
binding mode to DNA and plays critical roles in many DNA-related processes. 61 HMG1 and
HMG2 later were re-designated as HMGB1 and HMGB2.6 2
The HMG box consists of three alpha helices that can cooperatively wrap around DNA
from the minor groove, in an L-shape (Figure 1.4). 63 Binding of HMGB proteins induces
substantial distortion in DNA and produces partially unwound duplexes that facilitate proteinDNA complex formation. 64,65 HMGB proteins are classified into two different groups, according
to their binding specificity. 64 Proteins in the first category have very weak sequence-specificity
but demonstrate structure-specific binding to DNA; they have a strong binding preference for
unwound or bent DNA over linear DNA. 66,67 They usually have multiple HMG boxes with
highly conserved amino acid sequences and are present in higher intracellular levels than
proteins in the second category. For example, HMGB1, the most abundant mammalian HMGB
protein, is expressed at a density of approximately 1 molecule per 10-15 nucleosomes.68
Mammalian HMGB 1-4, mitochondrial transcription factor A (mtTFA), and upstream
transcription factor (UBF) all belong to this first group. On the other hand, HMGB proteins in
27
the second group exhibit strong sequence-selectivity. They have relatively low expression levels
or specific tissue-restricted expression patterns, and usually have only one HMG box with an
amino acid sequence optimized to recognize a specific DNA sequence. Lymphoid enhancerbinding factor 1 (LEF-1) and sex-determining region Y (SRY) proteins fall into this group of
HMG box proteins.
(B)
(A)
Helix III
Helix III
Phe'102
Phe37
Al
Ile'120
elix
I
Helix 11
Helix 11
Helix I
Figure 1.4. The solution NMR structures of HMGB1 domain A (A) and domain B (B). Residues in intercalating positions
are shown in green.
1.3.3. HMGB Proteins and Cisplatin-Modified DNA
Western blot and cDNA library screening analyses uncovered three candidates of
cisplatin-DNA damage recognition proteins. One of them had a molecular weight of ~100 kDa
and the others were about 28 kDa in size. 59 Subsequent studies identified the larger protein as
structure-specific recognition protein 1 (SSRP1),28 which has a HMG box at the middle of the
sequence, and the smaller ones as HMGB1 and HMGB2.6 97, 0 Interestingly, there is almost no
binding preference of HMGB proteins for DNA containing a 1,3-intrastrand cross-link compared
28
to unmodified DNA. These proteins have a strong binding preference for cisplatin 1,2-intrastrand
cross-linked DNA.69 HMGB proteins also bind preferentially to cisplatin interstrand cross-linked
DNA, although this binding interaction is much weaker than that between HMGB proteins and
1,2-intrastrand cross-linked DNA.71 An in vitro study using a cisplatin analog capable of forming
photo-activated cross-links between DNA and DNA-bound proteins revealed that more than 70%
of the total fraction of proteins in nuclear extracts bound to 1,2-intrastrand platinum lesions are
HMGB proteins, including HMGB1, 2, 3, and UBF. Only 20% and 6%, respectively, of the
bound proteins were identified as HMGB proteins when 1,3-intrastrand cross-linked and
interstrand-cross-linked DNA were used. 7 2 ,73 Because 1,2-intrastrand cross-links comprise more
than 90% of cisplatin-DNA adducts in vivo, the binding preference of HMGB proteins to 1,2intrastrand cross-links implies its importance in the mechanism of action of the drug.
The first evidence suggesting that HMGB proteins affect cisplatin cytotoxicity was
obtained from an in vivo yeast study. 74 A Saccharomyces cerevisiae strain deficient in
interstrand cross-link recognition protein I (IxrI), an HMGB protein that regulates transcription
of the yeast genome, was two-fold less sensitive to cisplatin than its IxrI-positive counterpart.
The trans-DDP sensitivity of these two strains was not significantly different. From this result
and the nature of the binding of HMGB proteins to platinated DNA, two initial hypotheses were
proposed concerning the role of HMGB proteins. One, the "repair-shielding hypothesis",
suggests that HMGB proteins bound to platinated DNA may shield the lesion from repair
proteins. Alternatively, in the "hijacking hypothesis", sequestration of HMGB proteins by
platinated lesions prevents them from executing their normal cell functions essential for
survival.75
29
Subsequent in vitro repair studies using site-specifically platinated DNA probes shed
further light on the question of the function of HMGB proteins. The addition of two
recombinantly expressed HMGB proteins, HMGB1 and human mitochondrial transcription
factor A (mtTFA), dramatically decreased repair of cisplatin-modified DNA substrates by repairactive CFEs. 40 Interestingly, this repair inhibition was observed for a 1,2-intrastrand cross-linked
substrate, but not for a 1,3-intrastrand cross-linked substrate. This finding reflects the adductspecific binding preference of HMGB proteins. The repair inhibition efficiency varied depending
on both the concentration and binding affinity of the HMGB proteins. This study strongly
supported the repair-shielding hypothesis.
1.4. Nature of HMGB1 and its HMG Boxes Binding to Cisplatin-Modified DNA
1.4.1. Structure of HMGB1 and Homologous HMGB proteins
Three canonical mammalian HMGB proteins, HMGB1, 2, and 3, have highly conserved
amino acid sequences. They have two HMG boxes, domain A and domain B, and a C-terminal
acidic tail consisting of aspartates and glutamates (Figure 1.5). The amino acid sequences of the
HMG boxes of these three proteins are nearly identical. This sequence information suggests that
the DNA binding properties of these proteins may be quite similar. In vivo experiments using
HMGB 1 or HMGB2 knockout mice, however, indicated that there are non-redundant biological
functions for each protein. 76 A more recently discovered protein, HMGB4, shares this sequence
similarity, but lacks a C-terminal acidic tail. 77 Also, three of four intercalating residues - Phe37
in domain A being the exception - are different in HMGB4. The other three mammalian HMGB
proteins share identical intercalating residue (Figure 1.5).
30
-- -Ha f 1
Ade -aim
...............................
|
HMGB1
HEGB2
HKGB3
HMGB4
GKGDPKKPRGKMSS
GKGDWPNKPRGKKSS
AKGDPKKPGEMS&U
GKEZQLKPKANVSS !LXP=
|Ada...En:Hex
HMGB1
KMGB2
HMGB3
HMGB4
HMGB4
SEFS
EKKK:Ds
EH
ZFS
4EEKKKE:PDS *
EFS
EIE EKREPEY
NKEQQPET*A KEFS
B
e
2
E RWKTNSAKEKGKFE
ERWKTMSIKEKSKFE
ERWKTMSGKEKSKFD
EKWRSISKHKKKYE
eaminHefx1
|
|
60
60
60
60
---
LFCSEYRPKXKGEHPGLS
LFCSEHRPKIKSEHPGLS
LFCSEFRPKIKSTNPGIS
LFCQDHY&QLKRENPNWS
120
120
118
118
man: Heffx3
DVKKLGEMWNNTlADDKQPYEKKKRNT-KEKYEKDIAYRAKGKPDAArGVVKEEKS
DTKKLGEMWSEQSAEDKQPYEQKATKLRKYEKDkIAYRAKGKSEAGKKGPGRPTGS
DVTKKLGEMNNLNDSEKQPYITKAkLMKEKYEEDVADYKSKGKDGG----PAKV
*QVAKLTGKMSTATDLEKEPYEQRVALLRKKYFEELELYRKQCN--------------
HMGBi
HMGB2
HMGB3
HKGB4
A demmatn:
DaKADKRYEREMKTYIPPKGETKKKrDPNPKRPPS
DMAKSDEARYDREKKEYVPPKGDEKGEKDENAPKRPPS
EEKYRYDREMEDYGPAEGGEKK--DPNPKRPPSG
KILWTKnARYQEENMEYGKR--KKRRKRDPQAPRRPPSS
A
HeA
| EdaumdkIiniI1:11
KKGB1
KMGB2
DMGB3
-
AREKYRMSARNDCRGKRVRQS-------------
180
180
174
164
214
208
199
185
Figure 1.5. Sequences of human HMGB1-4. The dark blue background denotes the residues at the positions
corresponding to intercalating residues in HMG boxes. The additional two amino acids at the N-terminus of domain A
helix 11 is a conserved feature of HMGB1-4 domain As. The C-terminal acidic tails of HMGB1-3, in light blue
background, vary in length among proteins. HMGB4 does not have the acidic tail. Two cysteines in red boxes can form
intra-domain disulfide bond under the mildly oxidizing conditions. Cys22 is replaced with tyrosine in HMGB4. The
consequence of the formation of disulfide linkage will be described in Chapter 2.
1.4.2. Two HMG Box Domains in HMGB1: Different Nature of Binding to DNA
Structures of the HMG boxes of HMGB1 have been determined by NMR spectroscopic
and X-ray crystallographic studies (Figure 1.4).78'79 In their free state, they retain an L-shape
similar to that in the DNA-bound form, with helices I and II as the longer arms and helix III as
the shorter one. The structure of HMGB1 domain B is substantially similar to other HMG boxes,
such as those in HMG-D or LEF-1. 80 HMGB1 domain A, despite its sequence similarity to
domain B, has some unique structural features not found in domain B or other HMGB proteins.
Notable are the linker region between the first two helices and the position of the main
intercalating residue, which will be described in the next section. Both domains A and B bind
with high affinity to bent DNA compared to linear DNA.-'
31
Domain A has a particularly high
binding affinity for bent DNA, whereas it cannot bend DNA effectively especially compared to
other HMG boxes. 8 ' On the other hand, binding of domain B is less selective for distorted DNA
but it induces more DNA bending than domain A. 80
HMGB 1 domain A binds to cisplatin 1,2-intrastrand cross-linked DNA better than
domain B by a factor of 2.5-130, depending on the DNA sequence to the 3' side of the platinated
lesion, which forms specific hydrogen bonds with HMG boxes. 81,84 Both domains of HMGB 1
also bind with higher affinity to cisplatin interstrand cross-linked DNA compared to unmodified
DNA.7 Domain A is the main DNA binding domain of HMGB 1. It interacts with the platinated
lesion of 1,2-intrastrand or interstrand cross-linked DNA. Domain B may simultaneously
stabilize the binding interaction and further distort the DNA. 85
1.4.3. Structure of Complexes of HMG Boxes with Cisplatin-Modified DNA
Complexes of HMG boxes with DNA are stabilized by hydrophobic interactions, specific
hydrogen bonds, and electrostatic interactions that involve water molecules at the interface of the
alpha helices and the DNA backbone. 84' 86 There are also intercalating residues at the specific
positions within the HMG boxes. These residues play an indispensible role in the binding
interaction. 80' 87 Cooperating with adjacent amino acids, the intercalating residues form a
hydrophobic wedge protruding into the minor groove of the DNA and inducing a significant
bend of the duplex. There are two possible positions for intercalating residues in each HMG box.
The first intercalating residue is located in the first alpha helix and is a large aliphatic residue
such as methionine in non-histone chromosomal protein 6A (NHP6A) and HMG-D or
phenylalanine of HMGB 1 domain B. The intercalation induces a major kink in the bound duplex.
The second intercalating residue is located at the N-terminus of the second helix. The
32
intercalation of the second residue induces another kink in the duplex a few base steps away
from the first kink. Unlike the first intercalating residue, the properties of residues at second
position vary for sequence-specific versus sequence-neutral HMG boxes, suggesting a role in
structure-specific binding preference. In the HMGB1 domain A, Alal6 is at the position that is
corresponding to the first intercalating residues of other HMG boxes. The methyl side chain of
alanine is not large enough to properly intercalate into the minor groove of the duplex. A lack of
the first intercalation explains the weak DNA bending ability of HMGB1 domain A. The
position of the second intercalating residue in HMGB1 domain A is occupied by Phe37, which
has a benzyl side chain that can insert between DNA bases. The strong binding preference of
HMGB1 domain A for distorted DNA is attributed to the unique character of Phe37. There is,
84 86
however, no reported structure of domain A bound to an unmodified DNA. ,
The structure of HMGB1 domain A bound to a short DNA duplex containing a central
cisplatin 1,2-intrastrand d(GpG) cross-link has been determined (Figure 1.6). It has revealed
structural features that differ from those of other HMG boxes bound to DNA.8 4 The most
interesting feature is that of the side chain of Phe37, which projects into the hydrophobic notch
between platinum-bound guanines. The n-n stacking and face-to-edge interactions between the
phenyl ring and the two guanine moieties stabilize the complex significantly. Replacing
phenylalanine with alanine at this position reduces the binding affinity of domain A to the
platinated DNA by a factor of more than 1000.86,88 Because this strong interaction holds Phe37 at
the platinated lesion, the binding mode of domain A, revealed by hydroxyl radical footprinting
analysis, is asymmetric with respect to the lesion, extending to the 3' side of modified guanines
(Figure 1.6.B). 84 Even though it does not intercalate much into the DNA duplex, Ala1 6
33
constructs a hydrophobic wedge with Tyr15, Val19, Val35 and Phe40, which protrudes into the
minor groove of the dulex. 86
(A)
(B)
{Pt(NH
}2
Helix I
Helix II
Helix III
Figure 1.6. The crystal structure of the complex of HMGB1 domain A (red) with cisplatinl,2-d(GpG) intrastrandcrosslinked DNA (blue). 81 (A) The benzyl ring of Phe37 is inserted between two platinum-modified guanines. (B) The
molecular surface of the complex shows the clear asymmetric binding of HMGB1 domain A to the cisplatin-modified
duplexes.
One more unique feature of domain A that differs from those of other HMG boxes is that
the linker region between helix I and helix II is elongated by two amino acids, Val35 and Asn36,
located just before Phe37 (Figure 1.5). Not only do they allow Phe37 more flexibility to form an
optimal structure for intercalating into the hydrophobic notch, but the valine residue also
interacts directly with the DNA backbone, as revealed in the crystal structure. Hydrogen bonding
between Ser4l and the base to the 3' side of the cross-link is also considered to be one of the
reasons for the strong binding affinity of domain A to the platinated DNA. This serine residue is
conserved in domain A of all four mammalian HMGB proteins. There is no available structure of
domain B bound to platinated DNA, but a mutagenesis study accompanied by hydroxyl radical
34
footprinting analysis revealed that Phe 102, at the position corresponding to Alal 6 in domain A,
interacts with the modified guanines, resulting in very symmetrical binding to cisplatin-modified
DNA, in contrast to the asymmetric binding of domain A. 86
Unlike the dramatic decrease in the binding affinity of the F37A variant of domain A, a
decrease of only 2-3 fold in the binding affinity was observed in the same variants of either
didomain AB or full-length HMGBl.
The binding modes of these variants were quite
symmetric about the platinum cross-link, similar to those of domain B, suggesting that domain B
but not domain A interacts with the damaged region of DNA. Based on the slight decrease of
binding affinity in these variants, it appears that domain A strengthens the binding interaction
either by weakly binding to some other site of the DNA or by electrostatically interacting with
the negatively charged phosphate backbone.
1.4.4. Function of the C-terminal Acidic Tail
The C-terminal acidic tail is an unusual motif observed in HMGB 1, 2, 3 and a few other
HMG proteins. This tail is highly negatively charged at physiological pH and might interact with
the positively charged HMG boxes by electrostatic interactions and impede DNA binding. 64
Such competitive binding is believed to increase the structure-specific binding selectivity of
HMGB proteins. As a result, the binding affinity of full-length HMGB 1 to DNA is much weaker
than that of a variant lacking the tail.88 ~90 An NMR mapping study of HMGB1 with truncated
tails showed that the tail binds to both HMG boxes and that shortening the tail diminishes this
interaction, especially with domain A, which is farther away from the C-terminus. 91 In addition,
the acidic tail participates in interactions with other proteins having positively charged motifs,
such as the histone protein H1 .92
35
An in vitro repair study using CFEs revealed that HMGB proteins other than HMGB 1
can inhibit the repair of DNA having a 1,2-intrastrand d(GpG) platinum cross-link similarly to
HMGB 1. Proteins with stronger binding affinities displayed better repair inhibition than HMGB 1
at equimolar concentrations, suggesting that the binding affinity is directly correlated with the
repair-shielding capability of the protein.40 9, 3 The relative inhibition efficiency of HMGB 1 versus
HMGB4, the homologue lacking an acidic tail, will be reported in Chapter 4.
For repair of damaged nucleosomal DNA, however, the tail-truncated HMGB1 displays
much weaker repair inhibition than wild type HMGB1. This result suggests that the acidic tail
may facilitate binding of the protein to the lesion by disrupting the electrostatic interactions with
histone proteins comprising the octamer core of the nucleosome. 94 There is, however, no direct
evidence that the tail of HMGB 1 is essential for intracellular repair.
1.5. Cellular Studies
Many studies of the influence of HMGB proteins on cisplatin cytotoxicity support the
repair-shielding hypothesis. Overexpression of HMGB 1 induced by estrogen treatment of MCF7
cells results in an increase in cisplatin cytotoxicity. 95 Transient expression of other HMGB
proteins such as HMGB2 or testis-specific HMG (tsHMG) from transfected plasmids also
similarly sensitizes cells to cisplatin. 96' 97 The expression level of HMGB1 appears to correlate
positively with the cytotoxicity of platinum-based anticancer compounds. 98 Not all studies,
however, support the hypothesis. There is no difference in cisplatin cytotoxicity for HMGB1+/+
and HMGB~' mouse embryonic fibroblast (MEF) cells. 99 A decrease in HMGB1 expression in
various cancer cell lines by using short hairpin RNAs did not yield consistent results. Some
knockdown cells were sensitized to cisplatin, whereas others were not. In a few cases,
36
knockdown even made the cells slightly resistant to cisplatin (D.Xu and S.J.Lippard, unpublished
observation).
These conflicting results in cultured-cell studies suggest that, unlike simple in vitro repair
systems, the intracellular milieu has cell-type dependent factors that can regulate the interaction
of HMGB proteins to cisplatin-modified DNA. One plausible such factor is the intracellular
redox potential. The intracellular redox potential is meticulously regulated by multiple redox
components, such as reduced (GSH) and oxidized (GS-SG) glutathione. 100' 101 HMGB1 and its
homologues, HMGB2 and HMGB3, have a conserved pair of cysteines in domain A that can
form disulfide bonds. In Chapter 3, the redox-dependent binding of HMGB1 domain A and fulllength HMGB1 will be described together with a discussion of its potential biological
importance.
Another plausible cause of the discrepancy between in vitro repair studies and those
carried out in cells is that the major sub-pathway of NER in the two systems could be quite
different. As described above, the repair assay system using CFEs and linear DNA without a
transcription promoter predominantly reflects GGR, for almost no TCR can occur under these
conditions. Therefore, if the TCR sub-pathway significantly contributes to repair of platinum
damage, there should be large differences between what has been previously reported for in vitro
repair assays and cellular repair. An attempt to evaluate this hypothesis, using a new method that
can easily measure repair of the cisplatin-DNA adducts in live cells, will be described in
Chapter 2.
1.6. Functions of HMGB1 in Processes Other than Repair
37
Necrotic
Cells
Z1000~
Plasma Membrane
Gay
Overepr
sepsis
Metastasis
MAPKINF-IBActivatio
Promt MMR
enhancer
NER Inhibition
Activate HDR
+.*
p53-DNA Binding
p53 Activation
Activate NHEJ
Nulcear Membrane
Figure 1.7. Cellular functions of HMGB1 presumably related to cisplatin-triggered cell death. HMGB1 can influence
many damage-triggered mechanisms as either an intranuclear interaction partner for DNA/proteins or an extracellular
cytokine.
HMGB proteins play essential roles in many important cellular processes by interacting
with DNA and other proteins. 4 6 s Some of these processes are important in cancer biology.'
Thus far we have focused on the repair-shielding model, based on the structure-specific binding
preference of HMGB proteins. There are, however, other natural functions of HMGB proteins
that are involved in the antitumor activity of various types of drugs regardless of their DNA
distorting capabilities. Because cell death is an ultimate result caused by multiple cisplatintriggered processes, it is important to examine the putative influence of HMGB proteins on each
of these. In this section, the interactions of HMGB with other biomolecules and the possible
consequences are briefly described. We focus on the most well-investigated HMGB protein,
38
HMGB1. Not only does it play many roles in the nucleus of normal cells, but it can also function
as a cytokine when released from cells under special conditions. The intracellular and
extracellular functions of HMGB 1 are delineated in the following two sections.
1.6.1. Interaction of HMGB1 with Intracellular Proteins
HMGB1 interacts with various partner proteins in the cells, many of which play their
own roles in the molecular mechanisms of antineoplastic compounds.io2 It can activate a
damage-triggered apoptotic signal that arises when cells are treated with cisplatin. Not only
cisplatin, but also several other antimetabolite drugs, such as 5-fluorouracil, cytarabin, or
mercaptopurine, can trigger an apoptotic response that is mediated by HMGB 1. It is noteworthy
that these drugs do not induce a severe distortion in DNA, and hence we can assume that this
function of HMGB1 is not relevant to the binding specificity. An RNAi study of multiple
proteins revealed that glyceraldehyde 3-phosphate dehydrogenase (GAPDH), protein disulfide
isomerase family member 3 (PDIA3), heat shock protein A8 (HSPA8), and HMGB2 associate
with HMGB1 in this process. 10410 5 The resulting multi-protein assembly activates p53 by
promoting the phosphorylation of serine residues. As a consequence, p53 functions as a
transcription activator that promotes the expression of proteins involved in reactions occurring at
downstream of the damage-triggered apoptotic cascade.
06 107
'
Several in vitro studies revealed
that HMGB1 regulates the binding of p53 to its target promoter sequence, probably by
optimizing DNA structure for binding.108-111 Similar DNA-binding facilitation by HMGB1 is
observed in the case of the p53 homologues, such as p73.
12
Although HMGB1 is generally expressed in most cell types, its interactions with specific
proteins suggest that it has functions specific to a given cell. The human HMGB1 gene has two
39
estrogen responsive elements (ERE), which explains the overexpression of HMGB 1 when
estrogen-receptor (ER) positive cells are treated with estrogen. 95 Treatment of MCF7 cells with
estrogen doubled the expression levels of HMGB 1 and increased the cisplatin sensitivity by a
factor of 4.95 In addition, HMGB 1 facilitates the binding of ER to ERE.'11 '
4
Given that a
significant fraction of breast tumors overexpress the ER, this bi-directional interaction between
ER and HMGB 1 implies a unique role of HMGB 1 in breast cancer and its treatment. 95
HMGB1 also interacts with several other nuclear proteins that are involved in cell death
mechanisms, either facilitating or inhibiting cell death pathways depending on the function of its
binding partners. For instance, HMGB1 increases the ability of retinoblastoma (RB) to induce
GI arrest or apoptosis following drug treatment.115 In some colon tumors, enhanced expression
levels of HMGB1 are closely related to the increased expression level of the inhibitor of
apoptosis (IAP) proteins.ii1
HMGB 1 is also a critical factor for the recognition of platinated damage during mismatch
repair (MMR),' 1 7 base excision repair (BER), 118 and double strand break repair (DSBR). 119 Even
though most of the evidence of HMGB1 participation in these repair mechanisms has been
obtained in vitro, as for NER, the proposed roles of HMGB 1 in these repair mechanisms must be
considered when interpreting the consequences of altering the expression levels of HMGB1 in
cellular system on the cellular response to cisplatin. Experimental results using HMGB1
knockdown cell lines of relevance to this discussion will be described in detail in Chapter 2.
It is not surprising, given the sequence and structure similarity between HMGB1 and
HMGB2, that HMGB2 can also interact with the cellular proteins mentioned above and induce
similar consequences. Even though it has not been definitively shown that HMGB4 can interact
40
with p53 and other proteins in the same manner as HMGB1, it is plausible that this testisrestricted protein, or HMGB2 which is overexpressed in testis, 76 is at least partially responsible
for the p53-mediated hypersensitivity of testicular cancer to cisplatin or other chemotherapeutic
reagents.
1.6.2. HMGB1, an Extracellular Damage Associated Molecular Pattern (DAMP)
HMGB1 is localized in normal cell nuclei, but it can be secreted to the extracellular
matrix following activation by macrophages in the event of severe inflammation.120 ,21 In
addition to such active release, it is passively released during necrotic, but not in apoptotic, cell
death.12 Released HMGB 1 has several post-translation modifications, including phosphorylation
of specific residues and intramolecular disulfide bond formation.123,124 The three cysteine
residues in HMGB1 released from necrotic cells are predominantly oxidized, which is assumed
to promote release from the nucleus by lowering DNA binding affinity. Extracellular HMGB1
can function as a cytokine, interacting with several trans-membrane proteins such as the receptor
for advanced glycation end products (RAGE) and toll-like receptors (TLRs).125
Interaction between such receptors and extracellular HMGB1 can activate several
downstream pathways. In particular, binding of HMGB1 to RAGE activates intracellular
apoptosis-triggering processes like the nuclear factor-KB (NF- KB) signaling pathway and the
mitogen-activated proteins kinase (MAPK) pathways. 12 6, 127 Therefore, both nuclear and
extracellular HMGB 1 can simultaneously participate in cell death-triggering mechanisms, albeit
through totally different pathways. Extracellular HMGB1 also influences metastasis of
cancers. 128
41
In addition to interacting with receptor proteins on the cell surface, extracellular HMGB 1
can enter cells by endocytosis. A previous in vitro study carried out in cultured cells described
the increase of intracellular levels of HMGB1 when the cells were treated with recombinant rat
HMGB1 (K.R.Barnes, D.Xu, and S.J.Lippard, unpublished observation). The change in the
HMGB 1 level influenced the cytotoxicity of cisplatin and trans-DDPas a result of the combined
effects of interaction with trans-membrane receptors and binding to intracellular proteins and
DNA after internalization to the nucleus. It remains to be determined whether HMGB1 released
from cells, which is post-translationally modified, can interact with intranuclear factors in the
same manner as intracellular or recombinant HMGB 1.
1.7. Testicular Cancer, Cisplatin, and HMGB Proteins
1.7.1. Cisplatin Hypersensitivity of Testicular Germ Cell Tumors (TGCTs)
There are some common cancers such as prostate, breast, lung, and colon/rectum cancers
that comprise more than half of the total cases in the US. Compared to these, testicular cancer is
not common, only accounting for about 1% of cancer in males. Despite its rarity, there are
several reasons that this particular type of cancer has received much attention. It most frequently
occurs in young males between the ages of 15 and 35,129 whereas the incidence of most other
cancers is strongly correlated with aging. Also, unlike the frequent cancers mentioned above,
which are of decreasing incidence nowadays, the incidence of testicular cancer is continuously
increasing.130 Moreover, testicular cancer has very distinctive features in terms of its response to
cancer treatment and its biochemistry. It is generally sensitive to chemotherapy, especially to
cisplatin-based treatments. A clinical trial carried out on TGCT patients in the early days of
cisplatin chemotherapy revealed a dramatic increase in cure rate, from 5% to 60%.m' It is
42
noteworthy that cisplatin chemotherapy has cured a significant fraction of metastatic TGCT
cases, whereas other types of metastatic cancers are generally considered incurable.13 2 Based on
this distinctive
feature,
identifying the biochemical
mechanisms
that
account
for the
hypersensitivity of TGCT would permit a better understanding of TGCT and provide clues for
the possible improvement in the treatment of other types of cancer.
There are several hypotheses that explain the cisplatin hypersensitivity of TGCT. The
expression profiles of germ cells are quite different from those of other cells, and many of the
hypotheses focus on the putative functions of proteins with unique expression profiles in TGCT.
For example, most TGCTs express wild type p53 at unusually high levels whereas a considerable
fraction of other cancer cells express mutated p53. 107 p 5 3 is a tumor repressor protein and a key
factor in damage-triggered apoptosis. It is well known from several molecular biology studies
that conservation of wild type p53 is highly correlated with the efficiency of chemotherapeutic
agents. 3 3 TGCTs also have relatively low expression levels of some repair proteins, such as
XPA and ERCC1/XPF. 13 4 ,1 3' The low intracellular concentration of glutathione observed in some
case of TGCT suggests that the detoxification of cisplatin by glutathione in TGCT is not as
efficient as in other tumors. This low glutathione level is especially interesting considering that
significant fractions of platinum-based anticancer drugs are deactivated by biological thiol
molecules before they reach the DNA.1 36
1.7.2. HMGB Proteins in TGCT
Unlike HMGB 1, which is abundant in most tissues, some HMGB proteins have tissue or
organ-specific expression patterns. Like HMGB 1, HMGB2 is expressed in various types of cells,
but the expression levels are especially high in lymphoid organs and testes.7 6'13 7 An in vivo study
43
using HMGB2 knockout mice showed that the high expression level of HMGB2 in testicular
tissue is relevant to its essential role in germ cell differentiation. HMGB2 and HMGB1 are
highly similar in amino acid sequence, except for a shorter C-terminal acidic tail of the former.
Given that overexpression of HMGB2 in non-testis cells can substantially sensitize cells to
cisplatin, 96 HMGB2 might participate in the mechanism of action of cisplatin and be one of the
factors causing the cisplatin hypersensitivity of TGCT.
HMGB2 is not the only HMGB protein with testis-specific expression: For instance,
HMG-box transcription factor 1 (HBP 1) and SRY are also present at high concentrations in testis
cells and this expression pattern is due to the unique gene regulation that needs for the germ cell
differentiation. 138,139 Murine tsHMG also shows promising properties as a repair-shielding
protein.93,97 This protein is a variant of the mitochondrial protein mtTFA and expression in
nucleus is only observed in testis.140 Not only did this protein show relatively high binding
affinity for, and repair inhibition of, cisplatin-DNA adducts in vitro, 93 but transfection of tsHMG
expression vector into HeLa cells also sensitizes them to cisplatin. 97 Even though there is no such
variant in human mtTFA,14 1 these studies of tsHMG strengthen the hypothesis that testis-specific
HMGB proteins contribute to the biochemistry that underlies the efficacy of cisplatin treatment
for TGCT.
Recently, HMGB4, a new member of mammalian HMGB proteins that shares
considerable similarity with the three other canonical HMGB proteins was identified.77 Despite a
lack of structural information, its sequence indicates that it might favorably interact with
cisplatin-platinated DNA. The characterization and expression profiles of this protein will be
described in Chapter 4.
44
1.8. Summary
Cisplatin is an antineoplastic compound used in the treatment of a wide variety of cancers.
The binding of activated cisplatin to DNA produces various adducts that trigger either repair of
damage or death in cancer cells. High mobility group box proteins, which mainly function as
architectural transcription factors, were the first identified cisplatin damage recognition proteins
and shown to function as a repair shield of cisplatin-DNA adducts. Testicular germ cell tumors
have some testis-specific HMGB proteins that might be related to their cisplatin sensitivity.
Various in vivo and in vitro studies indicate that altering the expression level of HMGB proteins
can modulate the cytotoxicity of cisplatin in a cell-type dependent manner. Other studies show
results that conflict with the repair-shielding hypothesis and raise questions about a role for
HMGB proteins in the molecular mechanism of cisplatin. Confirmation of a role for HMGB
proteins in the mechanism of action of cisplatin, and identifying intracellular regulating factors
that account for cell-dependent trends in previous studies, are important objectives that inspired
the work in this thesis.
45
1.9. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Rosenberg, B.; Vancamp, L.; Krigas, T. Nature 1965, 205, 698-699.
Jung, Y.; Lippard, S. J. Chem. Rev. 2007, 107, 1387-1407.
Orvig, C.; Abrams, M. J. Chem. Rev. 1999, 99, 2201-2204.
Hannon, M. J. PureAppl. Chem. 2007, 79, 2243-2261.
Loehrer, P. J.; Einhorn, L. H. Ann. Intern. Med. 1984, 100, 704-713.
Wong, E.; Giandomenico, C. M. Chem. Rev. 1999, 99, 2451-2466.
Galluzzi, L.; Senovilla, L.; Vitale, I.; Michels, J.; Martins, I.; Kepp, 0.; Castedo, M.;
Kroemer, G. Oncogene 2012, 31, 1869-1883.
Fuertes, M. A.; Alonso, C.; Perez, J. M. Chem. Rev. 2003, 103, 645-662.
(9)
Shlebak, A. A.; Clark, P. I.; Green, J. A. Cancer Chemother. Pharmacol.1995, 35, 349351.
(10)
(11)
(12)
Knox, R. J.; Friedlos, F.; Lydall, D. A.; Roberts, J. J. CancerRes. 1986, 46, 1972-1979.
Canetta, R.; Rozencweig, M.; Carter, S. K. Cancer Treat. Rev. 1985, 12 Suppl A, 125-136.
Calvert, A. H.; Harland, S. J.; Newell, D. R.; Siddik, Z. H.; Jones, A. C.; McElwain, T. J.;
Raju, S.; Wiltshaw, E.; Smith, I. E.; Baker, J. M.; Peckham, M. J.; Harrap, K. R. Cancer
(13)
(14)
(15)
Bokemeyer, C.; Kohrmann, 0.; Tischler, J.; Weissbach, L.; Rath, U.; Haupt, A.;
Schoffski, P.; Harstrick, A.; Schmoll, H. J.Ann. Oncol. 1996, 7, 1015-102 1.
Ehrsson, H.; Wallin, I.; Yachnin, J. Med. Oncol. 2002, 19, 261-265.
Wheate, N. J.; Walker, S.; Craig, G. E.; Oun, R. Dalton Trans. 2010, 39, 8113-8127.
(16)
Dhar, S.; Lippard, S. J. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 22199-22204.
(17)
(18)
Hollis, L. S.; Amundsen, A. R.; Stem, E. W. J.Med. Chem. 1989, 32, 128-136.
Hall, M. D.; Mellor, H. R.; Callaghan, R.; Hambley, T. W. J. Med. Chem. 2007, 50,
3403-3411.
Barnes, K. R.; Kutikov, A.; Lippard, S. J. Chem. Biol. 2004, 11, 557-564.
Bauer, C.; Peleg-Shulman, T.; Gibson, D.; Wang, A. H. J. Eur. J. Biochem. 1998, 256,
253-260.
Lovejoy, K. S.; Serova, M.; Bieche, I.; Emami, S.; D'Incalci, M.; Broggini, M.; Erba, E.;
Gespach, C.; Cvitkovic, E.; Faivre, S.; Raymond, E.; Lippard, S. J. Mol. Cancer Ther.
2011,10, 1709-1719.
Chemother. Pharmacol.1982, 9, 140-147.
(19)
(20)
(21)
(22)
(23)
Kelland, L. Nat. Rev. Cancer2007, 7, 573-584.
Siddik, Z. H. Oncogene 2003, 22, 7265-7279.
(24)
Jamieson, E. R.; Lippard, S. J. Chem. Rev. 1999, 99, 2467-2498.
(25)
Eastman, A. Pharmacol.Ther. 1987, 34,155-166.
(26)
(27)
(28)
Zlatanova, J.; Yaneva, J.; Leuba, S. H. FASEB J.1998, 12, 791-799.
Fuertes, M. A.; Castilla, J.; Alonso, C.; Perez, J. M. Curr.Med. Chem. 2003, 10, 257-266.
Bruhn, S. L.; Pil, P. M.; Essigmann, J. M.; Housman, D. E.; Lippard, S. J. Proc. Natl.
(29)
Acad. Sci. U. S. A. 1992, 89, 2307-2311.
Miller, S. E.; House, D. A. InorganicaChimica Acta 1990, 173, 53-60.
(30)
Bancroft, D. P.; Lepre, C. A.; Lippard, S. J.J.Am. Chem. Soc. 1990, 112, 6860-6871.
(31)
Kartalou, M.; Essigmann, J. M. Mutat. Res. Fundam. Mol. Mech. Mutagen. 2001, 478, 1-
21.
46
Fichtinger-Schepman, A. M. J.; van der Veer, J. L.; den Hartog, J. H. J.; Lohman, P. H.
M.; Reedijk, J. Biochemistry 1985, 24, 707-713.
Gelasco, A.; Lippard, S. J. Biochemistry 1998, 37, 9230-9239.
(33)
(34)
van Garderen, C. J.; van Houte, L. P. Eur. J. Biochem. 1994, 225, 1169-1179.
(35)
Takahara, P. M.; Rosenzweig, A. C.; Frederick, C. A.; Lippard, S. J. Nature 1995, 377,
649-52.
(36) Yang, D.; van Boom, S. S.; Reedijk, J.; van Boom, J. H.; Wang, A. H. Biochemistry 1995,
34, 12912-12920.
(37)
Huang, H.; Zhu, L.; Reid, B. R.; Drobny, G. P.; Hopkins, P. B. Science 1995, 270, 18421845.
Lovejoy, K. S.; Todd, R. C.; Zhang, S.; McCormick, M. S.; D'Aquino, J. A.; Reardon, J.
(38)
T.; Sancar, A.; Giacomini, K. M.; Lippard, S. J. Proc. Natl. Acad Sci. U. S. A. 2008, 105,
8902-8907.
(39)
Szymkowski, D. E.; Yarema, K.; Essigmann, J. M.; Lippard, S. J.; Wood, R. D. Proc.
Natl. Acad Sci. U. S. A. 1992, 89, 10772-10776.
(40) Huang, J. C.; Zamble, D. B.; Reardon, J. T.; Lippard, S. J.; Sancar, A. Proc. Natl. Acad
Sci. U. S. A. 1994, 91, 10394-10398.
(41)
Zamble, D. B.; Mu, D.; Reardon, J. T.; Sancar, A.; Lippard, S. J. Biochemistry 1996, 35,
10004-10013.
Bessho, T.; Mu, D.; Sancar, A. Mol. Cell. Biol. 1997, 17, 6822-6830.
(42)
(43)
Furuta, T.; Ueda, T.; Aune, G.; Sarasin, A.; Kraemer, K. H.; Pommier, Y. Cancer Res.
2002, 62, 4899-4902.
(44) Lehmann, A. R. Biochimie 2003, 85, 1101-1111.
(45)
Costa, R. M.; Chigancas, V.; Galhardo Rda, S.; Carvalho, H.; Menck, C. F. Biochimie
2003, 85, 1083-1099.
(46) Sugasawa, K.; Ng, J. M.; Masutani, C.; Iwai, S.; van der Spek, P. J.; Eker, A. P.; Hanaoka,
F.; Bootsma, D.; Hoeijmakers, J. H. Mol. Cell 1998, 2, 223-232.
(47)
Svejstrup, J. Q. Nat. Rev. Mol. Cell. Biol. 2002, 3, 21-9.
Tantin, D.; Kansal, A.; Carey, M. Mol. Cell. Biol. 1997, 17, 6803-6814.
(48)
(49)
Matsunaga, T.; Mu, D.; Park, C. H.; Reardon, J. T.; Sancar, A. J. Biol. Chem. 1995, 270,
20862-20869.
(50) Shivji, M. K.; Podust, V. N.; Hubscher, U.; Wood, R. D. Biochemistry 1995, 34, 50115017.
(51)
Mu, D.; Sancar, A. Prog.Nucleic Acid Res. Mol. Biol. 1997, 56, 63-8 1.
(52)
Selby, C. P.; Sancar, A. Methods Enzymol. 2003, 371, 300-324.
(53)
Ciccarelli, R. B.; Solomon, M. J.; Varshavsky, A.; Lippard, S. J. Biochemistry 1985, 24,
7533-7540.
(54) Roberts, J. J.; Friedlos, F. CancerRes. 1987, 47, 3 1-36.
(55)
Hey, T.; Lipps, G.; Sugasawa, K.; Iwai, S.; Hanaoka, F.; Krauss, G. Biochemistry 2002,
41, 6583-6587.
(56)
Jung, Y.; Mikata, Y.; Lippard, S. J. J. Biol. Chem. 2001, 276, 43589-43596.
(57)
Patrick, S. M.; Turchi, J. J. J.Biol. Chem. 1999, 274, 14972-14978.
(58)
Mello, J. A.; Acharya, S.; Fishel, R.; Essigmann, J. M. Chem. Biol. 1996, 3, 579-589.
(59)
Toney, J. H.; Donahue, B. A.; Kellett, P. J.; Bruhn, S. L.; Essigmann, J. M.; Lippard, S. J.
Proc.Natl. Acad. Sci. U. S. A. 1989, 86, 8328-8332.
(60)
Goodwin, G. H.; Johns, E. W. Eur. J. Biochem. 1973, 40, 215-219.
(32)
47
(61)
Hamada, H.; Bustin, M. Biochemistry 1985, 24, 1428-1433.
(62)
Bustin, M. Trends Biochem. Sci. 2001, 26, 152-153.
(63)
Thomas, J. 0.; Travers, A. A. Trends Biochem. Sci. 2001, 26, 167-174.
(64)
(65)
Stros, M. Biochim. Biophys. Acta 2010, 1799, 101-113.
Stros, M.; Launholt, D.; Grasser, K. D. Cell. Mol. Life Sci. 2007, 64, 2590-2606.
(66)
(67)
Bianchi, M. E.; Beltrame, M.; Paonessa, G. Science 1989, 243, 1056-1059.
Jaouen, S.; de Koning, L.; Gaillard, C.; Muselikova-Polanska, E.; Stros, M.; Strauss, F. J.
Mol. Biol. 2005, 353, 822-837.
(68)
(69)
(70)
Stros, M.; Polanska, E.; Struncova, S.; Pospisilova, S. Nucleic Acids Res. 2009, 37, 20702086.
Pil, P. M.; Lippard, S. J. Science 1992, 256, 234-237.
Billings, P. C.; Davis, R. J.; Engelsberg, B. N.; Skov, K. A.; Hughes, E. N. Biochem.
Biophys. Res. Commun. 1992, 188, 1286-1294.
(71)
(72)
Kasparkova, J.; Delalande, 0.; Stros, M.; Elizondo-Riojas, M. A.; Vojtiskova, M.;
Kozelka, J.; Brabec, V. Biochemistry 2003, 42, 1234-1244.
Guggenheim, E. R.; Xu, D.; Zhang, C. X.; Chang, P. V.; Lippard, S. J. Chembiochem
2009, 10, 141-157.
(73)
Zhu, G.; Lippard, S. J. Biochemistry 2009, 48,4916-4925.
(74)
(75)
Brown, S. J.; Kellett, P. J.; Lippard, S. J. Science 1993, 261, 603-605.
Cepeda, V.; Fuertes, M. A.; Castilla, J.; Alonso, C.; Quevedo, C.; Perez, J. M. Anticancer
Agents Med. Chem. 2007, 7, 3-18.
(76)
(77)
Ronfani, L.; Ferraguti, M.; Croci, L.; Ovitt, C. E.; Scholer, H. R.; Consalez, G. G.;
Bianchi, M. E. Development 2001, 128, 1265-1273.
Catena, R.; Escoffier, E.; Caron, C.; Khochbin, S.; Martianov, I.; Davidson, I. Biol.
Reprod. 2009, 80, 358-366.
(78)
Hardman, C. H.; Broadhurst, R. W.; Raine, A. R. C.; Grasser, K. D.; Thomas, J. 0.; Laue,
E. D. Biochemistry 1995, 34, 16596-16607.
(79)
(80)
(81)
(82)
(83)
(84)
(85)
(86)
(87)
(88)
(89)
(90)
(91)
(92)
Weir, H. M.; Kraulis, P. J.; Hill, C. S.; Raine, A. R.; Laue, E. D.; Thomas, J. 0. EMBO J.
1993, 12, 1311-1319.
Thomas, J. 0. Biochem. Soc. Trans. 2001, 29, 395-401.
Dunham, S. U.; Lippard, S. J. Biochemistry 1997, 36, 11428-11436.
Chow, C. S.; Whitehead, J. P.; Lippard, S. J. Biochemistry 1994, 33, 15124-15130.
Webb, M.; Thomas, J. 0. J. Mol. Biol. 1999, 294, 373-387.
Ohndorf, U. M.; Rould, M. A.; He, Q.; Pabo, C. 0.; Lippard, S. J. Nature 1999, 399, 708712.
McCauley, M. J.; Zimmerman, J.; Maher, L. J., 3rd; Williams, M. C. J. Mol. Biol. 2007,
374, 993-1004.
He, Q.; Ohndorf, U. M.; Lippard, S. J. Biochemistry 2000, 39, 14426-14435.
Murphy, F. V. t.; Sweet, R. M.; Churchill, M. E. EMBO J. 1999, 18, 6610-6618.
Jung, Y.; Lippard, S. J.Biochemistry 2003, 42, 2664-2671.
Stros, M.; Stokrova, J.; Thomas, J. 0. Nucleic Acids Res. 1994, 22, 1044-1051.
Stott, K.; Watson, M.; Howe, F. S.; Grossmann, J. G.; Thomas, J. 0. J. Mol. Biol. 2010,
403, 706-722.
Watson, M.; Stott, K.; Thomas, J. 0. J. Mol. Biol. 2007, 374, 1286-1297.
Cato, L.; Stott, K.; Watson, M.; Thomas, J. 0. J. Mol. Biol. 2008, 384, 1262-1272.
48
(93)
(94)
(95)
(96)
(97)
(98)
(99)
(100)
(101)
(102)
(103)
(104)
(105)
(106)
(107)
(108)
(109)
(110)
(111)
(112)
(113)
(114)
(115)
(116)
(117)
(118)
(119)
(120)
Ohndorf, U. M.; Whitehead, J. P.; Raju, N. L.; Lippard, S. J. Biochemistry 1997, 36,
14807-14815.
Ugrinova, I.; Zlateva, S.; Pashev, I. G.; Pasheva, E. A. Int. J. Biochem. Cell Biol. 2009,
41, 1556-1562.
He, Q.; Liang, C. H.; Lippard, S. J. Proc.Natl. Acad. Sci. U. S. A. 2000, 97, 5768-5772.
Arioka, H.; Nishio, K.; Ishida, T.; Fukumoto, H.; Fukuoka, K.; Nomoto, T.; Kurokawa,
H.; Yokote, H.; Abe, S.; Saijo, N. Jpn. J. CancerRes. 1999, 90, 108-115.
Zamble, D. B.; Mikata, Y.; Eng, C. H.; Sandman, K. E.; Lippard, S. J. J. Inorg. Biochem.
2002, 91, 451-462.
Robert, J.; Laurand, A.; Meynard, D.; Le Morvan, V. In 10th InternationalSymposium on
Platinum CoordinationCompounds in Cancer Chemotherapy Verona, Italy, 2007.
Wei, M.; Burenkova, 0.; Lippard, S. J. J. Biol. Chem. 2003, 278, 1769-1773.
Schafer, F. Q.; Buettner, G. R. Free Radic.Biol.Med. 2001, 30, 1191-1212.
Hogg, P. J. Trends Biochem. Sci. 2003, 28, 210-214.
Tang, D.; Kang, R.; Zeh, H. J., 3rd; Lotze, M. T. Biochim. Biophys. Acta 2010, 1799,
131-140.
Lotze, M. T.; DeMarco, R. A. Curr. Opin. Investig. Drugs 2003, 4, 1405-1409.
Krynetskaia, N.; Xie, H.; Vucetic, S.; Obradovic, Z.; Krynetskiy, E. Mol. Pharmacol.
2008, 73, 260-269.
Jayaraman, L.; Moorthy, N. C.; Murthy, K. G.; Manley, J. L.; Bustin, M.; Prives, C.
Genes Dev. 1998, 12, 462-472.
May, P.; May, E. Oncogene 1999, 18, 7621-7636.
Gutekunst, M.; Oren, M.; Weilbacher, A.; Dengler, M. A.; Markwardt, C.; Thomale, J.;
Aulitzky, W. E.; van der Kuip, H. PLoS One 2011, 6, e19198.
McKinney, K.; Prives, C. Mol. Cell. Biol. 2002, 22, 6797-6808.
Imamura, T.; Izumi, H.; Nagatani, G.; Ise, T.; Nomoto, M.; Iwamoto, Y.; Kohno, K. J.
Biol. Chem. 2001, 276, 7534-7540.
Banerjee, S.; Kundu, T. K. Nucleic Acids Res. 2003, 31, 3236-3247.
Zamble, D. B.; Jacks, T.; Lippard, S. J. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 61636168.
Stros, M.; Ozaki, T.; Bacikova, A.; Kageyama, H.; Nakagawara, A. J. Biol. Chem. 2002,
277, 7157-7164.
Boonyaratanakornkit, V.; Melvin, V.; Prendergast, P.; Altmann, M.; Ronfani, L.; Bianchi,
M. E.; Taraseviciene, L.; Nordeen, S. K.; Allegretto, E. A.; Edwards, D. P. Mol. Cell.
Biol. 1998, 18, 4471-4487.
Das, D.; Peterson, R. C.; Scovell, W. M. Mol. Endocrinol.2004, 18, 2616-2632.
Jiao, Y.; Wang, H. C.; Fan, S. J. Acta Pharmacol.Sin. 2007, 28, 1957-1967.
Volp, K.; Brezniceanu, M. L.; Bosser, S.; Brabletz, T.; Kirchner, T.; Gottel, D.; Joos, S.;
Zornig, M. Gut 2006, 55, 234-242.
Yuan, F.; Gu, L.; Guo, S.; Wang, C.; Li, G. M. J. Biol. Chem. 2004, 279, 20935-20940.
Prasad, R.; Liu, Y.; Deterding, L. J.; Poltoratsky, V. P.; Kedar, P. S.; Horton, J. K.;
Kanno, S.; Asagoshi, K.; Hou, E. W.; Khodyreva, S. N.; Lavrik, 0. I.; Tomer, K. B.;
Yasui, A.; Wilson, S. H. Mol. Cell 2007, 27, 829-841.
Stros, M.; Cherny, D.; Jovin, T. M. Eur. J. Biochem. 2000, 267, 4088-4097.
Lotze, M. T.; Tracey, K. J. Nat. Rev. Immunol. 2005, 5, 331-342.
49
Bonaldi, T.; Talamo, F.; Scaffidi, P.; Ferrera, D.; Porto, A.; Bachi, A.; Rubartelli, A.;
Agresti, A.; Bianchi, M. E. EMBO J. 2003, 22, 555 1-5560.
(122) Rovere-Querini, P.; Capobianco, A.; Scaffidi, P.; Valentinis, B.; Catalanotti, F.; Giazzon,
M.; Dumitriu, I. E.; Muller, S.; lannacone, M.; Traversari, C.; Bianchi, M. E.; Manfredi,
(121)
A. A. EMBO Rep. 2004, 5, 825-830.
(123) Youn, J. H.; Shin, J. S. J. Immunol. 2006, 177, 7889-7897.
(124) Urbonaviciute, V.; Meister, S.; Fiirnrohr, B. G.; Frey, B.; Gnckel, E.; Schett, G.;
Herrmann, M.; Voll, R. E. Autoimmunity 2009, 42, 305-307.
(125)
(126)
(127)
Bianchi, M. E.; Manfredi, A. A. Immunol. Rev. 2007, 220, 35-46.
Kang, R.; Tang, D.; Lotze, M. T.; Zeh, H. J., 3rd Autophagy 2011, 7, 442-444.
Kang, R.; Livesey, K. M.; Zeh, H. J., 3rd; Lotze, M. T.; Tang, D. Autophagy 2011, 7,
(128)
Campana, L.; Bosurgi, L.; Rovere-Querini, P. Curr. Opin. Immunol. 2008, 20, 518-523.
(129)
Bosl, G. J.; Motzer, R. J. N. Engl. J. Med. 1997, 337, 242-253.
(130)
McGlynn, K. A.; Devesa, S. S.; Sigurdson, A. J.; Brown, L. M.; Tsao, L.; Tarone, R. E.
1256-1258.
Cancer2003, 97, 63-70.
(131) Einhorn, L. H.; Donohue, J. Ann. Intern. Med 1977, 87, 293-298.
(132) Masters, J. R.; Koberle, B. Nat. Rev. Cancer2003, 3, 517-525.
(133) Lowe, S. W.; Ruley, H. E.; Jacks, T.; Housman, D. E. Cell 1993, 74, 957-967.
(134) Usanova, S.; Piee-Staffa, A.; Sied, U.; Thomale, J.; Schneider, A.; Kaina, B.; Koberle, B.
Mol. Cancer2010, 9, 248-258.
Koberle, B.; Masters, J. R.; Hartley, J. A.; Wood, R. D. Curr.Biol. 1999, 9, 273-276.
Sark, M. W.; Timmer-Bosscha, H.; Meijer, C.; Uges, D. R.; Sluiter, W. J.; Peters, W. H.;
Mulder, N. H.; de Vries, E. G. Br. J. Cancer 1995, 71, 684-690.
(137) Muller, S.; Ronfani, L.; Bianchi, M. E. J.Intern. Med. 2004, 255, 332-343.
(138) Smith, J. M.; Bowles, J.; Wilson, M.; Koopman, P. Dev. Dyn. 2004, 230, 366-370.
(139) Hacker, A.; Capel, B.; Goodfellow, P.; Lovell-Badge, R. Development 1995, 121, 16031614.
(140) Alami-Ouahabi, N.; Veilleux, S.; Meistrich, M. L.; Boissonneault, G. Mol. Cell. Biol.
(135)
(136)
1996, 16, 3720-3729.
(141) Larsson, N. G.; Oldfors, A.; Garman, J. D.; Barsh, G. S.; Clayton, D. A. Hum. Mol. Genet.
1997, 6, 185-191.
50
Chapter 2. Effect of HMGB1 on NER of Platinated DNA in Cells
51
2.1. Introduction
Because of their strong and unique binding interactions with cisplatin-damaged DNA,
proteins having a high mobility group (HMG)-box domain are possible regulators in the cellular
response to cisplatin.1 3 After the "repair-shielding hypothesis" was suggested based on in vitro
repair assays, which revealed repair inhibition by HMGB1 of a cisplatin 1,2-intrastrand d(GpG)
cross-linked DNA substrate in cell free extracts (CFEs),4 attempts were made to examine the
hypothesis in live cells and, ultimately, in vivo.
Some experiments in cultured cells showed promising results in agreement with the
repair-shielding model.5 6' Other studies indicated a negligible, or even negative, correlation
between HMGB 1 expression levels and cisplatin sensitivity. No noticeable difference in cisplatin
cytotoxicity was observed for HMGBl1' mouse embryonic fibroblast (MEF) cells versus wild
8
type HMGB1'* cells.7 HMGB1 knockdown in osteosarcoma cells promoted cisplatin resistance.
Such conflicting results can be attributed to the use of a simple in vitro model, which fails to
appropriately replicate a complex intracellular environment. The main difference between CFE
and intracellular systems is that the former measures the results of a specific repair mechanism,
whereas the latter reports the outcome of numerous cellular processes, several of which may
involve HMGB 1.
Nucleotide excision repair (NER) is generally accepted as the major mechanism by which
cisplatin 1,2-intrastrand adducts are repaired. NER involves assembly of a multi-protein complex
around the damaged site. Many of the proteins participating in NER are mutated in xeroderma
pigmentosum (XP) and Cockayne syndrome (CS) patients, and cells derived from such patients
are highly sensitive to cisplatin.9 Depending on how the damage is recognized, NER is divided
52
into global genomic (GGR) and transcription coupled (TCR) repair. The GGR pathway is
initiated following DNA damage recognition by the xeroderma pigmentosum group C
(XPC)/human homologue of RAD23B (hHR23B) complex, which induces a distortion of the
damaged site and recruits damaged-DNA binding (DDB) factors.10"' On the other hand, the
stalling of RNA polymerase II (RNA pol II) by DNA damage is assumed to trigger the TCR
pathway.' 2 Cockayne syndrome A (CSA) and B (CSB) participate exclusively in the initiation
step of TCR, and not that of GGR. This process is XPC-independent.
3
After recognition and
formation of the initial "sensing" complexes, the damage is repaired by partial unwinding of
DNA, cleavage of the damaged strand on both sides of the lesion, synthesis of gap-filling DNA,
and ligation of the final products.14
The limitation of analysis using CFE is that repair is predominantly accomplished by the
GGR pathway; TCR is negligible under these conditions, 5 because transcription activity is
significantly lower in CFE than in live cells. Furthermore, when a linear substrate lacking a
transcription promoter is used, as in previous HMGB1 repair inhibition studies, GGR is the only
NER process in operation. Methods that can observe the repair kinetics in live cells are needed in
order to understand repair of cisplatin damage and test the repair-shielding hypothesis.
Previously, Wee Han Ang developed a new method to measure the intracellular repair of
platinated DNA (Figure 2.1).16 In this assay, the Gaussia Luciferase (GLuc) expression vector is
used as a transcription probe. Upon expression, GLuc is secreted into the cell medium making it
possible to measure its time-dependent expression. Once the platinated plasmid probe is
transfected into the cells, the cellular repair processes can be monitored by quantifying GLuc
expression. This method has two main advantages; i) The time-dependent repair data can be
53
gained from the live cells without cell lysis. ii) Repair data for different platinum-adducts can be
obtained by site-specifically incorporating a compound of interest into the probe.
Nulcear Membrane
Plasma
Membrane
C)
Transfecdon
4raJSK4r
pGLuc pobe
with Pt-damages
r
Repair
(4)
Transcripion
= Platinated adduct
=GLuc
mRNA
coding region
Substrate-0
Translation
Oxidation
Figure 2.1. Diagram of GLuc transcription assay using platinated substrates.
Here we describe studies of the cellular response to cisplatin, with a focus on repair of PtDNA. Cells transfected with an HMGB 1 short hairpin RNA (shRNA) plasmid, to reduce the
expression level of HMGB 1, were used to investigate the influence of HMGB 1 on intracellular
repair kinetics and cisplatin cytotoxicity. The influence of HMGB 1 on cellular response to
cisplatin was examined by an MTT assay and a GLuc transcription assay in knockdown cell lines.
The MTT assay represents the global response resulting from multiple intracellular mechanisms
that are triggered by cisplatin treatment. The GLuc transcription assay dominantly reflects the
repair of the platinated DNA. Using these assays in conjunction, it is possible to investigate the
role for HMGB 1 in the repair of platinated DNA as well as the overall cisplatin-triggered
apoptosis. We expect that the results of these studies can explain the conflict between the
54
previous HMGB 1 studies conducted in the culture cells and verify the repair-shielding model in
live cells. In addition to work using HMGB 1 knockdown cell lines, the repair of specific types of
cisplatin adducts was investigated in fibroblast cell lines lacking either TCR, GGR, or both, to
estimate the relative contributions of each NER sub-pathway to the repair of each adduct.
2.2. Experimental
2.2.1. Cell Culture
Table 2.1. Fibroblast Cell Lines Used for Repair Studies
Cell ID
Phenotype
Immortalizationa
TCR
GGR
GM00637
Normal
SV40 virus
+
+
GM14930
XPG
SV40 virus
GM08437
XPF
SV40 virus
GM16095
CSB
pSV3gpt
+
+
SV40 virus
XPC
GM15983
aAll cell lines are transformed by either SV40 viral infection or pSV3gpt
plasmid transfection.
HeLa and A549 cells were purchased from ATCC. NER proficient/deficient fibroblast
cell lines were obtained from the Coriell Institute and the deficient sub-pathway of each cell line
is shown in Table 2.1. HeLa, A549, and fibroblast cell lines were grown in Dubecco's modified
eagle medium (DMEM) containing 10% fetal bovine serum (FBS), 100 units/mL of penicillin,
and 100 ptg/mL of streptomycin. HMGB1 knockdown HeLa and A549 were grown in the
medium supplemented with 0.25 ptg/mL and 0.35 pg/mL of puromycin (Invitrogen) respectively.
All cells were incubated at 37 'C under a 5% CO 2 atmosphere.
2.2.2. Knockdown of HMGB1 in HeLa and A549 Cells by Transfection of shRNAExpression Vector
55
To prepare a plasmid that is able to express HMGB1 shRNA, target sequences 1 (5'TTGGTGATGTTGCGAAGAAAC-3') and 2 (5'-GAAGAAGATGAAGATGAAGAA-3'),
with
low cross reactivity to other mRNA, were selected from the RNAi Consortium (TRC) database
of the Broad Institute. The vector pLKO1.TRC (Addgene) has a hU6 promoter for shRNA
expression, a 1.9 kb stuffer replaced by shRNA during construction of the plasmid, and the
puromycin resistance gene. The vector was digested by restriction enzymes Agel and EcoRI at
37 'C for 1 h and purified by agarose gel electrophoresis. The insertion strand was
phosphorylated at the 5' end by T4 PNK (New England Biolabs) at 37 'C for 1 h, and inserted
into the cut vector by overnight reaction with T4 DNA ligase (New England Biolabs) at 16 'C.
Ligated products were transformed into DH5a cells for amplification and then purified with a
QIAGEN HiSpeed* Maxi Kit. Purified plasmids were designated as pLKO1 HMGBl(1) and
pLKO 1 HMGB 1(2) respectively. HeLa or A549 cells were grown in a 60 mm cell culture dishes
until -60% confluent prior to transfection. For the formation of the transfection complex, a 1 pg
amount of each plasmid was mixed with 8 gL of enhancer (QIAGEN) in 150 gL of buffer EC
(QIAGEN) and incubated for 5 min at room temperature, and then 25 gL of effectene (QIAGEN)
was added to the mixed solution. The DNA-effectene solution was incubated for additional 10
min and then added to cells in fresh growth medium. The medium was replaced by puromycinsupplemented medium 48 h after transfection for antibiotic selection. Incubation was continued
until all non-transfected cells died and only puromycin-resistant cells remained. Since a
heterogeneous population of A549 cells did not demonstrate sufficient RNAi efficiency,
homogeneous cloning was carried out by incubating cells at low density until well-isolated
homogeneous clones were produced. Clones were collected using sterile glass cylinders. Cells
were transfected with the pLKO1 vector containing a random insertion sequence (5'-
56
GAGAGGACAAGAGATGTACTT-3') to produce control puromycin resistant cell lines. After
the transfection and selection, cells were continuously grown in the puromycin-supplemented
medium to maintain persistent transfected gene expression.
2.2.3. Western Blot
Cells were grown to 90% confluence in T25 cell culture flasks and collected by
trypsinization followed by centrifugation. After centrifugation, cell pellets were incubated on ice
for 30 min in 4 cell pellet volumes of whole cell extract buffer (20 mM HEPES, pH 7.5, 20%
glycerol, 1% Nonidet P-40, 1 mM MgCl 2 , 0.5 mM EDTA, 1 mM PMSF, 1% protease inhibitor
cocktail (Sigma-Aldrich), 1 mM DTT), and then centrifuged to separate the insoluble pellet from
the protein solution. The protein concentrations of whole cell extracts were measured with
Bradford Assay (Thermo Scientific). Whole cell extracts (10-20 pg) were resolved by SDSPAGE on a 4-20% Tris-HCl Ready-Gel (Bio-Rad) for 35 min at 200 V. After the electrophoresis,
the SDS gel was equilibrated in western transfer buffer (48 mM Tris-HCl, 39 mM glycine, pH
9.2) for 15 min before the proteins were transferred to 0.4 gm PVDF membrane (Pall
corporation) using a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad), for 2 h at 100 V,
0.35 A, 40 W conditions. After the membrane with protein on it was incubated in 5% milk
solution in TBS buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl) to block non-specific antibody
binding, it was incubated with a 1:1000 dilution of rabbit HMGB 1 antibody AB 18256 (Abcam),
for 1 h at room temperature. As internal standards, a 1:1000 dilution of rabbit Actin antibody
A2103 (Sigma-Aldrich) or a 1:200 dilution of rabbit GADPH antibody PA000275 (Syd Lab)
were used simultaneously with the HMGB1 antibody. The membrane was washed with TTBS
solution (TBS + 0.1% Tween 20) and incubated with a 1:4000 dilution of AB6721 goat antirabbit IgG antibody (Abcam) for 1 h at room temperature. After a final wash in TTBS, the blot
57
was exposed to SupersignalV West Pico substrate (Thermo Scientific) for 1 min and exposed to
BioMax film (Kodak Co). For quantitative estimation of RNAi efficiency, whole cell extracts of
HMGB 1 knockdown cells were loaded with serial dilutions of the parental cell extracts.
2.2.4. Immunofluorescence
Parental/control and knockdown HeLa clones in their homogeneous populations were
grown to 30-50% confluence on a glass-bottom cell culture dish. Cells were fixed with 4% paraformaldehyde in 50 mM phosphate buffer (pH 7.5) for 1 h at room temperature, washed with
PBS, and permeabilized with 0.1% Triton X-100 in PBS for 10 min at room temperature. Before
antibody treatment, permeabilized cells were incubated in a 1%v/v goat serum solution in PBS
for 30 min to prevent non-specific binding of antibodies and then incubated in the primary
HMGB1 antibody AB18256 (1:1000 dilution in PBS) at 37 'C for 1 h. After a PBS wash, cells
were treated with secondary antibody alexa Fluoro@ 546 (1:50 dilution in PBS). Finally, cells
were incubated in 4 ng/L of Hoechest for nuclear staining. Antibody-treated cells were visualized
with a fluorescent light microscope.
2.2.5. Cytotoxicity Assay
The colorimetric MTT assay was carried out to compare the cytotoxicity of platinumbased anticancer drugs in each cell line. HMGB 1 knockdown and control cells were seeded on a
96-well plate at cell densities that varied depending on cell line; 1000 cells/well for HeLa and
A549, 800-2000 cells/well for NER proficient/deficient fibroblast cell lines depending on
doubling time. Platinum compounds were dissolved in PBS immediately before drug loading,
and the concentration was measured by atomic absorption spectroscopy. The medium in each
well was replaced by 100 pL of new medium containing various concentrations of platinum
58
compounds. Cells were incubated at 37 'C, 5% CO 2 for an additional 72 h, and then incubated in
100 pL of 0.5 mg/mL MTT solution in DMEM without phenol red for 4 h. Precipitated formazan
was dissolved in 100 [tL of DMSO, and the absorbance at 550 nm was measured in a 96-well
plate reader (Bio-Tek). The IC50, the concentration required to kill 50% of treated cells, was
calculated by interpolation of the experimental data.
2.2.6. Preparation of Platinated pGLuc Plasmids
Transcription assays were carried out using both globally and site-specifically platinated
GLuc expression plasmids. The pGLuc vector for global platination is derived from pCMVGLuc (New England Biolabs) by removing the pSV40 gene to prevent the duplication of
plasmids in mammalian cells. Plasmids were incubated with 0, 5, 10, 20, 40, 80 molar equivalent
amounts of cisplatin in binding buffer (10 mM HEPES, pH 7.4, 100 mM NaCl) at 37 'C
overnight. Reaction solutions were dialyzed in 3500 kDa MWCO membranes against TE buffer
(10 mM Tris-HCl, pH 7.5, 1 mM EDTA) to remove residual platinum compounds.
Concentrations of DNA and platinum were measured by UV-Vis and atomic absorption
spectroscopy, respectively. The site-specifically platinated plasmids were produced by Guangyu
Zhu. These probes have a platinum-DNA adduct on the template strand between the CMV
promoter and the coding region of GLuc. Three types of cisplatin platinated plasmids, pGLucd(GpG)-PtA 2, pGLuc-d(GpTpG)-PtA 2 , and pGLuc-ICL-PtA 2 were used as transcription assay
probes of the repair of the 1,2-d(GpG) intrastrand, 1,3-d(GpTpG) intrastrand, and interstrand
cross-links, respectively. In addition, a phenanthriplatin platinated pGLuc-dG-PtPhen was used
to assay the repair of a monofunctional adduct. The concentration of site-specifically platinated
plasmids was measured by both UV-Vis spectroscopy and the Pico-green assay (Invitrogen).
59
2.2.7. Transcription Assay
The time-dependent change in the transcription level of the platinated plasmids was
measured by a transcription assay using globally/site-specifically platinated pGLuc-derivatives
transiently transfected into the cultured cells. Cells were plated on 96-well plates at densities of
2000-4000 cells/well depending on their growth rate in DMEM supplemented with 10% FBS.
They were allowed to grow for 48 h at 37 'C, under 5% CO 2 before transfection. In the case of
globally platinated pGLuc, each platinated and unplatinated control plasmid was mixed with
lipofectamine 2000 (1:2.5 w/v) in serum-free Opti-MEM and then incubated at room temperature
for 20 min to form transfection complexes. Aliquot of 25 pL Opti-MEM containing 50 ng of
complexed pGLuc were mixed with 50 iiL of DMEM and added into each well, and the medium
was replaced after 2 h. The medium containing secreted GLuc was collected at 12 h, 24 h, 36 h,
48 h, and 60 h or 8 h, 16 h, 24 h, 32 h, 40 h, 48 h after the first medium change. The amounts of
GLuc collected in the medium were measured in a 96-well plate luminometer (Bio-Tek) using
coelenterazine (Nanolight) as a substrate. The luminescence values measured from each well
were normalized by the average luminescence measured from the cells transfected with the
unplatinated pGLuc. Transcription assays of site-specifically platinated plasmids were carried
out in a similar manner as for the globally platinated plasmids with some modifications: 10 ng of
plasmids were transfected into each well and, for each type of platinated plasmid, the
unplatinated plasmids of each pGLuc derivative were transfected simultaneously as controls. The
luminescence values obtained from site-specifically platinated plasmid-transfected cells were
normalized by the value acquired from cells transfected by unplatinated pGLuc derivatives.
60
2.3. Results
2.3.1. Establishment of HMGB1 Knockdown HeLa and A549
Stable HMGB1 knockdown cell lines were established by both heterogeneous and
homogeneous antibiotic selection. The parental HeLa or A594 cells were incubated in the
puromycin-supplemented medium for antibiotic selection simultaneously with transfected cell
lines, and complete cell death of untransfected cells was observed in 7-8 days. Without single
colony isolation, the selected cells are heterogeneous in genetic information. After the initial
selection, both HeLa KD1 and HeLa KD2 presented noticeably reduced expression levels of
HMGB 1 in western blot analysis, with much higher RNAi efficacy in HeLa KD 1 (Figure 2.1 .A).
On the other hand, both shRNA constructs were unable to significantly block HMGB1
expression in a heterogeneous population of the A549 cells after puromycin selection (Figure
2.1.A). The western blot analysis of HMGB1 revealed a less than 30% decrease of HMGB1
expression levels in A549 KD1 and A549 KD2.
Because of its better RNAi efficiency, pLKO1 HMGB1(1) was used to establish
homogeneous clones. Four clones were successfully isolated from both HeLa and A549, and the
expression level of HMGB1 was significantly reduced by >90% in two of them, whereas no
noticeable change was observed in the other two clones of both cell lines (Figure 2.1 .B,C).
Expression of HMGB1 in HeLa clones was also tested by immunofluorescence (Figure 2.2),
which gave results in agreement with the western blot analysis. The established cell lines are
listed in Table 2.2 along with their designation and RNAi efficiency.
61
(A)
1
2
3
5
4
(B)
6
2
3
4
5
Actin -
Actin HMGBI-
(C)
1
HMGBI
1
2
3
4
5
6
-
7
Actin HMGBI Figure 2.2. HMGB1 knockdown efficiency analyzed by western blotting. (A) Expression level of HMGB1 in heterogeneous
populations of A549 KD1 (lane 1), A549 KD2 (lane 2), HeLa KD1 (lane 4), and HeLa KD2 (lane 5). A549 knockdown cell
lines showed moderate RNAi efficacy compared to HeLa knockdown cell lines. Parental A549 (lane 3) and HeLa (Lane 6)
were used as controls. (B) Expression level of HMGB1 in homogeneous populations of A549 control (lane 1) and C1-4
(lane2-5) cells. The RNAi efficiency in each clone is C11, C12, C14, and C13 in descending order. (C) Expression level of
HMGB1 in parental (lane 1), control (Lane 2), and homogeneous populations of C11-4 (Lane 3-6) HeLa cell lines. The
RNAi efficiency in each clone is Cl1, C14, C12, and C3 in descending order.
Table 2.2. List of Stably Established HeLa and A549 Cell Lines and Their RNAi Efficiency
Designation
Transfection
RNAi
No
-
A549
pLKO 1. TRC
-
A549 control
HeLa KD1
pLKO1HMGB1(1)
High
HeLa KD2
pLKOlHMGBl(2)
HeLa ClI
HeLa
Designation
Transfection
RNAi
No
pLKO 1. TRC
-
A549 KD1
pLKOlHMGB1(1)
Low
High
A549 KD2
pLKO1HMGB1(2)
Low
pLKOlHMGBl(1)
High
A549 C11
pLKOlHMGB1(1)
High
HeLa C12
pLKOlHMGBl(1)
Low
A549 C12
pLKOlHMGB1(1)
High
HeLa C13
pLKO1HMGBl(1)
Low
A549 C13
pLKO1HMGBl(1)
Low
HeLa C14
pLKO1HMGBl(1)
High
A549 C14
pLKO1HMGBl(1)
Low
HeLa control
62
Parental
Control
CI1
C2
C13
C14
Figure 2.3. Immunofluorescence results of HMGB1 knockdown HeLa in homogeneous populations. After sequential
treatment of antibodies, parental/control HeLa present stronger overlapping signals from HMGB1 and nuclear staining.
HeLa C11 and C14, showing high RNAi efficiency in western blot analysis, do not show significant HMGB1 signals
compared to the background signal. Strong HMGB1 signals, comparable to that of control/parental cells, are observed in
C12 and C13.
2.3.2. Cisplatin Sensitivity of HMGB1 Knockdown Cell Lines
Cisplatin cytotoxicity in HMGB1 knockdown and parental/control cell lines was
analyzed by the MTT assay (Table 2.3, Figure 2.4). IC50 values of HeLa KD1 and HeLa KD2
were larger than those of parental/control HeLa cells by a factor of 2 and 1.5, respectively. The
63
IC50 of HeLa KD1 increased more than that of HeLa KD2, which is consistent with the
difference of RNAi efficiency found in the western blot analysis (Figure 2.2.A). Genetically
homogeneous knockdown HeLa cells exhibited a similar correlation between the expression
levels of HMGB1 and cisplatin cytotoxicity. HeLa Ci1 and 4, with largely reduced HMGB1
expression levels, presented a 2.7 or 4 fold increase of IC50 values from parental or control
HeLa. IC50 values of HeLa clones with poor RNAi efficacy were only 1.5-fold larger than that
of parental HeLa or 1.34-fold larger than that of the HeLa control cell line.
(A)
(B)
100
100
-0-A49
-""Hela
-*-HeLa KD1
75
-+-A549KD1
75
-4--HeLa KD2
.
-4-A549KD2
50
0
25 -25
0
8
16
24
32
0
8
16
24
32
cDDP (paM)
cDDP (pM)
(C)
(D)
100
100
-+A5493
'"4"A549 Control
+A549 C1
-+*-Hela control
-*-Hela C11
5-+-HeLa C12
7S
-.
75
-4-Hela C13
5so
-+A549 Cl2
-4Hea C14
50
2S
25
0
0
0
4
8
12
16
0
cDDP (p&M)
4
8
12
16
cDDP (pLM)
Figure 2.4. Cisplatin cytotoxicity assay of HMGB1 knockdown cell lines. Cisplatin kill curves of (A) heterogeneous
knockdown HeLa, (B) heterogeneous knockdown A549, (C) homogeneous knockdown HeLa, and (D) homogeneous
knockdown A549 were acquired by MTT assay. For each individual experiment, a cytotoxicity assay of the parental
and/or control cell lines was conducted simultaneously.
A549 KD1 and KD2 presented neither a clear correlation nor a significant difference in
cisplatin cytotoxicity from parental/control A549 cells. This negligible change in cytotoxicity is
in agreement with the low RNAi efficiency in A549 KD1 and KD2 cells (Figure 2.2.A). In the
64
case of A549 CI1 and C12, which presented high RNAi efficiency in western blot analysis
(Figure 2.2.B), the cisplatin cytotoxicity decreased by a factor of 1.5 or 2.4, respectively.
Transfected A549 clones without a noticeable knockdown of HMGB 1 presented similar cisplatin
cytotoxicity to parental/control A549 cells.
Table 2.3. IC50 Values of HMGB1 Knockdown HeLa and A549 Cell Lines
aMTT
Cell ID
HeLa (pM)
Parental
1.26 ±0.55
Control
IC50/IC50parentai
A549 (iM)
IC50/IC50parenta
1
4.27 ±1.51
1
1.40± 0.59
1.11
3.91 ±0.37
0.91
C11
3.45 ±0.67
2.74
1.78 ±0.18
0.42
C12
1.88± 1.00
1.50
2.76± 0.31
0.65
C13
1.91 ± 0.98
1.52
5.00± 2.63
1.17
C14
5.16 ±2.54
4.09
4.54
1.06
Parentala
N/A
N/A
2.53
1
KD1
2.42 ±0.50
1.92
3.44
1.36
KD2
1.85+0.58
1.50
1.78
0.70
0.98
assay of heterogeneous HMGB1 knockdown A549 was carried out only once.
Because of the large difference in IC50 values between the single MTT assay of
heterogeneous A549 and the average of MTT assay of homogeneous A549, the results were
analyzed separately. All other values are the averages of at least two independent
experiments.
2.3.3. Restoration of the Transcription Level of Platinated Plasmid in Knockdown Cells
Global platination of pGLuc was carried out by incubation of the plasmid with cisplatin
at pH 7.4 for 24 h. Under these conditions, almost all of the platinum added in the initial reaction
mixture bound to DNA, as judged by the platinum to plasmid ratio measured by UV-Vis and
65
atomic absorption spectroscopy. The rate of the restoration of transcription levels, or the slope of
the time versus normalized transcription level plot, is assumed to reflect the repair efficiency in
each cell line. Both HeLa KD1 and HeLa KD2 exhibited a substantially slower increase of the
transcription level of globally platinated pGLuc than parental HeLa cells (Figure 2.5). The
different recovery rates of knockdown and parental cell lines were more noticeable at high
platinum per plasmid ratios. Similar to changes in IC50 values measured by the MTT assay, the
difference between HMGB 1 knockdown and parental HeLa in rate of restoration of transcription
level is more significant in HeLa KD1 than in HeLa KD2.
(A
-- 1A29*.2
W
r-0-1a
.Het
HnaIKD1HeLa2
-i
I
j4D
-4U94U
A
20
=10.9
PUPkla
puPupplm
PUPhasm= UI
20.9
se
so
sosee
0
U
4U
24
W
60
1
0
72
Uo
s
43
n
a
a
a
M
0
O
24
PUP*am=
1
PUPMasnM= 43.6
72
4 3so
0
U
24
M
a
U
4 a
72
78.
I.e
19.2
0*
U
24
24
6
a
a
os
0
n
U
24
s
N4
Tm
a
TmeM
Figure 2.5. Transcription profile of globally platinated pGLuc probe in heterogeneous knockdown HeLa. (A) Plot of the
transcription levels of platinated pGLuc probes normalized by the transcription level of unmodified pGLuc versus the
average platinum to plasmid ratio of the probe in different cell lines. (B) Plot of the normalized transcription level of
platinated pGLuc probe versus the time points at which the medium was collected for each platinum to plasmid ratio.
66
Compared to HeLa, heterogeneous A549 knockdown cell lines did not show a drastic
difference in the restoration rate of transcription of globally platinated DNA at low Pt:DNA
ratios. A549 KD2 demonstrated a relatively slower restoration rate than parental or control A549
at Pt:DNA ratios higher than 40. Transcription restoration in A549 KD1 and parental A549 did
not differ (Figure 2.6). In contrast to heterogeneous knockdown A549, homogeneous A549 C11
revealed significantly slower transcription restoration of globally platinated pGLuc, even at low
Pt:Plasmid ratios (Figure 2.7).
10
In.
0
Zj1
20
0
a
a
a
40
20
a
a
(B)
A5-Me-- Me..M
a
a
20
a
s
M
PM:Fawl
-4Me.MM A4
-- MeU2
F.
W
2
1129
L
0
12
24
36
4
a
72
a
12
as
36
a
4
72
S
12
24
KZI
4
so
72
OW N,
--PM49 -P-aM= -4-AS.
20
I --WA5mi
Ph~inid = 43A6
-- 74KD1
PtiPaid = 78.5
=6
ii'
.14
jiS
0
S
12
a
a
T h)
a
a
72
S
12
24
K
tWAw0
a
a
72
Figure 2.6. Transcription profile of globally platinated pGLuc probe in heterogeneous knockdown A549. (A) Plot of the
transcription levels of platinated pGLuc probes normalized by the transcription level of unmodified pGLuc versus the
average platinum to plasmid ratio of the probe in different cell lines. (B) Plot of the normalized transcription level of
platinated pGLuc probe versus the time points at which the medium was collected for each platinum to plasmid ratio.
67
(A)
100
100
A549
75
A549 C1
n
-. 1-
+
10+024h
+24h
-032
jso
a
t0
20
--A2h
10
0
50
40
30
50
(B)
-- As91
-a-AS49
-A49 CI
Pt/Plasmid a 1
+AS49
+ASI
Pt/Plasmid = 20
3
1
24
46
36
Time(h)
60
72
50
As
PtIPlasmid a 40
0
20
0
40
30
20
Ptpasmid
Pt:PAnnid
0
12
24
4h
36
n"u (h)
60
72
0
12
24
36m a
Time(h)
60)
Figure 2.7. Transcription profile of globally platinated probes in homogeneous HMGB1 knockdown A549 CI1 and
parental A549. Transcription assay was carried out with average Pt to plasmid ratios of 10, 20, and 40 respectively. (A)
Plot of the normalized transcriptionlevels of platinated pGLuc probes versus the average platinum to plasmid ratio of the
probes in parental A549 and A549 C11. (B) Plot of the normalized transcription level of platinated pGLuc probe versus
the time points at which the medium was collected for each platinum to plasmid ratio.
2.3.4. Cisplatin and Phenanthriplatin Cytotoxicity in NER Proficient/Deficient Fibroblast
Cell Lines
The cellular response of fibroblast cell lines established from CS or XP patients were
investigated for improved understanding of the participation of NER in the antitumor mechanism
of different types of platinum adducts (Table 2.4). Along with that of cisplatin, a traditional
bifunctional platinum compound, the cytotoxicity of phenanthriplatin, a recently reported
monoftunctional compound, was measured in NER proficient/deficient fibroblast cell lines.
Phenanthriplatin is a pyriplatin analogue with phenanthridine instead of pyridine replacing one of
the chloride ligands of cisplatin (Figure 2.8.B). Unlike most other monofunctional platinum
compounds, phenanthriplatin is much more potent than cisplatin in all tested cancer cell lines, by
a factor of 3-30.17 The cisplatin IC50 values of all fibroblast cell lines were around 1-2 pM,
except for CSB, which is 3-4 fold more sensitive to cisplatin than the others (Figure 2.8.A). In
the case of phenanthriplatin, normal fibroblast cells displayed the highest IC50 value (0.41 pM),
68
followed by XPC, XPF, XPG, and CSB in descending order (Figure 2.8.B). The ratio of cisplatin
IC50 value to phenanthriplatin IC50 value was smallest in normal fibroblast cells, 2.32. The
IC50 ratios of CSB and XPC were similar to each other, around 3-4, and those of XPG and XPF
were about 6-8, the largest among the fibroblast cell lines.
(B)
(A)
10
ino
100
60
HsN,
Cl
-+*-Nonn
HsN
Cl
-0---xcS
100
-0-
-.- xc
60
.0
--- XPF
0
--
20
XPG
20
0
0
4
12
8
Noml
-- $XPG
0
16
0
1
[cDDP pM
3
2
4
PhenPtIpM
Figure 2.8. Cisplatin (A) and phenanthriplatin (B) cytotoxi city of NER proficient/deficient fibroblast cell lines and control
fibroblast cell line.
Table 2.4. Cisplatin and Phenanthriplatin IC50 Values of Each Fibroblast Cell Line
Cell ID
IC50 cDDP
IC50 PhenPt
Ratio (cDDP/PhenPt)
Normal
0.99 ± 0.04 gM
0.426 ± 0.020 jM
2.32
XPG
1.17 ±0.05 pM
0.182 L 0.003 jM
6.46
XPF
1.71 ±0.35 gM
0.221 ± 0.016 pM
7.75
CSB
0.37 ± 0.06 jM
0.094 ± 0.003 pM
3.96
XPC
1.11 ±0.09 gM
0.319 ±0.011 pM
3.60
2.3.5. Repair Profiling of Different Platinum-DNA Adducts in NER Proficient/Deficient
Cell Lines
For all different types of platinated pGLuc probes, transcription was restored most
rapidly in the normal fibroblast cell line. In XPF and XPG, there was very slow restoration of
transcription for all the tested plasmids. The normalized expression levels from platinated pGLuc
in XPG were significantly lower than in other cell lines at every time point, for all the different
69
adducts. There was almost no recovery of transcription level of pGLuc-d(GpG)-PtA 2 in XPF,
XPG, and CSB, whereas in XPC there was a considerable amount of restoration (Figure 2.9.A).
In the case of pGLuc-d(GpTpG)-PtA 2 and pGLuc-ICL-PtA 2, CSB appeared to repair the damage
as fast as normal fibroblast, whereas the restoration curve of XPC is rather similar to that of XPF
(Figure 2.9.B,C). Unlike cisplatin-platinated plasmids having extremely slow recovery of
transcription levels in XPF and XPG, significant increase of transcription levels of pGLuc-dGPtPhen was observed in two cell lines. Except for normal fibroblasts, which have the fastest
recovery rate, the restoration plots for pGLuc-dG-Phen had similar slopes with XPF, XPG, CSB,
and XPC cell lines (Figure 2.9.D).
(B)
(A)
100
120
cDDP-1,2-d(GpG)
80
cDDP-1,3-d(GpTdG)
100
-+-Normal
ONormal
-+-CSB
o
-+-CSB
60
-4--XPC
XPC
20
--
XPF
40
--
XPF
--
XPG
20
--
XPG
0
0
16
8
0
24
32
40
48
8
0
56
16
(C)
(D)
100
90
so0
24
32
40
48
56
Time (h)
Time (h)
cDDP-ICL
-4-Normal
-+
.2 60
PhenPt-mono
7
-*
-
Normal
Cs1.-"CSB
-4-CS
--
XPF
5
-+-XPF
XPG
30
-4G
*XP
'-20
10
0
0
8
16
24
32
40
48
0
56
8
16
24
32
40
48
56
Time (h)
Time (h)
Figure 2.9. Transcription profile of site-specifically platinated probes in NER proficient/deficient cell lines. pGLuc
derivatives with a 1,2-intrastrand cross-link (A), a 1,3-intrastrand cross-link (B), an interstrand cross-link (C), and
phenanthriplatin monofunctional adduct (D) were used as transcription assay probes.
70
2.4. Discussion
2.4.1. Cellular Response to Cisplatin in HMGB1 Knockdown HeLa and A549 Cells
The repair-shielding model of HMGB 1 suggests that binding of HMGB 1 inhibits the
assembly of the repair machinery around the lesions and enhances the antitumor activity of
cisplatin. According to this model, reduction of intracellular levels of HMGB 1 by shRNA would
increase the repair rate of platinated adducts to which HMGB1 has strong binding preference,
and ultimately desensitize cells to cisplatin. The transcription assay of globally platinated pGLuc
probes, however, revealed the opposite trend. In both HeLa and A549 cells, HMGB 1 knockdown
retards transcription restoration of platinated plasmids. In the case of cytotoxicity assays,
HMGB1 knockdown HeLa showed a higher IC50 value than parental HeLa whereas knockdown
A549 is more sensitive to cisplatin than its parental cell line (Table 2.5). Assuming that any
change that slows down the repair processes enhances drug efficiency, the outcome of the
transcription assays and cytotoxicity assays in A549 were in agreement with each other. They do
not, however, fit with the repair-shielding hypothesis at all. On the other hand, the two assays
conducted in HeLa cell lines gave inconsistent results.
Table 2.5. The Change of Cellular Response to Cisplatin in HMGB1 Knockdown Cell Lines and
Expected Results Based on the Repair-Shielding Model
Knockdown Cells
Repair-Shielding Model
Recovery of Transcription Level
Cisplatin Cytotoxicity
T
t
A549
HeLa
71
The results of the transcription assay clearly revealed that HMGB 1 does not inhibit but
rather facilitates repair of cisplatin-modified DNA in intracellular environments. Since the
repair-shielding hypothesis is largely based on the results of in vitro repair but not cell-based
assays, the model potentially has limitations. Several cell-based studies suggest that TCR is the
dominant NER sub-pathway of cisplatin adducts in vivo, whereas GGR is the main sub-pathway
at work in CFE system. 9" 3 "8 Considering that the repair-shielding model assumes that HMGB
proteins might be involved in the recognition of DNA damage, the lack of the ability of the assay
to reflect the contribution of TCR brings into question the relevance of the repair-shielding
hypothesis.
In contrast to the transcription assays that presented consistently negative correlation
between the repair efficiency and the expression levels of HMGB 1 in both cell lines, the MTT
assay indicated the opposite trend in cisplatin sensitivities in HeLa and A549 cell lines upon
HMGB 1 knockdown. This opposite trend is assumed to originate from cisplatin-triggered
mechanisms other than DNA repair. Some of the partner proteins interacting with HMGB 1 and
the consequences of those interactions were described in the Chapter 1. It should be pointed out
that some of the interaction partner proteins show cell-specific expression profiles. For instance,
HeLa cannot express p53 because of human papillomavirus (HPV) infection, whereas A549
expresses wild type p53 protein in reasonably high levels. Considering how important p53 is in
damage-triggered apoptosis1 9 ,20 and the previous studies demonstrated the cell-specific influence
of HMGB1 on the activation of p53,-' the different expression profiles of p53 can be assumed
to be a factor of importance in regulating the consequences of HMGB 1 knockdown.
2.4.2. Damage Repair Processes and HMGB1
72
Cellular systems have well-organized DNA repair processes to maintain the integrity of
their genome. The efficiency of each repair apparatus is largely dependent on the type of damage.
Although NER is the major cisplatin repair mechanism, there are several other important repair
mechanisms that participate in the repair of platinated DNA. HMGB1 appears to participate in
those major repair mechanisms of cellular systems.
Mismatch repair (MMR) is a process that recognizes and corrects bases that are
incorrectly incorporated during DNA replication. It is particularly active in the repair of cisplatin
1,2-d(GpG) intrastrand cross-links, but not 1,3-d(GpNpG) or 1,2-d(ApG) interstrand cross-links
in vitro, comprising reconstituted MMR components.
24
HMGB 1 is a critical factor for in vitro
MMR systems, probably through physical interaction with MutSa, 25 ,2 6 a protein that plays a
critical role in base substitution, prior to excision of the damaged nucleotide.
Double strand break repair (DSBR) is also one of the main repair processes in cellular
systems. 27,28 Unlike NER and MMR, cleavages occur in both strands during DSBR. There are
two types of DSBR for cisplatin-modified DNA. Homology directed repair (HDR) has a lower
error because it uses the whole homologous DNA as a template. On the other hand, non-
homologous end joining (NHEJ) mechanism shows a higher error due to lack of the proper
template. The initiation step of NHEJ is recognition of damage by DNA-dependent protein
kinase (DNA-PK) and Ku proteins, both of them are activated by HMGB 1 in vitro.29 3 0 In
addition, these proteins are weakly involved in the HDR process.31 HMGB 1 also interacts with
and facilitates the function of RAG 1/2 nuclease, which cleaves DNA in HDR processes. 32
Most of the studies of the function of HMGB 1 in cellular repair processes are carried out
in vitro, and the behavior of HMGB 1 and repair proteins can be quite different in cells. The fact
73
that HMGB 1 functions to enhance both MMR and DSBR, however, suggests a reasonable
explanation for the outcome of the transcription assay in the HMGB 1 knockdown cell lines. The
contribution of each repair mechanism to the intracellular repair of cisplatin damage and the
importance of the presence of HMGB 1 for each repair process in cells needs further investigation
to confirm this hypothesis.
2.4.3. NER of Different Types of Platinum Adducts
The bifunctional nature of cisplatin and its analogues gives them the ability to form
various kinds of cross-links. Among them, the 1,2-intrastrand cross-link is predominant,
accounting for more than 90% of total platinated adducts. 33 The 1,3-intrastrand cross-link is the
second-most frequent adduct, and very small amounts of interstrand cross-links
or
monofunctional adducts have been reported. From in vitro studies using nuclear extracts, each
type of platinated adduct presents a unique profile for damage recognition proteins. 34 ,31 On the
other hand, monofunctional platinum-based anticancer drugs are able to form only one covalent
bond to the DNA, mainly on the N7 position of guanine residues. The crystal structure of DNA
bearing a central monofunctional adduct of pyriplatin, one of the most thoroughly investigated
monofunctional platinum compounds, revealed no significant distortion of DNA as occurs at
cross-linked adducts of bifunctional platinum compounds.36 The crystal structures of RNA pol II
and pyriplatin-modified DNA complex revealed that monofunctional lesions are capable of
stalling RNA pol II during the elongation process.37 Transcription assays of different
transcription probes containing a cisplatin 1,2-intrastrand, 1,3-intrastrand, or interstrand crosslinks were used to determine the different contributions of NER to the total repair of each adduct
in a set of fibroblast cell lines consisting of TCR/GGR active, TCR/GGR inactive, only TCR
active, and only GGR active cells. In addition to cisplatin-platinated transcription probes, a
74
pGLuc probe with phenanthriplatin-modified adducts was used to assess the repair of
monofunctional adducts. In cytotoxicity assays carried out prior to the transcription assay, no
concrete correlation between cisplatin cytotoxicity in each cell line and repair deficiency was
noted (Table 2.4). Only CSB cell lines showed distinctively low IC50 values. Even XPG and
XPF, in which both NER sub-pathways were blocked, presented no obvious difference in
cisplatin sensitivity when compared to normal fibroblasts. This lack of correlation is not
surprising, because each cell line was acquired from a different patient with a totally different
genomic information and medical history. Moreover, they all went through an immortalization
step that can influence cisplatin resistance. The ratio of cisplatin to phenanthriplatin IC50 values,
however, tells a more interesting story. The ratio is biggest in NER deficient cell lines, smallest
in normal fibroblast cell lines, and intermediate for TCR deficient or GGR deficient cell lines.
Even though this result is not definitive, it supports the hypothesis that the monofunctional
adducts of phenanthriplatin might be repaired by NER, and that the contributions of each subpathway are similar to those of cisplatin.
The restoration of normalized transcription levels is slower in all TCR and/or GGR
deficient cell lines compared to that in normal fibroblasts. This trend reflects the critical
contribution of NER to the repair of each type of platinated adduct. The order of the restoration
rate in repair deficient cell lines varies among different platinated adducts. The repair of a 1,2intrastrand cross-link is much faster in XPC than in CSB. The slope of the restoration curve of
CSB is almost the same as that of XPF and XPG, which suggests that TCR is a dominant NER
sub-pathway for the 1,2-intrastrand cross-link. On the other hand, a transcription assay of the
1,3-intrastrand cross-link showed considerable repair in CSB, much faster than the repair in XPC,
suggesting that GGR is the more important repair pathway for this cross-link. Given that the
75
repair curves of the 1,3-intrastrand cross-link for normal fibroblast and CSB are almost identical,
it is assumed that TCR barely influences the repair of the 1,3-intrastrand cross-link. The first in
vitro repair assay using repair active cell extracts showed that a linear probe bearing a cisplatin
1,3-intrastrand cross-link was repaired much more efficiently than that bearing a cisplatin 1,2intrastrand cross-link. 4 Studies on cultured cells, however, revealed that TCR is the major NER
sub-pathway of total cisplatin damage. This cellular study is in agreement with the results of the
transcription assays, considering about 90% of cisplatin adducts are 1,2-intrastrand cross-links.
Given the fact that the in vitro repair assay dominantly reflects the influence of GGR, the
conclusion that 1,3-intrastrand cross-linked DNA is a better substrate for NER than 1,2intrastrand cross-linked DNA, is only applicable under in vitro repair system conditions.
The restoration pattern for the cisplatin interstrand cross-link is similar to that of the
cisplatin 1,3-intrastrand cross-link, demonstrating faster recovery in TCR deficient cell lines than
in GGR deficient cell lines. Compared to the other types of cisplatin damage, not many details
are known about the repair of interstrand cross-links. It is, however, expected to be very different
from that of intrastrand cross-links because the interstrand cross-linked damage spans the duplex.
NER plays a considerable role in the repair of representative interstrand cross-link producing
agents like psoralen. One possible way to explain the contribution of NER to the repair of
cisplatin interstrand cross-links is based on the E.coli repair model, a combination of HDR and
NER, established by Ronald Cole. 38 In this repair model, one of the strands is first cleaved by an
NER apparatus and the gap is filled using homologous DNA as a template, as in HDR. This step
is followed by a second NER cleavage on the remaining strand. This repair model requires an
undamaged chromosomal DNA homologous to that on which the DNA is damaged. It is believed
that, in the transcription assay of the interstrand cross-link, a small fraction of undamaged
76
plasmid can function as the template. 3 9 The fraction of undamaged plasmid increases as time
passes and the damage is repaired. The transcription assay revealed that NER, especially the
GGR sub-pathway, is critical in the repair of interstrand cross-links. Several previous studies
also presented a large influence of the XPF/ERCC1 complex in the repair of the interstrand
cross-links as compared to other NER proteins. 39 Such an effect, however, was not observed in
this transcription assay, most likely because the restoration rates in XPC, XPF, and XPG are all
too slow to observe notable differences.
Although XPF and XPG showed very weak repair efficiency of cisplatin cross-links,
most likely because both sub-pathways of NER are blocked, these cell lines showed considerable
restoration of transcription from phenanthriplatin monofunctional adducts. It seems as though
other cellular repair mechanisms in addition to NER play important roles in the repair of
monofunctional adducts. In addition, for the phenanthriplatin adduct, no significant difference
was observed between TCR deficient and GGR deficient cell lines. The function of HMGB 1 in
each sub-pathway remains to be uncovered by future RNAi studies in these cell lines.
2.5. Summary
The cellular responses to cisplatin were examined in cultured mammalian cells. The
cytotoxicity studies carried out in HMGB1 knockdown HeLa and A549 showed opposite
correlations between the HMGB1 expression level and cisplatin cytotoxicity: HMGB1
knockdown increased the cisplatin resistance of A549 but sensitized HeLa to cisplatin. The
repair of globally platinated plasmids in HMGB1 knockdown cells was slower than in parental
cells. This result is opposite to what is expected from the repair-shielding theory. The
intracellular repair assay of site-specifically platinated plasmid in fibroblast cells deficient in
77
specific sub-pathways of NER showed different contributions of TCR and GGR in the repair of
different adducts. This different contribution is at least partially responsible for the conflicting
results observed in previous in vitro repair assays and intracellular repair assays. Further repair
studies in fibroblast cells combined with HMGB1 knockdown is expected to give clues
explaining these conflicting results acquired from the two systems.
78
2.6. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
Bruhn, S. L.; Pil, P. M.; Essigmann, J. M.; Housman, D. E.; Lippard, S. J. Proc. Natl.
Acad Sci. U. S. A. 1992, 89, 2307-2311.
Hughes, E. N.; Engelsberg, B. N.; Billings, P. C. J. Biol. Chem. 1992, 267, 13520-13527.
Brown, S. J.; Kellett, P. J.; Lippard, S. J. Science 1993, 261, 603-605.
Huang, J.-C.; Zamble, D. B.; Reardon, J. T.; Lippard, S. J.; Sancar, A. Proc. Natl. Acad.
Sci. U. S. A. 1994, 91, 10394-10398.
He, Q.; Liang, C. H.; Lippard, S. J. Proc.Natl. Acad. Sci. U. S. A. 2000, 97, 5768-5772.
Zamble, D. B.; Mikata, Y.; Eng, C. H.; Sandman, K. E.; Lippard, S. J. J. Inorg. Biochem.
2002, 91, 451-462.
Wei, M.; Burenkova, 0.; Lippard, S. J. J Biol. Chem. 2003, 278, 1769-1773.
Huang, J.; Ni, J.; Liu, K.; Yu, Y.; Xie, M.; Kang, R.; Vernon, P.; Cao, L.; Tang, D.
CancerRes. 2012, 72, 230-238.
Furuta, T.; Ueda, T.; Aune, G.; Sarasin, A.; Kraemer, K. H.; Pommier, Y. Cancer Res.
(29)
2002, 62, 4899-4902.
Sugasawa, K.; Ng, J. M.; Masutani, C.; Iwai, S.; van der Spek, P. J.; Eker, A. P.; Hanaoka,
F.; Bootsma, D.; Hoeijmakers, J. H. Mol. Cell. 1998, 2, 223-232.
Sancar, A. Annu. Rev. Genet. 1995, 29, 69-105.
Svejstrup, J. Q. Nat. Rev. Mol. Cell. Biol. 2002, 3, 21-29.
Mu, D.; Sancar, A. J. Biol. Chem. 1997, 272, 7570-7573.
Costa, R. M.; Chigancas, V.; Galhardo Rda, S.; Carvalho, H.; Menck, C. F. Biochimie
2003, 85, 1083-1099.
Mu, D.; Sancar, A. Prog.Nucleic Acid Res. Mol. Biol. 1997, 56, 63-8 1.
Ang, W. H.; Myint, M.; Lippard, S. J. J. Am. Chem. Soc. 2010, 132, 7429-7435.
Park, G. Y.; Wilson, J. J.; Song, Y.; Lippard, S. J. Proc.Natl. Acad. Sci. U. S. A. in press
Bulmer, J. T.; Zacal, N. J.; Rainbow, A. J. Cancer Chemother. Pharmacol. 2005, 56,
189-198.
Lowe, S. W.; Ruley, H. E.; Jacks, T.; Housman, D. E. Cell 1993, 74, 957-967.
May, P.; May, E. Oncogene 1999, 18, 7621-7636.
Imamura, T.; Izumi, H.; Nagatani, G.; Ise, T.; Nomoto, M.; Iwamoto, Y.; Kohno, K. J.
Biol. Chem. 2001, 276, 7534-7540.
Krynetskaia, N.; Xie, H.; Vucetic, S.; Obradovic, Z.; Krynetskiy, E. Mol. Pharmacol.
2008, 73, 260-269.
Stros, M.; Ozaki, T.; Bacikova, A.; Kageyama, H.; Nakagawara, A. J. Biol. Chem. 2002,
277, 7157-7164.
Fourrier, L.; Brooks, P.; Malinge, J. M. J. Biol. Chem. 2003, 278, 21267-21275.
Yuan, F.; Gu, L.; Guo, S.; Wang, C.; Li, G. M. J. Biol. Chem. 2004, 279, 20935-20940.
Mello, J. A.; Acharya, S.; Fishel, R.; Essigmann, J. M. Chem. Biol. 1996, 3, 579-589.
Chu, G. J. Biol. Chem. 1997, 272, 24097-24 100.
Iliakis, G.; Wang, H.; Perrault, A. R.; Boecker, W.; Rosidi, B.; Windhofer, F.; Wu, W.;
Guan, J.; Terzoudi, G.; Pantelias, G. Cytogenet. Genome Res. 2004, 104, 14-20.
van Gent, D. C.; Hiom, K.; Paull, T. T.; Gellert, M. EMBO J. 1997, 16, 2665-2670.
(30)
Yumoto, Y.; Shirakawa, H.; Yoshida, M.; Suwa, A.; Watanabe, F.; Teraoka, H. J
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
Biochem. 1998, 124, 519-527.
79
(31)
(32)
(33)
(34)
(35)
(36)
(37)
Sawchuk, D. J.; Mansilla-Soto, J.; Alarcon, C.; Singha, N. C.; Langen, H.; Bianchi, M. E.;
Lees-Miller, S. P.; Nussenzweig, M. C.; Cortes, P. J. Biol. Chem. 2004, 279, 2982129831.
West, R. B.; Lieber, M. R. Mol. Cell. Biol. 1998, 18, 6408-6415.
Eastman, A. Pharmacol.Ther. 1987, 34,155-166.
Guggenheim, E. R.; Xu, D.; Zhang, C. X.; Chang, P. V.; Lippard, S. J. Chembiochem
2009, 10, 141-157.
Zhu, G.; Lippard, S. J. Biochemistry 2009, 48, 4916-4925.
Lovejoy, K. S.; Todd, R. C.; Zhang, S.; McCormick, M. S.; D'Aquino, J. A.; Reardon, J.
T.; Sancar, A.; Giacomini, K. M.; Lippard, S. J. Proc. Nati. Acad. Sci. U. S. A. 2008, 105,
8902-8907.
Wang, D.; Zhu, G.; Huang, X.; Lippard, S. J. Proc. Nati. Acad Sci. U. S. A. 2010, 107,
(38)
9584-9589.
Cole, R. S. Proc.NatL. Acad. Sci. U. S. A. 1973, 70, 1064-1068.
(39)
Kuraoka, I.; Kobertz, W. R.; Ariza, R. R.; Biggerstaff, M.; Essigmann, J. M.; Wood, R. D.
J. Biol. Chem. 2000, 275, 26632-26636.
80
Chapter 3. Redox State-Dependent Interaction of HMGB1 and Cisplatin-Modified DNA
Research in this chapter was published in Biochemistry, 2011, 50, 2567-2574.
81
3.1. Introduction
Cisplatin is one of the most widely used anticancer drugs, being effective against a range
of tumors including ovarian, genitourinary, lung, head and neck cancers.1 Most noteworthy is the
cure rate of testicular cancer, which has increased dramatically following the introduction of
2
cisplatin-based chemotherapy. Cisplatin and its second generation analogues carboplatin and
oxaliplatin bind to DNA preferentially at the N7 position of guanine bases,1,3 ,4 inhibiting
replication 5'6 and transcription.7~9 The inhibition of those critical DNA-related processes triggers
0
subsequent intracellular events that activate necrotic and apoptotic pathways.1 Despite its
outstanding antitumor activity, cisplatin chemotherapy is limited by intrinsic and acquired
resistance of certain tumors." Several major mechanisms of cisplatin resistance have been
discovered,12, 13 and there have been numerous attempts to overcome the problems based on these
mechanisms. One strategy has been to develop analogs of cisplatin with different antitumor
profiles, like carboplatin and oxaliplatin.14" 5 Another is to identify intracellular factors that
participate in the mechanisms of cisplatin resistance, and to discover ways to control these
factors.
HMGB1 is a highly abundant and ubiquitously expressed nuclear protein, the defining
member of the HMG superfamily. This small, -25 kDa protein functions as a transcription factor
that bends DNA and assists other DNA-binding proteins to form their recognition complexes.16'
7
HMGB 1 consists of two tandem DNA binding domains, high mobility group box A and B, and a
highly acidic C terminal tail composed of a string of aspartate and glutamate residues. Both
domains A and B consist of three a-helices that cooperatively wrap around DNA, approaching
from the minor groove, in the DNA-HMG box complex. HMG boxes in HMGB 1 bind to DNA
82
in a structure-specific but sequence-independent manner, with greater binding affinity for
18 19
nonlinear DNA such as bent, kinked, or unwound duplexes. ,
HMGB 1 as well as the A and B domains binds to platinated lesions on DNA with
specificity for 1,2-intrastrand cross-links,2 0 which account for about 90% of the cisplatin-DNA
adducts formed in vivo.21,22 HMGB 1 also has binding selectivity for interstrand cross-linked
(ICL) versus undamaged DNA, but not 1,3-intrastrand cross-links. 23 HMGB1 domain A binds
with greater specificity to platinated DNA than does domain B. In full-length HMGB 1, it is the
A domain that mainly binds to platinated lesions, although domain B also plays a role in
strengthening the interaction.2 4
An X-ray crystallographic analysis and binding assay revealed details of the interaction
between HMGB1 domain A and DNA harboring a cisplatin 1,2-intrastrand d(GpG) cross-link
(Figure 3.1).25 A critical feature that dramatically enhances the binding preference of the domain
for the platinated DNA is the occurrence of an intercalating residue, Phe37. The phenyl ring of
Phe37 inserts into a hydrophobic notch formed in the minor groove across from the two
platinum-modified guanine bases, and this stacking interaction significantly stabilizes the overall
protein-DNA complexes and leads to an asymmetric positioning of the protein with respect to the
platinum cross-link.24 25
The sensitivity of the cells to cisplatin can be altered by changing the expression levels of
HMGB1 and other HMG-domain proteins.26-28
Moreover, in vitro repair assays reveal that
HMGB1 impedes nucleotide excision repair (NER), the major mechanism by which platinated
lesions are removed from DNA. 29 -3'Taken together, these studies suggest that HMGB1 bound to
cisplatin-modified DNA shields the platinated lesion from recognition by the repair proteins. A
83
correlation between HMGB1 and cisplatin cytotoxicity, however, is not universally obtained. A
cytotoxicity study revealed that the cisplatin sensitivity in genetically modified HMGB 1
knockout mouse embryonic fibroblasts (MEFs) is not substantially different from that of the
parental MEF cell line.32 The failure to correlate cell sensitivity to cisplatin with HMG box
protein levels indicates either that there is no such correlation or that one or more variable factors
among different cell lines alter the HMGB 1-platinated DNA interaction. The present study was
undertaken to test one possibility for the latter explanation.
Helix I
Pe
N 'IS
L
SHelix I
Figure 3.1. X-ray crystal structure of the complex between HMGB1 domain A and a 16-bp cisplatin-modified DNA under
reducing conditions. 2 5 Phe37 (cyan) inserts between two guanine bases bound to platinum (magenta). Two thiol groups
from Cys22 and Cys44 are shown as van der Waals spheres. The hydrophobic core at the interface between helix I and
helix H is shown as a bright pink surface.
HMGB 1 domain A contains two cysteine residues, in positions 22 and 44, which can
form a disulfide bond under mildly oxidizing conditions.
84
Previous work revealed that oxidation
or modification of these cysteine thiols decreases the binding affinity of HMGB 1 to various
DNA probes, including a cisplatin-modified one.34-36 Recently, the redox properties of HMGB1
domain A were investigated by NMR spectroscopy.3 7 The calculated redox potential of domain
A falls within the physiological intracellular redox potential range, which suggests that a
significant fraction of HMGB1 will exist in the oxidized form within cells. It is therefore
possible that the variability in cellular response to cisplatin as a function of HMG box proteins
will reflect the redox state of the cells.
In this chapter, the redox-dependent binding of full-length HMGB1 or HMGB1 domain
A (HMGBla) to DNA bearing a cisplatin 1,2-d(GpG) intrastrand cross-link will be delineated.
The binding properties of HMGB1, HMGBla, and their variants modified to prevent disulfide
bond formation were investigated under different redox conditions using electrophoretic mobility
shift assays (EMSAs) and hydroxyl radical footprinting analyses. The detailed changes in the
nature of binding and their dependence upon redox conditions are discussed in the light of the
previously reported crystal structure of HMGB 1 domain A bound to platinated DNA probe under
reducing conditions. This study allows an assessment of the importance of the HMGB 1 redox
state for the efficacy of the anticancer action of cisplatin and other bifunctional platinum-based
drugs.
3.2. Experimental
3.2.1. Expression and Purification of HMGB1 Proteins
Wild type HMGBla and its variants were expressed in E. coli from pET32 Xa/LIC.24
Codons for one or both of the cysteine residues in HMGB 1a were replaced with those for alanine
or serine by the Stratagene QuikChange site-directed mutagenesis kit protocol, and the modified
85
plasmids were transformed into BL21 (DE3) cells. The transformed cells were grown in LB
medium with 100 pg/mL ampicillin at 37 'C and 100 rpm. When OD600 was 0.5-1.0, IPTG was
added at a final concentration of 50 tM to induce gene expression. After induction, cells were
incubated at 28 'C for 14-16 h and harvested by centrifugation. Proteins were extracted by
sonication and purified with a Ni-NTA column (Novagen). N-terminal peptides including a
histidine tag were proteolytically cleaved from the eluted HMGBla protein variants by Factor
Xa and removed by passage down a Macro-prep high S cation exchange column (Bio-Rad).
Purified HMGBla and its variants were dialyzed against storage buffer (10 mM Tris-HCl, pH
8.0, 50 mM NaCl, 0.1 mM EDTA, 0.1 mM PMSF). For disulfide bond formation, protein
solutions were dialyzed overnight against storage buffer containing 5 IM CuCl 2 and then
redialyzed against buffer lacking CuCl 2. The absence of free thiols in wild type HMGB 1a was
confirmed by a Thiostar assay (Arbor Assays). For the expression of the full-length protein, an
HMGB 1 coding gene with an N-terminal His-tag in the pET30 Xa/LIC vector2 4 was transformed
into pET22b (+) for increased efficiency of expression and purification. In addition to mutations
at Cys22 and Cys44 in domain A, Cys105 was replaced by serine to prevent formation of
undesired disulfide bonds during oxidative dialysis. Proteins were expressed analogously to
HMGBla proteins, except that induction was carried out for 4 h at 37 'C. Following initial
purification on a Ni-NTA column, the N-terminal tag was removed by Factor Xa treatment, as
indicated above. Full-length HMGB1 proteins were isolated by using DEAE anion exchange
column. Purified proteins were dialyzed as for HMGB1a proteins. The concentrations of proteins
were determined by UV spectroscopic and BCA assays.
3.2.2. Platination, Purification, and Characterization of 25-bp DNA Probes
86
25-bp DNA duplexes with central platinum-binding sequences were used as DNA probes
for electrophoretic mobility shift and footprinting assays. To compare the binding affinity of
HMGBla to DNA probes with different platinated adducts, cisplatin 1,2-d(GpG) intrastrand
cross-linked, cisplatin 1,3-d(GpTpG) intrastrand cross-linked, and pyriplatin dG probes were
prepared (Table 3.1).
Table 3.1. Sequence of 25mer Oligonucleotides with Different Types of Cisplatin Binding
Sequences
Sequence
Probe (Type)
Oligonucleotide
25TGGA-PtA 2a
25TGGAts
5'- CCTCTCCTCTCTGGATCTTCTCTCC - 3'
(1,2-d(GpG))
25TGGAbs
5' - GGAGAGAAGATCCAGAGAGGAGAGG - 3'
25TGTG-PtA 2
25TGTGts
5'- CCTCTCCTCTCTGTGATCTTCTCTCC - 3'
(1,3-d(GpTpG))
25TGTGbs
5' - GGAGAGAAGACACAGAGAGGAGAGG - 3'
25TGTC-PtPy
25TGTCts
5'- CCTCTCCTCTCTGTCATCTTCTCTCC - 3'
(mono-dG)
25TGTCbs
5' - GGAGAGAAGAGACAGAGAGGAGAGG -3'
aBold
Gs are the platination sites.
All platinated top strands and their complementary bottom strands were purchased from
Integrated DNA Technologies and purified by preparative ion exchange HPLC on a Dionex
DNApac PA-100 (9 x 250 mm). To prepare for platination, cisplatin and pyriplatin were first
activated by incubating with 1.95 molar equiv or 0.95 molar equiv of AgNO 3, respectively, for 4
h at room temperature. Solid AgCl was removed via a 0.45 pm syringe filter. Purified top strands
were incubated with 1.2 molar equiv of activated platinum compounds at 37 'C overnight.
Platinated oligonucleotides were then purified by ion exchange HPLC (Figure 3.2). The DNA
concentration was determined by UV-Vis spectroscopy and the Pt/DNA ratio was calculated
based on the platinum concentration measured by flameless atomic absorption spectroscopy. The
nucleosides
and platinated guanosines
were quantitated by analytical
87
HPLC on a
SUPERCOSILMLC- 18-S (4.6
x
250 mm) column of SI-nuclease digestion (Promega) products
after further treatment by calf-intestinal phosphatase (CIP) reaction (New England Biolabs)
(Figure 3.3).
(A)
25TGGA-PtA 2
[NaCq= 0A56 M
**0
60-
25TGGA
= 0.480 M
D-[NaC
0
0
16
0
20
2-3mhvi
(B)
(C)
Figure 3.2. Ion exchange HPLC purification of platinated 25mer oligonucleotides. The platinated oligonucleotides
containing cisplatin 1,2-intrastrand cross-link (A), pyriplatin monofunctional adduct (B), or cisplatin 1,3-intrastrand
cross-link were separated by running a linear gradient of [NaCl] in 25 mM Tris-HCI (pH 7.4) buffer system. The peaks of
desired platinated product and non-platinated oligonucleotides are labeled together with the estimated NaCl
concentration at the time of elution. There are several multi-platinated byproducts in addition to the major product.
88
(A) mAU
200
150
100
50
0
0
(B)
mAu
120
30 min
C
T
Base
Calc
Obs
100
s8
20
40
20
0
C
13
13.1
G T
0 11
0 10.9
A
0
0
-1
0
10
20
()mAU
12U
(C)
Base
CCalc
100
so
Obs
W
30 min
C
G
T
A
12
0
11
0
12.1
0
10.9
0
40200
10
20
30 min
Figure 3.3. S1 nuclease digestion of platinated 25mer oligonucleotides. 25TGGAts-PtA 2 (A), 25TGTCts-PtPy (B), and
25TGTGts-PtA 2 (C) were digested by S1 nuclease and then dephosphorylation by CIP. Deoxynucleosides were separated
by running a linear gradient of methanol to 100 mM sodium acetate (pH 5.2) solution. The ratio of each deoxynuclesoide
was calculated based on the area of each peak.
3.2.3. Electrophoretic Mobility Shift Assays
The single-stranded complements of the strands containing a central platinated lesion
(~10 pmol) were radiolabeled at the 5' end with [y- 32 P]ATP by using polynucleotide kinase
(PNK). PNK was heat-deactivated at 70 *C for 30 min and removed by phenol extraction.
Radiolabeled oligonucleotides were annealed with cisplatin-modified or unmodified strand in
annealing buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1mM MgCl 2 ) by cooling the
temperature slowly from 95 'C to 4 'C over 3 h. Differing amounts of HMGB1 proteins were
89
incubated in a final volume of 15 ptL with the annealed probe in binding buffer (10 mM HEPES
pH 7.5, 10 mM MgCl 2 , 50 mM LiCl, 100 mM NaCl, 0.05% Nonidet P-40) for approximately 30
min on ice. DTT or
P-mercaptoethanol
was used as a reducing reagent at concentrations varying
from 100 pM to 10 mM. After incubating on ice, 1.5 tL of loading buffer (7% sucrose, 0.17%
ficoll 400, xylenol cyanol) was added prior to loading onto a 10% native polyacrylamide gel
(37.5:1 acrylamide:bisacrylamide). The gel was run at 300 V for 1.33 h with cooling by cold
water in 0.5x TBE, dried at 80 'C for 1 - 2 h, and visualized on a Storm Phosphorimager
(Amersham Bioscience). The dissociation constant Kd, defined
K
_
=
[P][D]
___
[P-D]
(1)
1
in eq 1, was computed, where [P], [D], and [P D] are the concentrations of protein, DNA, and
protein-DNA complex, respectively. The fraction of the protein-bound oligonucleotide, 0, was
derived from eq (1) as indicated in eq 2. The dissociation constant Kd was approximated as the
total protein concentration at the point in the titration where the fraction of protein-bound DNA
is 0.5. The Kd values for the different variants were determined as an average of at least three
independent trials.
6=
P
Ka + P
(2)
3.2.4. Hydroxyl Radical Footprinting Assay
Approximately 15 nM of DNA probe containing a radiolabel at the 5' end of the
complementary strand were incubated with the indicated amounts of protein on ice for 30 min 1 h in 20 ptL of binding buffer with or without added reducing reagent. A 5 pL aliquot of the
90
reaction solution was loaded onto a 10% native polyacrylamide gel and electrophoresis was
performed to confirm that most of the labeled DNA was bound to protein. Footprinting was
initiated by adding a mixture of 2 pL 100 mM sodium ascorbate, 2 ptL 1.5% H2 0 2 , and 2 pL 25
mM/50 mM Fe(II)(NH4) 2(SO4) 2/EDTA. After 2 min, the reaction was quenched by addition of
10 pL of a 1 M thiourea solution. Proteins and salts were removed by phenol extraction and
ethanol precipitation. A pellet containing the cleaved product was dissolved in 8 PL of loading
buffer (100% formamide, bromophenol blue) and resolved in a 15%
denaturing urea
polyacrylamide gel (19:1 acrylamide:bisacrylamide).
3.3. Results
3.3.1. Binding Affinity of Wild Type HMGB1a to Platinated DNA Under Different Redox
Conditions
The binding affinity of wild type HMGBla to DNA containing a cisplatin d(GpG)
intrastrand cross-link was examined under different redox conditions by EMSA. The absence of
free thiol in recombinant HMGBla was confirmed by a Thiostar assay. The fraction of the
reduced form of HMGBl a was less than 1/1000 of total protein. Purified wild type HMGBl a had
greater than 1000-fold higher binding affinity to cisplatin-1,2-d(GpG) intrastrand cross-linked
DNA compared to unmodified DNA under reducing conditions (Figure 3.4.A). In addition, there
was no significant binding of HMGBla to a platinated DNA probe bearing a 1,3-intrastrand
cross-link or a pyriplatin monofunctional adduct, even at micromolar concentrations of HMGB 1a
(Figure 3.4.B).
91
(A)
1
3
2
(B)
4
1,2-d(GpG)
1
2
3
4
6
5
8
7
9
1,3-d(GpTpG)
5
6
7
8
9
10
mono-dG
10
11
12
13
14 15
Figure 3.4. EMSA analysis of wild type HMGBla binding to the different DNA probes. (A) EMSA analysis of HMGBla
bound to cisplatin-1,2-intrastrand cross-linked (lanes 1,3,5,7,9) and unmodified (lanes 2,4,6,8,10) 25TGGA DNA probe
under reducing conditions (10 mM DTT). Protein concentrations were 0.3 nM (lanes 1,2), 3 nM (lanes 3,4), 30 nM (lanes
5,6), 300 nM (lane 7,8), and 3 stM (lanes 9,10). (B) HMGBla is incubated with 1 nmol of cisplatin-1,2-intrastrand crosslinked (lanes 1-5), cisplatin 1,3-intrastrand cross-linked (lanes 6-10), or pyriplatin monofunctionally platinated (lanes 1115) 25-bp DNA probe and analyzed by EMSA. Protein concentrations were I nM (lanes 1, 6, 11), 10 nM (lanes 2, 7, 12),
100 nM (lanes 3, 8, 13), 1 pM (lane 4, 9, 14), and 10 pM (lanes 5, 10, 15).
Under non-reducing conditions, the dissociation constant of oxidized HMGB 1a bound to
the platinated probe was computed to be 7.6 nM. This value did not change significantly when
H2 0 2 was added to the binding buffer, as expected from the Thiostar assay result. Under
reducing conditions, HMGB 1a exhibited a greater than 10-fold increase in binding affinity (Kd =
0.72 nM) compared to that observed under non-reducing conditions (Figure 3.5). There was no
significant change in binding affinity with different concentrations of the reducing agent
dithiotreitol (DTT) in the range from 0.1 to 10 mM, or when p-mercaptoethanol was used instead
92
of DTT (Figure 3.6.A). HMGB1a has a high binding specificity for cisplatin-modified DNA over
unmodified DNA.
Oxidized domain A
am
dot
1
2
3
4
Reduced domain A
Offt
oo
am
o.
.~ ~
5
6
~
7
8
9
10
11
12
13
14
1Z~
0.8 -
Wo
0.6OA 0.2-
o.-.00 .
0.1
.....
................
10
1
100
[Protln) (nM)
Figure 3.5. EMSA analysis of wild type HMGBla bound to the cisplatin-modified DNA probe under different redox
conditions. Various amounts of proteins were incubated under non-reducing condition or in the presence of 1 mM DTT
with a radiolabeled 25-bp probe containing a central 1,2-intrastrand cross-link (-0.1 nM) (top). Protein concentrations
were 0.25 nM (lanes 1,8), 0.5 nM (lanes 2,9), 1 nM (lanes 3,10), 2.5 nM (lane 4,11), 5 nM (lanes 5,12), 10 nM (lanes 6,13),
and 25 pM (lanes 7,14). The fraction of DNA probe bound to oxidized (blue) or reduced (red) domain A is plotted as a
function of protein concentration and the solid line is the fit of the data to eq 2 (bottom).
2.3.2. Site-Directed Mutagenesis of Cysteine Residues
To examine the role of cysteine residues on the binding of HMGB 1a to platinated DNA,
assays of singly or doubly mutated recombinant HMGBla proteins C22A, C44A, C44S,
C22A/C44A, and C22S/C44S were carried out. There was no evidence of dimer formation by
intermolecular disulfide bond formation involving two singly modified variants on non-reducing
SDS-PAGE. Unlike the wild type HMGB 1a, no variant displayed a significant difference in
93
binding affinity for reducing versus non-reducing conditions (Figure 3.6.B,C), due to an inability
to form the intramolecular disulfide bonds. All variants exhibited lower binding affinity
compared to wild type HMGBla under reducing conditions (Table 3.2).
(A)
no
tUincing agent
3
2
1
4
5
6
5 mM
5 mM
DTT
P-enercptoetanol
7
8
9 10
11 12 13 14 15
(B)
5 mM
2
1
4
3
5 mM
IB-erc ateha
DTT
no reduidng agent
5
9 10 1112 13 1415
8
7
6
(C)
no redudng agent
5 mM DTT
1
2
3
4
5
6
7
8
9
10 11 12 13 14
Figure 3.6. The influence of reducing reagent on the binding of HMGB1a to cisplatin-modified DNA. EMSA assays of 0.1
nM cisplatin-modified DNA probe with (A) wild-type HMGB1a at the concentrations of 0.16 nM (lanes 1, 6, 11), 0.8 nM
(lanes 2, 7, 12), 1.6 nM (lanes 3, 8, 13), 8 nM (lanes 4, 9, 14), and 16 nM (lanes 5, 10, 15), incubated with labeled probe in
binding buffer without reducing agent (lanes 1-5), in the presence of 5 mM DTT (lanes 6-10), or with 5 mM Pmercaptoethanol (lanes 11-15), (B) C22S/C44S doubly mutated HMGBla with the same concentration and redox
conditions as in (A), or (C) C44S variant of HMGB1a at the concentrations of 0.28 nM (lanes 1, 8), 0.56 nM (lanes 2, 9),
0.92 nM (lanes 3, 10), 1.84 nM (lanes 4, 11), 2.76 nM (lanes 5, 12), 4.6 nM (lanes 6, 13), and 9.2 nM (lanes 7, 14) under
reducing (lanes 1-7) or non-reducing (lanes 8-14) conditions.
Variants with serines replacing cysteines showed slightly higher binding affinity than
those having alanines, presumably because of the more conservative nature of the replacement.
94
Singly mutated variants have higher binding affinity than doubly mutated ones. The doubly
mutated variant C22A/C44A showed the lowest relative binding affinity, however, compared to
that of fully oxidized wild type HMGBla, its binding affinity is approximately 2-fold higher.
The dissociation constant of the C22A variant represented the strongest binding affinity to
cisplatin-modified DNA among domain A variants, although approximately 1.5-fold weaker than
wild type HMGBla under reducing conditions. Also, the C22S/C44S doubly mutated variant
showed similar binding affinity to the C44S singly mutated variant. These differences in binding
affinity indicate that the mutation of each cysteine into a different amino acid slightly decreases
the binding interaction between protein and DNA. This effect is not as strong as that of domain
A oxidation; presumably the influence of modifying Cys22 is slightly less than that of altering
Cys44.
Table 3.2. Dissociation Constants of Wild-Type and Variants of Domain A Binding to a 25-bp
Double-Stranded DNA Probe Containing a Site-Specific 1,2-d(GpG) Cross-Link as Determined
by Electrophoretic Mobility Shift Assays
Protein
Kd
(nM)
Wild type (red)
0.70
Wild type (ox)
7.58 ± 0.74
C44S
1.60
0.07
0.44
C22S/C44S
1.65 ± 0.37
0.42
C22A
1.12 ±0.14
0.63
C44A
2.41 ±0.31
0.29
C22A/C44A
3.40 ±0.74
0.21
Kd(WT,red) /Kd
1
0.10
0.092
95
3.3.3. Hydroxyl Radical Footprinting
The intercalating residue Phe37 is positioned at the start of helix II of domain A and
clamped by two guanine bases of the platinum cross-link. Because of this interaction, HMGB 1
domain A interacts with the DNA probe mainly downstream of the cisplatin lesion. The hydroxyl
radical footprinting pattern of HMGB1 extends to the 3' side of the 1,2-d(GpG) cross-link site,
reflecting this asymmetry. The resulting footprint provides a unique signature of HMGB1
domain A binding to platinated DNA that is not observed for the platinated DNA bound to most
of other HMG proteins, which display more symmetrical footprinting patterns with respect to the
platinated lesion.
The hydroxyl radical footprinting assays of wild type HMGB1a and its variants were
carried out in order to investigate possible changes in the protein-DNA interface induced by
formation of a disulfide bond or replacement of cysteines in the protein. Analysis of the EMSA
patterns from aliquots of DNA-protein mixtures prior to initiating the footprinting reaction
chemistry showed that greater than 95% of the DNA probe exists in a protein-bound state. The
hydroxyl radical footprinting patterns of wild type HMGB1a both under non-reducing and
reducing conditions are indistinguishable and display the asymmetry of the previous HMGB 1
footprint. In addition, none of the variants of domain A produced a footprinting pattern different
from that of wild type HMGBla (Figure 3.7). In a control study, a mobility shift assay of
oxidized domain A incubated with platinated DNA probes in binding buffer was performed in
the presence of additional hydrogen peroxide or sodium ascorbate, reagents used to generate the
hydroxyl radical (Figure 3.8). The results confirmed that these redox active agents do not alter
the redox state of the protein significantly during footprinting reactions.
96
Lane
1
2
3
Counts
5
T
C
T
C
A
T
4
Ol*
200mm
*
C
T
Counts
C
T
C
T
3
100
200mm
Figure 3.7. Footprinting analysis of HMGB1 domain A bound to the cisplatin-modified DNA probe. (A) Single-stranded
DNA was radiolabeled at the 5' end with [y-P3 2pATP and annealed to its complementary strand containing a 1,2intrastrand d(GpG) cisplatin cross-link. A 15 nM portion of double-stranded probe was incubated without any protein
(lane 1), with ~ 600 nM of fully oxidized wild type domain A under non-reducing conditions (lane 2), with 1 mM DTT
(lane 3), or C22A/C44A doubly mutated domain A (lane 4) followed by hydroxyl radical cleavage. (B) Plot of the intensity
of each cleaved DNA band from the hydroxyl radical footprinting experiments in A: lane 1 (black), lane 2 (blue), lane 3
(red), and Land 4 (green). All footprinting patterns matched that of a previously reported footprinting pattern of wild
type domain A under reducing conditions. (C) Plot of the intensity of each cleaved DNA band from the hydroxyl radical
footprinting study of oxidized/reduced forms of HMGB1 A domain. The footprinting patterns of DNA probe bound to
oxidized and reduced wild type domain are nearly identical.
97
(A)
2mM If^
1 mM DiT
1
2
5
4
3
6
7
9
8
(B)
MD medx eent
2
10
11
0.1 mM
1
2
3
4
5
10 mM
uodiuunm nscrtu
MD redox rament
1 mM DIf
7
6
8
14 15
12 13
9
udio
acrbimtn*
10
n
12
Figure 3.8. EMSA analysis of wild type domain A binding to cisplatin-modified DNA probe, incubated in the presence of
the reagents used to generate hydroxyl radicals. (A) Protein-DNA mixtures were incubated in binding buffer with the
reducing reagent DTT, oxidizing reagent H 20 2 , or no added reagent. (B) Protein-DNA mixtures were incubated in
binding buffer with DTT, sodium ascorbate, or with no added reagent. *Sodium ascorbate was added to the binding
buffer at a final concentration of 10 mM about 5 min before loading.
3.3.4. Studies of Full-Length HMGB1
Following purification by Ni-NTA and DEAE column, full-length wild type HMGB1
was dialyzed against storage buffer including 5 pM CuCl 2 for disulfide bond formation between
Cys22 and Cys44. The SDS-PAGE analysis under non-reducing conditions revealed two bands
with molecular weights of ~50 kDa and ~25 kDa. Complete conversion of the 25 kDa band to
the 50 kDa band was observed when the purified proteins were incubated with CuCl 2 at a final
concentration of 5 pM for 3 h at 4 'C, whereas DTT treatment converted the 50 kDa band to the
25 kDa band (Figure 3.9.A). Because the same treatment of HMGB 1a yielded the oxidized form
but not a dimer, it appears that the 50 kDa band is a result of an intermolecular disulfide bond
involving Cys105 in domain B. Cys105 was therefore replaced by serine using site-directed
98
mutagenesis to prevent such undesired dimer formation. Purified C 105S HMGB 1 was dialyzed
in the presence of 5 pLM CuCl 2. After being resolved by 18% non-reducing SDS-PAGE, oxidized
C 105S HMGB1 presented a single band (Figure 3.9.B, lanes 1,2) with no trace of dimer. There
was no change in electrophoretic mobility when the protein was treated with 1 mM diamide, a
thiol oxidizing reagent (Figure 3.9.B, lanes 3,4). Treatment with 1 mM DTT converted most of
the protein to the one with slower electrophoretic mobility (Figure 3.9.B, lane 5), which we
attribute to reduction of the disulfide bond between Cys22 and Cys44. Most of the wild type
HMGB 1, prior to oxidative dialysis, existed as the reduced form (Figure 3.9.B, lanes 6,7). The
electrophoretic mobility of a C22A/C44A/C105S triple mutant full-length HMGB1 is the same
as that of reduced, wild type HMGB 1 (Figure 3.9.B, lanes 8,9).
(A) 1 2
3
4
5 6
(B) 1 2
7 8
3
4 5 6
7 8 9
Figure 3.9. SDS-PAGE analysis of full-length HMGB1 in different redox states. (A) Purified wild type full-length HMGB1 was
resolved on a 4-20% gradient SDS gel containing no redox reagent (lanes 1,2), 10 mM DTT (lanes 3,4), 10 mM diamide (lanes
5,6), and 5 pM CuC12 (lanes 7,8). For each sample, 20 pmol (lanes 1, 3, 5, 8) or 40 pmol (lanes 2, 4, 6, 7) were loaded into the
wells. There is a band with higher molecular weight that is assumed to be the HMGB1 dimer observed in lanes 7 and 8. (B)
Various full-length HMGB1 proteins under different redox conditions were resolved on an 18% SDS gel. A C1O5S variant
dialyzed against buffer containing 5 pM CuC12 was loaded without any redox reagent (lanes 1,2), after incubation with 1 mM
diamide (lanes 3,4), or 1 mM DTT (lanes 5). For comparison, wild type full-length HMGB1 (lanes 6,7) and triply mutated
C22A/C44A/C1OSS protein (lanes 8,9) were resolved together. There are two bands with different electrophoretic mobilities.
3.3.6. Redox-Dependence of the Binding of Full-length HMGB1 to Cisplatin-Modified DNA
The binding of full-length HMGB1 and its variants was investigated under different
redox conditions. The binding of full-length HMGB1 to the 25TGGA-PtA 2 probe was
determined by EMSA analyses (Table 3.3). The
Kd
value of wild type HMGB1 was calculated to
be 110 nM in the presence of 10 mM DTT, in agreement with the previously reported value
99
(Figure 3.10.A).2 The Kd value of the C105S variant was determined to be 120 nM in the
presence of DTT, which is comparable to that of wild type protein. This variant exhibited a
greater than 10-fold decrease in binding affinity under non-reducing conditions (Figure 3.10.B).
To prevent precipitation of DNA due to a large protein-to-DNA ratio, an increased concentration
of 25TGGA-PtA 2 probe was used for the binding analysis of C105S under non-reducing
conditions. No significant change in binding affinity of HMGB1 for the 25TGGA-PtA 2 probe
was observed over a 1 to 10 mM DTT concentration range. The dissociation constant of triply
mutated C22A/C44A/C105S HMGB1, corresponding to the C22A/C44A doubly mutated variant
of HMGBla, was 240 nM. There is no redox-dependence in the case of C22A/C44A/Cl05S
HMGB1 (Figure 3.10.C).
Table 3.3. Dissociation Constants of Wild-Type and Variants of Full-Length HMGB1 Binding
to a 25-bp Double-Stranded DNA Probe Containing a Site-Specific 1,2-d(GpG) Cross-Link as
Determined by Electrophoretic Mobility Shift Assays
Protein
Kd
C105S (red)
120 ±40
1
C105S (ox)
1390±330
0.086
Wild type (red)
110 ±20
1.09
C22A/C44A/C105S
240± 50
0.50
(nM)
Kd(C1oSs,red)
100
/Kd
0.7
-
0.6
CIO5S variant
Wild Type HMGBI
(A)
0.s
inooqm
-
OA
Z0.3
0.2
0.1
1
2
3
5
4
6
7
8
10 U1213 14
9
IProli ngM)
(B)
No DTT
(C10%S, 1
10mMDTT
1mMDTT
0.8
I.e
006
OA
0.2
1
2
3
4
6
5
7
8
9
10 11
12 13 14 15
0
1
(C)
No DTT
mW
1
5 mM OTT
0.1 mM OTT
go
a
N4
0.8
UWPQq0.6
0.2
23
45
6
78
9 101112 13
14 15
0
[ProtIn](nM)
Figure 3.10. EMSA analysis of full-length HMGB1 and its variants bound to the cisplatin-modified DNA probe. (A) Wild
type and C105S HMGB1 were incubated with the 25TGGA-PtA 2 probe (-1 nM) in the presence of 10 mM DTT (left). The
protein concentrations of wild type HMGB1 were 7.2 nM (lanes 1), 24 nM (lane 2), 48 nM (lane 3), 120 nM (lane 4), 240
nM (lane 5), 480 nM (lane 6), and 720 nM (lane 7), and the concentrations of C105S variant were 9.6 nM (lane 8), 24 nM
(lane 9), 48 nM (lane 10), 96 nM (lane 11), 120 nM (lane 12), 240 nM (lane 13), and 480 nM (lane 14). The fraction of
protein-bound probe is plotted as a function of protein concentration (right). (B) After oxidative dialysis, C105S HMGB1
was incubated with 25TGGA-PtA2 (10 nM) in the non-reducing conditions (lanes 1-5) or in the presence of 1 mM (lanes 610) or 10 mM (lanes 11-15) DTT and analyzed by EMSA (left). The protein bound fraction of DNA is plotted as a function
of protein (right). Proteins concentrations were 16 nM (lanes 6, 11), 28 nM (lanes 7, 12), 80 nM (lane 8, 13), 120 nM (lane
9, 14), 160 nM (lanes 10, 15), 500 nM (lane 1), 1pM (lane 2), 2pM (lane 3), 3pM (lane 4), and 4pM (lane 5). (C) The
binding affinity of the triply mutated C22A/C44A/C105S HMGB1 to 25TGGA-PtA 2 (1 nM) was analyzed by EMSA in the
non-reducing conditions (lanes 1-5) or in the presence of 0.1 mM (lanes 6-10)/5 mM (lanes 11-15) DTT (left) and the
bound fraction of DNA is plotted as a function of protein (right). Proteins concentrations were 40 nM (lanes 1, 6, 11), 100
nM (lanes 2, 7, 12), 200 nM (lane 3, 8, 13), 400 nM (lane 4, 9, 14), and 600 nM (lanes 5, 10, 15).
101
3.4. Discussion
3.4.1. Interaction of Cisplatin-Modified DNA with HMGB1a in Different Redox States
HMGB1 plays critical roles in many important biological mechanisms, being active not
only within the nuclei of cells but also as a cytokine when released to the extracellular
matrix.1',,
HMGB1 contains two tandem HMG boxes, different in sequence but similar in
tertiary structure to one another. Both HMG box domains have a strong structure-specific
binding preference for non-linear DNA, including cisplatin-modified DNA. Interestingly, the
nature of the binding interaction of the two domains to DNA bearing a cisplatin 1,2-d(GpG)
intrastrand platinum cross-link is completely different, owing to the positions of the main
intercalating residues.4 0 '4 1 Intercalating residues are a conserved feature of HMG boxes that play
critical roles in stabilizing the complex and distorting DNA structure. In the case of cisplatinmodified DNA, one of the intercalating residues strongly interacts with the two platinated
guanines. When full-length HMGB1 binds to the cisplatin-modified DNA, it is mainly domain A
that interacts with the platinum lesion. 24 The Phe37 intercalating residue in domain A is
positioned at the beginning of helix II of the domain and, therefore, HMGB1 domain A mainly
interacts at the 3' side of the platinated lesion.'
40
The main intercalating residue in domain B,
and in most other HMG box proteins, is located in the middle of helix I, resulting in an overall
symmetrical binding pattems.
Under physiological conditions, HMGB1 can exist in two different redox states,
depending upon the presence or absence of a disulfide bond between Cys22 and Cys44 in
domain A.33 With the use of a precipitation assay using globally platinated DNA, an early study
established that the binding affinity of HMGB 1 decreases under oxidizing conditions and that the
102
oxidation involves cysteine residue not specifically identified in the protein.36 Our results
establish that the conversion between reduced and oxidized forms of domain A, involving Cys22
and Cys44 residues, regulates the binding interaction between HMGB 1 and the major cisplatinDNA adduct, a 1,2-d(GpG) intrastrand cross-link. Mutagenesis experiments involving these
cysteine residues provide clues about their roles in regulating the DNA binding properties of
HMGB 1 to cisplatin-modified DNA in the cellular context, as discussed below.
3.4.2. Change of HMGB1a Conformation Induced by Disulfide Bond Formation and
Protein Interaction with DNA
Cys22 and Cys44, which can form an intramolecular disulfide bond, are located at the
center of the helix I and helix II of HMGB1 domain A, respectively (Figure 3.1). The X-ray
crystal structure of the protein-DNA complex shows that those cysteines are located at the inner
side of the interface between two helices and face each other. The distance between the two
sulfur atoms in this structure is 4.25 A, which is approximately twice the length of a disulfide
bond.2 s This short distance implies that the overall conformation of the complex should not be
dramatically altered by disulfide bond formation. The footprinting patterns of oxidized wild type
domain A and domain A variants are identical to that of reduced HMGB1 domain A (Figure 3.7).
The similarity of footprinting patterns indicates a high degree of conservation of the interactions
between the residues in HMGB 1 domain A and the DNA probe in the complex, including the
critical n-n stacking and face-to-edge interactions between the Phe37 side chain and the two
platinated guanine bases.
Despite the similarity of the protein-DNA footprinting, there is a 10-fold weakening of
the binding affinity of HMGB 1a upon intramolecular disulfide bond formation. One possible
103
explanation for this phenomenon is that the two cysteine residues interact directly with the
platinated DNA and that lack of such an interaction in the oxidized form causes a decrease in
binding affinity. The distances between the cysteine residues and atoms in the DNA probe,
however, are too long for any direct interaction as revealed by the crystal structure of the
complex. Another possibility is that formation of a disulfide linkage alters the conformation of
domain A, possibly changing the alignment of the first and second helices and preventing
formation of the optimal conformation for DNA binding. There are several amino acid residues
close to two cysteines that interact both directly and indirectly with the DNA probe. A change of
relative positions of these residues with respect to the DNA, induced by disulfide bond formation
between the two cysteines, might be expected to destabilize the protein-DNA complex. Previous
studies revealed that the higher binding affinity of HMGB 1 under reducing conditions is not
limited to cisplatin-modified DNA, but is generally applicable to other non-linear DNA
probes.14 3,' Thus disulfide bond formation may influence the binding of HMGB 1 to other forms
of DNA probes in a similar manner. Unlike the case of wild type domain A, the presence of a
reducing reagent does not influence the binding affinity of domain A variants to cisplatinmodified DNA probes, which suggests that no other redox chemistry except the breaking of a
disulfide bond occurs under the conditions used in the present study. Cys22 and Cys44 are
conserved in three main HMGB proteins, HMGB1, 2, and 3, which suggests the disulfide bond
formation may play a pivotal role in regulating functions of HMGB proteins as transcription
factors.
3.4.3. Influence of Cysteines on Conformation of HMGB1a
The variants of domain A do not exhibit redox-dependent changes in binding affinity for
platinated DNA, but their interaction with the cisplatin-modified duplex is not as strong as that of
104
the reduced wild type protein. The latter property is presumably a consequence of the lack of
stabilizing interactions between the cysteine mercaptomethyl side chains and the neighboring
amino acids with the platinated DNA. There is a hydrophobic core at the interface between helix
I and helix II, and two cysteines are positioned at the center of the core (Figure 3.1).40 It is not
uncommon that replacement or modification of cysteines by other residues in a hydrophobic
environment decreases the stability and activity of proteins.42
Both a previous NMR study of
the C22S variant of domain A45 and the X-ray crystal structure of wild type domain A25 reveal
the side chain of Cys44 to be buried in the hydrophobic core of the protein, whereas that of
Cys22 in the unmodified protein and Ser22 in the C22S variant are more exposed to solvent and
lie outside of the helix I/helix II interface. The relatively small decrease in binding affinity
induced by the replacement of the Cys22 supports this analysis. Because the effect of the
mutations on binding affinity is less than that of oxidizing the cysteine residues, we conclude that
disruption of hydrophobic interactions between the protein and platinated DNA is less important
than formation of a disulfide linkage.
3.4.4. Binding Properties of Reduced and Oxidized Full-Length HMGB1 and its Variants
The effect of the three cysteine residues in HMGB 1 on DNA binding were investigated in
previous in vitro studies. 3 3 A significant increase of cytosolic HMGB1 was observed when
Cys 105 was replaced by serine, suggesting that this residue is an essential factor for the nuclear
transport of HMGBl. The effect of replacing Cys22 or Cys44 by serine on the cellular
distribution of HMGB1 was negligible. On the other hand, Cys22 and Cys44 readily formed a
disulfide bond when the cells were incubated with 0.5 mM diamide prior to protein extraction,
whereas intra/intermolecular disulfide formation involving Cys105 was not observed under the
same conditions. This intramolecular disulfide bond increases the electrophoretic mobility of
105
HMGB 1 on denaturing gels, presumably by preventing complete disruption of the tertiary
structure (Figure 3.9).
Cys 105 is located in the middle of helix I in domain B, close to the intercalating residue
Phe102. The affinities of wild type HMGB1 and its C105S variant to platinated DNA were not
significantly different, implying that Cys105 is not critical for interaction between full-length
HMGB 1 and platinated DNA. Under non-reducing conditions, the C105 S variant of full-length
HMGB 1 exhibits micromolar binding affinity that is more than 10 times weaker than the value
measured under the reducing conditions. This trend is similar to that for HMGBl a. This result
suggests that the decrease of binding affinity of domain A by oxidation is directly reflected in
that of full-length HMGB 1.
3.4.5. In Vivo Redox Chemistry of HMGBI
HMGB1 is a DNA-binding protein that recognizes cisplatin 1,2-intrastrand cross-links
with a high preference over unmodified DNA. In vitro repair assays revealed a dramatic
reduction in NER efficacy in the presence of excess HMGB1, and binding of HMGB1 might
function to increase the anticancer efficacy of cisplatin by shielding the lesions from repair
proteins. Unlike studies in vitro, however, attempts to alter the sensitivity of cells to cisplatin by
controlling HMGB1 expression levels have not been uniformly successful, with the results
depending on conditions and the types of cell lines used in the individual studies.
Disulfide bonds are the most common covalent linkages between protein side chains.
Thiol-disulfide conversions can change important properties of proteins, including local/global
conformations and metal-binding affinity in a redox-dependent manner and thereby regulate
many critical intracellular functions. 47' 48 The stability of the disulfide bond under physiological
106
conditions is highly sensitive to the redox potential of the cysteine/cystine redox pair, which is
determined by the protein local environment.
HMGB1 domain A has two cysteine residues adjacent to each other in its tertiary
structure (Figure 3.1). The standard redox potential of the intramolecular disulfide bond of
domain A is - 237 ± 7 mV (vs. NHE), as determined in an NMR study investigating the redox
properties of the protein.37 This value is in the normal cellular range of redox potentials as
determined by the redox state of the GSH/GSSG couple. 49 Therefore, if the redox potential of
full-length HMGB1 does not substantially differ from that of domain A, even subtle alterations
of intracellular redox potential at different stages in the cell cycle or in different types of the cell
will dramatically alter the ratio of oxidized to reduced levels of HMGBl. An increase of the
fraction of oxidized HMGB1 will decrease the binding strength of HMGB1 to the platinated
lesion and consequently diminish its ability to shield the damaged duplex from the excision
repair protein. It is therefore conceivable that variations in the redox state of the HMGB1 can
explain the controversial results in previous attempts to establish a correlation between
expression levels of HMGB1 and cisplatin cytotoxicity.
Recently, HMGB4, a new member of HMGB protein superfamily, was discovered.
HMGB4 is expressed exclusively in testis. Unlike other HMGB proteins, there is a tyrosine
instead of a cysteine in position 22 in human HMGB4. Because HMGB4 does not have a Cterminal acidic tail which decreases the binding affinity of other HMGB proteins to DNA, we
expect that HMGB4 would display stronger, redox-independent binding to platinated DNA. The
interaction of HMGB4 and platinated DNA, as well as its possible role in promoting cisplatin
chemotherapy of testicular germ cell tumors, will be described in Chapter 4.
107
3.5. Conclusion
In this investigation the influence of a change in the Cys/Cys-Cys redox state on the
interaction of the main DNA-binding element of HMGB1 and cisplatin-modified DNA was
evaluated by EMSA and DNA footprinting analyses. HMGB1 domain A exhibits notably strong,
nanomolar binding affinity for a 1,2-intrastrand d(GpG) cisplatin cross-linked DNA probe.
Intramolecular disulfide bond formation between Cys22 and Cys44 decreases the binding
affinity by an order of magnitude. This decrease in the binding affinity was proved to arise
exclusively from disulfide bond formation. The dissociation constant of domain A variants
having one or both of the cysteines replaced by alanine or serine was unaffected by the external
redox environment. Compared to the reduced, wild type domain A, the binding affinities of the
variants decreased, presumably through disruption of hydrophobic interactions between two
helices I and II involved in DNA binding. Hydroxyl radical footprinting patterns of platinated
probes bound to domain A under different redox conditions reveal that its overall positioning on
the platinated duplex does not change significantly with redox state. Given the results of these
binding studies and the structural properties of the domain A/platinated DNA complex,
restriction of the relative alignment of helices I and II caused by disulfide bond formation
appears to be the major factor affecting the binding affinity.
108
3.6. References
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
(29)
(30)
(31)
(32)
Jamieson, E. R.; Lippard, S. J. Chem. Rev. 1999, 99, 2467-2498.
Bosl, G. J.; Motzer, R. J. N. Engl.J. Med. 1997, 337, 242-253.
Cohen, G. L.; Ledner, J. A.; Bauer, W. R.; Ushay, H. M.; Caravana, C.; Lippard, S. J. J.
Am. Chem. Soc. 1980, 102, 2487-2488.
Davies, M. S.; Berners-Price, S. J.; Hambley, T. W. J. Inorg. Biochem. 2000, 79, 167-172.
Shimizu, M.; Rosenberg, B. J. Antibiot. 1973, 26, 243-245.
Harder, H. C.; Smith, R. G.; Leroy, A. F. Cancer Res. 1976, 36, 3821-3829.
Corda, Y.; Job, C.; Anin, M. F.; Leng, M.; Job, D. Biochemistry 1993, 32, 8582-8588.
Mello, J. A.; Lippard, S. J.; Essigmann, J. M. Biochemistry 1995, 34, 14783-14791.
Mymryk, J. S.; Zaniewski, E.; Archer, T. K. Proc. NatL. Acad. Sci. U. S. A. 1995, 92,
2076-2080.
Siddik, Z. H. Oncogene 2003, 22, 7265-7279.
Dempke, W.; Voigt, W.; Grothey, A.; Hill, B. T.; Schmoll, H. J. Anti-Cancer Drugs 2000,
11, 225-236.
Richon, V. M.; Schulte, N.; Eastman, A. CancerRes. 1987, 47, 2056-2061.
Lai, G. M.; Ozols, R. F.; Young, R. C.; Hamilton, T. C. J. Natl. Cancer Inst. 1989, 81,
535-539.
Wong, E.; Giandomenico, C. M. Chem. Rev. 1999, 99, 2451-2466.
Shlebak, A. A.; Clark, P. I.; Green, J. A. Cancer Chemother. Pharmacol.1995, 35, 349351.
Giese, K.; Cox, J.; Grosschedl, R. Cell 1992, 69, 185-195.
Thomas, J. 0.; Travers, A. A. Trends Biochem. Sci. 2001, 26, 167-174.
Bianchi, M. E.; Beltrame, M.; Paonessa, G. Science 1989, 243, 1056-1059.
Thomas, J. 0. Biochem. Soc. Trans. 2001, 29, 395-401.
Pil, P. M.; Lippard, S. J. Science 1992, 256, 234-237.
Fichtinger-Schepman, A. M. J.; van der Veer, J. L.; den Hartog, J. H. J.; Lohman, P. H.
M.; Reedijk, J. Biochemistry 1985, 24, 707-713.
Eastman, A. Pharmacol.Ther. 1987, 34, 155-166.
Kasparkova, J.; Delalande, 0.; Stros, M.; Elizondo-Riojas, M. A.; Vojtiskova, M.;
Kozelka, J.; Brabec, V. Biochemistry 2003, 42, 1234-1244.
Jung, Y.; Lippard, S. J. Biochemistry 2003, 42, 2664-2671.
Ohndorf, U. M.; Rould, M. A.; He, Q.; Pabo, C. 0.; Lippard, S. J. Nature 1999, 399, 708712.
He, Q.; Liang, C. H.; Lippard, S. J. Proc.Natl. Acad. Sci. U. S. A. 2000, 97, 5768-5772.
Zamble, D. B.; Mikata, Y.; Eng, C. H.; Sandman, K. E.; Lippard, S. J. J. Inorg.Biochem.
2002, 91, 451-462.
Brown, S. J.; Kellett, P. J.; Lippard, S. J. Science 1993, 261, 603-605.
Huang, J. C.; Zamble, D. B.; Reardon, J. T.; Lippard, S. J.; Sancar, A. Proc. Natl. Acad.
Sci. U. S. A. 1994, 91, 10394-10398.
Ugrinova, I.; Zlateva, S.; Pashev, I. G.; Pasheva, E. A. Int. J. Biochem. Cell Biol. 2009,
41, 1556-1562.
Zamble, D. B.; Mu, D.; Reardon, J. T.; Sancar, A.; Lippard, S. J. Biochemistry 1996, 35,
10004-10013.
Wei, M.; Burenkova, 0.; Lippard, S. J. J. Biol. Chem. 2003, 278, 1769-1773.
109
(33)
Hoppe, G.; Talcott, K. E.; Bhattacharya, S. K.; Crabb, J. W.; Sears, J. E. Exp. Cell Res.
(34)
2006, 312, 3526-3538.
Sheflin, L. G.; Fucile, N. W.; Spaulding, S. W. Biochemistry 1993, 32, 3238-3248.
(35)
(36)
Urbonaviciute, V.; Meister, S.; Ffrnrohr, B. G.; Frey, B.; G ckel, E.; Schett, G.;
Herrmann, M.; Voll, R. E. Autoimmunity 2009, 42, 305-307.
Billings, P. C.; Davis, R. J.; Engelsberg, B. N.; Skov, K. A.; Hughes, E. N. Biochem.
Biophys. Res. Commun. 1992, 188, 1286-1294.
(37)
Sahu, D.; Debnath, P.; Takayama, Y.; Iwahara, J. FEBS Lett. 2008, 582, 3973-3978.
(38)
(39)
Stros, M. Biochim. Biophys. Acta 2010, 1799, 101-113.
Bianchi, M. E.; Manfredi, A. A. Immunol. Rev. 2007, 220, 35-46.
(40)
(41)
(42)
(43)
He, Q.; Ohndorf, U. M.; Lippard, S. J. Biochemistry 2000, 39, 14426-14435.
Murphy, F. V. t.; Sweet, R. M.; Churchill, M. E. EMBOJ. 1999, 18, 6610-6618.
Jaiswal, R.; Panda, D. ProteinSci. 2008, 17, 846-854.
Schuman, B.; Persson, M.; Landry, R. C.; Polakowski, R.; Weadge, J. T.; Seto, N. 0. L.;
Borisova, S. N.; Palcic, M. M.; Evans, S. V. J. Mol. Biol. 2010, 402, 399-411.
(44)
Tammam, S. D.; Rochet, J. C.; Fraser, M. E. Biochemistry 2007, 46, 10852-10863.
(45)
Hardman, C. H.; Broadhurst, R. W.; Raine, A. R. C.; Grasser, K. D.; Thomas, J. 0.; Laue,
(46)
(47)
(48)
(49)
E. D. Biochemistry 1995, 34, 16596-16607.
Dunham, S. U.; Lippard, S. J. Biochemistry 1997, 36, 11428-11436.
Hogg, P. J. Trends Biochem. Sci. 2003, 28, 210-214.
Maret, W. Proc.NatL. Acad. Sci. U. S. A. 1994, 91, 237-24 1.
Schafer, F. Q.; Buettner, G. R. FreeRadic. Biol. Med. 2001, 30, 1191-1212.
110
Chapter 4. HMGB4, a Potential Enhancer of Cisplatin in Testicular Germ Cell Tumors
111
4.1. Introduction
The introduction of cisplatin in the 1970's brought dramatic changes to cancer
chemotherapy.1 The most revolutionary change was for treatment of testicular germ cell tumors
(TGCTs). The 5-year survival rate of TGCT patients increased from 72% in 1970-1973 to 91%
in 1983-1985 after the introduction of cisplatin. 2 Today, the cure rate of testicular cancer is
nearly 100% if discovered at an early stage. 3 Even in highly metastatic stage 3 of TGCT, the 5year survival rate is about 74%, compared to less than 30% for most other metastatic cancers. 3
The origin of the hypersensitivity of TGCT to cisplatin is not fully understood. Previous attempts
to understand this phenomenon have focused on proteins that are uniquely expressed in TGCT
and therefore may give rise to this hypersensitivity to cisplatin.
HMGB proteins recognize and bind to cisplatin-damaged DNA.'
In the mouse, the
expression profile of HMGB2 depends on age and type of tissue. For example, HMGB2 is highly
expressed in mouse embryos, but significantly decreases in adult mice, where a high expression
level is maintained only in the lymphoid organs and testes. 6 The dramatic increase of HMGB2
expression during the differentiation of testis germ cells 7 '8 implicates it as a possible reason for
cisplatin hypersensitivity in testis cells. This hypothesis is supported by a study using a human
non-small cell lung cancer cell line that showed a significant increase in cisplatin sensitivity
when the HMGB2 expression level was artificially enhanced by transfection of an HMGB2
expression vector.9 It is noteworthy, however, that the expression levels of HMGB2 are high in
many human tumors compared to non-cancerous cells,10'11 and therefore it is uncertain whether
HMGB2 plays a unique role in TGCT. Moreover, the prevalence of the closely related DNAdamage binding protein, HMGB 1, further calls into question the possibility that HMGB2 induces
cisplatin hypersensitivity in TGCT.
112
Recently, HMGB4 was discovered as a new member of mammalian HMGB protein
family.12 HMGB4 is preferentially expressed in the testis as revealed by profiling mouse
HMGB4. It is of considerable interest that this new HMGB protein is expressed only in tissues
that are hypersensitive to cisplatin. Although there is currently no structural information
available for HMGB4, its amino acid sequence offers several clues about its likely binding
interaction to platinated DNA (Figure 4.1.A). Most notably, HMGB4 lacks the long, acidic Cterminal tail present in HMGB1-3, homologous proteins of HMGB4, which significantly
weakens the DNA-binding affinity of HMG box domains (Figure 4.1 .B).13 Given this distinctive
structural characteristic, we anticipated that HMGB4 would bind with greater affinity to
cisplatin-modified DNA, which in turn might enhance its repair shielding properties.
(A)
Heft
I
11
EGlaPQDHKGB4a
|
Helix 2!
31
21
t
Henx 3
61
41
0
In
71
*
(B)
0
0
165 185
35 39
214
HMGB1domain B
HMGB1 domain A
HMGBI
181
3587
B
HMGB4domain
0
8
HMGBdomain A
165
87
HMGB4doma
B
Bdoai
Figure 4.1. Sequences and structures of DNA binding motifs in HMGB1 and HMGB4. (A) Sequence alignment of the
DNA-binding domains of HMGB1 and HMGB4. All domains share the same basic structure consisting of three alpha
helices. Residues with brown or light orange background represent identical or homologous amino acids, respectively.
Two possible positions for the intercalating residues are marked with red boxes. (B) Schematic representation of HMGB1,
HMGB4, and two DNA-binding domains of HMGB4 used in this study.
In this chapter, the nature of the binding interaction between HMGB4 and cisplatinmodified DNA as well as the inhibitory function of HMGB4 on repair of the platinated lesion
will be described. HMGB4 and its two DNA binding domains were examined using recombinant
113
mouse HMGB4 and a site-specifically platinated DNA probe carrying different platinum-DNA
adducts. For these experiments we employed electrophoretic mobility shift (EMSAs) and
footprinting assays. The repair inhibitory effects were measured in vitro with repair-active cell
free extracts (CFEs) in the presence of HMGB4 proteins. The difference between HMGB4 and
HMGB 1, the most intensively investigated HMGB protein, are discussed. These in vitro studies
strongly suggest an intracellular function of HMGB4 in processing cisplatin-DNA adducts, with
consequences for the mechanism of action of the drug including its remarkable ability to cure
testicular germ cell cancer.
4.2. Experimental
4.2.1. Cloning and Expression of HMGB4 Proteins
cDNAs encoding full-length mouse HMGB4, HMGB4 domain A (HMGB4a), and
HMGB4 domain B (HMGB4b) (Figure 4.1.B) were amplified by PCR from pXJ41-HMGB4
plasmid,12 kindly provided by Dr. Irwin Davidson, Institut de Gendtique et de Biologie
Moleculaire et Cellulaire, Illkirch, France, with an NdeI and XhoI restriction sites at the 5' and 3'
ends, respectively. Amplified genes were digested by these enzymes and inserted into a
pET22b(+) vector. For protein purification, HMGB4 genes were cloned with (His) 6 tags at their
C-termini. A full-length HMGB4 gene lacking the C-terminal (His) 6 tag was also prepared to
evaluate potential influence of (His) 6 tag on the platinated DNA-binding properties of HMGB4.
Site-directed mutagenesis was performed according to the Strategene Quick change protocol to
create an F37A variant of full-length HMGB4 and HMGB4a. Cloned plasmids were transformed
into Rosetta (DE3) cells and grown in LB medium at 100-150 rpm at 37 'C until the OD 60 0
reached 0.8-1.0. After the addition of IPTG to a final concentration of 50 pM, cells were
114
incubated at 25 'C for 12-15 h, then harvested by centrifugation. Proteins were extracted by
sonication in lysis buffer (25 mM Tris-HC, pH 7.5, 50 mM NaCl, 1 mM MaCl 2, 10% v/v
glycerol). The (His) 6 tag proteins were purified first with a Ni-NTA column (Novagen) and then
on a Macro-Prep High S cation exchange column (Bio-Rad). Full-length HMGB4 without a
(His) 6 tag was first purified using a Macro-Prep High S column followed by an S75 size
exclusion column. Purified proteins were dialyzed against storage buffer (10 mM Tris-HCl, pH
7.5, 50 mM NaCl, 0.1 mM PMSF) and kept at -80 'C until use. Protein concentrations were
determined by UV-Vis spectroscopy and the Bradford assay.
4.2.2. Preparation of 25-bp Site-Specifically Platinated DNA Probes
A series of 25mer oligonucleotide probes containing a cisplatin 1,2-intrastrand d(GpG),
1,3-intrastrand d(GpTpG) cross-link, or pyriplatin-dG adduct were synthesized and purified as
described in Chapter 3. Strands complementary to the platinated oligonucleotides were
radiolabeled at their 5' ends with [y- 32P]ATP using PNK (New England Biolabs). Platinummodified oligonucleotides, 25TGGAts-PtA 2 , 25TGTGts-PtA 2 , 25TGTCts-PtPy, and unplatinated
25TGGAts were annealed to their complementary, radiolabeled strands by heating to 90 'C and
cooling to 4 'C over 4 h. The annealed products were precipitated with ethanol precipitation, redissolved in dH 20, and stored at -20 'C.
4.2.3. Preparation of a Repair Assay Substrate
115
(A)
63mer
5' -ATCAATATCCACCTGCAGATTCTACCAAAAGTGTATTTGGAAACTGCTCCATCAAAAGGCATG-3'
14ezr
5' -TTCACCGGAATTCC-3'
69uer
5'-CCTCAACATCGGAAAACTACCTCGTCAAAGGTTTATGTGAAAACCATCTTAGACGTCCACCTATAACTA-3'
86mer
5'-ATGTTGAGGGGAATTCCGGTGAACATGCCTTTTGATGGAGCAGTTTCCAAATACACTTTTGGTAGAATCTGCA
GGTGGATATTGAT-3'
60mer
5' -TAGTTATAGGTGGACGTCTAAGATGGTTTTCACATAAACCTTTGACGAGGTAGTTTTCCG-3'
(B)
14mer-Pt
*
63mer
69mer
i
LiUgation
86mer
5'
60mer
5'
NER Product
*
Exicison of damaged site
5'
5
DNA synthesis
5'
Figure 4.2. Preparation of in vitro NER substrates. (A) Sequences of oligonucleotides used to prepare a 146-bp repair
substrate. (B) Schematic diagram of the synthesis and repair of the linear DNA substrate with a central-1,2-intrastrand
cisplatin cross-link.
The site-specifically modified and unmodified 146-bp DNA substrates for use in repair
assays were prepared as previously described (Figure 4.2).14 The 14mer oligonucleotides with
and without a central cisplatin 1,2-d(GpG) cross-link were radiolabeled at their 5' ends with [y32P]ATP.
Labeled oligonucleotides were annealed with 2 molar equiv of the remaining,
unlabeled oligonucleotides in annealing buffer (50 mM Tris-HCl, pH 7.9, 100 mM NaCl, 10 mM
MgCl 2, 1 mM DTT), as described in the previous section. Annealed products were ligated at
16 *C overnight and purified by 8% urea-PAGE. The full-length 146mer substrate was isolated
from the gel in extraction buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA),
116
ethanol-precipitated, and re-annealed. The purity of the substrate was confirmed by 8% ureaPAGE and 5% Native-PAGE.
4.2.4. Nucleotide Excision Repair Assay
Approximately 10-20 fmols of 146-bp substrate was incubated with the protein of interest
in 20 jiL of NER reaction buffer (20 mM HEPES, pH 7.9, 24 mM KCl, 2 mM MgCl 2 , 4 mM
ATP, 20 pLM dNTP, and 200 jig BSA) on ice for 30 min and at 30 'C for 10 min. A 5 ptL portion
of CHO AA8 CFE solution including 50 jig of protein was added to each sample. The reactions
were performed at 30 'C for the indicated time period and then quenched by addition of stop
solution, which included 1 pig of proteinase K and SDS at a final concentration of 0.34% w/v; the
solution was then incubated at 60 *C for 15 min. DNA in the reaction solution was purified by
phenol/chloroform extraction followed by ethanol precipitation. Repair products were separated
at 1800 V for 1.5 h in 8%urea-PAGE in 1xTBE, dried at 80 'C for 1-2 h and visualized on a
Storm phosphorimager (Amersham Bioscience).
4.2.5. Electrophoretic Mobility Shift Assay
The binding affinity of HMGB4 proteins was measured by the electrophoretic mobility
shift assay (EMSA) as previously described. 15 Briefly, a radiolabeled 25-bp DNA probe and
different amounts of HMGB4 proteins were incubated in 15 jiL of binding buffer (10 mM
HEPES, pH 7.5, 10 mM MgCl 2 , 50 mM LiCl, 100 mM NaCl, 0.05% Nonidet P-40, 0.2 mg/mL
BSA) containing 50 ng of poly d(GC) as a competitor. The unbound DNA probe and proteinDNA complexes were separated by 10% Native-PAGE for 1.5 h with cooling to about 13 'C in
0.5x TBE at 300 V, dried, and visualized. The apparent dissociation constant, Kd, of each protein
was calculated as previously described in Chapter 3.
117
4.2.6. Hydroxyl Radical Footprinting
Footprinting analyses were carried out by a previously described method 15 with some
modifications. Different amounts of protein were mixed with 10-15 nM of platinated 25TGGA
probe for 1 h on ice in 20 pL of 1xEMSA binding buffer containing 100 ng poly d(GC) and 10
gg BSA. After incubation, 5 gL of each sample was analyzed by 10% Native-PAGE to confirm
that > 90% of the DNA existed in the form of a DNA-protein complex. The footprinting reaction
was initiated by adding 2 gL of 25 mM/50 mM Fe(NH 4)2 (SO4)2/EDTA, 100 mM sodium
ascorbate, and 1.5% H20 2. After 3 min at room temperature, the reaction was quenched by
adding 10 pL of 1 M thiourea solution. A 5 pL solution of 10% SDS was added to each
footprinting sample, which was incubated at 37 'C for 20 min, and then the DNA was extracted
by phenol/chloroform and precipitated by ethanol addition. The samples were subsequently airdried, re-dissolved in 8-RL formamide solution containing bromophenol blue, and resolved by 20%
urea-PAGE in 1x TBE buffer for 3.5 h at 1800 V.
4.3. Results
4.3.1. Expression and Solubility of HMGB4 Proteins
E. coli strain Rosetta (DE3) (Novagen) was used to express the HMGB4 proteins because
of the high frequency of rare codons in the HMGB4 cDNA. After purification, HMGB4 proteins
were resolved on 4-20% gradient SDS gels and stained with Comassie Blue to confirm the purity.
The full-length HMGB4, with or without a His tag, electrophoresed with a mobility
corresponding to a molecular weight of around 22 kDa. The two DNA-binding domain HMGB4a
and HMGB4b bands were also observed in a region corresponding to 10-11 kDa (Figure 4.3).
Full-length HMGB4 without the C-terminal (His)6 tag exhibited slightly faster electrophoretic
mobility than the protein with the tag (Figure 4.3, lane 2). HMGB4 proteins are quite soluble in
118
storage solution with low salt concentration. No significant precipitation or turbidity was
observed when the protein was concentrated to 100 gM.
M
1
2
3
4
5
6
40kWe -+w
30kWe -+o
25kWe -+P
20 kDa .*
10+
Figure 4.3. Purified full-length HMGB4 and its DNA binding domains resolved in 4-20% gradient SDS gels. Full-length
mouse HMGB4 (lane 1), human HMGB4 (lane 2), mouse HMGB4 with C-terminal (His)6 tag (lane 3) displayed
electrophoretic mobility corresponding to the molecular weight close to 20 kDa on the gel. Mouse HMGB4 without the tag
(21.6 kDa) displayed slightly faster mobility than human HMGB4 (22.4 kDa) or mouse HMGB4 with (His) 6 tag (22.4 kDa).
F37A (lane 4, 11.4 kDa) and wild type (lane 5, 11.4 kDa) HMGB4a or HMGB4b (lane 6, 10.2 kDa) with (His) 6 tag were
observed around the 10 kDa marker band.
4.3.2. Binding of HMGB4 Proteins to Platinated DNA
The binding affinities of full-length HMGB4 and its DNA binding domains were
investigated by EMSAs. Full-length HMGB4 binds to the 25-bp DNA probes with a strong
preference for cisplatin 1,2-intrastrand cross-linked adducts, as revealed by its nanomolar
binding affinity (Kd = 4.35 nM). Very little binding occurs at the cisplatin 1,3-intrastrand crosslinked adduct (Kd > 1 VM) (Figure 4.4.A). No binding of HMGB4a to the DNA probe with a
pyriplatin monoadduct was observed below protein concentrations of 5 IM (Figure 4.4.B). There
is no significant difference in binding affinities for full-length HMGB4 containing a C-terminal
(His) 6 tag compared to protein lacking the tag for the 1,2-intrastrand cross-linked DNA (Figure
4.5.A).
119
(A)
1,3-intrastmnd
I ,2-intrastrand
1
2
4
3
5
6
7
8
9
10
Unmodified
11 12
13
14
15
Mono-dG
1
2
3
4
5
Figure 4.4. EMSA analysis of HMGB4 and HMGB4a bound to different probes. (A) Binding preference of full-length
HMGB4 to 1,2-intrastrand versus 1,3-intrastrand cross-linked DNA as revealed by EMSAs. Variable amounts of fulllength HMGB4 protein were incubated with site-specifically platinated 25TGGA (lane 1-5), 25TGTG (lane 6-10), or
unplatinated 25TGGA DNA (lanes 11-15) probes (~1 nM). Protein concentrations in each set were 10 nM (lanes 1, 6, 11),
20 nM (lanes 2, 7, 12), 40 nM (lanes 3, 8, 13), 100 nM (lanes 4, 9, 14), and 200 nM (lanes 5, 10, 15). (B) Binding preference
of HMGB4a to pyriplatin monofunctional DNA adducts. Whereas a significant fraction of cisplatin-modified 25TGGA
exists as protein-bound form after it was incubated with 5 nM of HMGB4a (lane 1), no complex was observed when
pyriplatin-platinated 25TGTC was incubated with 5 nM (lane 2), 50 nM (lane 3), 500 nM (lane 4), and 5 pM (lane 5) of
HMGBla.
The dissociation constant for the HMGB4a/platinated DNA complex is about twice that
of the full-length HMGB4/platinated DNA complex (Table 4.1). Moreover, the binding
interaction of HMGB4a does not change in the presence of 1-10 mM DTT (Figure 4.5.B), unlike
HMGB 1 domain A, which has a redox-dependent binding affinity owing to the presence of
adjacent cysteine residues. 16 One of these residues is absent in HMGB4 proteins (Figure 4.1.A).
The binding affinity of HMGB4a is more than 50-fold greater than that of HMGB4b (Figure
4.5.C and Table 4.1). Both DNA binding domains show a specificity for DNA carrying a 1,2-
120
intrastrand d(GpG) cross-link, but not a 1,3-intrastrand cross-link, as is the case for full-length
HMGB4.
(A)
HMGB4+(His)g'
0.8
HMGB4
-
0
0
02
0
1
2
3
4
5
6
7
8
9
10
!ProteinI(nM)
1
(B) No DTT (HMGB4
ox)
5mM DTT (HMGB4a red)
C
0
C
0
m
-mn.
1
0
0.
2
3
4
5
6
7
8
9
10
11
12 13 14
HMGB4b
(C)
ProteIn]
1
2
3
4
5
6
7
8
Figure 4.5. Binding preference of full-length HMGB4 and its DNA-binding domains to the cisplatin-modified DNA probe.
(A) EMSA analysis of the titration of platinated 25TGGA (1 nM) with full-length HMGB4 with or without a (His) 6 tag at
concentrations of 2 nM (lanes 1, 6), 6 nM (lanes 2, 7), 20 nM (lanes 3, 8), 60 nM (lanes 4, 9), and 100 nM (lanes 5, 10) (left)
and a plot of the protein-bound DNA fraction versus protein concentration (right). (B) EMSA analysis of HMGB4a
bound to the platinated 25TGGA (1 nM) under reducing (5 mM DTT) or non-reducing conditions (left) and a plot of
protein-bound DNA fraction versus concentrations of HMGB4a (right). Protein concentrations were 0.6 nM (lanes 1, 8), 2
nM (lanes 2, 9), 6 nM (lanes 3, 10), 10 nM (lanes 4, 11), 20 nM (lanes 5, 12), 60 nM (lanes 6, 13), and 100 nM (lanes 7, 14)
(C) EMSA analysis of IMGB4b bound to a cisplatin-modified DNA probe (15 nM). Protein concentrations were 31.25,
62.5, 125, 250, 500, 1000, 2000, and 4000 nM (lanes 1-8).
121
Table 4.1. Dissociation Constants of HMGB Proteins and Their Variants for DNA Probes
Containing a Central Cisplatin 1,2-d(GpG) Intrastrand Cross-Linka
Protein
Kd (nM)
Kd(F37A) /Kd(WT)
HMGB4
4.35
0.16
-
HMGB4 F37A
11.74
0.33
2.7
HMGB4a
7.91
0.63
-
HMGB4a F37A
>> 20 [tM
>2000
HMGB4b
422-32
-
HMGBlac
0.7 ±0.1
-
HMGB1 AB165b
0.5 ± 0.2
-
HMGBlb
120 ±10
-
HMGB1 F37Ab
210± 15
1.75
are an average of at least three independent experiments.
aValues
bRef
15
cRef 16
4.3.3. Binding Affinities of Wild Type HMGB4 and its F37A Variant
To investigate the importance of Phe37 in modulating the interaction of cisplatinmodified DNA with full-length HMGB4 or HMGB4a, F37A variants were produced by sitedirected mutagenesis. HMGB4a had significantly lower binding affinity for platinated DNA
when Phe37 was replaced by alanine (Figure 4.6.A); no protein/platinated DNA complex was
observed up to concentrations of 20 pM, indicating that the binding affinity of this variant is >
2000-fold less than that of wild type HMGB4a. On the other hand, the dissociation constant of
the F37A variant of full-length HMGB4 is only 2.7-fold greater than that of wild type protein
(Figure 4.6.B).
122
F37A HMGB4a
WId type HMGB4a
(A)
1
2
3
(B)
4
5
8
7
6
9 10
Wild Type HMGB4
1
2
3
4
5
6
11 12 13
14 15
F37A vadant
7
8
9
10
11
12
13
14
Figure 4.6. EMSA analysis of HMGB4 proteins and their F37A variants binding to platinated DNA. (A) Wild type
HMGB4a and its F37A variant bound to the cisplatin- 1,2-d(GpG) intrastrand cross-linked DNA probe (1 nM). Protein
concentrations were 2 nM (lanes 1, 8), 6 nM (lanes 2, 9), 20 nM (lanes 3, 10), 60 nM (lanes 4, 11), 200 nM (lanes 5, 12), 600
nM (lanes 6, 13), 2 pM (lanes 7, 14), and 6 pM (lane 15, F37A variant only). (B) Wild type and F37A variant of the fulllength HMGB4 bound to the cisplatin 1,2-d(GpG) intrastrand cross-linked DNA probe (1 nM). Protein concentrations
were 1 nM (lanes 1, 8), 2 nM (lanes 2, 9), 6 nM (lanes 3, 10), 10 nM (lanes 4, 11), 20 nM (lanes 5, 12), 60 nM (lanes 6, 13),
and 100 nM (lanes 7, 14).
4.3.4. Weak Asymmetry of HMGB4 Binding to a 1,2-Intrastrand Cross-Link
Wild type full-length HMGB4 and HMGB4a share very similar footprinting patterns,
which are marked by very weak asymmetry with respect to the site of the platinum cross-link
(Figure 4.7). A comparison between domain A of HMGB1 and that of HMGB4 reveals a
difference of intensities for DNA cleaved at 1-2 bps upstream from the cisplatin-d(GpG) lesion
(Figure 4.7). The footprinting results for HMGB4a (Figure 4.7) and full-length HMGB4 (Figure
4.8) are rather similar to those of the HMGB I domain B
15
. Moreover, even though most of the
DNA exists as a DNA-protein complex in EMSA analyses carried out simultaneously with the
footprinting study, binding of HMGB 1 domain A blocks the cleavage at platinated lesions much
better than HMGB4 domain A. The hydroxyl radical cleavage pattern of HMGB4b is quite close
123
to that of HMGB4a, with slightly better footprinting upstream of the platinated lesion. The
footprinting patterns of wild type HMGB4 and its F37A variant under the same conditions are
almost identical (Figure 4.8)
HMGBI
HMGB4
HMGB4
(A)
DomainA DomainB
u
... a|i
- uu
- No Protein (Control)
- HMGB1 domain A (600 nM)
1L23 56A 1 A
- HMGB4 domain A (16 M)
Counts
- HMGB4 domain 8 (16 pM)
21ese
410be 45M
T
C
T
*oo
41
A
T
C
mo
-
M
T
e
Wss
-
agom nM
-a
mosn -
---
C
G
--
AT
owlow
T
500
ss
--
0
_
,
,
_
5W
S-
domain B(16 pM)
-HMGB4
C
1000
T
C
HMGB4 domain A (1 pM)
-
t4
T
.
.
.'
10
ixel
,
Figure 4.7. Footprinting analysis of DNA-binding domains of HMGB1 and HMGB4 bound to the cisplatin-modified DNA
probe. (A) A 15 nM amount of platinated 25TGGA DNA was cleaved by hydroxyl radical footprinting reagents after
incubation with no protein (lane 1) or in the presence of 600 nM (lane 2) or 1 ptM (lane 3) HMGB1 domain A, 1 pM (lane
4) or 2 pM (lane 5) HMGB4 domain A, 16 pM (lane 6) or 32 pM (lane 7) HMGB4 domain B on ice for 30 min.
Footprinting products were separated using 20% urea-PAGE. (B) Plot of the intensity of the cleaved DNA bands from the
footprinting experiment in A: lane 1 (black), lane 2 (green), lane 4 (blue), and lane 6 (red). The cleaved bands upstream of
the cisplatin-modified guanines in the lane containing HMGB4 domains A and B are weaker in intensity compared to
those from HMGB1 domain A. (C) Plot of intensity of cleaved DNA corresponding to lanes 4 and 6 in A. The footprinting
patterns of HMGB4 domain A and domain B are similar to each other, except that domain A displays weaker
footprinting at the cytosine base located 2-bp upstream from the platinated guanines.
124
1
(A)
2
3 4
(B)
Counts
T
C
T
"M-
A
T
Uvw
(C)
C
Counts
T
T
-
T
C
T
m
qu
ess
HMGB4 Wild Type (1 pM)
- HMGB4 F37A (1 pM)
. C
-
TGGAT
C T
Figure 4.8. Footprinting analysis of HMGB1 domain A and full-length HMGB4 proteins bound to a cisplatin-modified
DNA probe. (A) A 15 nM solution of platinated 25TGGA DNA was cleaved by hydroxyl radicals after incubation without
any protein (lane 1) or in the presence of 600 nM HMGB1 domain A (lane 2), 1 piM wild type HMGB4 (lane 3), 1 PM
F37A HMGB4 (lane 4). (B) Plot of the intensity of each cleaved DNA band from the footprinting experiment shown in A:
lane 1 (black), lane 2 (green), lane 3 (blue), and lane 4 (red). (C) Plot of the intensity of each cleaved DNA band from the
footprinting reaction in the presence of wild type HMGB4 (blue) or F37A HMGB4 (red). The cleavage patterns of wild
type and F37A full-length HMGB4 proteins are almost identical, which indicates much weaker asymmetric binding
compared to HMGB1 domain A.
4.3.5. Repair Inhibitory Capacity of HMGB4 Proteins
To investigate the influence of HMGB4 proteins on the repair of cisplatin damaged DNA,
the repair of a 146-bp linear DNA substrate with a radiolabel close to the platinated lesion was
studied in the presence of HMGB proteins. After 1.5 h of incubation under standard conditions,
described in Experimental, 20-30mer strands having 4-6% of full-length DNA substrate were
observed by autoradiography (Figure 4.9, lane 3). There were no such short strands generated
125
when 5 gL of reaction buffer was added instead of CFE (Figure 4.9, lane 2) or when a nonplatinated 146-bp probe was used (Figure 4.9, lane 1). This result indicates that the short strands
are products of the repair of the platinated lesion.
M
1
2
3
200
150
100
25
Figure 4.9. Nucleotide excision repair products resolved on a 8% urea-PAGE gel. 146-bp DNA substrates were incubated
for 1.5 h at 30 *C, quenched by addition of stop solution, and resolved for 1.5 h at 1800 V. Incubation of an unplatinated
146-bp DNA probe with CFE (lane 1) and platinated 146-bp DNA probe without CFE (lane 2) showed no excision
products, whereas incubation of the platinated 146-bp DNA probe with CFE (lane 3) produced short excision products.
The abilities of full-length HMGB1 and HMGB4 to block repair of a DNA substrate
bearing a cisplatin 1,2-intrastrand d(GpG) cross-link are presented in Figure 4.10. At 1 pM
concentration, HMGB1 reduces repair by 45% compared to control, whereas HMGB4 at the
same concentration inhibits repair by > 90% (Table 4.2). In the presence of 125 nM HMGB4,
repair is inhibited by 70%, whereas no significant decrease in the repair occurs in the presence of
HMGB1 at the same concentration. When the incubation time was extended to 3 h, the amount
of repair products increased to ~8%, with a notable increase in repair products shorter that 25
bases, presumably resulting from non-specific nuclease digestion of primary repair products
126
(Figure 4.1 O.B). The relative excision in the presence of HMGB proteins normalized by the
excision level in the absence of the proteins does not change appreciably upon extension of the
reaction time (Figure 4.1 O.C). HMGB4a also inhibits repair as efficiently as full-length HMGB 1.
There is no significant repair inhibition observed in the presence of the F37A variant of
HMGB4a (Figure 4.11).
(A)
1
HMGB4
2
3
4
5
6
7
(B)
HMGBI
8
1 h Incubation
3 h Incubation
9 10 11 12 13 14 15 16
(C)
125
-+-HMGB4 3h
HMGB1 3h
-+-HMGB4 1h-4HMGB1 1h
100
5
25
0
0
2
4
6
8
[Protein] (pM)
Figure 4.10. Excision of a cisplatin-1,2-intrastrand d(GpG) cross-link by active repair extracts in the presence of HMGB1
or HMGB4. (A) Platinated or unplatinated 146-bp repair substrates were incubated in the presence of added HMGB1 or
HMGB4 at 30 *C. The samples were collected after 1 h or 3 h. The repair products were resolved by 8% urea-PAGE. In
control experiments, no excision products were observed when an unplatinated probe was incubated with repair active
CFE (lane 1). Concentrations of HMGB proteins were increased, from 125 nM (lane 3 and lane 10) to 8 pM (lane 9 and
lane 16). (B) The excision products from a platinated repair substrate from 1 h (left) or 3 h (right) incubation. Extended
incubation periods caused a noticeable increase of minor, lower molecular weight, products, that might be the
consequence of exonucleolytic degradation of primary repair products. (C) Plot of the fraction of repaired substrate
versus amount of added HMGB1 or HMGB4 protein after 1 h and 3 h incubation. The amounts of repair products are
normalized to the fraction of repair product in a reaction lacking any HMGB protein (lane 2 of A). Longer incubation
times increased the total amounts of repair products, but the concentration-dependent repair inhibition ability of HMGB
proteins was not influenced by extending the incubation time.
127
(A)
F37A HMGB4a
Wild type HMGB4a
" -
a
V a
a
4t 4
-
(B)
-+-Wild type HMGB4a
120
0 4
100
0
80
60
40
20
0
0
,-
---
1
2
3
4
5
6
7
8
[Protein] (pM)
Figure 4.11. Effect of HMGB4 domain A on excision of a cisplatin 1,2-intrastrand d(GpG) cross-link. (A) Repair
inhibitory effects of wild type HMGB4a and F37A HMGB4a. An unplatinated 146-bp substrate (lane 1) or platinated 146bp substrate (lanes 2-14) was incubated with CFE in the presence of HMGB4a proteins for 1.5 h at 30 *C and the NER
products were analyzed on 8% urea-PAGE gel. Concentrations of protein were 0 pM (lanes 2), 0.25 ptM (lanes 3 and lane
9), 0.5 ptM (lanes 4 and lane 10), 1 ptM (lanes 5, and lane 11), 2 pM (lanes 6 and lane 12), 4 ptM (lanes 7 and lane 13), and 8
ptM (lanes 8 and lane 14). (B) Plot of the relative amount of repair product normalized to the fraction of repair product in
a reaction lacking any HMGB protein (lane 2 of A) versus concentration of added HMGB4a proteins. Wild type HMGB4
shows significant repair inhibition, whereas no significant decrease of repair products was observed when F37A HMGB4a
was added to the reaction.
Table 4.2. Fraction of Repair Product of a Cisplatin-1,2-Intrastrand d(GpG) Cross-Linked
DNA in the Presence of Different Concentrations of HMGB Proteinsa
[Protein]
HMGB1
HMGB4
HMGB4a
125 nM
N/Ab
69%
N/Ab
1 RM
45%
92%
56%
8 9M
86%
>95%
86%
aValues
are an average of at least two independent experiments.
"No significant decrease of repair products were observed.
128
4.4. Discussion
4.4.1. Natural Functions of Testis-Specific HMGB Proteins and Their Interaction with
Cisplatin-Modified DNA
Since the discovery that HMGB proteins bind to cisplatin-damaged DNA, 4 5' several in
vivo and in vitro studies have been carried out in an attempt to correlate the expression level of
HMGB proteins with cisplatin cytotoxicity. 9,17,18 Members of the HMGB protein family that
exhibit high or selective expression in the testis are of particular interest. Murine tsHMG was the
focus of several investigations as a member of HMGB proteins that may contribute to the
hypersensitivity of TGCT to cisplatin. Murine tsHMG is the nuclear isoform of a mitochondrial
transcription factor A (mtTFA), which is expressed only in testis.1 9 This protein has a stronger
binding affinity and greater in vitro repair inhibition of cisplatin-modified DNA than HMGB.2 0
Additionally, overexpression of tsHMG in HeLa cells significantly increased cisplatin
sensitivity.
Despite the fact that the corresponding nuclear isoform of the human mtTFA is not
expressed in human testis, and therefore cannot be correlated with the hypersensitivity of human
testicular cancer to cisplatin, 22 studies of tsHMG demonstrated that HMGB proteins with higher
binding preferences for platinated DNA are able to increase the sensitivity of tumor cells to
cisplatin.
In 2008, the 4th member of mammalian canonical HMGB proteins, called HMGB4, was
identified. 12 Expression profiling of mouse HMGB4 revealed that HMGB4 is expressed
preferentially
in the testis, where
it is proposed to participate in the mechanism
of
spermatogenesis, similar to HMGB2, which helps differentiation of mouse germ cells. HMGB 1,
2, and 3 have very similar amino acid sequences, with >70% sequence similarity. The sequence
of HMGB4 resembles these, but several regions differ significantly. The most notable difference
129
is lack of the C-terminal acidic tail. As a result, HMGB4 has a net positive charge under
physiological conditions (+18.8 at pH 7.5), whereas HMGB 1 is negatively charged (- 5.4 at pH.
7.5).
4.4.2. Information about the Sequence of HMGB4 Relevant to its Binding Interaction with
Cisplatin-Modified DNA
There are several aspects of the HMGB4 sequence that might promote protein binding to
cisplatin-modified DNA. The highly negatively charged C-terminal tail of HMGB proteins
functions to regulate the interaction between HMG boxes and DNA by electrostatically
interacting with their positively charged amino acids, competitively with negatively charged
DNA.
The HMGB 1 didomain AB, which lacks the tail, and domain A of HMGB 1 both have
a much higher binding affinity for platinated DNA than full-length HMGB1. 15 The binding
affinity of HMGB4, which is deficient in the C-terminal acidic tail, more closely resembles that
of the HMGB1 didomain AB than that of the full-length protein. Apart from the absence of the
C-terminal tail, there are relatively large variations in the amino acids of the DNA-binding
domains for HMGB4 compared with other HMGB proteins. Sequence alignment analysis of
HMGB4 and HMGB 1 (Figure 4.1 .A) reveals that many amino acid residues that interact directly
with the DNA backbone in the X-ray crystal structure of a complex between platinated DNA and
HMGB1 domain A (Figure 4.12)24 are conserved in HMGB4 domain A. Notably, the
intercalating residue Phe37, which plays a significant role in the binding of HMGB1 to
platinated DNA, 24,25 is conserved in HMGB4, whereas the position corresponding to Ala16 in
HMGB1-3 is occupied by Ile16. Two putative intercalating residues in domain B, Phe102 and
Ile 121, are replaced by Leul 00 and Val 119 in HMGB4, which implies that the binding nature of
HMGB4 domain B might differ significantly from that of HMGB 1 domain B. The most
130
remarkable sequence difference is the replacement of cysteine at position 22 in HMGB 1 with a
phenylalanine in rat and mouse HMGB4 or a tyrosine in human HMGB4. Cys22 in other HMGB
proteins can form an intradomain disulfide bond with Cys44 under mildly oxidizing conditions.26
Previous redox studies of HMGB 1 and its domain A demonstrated that the binding affinity of the
protein decreases significantly upon formation of the disulfide bond, suggesting that the
interaction between DNA and HMGBI is regulated by the intracellular redox environment. 16
HMGB4 cannot form an intradomain disulfide bond, and hence the DNA-binding capacity of
HMGB4 is redox-independent, unlike the other three HMGB proteins.
L:
A
Figure 4.12. X-ray crystal structure of the complex of the HMGB1 domain A and a 16-bp DNA probe containing a
cisplatin 1,2-intrastrand cross-link (pdb accession number 1CKD). The r-r stacking and edge-to-face aromatic
interactions between the side chain of Phe37 and the two platinated guanine rings are designated by double-headed
arrows.
4.4.3. Binding Affinity and Binding Specificity of HMGB4 to Platinated DNA
131
The overall platinated DNA-binding properties of mouse HMGB4 are quite similar to
those of HMGB1. HMGB4 binds specifically to a DNA probe with a cisplatin 1,2-intrastrand
d(GpG) cross-link. This binding preference was not observed when a probe containing a cisplatin
1,3-intrastrand d(GpTpG) cross-link was used, similar to that of HMGB l. The binding affinity
of full-length HMGB4 is about 9-fold weaker than that of HMGB 1 didomain AB, but about 28fold stronger than that of full-length HMGB 1, a result that is consistent with the prediction based
on the lack of a C-terminal acidic tail. Unlike HMGB1, however, the binding of HMGB4a or
full-length HMGB4 is redox-independent because it cannot form an intradomain disulfide bond.
Not only the full-length protein, but also each DNA binding motif, binds more weakly to
platinated DNA than its HMGB 1 counterpart. These differences may be the result of differences
in the amino acid residues of the DNA binding motifs in HMGB 1 versus HMGB4. The two
putative intercalating residues of HMGB 1 domain B are different, whereas the Phe37 in domain
A is conserved in HMGB4 from all species. The importance and putative roles of the
intercalating residues are discussed below. Even though the absolute values of dissociation
constants differ for the DNA-binding domains of HMGB 1 and HMGB4, the relative ratios of the
binding affinities of HMG domains of HMGB4 are similar to those of HMGB 1 (Table 4.1). As
for HMGB 1, domain A of HMGB4 is the main DNA-binding domain because its binding
affinity is much greater than that of domain B. Full-length HMGB4 displays about 2-fold higher
binding affinity than HMGB4a, and is similar to that of HMGB 1; similarly, HMGB 1 didomain
AB exhibits about twice the binding affinity of HMGB 1 domain A. Presumably, as in the
binding of HMGB 1, HMGB4 domain B increases DNA bending and facilitates binding of fulllength HMGB4 instead of interacting directly with the platinum cross-links, whereas domain A
binds directly to the Pt binding site.
132
4.4.4. Function of Intercalating Residues and Their Influence on the Structure of the
Platinated DNA-Protein Complex
Unlike results obtained with most other HMG box proteins, hydroxyl radical footprinting
analyses of full-length HMGB1 or its domain A with DNA bearing a cisplatin 1,2-intrastrand
d(GpG) cross-link reveal highly asymmetric binding with respect to the site of platination.
The X-ray crystal structure of the complex revealed this asymmetry to be the consequence of
insertion of the side chain of Phe37 from domain A into the hydrophobic notch in the minor
groove formed by the two platinum-modified guanosine residues (Figure 4.12) in the proteinDNA complex. A combination of iu-7c stacking and edge-to-face aromatic interactions between
the phenyl ring side chain and the two guanine bases significantly stabilizes the complex. As a
result, helix II of HMGB1 domain A, with Phe37 at its N-terminus, binds mainly to the 3' side of
the platinum lesion, causing the binding asymmetry. There is one more amino acid position in
HMGB1 domain A that is capable of intercalating into the platinated DNA, Alal6 in helix I.
Removal of Phe37 and mutating Ala16 to Phel6 completely altered the binding mode from
asymmetric to symmetric with respect to the cisplatin binding site.2
The binding of HMGB1
domain B produces a symmetric footprint because the main intercalating residue in this protein,
Phel02, is at the site corresponding Alal6 of domain A and mainly interacts with the cisplatinDNA cross-link. Replacement of Phe102 with Alal02 alters the binding of domain B to an
asymmetrical one, probably through the induction of a weak interaction between the modified
guanine and Ile121, the residue at the position corresponding to Phe37 in domain A.25
The intercalating residue Phe37 is conserved in domain A of HMGB4. Replacement of
Phe37 with Ala37 decreases the binding affinity of domain A to the platinated DNA by more
than a factor of 2000, demonstrating its critical role in stabilizing the interaction. The
133
footprinting analysis, however, surprisingly revealed very weak asymmetry compared to that of
HMGB1 domain A, even close to the symmetric footprinting pattern of HMGB1 domain B
(Figure 4.7). This footprinting pattern, and the fact that the binding affinity of wild type HMGB4
domain A is weaker than that of HMGB 1 domain A by more than a factor of 10 (Table 4.1),
clearly reveal that some of HMGB1 residues interacting with platinated DNA backbone are
missing in HMGB4. The differences lie more toward the N- than the C-terminal side of Phe37
(Figure 4.1.A) and must in some manner override the influence of the inserting phenylalanine
side chain on the binding symmetry.
Another possible intercalation position, corresponding to Alal 6 in HMGB 1, is a valine in
human HMGB4 or isoleucine in mouse HMGB4. The side chain of alanine is too short to
intercalate into DNA, but a weak hydrophobic contact between Alal 6 and one of the bases in the
unmodified strand was observed in the crystal structure of HMGB 1 domain A bound to cisplatinmodified DNA.
The side chain of isoleucine or valine is presumably long enough to reach and
insert at least partially between the base pairs, providing a second intercalation on the
unmodified strand.
Compared to full-length HMGB4 or HMGB4a, HMGB4b displays a very low binding
affinity for platinated DNA. In HMGB4b, LeulOO is at the position corresponding to
intercalating residue Phe102 of HMGB1 domain B and Ala16 in HMGB1 domain A (Figure
4.1 .A). Another possible intercalating residue, corresponding to Phe37 of HMGB4 domain A and
Ile 121 of HMGB 1 domain B, is Val 119. The footprinting pattern of HMGB4b is quite similar to
that previously reported for HMGB1 domain B, in which Phe102 mainly plays a role as an
intercalating residue.2 s This result suggests Leul 00, rather than Val 119, may function as an
intercalating residue in HMGB4b when binding to platinated-DNA.
134
4.4.5. Repair Inhibition by HMGB Proteins
One hypothesis for HMGB1 participation in the mechanism of the action of cisplatin is
repair shielding.2 7 According to this model, binding of HMGB 1 to the platinated lesion inhibits
damage repair either by preventing its recognition by cellular proteins or by disrupting formation
of the repair apparatus, thereby increasing the efficacy of the drug. One of the strongest pieces of
evidence in support of this model is the in vitro repair study that reveals a dramatic loss in repair
of cisplatin 1,2-intrastrand d(GpG) cross-links in the presence of increasing and excess amounts
of HMG proteins.20,27,28 Addition of HMGB4 to a repair reaction similarly decreases the level of
repair of the cisplatin-1,2-intrastrand cross-link. Impressively, the inhibition efficiency of
HMGB4 is much greater than that of HMGBl; only 62.5 nM of HMGB4 is sufficient to decrease
repair by > 50%, whereas 16 times more HMGB 1 had to be added to reach a similar degree of
repair inhibition. This repair shielding is even greater than that reported for tsHMG, a mouse
testis-specific protein, which has a very high repair inhibitory efficiency.20 The F37A variant of
HMGB4a does not show any repair inhibition because it cannot bind to platinated lesions. The
repair inhibition efficiency of HMGB4a, which is smaller than full-length HMGB1 but has a
higher binding affinity, is similar to that of HMGB 1, which suggests that the repair inhibition by
HMGB proteins can be influenced by both their size and binding affinity.
4.4.6. HMGB4, Cisplatin, and Testicular Cancer
It is well known that testicular germ cell tumors (TGCTs) are hypersensitive to cisplatin
chemotherapy. Molecular biology studies of cisplatin demonstrate rapid induction of apoptosis in
TGCT after exposure to cisplatin, in agreement with clinical studies. 3,29 There are several
hypotheses to explain the hypersensitivity of TGCT, such as the low expression level of certain
30p31
proteins that are indispensible for detecting or repairing DNA damages in testis tissue.
135
We propose that HMGB4 may be one of the factors responsible for the high cisplatin
efficacy against testicular cancer. The in vitro studies carried out in this work reveal that
HMGB4, preferentially expressed in testis, not only interacts with cisplatin cross-linked DNA
with much higher binding affinity than HMGB1, but also inhibits excision repair to a
significantly greater degree. Moreover, its inability to form intradomain disulfide bonds removes
the possible diminution of binding affinity for platinated DNA under oxidizing conditions.
Whether or not HMGB4 participates in the cisplatin mechanism in vivo, as well as we have
observed in vitro, and is responsible for the efficacy of cisplatin in treating testicular cancer
remains to be determined.
4.5. Conclusion
This work describes the nature of the binding interaction between the testis-specific
protein HMGB4 and cisplatin-modified DNA. Despite differences in the amino acid sequence,
the binding properties of HMGB4 are quite similar to those of the more widely distributed
homologue HMGBl. In particular, HMGB4: (i) binds to the 1,2-intrastrand cross-linked DNA
with high preference; (ii) Interacts mainly through its HMG domain A; and (iii) Binds more
weakly when Phe37 is mutated to alanine, with the domain A exhibiting dramatically reduced
binding affinity for platinated DNA. The binding affinities of HMGB4 and its DNA-binding
domains to platinated DNA are less than those of their HMGB 1 counterparts. The binding mode
of HMGB4, unlike that of HMGB1, is very weakly asymmetric with respect to the platinated
cross-link. In vitro assays reveal that HMGB4 can inhibit excision repair of platinated lesions
much more efficiently than HMGBl. These results are consistent with our hypothesis that
HMGB4 may function in the mechanism of action of cisplatin like HMGB1, but much more
136
effectively. Because HMGB4 is expressed preferentially in testis, it may be an important factor
contributing to the hypersensitivity of TCGTs to cisplatin.
137
4.6. References
(1)
(2)
Boulikas, T.; Vougiouka, M. Oncol. Rep. 2004, 11, 559-595.
Biggs, M. L.; Schwartz, S. M. In SEER Survival Monograph: Cancer Survival Among
Adults: U.S. SEER program, 1988-2001, Patient and Tumor Characteristics;Ries, L. A.
G., Yong, J. L., Kell, G. E., Eisner, M. P., Lin, Y. D., Homer, M. J., Eds.; NIH Pub.:
Bethesda, MD, 2007.
(3)
Masters, J. R.; Koberle, B. Nat. Rev. Cancer 2003, 3, 517-525.
(4)
Bruhn, S. L.; Pil, P. M.; Essigmann, J. M.; Housman, D. E.; Lippard, S. J. Proc. Natl.
(5)
Acad. Sci. U. S. A. 1992, 89,2307-2311.
Pil, P. M.; Lippard, S. J. Science 1992, 256, 234-237.
(6)
(7)
(8)
Bianchi, M. E.; Beltrame, M.; Paonessa, G. Science 1989, 243, 1056-1059.
Seyedin, S. M.; Kistler, W. S. J. Biol. Chem. 1979, 254, 11264-11271.
Ronfani, L.; Ferraguti, M.; Croci, L.; Ovitt, C. E.; Scholer, H. R.; Consalez, G. G.;
Bianchi, M. E. Development 2001, 128, 1265-1273.
(9)
Arioka, H.; Nishio, K.; Ishida, T.; Fukumoto, H.; Fukuoka, K.; Nomoto, T.; Kurokawa,
(10)
Kwon, J. H.; Kim, J.; Park, J. Y.; Hong, S. M.; Park, C. W.; Hong, S. J.; Park, S. Y.; Choi,
Y. J.; Do, I. G.; Joh, J. W.; Kim, D. S.; Choi, K. Y. Clin. CancerRes. 2010, 16, 5511-
(11)
Balani, P.; Boulaire, J.; Zhao, Y.; Zeng, J.; Lin, J.; Wang, S. Mol. Ther. 2009, 17, 10031011.
Catena, R.; Escoffier, E.; Caron, C.; Khochbin, S.; Martianov, I.; Davidson, I. Biol.
H.; Yokote, H.; Abe, S.; Saijo, N. Jpn. J. Cancer Res. 1999, 90, 108-115.
5521.
(12)
Reprod. 2009, 80, 358-366.
(13)
(14)
Wang, Q.; Zeng, M.; Wang, W.; Tang, J.Biochem. Biophys. Res. Commun. 2007, 360,
14-19.
Lovejoy, K. S.; Todd, R. C.; Zhang, S.; McCormick, M. S.; D'Aquino, J. A.; Reardon, J.
T.; Sancar, A.; Giacomini, K. M.; Lippard, S. J. Proc.Natl. Acad. Sci. U. S. A. 2008, 105,
8902-8907.
(15)
(16)
Jung, Y.; Lippard, S. J. Biochemistry 2003, 42, 2664-2671.
Park, S.; Lippard, S. J. Biochemistry 2011, 50, 2567-2574.
(17)
(18)
(19)
Brown, S. J.; Kellett, P. J.; Lippard, S. J. Science 1993, 261, 603-605.
He, Q.; Liang, C. H.; Lippard, S. J. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 5768-5772.
Larsson, N. G.; Garman, J. D.; Oldfors, A.; Barsh, G. S.; Clayton, D. A. Nat. Genet. 1996,
13, 296-302.
Ohndorf, U. M.; Whitehead, J. P.; Raju, N. L.; Lippard, S. J. Biochemistry 1997, 36,
(20)
14807-14815.
(21)
(22)
Zamble, D. B.; Mikata, Y.; Eng, C. H.; Sandman, K. E.; Lippard, S. J. J. Inorg. Biochem.
2002, 91, 451-462.
Larsson, N. G.; Oldfors, A.; Garman, J. D.; Barsh, G. S.; Clayton, D. A. Hum. Mol. Genet.
(23)
1997, 6, 185-191.
Watson, M.; Stott, K.; Thomas, J. 0. J. Mol. Biol. 2007, 374, 1286-1297.
(24)
Ohndorf, U. M.; Rould, M. A.; He,
Q.; Pabo, C. 0.;
712.
(25)
Lippard, S. J. Nature 1999, 399, 708-
He, Q.; Ohndorf, U. M.; Lippard, S. J. Biochemistry 2000, 39, 14426-14435.
138
(26)
(27)
(28)
(29)
(30)
(31)
Hoppe, G.; Talcott, K. E.; Bhattacharya, S. K.; Crabb, J. W.; Sears, J. E. Exp. Cell Res.
2006, 312, 3526-3538.
Huang, J.-C.; Zamble, D. B.; Reardon, J. T.; Lippard, S. J.; Sancar, A. Proc.Natl. A cad.
Sci. U. S. A. 1994, 91, 10394-10398.
Zamble, D. B.; Mu, D.; Reardon, J. T.; Sancar, A.; Lippard, S. J. Biochemistry 1996, 35,
10004-10013.
Masters, J. R.; Osborne, E. J.; Walker, M. C.; Parris, C. N. Int. J.Cancer 1993, 53, 340346.
Koberle, B.; Masters, J. R.; Hartley, J. A.; Wood, R. D. Curr.Bio. 1999, 9, 273-276.
Usanova, S.; Piee-Staffa, A.; Sied, U.; Thomale, J.; Schneider, A.; Kaina, B.; Koberle, B.
Mol. Cancer 2010, 9, 248-258.
139
Chapter 5. Summary and Future Directions
140
5.1. Summary and Future Directions
In the previous chapters, studies related to the binding of HMGB proteins to cisplatinmodified DNA were described. As presented in Chapter 1, studies using live cells produced
contradictory results about the correlation between expression levels of HMGB proteins and
cisplatin sensitivity. This inconsistency was also observed in HMGB1 knockdown studies
discussed in Chapter 2. Two different human cancer cell lines, HeLa and A549, respond
differently to cisplatin following HMGB 1 knockdown by shRNA. Moreover, the repair of
platinated DNA in both knockdown cells was slower than that in parental cell lines, which does
not support the expectation from the repair-shielding model.
HMGB 1 interacts with many proteins involved in cisplatin-triggered apoptosis. We
hypothesize that HMGB 1 can promote the repair of platinated DNA via those interactions and
under some conditions this repair-promoting effect overcomes the repair-shielding effect of
HMGB 1. In Chapter 4, the binding interaction of HMGB4, an HMG protein expressed
exclusively in testis that may function as repair-shielding protein to platinated DNA, was
established. The strong binding affinity and efficient repair inhibition of HMGB4 suggested that
HMGB4 could act as a repair shield for cisplatin platinated DNA. Whether or it contributes to
cisplatin sensitivity in live testes cells remains to be investigated. HMGB4 expression levels can
be up-regulated by transfection with an HMGB4-encoding gene, especially in cells that do not
naturally express HMGB4, or they can be down-regulated by use of RNAi method. The function
of HMGB4 can be examined by investigating the cellular response to cisplatin of these modified
cells. Preliminary results of HMGB4 studies in live cells are described in Appendices A and B.
141
It is also possible that some intracellular factors that vary among the different cell lines
affect the ability of HMGB 1 to alter the repair of platinated DNA. Among such possible factors,
we focused on the intracellular redox potential. The study presented in Chapter 3 reveals that
the binding affinity of HMGB1 to platinated DNA decreases by a factor of more than 10 upon
the formation of a disulfide bond between Cys22 and Cys44 in HMGB1 domain A. To verify
that a difference in intracellular redox potential among cells is in fact an important factor in
regulating the DNA binding of HMGB1, it would be interesting to determine whether a
significant fraction of HMGB1 exists in its oxidized form in the intracellular environment and
that the redox potential of cells is meaningfully correlated to the capability of HMGB1 to
sensitize cells to cisplatin. A classical way of measuring the redox potential is to determine the
ratio of oxidized and reduced forms of glutathione that exist in a cell extract.' There are also the
redox assays that use redox-sensitive GFP probes to measure the intracellular redox potential
without cell lysis. 2 It is possible to measure the redox potential in nuclei, where HMGB1
interacts with DNA, by expressing GFP having a nuclear targeting signal. To prove that redox
potential is the regulatory factor in the function of HMGB1 in the mechanism of action of
cisplatin, one might first determine the intracellular redox potential of various cell lines, and then
investigate the influence of HMGB 1 in the cellular response to cisplatin via the knockdown of
HMGB 1 in each cell line.
In summary, the in vitro results presented in this thesis suggest a hypothesis that can
explain the controversial results observed previously. Studies on live cells remain for further
interrogation of these hypotheses. Once the repair-shielding ability of HMGB proteins on
platinated DNA is determined in live cells, or ultimately, in vivo, we expect that our knowledge
142
of HMGB proteins will help predict the cisplatin sensitivity of tumors and provide further insight
into the mechanism of action of cisplatin.
5.2. References
(1)
(2)
Schafer, F. Q.; Buettner, G. R. Free Radic. Biol. Med. 2001, 30, 1191-1212.
Hanson, G. T.; Aggeler, R.; Oglesbee, D.; Cannon, M.; Capaldi, R. A.; Tsien, R. Y.;
Remington, S. J. J. Biol. Chem. 2004, 279, 13044-13053.
143
Appendix A. Expression Profiling of Human HMGB4
A.1. Purpose
In Chapter 4, the binding properties of HMGB4 to platinated DNA and the repair
inhibitory capability of HMGB4 in cell free extract (CFE) systems were described. The results
showed that HMGB4 might act as a repair-shielding protein, contributing to the cisplatin
hypersensitivity in testicular germ cell tumors (TGCTs). Previous reports on the testis-restricted
expression of HMGB4 were based on western blot and RT-PCR studies using mouse tissues.1 To
date, no expression profile of human HMGB4 has been established. Herein, we investigate
whether human HMGB4 exhibits the same selective expression in human testis as reported for
the mouse model. To verify our model, the expression levels of HMGB4 in human cancer cell
lines originating from various tissues were investigated by western blot analysis.
A.2. Experimental
A.2.1. Cell Culture
HeLa, A549, NTera2, MCF7, PC-3, 293/T17, U2OS, and Jurkat cell lines were obtained
from ATCC. Cells were grown in DMEM (GIBCO/BRL) containing 10% FBS (GIBCO/BRL),
100 units/mL of penicillin, and 100 ptg/mL of streptomycin. All cells were incubated at 37 'C
under a 5% CO 2 atmosphere.
A.2.2. Western Blot Analysis
Adherent cells were grown to 90% confluence in T75 cell culture flasks and collected
after trypsinization followed by centrifugation. Jurkat cells were grown in 6 mL of medium in a
T25 cell culture flask to a density of ~ 5 x 105 cells/mL and harvested by centrifugation. Protein
extraction was carried out as previously described in Chapter 2. Whole cell extracts (40-50 pg)
were separated by SDS-PAGE on a 4-20% Tris-HCl Ready-Gel (Bio-Rad) and then transferred
144
to the PVDF membrane in western transfer buffer (48 mM Tris-HCl, 39 mM glycine, pH 9.2,
0.0 1% SDS). After the transfer, the PVDF membrane was incubated in Tris buffered saline (TBS)
buffer supplemented by 5% milk powder to block non-specific binding of antibodies. The
subsequent antibody treatment and washing steps are identical to those described in Chapter 2.
Anti-HMGB4 rabbit IgG SAB2104592 (Sigma Aldrich) and anti-Actin rabbit IgG (Sigma
Aldrich) were used as a primary antibody and anti-Rabbit IgG goat antibody (Abcam) was used
as a secondary antibody.
A.3. Results and Discussion
A.3.1. Expression Profile of Human HMGB4 in Cancer Cell Lines
(A)
(B)
0
Actin
25 kDa
20 kDa -+
HMGB4 1
Figure A.1. Western blot analysis of human HMGB4. (A) NTera2 (testicular cancer) and MCF7 (breast cancer) cells
presented HMGB4 bands between 20 kDa and 25 kDa, whereas A549 cells lacked HMGB4 expression. (B) Among human
cancer cell lines originating from different tissues, only NTera2 presented HMGB4 expression. In this particular western
blot analysis, MCF7 did not show a HMGB4 band with observable intensity.
In the western blot analysis, bands ascribed to HMGB4 were observed between 20 kDa
and 25 kDa bands in protein molecular weight ladder, in agreement with the molecular weight of
human HMGB4 (Figure A. L.A). The whole cell extracts of testicular (NTera2) and breast cancer
cells (MCF7) showed a positive HMGB4 band (Figure A. L.A). Compared to the relatively strong
signal observed in whole cell extracts of NTera2, a very weak HMGB4 band was observed in
whole cell extracts of MCF7. No HMGB4 expression was observed in the whole cell extracts of
lung (A549), prostate (PC-3), kidney (293/17), cervical (HeLa), and T-lymphocyte (Jurkat)
145
cancer cell lines. These results revealed that, as in the case of mouse HMGB4, human HMGB4 is
expressed in a testis-preferred manner. The weak signal of HMGB4 in cell extracts from breast
cancer cells indicates the influence of HMGB4 on the intracellular processes of breast cancer
cells is probably not as strong as in testicular cancer cells.
A.3.2. Future Directions
NTera2 is derived from embryonal carcinomas (EC). HMGB4 expression levels in TGCT
cell lines other than EC have to be investigated. In addition, considering that expression levels of
many proteins alter during the immortalization process of primary cells, primary tissue from
TGCT patients should be tested to confirm that expression of HMGB4 occurs in actual testicular
tumor cells.
A.4. References
(1)
Catena, R.; Escoffier, E.; Caron, C.; Khochbin, S.; Martianov, I.; Davidson, I. Biol.
Reprod. 2009, 80, 358-366.
146
Appendix B. Attempts to Knockdown of HMGB4 in NTera2
B.1. Purpose
The western blot analysis presented in Appendix A revealed that NTera2, an embryonal
carcinomas cell line, expresses high levels of HMGB4, whereas other cell lines either do not
express HMGB4 or present only very low levels of HMGB4. Here we describe attempts to
establish HMGB4 knockdown NTera2 to investigate the influence of HMGB4 on the mechanism
of action of cisplatin in TGCT.
B.2. Experimental
B.2.1. HMGB4 Knockdown in NTera2 by Transfection of shRNA Expression Vector
We attempted to knockdown HMGB4 in NTera2 by transfection of HMGB4 short hairpin
RNA (shRNA) expression vectors. The shRNA expression vectors were synthesized by inserting
three HMGB4 shRNA target sequences into the pLKO1 vector as described in Chapter 2.
Transfection of pLKO1 plasmids into NTera2 cells was carried out as described in Chapter 2.
After transfection, cells were incubated in growth medium supplemented by 200 ng/mL of
puromycin to select for cells containing the transfected DNA. Information for each stably
transfected NTera2 cell line is described in Table B. 1. Homogeneous cloning was also attempted,
but aborted due to observed changes in NTera2 morphology.
Table B. 1. List of pLKO 1-Transfected NTera2 Cell Lines
Designation
Transfection
Target Sequence
NTera2 KD1
pLKO1 HMGB4 (1)
5'- GCTAAGTACTTCGAGGAACTT-3'
NTera2 KD2
pLKOl HMGB4 (2)
5'- GCAAATGTCTCTTCTTACGTT-3'
NTera2 KD3
pLKO1 HMGB4 (3)
5'- GCACCCTTATGAGCAAAGAGT-3'
B.2.2. HMGB4 Knockdown in NTera2 by Viral Transduction
147
As an alternative approach to transfection of shRNA expression plasmids, viral
transduction was attempted to acquire knockdown cell lines having better RNAi efficiency. The
production of viral particles was carried out by Peter Bruno in the Hemann Lab at the Koch
Institute as previously described.1 Four different target sequences were selected and cloned into
the MSCV-LTRmirR30-SV40GFP vector. This vector has an RNA polymerase II promoter for
shRNA expression, a GFP expression gene, a puromycin resistance gene, and a retroviral long
terminal repeat (LTR) promoter. For transduction, 3 x 105 NTera2 cells were plated in 2 mL
growth medium in 6-well plates. After overnight incubation at 37 'C under a 5% CO 2
atmosphere, 10 ptL of viral particle solution were directly added into each well. 48 h after the
transduction, the medium was replaced by fresh medium supplemented with 200 ng/mL
puromycin. After antibiotic selections, GFP-positive cells were sorted at the MIT fluorescence
activated cell sorting (FACS) facility. The expression levels of HMGB4 in transduced cell lines
were measured by western blot analysis as described in Appendix A. Information about each
transduced NTera2 cell lines is presented in Table B.2.
Table B.2. List of Viral-Transduced NTera2 Cell Lines
aThe
Designation
Target Sequence
NTera2 50a
5' - TCCATGTTCTATTCATTAGTA -3'
NTera2 211
5' - CCCAAAGTTACTCAGCTAGTA - 3'
NTera2 1115
5'- CACAGCTGTCATACCTGATTA - 3'
NTera2 1757
5' - TCCAGCTAAAGCCTAAGGCAA - 3'
number represents the start position of the target sequence in HMGB4 mRNA.
B.3. Results and Discussion
B.3.1. Expression Levels of HMGB4 in Stably Transfected Cell Lines
148
The HMGB4 expression levels in NTera2 cell lines transfected by pLKO 1 shRNA
expression vectors were analyzed by western blots. Expression levels in heterogeneous NTera2
KD 1, KD2, and KD3 cells were not significantly different from that of parental NTera2 cells
(Figure B. 1). Cells plated at low density for homogeneous cloning displayed a significantly
altered neuron-like morphology, probably because of cell differentiation induced by low plating
density.2
Actin
HMGB4
Figure 8.1. Western blot analysis of HMGB4 in NTera2 cells transfected by pLKO1 shRNA expression vectors. No significant
decrease of HMGB4 expression level is observed in all three knockdown cell lines as compared to parental NTera2 cell line.
B.3.2. Viral Transduction for HMGB4 Knockdown
Because the pLKO1 transfection followed by antibiotic selection was not effective for
inducing significant knockdown of HMGB4 expression, a viral transduction was attempted as an
alternative approach. Flow cytometry analysis of GFP expression after the antibiotic selection
revealed that only 20-30% of the puromycin-resistant population was GFP positive (Figure B.2).
To improve RNAi efficiency, the cells were sorted by the GFP signal. No considerable
knockdown of HMGB4 was observed by western blot analysis of GFP positive cells (Figure B.3).
149
10111
s
l0l1'
10 10
10
FITC-A
FITC-A
NTera21115
23.4%
10
10
10s
0
NTera21757
25.1%
3:
10
FITC-A
FITC-A
Figure B.2. Flow cytometry analysis of GFP expression in viral transduced NTera2 cells. Even after puromycin selection,
less than half of cells express GFP. The ID and percentage of GFP positive population were noted in each histogram.
Actin
HMGB4
Figure B.3. Western blot analysis of NTera2 cells transduced by viral particle with HMGB4 shRNA gene. No significant
decrease of HMGB4 expression was observed in any of transduced cells compared to parental NTera2 cells.
B.3. Future Directions
Neither transfection nor viral transduction of shRNA genes into NTera2 afforded
knockdown. Establishing an HMGB4 knockdown NTera2 cell line with high RNAi efficiency is
critical for investigating the influence of HMGB4 on the response of TGCT to cisplatin. Two
alternatives to the attempted methodologies are i) using siRNA instead of shRNA expression
150
plasmids or ii) using a TGCT cell line that is not established from embryonic carcinomas, thus
avoiding differentiation during development of homogeneous knockdown clones. Once the
HMGB4 knockdown TGCT cell line has been established, the cellular response to cisplatin in
HMGB4 knockdown TGCT cells can be tested by cytotoxicity and transcription assays.
B.4. References
(1)
(2)
Dickins, R. A.; Hemann, M. T.; Zilfou, J. T.; Simpson, D. R.; Ibarra, I.; Hannon, G. J.;
Lowe, S. W. Nat. Genet. 2005, 37, 1289-1295.
Rees, A. R.; Adamson, E. D.; Graham, C. F. Nature 1979, 281, 309-311.
151
BIOGRAPHY
Semi Park was born in Seoul, Korea in 1985, to Myungsun Park and Hyejung Kim. She
attended Seoul science high school where she became interested in Chemistry. She graduated
from Seoul National University in 2007 with a B.Sc. in chemistry under the direction of
Professor Seokmin Shin and Professor Seung Keun Kim. She received the Presidential Scholarship
from Science and Engineering Foundation of Republic of Korea from 2003 to 2007. She began to study
inorganic chemistry at MIT in 2007 and joined the laboratory of Professor Stephen J. Lippard,
where she studied the role of high mobility group box proteins in the mechanism of action of
cisplatin.
152
Semi Park
Phone: 617-455-8844
E-mail: semipark@mit.edu
77 Massachusetts Avenue
Cambridge, MA, 02139
EDUCATION
Cambridge, MA
Massachusetts Institute of Technology
Ph.D. Candidate in Inorganic Chemistry
September 2007 - August 2012 (expected)
Seoul, Republic of Korea
Seoul National University
Bachelor of Science in Chemistry
March 2003 - February 2007
EXPERIENCE
Lippard Lab, MIT
January 2008 - current
Advisor: Prof. Stephen J. Lippard
" Examined the function of proteins involved in the action of the mechanism of platinum-based
anticancer drugs.
* Identified the factors that regulate the interaction of high mobility group box 1 protein and
cisplatin modified DNA, with special focus on intracellular redox potential.
* Delineated the repair of the platinum-based anticancer drugs in repair-deficient system
* Investigated the repair inhibitory function of HMGB proteins at both the intracellular and in vitro
levels
* Identified membrane proteins facilitating the uptake of cisplatin
Molecular Biophysics and Material Lab
March 2006 - August 2006
Advisor: Prof. Kim, Seong Keun
. Observed behavior of photo-excited hot electrons on metal surfaces for investigating electron
transfer across the interface.
SKILLS
Biochemistry: Protein expression in bacterial cells, Protein engineering by site-directed mutagenesis,
Plasmid modification, PCR, Mass spectroscopy of oligonucleotides and proteins, Oligonucleotide
sequencing, Radiolabeling of biomolecules, Protein purification, FPLC chromatography, Synthesis and
purification of oligonucleotide, Chemical modification of DNA, Methods for studying DNA-protein
interaction (EMSA, Hydroxyl radical/DNAse I footprinting, ITC), Western Blot, Native/Urea/SDS
PAGE, 2D-protein PAGE, DNA repair assay in vitro
Cell biology and Pharmaceutical Studies: Mammalian cell culture, Immunoprecipitation,
Immunofluorescence, Establishment of modified cells by transient/stable transfection, Viral/Non-viral
transfection, Cell sorting, Cytotoxicity assay of drugs, DNA repair assays in cultured cells, RNA
purification and RT-PCR, Spheroid generation
153
Instruments: HPLC and FPLC, MALDI-TOF, LC-MS, UV/Vis, CD, fluorescence/luminescence
spectroscopy, phosphoimager, DNA synthesizer
Computer Skills: Office software packages, such as MS Word, Excel, PowerPoint; Chemistry-based
software, such as Chemdraw, Scifinder Scholar, PyMol, Endnote; data processing software, such as
Mathematica, Kaleidagraph, OriginLab, Illustrator, Prism.
Languages: English, Korean (native speaker)
AWARD/SCHOLARSHIPS
IPMI Student Award: International Precious Metals Institute. 2010
MIT Presidential Fellowship. 2007
32nd Fellowship Graduate Student of the Korean Foundation for Advanced Studies
Presidential Scholarship: Science and Engineering Foundation of Republic of Korea
PUBLICATION
Park, S. and Lippard, S.J. (2011) Redox State-Dependent Interaction of HMGB 1 and Cisplatin-Modified
DNA. Biochemistry 50, 2567-2574
Park, S. and Lippard, S.J. Binding Interaction of IMGB4 with Cisplatin-Modified DNA, Manuscript
submitted
TEACHING EXPERIENCE
Teaching assistant of MIT Course 5.111, Principle of Chemical Science: 2007. Fall
Teaching assistant of MIT Course 5.310, Laboratory Chemistry: 2008. Spring
PROFESSIONAL AFFILIATION
Member of American Chemical Society
Member of Society of Bioinorganic Chemistry
154