Post on 18-Sep-2020
Nº D’ORDRE : 896 .
THESE présentée en vue de l’obtention de grade de Docteur de l’Université de Toulouse délivré par l’Institut National des Sciences
Appliquées de Toulouse
Spécialité : Ingénieries microbienne et enzymatique
Par Mlle. Adriana MARQUEZ-QUIÑONES INTITULE :
REACTIVE OXYGEN SPECIES, HEPATITIS AND CARCINOGENESIS INITIATION: AN INTEGRATIVE APPROACH COMBINING TRANSCRIPTOMIC AND
METABONOMIC PROFILINGS
Espèces réactives de l’oxygène, hépatite et initiation du cancer : une approche intégrée qui combine le profilage transcriptomique et métabonomique.
DATE DE SOUTENANCE : 21 NOVEMBRE 2007 JURY : FONCTION M. CORPET Denis Président M. FRANÇOIS Jean-Marie Directeur de thèse Mme. GUERAUD Françoise Codirectrice de thèse M. EZAN Eric Rapporteur M. GRUNE Tilman Rapporteur M. Del RIO-GUERRA Gabriel Rapporteur M. PARIS Alain Examinateur M. VILLA-TREVIÑO Saúl Examinateur ECOLE DOCTORALE : Sciences Ecologiques, Vétérinaires, Agronomiques et
Bioingénieries (SEVAB) LABORATOIRES : - UMR 5504 & 792-CNRS-INSA-INRA Ingénierie des Systèmes
Biologiques et des Procédés. - UMR 1089-Xénobiotiques INRA-ENVT
2
3
This work was part of a collaboration between France and Mexico. The work presented here was done in Toulouse France at both the UMR-1089 Xénobiotiques, INRA and at the UMR 5504 & 792 d’Ingénierie des Systèmes Biologiques et Procédés, CNRS-INRA-INSA, and in Mexico City at the laboratory number 50 of the Departamento de Biologia Celular del Centro de Investigación y de Estudios Avanzados del IPN (CINVESTAV). This thesis was directed by Pr. Jean-Marie François and Dr. Françoise Guéraud and advised by Dr. Saúl Villa-Treviño. This work was supported by the Program Alβan, the European Union Program of High Level Scholarships for Latin America, scholarship No. E04D03532MX, ECOS-ANUIES grant M02-S01, CONACyT grant 39525-M, COST B35 and fellowship from CONACyT AMQ165505.
4
5
To my family: my mother, Sues, Rafa, to Syndy, Josué, Sofi To my grandmother… We are the champions! To my grandfather… I miss you To my father… I got it!!! We have found that where science has progressed the furthest, the mind has regained
from nature that which mind has put into nature. We have found a strange footprint on
the shores of the unknown. We have devised profound theories, one after another, to
account for its origin. At last, we have succeeded in reconstructing the creature that
made the footprint. And lo!... It is our own
Sir Arthur Eddington
In What the Bleep do we know?
6
7
ACKNOWLEDGMENTS I would like to express my gratitude to my supervisor, Dr. Françoise Guéraud, whose expertise, patience and friendship added considerably to my graduate experience and to my personal life. I am also extremely grateful to Prof. Jean-Marie François and Dr. Saul Villa-Treviño for the direction of this thesis. Special thanks to Dr. Alain Paris for all the indispensable notes and remarks in the building and development of this ambitious project. I appreciate his vast knowledge in many areas (biology, statistics, physiology, etc) and his patience to share a little bit of it with me. I want to thank Dr. Eric Ezan, Research director at CEA, Saclay-France; Prof. Tilman Grune, professor at the University of Hohenheim, Stuttgart-Germany; and Dr. Gabriel del Rio-Guerra, professor at the Universidad Nacional Autónoma de México, UNAM-Mexico for having accepted of being my external reviewers. Thanks also to Prof. Denis Corpet for accepting to be part of my thesis commitee. Thanks to Balbine Roussel for the valorous work in data analysis and statistics, especially for her friendship and her excellent mood that made my stay in France a really nice time. Especially I want to thank Dr. Julio Pérez-Carreón for being a very good friend and an excellent collaborator and for all the ideas and debates that we share in the development of this work (You would be an excellent psychologist!!!). Thanks to Dr. Neven Žarkovic and Dr. Kamelija Žarkovic for giving us the monoclonal antibody anti-HNE-his, and for their collaboration in histological analyses and interpretation. Thanks also to Ana Čipac for her helping in immunohistochemistry of HNE-his. I am also grateful to all the people of the laboratory for oxidative stress at the Rudjer Boskovic Institute; Zagreb, Croatia, for their welcome in my stay in Zagreb. Thanks to Dr. Jean-Pierre Cravedi, Director of the UMR-1089 Xénobiotiques for receiving me at the laboratory. This thesis would not be possible without all the technical assistance: thanks to Samia Fattel-Fazenda for her help in histological methodologies; to Dr. Cécile Canlet for NMR analyses. Thanks to Céline Domange and Julie Labat for technical helping in liver extraction and sample preparation for NMR analyses. I am also grateful to the Biopuce plateform staff (Dr.Véronique Le Berre, Dr. Serguei Sokol, Lidwine Trouilh, and Delphine Labourdette) for their assistance in transcriptomic work. Thanks to Florence Blas-y-Estrada and Raymond Gazel for animal care. Thanks to Elisabeth Perdu and Dr. Patrick Rouimi for their assistance in biochemical procedures. I am very grateful to Pascal Martin for his remarks and help in data treatment and statistics. I would like to thank all the people of the UMR-1089 INRA/ENVT for their welcome during my stay in France, where I met very good friends and excellent people that I will remember with a lot of happiness. Thanks to: Thierry (The best french teacher I could ever have!), Fabrice, Sylviane, Denis, Nathalie Naud, Nathalie Priymenko, Julian, Maysaloun, Céline Dargent, Céline Domange, Maryse, Jerôme, Delphine, Emilien, Anne, Haïfaa, Chary, François, Suzette, Marie-Hélène, Cécile, Patrick, Laurence and all the others. Finally, I would like to thank my family and my friends for encouraging me to make my PhD, even when we were far away for a while.
8
9
PUBLICATIONS:
• Marquez-Quiñones A, Paris A, Roussel B, Perez-Carreon J, LeBerre V, François JM, Villa-Treviño S, Guéraud F. Proteasome activity deregulation in LEC rat hepatitis: following the insights of transcriptomic analysis. OMICS: A Journal of Integrative Biology. In press
• Marquez A, Villa-Treviño S, Guéraud F. The LEC rat: a useful model for
studying liver carcinogenesis related to oxidative stress and inflammation. Redox Report 2007;12(1):35-9
• García-Roman R, Pérez-Carreón JI, Márquez-Quiñones A, Salcido-Neyoy ME,
Villa-Treviño S. Persistent activation of NF-kappaB related to IkappaB's degradation profiles during early chemical hepatocarcinogenesis. J Carcinogenesis 2007; 6:5
COMMUNICATIONS:
• HNE-Club Meeting; Genoa, Italy (June16/18, 2006); oral communication: “Correlation between oxidative stress and tumor marker expression in the onset of hepatocarcinogenesis in LEC rat”.
• International Free radical Summer School 2006: Biomarkers of oxidative stress, Spetses Island, Greece (September 30-October 6, 2006); poster presentation: “Gene expression profiles from oxidative stress-related hepatocarcinogenesis model: The LEC rat model”. With this presentation we have won the poster award.
• Rudjer Boskovic Institute; Zagreb, Croatia (November 16, 2006); invited oral communication: “Reactive Oxygen Species and Carcinogenesis: An integrative approach combining genetic and metabonomic profiles”.
• Deuxièmes journées scientifiques du Réseau Français de Métabonomique et Fluxomique, Saint-Sauves d'Auvergne, France (13-15 décembre 2006); poster presentation: “The LEC rat hepatocarcinogenesis model: Relationship between transcriptomic and metabonomic signatures”.
• SFRR-europe 2007 Meeting; Vilamoura, Portugal (October 10-13, 2007); oral communication in the “Young Investigator Session”: “Proteasome activity deregulation in LEC rat hepatitis: following the insights of transcriptomic analysis”.
10
11
ABBREVIATIONS: 1H NMR Proton nuclear magnetic resonance
8-IsoPGF2α 8-Isoprostaglandine F2α
8-OHG 8-Hydroxyguanine
ALT Alanine aminotransferase
ARE Antioxidant responsive element
AST Aspartate aminotransferase
BHT Butylhydroxytoluene
CAT Catalase
ChT Chymotrypsin-like
Cu-Mt Copper-metallothionein complexes
DHN-MA 1,4-dihydroxynonane-mercapturic acid
FDA Factorial discriminant analysis
G6Pase Glucose-6-phosphatase
GGT γ-glutamyltranspeptidase GO Gene ontology
GPX Glutathione peroxidase
GSH Reduced glutathione
GSSG Oxidized glutathione
GST Glutathione S-transferase
GSTP Glutathione S-transferase type pi
HBC Hepatitis B virus
HCC Hepatocellular carcinoma
HCV Hepatitis C virus
HNE 4-Hydroxy-2-nonenal
LE Long Evans
LEC Long Evans Cinnamon-like color
MDR Multi-drug resistant protein
NRS Nitrogen reactive species
OH· Hydroxyl radical
OSC Orthogonal signal correction
PGPH Peptidylglutamyl peptide hydrolase
PLS Partial least square
PLS-DA Partial least square-Discriminant analysis
qRT-PCR Quantitative RT-PCR
ROS Reactive oxygen species
SAGE Serial analysis of gene expression
SOD Superoxide dismutase
t-bil Total bilirubin
TL Trypsin-like
VIP Variable Importance Projection
WND Wilson's protein
12
13
FIGURE LIST Figure 1 Sources of cell ROS production 25 Figure 2 Lipid peroxidation process 28 Figure 3 Clinico-pathological description of LEC rat spontaneous hepatitis
and hepatocarcinogenesis 42
Figure 4 Histological and biochemical features of LEC rats at different hepatitis stages
72
Figure 5 Tumor marker expression in liver of LEC rats at different hepatitis stages
74
Figure 6 HNE-protein adduction in LEC rat liver at different hepatitis stages
76
Figure 7 Lipid peroxidation products excretion in urine of LEC rats at different hepatitis stages
77
Figure 8 Diet consumption at different stages of hepatitis in LEC rats 78 Figure 9 Effect of diet consumption on lipid peroxidation products’ urinary
excretion 78
Figure 10 Effect of D-penicillamine on hepatitis development, carcinogenesis initiation and oxidative stress
80
Figure 11 Histological and biochemical parameters comparison between LEC rats groups (Normal 6w and D-penicillamine) and Long Evans rats
82
Figure 12 Ratio/Intensity scatter plots 92 Figure 13 Relation between hepatitis progression and gene expression 93 Figure 14 Factorial Discriminant Analysis (FDA) of transcriptomic data 95 Figure 15 Confirmation of the differential expression of GSTP by
quantitative RT-PCR 96
Figure 16 Gene ontology classification of 59 known genes 97 Figure 17 Metabolism ontology sub-classification 98 Figure 18 Proteasome structure 101 Figure 19 Heat map of proteasome related genes in the liver of LEC rats at
different disease stages 102
Figure 20 Changes in proteasome activities in the liver of LEC rats at different disease stages
104
Figure 21 Immunodetection of proteasome in liver of LEC rats at different disease stages and treatments
105
Figure 22 Effect of D-penicillamine on gene expression patterns 106 Figure 23 FDA of 1H NMR spectra of urine samples 117
Figure 24 PLS-2 regression between 1H NMR metabolic profiles (t[1] axis) and plasma levels of hepatitis markers (u[1] axis)
118
Figure 25 Pattern recognition analysis of 1H NMR spectra of lipophilic extracts from liver samples of LEC rats at different hepatitis stages
120
Figure 26 Pattern recognition analysis of 1H NMR spectra of aqueous extracts from liver samples of LEC rats at different hepatitis stages
122
Figure 27 PLS-2 regression between transcriptomic and metabonomic data 125 Figure S1 1D 1H NMR spectra of liver aqueous fractions 171
Figure S2 1D 1H NMR spectra of liver aqueous fractions 172
14
TABLE LIST Table I Principal ROS studied in biological systems 24 Table II Array hybridization design 90 Table III ∆2 Mahalanobis distance between all different groups
represented in FDA test 95
Table IV Differential genes classified in GO: Protein Metabolism 99 Table V Differential genes classified in GO: Nucleobase, Nucleotide
and Nucleic acid Metabolism 100
Table VI Identified metabolites that were related to hepatitis development in LEC rats
118
Table VII ∆2 Mahalanobis distance between groups from FDA analysis of 1H NMR spectra of liver lipophilic extracts
120
Table VIII
∆2 Mahalanobis distance between groups from FDA analysis of 1H NMR spectra of liver aqueous extracts
122
Table IX Discriminant variables selected by ANOVA (P < 0.05) in a FDA
123
Table A List of differential genes that participate in PLS-DA axis construction
153
Table B Correlation between transcriptomic and metabonomic data 167
15
PROLOGUE 19 A. INTRODUCTION 21
A.1 THE CONCEPT OF OXIDATIVE STRESS 23
A.1.1 Oxidative DNA damage 26
A.1.2 Lipid peroxidation 27
A.1.3 Protein modification by ROS 29
A.1.4 Cellular adaptive responses to oxidative stress 30
A.2. OXIDATIVE STRESS AND LIVER DISEASES 32
A.2.1 Wilson’s disease 32
A.2.2 Hepatocarcinogenesis 33
A.3 OMICS TECHNOLOGIES AND SYSTEMS BIOLOGY 35
A.3.1 Omics approaches, what do they mean? 35
A.3.2 Transcriptomics 36
A.3.3 Proteomics 37
A.3.4 Metabonomics 38
A.3.5 Systems biology: Integrating the puzzle 38
A.4 THE LEC RAT MODEL 40
A.4.1 Copper accumulation in LEC rats 40
A.4.2 Clinicopathological characteristics of LEC rats 41
A.4.3 Mechanism of hepatocarcinogenesis in LEC rats 43
B. AIMS OF THE STUDY 47
C. MATERIAL AND METHODS 53
C.1 ANIMALS AND TREATMENTS 55
C.2 URINE COLLECTION AND STORING 55
C.3 TISSUE PREPARATION 55
C.4 PLASMA HEPATITIS MARKERS DETERMINATION 56
C.5 OXIDATIVE STRESS DETERMINATION 56
C.5.1 8-Isoprostaglandine F2α (8-IsoPGF2α) urine determination by Enzyme
Immunoassay (EIA) 56 C.5.2 1,4-dihydroxynonane-mercapturic acid (DHN-MA) urine determination by
Enzyme Immunoassay (EIA) 57
16
C.6 HISTOLOGY ANALYSES 57
C.6.1 Hematoxylin and eosin staining 57
C.6.2 Immunohistochemistry of Glutathione S-transferase, placental
form (GSTP) 58
C.6.3 Immunohistochemistry of 4-hydroxynonenal (HNE) protein adducts 58
C.6.4 Histological detection of gamma-glutamyl transpeptidase (GGT)
positive hepatocytes 59 C.7 PROTEASOME PEPTIDASE ACTIVITIES 59
C.7.1 Cytosol extraction 59
C.7.2 Proteasome activity assay 60
C.8 SDS-PAGE AND IMMUNOBLOTTING 60
C.9 RNA EXTRACTION 61
C.10 QUANTITATIVE RT-PCR (QRT-PCR) 61
C.11 MICROARRAY ANALYSIS 61
C.11.1 DNA microarray slides 61
C.11.2 Preparation of labeled cDNA 62
C.11.3 Array hybridization and scanning 62
C.11.4 Data analysis and public access 63
C.12. METABONOMIC ANALYSES 63
C.12.1 Samples preparation 63
C.12.2 1H NMR spectroscopic analysis 64
C.12.3 Spectra analysis 65
C.13 STATISTICAL ANALYSES 65
D. RESULTS AND DISCUSION 67
D.1 CHAPTER 1: 69
CLINICAL AND BIOCHEMICAL CHARACTERISATION OF LEC RAT
MODEL
D.1.1 INTRODUCTION AND AIMS 71
D.1.2 EXPERIMENTAL PROTOCOL 71
D.1.3 RESULTS 73
D.1.3.1 Hepatitis development in LEC rats 73
D.1.3.2 GGT and GSTP expression in LEC rat hepatocytes 75
D.1.3.3 Oxidative stress induction in LEC rats 75
17
D.1.3.4 Effect of D-penicillamine on hepatitis development 79
D.1.3.5 Comparison between D-penicillamine-treated and normal rats 81
D.1.4 DISCUSSION 81
D.2 CHAPTER 2: 87
TRANSCRIPTOMIC ANALYSIS OF LEC RAT HEPATITIS DEVELOPMENT
D.2.1 INTRODUCTION AND AIMS 89
D.2.2 EXPERIMENTAL PROTOCOL 89
D.2.3 RESULTS 91
D.2.3.1 Microarray analysis and validation 91
D.2.3.2 Functional classification of VIP genes 96
D.2.3.3 Proteasome activity 101
D.2.3.4 Effect of D-penicillamine on gene expression 103
D.2.4 DISCUSSION 106
D.3 CHAPTER 3 113
METABONOMIC ANALYSIS AND ITS RELATION WITH
TRANSCRIPTOMIC DATA
D.3.1 INTRODUCTION AND AIMS 115
D.3.2 EXPERIMENTAL PROTOCOL 116
D.3.3 RESULTS 116
D.3.3.1 1H NMR OF URINARY SAMPLES OF LEC RATS BEFORE AND DURING
THE ONSET OF HEPATITIS 116
D.3.3.2 METABOLIC CHANGES IN LIVER OF LEC RATS AT DIFFEENT
HEPATITIS STAGES 119
D.3.3.2.1 Metabonomic analysis of lipid extracts of LEC rat liver samples 119
D.3.3.2.2 Metabonomic analysis of aqueous extracts of LEC rat liver samples
121
D.3.3.3 QUANTITATIVE RELATIONSHIP BETWEEN METABONOMIC AND
TRANSCRIPTOMIC ANALYSES 124
D.3.4 DISCUSSION 124
18
E. GENERAL DISCUSSION AND PERSPECTIVES 131
F. REFERENCES 139 G. ANNEXES 151 H. PUBLICATIONS 173
19
PROLOGUE
This project was part of a collaboration between France and Mexico that has been
supported by the ECOS-ANUIES-CONACyT program (project: M02-S01 “Education
in DNA array technology and practical application in functional genomic in eukaryotic
cells”). The interest in studying the mechanism related to degenerative diseases is to
search novel approaches to early detection and treatment of those diseases.
The aim of this project was to define the gene expression and metabolic changes that
take place during the oxidative stress-related hepatitis and hepatocarcinogenesis. In
Toulouse (France) we found both the animal model and the technology to achieve this
objective.
The LEC rat model is considered a model of interest, because it is shown to generate
liver cancer related to inflammation and oxidative stress. LEC rats are bred in the
Institut National de la Recherche Agronomique (INRA), where the metabolism of lipid
peroxidation products (reflects of oxidative stress) and their biological importance are
studied. Using the “omics” technologies, it was possible to evaluate the global mRNA
expression and metabolic end products. These “omics” technologies, transcriptomics
and metabonomics, have been successfully developed in the institutions in where the
project took place: the microarray technology at the Institut National des Science
Appliquées (INSA), within the Toulouse Genopôle; and the metabonomics at INRA.
The knowledge and the experience of these groups in biochemical and functional
genomics areas, allowed me the use of these new approaches to formulate some new
hypothesis of oxidative stresss-induced hepatitis that could be further analyzed in the
effort to better understand pathologies as hepatitis and hepatocarcinogenesis.
20
21
A. INTRODUCTION
22
23
A.1 THE CONCEPT OF OXIDATIVE STRESS The redox state of a cell is defined by the reduction potential (Eh) of all the redox
couples present in the cell, and because of its high abundance, glutathione (GSH)
represents one of the principal molecules that define the cellular redox potential
(Schafer et al., 2001). The redox environment of the cell can be estimated through the
redox state of the glutathione couple (GSSG/2GSH). Modifications in the reduction
potential of the GSSG/2GSH couple have been associated with the biological status of
the cell, for example, cell proliferation takes place at a Eh ≈ -240mV, differentiation at a
Eh ≈ -200mV; or apoptosis at a Eh ≈ -170mV (Schafer et al., 2001). These data point out
the importance of redox balance in biological systems, and is the basis of the oxidative
stress concept.
By definition oxidative stress is an increase in the reduction potential or a large decrease
in the reducing capacity of the cellular redox couples (Genestra, 2007). In other words,
oxidative stress is the result of an increased oxidized molecule production and a
diminution of their elimination. One group of these oxidized molecules responsible for
oxidative stress is the reactive oxygen species (ROS) family. As defined by their name,
ROS are species of oxygen which are in a more reactive state than molecular oxygen,
and by which therefore, oxygen is reduced to varying degrees. ROS compile a big
family of molecules, the most studied in biological systems are described in Table I.
ROS are produced by all aerobic cells. Under physiological conditions mitochondrial
production of ROS has been estimated for about ~2% of the total oxygen uptake by the
organisms (Inoue et al., 2003).
ROS can be produced by endogenous and exogenous sources (Figure 1). Endogenously,
ROS arise from mitochondria, cytochrome P450 metabolism, peroxisomes and
24
OXIDANT REPRESENTATION CHARACTERISTICS
Superoxide anion O2-
One-electron reduction state of O2. Formed in many autoxidation reactions and by the electron transport chain. Rather unreactive but can release Fe2+ from iron-sulphur proteins and ferritin
Hydrogen peroxide H2O2
Two-electron reduction state, formed by dismutation of O2
- or by direct reduction of O2. Lipid soluble and thus able to diffuse across membranes. In aqueous solution can oxidize or reduce a variety of inorganic ions
Hydroxyl radical OH·
Three-electron reduction state, formed by Fenton reaction and decomposition of peroxynitrite. Has a very short in vivo half-life of approximately 10-9 s and a high reactivity, will attack most cellular components. It can damage virtually all types of macromolecules: carbohydrates, nucleic acids, lipids and amino acids
Organic hydroperoxide ROOH
Formed by radical reactions with cellular components such as lipids and nucleobases. Specific functional group or a molecule containin and oxygen-oxygen single bond (R-O-O-R)
Alkoxy, perory radicals RO·, ROO·
Oxygen centred organic radicals. Lipid forms participate in lipid peroxidation reactions. Produced in the presence of oxygen by radical addition to double bonds or hydrogen abstractions
Hypochlorite HOCL
Formed fron H2O2 by myeloperoxidase. Lipid soluble and highly reactive. Will readily oxidize protein constituents, including thiol groups, amino groups and methionine.
Peroxynitrite ONNO-
Is an oxidant and nitrating agent, formed in a rapid reaction betwen ·O2 and NO·. Lipid soluble and similar in reactivity to hypochlorite. Can damage a wide array of molecules in cells, including DNA and proteins: Protonation forms peroxynitrous acid, which can undergo homolytic cleavage to form hydroxyl radical and nitrogen dioxide
Table I. Principal ROS studied in biological systems. Adapted from Genestra (2007).
25
inflammatory cell activation (Conner et al., 1996; Valko et al., 2006). Xenobiotic
metabolism, ionizing radiation and cigarette smoking are examples of exogenous
sources for ROS production. The generation of some ROS is closely linked to cellular
redox-active metals, like iron and sometimes copper. Iron and copper content in cells is
a strictly controlled mechanism that ensures that there is no free intracellular iron or
copper ions. However under some stress conditions, release of copper or ion leads to an
uncontrolled ROS production (Leonard et al., 2004).
Figure 1. Sources of cell ROS production
26
Due to their high reactivity, ROS can react very quickly between each other and with
other molecules present in the cells. Interaction of ROS with cellular lipids, proteins and
nucleic acids are thought to be mechanisms by which oxidative stress modulates cellular
faith (Droge, 2002). Here I will describe some of the reactions of ROS with
biomolecules and the consequence of these interactions on cellular damage and
carcinogenesis. However, our current knowledge of ROS biochemistry show that they
have not only deleterious effects but they actually can act as second messengers in cell
signaling (Genestra, 2007).
A.1.1 Oxidative DNA damage
Among ROS, the hydroxyl radical (OH·) is one of the most reactive. OH· is known to
react with all components of the DNA molecule, including purine and pyrimidine bases
and the deoxyribose backbone (Dizdaroglu et al., 2002). Fixation of DNA oxidative
modifications may be involved in mutagenesis and hence in carcinogenesis. Different
types of tumor have been shown to present oxidative-related DNA damage. ROS-induce
DNA damage involves single- or double-stranded DNA breaks, purine and pyrimidine
oxidation and DNA cross-links. Other oxidative stress DNA-modifications include lipid
peroxidation-DNA adducts. Consequences of DNA modifications include transcription
modifications, induction of signal transduction pathways, replication errors and
genomic instability (Valko et al., 2006). Evidences of DNA-ROS interaction in tumors
have been demonstrated by measuring the levels of 8-hydroxyguanine (8-OHG) (Floyd,
1990). 8-OHG is actually considered a potential marker of carcinogenesis (Olinski et
al., 2003). Both mitochondrial and genomic DNA are susceptible of oxidative-related
DNA damage, and both are also related to tumor development (Klaunig et al., 2004).
27
Repair mechanisms of DNA modifications are well studied. The major pathway for the
repair of ROS-induced DNA modifications is considered to be the base excision repair
(BER) pathway (Cadet et al., 2000). Alterations in BER pathway caused by diet
modifications, inflammation or other factors are associated to human cancers, as seen by
a diminution in BER enzymes activities in lung and tobacco-associated cancer patients
(Tudek et al., 2006; Tudek, 2007), enforcing the importance of ROS-induced DNA
alterations in cancer development.
A.1.2 Lipid peroxidation
One of the well recognized targets of oxidative stress are cellular membranes.
Polyunsaturated fatty acid residues of phospholipids are extremely sensitive to
oxidation. Lipid peroxidation refers to the oxidative degradation of lipids. It is the
process by which lipids are attacked by reactive molecules (i.e. ROS) abstracting a
hydrogen atom from a methylene carbon in their side chain. The overall process of lipid
peroxidation consists of three stages: initiation, propagation and termination. Lipid
peroxidation products include alcanes like pentane and ethane (Valko et al., 2006),
aldehydes like 4-hydroxy-2-nonenal (HNE) and malondialdehyde (MDA) (Esterbauer et
al., 1991; Comporti, 1998), and isoprostanes like the 8-IsoPGF2α (8-IsoProstaglandin
F2α) (Pratico et al., 2004) (Figure 2).
Both MDA and HNE have the ability to interact with biomolecules as proteins and
nucleic acids. MDA is mutagenic in bacterial and mammalian cells, it reacts with DNA
to form the premutagenic pyrimido[1, 2-a]purin-10(3H)-one (M1G) adduct (Plastaras et
al., 2000). M1G adduct is mutagenic in E. coli, inducing transversions to T and
transitions to A (Fink et al., 1997). Similarly, HNE is genotoxic for hepatocytes and
cerebral endothelial cells. Treatment of hepatocytes with HNE leads to a spectrum of
28
Figure 2. Lipid peroxidation process. Figure adapted from Young and McEneny (2001) (Young et al., 2001).
29
DNA alterations from sister chromatide exchanges to microsomal aberrations (Eckl,
2003). HNE is considered as a second messenger of oxidative stress (Zarkovic et al.,
1999; Zarkovic, 2003). Cellular signaling leading to apoptosis, proliferation or
differentiation can be modulated by HNE depending on its intracellular concentration
(Yang et al., 2003). HNE has been associated to cancer initiation by direct DNA
adduction (Hu et al., 2002) and by inhibiting the enzymes related to BER (Feng et al.,
2004).
A.1.3 Protein modification by ROS
Protein oxidation results from the reaction of ROS with both amino acid side chains and
peptidic backbone. Oxidative damage to proteins can take place within almost all the
amino acids, being the most reactive with sulfur amino acid-containing proteins (Berlett
et al., 1997). Certain oxidations of proteins are reversible, mainly oxidative
modification of cysteine and methionine can be reversed by enzymatic systems like
glutaredoxin/glutathione/glutathione reductase systems or thiorredoxin/thioredoxin
reductase systems (Holmgren et al., 2005). This reversibility of protein oxidation is very
important in cell signaling pathways and it is considered one of the main mechanisms of
redox signaling (Droge, 2002). Changes in the local redox state of protein sulfhydryls
leads to conformational changes that, depending on the protein, can either diminish or
augment DNA binding activity, or promote protein complex formations which are
necessary for signal transduction or transcription to proceed (Allen et al., 2000).
Irreversible oxidized proteins are hydroxylated and carbonylated amino acid derivates.
These moieties are chemically stable; hence, protein carbonyl groups are used as
biomarkers of oxidative stress. Accumulation of protein carbonyls has been observed in
Alzheimer’s disease, diabetes, inflammatory bowel disease, among others (Dalle-Donne
30
et al., 2003). Oxidized proteins are generally less active, and are exposing hydrophobic
amino acids at their surface, leading to changes in protein conformation and activity
(Friguet, 2006).
Elimination of mild oxidized proteins is preferentially done by the 20S proteasome in a
ATP- and ubiquitin-independent manner (Grune et al., 1995; Grune et al., 2003b).
However, strong protein oxidation or/and under strong oxidative stress condition lead to
proteasome inhibition (Grune et al., 1995). Accumulation of damaged proteins due to
proteasome inhibition has been associated to metabolic perturbations, aging and
different pathologies including Parkinson and Alzheimer diseases (Hayashi et al., 1998;
Ciechanover, 2005; Reed et al., 2007), but little is known about proteasome inhibition in
carcinogenesis.
A.1.4 Cellular adaptive responses to oxidative stress
Living aerobic organisms have developed elaborate sequences of adaptative
mechanisms to regulate oxygen homeostasis. The “antioxidant network” as defined by
Chaudiere et al. includes non-enzymatic and enzymatic antioxidant defenses, cofactors
and metabolic pathways that act in an orchestrated manner in order to minimize the
damaging effects of ROS (Chaudiere et al., 1999).
Non-enzymatic antioxidants include small hydrophobic and hydrophilic molecules.
They act as chemical traps of oxidizing free radicals and ROS. Hydrophilic scavengers
are mainly represented by GSH and ascorbic acid, they are found in cytosolic,
mitochondrial and nuclear aqueous compartments. Hydrophobic antioxidants are found
in lipoproteins and cell membranes where they inhibit or interrupt chain reactions of
lipid peroxidation (Halliwell et al., 1993). They include vitamin E (α-tocopherol),
31
carotenoids and ubiquinol. Essentially, non-enzymatic antioxidants are a first defense
against free radical formation by quenching nonspecifically free radicals.
More sophisticate mechanisms for ROS control include enzymatic antioxidants.
Antioxidant enzymes include superoxide dismutases (SODs), which ensure the
enzymatic degradation of superoxide ion. Hydroperoxides are degraded by catalase
(CAT) and glutathione peroxidase (GPX) or ascorbate peroxidase. GPX and ascorbate
peroxidases are reductases whose reducing coenzymes are regenerated by NADPH
equivalents produced in metabolic pathways (Chaudiere et al., 1999). Thiol-containing
proteins such as thioredoxins, thioredoxin reductase, peroxyredoxins and glutaredoxins
are mainly implicated in reduction of oxidized proteins (Holmgren, 2000). Other
enzymatic systems are designed for ROS and ROS-oxidized molecules elimination.
Glutathione S-transferases (GSTs) inactivate endogenous lipid peroxidation products
(aldehydes such as HNE), quinines, epoxides and hydroperoxides formed as secondary
metabolites during oxidative stress (Hayes et al., 2005).
Cellular defense mechanisms of oxidative stress might be activated by cellular sensors
of oxidative damage. Activation of gene expression in response to oxidative stress is
related to transcription modulation by transcription factors like NFκB, Nrf2 and AP-1
(Allen et al., 2000; Scandalios, 2005). Direct signaling of H2O2 in the differential
regulation of antioxidant genes likely occurs via protein-DNA interactions in the Nrf2-
antioxidant responsive element (ARE) and via NFκB in the promoters of these genes.
For example, HNE induces thioredoxin system gene expression through Nrf2 signaling
in neural cells as an adaptive response to mild oxidative stress (Chen et al., 2005; Tanito
et al., 2007).
Cellular response face to oxidative stress is complex and depends on levels of ROS
production. While mild oxidative stress is related to physiological process such as
32
proliferation and differentiation, strong and uncontrolled oxidative stress leads to cell
death by apoptosis or necrosis (Droge, 2002). Whether ROS effect on cells is beneficial
or pathological depends on complex factors that are not fully understood. Nevertheless,
the role of free radicals, mainly ROS, in pathophysiology is evident. The aim of this
work is to focus on ROS-induced hepatitis as a model for ROS modulation of cell faith
in relation to hepatocarcinogenesis.
A.2. OXIDATIVE STRESS AND LIVER DISEASES
In humans, chronic liver diseases are generated by different causes. Chronic alcohol
consumption, viral infections, metabolic dysfunctions, xenobiotic exposition and metal
overload are factors associated with hepatitis and liver tumor development. Oxidative
stress seems to be a common mechanism involved in all these different hepatitis causes
(Vendemiale et al., 2001; Nagasaka et al., 2006; Seitz et al., 2006). Patients suffering
from hepatitis C manifest hepatic oxidative stress, condition that is exacerbated by
alcohol consumption (Wang et al., 2006). 8-Nitroguanine, a marker of DNA oxidation,
is highly formed in patients with hepatitis C. Furthermore, some hepatitis C virus
(HCV) carriers with normal alanine aminotransferase (ALT) levels in serum have
elevated levels of lipid peroxidation products and low levels of reduced glutathione
(GSH) in plasma. In these patients, a greater degree of oxidative stress markers
correlates with a more severe status of the disease (Vendemiale et al., 2001).
A.2.1 Wilson’s disease
Wilson’s disease is an autosomal recessive disorder of copper transport. Affected
individuals exhibit excessive copper accumulation in the liver and brain, deficient
holoceruloplasmin biosynthesis and a marked impairment in biliary copper excretion
33
(Linder et al., 1996). The gene defective in Wilson’s disease is the ATP7B gene which
encodes a P-type ATPase, the Wilson protein (WND) (Tanzi et al., 1993). WND is
responsible for copper transport from the cytoplasm to the trans-golgy network, where
copper is bounded to apo-ceruloplasmin to be excreted into plasma (Pfeil et al., 1999).
Mutations in WND results in misdistribution of the protein and in a defective
intracellular copper transport. As a consequence copper accumulates in cells, generating
ROS production and hence oxidative stress. Wilson’s disease patients suffer of recurrent
or fatal hepatitis and of neurological disorders. Wilson’s disease patients die either due
to progressive liver failure or neurological deterioration. Hepatocelullar carcinoma
(HCC) is a rare condition in Wilson’s disease, but some cases have been reported
(Iwadate et al., 2004). The low incidence of HCC in patients with Wilson’s disease may
be attributable to the significantly shortened life in untreated patients that does not allow
time for cancer to develop (Pfeil et al., 1999).
A.2.2 Hepatocarcinogenesis
Liver carcinogenesis is associated with chronic inflammation and cirrhosis. Persistent
infection with HCV is a major risk factor for development of HCC. Mechanisms
associated to HCV-related HCC include mitochondrial oxidative stress and metabolic
alterations in lipid and glucose metabolism together with continued cell death and
regeneration associated with inflammation (Koike, 2007). In alcohol-related
hepatocarcinogenesis, oxidative stress generated from alcohol metabolism,
inflammation, and increased iron storage are important mechanisms (Seitz et al., 2006).
Carcinogenesis is a complex multistage process in which both proliferation and cell
death are altered. According to observations made in models of chemical carcinogenesis
(mainly in liver) and in epidemiological findings, the process of neoplastic development
34
has been operationally divided into three defined stages: initiation, promotion and
progression (Farber et al., 1986; Trueba et al., 2004; Anisimov, 2007). Initiation follows
irreversible changes in the genotype of cells (i.e. mutations in critical genes such as
oncogenes and tumor suppressor genes). The promotion stage is characterized by the
clonal expansion of initiated cells by the induction of cell proliferation and/or inhibition
of cell death. This process results in the formation of identifiable focal lesions
(preneoplastic lesions). Progression involves cellular and molecular changes that occur
from the preoneoplastic to the neoplastic state. This stage is mainly characterized by
genetic instability and disruption of chromosome integrity (Klaunig et al., 2004).
During the first stages of carcinogenesis, cells acquire a resistant phenotype, by which
cells are resistant to physiological mechanism of cell growth regulation and also to
cytotoxic environments (Laconi et al., 2000). Hence, initiated cells might be more able
to proliferate and resist to apoptotic physiological stimulus. Such phenotype is seen in
cancer cells, and it is explained by overexpression of genes related to toxic metabolite
excretion like multi-drug resistant protein (MDR), GSTs and GSH-associated enzymes
(Tew, 1994). Liver carcinogenesis is a good example of the phenotypic resistance in the
pathogenesis of neoplasia hypothesis. During initiation, hepatocytes acquire the
expression of proteins related to GSH metabolism. Moreover two of these proteins, the
glutathione-s-transferase type pi (GSTP) and the γ-glutamyltraspeptidase GGT, are
actually considered as early tumor markers (Pitot et al., 1996).
Initiation stage of liver carcinogenesis is closely related to liver injury and consequent
cellular turnover caused by the different etiologies of hepatitis (McGlynn et al., 2005;
El-Serag et al., 2007). Mechanisms by which hepatocytes are transformed into
preneoplastic and neoplastic cells are not well understood, but the influence of ROS and
oxidative stress in this process becomes more evident. Because of the complexity of
35
carcinogenesis, it is difficult to explore the mechanisms of ROS-related
hepatocarcinogenesis in a gene by gene or isolated pathway matter, and a global
analysis of cellular and liver behavior under oxidative stress is necessary.
A.3 OMICS TECHNOLOGIES AND SYSTEMS BIOLOGY
A.3.1 Omics approaches, what do they mean?
From many years science used “reductionism” as tool to explain global phenomena.
Reductionism refers to the attempt of explaining complex phenomena by studying
functional proprieties of the individual components that compose a multicomponent
system (Strange, 2005). Even if these kind of studies remains and will remain essential
for biological understanding of specific situations in a phenomenon, they are not longer
effective in explaining the whole phenomena. Thanks to advances in biotechnology,
analytical chemistry and bioinformatics it is now possible to study, in a large scale, the
principal components of a biological system (Ivakhno, 2007). Genomics is defined as
the study of genes and their products in an organism as a whole. Functional genomics
goals include the integral study of the molecules that form part of the biological
systems, i.e. RNA, proteins and metabolites and which interactions and function lead
life to cells (Hocquette, 2005; Strange, 2005).
In order to understand functional genomics some definitions are necessary:
Genome: The complete collection of hereditary information of an individual organism.
This includes both genes and non-coding sequences of the DNA.
Transcriptome: Set of all messenger RNA (mRNA) molecules or transcripts produced
in one or a population of cells. It reflects the genes that are being actively expressed at
any given time.
36
Proteome: The entire complement of proteins expressed by a genome, cell, tissue or
organism. Specifically this term has been used to define the whole expressed proteins at
a given time point under defined conditions.
Metabolome: Complete set of small-molecule metabolites to be found within a
biological sample.
The global study of each of these fields has lead to new disciplines in biology:
Transcriptomics, proteomics and metabonomics for the study of the transcriptome, the
proteome and the metabolome, respectively (Hocquette, 2005). Together this new
approaches allow the description and comprehension of complex cellular networks in
which communication between genes, proteins and physiological adaptation through
metabolic profile changes are taken in account (Ivakhno et al., 2007). Omic approaches
are promising in understanding disease mechanisms and in the research of new early
markers for the detection of diseases.
A.3.2 Transcriptomics
The study of transcriptomics examines the expression level of mRNAs in a given cell
population, often using high-throughput techniques including differential display, serial
analysis of gene expression (SAGE) and cDNA or oligonucleotid arrays (Chittur, 2004).
The principle of arrays is simple: thousands of probes (DNA fragments corresponding
to different genes) are immobilized on a solid support (nylon membranes or glass slides)
and a specific labeled target made from RNA isolated from the sample of interest is
hybridized with the probes. The signal intensity of each individual probe should
correlate with the abundance of the mRNA complementary to that particular probe
(Brentani et al., 2005).
37
DNA microarrays have been used for the study of several pathologies. In the liver, the
principal objectives imply the research for markers that would allow early diagnosis of
diseases like cirrhosis or HCC to proceed (Chen et al., 2003; Nakatsura et al., 2003).
Using DNA microarrays coupled to unsupervised hierarchical clustering, Lee et al.
classified patients suffering of HCC, their results showed the power of transcriptomic
analyses in tumor classification and prognosis (Lee et al., 2004). Similarly, Perez-
Carreon et al. described a list of differential expressed genes implicated in experimental
chemical hepatocarcinogenesis (Perez-Carreon et al., 2006).
A.3.3 Proteomics
The main objective of proteomics is to quantify protein levels and their dynamic
changes. Proteins are mainly studied by physical separation using two-dimensional (2D)
electrophoresis. The resulting gel contains the proteome separated by the
physicochemical properties of proteins, molecular weight and isoelectric point. Most
current studies in proteomics are based on mass spectrometry to detect and identify
proteins (Spahn et al., 2004). Applications of proteomics for the discovery of
serological tumor markers for clinical use are promising (Poon et al., 2001). More
interestingly, using this kind of analyses, researches have been able to define set of
proteins sensitive to oxidative modifications caused by ischemia/reperfusion damage
(Cesaratto et al., 2005).
Methodological limitations such as difficulties with membrane-associated, very acidic
or very basic, very low or very high molecular weight and very low abundance proteins
are still causing difficulties in proteome study (Humphery-Smith, 2004); nevertheless,
technical improvements are promising (Bunai et al., 2005).
38
A.3.4 Metabonomics
The analysis of the final metabolic products is possible grace to high-output analytical
techniques like 1H nuclear magnetic resonance (1H NMR) and mass spectrometry (MS).
These technologies have lead to the characterization of metabolic modifications caused
by different stimulus such as drug treatment, diseases or xenobiotic exposition (Griffin,
2003). Metabonomics is defined as “the quantitative measurement of the dynamic
multiparametric metabolic response of living systems to pathophysiological stimuli or
genetic modification" (Nicholson et al., 2002). The objective of metabonomics is to
study all metabolites present in a biological system without any bias associated with the
choice of the metabolites to be studied. This helps the identification of metabolic
networks (Lindon et al., 2005; Teahan et al., 2006). One advantage of metabonomics is
the possibility of studying the effects on general metabolism by analyzing urine or
serum samples. This would make possible to follow the time-related metabolic
responses under specific conditions. This is a revolutionizing approach for clinical and
pharmacological applications (Fan et al., 2004).
Metabonomics has been used in the investigation of drug-induced hepatotoxicity, and it
is nowadays considered as a useful tool in pre-clinical and clinical studies in drug
development (Kleno et al., 2004; Mortishire-Smith et al., 2004). In an integrative
clinical approach, using transcriptomic and metabonomic analyses, factors involved in
bad prognosis of neuroendocrine cancers have been described (Fan et al., 2004; Ippolito
et al., 2005).
A.3.5 Systems biology: Integrating the puzzle
The function of living organisms cannot be addressed satisfactorily by looking at
molecules alone. Classical isolated molecule studies are very useful to explain how
39
molecules work, but are very poor in information of how molecules interact in complex,
multicomponent biological processes.
Systems biology is an emerging field of biological science. Its goal is to understand the
integrated function of the different parts of an organism in order to identify molecular
interaction networks on the basis of correlated molecular behavior observed in omic
studies (Strange, 2005). In order to achieve this ambitious aim, it is necessary to link
quantitative data from omic approaches with phenotypic observations. One of the
challenges of omic experiments is how to detect meaningful patterns in the massive
amounts of data generated and how to deduce experimentally testable functions from
those patterns.
Bioinformatics is a tool for biologists useful to aboard this problem; it merges biology
with dedicated statistics methods and databases to develop algorithms and statistical
methods necessary to access, manage and analyze large data sets and assess their
interrelationships (Werner, 2007). In order to better explore omic data, multivariate
statistical analyses have been successfully used (Perez-Enciso et al., 2003). It is out of
scope in this work to deep into the mathematical bases of multivariate statistical
analyses; a basic explanation of the methods used in this thesis is given in when used in
chapter D.2.
The integrative system approaches have great advantages. One is to bring the research
into a new vision of without an a priori formulation of hypothesis. In the past,
hypothesis-driven research was conducted with the aim of testing biological hypothesis
which emerged from previous studies or theoretical considerations. Nowadays, with
omic technologies, new hypothesis are generated with the scope of the key role of some
unknown molecules (Dmitrovsky, 2004). The idea is to search molecular signatures
40
(cluster of genes, of proteins, of metabolites, etc) which would characterize a biological
process or a specific phenotype.
A.4 THE LEC RAT MODEL
The general aim of this work was to define in an integrative approach the gene
expression and metabolic changes during the onset of oxidative stress-induced
hepatocarcinogenesis. In order to achieve this objective we chose Long Evans
Cinnamon-like color (LEC) rats as model.
LEC rats generate spontaneous acute hepatitis, fibrosis and liver tumors as a
consequence of an abnormal liver copper (Cu) accumulation and subsequent oxidative
stress (Mori et al., 1994). This rat strain shows chronic liver damage, hepatocyte death
and regeneration, at the promotion stage of carcinogenesis. Such a natural history of
HCC development in LEC rats is similar to that of human HCC.
A.4.1 Copper accumulation in LEC rats
LEC rat is an inbred strain of a mutant rat that was originally isolated from a closed
colony of Long Evans rats. LEC rats are characterized by excessive Cu accumulation in
the liver and impaired biliary Cu excretion (Li et al., 1991). Cu accumulation produces
great quantities of ROS, mainly the hydroxyl radical (OH·), which is believed to be the
origin of the acute hepatitis and the subsequent HCC that is observed in this rat strain
(Yamamoto et al., 2001).
LEC rats have a deletion in the gene homologous to the Wilson’s disease gene, the
ATP7B gene (Yamaguchi et al., 1994). Wu et al. identified the LEC Atp7b gene
mutation. The disease causing mutation is a deletion of 900 bp in the 3’ terminus which
removes the critical ATP binding domain of the wild type gene leading to a non
functional protein (Wu et al., 1994). Introduction of human ATP7B gene into LEC rats
41
restores biliary Cu excretion and prevents hepatic abnormalities, showing that Atp7b
gene mutation is solely responsible for LEC rat pathologies (Meng et al., 2004).
Hepatocyte Cu distribution in LEC rats progresses from being initially well distributed
in cytoplasm, bound to metallothionein (Cu-Mt complexes) to an accumulation of the
Cu-Mt complexes in the lysosome just before acute hepatitis. The acid conditions in the
lysosome result in the degradation of these complexes resulting in the formation of a
partially polymerized Cu-Mt containing reactive copper (Klein et al., 1998). This Cu
initiates lysosomal lipid peroxidation, leading to hepatocyte necrosis and fulminant
hepatitis.
A.4.2 Clinicopathological characteristics of LEC rats
LEC rats present elevated hepatic Cu levels, reduced biliary Cu excretion, haemolysis,
ceruloplasmin deficiency and increased hepatic iron levels. This mutant strain also
possesses reduced hepatic selenium that may induce a reduction in the antioxidant
capacity against copper induced free-radical damage (Downey et al., 1998).
Around 11-16 weeks old LEC rats suffer from an acute hepatitis period, with symptoms
of jaundice. Kasai et al. have described the clinicopathological characteristics of the
acute hepatitis in the LEC rat (Kasai et al., 1990). Around one week from the first
hepatitis sign, LEC rats suffer from a fulminant hepatitis period in which 40-50% of rats
die. The remaining animals can survive for more than 1 year with chronic hepatitis and
develop preneoplastic and neoplastic lesions of the liver such as hepatocellular
carcinoma (HCC) together with cholangiofibrosis (Figure 3). Most of the liver cancers
are histologically classified as well-differentiated HCC. Sex differences exist in
hepatitis index and liver tumor formation. Male LEC rats have more HCC formation
42
Figure 3. Clinico-pathological description of LEC rat spontaneous hepatitis and hepatocarcinogenesis. Photo: LEC rat (Cinnamon-like color rat) and Long Evans (LE) rat (Black and white color rat). Scheme adapted from Masuda, et. al. 1988 (Masuda et al., 1988).
43
frequency while females develop cholangiocarcinoma more frequently. Moreover,
hepatitis in female LEC rats takes place earlier than in males (Kasai et al., 1990).
The study of liver lesion incidence shows a sequential development of liver foci,
nodules and HCC similar to those seen in chemical hepatocarcinogenesis models.
Furthermore, the phenotype of preneoplastic and neoplastic lesions in LEC rat is the
same as in chemical hepatocarcinogenesis, with increased levels of the positive tumor
markers GSTP, GGT and α-fetoprotein (AFP) and decreased levels of the negative
markers glucose-6-phosphatase (G6Pase) and adenosine triphosphatase (ATPase)
(Sawaki et al., 1990).
A.4.3 Mechanism of hepatocarcinogenesis in LEC rats
Hepatocarcinogenesis in LEC rats is related to liver Cu and iron accumulation, since in
rats fed a low Cu and iron diet or treated with copper-chelating agents, preneoplastic
lesions almost disappear (Jong-Hon et al., 1993; Hayashi et al., 2004). Copper-
associated liver injury is regarded as resulting ultimately from oxidative stress. OH·
production in rats suffering from hepatitis is increased compared with Wistar rats and
LEC rats showing no signs of hepatitis (Yamaguchi et al., 1994). When LEC rats were
treated with the OH· scavenger D-mannitol, a decrease in ALT and bilirubin
concentrations together with a reduction in mitochondrial lipid peroxidation was
observed.
Lipid peroxidation products like HNE and MDA may be implicated in
hepatocarcinogenesis initiation in LEC rats. Increased mutagenic exocyclic DNA
adducts were observed in the liver of LEC rats. These adducts come from the addition
of lipid peroxidation products to DNA (Nair et al., 1996). Levels of etheno-DNA
adducts, 1,N6-ethenodeoxyadenosine and 3,N4-ethenodeoxycytidine, increase with age
44
reaching a peak at 8 and 12 weeks in nuclear and mithochondrial DNA, respectively
(Nair et al., 2005). Acute hepatitis in LEC rats impairs the repair of oxidative DNA base
damage by altering the expression of DNA glycosylases, endonuclease II and 8-
oxoguanine DNA-glycosylase which initiate the BER pathway (Choudhury et al.,
2003).
During hepatitis, p53 expression and hepatocyte apoptosis are higher in LEC rats than
in age matched control rats (Jia et al., 2002). Although p53 expression is increased, Ba
et al., using a yeast-based functional assay, demonstrated the presence of p53 mRNA
mutations in LEC rat liver. The authors suggest that during hepatitis, the cellular
damage degrades transcriptional fidelity and that mutations in p53 may have an effect in
p53 function and hence in cell cycle control (Ba et al., 2000). During chronic hepatitis,
there is a continuous cellular turnover. However the proliferation/apoptotic index ratio
indicates an imbalance in favor of cellular proliferation (Jia et al., 2002). The analysis of
G1 phase-related cell cycle cyclines suggest that while cycline D1 may be involved in
the regeneration of hepatocytes during chronic hepatitis, cyclin dependent kinase 4
(Cdk4) may play an important role in the development of HCC since it is significantly
increased in HCC compared to precancerous and chronic hepatitis LEC rat livers (Kita
et al., 2004). Together theses results indicate that during HCC development in LEC rats,
there is an increase in hepatocyte proliferation rather than a diminution in apoptosis, as
is the case in human HCC.
LEC rat is accepted as good model for Wilson’s disease. Several studies using different
compounds have been made in order to test their efficiency in diminishing hepatic
failure and HCC development. Cu-chelating agents like D-penicillamine, trientine
dihydrochloride and N-benzyl-D-glucamine dithiocarbamate show good results in
inhibiting hepatitis and HCC formation by preventing the Cu-dependent ROS
45
production in LEC rats (Jong-Hon et al., 1993; Hayashi et al., 2004; Shimada et al.,
2005). D-penicillamine is one of the treatments of choice for Wilson’s disease patients.
D-penicillamine not only prevents the development of hepatitis, but it can reverse the
hepatitis evolution by dissolving the polymerized Cu-Mt complexes and diminishing
ROS production (Klein et al., 2000). Antioxidant treatment has successfully diminished
the incidence of hepatic failure in LEC rats. N-acetylcysteine (Kitamura et al., 2005),
proline, ascorbic acid, thioredoxin in combined administration have significantly
delayed the appearance of jaundice and rat mortality (Hawkins et al., 1995).
46
47
B. AIMS OF THE STUDY
48
49
The interest of studying the hepatitis process using global approaches comes from the
idea to evaluate the changing phenotype that leads to hepatitis and cancer initiation
without any bias that could be produced by pre-formulated hypothesis. This would
allow us to search for new markers for early detection and treatment of hepatitis and to
increase the knowledge of oxidative stress-related mechanisms of disease and cell
regulation.
The general aim of this work was to define the oxidative stress-induced gene
expression and metabolic changes during the hepatitis development and the onset
of the hepatocarcinogenesis process.
In order to achieve this objective we have chosen the LEC rat model, in which ROS
induction is closely related to hepatitis and hepatocarcinogenesis development. As
hepatitis in this rat strain is spontaneous, its development is not synchronic. The first
goal of the project was then to define some biochemical parameters in order to get a
good classification of rats according to their hepatitis state and their degree of
oxidative stress.
Hepatocarcinogenesis in LEC rats has been well studied and it has been established as a
multistep process that follows an initiation-promotion-progression development of
preneoplastic and neoplastic lesions. Hepatocyte initiation is thought to take place
during hepatitis, but there are no clear results of it and some controversial data point out
hepatocyte initiation before hepatitis development as shown by increased levels of
tumor marker expression in hepatocytes in LEC rats without hepatitis. In order to define
the period of hepatocytes initiation we aimed to explore the tumor marker expression
in different LEC rat groups before and during hepatitis development.
The election of a suitable control group to compare both ROS-induced gene expression
and metabolic perturbations that lead to hepatitis and cancer initiation was a critical
50
stage in this work. We explored the use known preventive strategies to obtain a group of
rats that behave as normal rats without oxidative stress or hepatitis induction. To
validate D-penicillamine treated LEC rats as a suitable control group for
transcriptomic and metabonomic approaches, we evaluated the biochemical and
histological parameters of hepatitis development in these rats and we compared them
against non diseased LEC rats and Long Evans (LE) rats.
Transcriptomic changes in LEC rats towards hepatitis were defined using DNA
microarray approaches comparing those LEC rats at different hepatitis stages against the
control group of D-penicillamine-treated LEC rats. In order to explore the high amount
of data coming from transcriptomic data, multivariate statistical analyses were
performed. To define those changes in gene expression that might have an impact
in the development of hepatitis we used pattern recognition methods that allowed us to
discriminate between non diseased and diseased animals and to correlate the known
biochemical markers of hepatitis. Those selected genes would be useful for making
some new hypotheses concerning ROS-induced cell damage and cancer induction.
Changes in gene expression can lead to metabolic perturbations by changing enzyme
cell content. Moreover, ROS can directly modulate general metabolism by altering
enzyme and protein function. We were then interested in studying the metabolic end
product changes in a global context using metabonomic approaches. The evaluation
of small molecules in urine that might be related to hepatitis development may lead to
the identification of new markers for early detection that could be analyzed in a non
invasive way. On the other hand, the evaluation of changes in liver metabolome could
be useful for studying the mechanism of ROS-induced inflammation and cancer
initiation.
51
Systems biology refers to the integration of omic data in the objective of better
understanding the global function of cells and organisms. In an effort of better
understanding hepatitis and cancer processes, we aimed to integrate the data coming
from transcriptomic analyses to those coming from metabonomics. The correlated data
would be useful to define some of the cellular networks associated to ROS-induced liver
damage.
52
53
C. MATERIAL AND METHODS
54
55
C.1 ANIMALS AND TREATMENTS
Long Evans Cinnamon-like (LEC) rats were bred in our Institution animal facility from
rats kindly given by Dr. Matsumoto from Tokushima University (Japan). Male LEC rats
were maintained in metabolic cages from 6 weeks old until sacrifice at different
hepatitis stages. Male 5 weeks old Long Evans rats were purchased from Janvier (Le
Genest-Saint-Isle, France) and maintained for acclimatizing for 1 week before the
beginning of experiments at 6 weeks old. Untreated Long Evans rats were sacrified at
13 weeks old. For D-penicillamine-treated rats group, 6 LEC and 6 Long Evans rats
were administered with D-penicillamine in drinking water (100 mg/kg/day) for 6 weeks
until sacrifice at 13 weeks old. D-penicillamine (Sigma, Saint-Quentin Fallavier,
France) concentration was adjusted every three days according to water consumption.
C.2 URINE COLLECTION AND STORING
Urine samples were collected daily from rats housed in plastic metabolic cages. 0.5 ml
of a 360 mM butylhydroxytoluene (BHT) ethanolic solution was added to the urine
collection tubes that were placed in a unit maintained at 0 ºC during urine collection.
24-h collected urine samples were stored at -20 ºC until analysis.
C.3 TISSUE PREPARATION
Rats were sacrified by exsanguination under ether anesthesia. Blood was taken from the
aorta vein for plasma separation. Liver was excised, washed in physiological saline
solution. For RNA isolation, liver samples were immersed in RNAlater (Qiagen,
Courtaboeuf, France) following manufacturer’s instructions and stored at -80 ºC. For
histological examination, liver samples were fixed in 4% buffered formaldehyde,
embedded in paraffin. For GGT activity histochemistry, liver fractions were frozen by
56
immersion in liquid nitrogen-cooled 2-methylbutane and stored at -80 ºC. The rest of
liver samples were immediately frozen in liquid nitrogen and stored at -80 ºC.
C.4 HEPATITIS MARKERS IN PLASMA DETERMINATION
Aspartate aminotransferase (AST), alanine aminotransferase (ALT) activities and total
bilirubin (t-bil) levels in plasma were determined by the Toulouse Genopôle,
“Exploration physiopathologique”.
C.5 OXIDATIVE STRESS DETERMINATION
C.5.1 8-Isoprostaglandine F2α (8-IsoPGF2α) urine determination by Enzymatic
Immunoassay (EIA)
This assay is based on the competition between 8-IsoPGF2α and 8-IsoPGF2α-
acetylcholinesterase (AChE) conjugate (8-IsoPGF2α tracer) for a limited number of 8-
IsoPGF2α-specific rabbit antiserum binding sites (Pradelles et al., 1985; Wang et al.,
1995). Because the concentration of the 8-IsoPGF2α tracer is held constant while the
concentration of 8-IsoPGF2α varies, the amount of 8-IsoPGF2α tracer that is able to bind
to the rabbit antiserum will be inversely proportional to the concentration of 8-IsoPGF2α
in the sample. This rabbit antiserum-8-IsoPGF2α (either free or tracer) complex binds to
the anti-rabbit IgG mouse monoclonal antibody that is attached to the well. A
calibration curve from 7.2 pg/ml to 1 ng/ml is used. Samples were assayed in duplicates
with a 1:20 dilution. After an incubation period of 18 h with the specific antibody, tracer
is added for another incubation period during 4 h. Then, the plate is washed to remove
any unbound reagent and Ellman’s Reagent (which contains the AChE substrate) is
added to the well. The product of this enzymatic reaction is determined
spectrophotometrically at 414nm using a plate reader (Multiskan Ascent, Thermo
57
Labsystem, Cergy Pontoise, France). The color intensity is proportional to the amount
of 8-IsoPGF2α tracer bound to the well, which is inversely proportional to the amount of
free 8-IsoPGF2α present in the well during the incubation. All results were expressed on
the basis of 24 h excretion.
C.5.2 1,4-dihydroxynonane-mercapturic acid (DHN-MA) urine determination by
Enzymatic Immunoassay (EIA)
This competitive immunoenzymatic assay follows the same principle of that of the 8-
IsoPGF2α described above. EIA analyses were performed according to Gueraud et al
(Gueraud et al., 2006). Urine samples were assayed in duplicates with a 1:200 dilution.
A calibration curve was performed with a standard solution (32 pg/ml to 10 ng/ml of
DHN-MA). After an incubation period of 18 h at 4ºC with the specific antibody and
tracer, the plate was washed and the tracer fraction bound to the antibody was reacted
with Ellman’s reagent, which is the acetylcholinesterase enzyme substrate (tracer). The
resultant colored reaction was read at 414 nm using a plate reader (Multiskan Ascent),
color development being inversely proportional to the concentration of DHN-MA
measured. All results were expressed on the basis of 24 h excretion.
C.6 HISTOLOGY ANALYSES
C.6.1 Hematoxylin and eosin staining
5 µm liver sections were obtained from formalin-fixed paraffin samples with a
microtome, and sections were attached to silane-coated microscope slides. Slides were
deparaffinized and rehydrated as follows: two incubations of 5 minutes in xylene
followed by two incubations of 4 minutes in 100% ethanol, one incubation of 4 minutes
in 96% ethanol and one incubation of 4 minutes in 70% ethanol. Slides were then
58
washed in double distilled water. Sections were then stained with Mayer’s hematoxylin
for 6-10 seconds and immediately washed with tap water. Then slides were immersed in
acid ethanol (1% HCl and 70% Ethanol) for 2 minutes and then washed with tap water.
Subsequently, slides were stained with eosin for 1-3 minutes and then washed in
distilled water. After staining, slides are dehydrated and mounted with Canada balsam.
C.6.2 Immunohistochemistry of Glutathione S-transferase, placental form (GSTP)
Deparaffinized sections were blocked for 1 hour in 0.1% H2O2 in phosphate-buffered
saline (PBS), pH 7.4. Subsequently, they were incubated with commercial monoclonal
antibodies specific to glutathione S-transferase, placental form (GSTP,
DakoCytomation, Glostrup, Denmark) diluted 50 times in blocking buffer overnight.
After washing with PBS, the primary antibody was detected using an avidin-biotin
complex immunoperoxidase technique (Zymed Laboratories, Inc., Carlsbad, CA).
Slides were counterstained with hematoxyline and observed in light microscopy.
C.6.3 Immunohistochemistry of 4-hydroxynonenal (HNE)
In order to diminish the in situ formation of HNE, slides were incubated with H2O2 as
substrate for peroxidase only after antibodies attachment. Hence, deparaffinized
sections were incubated in methanol for two minutes then they were washed in PBS
three times of five minutes each. Subsequently, slides were incubated overnight with
monoclonal antibodies specific to HNE-histidine adducts diluted (1:5) in 1% bovine
serum albumin diluted in PBS. After washing with PBS, slides were incubated with 3%
H2O2 in PBS for 15 minutes in the dark. Then, slides were washed in TBS three times
for 5 minutes. Subsequently the primary antibody was detected using a Dako LSBA2
59
avidin-biotin complex immunoperoxidase system (DakoCytomation). Slides were
counterstained with hematoxyline and observed with a light microscopy.
C.6.4 Histological detection of gamma-glutamyl transpeptidase (GGT) positive
hepatocytes
GGT activity can be detected in frozen liver histological sections according to
Rutenburg et al. (Rutenburg et al., 1969). Representative 5 µm thick sections from 2-
methylbutane-protected frozen liver slices were obtained with a cryostat at -15ºC. Slides
were fixed in 96% ethanol for 5 minutes at -20ºC. Slides were then incubated for 20
minutes at room temperature in a reaction buffer containing 125 µg/ml gamma-
glutamyl-1-4-methoxy-2-nafthylamide (GMNA), 0.5 mg/ml glycil-glycine and 0.5
mg/ml fast-blue in 100 mM Tris pH 7.5. A red precipitate is then formed in GGT
positive cells. Slides were then rinsed with distilled water and incubated in 0.1M cupric
sulfate for 5 minutes at room temperature. Finally, slides were rinsed with distilled
water and dried. Slides were counterstained with hematoxyline and observed with a
light microscopy.
C.7 PROTEASOME PEPTIDASE ACTIVITIES
C.7.1 Cytosol extraction
10% (w/v) liver homogenates were prepared in 50 mM Tris-HCl pH 8.0 containing 0.1
mM EDTA and 1 mM β-mercaptoethanol. Homogenates were centrifuged at 100,000 g
for 1 h at 4 ºC; supernatants were stored at -80 ºC until use.
60
C.7.2 Proteasome activity assay
Chymotrypsin-like (ChT), trypsin-like (TL) and peptidylglutamyl peptide hydrolase
(PGPH) proteasome activities were determined with fluorogenic synthetic peptides N-
Succinyl-Leu-Leu-Val-Tyr-7-amido-4-methylcoumarin (LLVY-AMC), BOC-Leu-Ser-
Thr-Arg-7-amido-4-methylcoumarin (BOC-AMC) and Z-Leu-Leu-Glu-β-
naphthylamide (Z-LLG-NA), respectively. Assays were performed as described by
Fataccioli et al. (Fataccioli et al., 1999). Briefly, cytosolic fractions containing 150 µg
of protein were incubated with fluorogenic substrates (50 mM) and SDS 0.06 % in 150
mM Tris-HCl pH 8.0 for 30 minutes at 37 ºC. The reaction was stopped by adding 1 ml
SDS 1% and 2 ml 0.1 M sodium borate pH 9.1. The peptidase activity was determined
fluorometrically by measuring the release of 7-amino-4-methylcoumarin (λexc = 370 nm,
λem = 430 nm) and 2-naphthylamine (λexc = 323nm, λem = 400 nm) in a Jobin Yvon
spectrofluo JY3D fluorometer (Horiba Jobin Yvon, Longjumeau, France). A standard
curve of fluorescence for 7-amino-4-methylcoumarine and 2-naphthylamine was used to
calculate the concentration of liberated products in the assay.
C.8 SDS-PAGE AND IMMUNOBLOTTING
Cytosolic protein fractions were separated by SDS-PAGE using a mini-PROTEAN II
electrophoresis cell (Bio-Rad, Marnes la Coquette, France). Following transfer, the
nitrocellulose membranes were probed with antibodies to proteasome C2 subunit,
PROS-30 (Santa Cruz Biotechnology, Inc., Tebu-Bio, Le Perray en Yvelines, France),
or α-Tubulin (Clone B-5-12) (Sigma) diluted 1:300 and 1:1000 respectively. Antibodies
were visualized using the ECL chemiluminescent system from Amersham Biosciences
(GE Healthcare, Ramonville Saint-Agne, France).
61
C.9 RNA EXTRACTION
Total RNA extraction from RNAlater-stabilized liver samples was performed following
the instructions from Qiagen RNeasy mini kit using the On-Column DNase digestion
set (Qiagen, Courtaboeuf, France). Quality and quantity were determined by capillary
electrophoresis on the bioanalyzer from Agilent (RNA 6000 Nano Assay, Agilent,
Massy, France) and by spectrometric analysis on a NanoDrop ND-1000
spectrophotometer (Nyxor Biotech, Labtech, Palaiseau, France). Only samples with λ260
nm/λ280 nm ratio above 1.9 and rRNA28S/ rRNA18S ratio above 1.7 were taken for DNA
microarray and quantitative RT-PCR assays.
C.10 QUANTITATIVE RT-PCR (QRT-PCR)
2 µg of total RNA were reverse transcribed using SuperScript II-reverse transcriptase
system (Invitrogen, Cergy Pontoise, France) following manufacturer’s instructions.
Quantitative PCR assay for cDNA was carried out using TaqMan Universal PCR master
mix in a AB PRISM 7000 sequence detection system (Applied Biosystem, Courtaboeuf,
France). The fluorescent probe for rat glutathione S-transferase pi (GSTP) gene, probe
code: Rn00821792_g1, was used. Data were normalized against beta actin gene
expression, probe Rn00667869-m1 (Applied Biosystem). Signals were quantified using
a standard curve made by serial dilutions of a cDNA pool from all samples.
C.11 MICROARRAY ANALYSIS
C.11.1 DNA microarray slides
DNA microarray glass slides were made at Biochips Platform of Toulouse Genopôle
(Toulouse, France). Rat 70mer oligos from Operon_V3_rat oligo set (Operon
Biotechnologies, Cologne, Germany) were loaded at 300 pmol/well. Oligos were
62
resuspended in 20 µl 3× SSC buffer (final concentration 20 µM). Spotting was
performed as single spot per probe on Ultra GAPS glass slides (Corning, Fontainebleau,
France) using Genetix Qarray mini robot (TeleChem International Inc., Saint Marcel,
France). Atmosphere moisture was set at 50% and plates were cooled to 15 ºC to
prevent drying. Spotting design was as follows: 48 grids, each grid containing 24×25
spots. Each slide bears 28000 spots corresponding to 27004 transcripts and 22012 rat
genes.
C.11.2 Preparation of labeled cDNA
Total RNA (5 µg) was used for cDNA preparation. Reverse transcription was made
using ChipShotTM reverse transcriptase from Promega (Promega, Charbonnieres Les
Bains, France) using Cy3- and Cy5-dCTP from Amersham Bioscience. cDNA
purification was made by the Promega ChipShotTM labeling Clean-up system. Dye
incorporation was checked by measuring the absorbance at 260 and 550 nm for Cy3 and
260 and 650 nm for Cy5 in a NanoDrop spectrophotometer (Nyxor Biotech).
C.11.3 Array hybridization and scanning
Equal amounts of Cy3- and Cy5-labeled cDNA from control and tested groups were
hybridized for 8 hours at 42ºC in the automatic Hybridization Machine Discovery from
Ventana (Ventana Medical Systems, Illkirch, France). After hybridization, slides were
washed twice in 2× SCC (Ribowash buffer, Ventana Medical Systems) for 5 minutes
with agitation and then once with 0.1× SSC buffer. After washing, slides were dried by
centrifugation. Slides were scanned with Axon 4000A Microarray Scanner (Molecular
Devices, Saint-Grégoire, France).
63
C.11.4 Data analysis and public access
All data reported here are available to public, via NCBI-GEO platform, as both original
and normalized data. Image analysis was made by Genepix Pro V6 software. Each spot
was evaluated and selected to be analyzed by Bioplot software (http://biopuce.insa-
toulouse.fr). Spot analysis consisted in background subtraction with Lowess allowed
value 10, log transformation and Lowess normalization. The generated data set has been
submitted to the National Center for Biotechnology Information (NCBI) Gene
Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/projects/geo/) database
(GSE7654).
Data analysis consisted of a variance analysis. The whole data set (27004 gene probes)
was subjected to an ANOVA using SPLUS 2000 software (MathSoft Inc., Seattle, WA).
Significance was set as P-value < 0.05. Partial Least Square (PLS)-2, PLS-Discriminant
Analysis (PLS-DA) and orthogonal signal correction (OSC) were made using SIMCA-P
8.0 software (Umetrics, Umea, Sweden). Oligos that contribute to axis building for both
PLS-2 and PLS-DA analyses were defined by the Variable Importance Projection (VIP)
criterion given by the SIMCA-P 8.0 software.
For biological organization of data, gene ontology (GO) classification was performed
by Gene Ontology Tree Machine from Vanderbilt University
(http://bioinfo.vanderbilt.edu/gotm) (Zhang et al., 2004).
C.12. METABONOMIC ANALYSES
C.12.1 Samples preparation for 1H NMR spectroscopy
Urine samples for NMR spectroscopy were prepared by mixing 500 µl of urine with
200 µl of phosphate buffer in D2O (pH 7.4) containing 10 mM deuterated
trimethylsilylpropionate as chemical shift reference (δ 0.0). The buffered urine samples
64
were then centrifuged at 10000 rpm for 10 min to remove any precipitates, and aliquots
of the resulting supernatant (600 µl) were placed in 5 mm NMR tubes.
Samples of liver tissue (~100 mg) were homogenized in 2 ml of 50/50 acetonitrile/H2O
containing 0.1% of BHT in an ice/water bath. The homogenates were centrifuged at
5000 g for 10 min at 4°C. The supernatants were removed and freeze dried before being
reconstituted in 1 ml D2O. 2 ml of 75/25 chloroform/methanol was added to the pellets
and extraction was followed by a subsequent centrifugation (5000 g for 15 min at 4°C).
The lipophilic supernatants were removed, dried under a stream of nitrogen and
reconstituted in 600 ml of CDCl3. The reconstitued solutions were transferred to 5 mm
(o.d.) NMR tubes.
C.12.2 NMR spectroscopic analysis
All 1H NMR spectra were obtained on a Bruker DRX-600 Avance NMR spectrometer
(Rheinstetten, Germany) operating at 600.13 MHz for 1H resonance frequency using an
inverse detection 5 mm 1H-13C-15N cryoprobe attached to a CryoPlatform (the
preamplifier cooling unit).
1H experiments of urine samples were recorded using the « Improved Watergate »
sequence for the suppression of the water resonance, accumulating 128 FIDs into 32 K
data points on a 12 ppm spectral width. All free induction decays were multiplied by an
exponential function equivalent to a 0.3 Hz line-broadening factor before Fourier
transformation.
The 1H NMR spectra of liver extracts were acquired at 300 K using a 1D single pulse,
accumulating 128 FIDs into 32 K data points on a 12 ppm spectral width. All free
induction decays were multiplied by an exponential function equivalent to a 0.3 Hz line-
broadening factor before Fourier transformation.
65
600 MHz 1H-1H correlation spectroscopy (COSY) NMR was performed on two aqueous
extracts samples to confirm the 1H NMR assignments. Transients were acquired into
1024 data points with 32 scans per increment, a spectral width of 7183 Hz, and 256
increments in the F1 axis, which was zero-filled to 512 prior to Fourier Transformation.
The relaxation delay between successive pulse cycles was 1.5 s. The data sets were
weighted using a sine-bell function in t1 and t2 prior to Fourier Transformation.
C.12.3 Spectra analysis
All NMR spectra were data-reduced using AMIX (version 3.1, Bruker Analytik,
Rheinstetten, Germany) to integrated regions 0.04 ppm wide corresponding to the
region δ 10.0 to 0.5. The regions δ 6.5-4.5 in the urine spectra and δ 5.0-4.7 in the
aqueous extracts were set to zero integral value to remove the variability of water
resonance suppression and cross-relaxation effects on the urea signal. For the lipophilic
extracts, the regions δ 7.5-6.8 (chloroform) and δ 3.6-3.3 (methanol) were removed
from all spectra.
Each integrated region (bucket) was considered as a variable for multivariate statistical
analyses.
C.13 STATISTICAL ANALYSES
Multivariate statistical analyses are described above. Univariate statistical analysis for
data from qRT-PCR, proteasome activities and immunoblotting are expressed as means
and standard deviations. Statistical differences were determined by ANOVA, followed
by Bonferroni’s post-hoc test.
66
67
D. RESULTS AND DISCUSION
68
69
D.1 CHAPTER 1:
CLINICAL AND BIOCHEMICAL CHARACTERISATION OF LEC RAT
MODEL
D.1.1 INTRODUCTION AND AIMS
D.1.2 EXPERIMENTAL PROTOCOL
D.1.3 RESULTS
D.1.3.1 HEPATITIS DEVELOPMENT IN LEC RATS
D.1.3.2 GGT AND GSTP EXPRESSION IN HEPATOCYTES OF LEC RATS
D.1.3.3 OXIDATIVE STRESS INDUCTION IN LEC RATS
D.1.3.4 EFFECT OF D-PENICILLAMINE ON HEPATITIS DEVELOPMENT
D.1.3.5 COMPARISON BETWEEN D-PENICILLAMINE-TREATED RATS
AND NORMAL RATS
D.1.4 DISCUSION
70
71
D.1.1 INTRODUCTION AND AIMS
In order to achieve a better understanding of the role of oxidative stress in cancer,
suitable models in which direct association between both phenomena could be
demonstrated are necessary. In this context, LEC rat model for spontaneous hepatitis
and hepatocarcinogenesis is of interest (Mori et al., 1994).
This first step in our work aimed to define the biochemical and histological features in
LEC rat model in order to achieve a good classification of rats that allowed us to
describe the hepatitis process and at the same time to validate the group of D-
penicillamine treated rats as a control group for further studies in global approaches.
LEC rats are known to develop liver tumors. Carcinogenesis initiation stage in LEC rats
is thought to be at the onset of hepatitis, nevertheless, there is some controversial data
that suggest the acute hepatitis as a promoter stimulus for the cancer process. This
would mean that before the acute hepatitis, some already initiated cells might be present
and that during hepatitis those cells will be promoted with the hepatitis-regeneration
stimulus. In order to define the LEC rat cancer initation stage, we analyzed the tumor
marker expression in the liver of each LEC rat group.
D.1.2 EXPERIMENTAL PROTOCOL
D.1.2.1 Hepatitis determination:
Plasma ALT and AST activity determination by automated biochemistry
Liver structure analysis by H & E staining
D.1.2.2 Oxidative stress parameters
Determination of lipid peroxidation metabolites in urine
8-IsoPGF2α by EIA
DHN-MA by EIA
72
A. Hematoxylin and Eosin (H & E) staining of liver slides
B. Biochemical features and group classification
Figure 4. Histological and biochemical features of LEC rats at different hepatitis stages: Rats were classified according to their disease state, age and treatment. Values refer to means (SD= Standard Deviation). Means without a common letter differ as determined by ANOVA and the Bonferroni post-hoc test
Normal 6w Normal 9w Slight jaundice Jaundice F3,20 P
Age (weeks) 6 9 11-13 11-13 - -
ALT (U/l) 62.83(SD 13.7)a 66.7 (SD 7.7)a 584.8 (SD 120.4)b 1184 (SD 344.7)c 50.94 < 0.0001
AST (U/l) 123.8 (SD 15.3)a 95.5 (SD 7.4)a 479 (SD 144.7)b 725.7 (SD 208)b 33.94 < 0.0001
T-bilirubin (µmol/l)
2.1 (SD 1.8)a 1.9 (SD 0.8)a 107 (SD 64.7)a 369.7 (SD 228)b 12.87 < 0.0001
Liver histology Normal Normal architecture,
Slight hydropic change.
Abnormal. Conserved liver architecture; cholestasis;
presence of micronecrosis and apoptosis; inflammatory
infiltrates.
Abnormal. Necrosis, apoptosis, inflammation,
cholestasis, nuclear polymorphism, fat droplets.
- -
73
Determination of HNE-modified proteins by immunohistochemistry using the
monoclonal antibody anti HNE-His
D.1.2.3 Tumor marker expression
Determination of GGT activity in liver sections
Determination of GSTP protein by immunohistochemistry
D.1.3 RESULTS
D.1.3.1 Hepatitits development in LEC rats.
LEC rats were sacrified at different hepatitis stages, from normal to strong hepatitis
stage. This rat strain suffers of a period of acute hepatitis between 11-13 weeks with
symptoms of jaundice. Clinical symptoms of hepatitis included decrease in body
weight, yellowish skin on ears, tail and feet. During this period, activities of plasma
enzymes, alanine aminotransferase (ALT), aspartate aminotransferase (AST), as well as
total bilirubin levels, were significantly increased (Figure 4B). Contrary to “normal”
groups in which liver architecture was conserved (Figure 4), liver structure was strongly
altered during hepatitis periods with presence of cholestasis, apoptosis and
inflammation for the slight jaundice group and strong inflammation, hepatocytes
nuclear polymorphism and flat droplets (steatosis) in jaundice group.
74
Figure 5. Tumor marker expression in liver of LEC rats at different hepatitis stages: Liver sections were used to detect gamma glutamyltranspeptidase (GGT) activity by histochemistry (HC) or glutathione S-transferase, placental type (GSTP) by immunohistochemistry (IH) at different hepatitis stages.
75
D.1.3.2 GGT and GSTP expression in hepatocytes of LEC rats
Glutathione S-transferase (GSTP) and γ-glutamyltranspeptidase (GGT) are two well
known tumor markers in liver. In order to know in which moment cancer initiation stage
takes place in LEC rats, we analyzed those tumor marker expression in the liver of LEC
rats in each hepatitis period. Figure 5 shows histological determination of GGT activity
and GSTP protein presence in liver sections of LEC rats at different hepatitis stages.
GGT showed a time-dependent increase, which correlated to the state of disease. GGT
expression in Normal 6w and Normal 9w rat groups was confined to cells around bile
caniliculi. As hepatitis developed, GGT activity was detected in almost all hepatocytes
with no specific topographical distribution.
The evaluation of GSTP protein in hepatocytes by immunohistochemistry showed that
GSTP positive hepatocytes appear at the beginning of hepatitis process. Interestingly,
GSTP was present not only in the cytoplasm but also in the nuclei of hepatocytes in
Slight jaundice group. This GSTP presence in nucleus is transitory, since in Jaundice
group, GSTP is present only in the cytoplasm of hepatocytes groups presumed as
“initiated” cells. These results indicated that the cancer initiation stage takes place just
before the acute hepatitis.
D.1.3.3 Oxidative stress induction in LEC rats
Copper-associated liver injury is regarded as resulting ultimately from oxidative stress.
The majority of the OH· in vivo comes from the Fenton reaction Mn+ + H2O2 � M(n+1)+
+ OH· + OH-. The formation of OH· has been demonstrated to initiate the oxidative
degradation of lipids known as lipid peroxidation. 4-hydroxy-2-nonenal (HNE) is one
of the major products during lipid peroxidation. HNE is a reactive molecule that can
76
Figure 6. HNE-protein adduction in LEC rat liver at different hepatitis stages:
Immunohistochemistry for HNE-histidine adducts in liver sections of LEC rats at different hepatitis stages: A. Normal 6w, B. Normal 9w, C. Slight jaundice, D. Jaundice.
interact with biomolecules as DNA and proteins. HNE-protein adducts were visualized
by immunohistochemistry (Figure 6). HNE-positive hepatocytes were evaluated in each
group of rats, rats from the group Normal 6w presented HNE positivity in hepatocytes
around central vein or in the stroma. Just before the hepatitis in Normal 9w group, HNE
positivity was diffuse, not intensive but in a lot of hepatocytes. Positivity was both
cytoplasmic and nuclear or perinuclear, and in some cases nuclear or perinuclear
positivity was stronger than the cytoplasmic one. In diseased rat groups HNE-modified
proteins were also present. In Slight jaundice group, HNE positivity is mostly
cytoplasmic and sporadically positive picnotic nuclei were present. During the acute
phase of hepatitis, in Jaundice group, HNE positivity was diffuse in hepatocyte
cytoplasm without any topographic distribution, some apoptotic cells were also positive.
The main urinary metabolite of HNE in rat is the 1,4-dihydroxynone-mercapturic acid
(DHN-MA) (Alary et al., 1995; Alary et al., 1998). DHN-MA was assessed as a good
77
marker of HNE formation under oxidative stress conditions (Peiro et al., 2005). We
have measured DHN-MA in the urine of rat at the day of sacrifice (Figure 7-B). As
expected, DHN-MA urinary excretion increased 3.6 times in the Slight jaundice group
compared to the Normal 6w group of rats. However, levels of DHN-MA drastically
diminished in Jaundice group of rats, even when HNE was constantly produced in liver
as seen in the HNE-his adducts immunohistochemistry. DHN-MA is formed by the
metabolism of HNE originated from both cellular oxidative stress and from exogenous
sources like diet. During the acute phase of hepatitis, rat food consumption drops down
(Figure 8), so HNE coming from diet might also decrease and consequently levels of
DHN-MA. Accordingly to this hypothesis, a positive correlation between DHN-MA
levels and diet consumption was found (Figure 9). These results showed that DHN-MA
Figure 7. Lipid peroxidation products’ excretion in urine of LEC rats at different
hepatitis stages: Urine samples were collected from LEC rats. 8-IsoPGF2α and DHN-MA were measured by competitive enzymatic immunoassay (EIA). Values refer to means and standard deviations. Means without a common letter differ as determined by ANOVA and the Bonferroni post-hoc test
78
Figure 8. Diet consumption at different stages of hepatitis in LEC rats. 24 hours diet consumption was measured the day of rat sacrifice. Values refer to MEANS and Standard Deviation. Means without a common letter differ as determined by ANOVA and the Bonferroni post-hoc test.
Figure 9. Effect of diet consumption on lipid peroxidation products urinary
excretion: Correlation between diet consumption and 8-IsoPGF2α or DHN-MA urine excretion in LEC rats.
79
levels are very sensitive to exogenous HNE and then it was not a suitable marker for
endogenous oxidative stress when food consumption is not controlled.
Isoprostanes are other products of lipid peroxidation. Among them, 8-IsoPGF2α has
been described as a good marker of oxidative stress and inflammation. 8-IsoPGF2α
determination in urine showed an increase as hepatitis developed (Figure 7A). Levels of
8-IsoPGF2α were 6.5 times higher in Jaundice group as compared to Normal 6w group.
In Figure 9 we show the relation between 8-IsoPGF2α, DHN-MA and diet consumption.
When rats bear a normal food consumption, corresponding to non-acute disease, levels
of 8-IsoPGF2α were low, while as rats stopped eating, levels of 8-IsoPGF2α were high
because of hepatic inflammation.The contrary is seen with DHN- MA excretion because
of the decrease of food originating HNE consumption. These observations allowed us to
consider 8-IsoPGF2α determination as a better marker of lipid peroxidation in situations
of acute endogenous oxidative stress processes as in the case of LEC rats.
D.1.3.4 Effect of D-penicillamine on hepatitis development
Figure 10 shows the effect of D-penicillamine administration on hepatitis development
in LEC rats. As indicated by hepatitis markers on plasma, D-penicillamine stopped
hepatitis development in LEC rats (Figure 10B). D-penicillamine treatment lead to rats
with normal liver architecture, low levels of GGT activity and no presence of GSTP in
hepatocytes (Figure 10A). HNE-modified proteins were only seen around portal veins,
and urinary levels of 8-IsoPGF2α were diminished by 7.9 times in D-penicillamine
group of rats compared to non treated rats suffering from hepatitis.
80
A. Histological comparisons between un-treated jaundice LEC rats and D-penicillamine-treated rats
B. Effect of D-penicillamine on plasma hepatitis markers and urinary 8-IsoPGF2α
Group Jaundice D-penicillamine P Age (weeks) 11/13 13 - ALT (U/l) 1184 (SD 344.7) 45.7 (SD 2.7) 0.00047 AST (U/l) 725.7 (SD 208) 81.8 (SD 2.4) 0.00063
T-bilirrubin (µmol/l) 369.7 (SD 228) 2.3 (SD 0.6) 0.01092 Liver histology Abnormal. Necrosis, apoptosis,
inflammation, cholestasis, nuclear polymorphism, fat droplets
Normal. Slight hydropic change -
8-IsoPGF2α (ng/24h) 50.98 (SD 21.24) 6.43 (SD 1.84) 0.00050 Figure 10. Effect of D-penicillamine on hepatitis development, carcinogenesis initiation and oxidative stress in LEC rats: Values refer MEANS (SD = Standard Deviation), P = p value obtained by two-tailed student t test.
81
D.1.3.5 Comparison between D-penicillamine-treated rats and normal rats
In order to establish if D-penicillamine-treated LEC rats are a suitable control group for
further comparisons in global approaches (transcriptomics and metabonomics), we
compared these rats with the Normal 6w LEC group and a group of Long Evans rats
which are normal rats. Figure 11 shows the histological and biochemical comparison
between these groups. In all groups normal liver architecture, negative immunological
staining of GSTP and HNE-protein adducts were seen (Figure 11A). Plasma levels of
classical hepatitis markers were also low in all groups (Figure 11B). Together all these
results allowed us to choose D-penicillamine-treated rats as suitable control group in
further experiments since they behave as normal rats.
D.1.4 DISCUSSION
This first part of our study aimed to characterize hepatitis development in LEC rats on
the basis of observations of jaundice, biochemical parameters of hepatitis, and changes
in liver histology and to correlate the hepatitis process to oxidative stress and cancer
initiation processes. Since hepatitis was spontaneous, it differs from each rat and its
apparition was not synchronic, so rat classification was made according to different
hepatitis markers (hepatic enzymes and bilirubin in plasma and histological features),
rather than to a time concerning classification. We were able to classify LEC rats in 4
groups according to hepatitis stage: Normal 6w, Normal 9w, Slight jaundice and
Jaundice in which we observed significant differences between groups concerning
histological features, oxidative stress and tumor marker expression.
82
A. Histological comparisons between LEC rats 6w, D-penicillamine groups and Long
Evans
B. Plasma hepatitis markers
GROUP ALT (U/l) AST (U/L) T-BILIRUBIN (µMOL/L))
Normal 6w 62.83 (SD 13.7)a 123.8 (SD 15.3)a 2.1 (SD 1.8) LEC
D-penicillamine 45.7 (SD 2.7)b 84.8 (SD 2.4)b 2.3 (SD 0.6)
Long Evans 28.17 (SD 2.14)c 55.5 (SD 3.57)c 2.06 (SD 1.74) F2,15 22.74 56.28 0.2052 P < 0.0001 < 0.0001 0.8167
Figure 11. Histological and biochemical parameters comparison between LEC rats
groups (Normal 6w and D-penicillamine) and Long Evans rats. Values refer MEANS (SD = Standard Deviation). Means without a common letter differ as determined by ANOVA and the Bonferroni post-hoc test.
83
Kasai et al. have described the clinicopathological characteristics of the acute hepatitis
in LEC rats (Kasai et al., 1990). Similarly to their observations, we have showed that
the onset of the acute phase of hepatitis was characterized by an increase of apoptosis,
large nuclei and hepatocytes in mitosis, and that in the fulminant hepatitis period, liver
necrosis and inflammation were found together with an elevation of plasma enzyme
activities like ALT, AST and symptoms of jaundice.
LEC rats have been described to spontaneously develop liver tumors. Tumor formation
in this rat strain follows a natural history similar to those seen in chemical
carcinogenesis in which initiation, promotion and progression stages are followed by
the differential expression of proteins, known as tumor markers, in preneoplastic and
neoplastic lesions (Sawaki et al., 1990). Two of these known tumor markers are GGT
and GSTP. Our results showed that both proteins are expressed in hepatocytes at early
stages of hepatitis. GGT is an ectoenzyme that is expressed in fetal liver and in bile
canaliculi cells; in the adult rat liver, GGT is normally expressed only in the bile
canaliculi cells an in the canaliculi membrane face of hepatocytes. GGT activity was
detected in higher levels in Normal 6w rats than in Normal 9w group, even when
topographic expression of GGT was normal in both groups. This transitional increase of
GGT expression was first described by Jain et al. (Jain et al., 1992). They showed an
increase of GGT mRNA in 4 weeks aged LEC rats that was then diminished in 8 weeks
aged LEC rats. Moreover, GGT expression was higher in LEC rats than in Long Evans
Agouti (LEA) rats. The overexpression of GGT in hepatocytes has been associated with
stress conditions, such as oxidative stress (Yamada et al., 2006) and hepatitis (Paolicchi
et al., 2005), so the increase seen in the liver of Slight jaundice and Jaundice rat groups
might correspond to the liver disease state, and because of no specific topographical
84
distribution of GGT activity, it is not possible to associate GGT expression to the cancer
initiation stage.
GSTP is considered as one of the most reliable tumor markers in liver (Aliya et al.,
2003). GSTP is normally not expressed in adult rat hepatocytes and it is expressed in
almost 100% of HCC cells independently of the initiation-promotion stimulus. GSTP
expression in hepatocytes at early stages of hepatitis enforces the idea of initiation of
carcinogenesis at these stages. GSTs are mainly expressed in the cytoplasm, and their
involvement in xenobiotic biotransformation, drug metabolism and thiol homeostasis
has been well documented (Seidegard et al., 1997; Tew, 2007). GSTP is nowadays
considered as important in tumorigenesis and in cancer drug resistance (Tew, 2007).
GSTP has been shown to be induced as a cellular response to HNE and its expression
and activity is closely related to oxidative stress modulation (Fukuda et al., 1997). The
fact that we have found GSTP expression in nuclei of hepatocytes in early stages of
hepatitis (Slight jaundice group) remarks the importance of oxidative stress in cancer
initiation, since nuclear localization of GSTP has been associated with prevention of
apoptosis during oxidative stress conditions by conjugating lipid peroxidation products
as 4-oxo-2-nonenal with GSH and by consequence diminishing the formation of
exocyclic DNA products (Kamada et al., 2004). This suggests the induction of GSTP as
an adaptive response to oxidative stress. Further investigations may be done in order to
clarify the importance of this transitional expression of GSTP in hepatocyte nucleus and
its relation with liver cancer initiation.
Oxidative stress was clearly related to hepatitis development as demonstrated both by
increased levels of 8-IsoPG2α in urine and, directly in liver, by formation of HNE-
protein adducts. HNE is considered as the major lipid peroxidation product that has
been shown to have not only deleterious effects but also an important role in
85
physiological processes as a second messenger of oxidative stress. Its presence is used
as a marker of oxidative stress in different tissues and cells during pathophysiological
processes (Zarkovic, 2003). HNE-modified proteins were visualized from young LEC
rats without signs of hepatitis (Normal 9w rat group). Modification of proteins by HNE
may be implicated in oxidative stress-mediated carcinogenesis, since it has been
demonstrated that HNE can inhibit some DNA damage repairing enzymes contributing
to DNA mutations and genome instability (Feng et al., 2004). HNE positivity was not
only in cytoplasm but also around the nuclei. This is the first report showing HNE-
modified proteins around hepatocyte nucleus. This interesting finding opens the door to
new insights of the HNE role in cellular homeostasis and carcinogenesis beyond to the
already known DNA-HNE adducts.
One of the aims of this first step of our work was to define a suitable control group of
rats that would be compared as normal rats in further experiments. LEA rat strain,
which is an Atp7b non mutant Long Evans deriving strain, is often used as control in
LEC rat studies, both rat strains deriving from a common ancestor (Long Evans rats).
However, LEA rats may have slightly evolved differently from LEC rats as indicated by
differences in epoxide hydrolase, GGT and UDP-glucoronyltransferase enzyme
activities and in cytochrome P-450 content between LE and LEA rats (Sugiyama et al.,
1988). Here we have used D-penicillamine in order to inhibit hepatitis development.
Histological and clinical features of these rats showed that they behave as normal rats,
according to the known effect of D-penicillamine to inhibit hepatitis and
hepatocarcinogenesis in LEC rats (Togashi et al., 1992; Klein et al., 2000).
Furthermore, we compared the histological features and the plasma levels of ALT, AST
and t-bilirrubin of D-penicillamine-treated LEC rats to those of untreated 6 weeks old
86
LEC rats and to 13 weeks old Long Evans rats. D-penicillamine-treated LEC rats
behaved as normal rats with normal liver histology and negative staining for GSTP and
HNE-protein adducts, as well as low levels of hepatitis markers in plasma. Even when
we found significantly difference between LE and D-penicillamine-treated rats in ALT
and AST levels, D-penicillamine-treated LEC rats have the same genetic background as
untreated LEC rats, they were considered as a suitable control for studying the
mechanisms involved in hepatitis and cancer initiation in LEC rats by transcriptomics
and metabonomics analyses.
As a conclusion, in this first stage of our work we defined the rat groups that were then
used in transcriptomic and metabonomic analyses. Four groups of untreated LEC rats
were defined: Two non-diseased groups (Normal 6w and Normal 9w) and two groups of
diseased animals showing different hepatitis stages (Slight jaundice and Jaundice).
Together these groups of rats allow us to study the hepatitis development and the cancer
initiation stage in LEC rats, in comparison to a control group formed of D-
penicillamine-treated rats that were shown to behave as normal rats with no hepatitis
symptoms and no oxidative stress induction.
87
D.2 CHAPTER 2:
TRANSCRIPTOMIC ANALYSIS OF LEC RAT HEPATITIS DEVELOPMENT
D.2.1 INTRODUCTION AND AIMS
D.2.2 EXPERIMENTAL PROTOCOL
D.2.3 RESULTS
D.2.3.1 Microarray analysis and validation
D.2.3.2 Functional classification of VIP genes
D.2.3.3 Proteasome activity
D.2.3.4 Effect of D-penicillamine on gene expression
D.2.4 DISCUSSION
88
89
D.2.1 INTRODUCTION AND AIMS
Nowadays DNA-microarray technology is a powerful tool to determine the repertoire of
genes that are differentially expressed in cells or tissues exposed to different stimuli like
xenobiotic exposure or disease development. Microarray studies have produced detailed
pictures of gene expression in a sample, information that is not apparent in conventional
studies. The data sets are large and encompass new described genes. The problem is to
find relationships or discriminate which expression changes are important in such an
important mass of data. In these regard, multivariate statistical methods are of interest.
Biological information from transcriptomic data sets can be captured by correlating
them with other significant biological information using dedicated statistical methods.
In this part of our work, we have used Partial Least Squares (PLS) regression and PLS-
Discriminant Analysis (PLS-DA) to select genes that correlated to hepatitis symptoms
and that discriminated LEC rats at different hepatitis stages, respectively.
D.2.2 EXPERIMENTAL PROTOCOL
D.2.2.1 cDNA microarray analysis
D.2.2.1a RNA extraction
D.2.2.1b Preparation of labeled cDNA
D.2.2.1c Hybridization array: In Table I the hybridization design is shown. In
order to diminish the dye bias, caused by unequal amount of dye incorporation, the
array hybridization followed the dye swap design. Within the same group, three
independent samples were labeled with Cy3-dCTP and the other three were labeled with
Cy5-dCTP.
90
Comparison
Group TEST CONTROL Image file name Chosen chanel
Ra0502 Ra0703 05-02Cy3-07-03Cy5 PMT550.grp green
Normal Ra0602 Ra0803 06-02Cy5-08-03Cy3 PMT550.grp red
6w Ra0702 Ra0903 07-02Cy3-09.03Cy3 PMT550.grp green
Ra2906 Ra1003 29-06Cy5-10-03Cy3 PMT600.grp red
Ra3006 Ra1103 30-06Cy3-11-03Cy5 PMT550.grp green
Ra3106 Ra1203 31-06Cy5-12-03Cy3 PMT550.grp red
Ra0101 Ra1203 01-01Cy3-12-03Cy5 PMT520.grp green
Normal Ra0201 Ra1103 02-01Cy5-11-03Cy3 PMT520.grp red
9w Ra2406 Ra1003 24-06Cy3-10-03Cy5 PMT520.grp green
Ra2506 Ra0903 25-06Cy5-09-03Cy3 PMT550.grp red
Ra3207 Ra0803 32-07Cy3-08-03Cy5 PMT520.grp green
Ra3407 Ra0703 34-07Cy5-07-03Cy3 PMT520.grp red
Ra0401 Ra0703 04Cy5 07Cy3 680b.grp red
Slight Ra0601 Ra0903 06Cy5 09Cy3 700b.grp red
Jaundice Ra1603 Ra0803 16-03Cy3-08-03Cy5 PMT500.grp green
Ra2106 Ra1003 21-06Cy3-10-03Cy5 PMT500.grp green
Ra2306 Ra1103 23-06Cy5-11-03Cy3 PMT500.grp red
Ra1303 Ra1203 13-03Cy3-12-03Cy5 PMT500.grp green
Ra0501 Ra0803 08Cy5 05Cy3 650.grp red
Jaundice Ra1503 Ra1103 15Cy5 11Cy3 650.grp red
Ra1403 Ra1003 10Cy5 14Cy3 700.grp green
Ra1703 Ra1203 12Cy5 17Cy3 650.grp green
Ra2606 Ra0903 26-06Cy3-09-03Cy5 PMT520.grp green
Ra2706 Ra0703 27-06Cy5-07-03Cy3 PMT520.grp red
Ra0105 Ra0703 01-05Cy3-07-03Cy5 PMT500.grp green
Long Evans Ra0205 Ra0803 02-05Cy5-08-03Cy3 PMT520.grp red
Ra0305 Ra0903 03-05Cy3-09-03Cy5 PMT520.grp green
Ra0405 Ra1003 04-05Cy5-10-03Cy3 PMT500.grp red
Ra0505 Ra1103 05-05Cy3-11-03Cy5 PMT500.grp green
Ra0605 Ra1203 06-05Cy5-12-03Cy3 PMT550.grp red
Ra0705 Ra1203 07-05Cy3-12-03Cy5 PMT520.grp green
Long Evans Ra0805 Ra1103 08-05Cy5-11-03Cy3 PMT520.grp red
D- Ra0905 Ra1003 09-05Cy3-10-03Cy5 PMT520.grp green
penicillamine Ra1005 Ra0903 10-05Cy5-09-03Cy3 PMT520.grp red
Ra1105 Ra0803 11-05Cy3-08-03Cy5 PMT520.grp green
Ra1205 Ra0703 12-05Cy5-07-03Cy3 PMT520.grp red
Table II. Array hybridization design: Control group refers to D-penicillamine-treated LEC rats. Image file names are shown as they are designed in Bioplot software. The channel color was chosen according to test labeling. All samples were compared against the same control group.
D.2.2.2 Quantitative RT-PCR of GSTP gene
D.2.2.3 Proteasome peptidase activity determination in cytosol fractions
D.2.2.4 Proteasome protein determination by western blotting
91
D.2.3 RESULTS
D.2.3.1 Microarray analysis and validation
The rat pangenomic microarray used in this study was made of 27004 oligo probes that
corresponded to 22012 rat genes. After scanning, images were analyzed by the Bioplot
software; Lowess-normalized data set with test/control (untreated/D-penicillamine
treated rats) ratios were obtained. In order to evaluate the difference between test groups
and the control group, means of ratio were calculated and after a student t-test, genes
were considered as differential genes when they showed a P < 0.05. In Figure 12 the
scatter plots of each comparison are shown. Each spot is represented as log10 of mean
ratio ± Standard Deviation as a function of the total intensity (Cy3 + Cy5). So the points
falling at the right corresponded to those genes with the higher expression in the
samples (high levels of mRNA) while those spots falling in the left were genes which
expression is more reduced. 1673 differential genes were found in Normal 6w/D-
penicillamine comparison, among them 880 genes were upregulated (log10 ratio > 0, red
spots) and 783 were downregulated (log10 ratio < 0, green spots) (Figure 12A). The
number of differential genes in the Normal 9w/D-penicillamine comparison was
diminished (777 differential genes), furthermore, the ratios of these differential genes
were small ratios, showing that the gene expression profiles in liver samples from
Normal 9w and D-penicillamine groups are quite similar (Figure 12B). As rats became
diseased, differences between test and control groups were stronger. 1702 differential
genes were founded in Slight jaundice/D-penicillamine comparison, among which 982
were upregulated and 720 downregulated. In the Jaundice/D-penicillamine comparison,
1817 differential genes were founded, 1035 upregulated and 782 downregulated (Figure
12C and D).
92
Figure 12. Ratio/Intensity scatter plots: In black: non-differential genes (P > 0.05); in red: differential upregulated genes (P < 0.05 and log10 ratio > 0). In green differential downregulated genes (P < 0.05 and log10 ratio < 0). A. Normal 6w vs D-penicillamine comparison. B. Normal 9w vs D-penicillamine comparison. C. Slight jaundice vs D-penicillamine comparison. D. Jaundice vs D-penicillamine comparison. P values were calculated by student-t test.
In order to describe in a global context the gene expression changes during hepatitis
development, multivariate statistical analyses were performed. Normalized data sets
were filtered by ANOVA test when considering the group factor. From 27004 oligo
probes, 696 probes with a P value < 0.05 were considered as significant variables.
Probes were then subjected to an orthogonal signal correction (OSC) filtering in order to
remove strong structured variation in X matrix (transcriptomic data) that is not
correlated to Y matrix (class group) (Trygg, 2002). 89.88% of the information remained
after this correction. In order to discriminate between genes related to hepatitis
progression and those mainly explained by a age effect, a PLS-2 regression between the
696 statistically significant genes and plasma hepatitis markers ALT, AST and t-bil was
performed (Figure 13A). The goal of this method is to analyze or predict a set of
93
dependent variables (matrix Y, i.e. plasma hepatitis markers) from a
Figure 13. Relation between hepatitis progression and gene expression. (A) 696 statistically significant variables from transcriptomic data (P < 0.05) were used for a PLS-2 regression against plasma hepatitis markers, ALT, AST and t-bil. (B) PLS-DA of transcriptomic data. Circles: rats within the same group. N6w: Normal 6w. N9w: Normal 9w. Sl-j: Slight jaundice. J: Jaundice.
set of independent variables (matrix X, i.e. microarray data) or predictors (Abdi, 2003).
PLS-2 regression showed a projection in which the Slight Jaundice and Jaundice groups
were well separated from the non diseased groups, Normal 6w and Normal 9w.
However, these two latter groups could not be separated between each other using this
analysis. The 696 genes were hence analyzed by a PLS-DA in order to refine the
separation between all groups of rats (Figure 13B). PLS-DA is a supervised
multidimensional statistical method that uses Principal Component Analysis (PCA)
94
principles, i.e. data projection into different dimensions, to find the components which
account for differences between groups. As shown in Figure 13B, the projection on the
first axis [t1] separated the non diseased rats (Normal 6w and Normal 9w groups) from
diseased rats, and even more, this mode of analysis was able to improve the separation
between groups of rats exhibiting a slight jaundice from those with a stronger disease.
The separation between groups was validated using another discriminant analysis, the
factorial discriminant analysis (FDA) (Figure 14), in which using only 10 variables
selected from the first 696 ones by a specific algorithm (Dumas et al., 2002), ∆2
Mahalanobis distances between all different groups were significantly different from
zero (P < 0.001) (Table III).
The oligo probes that contribute to axis building were defined using the Variable
Importance Projection (VIP) criterion given by the SIMCA-P 8.0 software. Statistical
VIP criterion is a measure which quantifies the influence on the response (on the latent
variable describing the projection of hepatitis development) of each variable summed
over all components and categorical responses relative to the total sum of squares of the
model (Perez-Enciso et al., 2003). It is a useful tool to select specific sets of genes that
distinguishes animals at different stages of hepatitis, making it an important parameter
to find possible pathways involved in disease development. From the 696 tested genes,
483 were considered as important for axis projection (VIP > 0.8) for the PLS-2 analysis
and 239 for de PLS-DA. 50% of the 239 genes found to be important for the PLS-DA
were also found to be important in the PLS-2 analysis, demonstrating that the separation
of groups in the PLS-DA is mainly due to hepatitis development. Table A
(supplementary data) provides the list of these 239 genes ordered by decreasing
importance of the VIP value.
95
Figure 14. Factorial Discriminant Analysis (FDA) of transcriptomic data: 10 variables from 692 significative variables (P < 0.05) were selected as important for axes building. The first axis (LD1) explains the 66.9% of the variance while the second axis explains the 19.3% of the variance.
Table III. ∆
2 Mahalanobis distances between all different groups represented in FDA
test. The probability of these distances to be null is represented.
GSTP was found among the list of genes that significantly discriminated groups. We
confirmed the expression changes of this gene by the qRT-PCR method using β-actin as
96
the reference gene (F3,23 = 54.82, P = 0.0104) (Figure 15). Therefore, this data together
with the fact that this gene is considered as a tumor marker known to be expressed in
LEC rats during hepatitis (Masuda et al., 1989) consolidated the validity of our
transcriptomic analysis of LEC rats during hepatitis development.
Figure 15. Confirmation of the differential expression of GSTP by quantitative RT-
PCR. Triangles: GSTP expression units from microarray data. Open circles: GSTP expression units from TaqMan QRTPCR assay. Data is expressed as ratio Test/Control ratio (means ± SD) where control is D-penicillamine-treated rat group. Means without a common letter differ as determined by a one-way ANOVA and the Bonferroni post-hoc test.
D.2.3.2 Functional classification of VIP genes
To gain more biological insights of our expression data, the list of 239 genes with a
significant VIP value (VIP > 0.8) were subjected to a functional classification based on
Gene Ontology annotation (GO). Only 40% of the total rat genome has a gene ontology
annotation (Lomax, 2005). This explained that only 59 genes had a GO annotation in
Biological Process ontology from the 239 selected genes (Figure 16). From these GO
97
Figure 16. Gene ontology classification for 59 known genes. Classification by biological process was made (1-10). Chart data are read as: Category name (Accession): number of tested genes annotated in this category, percent of tested genes annotated in this category against total tested genes having a GO annotation.
98
Figure 17. Metabolism ontology sub-classification. Genes being part of metabolism classification were sub-classified according to their specific activity (1-25). Chart data are read as: Category name (Accession): number of tested genes annotated in this category, percent of tested genes annotated in this category against total tested genes having a GO annotation.
99
TABLE IV. Differential genes classified in GO: Protein Metabolism Description Symbol Oligo_id ENSEMBLE Gene_id Genbank Ratio p value
Normal 6w
Normal 9w
Slight jaundice
Jaundice
Cathepsin D precursor (EC 3.4.23.5)
Cts. Rn30018601 ENSRNOG00000020206 NM_134334 1.14a 1.09a 1.86a,b 2.44b 0.0207
Glycine N-methyltransferase (EC 2.1.1.20)
Gnmt Rn30014947 ENSRNOG00000016349 X06150 0.73a 0.85a,b 0.55b 0.41b 0.0050
C-reactive protein precursor Crp R002545_01 ENSRNOG00000000053 DW387982 0.68a 0.89a,b 0.45b 0.35b 0.0344 H-2 class II histocompatibility antigen, gamma chain
Cd74 Rn30017236 ENSRNOG00000018735 CB576583 1.06a 1.15a 1.73b 2.36c 0.0125
Meprin A alpha-subunit precursor (EC 3.4.24.18)
Mep1a R004430_01 ENSRNOG00000011022 BC081834 0.97a 0.95a 0.61b 0.92a 0.0351
proteasome 26S non-ATPase subunit 12
Psmd12 Rn30002776 ENSRNOG00000003117 BC083758 1.10a,b 1.01a 1.05a,b 1.48b 0.0160
Calpain small subunit 1 (CSS1) Capns1 Rn30001369 ENSRNOG00000001503 BC098068 0.94a,b 0.98a 1.11a,b 1.39b 0.0408 Serum amyloid P-component precursor (SAP)
Apcs R000224_01 ENSRNOG00000009086 X55761 1.01a 0.84a,b 0.41b 0.54a,b 0.0071
Complement C1q subcomponent subunit A precursor
C1qa Rn30011801 ENSRNOG00000012807 BC086605 1.00a 0.95a 1.67a,b 1.91b 0.0439
TGFB inducible early growth response 3 (predicted)
Gna14 Rn30013625 ENSRNOG00000014840 BC090316 0.90a 0.96a 1.24b 0.94a 0.0465
Kit ligand precursor (C-kit ligand) Kilt R003524_01 ENSRNOG00000005386 NM_021843 1.09a 0.96a 0.90a 1.46b 0.0406 Legumain precursor (EC 3.4.22.34) Lgmn Rn30006415 ENSRNOG00000007089 BC087708 1.05a,b 0.93a,b 1.36a 1.57b 0.0169 Ubiquitin-conjugating enzyme E2 B (EC 6.3.2.19)
UBC2_HUMAN
Rn30004545 ENSRNOG00000005064 AF144083 1.01a,b 0.99a,b 1.17a 1.52b 0.019
Proteasome subunit beta type 8 precursor (EC 3.4.25.1)
Psmb8 R004525_01 ENSRNOG00000000456 NM_080767
1.11a 1.07a 1.16a 1.27b 0.0317
Proteasome subunit alpha type 6 (EC 3.4.25.1)
Psma6 Rn30006440 ENSRNOG00000007114 NM_017283 1.10a 1.01a 1.16a,b 1.44b 0.0132
Proteasome subunit beta type 5 precursor (EC 3.4.25.1)
Psmb5 Rn30012312 ENSRNOG00000013386 XM_341314 0.93a 1.05a 1.15a,b 1.43b 0.0213
proteasome 26S, non-ATPase regulatory subunit 6
Psmd6 Rn30006048 ENSRNOG00000006751 BC059159 1.03a 1.03a 0.89a 1.60b 0.0306
Aminopeptidase B (EC 3.4.11.6) Rnpep R002043_01 ENSRNOG00000006720 AY724503 0.86a 0.95a 0.62b 1.11a 0.0465
100
TABLE V. Differential genes classified in GO: Nucleobase, Nucleoside, Nucleotide and Nucleic acid Metabolism
Description Symbol Oligo_id ENSEMBLE Gene_id Genbank Ratio p value
Normal 6w
Normal 9w
Slight jaundice
Jaundice
Uricase (EC 1.7.3.3) Uox Rn30014939 ENSRNOG00000016339 M24396 0.69a 1.04a,b 0.57b 0.50b 0.0234 Growth arrest and DNA-damage-inducible protein GADD45 alpha
Gadd45a R001492_01 ENSRNOG00000005615 CO396252 1.00a 0.94a 1.06a,b 1.49b 0.0055
Signal transducer and activator of transcription 3
Sta3 Rn30018174 ENSRNOG00000019742 NM_012747 0.84a 0.86a 1.56b 1.72b 0.0498
Dynamin-2 (EC 3.6.5.5) Dnm2 Rn30006927 ENSRNOG00000007649 L25605 1.03a 1.06a 1.03a 1.58b 0.0151 Runt-related transcription factor 1, Acute myeloid leukemia 1 protein
Runx1 Rn30001533 ENSRNOG00000001704 L35271 0.93a 1.05a 1.27b 1.23b 0.0280
Cellular nucleic acid binding protein, Zinc finger protein 9
Cnbp1 R000749_01 ENSRNOG00000010239 CB743905 1.06a,b 1.09a 1.07a,b 1.45a,b 0.0298
DNA-repair protein XRCC1 Xrcc1 Rn30018333 ENSRNOG00000019915 AF290895 1.02a,b 0.98a 1.22a,b 1.54a,b 0.0338
101
annotated genes, 38 fall into Metabolism, 19 in Organismal Physiological Process and
19 in Regulation of Physiological Process categories. A sub-classification in Cellular
Metabolism category revealed that genes implicated in Protein Metabolism, and in
Nucleobase, Nucleoside, Nucleotide and Nucleic acid Metabolism were overrepresented
(Figure 17). Values of expression changes of these genes and their associated function
are described in Tables IV and V.
Figure 18. Proteasome structure: 26S proteasome is composed of two subcompexes; a 20S core particule (CP) in which catalytic activity is present; and two 19S (PA700) particules in which recognition of ubiquitin and protein defolding take place. D.2.3.3 Proteasome activity
The proteasome (Figure 18) is a large, 26S, multicatalytic protease that degrades
polyubiquitinated proteins to small peptides. It is composed of two subcomplexes: a 20S
core particle that carries the catalytic activity, and the regulatory 19S particle
(Ciechanover, 2005). The main function of the proteasome is to degrade superfluous
and/or damaged proteins by proteolysis. Our DNA microarray data revealed that several
genes in the Protein Metabolism sub-category whose expression was increased in LEC
rats during hepatitis are implicated in protein degradation by the proteasome (Table IV).
Furthermore, a heat map of proteasome-related genes showed a tendency of
102
upregulation of almost all genes that encode the 19S and 20S proteasome in Jaundice
group of rats (Figure 19). Among them, six genes were significantly increased. This
result suggested that the proteasome activity could be modified. To verify this
hypothesis, we determined the peptidase activities of 20S proteasome using specific
fluorogenic peptides as substrates n cytosolic fractions (Figure 20). Unexpectedly, there
was no significant modification of the chymotrypsin-like (ChT) proteasome activity
Figure 19. Heat map of proteasome-related genes in the liver of LEC rats at different
disease stages. Mean expression ratios of proteasome-related genes in each rat groups in respect to D-penicillamine-treated rats. Green and red represent downregulation and upregulation, respectively, as indicated in the “Ratio Scale” bar. (A) 20S proteasome-related genes. (B) 19S proteasome-related genes. * P < 0.05.
103
(Figure 20A) even when the gene encoding the proteasome subunit responsible for this
activity, the Pmsb5 gene, was significantly up-regulated (Table IV). Moreover, we
found a significant decrease in trypsin-like (TL) proteasome activity in Slight jaundice
and Jaundice groups of rats (F4, 25 = 58.62, P < 0.0001) (Figure 20B). Likewise, the
peptidylglutamyl peptide hydrolase (PGPH) proteasome activity was diminished by
about 30% in the Slight jaundice group (Figure 20C), but restored in the Jaundice rat
group (F4, 25 = 13.01, P = 0.0075). Both genes responsible for TL (Psmb2 gene) and
PGPH (Psmb1 gene) had a tendency of being overexpressed during hepatitis (Figure
19A).
In order to evaluate whether changes in proteasome activities were due to variations in
proteasome protein levels, we evaluated the 20S proteasome levels in cytosol by
Western blotting. As shown in Figure 21, 20S proteasome levels were increased by 1.5
and 1.8 fold, respectively in Slight jaundice and Jaundice groups with respect to control
D-penicillamine group (F4, 25 = 10.86, P = 0.015). Thus, a decrease in TL and PGPH
proteasome activities does not correspond to a decrease in 20S proteasome levels but
rather to an inhibition of the intrinsic protein activity.
D.2.3.4 Effect of D-penicillamine on gene expression
In order to define if the treatment with D-penicillamine changes gene expression, Long
Evans (LE) rats were submitted to a treatment with D-penicillamine in the same way as
D-penicillamine-treated LEC rats. Both groups, non treated LE rats (LE) and D-
penicillamine-treated LE rats (LE-D-penicillamine) were compared against D-
penicillamine-treated LEC rats. From the LE/D-penicillamine comparison 3213 genes
were considered as differential genes (P < 0.05), 1799 were upregulated (log10 ratio > 0)
and 1414 were downregulated (log10 ratio < 0) (Figure 22A). When
104
Figure 20. Changes in proteasome activities in the liver of LEC rats at different
disease stages. The proteasome activities in the liver of LEC rats in different groups were measured using the fluoropeptides, s-LLVY-AMC for the ChL activity (A), Boc-LSTR-AMC for the TL activity (B), and Z-LLE-βNA for the PGPH activity (C), as substrates. Each determination was performed at least in triplicate. Values refer to proteasome activity means ± SD. Means without a common letter differ as determined by a one-way ANOVA and the Bonferroni post-hoc test.
105
comparing LE rats treated with D-penicillamine against D-penicillamine LEC-rats only
413 differential genes were found, these genes might correspond to strain differences
(Figure 22B). Venn diagram allowed us to define those differential genes presumed to
be altered by D-penicillamine treatment. 3130 genes were modulated by D-
penicillamine (Figure 22C). Genes in VIP selection of the PLS-DA that might be altered
by D-penicillamine treatment are signaled in Table A (Annex 1).
Figure 21. Immunodetection of proteasome in liver of LEC rats at different disease stages and treatments. (A) Representative image of immunoblot analysis for 20S proteasome and alpha-tubulin proteins. (B) Densitometry analysis of immunoblot assay, values refer to proteasome/alpha-tubulin ratio means ± SD (n=6). Means without a common letter differ as determined by a one-way ANOVA and the Bonferroni post-hoc test.
106
Figure 22. Effect of D-penicillamine on gene expression patterns. A. Scatter plot of Long Evans/LEC D-penicillamine comparison. B. Scatter plot of D-penicillamine-treated Long Evans/LEC D-penicillamine comparison. C. Venn diagram of both comparisons, 3130 genes were assumed to be deregulated due to D-penicillamine treatment in Long Evans rats.
D.2.4 DISCUSSION
A first evaluation of gene expression in LEC rats was made by Klein et al. in 2003
(Klein et al., 2003). From their transcriptomic analysis, authors have only retained
genes with the highest expression changes, and from these data, concluded on a major
role of oxidative stress, inflammation and DNA damage in hepatitis progression.
However, this mode of data analysis is too restrictive and there is no obvious
relationship between the levels of gene expression and the intensity of the disease since
alterations in a given cellular activity are usually not the result of a single but often
depends on the coordinated effects of multiple genes (Valafar, 2002).
107
To optimize the exploration of large data sets such as transcriptomic data in a given
biological context, multivariate statistical methods followed by Gene Ontology
classification represent powerful tools of analysis. We considered D-penicillamine-
treated LEC rats as a suitable control group for transcriptomic analyses. We checked in
LE rats the effect of D-penicillamine in gene expression. Even when there are some
modifications in gene expression patterns due to D-penicillamine treatment, D-
penicillamine had no effect in the genes here discussed.
PLS-2 analysis between microarray and plasma hepatitis markers showed a projection
of group of animals regarding hepatitis, which means a strong correlation between gene
expression changes and the development of hepatitis in LEC rats. This method was
effective to separate between non-diseased groups from jaundice groups, but it was not
pertinent enough to discriminate the two non-diseased groups (Normal 6w and Normal
9w), probably because the number of hepatitis markers is much reduced and their values
are very similar between these groups. PLS-DA takes the advantage of prior class
information to attempt to maximize the separation between groups of observations
(Perez-Enciso et al., 2003). All groups were significantly separated from each other by
PLS-DA. Furthermore, more than 50% of VIP-positive genes from PLS-DA were
common to those coming from PLS-2 regression demonstrating that PLS-DA group
separation is also well correlated to animal hepatitis development.
Statistical VIP criterion serves as a powerful tool to select the variables that have a
strong impact on the building of the axes in PLS methods. A high VIP value implies a
strong influence of the variable in the axes building (Perez-Enciso et al., 2003). GSTP
presented one of the highest VIP value (VIP value = 3.902), suggesting that the increase
in expression of this gene was very important for separation between groups of rats in
the PLS-DA. This finding agrees with the fact that we had found GSTP in liver of LEC
108
rats during hepatitis. Furthermore, since GSTP is considered as one of the most reliable
tumor markers in liver (Sawaki et al., 1990), the idea of cancer initiation during
hepatitis in LEC rats is enforced.
During hepatitis progression in LEC rats, a strong oxidative stress takes place
(Yamamoto et al., 2001). Oxidation of biomolecules as DNA, lipids and proteins has
been extensively described (Yamamoto et al., 2001; Nair et al., 2005; Yasuda et al.,
2006). According to these observations, our analyses of microarray data show
enrichment of genes that belong to Protein Metabolism and Nucleobase, Nucleoside,
Nucleotide and Nucleic acid metabolism established by GO classification. Genes
involved in Nucleobase, Nucleoside, Nucleotide and Nucleic acid metabolism, and
genes involved in DNA reparation are up-regulated in Jaundice group. Growth Arrest
and DNA Damage 45-alpha (GADD45-alpha) is a nuclear protein involved in the
maintenance of genomic stability, DNA repair, and suppression of cell growth.
Overexpression of GADD45 gene has been reported in LEC rats at early stages of
hepatitis suggesting important DNA damage (Klein et al., 2003). The mechanisms by
which oxidized DNA bases are repaired include the base excision repair (BER)
pathway. M1G, etheno-DNA adducts, 8-hydroxyguanine and other oxidized base
lesions are removed by this pathway (Valko et al., 2006). Acute hepatitis in LEC rats
impairs the repair of oxidative DNA base damage by altering the activity of DNA
glycosylases, endonuclease II and 8-oxoguanine DNA-glycosylase, which initiate the
BER pathway (Choudhury et al., 2003). Interestingly, our results show overexpression
of DNA-repair protein XRCC1 gene. XRCC1 protein seems to play a central role in
coordinating the enzymes participating in BER pathway (Marsin et al., 2003).
Moreover, XRCC1 protein interacts with all enzymes of BER downstream of DNA-
glycosylase leading to the stabilization of the corresponding enzyme. Hence, up-
109
regulation of XRCC1 in LEC diseased groups may contribute to BER pathway recovery
in LEC rats. Further investigations are needed to determine the role of XRCC1 in BER
pathway in LEC rats.
Protein Metabolism was the one category with the highest amount of associated genes.
The results may indicate an increase in protein degradation, since genes falling in this
category are genes related to ubiquitin-proteasome system and to other peptidases.
Increase in protein degradation could reflect protein damage as protein oxidation (Grune
et al., 1995). According to our results showing overexpression of genes involved in both
19S and 20S proteasome system, our initial hypothesis was that proteasome activity
should be augmented in Slight jaundice and Jaundice groups. However, TL and PGPH
20S proteasome peptidase activities were diminished in these groups. Protein levels of
proteasome were significantly increased in Slight jaundice and Jaundice groups
compared to D-penicillamine group, suggesting that the diminution of proteasome
activities may not be due to less proteasome proteins in cell, but rather because of a
direct inhibition of protein activity.
If oxidized proteins are not eliminated through proteasome pathway, they are able to
accumulate and to aggregate (Grune et al., 2003b). Oxidized protein aggregates are very
difficult to eliminate because they can not be unfolded, an important step in protein
degradation, and they are able to cross-link with proteasome subunits enhancing
proteasome inhibition. HNE, a major end product of lipid peroxidation, modifies
proteins leading to the formation of such protein aggregates (Grune et al., 2003a).
Furthermore, HNE can inhibit directly the proteasome activity by covalent binding to
proteasome subunits. Okada et al. (1999) have found similar results for the inhibition of
TL and PGPH peptidase activities under oxidative stress conditions in kidney. Authors
concluded that HNE-proteasome adduction was partially responsible for the proteasome
110
inhibition under oxidative stress conditions (Okada et al., 1999). Although the detailed
mechanism of inactivation of proteasome with HNE is not fully understood, authors
suggest that the active site of the enzymes (TL and PGPH) may preferentially react with
HNE, leading inactivation. These findings lead us to propose, for the first time, the
inhibition of proteasome activities by oxidative stress during hepatitis progression in
LEC rats.
Inhibition of proteasome activities may induce the expression of proteasome genes in a
feedback mechanism that would compensate for the reduced protein degradation level.
Feedback regulation of proteasome genes have been described in Saccharomyces
cerevisiae. Proteasome gene expression was mediated by Rpn4 protein, which binds to
PACE (Proteasome associated control element) sequences that are found in the
promoters of proteasome subunits (London et al., 2004). Rpn4 protein is actually
degraded by the proteasome system, stabilization of Rpn4 by proteasome inhibition or
malfunction leads increased transcription of proteasome subunits by a negative feedback
circuit (Xie et al., 2001). In mammal cells, Meiners et al. (Meiners et al., 2003), have
studied the expression of proteasome genes after the inhibition of ChT proteasome
peptidase activity by specific proteasome inhibitors. They showed that there is a
concerted change in the expression of almost all the 20S proteasome subunits following
proteasome activity inhibition. Our results agreed with Meiners observations since even
when not all the genes were statistically significantly overexpressed, there was a clear
tendency for an upregulation of the majority of genes encoding for proteasome (Figure
5). The implication of a common transcription factor like Rpn4 in transcriptional
regulation of proteasome genes in mammal cells remains to be elucidated.
The proteasome is an essential organelle that participates in different cellular functions,
like apoptosis, survival and DNA reparation (Reed et al., 2007). Alterations in
111
proteasome activity have been related to metabolic perturbations, aging and different
pathologies including Parkinson and Alzheimer diseases (Hayashi et al., 1998;
Ciechanover, 2005; Reed et al., 2007). Oxidized and damaged protein accumulation
may alter cellular functions, giving rise to deleterious effects on cellular homeostasis
that could be involved in initiation and progression of carcinogenesis (Friguet, 2006;
Cecarini et al., 2007). Interestingly, proteasome inhibition, either by specific inhibitor or
by iRNA, induces the expression of GSTP gene in epithelial liver cells. Mechanisms of
GSTP induction by proteasome inhibition seem to be different to Nrf2/Keap1 pathway
(Usami et al., 2005). Further studies are necessary to know if there is an association
between proteasome inhibition and GSTP upregulation in LEC rats, however the idea of
proteasome inhibition in cancer initiation is proposed.
In conclusion, using the LEC rat model, our work presents for the first time a genome-
scale analysis of hepatitis development and cellular responses due to copper-induced
oxidative stress. In this study, multivariate statistical analysis coupled to a biological
interpretation achieved by GO classification of the transcriptomic data allowed us to
focus on proteasome system and on the consequences of proteasome inhibition as a
main deregulated pathway that might be a determining factor during oxidative stress-
induced hepatitis and further cancer initiation. Our work shows an original and
functional way of analysis that illustrates the power of multivariate statistical tools to
explore microarray data to raise rational biological hypothesis.
112
113
D.3 CHAPTER 3
METABONOMIC ANALYSIS AND ITS RELATION WITH
TRANSCRIPTOMIC DATA
D.3.1 INTRODUCTION AND AIMS
D.3.2 EXPERIMENTAL PROTOCOL
D.3.3 RESULTS
D.3.3.1 1H NMR OF URINARY SAMPLES OF LEC RATS BEFORE AND
DURING THE ONSET OF HEPATITIS
D.3.3.2 METABOLIC CHANGES IN LIVER OF LEC RATS AT DIFFEENT
HEPATITIS STAGES
D.3.3.2.1 Metabonomic analysis of lipid extracts of LEC rat liver samples
D.3.3.2.2 Metabonomic analysis of aqueous extracts of LEC rat liver samples
D.3.3.3 RELATIONSHIP BETWEEN METABONOMIC AND
TRANSCRIPTOMIC ANALYSES
D.3.4 DISCUSSION
114
115
D.3.1 INTRODUCTION AND AIMS
Transcriptomic approaches are very useful to make a first exploration of understanding
gene function and pathology. To fully use the information generated by these analyses,
it is necessary to have a description of the changing phenotype so that gene expression
can be further understood in a biological context. In this regard, metabonomics
approaches provide a final link between gene expression and cellular phenotype
(Nicholson et al., 2002).
In LEC rats, metabolic perturbations that may predispose this rat strain to hepatitis and
hepatocarcinogenesis have been described. Depleted liver selenium and deficiency in
the activity of selenium dependent enzymes have been found (Downey et al., 1998).
Lipid metabolism is also deregulated in LEC rats, as an increased content of
triglycerides, free cholesterol and a reduction of bile acid export has been described in
12 weeks old LEC rats (Levy et al., 2007). Other perturbations have been explained as
consequences of oxidative environment in LEC rat liver. Diminution of protein-thiols
and of reduced glutathione, together with oxidative modification of glycoprotein’s
oligosaccharides during hepatitis are some examples (Samuele et al., 2005; Yasuda et
al., 2006).
Whether metabolic perturbations in LEC rats are associated with gene expression
changes during hepatitis and hepatocarcinogenesis development is difficult to explore
by analyzing case by case. Our aims in this part of the work were first, to determine
whether metabolic profiles obtained by 1H NMR could explain hepatitis development in
LEC rats; and second, to define canonical correlations between transcriptomic data and
final metabolic changes.
116
D.3.2 EXPERIMENTAL PROTOCOL
D.3.2.1 Sample preparation for 1H NMR analysis: Urine and liver extracts were
prepared for 1H RMN analysis as described in the material and methods chapter.
D.3.2.2 Data analyses: 1H NMR spectra from 0.5 to 10 ppm were obtained and
cut in 0.04 ppm buckets. Quantitative variables were generated by integrating the area
under the curve of each bucket. From each 10 ppm spectra 220 variables were obtained.
Variables were then subjected to multivariate statistical methods. OSC was performed
to each group of data. Discriminant analyses (FDA and PLS-DA) were made in order to
evaluate the discrimination between metabolic profiles belonging to each LEC rat
group. PLS-2 regression between metabonomic data and hepatitis markers in plasma
was performed in order to associate metabolic changes to hepatitis development. Finally
a PLS-2 regression between liver metabolic and transcriptomic data was made with the
objective of explaining metabolic changes by changes in gene expression patterns in the
onset in LEC rats hepatitis.
D.3.3 RESULTS
D.3.3.1 1H NMR of urinary samples of LEC rats before and in the onset of hepatitis
Urine samples collected from non-treated and D-penicillamine-treated LEC rats were
analyzed by 1H NMR. Metabolic profiles were then analyzed and subjected to a
factorial discriminant analysis (FDA) (Figure 23). The first factorial axis given by this
analysis explained 41.9% of the variance and projection on this axis separates non-
treated from treated animals. The second axis explained 22.7% of the variance and
projection on this axis pointed out at the same time the development of hepatitis for the
non-treated animals and the metabolic changes due to the age of the animals.
117
Figure 23. FDA of 1H NMR spectra of urine samples: Urine samples from rats housed in metabolic cages were analyzed from 6 weeks old until sacrifice. 50 discriminant variables were selected with a P < 0.05. Black arrow shows the trajectory from untreated animals towards hepatitis. Grey arrow shows the trajectory of D-penicillamine-treated rats.
The fact that we were able to discriminate rats that had received D-penicillamine from
those rats without treatment lead us to ask if these differences would be related to
difference in hepatitis development among the groups or to metabolic changes due to
the age of the animals. In order to resolve this question a PLS-2 regression between
urine 1H NMR metabolic profiles and plasma levels of hepatitis markers (ALT, AST
and t-bil) was made. In figure 24, the relation between changes in metabolic profiles and
plasma levels of hepatitis markers is shown. There was a good separation between rats
treated with D-penicillamine and non-treated rats. The first component explained the
45.1% of the variance from metabonomic data while the second component explained
118
Figure 24. PLS-2 regression between 1H NMR metabolic profiles (t[1] axis) and
plasma levels of hepatitis markers (u[1]) axis
Table VI. Identified metabolites that were related to hepatitis development in LEC
rats: Variables were selected by a PLS-2 regression between metabonomics in urine and levels of hepatitis markers in plasma. cor: correlation coeficient
119
85.4% of the variance from biochemical data. These results mean that the differences in
metabolic profiles in urine samples were related to disease development. 21 variables
with a correlation coefficient (cor) above 0.8 between metabonomic and biochemical
data were found. From these variables, only 4 were identified from 1H MNR spectra.
Taurine, acetamide, and isoleucine were positively correlated with biochemical
parameters, that means that when ALT, AST and t-bil plasma level augmented these
variables also augmented, while citrate was negatively correlated to hepatitis markers
(Table VI).
D.3.3.2 Metabolic changes in liver of LEC rats at different hepatitis stages
Since one of the main objectives of this work was to define a relation between
transcriptomic and metabonomic data, we analyzed the changes in metabolome directly
in liver samples of LEC rats at different hepatitis stages. Aqueous and lipophilic
extracts of liver were obtained and analyzed by 1H NMR.
In order to achieve a better correspondence between data from metabonomic and
transcriptomic analysis, metabonomic data were expressed as the test/control ratio
similarly to transcriptomic data. Test rats corresponded to non-treated LEC rats (classed
according to hepatitis stages) and control rats corresponded to D-penicillamine-treated
LEC rats. Ratios were obtained by comparing animals identically as done in microarray
slides (See Table II).
D.3.3.2.1 Metabonomic analysis of lipophilic extracts of LEC rat liver samples
Figure 25 shows the pattern recognition analyses of metabonomic data from lipophilic
(chloroform:methanol, 2/1,v/v) extracts of LEC rat liver samples at different hepatitis
stages. Data were subjected to an OSC after which 51.4% of the information remained.
120
Figure 25. Pattern recognition analysis of
1H NMR spectra of lipophilic extracts from
liver samples of LEC rats at different hepatitis stages: A. PLS-DA. B. FDA of 10 discriminant variables selected by ANOVA with a P < 0.05. N 6w: Normal 6w; N 9w: Normal 9w; Sl j: Slight jaundice; J: Jaundice.
Table VII. Mahalanobis distances between groups from FDA analysis of
1H NMR
spectra of liver lipophilic extracts.
121
Discriminant analysis of the data showed the trajectory of groups from non diseased
(Normal 6w and Normal 9w groups) to hepatitis (Slight jaundice and Jaundice groups).
However, discrimination of groups was not so evident in the PLS-DA (Figure 25A).
AFD with 10 discriminant variables was better to oppose non-diseased rats from
diseased rats in the first axis which explained the 79.9% of the variance between groups.
The probability of Mahalanobis distance to be null were very low (P < 0.001) (Table
VII).
D.3.3.2.2 Metabonomic analysis of aqueous extracts of LEC liver samples
Aqueous (acetonitrile:H2O, 50/50, v/v) liver extracts were subjected to high resolution
1H RMN and changes in metabolic profiles were then analyzed by pattern recognition
analyses (Figure 26). 60.1% of information remained after an OSC. Both discriminant
analyses (PLS-DA and FDA) showed a good separation of each group of LEC rats with
a trajectory that explained hepatitis development in the first axis. Furthermore, FDA
analysis using 10 discriminant variables explained the 81.7% of the variance between
groups in the first axis. The probability of Mahalanobis distance to be null were very
low (P < 0.001) (Table VIII). Variables that explained the axes building are shown in
Table IX.
Interestingly, the trajectory in which hepatitis development was explained by
metabonomic data of aqueous liver extracts was very similar of that seen by
transcriptomic data (Figure 14). So we focused on these data coming from aqueous
extracts to explore if there was any relation between metabolic profile changes and gene
expression changes due to hepatitis development.
122
Figure 26. Pattern recognition analyses of
1H NMR of aqueous liver extracts of LEC
rats at different hepatitis stages: A. PLS-DA. B. FDA from 10 discriminant variables selected by ANOVA with a P < 0.05. N 6w: Normal 6w; N 9w: Normal 9w; Sl-j: Slight jaundice; J: Jaundice. Arrows indicate the trajectory on the axes from non-diseased to diseased rat groups.
Table VIII. Mahalanobis distances between groups from FDA analysis of
1H NMR
spectra of liver aqueous extracts
123
Table IX. Discriminant variables selected by ANOVA (P < 0.05) in a FDA: Metabolite associated gene ontology descriptions are shown.
124
D.3.3.3 Quantitative relationship between metabonomic and transcriptomic analyses
In order to define whether gene expression changes could explain the changes in
metabolic profiles seen in aqueous extracts of liver samples of LEC rats, a PLS-2
regression between transcriptomic data (as X matrix) and metabonomic data (as Y
matrix) was done (Figure 27A). The correlation between both data showed a linear
trajectory that explained hepatitis development. The first component for transcriptomic
data explained the 34.3% of the variance, while the first component for metabonomic
data explained the 30.6% of the variance between groups.
In order to know which variables were correlated between the two omics analyses, a
correlation matrix between both groups of data was done. From it, 39 variables with a
correlation coefficient above 0.85 were selected (Table B, annexes). The representation
of transcriptomic and metabonomic variables in a Cartesian plane showed positive and
negative correlation between variables (Figure 27B). This kind of analysis allowed us to
demonstrate that it was possible to associate changes in metabolic profiles (Y) due to
hepatitis development by changes in gene expression patterns (X). D.3.4
DISCUSSION
Metabolic perturbations related to copper accumulation and oxidative stress have been
extensively described. Copper induces ROS formation which in turns inhibit
mitochondrial enzymes like α-ketoglutarate dehydrogenase and pyruvate dehydrogenase
(Sheline et al., 2004). Deregulation in lipid metabolism was observed in a mouse model
of Wilson’s disease, in which the main modification was an inhibition of cholesterol
metabolism (Huster et al., 2007). Moreover, the production and metabolism of ROS is
highly regulated by metabolic pathways, the pentose phosphate cycle acting
125
Figure 27 . PLS-2 regression between transcriptomic and metabonomic data. A. PLS-2, transcriptomic data (X matrix) was used to explain metabolic changes (Y matrix) in aqueous extracts of LEC rat livers at different hepatitis stages. B. Correlation between variables from transcriptomic data (green circles) and metabonomic data (orange boxes). Variables with a correlation coefficient > 0.85 are represented. N 6w: Normal 6w; N 9w: Normal 9w; Sl-j; Slight jaundice; J: Jaundice.
126
actually as the main metabolic response under oxidative stress with the production of
NADPH, a metabolite indispensable to maintain cellular reductive capacity (Tuttle et
al., 2007). Because alterations in physiological status by oxidative stress can break
homeostasis, resulting in ultimate changes in metabolites, metabonomic analyses are of
interest in studying and understanding the molecular mechanism of diseases.
Using 1H NMR coupled with pattern recognition techniques, we were able to
discriminate LEC rats undergoing hepatitis in both urine and liver samples.
Metabonomic analysis of urine samples showed that differences between untreated and
D-penicillamine treated LEC rats were related to hepatitis development as seen by its
good correlation with hepatitis markers in plasma. Furthermore, from the identified
urine metabolites (Table VI), taurine and acetamide groups were positively correlated
with plasmatic markers. Interestingly, taurine has been reported to be depleted in the
liver under oxidative stress due to chronic alcohol administration to rats (Kerai et al.,
2001), while hypertaurinuria has been observed when hepatotoxins were administered
to rats (Sanins et al., 1990). In our study on liver metabonomic data, taurine variable
was not selected. The acetamide group might be directly related to oxidative
degradation of bilirubin, as oxidative metabolites of bilirubin bearing an acetamide
function have been evidenced under oxidative stress conditions such as subarachnoid
hemorrhage (Kranc et al., 2000). The fact that pattern recognition techniques
discriminates diseased animals from non diseased in urine samples is promising while
new markers of hepatitis could be further identified and used as non-invasive tests.
Analysis of liver samples revealed that in both lipophilic and aqueous liver extracts,
metabolic changes are associated with hepatitis development as seen by a trajectory
from non diseased LEC rats (Normal 6w and Normal 9w groups) towards diseased LEC
rats (Slight jaundice and Jaundice groups). However, in this work, we focused on
127
aqueous extracts to understand metabolic perturbations regarding hepatitis and
transcriptomic data because of similar trajectories we obtained between metabonomic
and transcriptomic data (Figure 14 and 26). However further studies should be done in
order to identify and to explore changes in non polar metabolites.
Aqueous metabolites that were of importance for axis construction in FDA showed that
metabolic perturbations associated with hepatitis development included energetic
metabolism and amino acid metabolism (Table IX). Decreased levels of α-glucose lead
to the hypothesis of accelerated glycolysis. Perturbation of amino acid metabolism was
evident with increased levels of lysine and glutamate in Slight jaundice and Jaundice
groups. Enhanced glycolysis has been proposed as a protection mechanism face to
oxidative stress in cancer cells (Kondoh et al., 2007). In proliferating cells, transitional
augmentation of aerobic glycolysis and reduced ROS formation have been explained by
two cooperative effects: (a) generation of antioxidant molecules like pyruvate, citrate
and glutamate, and (b) prevention of excessive ROS production by keeping the rate of
mitochondrial glucose oxidation at basal levels (Brand, 1997; Wu et al., 2004). The
hypothesis of glucose enhanced metabolism as an adaptive response to oxidative stress
in LEC rat hepatitis is questioned and more studies should be made to test this idea.
Increase in lysine and glutamate could reflect the increase in protein catabolism
occurring during hepatitis.
Integrative biology refers to the integration of biological data to elucidate the cellular
network as a whole (Joyce et al., 2006). Integration of omics data is not easy to achieve
and requires the use of dedicated statistics and bioinformatics tools. Here we used PLS-
2 regression in order to correlate the metabolic changes in liver of rats suffering of
hepatitis by changes in gene expression patterns. Very interestingly, we found a linear
correlation between metabonomic and transcriptomic data associated with hepatitis
128
development and progression in LEC rats. This result might mean that both gene
expression changes and metabolic perturbations are closely related face to hepatitis
development. The correlation matrix formed by this analysis allowed us not only to
select new genes that were modulated by oxidative stress-induced hepatitis and that
were not evident in a single transcriptomic analysis, but also to formulate new
hypothesis about cellular adaptation to oxidative stress.
In a general context, the metabolites found to be correlated to gene expression changes
were mainly related to oxidative stress and to cellular adaptation to it. Increased levels
of oxidized glutathione, and amino acids related to glutathione biosynthesis (glutamate
and glutamine) reflect the oxidative conditions under hepatitis. While, as mentioned
above, diminution of glucose might reflect cellular protection to oxidative stress.
Genes related to these conditions might reflect the cellular responses or consequences
on general metabolic perturbations and not necessarily to a direct metabolite/gene
relation. Diminution of cytochrome P450 gene expression, as it is the case in our study
with the progression of hepatitis (Table B, annexes), has been described under oxidative
stress and inflammation. Since basal activity of cytochromes P450 leads to a continuous
ROS production, downregulation of cytochrome P450 genes has been considered as a
mechanisms to protect cells of deleterious effects of ROS (Morgan, 2001). Mig-6 gene,
which expression is also decreased with the progression of hepatitis (Table B, annexes),
is coding for an adapter protein that has been associated with proliferation and
differentiation control and is defined as an endogenous inhibitor of epidermal grow
factor receptor (EGFR) signaling (Xu et al., 2005). Down-regulation of this gene might
be associated with liver regeneration via hepatocyte proliferation as compensative
response to acute hepatitis. Alpha-methylacyl-CoA racemase gene, which is down
regulated with hepatitis in our study (Table B, annexes), is involved in the catabolism of
129
branched-chain fatty acids. Deficiency in this gene leads to decrease of bile acid
synthesis and consequently to choleostasis (Setchell et al., 2003).
The understanding of complex process like hepatitis and hepatocarcinogenesis is a big
deal in which several processes should be examined at the same time: apoptosis,
necrosis due to toxic effects of free radicals and in the other hand, inflammation and
hepatocyte proliferation as compensative mechanisms of liver damage. Metabolic
perturbations related to these events are not easy to elucidate, and the analysis of great
amounts of information is required. Thanks to 1H NMR metabonomic analyses we were
able to select, with a systemic approach, some metabolites that are related to oxidative
stress-induced hepatitis. However, low sensitivity and difficulties in peak interpretation
and molecule identification made the analysis restrictive to some identified metabolites
present in huge amounts. Nevertheless, we can not discard the importance of
metabonomic data, mainly as an integrated approach to transcriptomic data in
formulating new questions about mechanisms that lead to hepatitis development and the
association of hepatitis to cancer initiation.
130
131
E. GENERAL DISCUSSION AND PERSPECTIVES
132
133
The main objective of this work was to explore in a global scale the changes of gene
expression and their relation with perturbations in metabolic profiles in a model of
oxidative stress-induced hepatitis. Since acute hepatitis is closely related to liver cancer
initiation in LEC rats, perturbations in gene expression and in metabolic profiles could
also be analyzed within the scope of early modifications that can be associated with
carcinogenesis.
Transcriptomic approach led us to the construction of a database with the expression
levels of 27004 oligo probes which correspond to almost all the rat genome. This
database was used for the study of gene expression pattern and for the formulation of
new hypotheses. One of the problems that we were confronted to was the biological
interpretation of data, since only around 40% of genome has a defined function and
hence our analysis was restricted to this. However, using our transcriptomic database
and as advances in gene function definition arise, further analyses could be made in
order to find new insights about mechanisms of hepatitis and hepatocarcinogenesis.
Proteasome activity inhibition in LEC rats during hepatitis was a major finding; this is
the first report that implicates proteasome inhibition during hepatitis and its possible
consequence in hepatocarcinogenesis. We were able to achieve this finding thanks to
the original analysis of data that we performed. Two points should be underlined from
this finding: 1) Data analysis: data coming from transcriptomic analyses were
performed without any restriction of pre-made hypotheses or bias that could arise from
the exploration of a gene by gene analysis, where the observation of data is restricted to
modifications in “expected” or “known” genes. And 2) validation of hypothesis raised
from transcriptomic data is indispensable, not only by measuring the mRNA levels by
other techniques but by testing the activity of the proteins coded by the selected genes.
As it was the case for proteasome genes, where our initial hypothesis was an
134
upregulation in protein degradation because we found genes related to ubiquitin-
proteasome system upregulated, contradictory results could arise.
Protein degradation by proteasome is a key mechanism in protein turnover. Proteasome
activity deregulation may result in damaged proteins accumulation leading to toxic
insults to cells and to modifications in signal pathways. It could be therefore interesting
to evaluate the accumulation of oxidized proteins during the onset of hepatitis and to
find a correlation between this damaged protein accumulation and cell perturbations that
could results in cell cancer initiation. Oxidative stress-related modifications in
proteasome activity have been described in vivo in a model of kidney carcinogenesis.
Lipid peroxidation products such as HNE are directly implicated for proteasome activity
inhibition. As HNE has been described as a key molecule for oxidative stress-related
regulation of cell faith, it is also of interest to study the implication of HNE in ROS-
related inhibition of proteasome in liver. This would add more insights in the
mechanisms of ROS induced carcinogenesis.
In this work we were also interested in defining the changes in metabolic profiles that
lead to hepatitis. Our analysis of the metabolome based on NMR-metabonomics showed
a good correlation between changes in metabolic profiles and the levels of hepatitis
markers in plasma that lead us to conclude that there should be a metabolic response
face to oxidative stress and inflammation. However, the biological interpretation of
metabonomic analyses was limited because of the sensitivity of the analytical method
used to define metabolic profiles, the 1H NMR, and because of limitations in variable
identification. Even when pattern recognition analyses showed a very good
discrimination between groups with a trajectory that defined hepatitis development, the
variable identification led to the definition of only some metabolites. Moreover, 1H
NMR allows only the analysis of compounds that are present in huge amounts. We can
135
not discourage the use of 1H NMR for metabonomic analyses because it gives a
complete overview of well represented metabolites, but complementary analytical
procedures might be done, as mass spectrometry, in order to define some other
metabolites that are not concentrated enough to be seen by NMR.
Systems biology is a new discipline in biology that aims to integrate omic data between
each other and with other biological observations. Here, in an effort to better understand
the mechanism that lead to a hepatitis phenotype, we integrated the data coming from
transcriptomic techniques with those data coming from 1H NRM metabonomic analyses
in liver samples. PLS-2 regression between these two different data sets showed a linear
correlation between transcriptomic and metabonomic data in explaining the hepatitis
development. This result is very important since it would mean that both changes in
gene expression patterns and in metabolic profiles were closely correlated.
It was exiting to find correlation between transcriptomic and metabonomic data because
this may allow to new biological signatures implicated in ROS-induced liver diseases.
Furthermore, this is a significant advance in the discovering of new insights on
mechanism of hepatitis and in the discovering of possible early markers of
hepatocarcinogenesis. However, there was no evident biological relation between both
data (mRNA levels and metabolic profiles). It is worth to say that these kind of
canonical analyses allow us to explore the relationship between to set of variables, but it
does not mean a direct relation gene: metabolite. From our analysis, only one direct
relationship between gene and metabolite has been detected where aldehyde
dehydrogenase 18 is involved in glutamine degradation (Table B). However, some
hypotheses can be made and further investigated, for instance concerning the role of
cytochromes P450 in the development of hepatitis.
136
The expression levels of the genes that are directly related to the metabolites selected by
the PLS2 (genes which code for enzymes involved in their metabolic pathways) were
not significantly modified. Hence, there were not selected by the multivariate statistical
analyses. The fact that there were metabolic perturbations that explain the progression
of hepatitis, but that these perturbations could not be biologically explained by changes
in gene expression patterns lead us to suppose that there should be modification in the
enzyme levels or activity. Therefore it should be important to analyze the changes in
protein profiles that would be related to hepatitis development. Proteomics could be
used then as a complementary tool in order to gain complete information of cellular
response face to oxidative stress and inflammation. Using the whole data coming from
these three principal omic approaches (transcriptomics, proteomics and metabonomics),
we would be able to integrate the puzzle of the relationship gene: protein � protein:
metabolite that may explain complex processes like hepatitis and cancer initiation.
Furthermore, using proteomic analyses it would be not only possible to evaluate directly
the accumulation of oxidized proteins caused by perturbations in proteasome activity
but also to identify possible targets of ROS-induced protein modifications implicated in
hepatitis and cancer initiation processes.
Finally and as a general comment of this work I want to say that omic analyses are
nowadays very popular in biological and pharmaceutical research, their applications in
clinical and pharmacological investigations are promising, since we will be able to
study in a global scale the physiological or pathological cellular responses face to
diseases or to drug exposition, new insights in mechanism of diseases would be
elucidated, new biological markers of disease or toxicological effect would be found.
This new vision in biology implies the management and the interpretation of high
amount of data, and the integration of disciplines of different fields such as chemistry,
137
biology, physiology, informatics, mathematics is indispensable in the objective of
diminishing the reductionism vision of individual component study for explain complex
phenomena or to avoid the bias of “what we want to find” that could arise from pre-
made hypothesis.
138
139
F. REFERENCES
140
141
ABDI, H. (2003). Partial Least Squares (PLS) Regression. In Encyclopedia of Social Sciences Research Methods. T. Futing, ed. (Sage, Thousand Oaks, CA).
ALARY, J., BRAVAIS, F., CRAVEDI, J.P., DEBRAUWER, L., RAO, D., and BORIES, G. (1995). Mercapturic acid conjugates as urinary end metabolites of the lipid peroxidation product 4-hydroxy-2-nonenal in the rat. Chem Res Toxicol 8, 34-39.
ALARY, J., DEBRAUWER, L., FERNANDEZ, Y., PARIS, A., CRAVEDI, J.P., DOLO, L., et al. (1998). Identification of novel urinary metabolites of the lipid peroxidation product 4-hydroxy-2-nonenal in rats. Chem Res Toxicol 11, 1368-1376.
ALIYA, S., REDDANNA, P., and THYAGARAJU, K. (2003). Does glutathione S-transferase Pi (GST-Pi) a marker protein for cancer? Mol Cell Biochem 253, 319-327.
ALLEN, R.G., and TRESINI, M. (2000). Oxidative stress and gene regulation. Free Radic Biol Med 28, 463-499.
ANISIMOV, V.N. (2007). Biology of aging and cancer. Cancer Control 14, 23-31. BA, Y., TONOKI, H., TADA, M., NAKATA, D., HAMADA, J., and MORIUCHI, T.
(2000). Transcriptional slippage of p53 gene enhanced by cellular damage in rat liver: monitoring the slippage by a yeast functional assay. Mutat Res 447, 209-220.
BERLETT, B.S., and STADTMAN, E.R. (1997). Protein oxidation in aging, disease, and oxidative stress. J Biol Chem 272, 20313-20316.
BRAND, K. (1997). Aerobic glycolysis by proliferating cells: protection against oxidative stress at the expense of energy yield. J Bioenerg Biomembr 29, 355-364.
BRENTANI, R.R., CARRARO, D.M., VERJOVSKI-ALMEIDA, S., REIS, E.M., NEVES, E.J., DE SOUZA, S.J., et al. (2005). Gene expression arrays in cancer research: methods and applications. Crit Rev Oncol Hematol 54, 95-105.
BUNAI, K., and YAMANE, K. (2005). Effectiveness and limitation of two-dimensional gel electrophoresis in bacterial membrane protein proteomics and perspectives. J Chromatogr B Analyt Technol Biomed Life Sci 815, 227-236.
CADET, J., BOURDAT, A.G., D'HAM, C., DUARTE, V., GASPARUTTO, D., ROMIEU, A., et al. (2000). Oxidative base damage to DNA: specificity of base excision repair enzymes. Mutat Res 462, 121-128.
CECARINI, V., GEE, J., FIORETTI, E., AMICI, M., ANGELETTI, M., ELEUTERI, A.M., et al. (2007). Protein oxidation and cellular homeostasis: Emphasis on metabolism. Biochim Biophys Acta 1773, 93-104.
CESARATTO, L., VASCOTTO, C., D'AMBROSIO, C., SCALONI, A., BACCARANI, U., PARON, I., et al. (2005). Overoxidation of peroxiredoxins as an immediate and sensitive marker of oxidative stress in HepG2 cells and its application to the redox effects induced by ischemia/reperfusion in human liver. Free Radic Res 39, 255-268.
CHAUDIERE, J., and FERRARI-ILIOU, R. (1999). Intracellular antioxidants: from chemical to biochemical mechanisms. Food Chem Toxicol 37, 949-962.
CHEN, L., GORYACHEV, A., SUN, J., KIM, P., ZHANG, H., PHILLIPS, M.J., et al. (2003). Altered expression of genes involved in hepatic morphogenesis and fibrogenesis are identified by cDNA microarray analysis in biliary atresia. Hepatology 38, 567-576.
CHEN, Z.H., SAITO, Y., YOSHIDA, Y., SEKINE, A., NOGUCHI, N., and NIKI, E. (2005). 4-Hydroxynonenal induces adaptive response and enhances PC12 cell
142
tolerance primarily through induction of thioredoxin reductase 1 via activation of Nrf2. J Biol Chem 280, 41921-41927.
CHITTUR, S.V. (2004). DNA microarrays: tools for the 21st Century. Comb Chem High Throughput Screen 7, 531-537.
CHOUDHURY, S., ZHANG, R., FRENKEL, K., KAWAMORI, T., CHUNG, F.L., and ROY, R. (2003). Evidence of alterations in base excision repair of oxidative DNA damage during spontaneous hepatocarcinogenesis in Long Evans Cinnamon rats. Cancer Res 63, 7704-7707.
CIECHANOVER, A. (2005). Intracellular protein degradation: from a vague idea thru the lysosome and the ubiquitin-proteasome system and onto human diseases and drug targeting. Cell Death Differ 12, 1178-1190.
COMPORTI, M. (1998). Lipid peroxidation and biogenic aldehydes: from the identification of 4-hydroxynonenal to further achievements in biopathology. Free Radic Res 28, 623-635.
CONNER, E.M., and GRISHAM, M.B. (1996). Inflammation, free radicals, and antioxidants. Nutrition 12, 274-277.
DALLE-DONNE, I., ROSSI, R., GIUSTARINI, D., MILZANI, A., and COLOMBO, R. (2003). Protein carbonyl groups as biomarkers of oxidative stress. Clin Chim Acta 329, 23-38.
DIZDAROGLU, M., JARUGA, P., BIRINCIOGLU, M., and RODRIGUEZ, H. (2002). Free radical-induced damage to DNA: mechanisms and measurement. Free Radic Biol Med 32, 1102-1115.
DMITROVSKY, E. (2004). Tissue microarrays for hypothesis generation. J Natl Cancer Inst 96, 248-249.
DOWNEY, J.S., BINGLE, C.D., COTTRELL, S., WARD, N., CHURCHMAN, D., DOBROTA, M., et al. (1998). The LEC rat possesses reduced hepatic selenium, contributing to the severity of spontaneous hepatitis and sensitivity to carcinogenesis. Biochem Biophys Res Commun 244, 463-467.
DROGE, W. (2002). Free radicals in the physiological control of cell function. Physiol Rev 82, 47-95.
DUMAS, M.E., CANLET, C., ANDRE, F., VERCAUTEREN, J., and PARIS, A. (2002). Metabonomic assessment of physiological disruptions using 1H-13C HMBC-NMR spectroscopy combined with pattern recognition procedures performed on filtered variables. Anal Chem 74, 2261-2273.
ECKL, P.M. (2003). Genotoxicity of HNE. Mol Aspects Med 24, 161-165. EL-SERAG, H.B., and RUDOLPH, K.L. (2007). Hepatocellular carcinoma:
epidemiology and molecular carcinogenesis. Gastroenterology 132, 2557-2576. ESTERBAUER, H., SCHAUR, R.J., and ZOLLNER, H. (1991). Chemistry and
biochemistry of 4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radic Biol Med 11, 81-128.
FAN, T.W., LANE, A.N., and HIGASHI, R.M. (2004). The promise of metabolomics in cancer molecular therapeutics. Curr Opin Mol Ther 6, 584-592.
FARBER, E., and SARMA, D.S. (1986). Chemical carcinogenesis: the liver as a model. Pathol Immunopathol Res 5, 1-28.
FATACCIOLI, V., ANDRAUD, E., GENTIL, M., FRENCH, S.W., and ROUACH, H. (1999). Effects of chronic ethanol administration on rat liver proteasome activities: relationship with oxidative stress. Hepatology 29, 14-20.
FENG, Z., HU, W., and TANG, M.S. (2004). Trans-4-hydroxy-2-nonenal inhibits nucleotide excision repair in human cells: a possible mechanism for lipid
143
peroxidation-induced carcinogenesis. Proc Natl Acad Sci U S A 101, 8598-8602.
FINK, S.P., REDDY, G.R., and MARNETT, L.J. (1997). Mutagenicity in Escherichia coli of the major DNA adduct derived from the endogenous mutagen malondialdehyde. Proc Natl Acad Sci U S A 94, 8652-8657.
FLOYD, R.A. (1990). The role of 8-hydroxyguanine in carcinogenesis. Carcinogenesis 11, 1447-1450.
FRIGUET, B. (2006). Oxidized protein degradation and repair in ageing and oxidative stress. FEBS Lett 580, 2910-2916.
FUKUDA, A., NAKAMURA, Y., OHIGASHI, H., OSAWA, T., and UCHIDA, K. (1997). Cellular response to the redox active lipid peroxidation products: induction of glutathione S-transferase P by 4-hydroxy-2-nonenal. Biochem Biophys Res Commun 236, 505-509.
GENESTRA, M. (2007). Oxyl radicals, redox-sensitive signalling cascades and antioxidants. Cell Signal 19, 1807-1819.
GRIFFIN, J.L. (2003). Metabonomics: NMR spectroscopy and pattern recognition analysis of body fluids and tissues for characterisation of xenobiotic toxicity and disease diagnosis. Curr Opin Chem Biol 7, 648-654.
GRUNE, T., and DAVIES, K.J. (2003a). The proteasomal system and HNE-modified proteins. Mol Aspects Med 24, 195-204.
GRUNE, T., MERKER, K., SANDIG, G., and DAVIES, K.J. (2003b). Selective degradation of oxidatively modified protein substrates by the proteasome. Biochem Biophys Res Commun 305, 709-718.
GRUNE, T., REINHECKEL, T., JOSHI, M., and DAVIES, K.J. (1995). Proteolysis in cultured liver epithelial cells during oxidative stress. Role of the multicatalytic proteinase complex, proteasome. J Biol Chem 270, 2344-2351.
GUERAUD, F., PEIRO, G., BERNARD, H., ALARY, J., CREMINON, C., DEBRAUWER, L., et al. (2006). Enzyme immunoassay for a urinary metabolite of 4-hydroxynonenal as a marker of lipid peroxidation. Free Radic Biol Med 40, 54-62.
HALLIWELL, B., and CHIRICO, S. (1993). Lipid peroxidation: its mechanism, measurement, and significance. Am J Clin Nutr 57, 715S-724S; discussion 724S-725S.
HAWKINS, R.L., MORI, M., INOUE, M., and TORII, K. (1995). Proline, ascorbic acid, or thioredoxin affect jaundice and mortality in Long Evans cinnamon rats. Pharmacol Biochem Behav 52, 509-515.
HAYASHI, M., MIYANE, K., HIROOKA, T., ENDOH, D., HIGUCHI, H., NAGAHATA, H., et al. (2004). Inhibitory effects of trientine, a copper-chelating agent, on induction of DNA strand breaks in hepatic cells of Long-Evans Cinnamon rats. Biochim Biophys Acta 1674, 312-318.
HAYASHI, T., and GOTO, S. (1998). Age-related changes in the 20S and 26S proteasome activities in the liver of male F344 rats. Mech Ageing Dev 102, 55-66.
HAYES, J.D., FLANAGAN, J.U., and JOWSEY, I.R. (2005). Glutathione transferases. Annu Rev Pharmacol Toxicol 45, 51-88.
HOCQUETTE, J.F. (2005). Where are we in genomics? J Physiol Pharmacol 56 Suppl 3, 37-70.
HOLMGREN, A. (2000). Antioxidant function of thioredoxin and glutaredoxin systems. Antioxid Redox Signal 2, 811-820.
144
HOLMGREN, A., JOHANSSON, C., BERNDT, C., LONN, M.E., HUDEMANN, C., and LILLIG, C.H. (2005). Thiol redox control via thioredoxin and glutaredoxin systems. Biochem Soc Trans 33, 1375-1377.
HU, W., FENG, Z., EVELEIGH, J., IYER, G., PAN, J., AMIN, S., et al. (2002). The major lipid peroxidation product, trans-4-hydroxy-2-nonenal, preferentially forms DNA adducts at codon 249 of human p53 gene, a unique mutational hotspot in hepatocellular carcinoma. Carcinogenesis 23, 1781-1789.
HUMPHERY-SMITH, I. (2004). A human proteome project with a beginning and an end. Proteomics 4, 2519-2521.
HUSTER, D., PURNAT, T.D., BURKHEAD, J.L., RALLE, M., FIEHN, O., STUCKERT, F., et al. (2007). High copper selectively alters lipid metabolism and cell cycle machinery in the mouse model of Wilson disease. J Biol Chem 282, 8343-8355.
INOUE, M., SATO, E.F., NISHIKAWA, M., PARK, A.M., KIRA, Y., IMADA, I., et al. (2003). Mitochondrial generation of reactive oxygen species and its role in aerobic life. Curr Med Chem 10, 2495-2505.
IPPOLITO, J.E., XU, J., JAIN, S., MOULDER, K., MENNERICK, S., CROWLEY, J.R., et al. (2005). An integrated functional genomics and metabolomics approach for defining poor prognosis in human neuroendocrine cancers. Proc Natl Acad Sci U S A 102, 9901-9906.
IVAKHNO, S. (2007). From functional genomics to systems biology. Febs J 274, 2439-2448.
IVAKHNO, S., and ARMSTRONG, J.D. (2007). Non-linear dimensionality reduction of signaling networks. BMC Syst Biol 1, 27.
IWADATE, H., OHIRA, H., SUZUKI, T., ABE, K., YOKOKAWA, J., TAKIGUCHI, J., et al. (2004). Hepatocellular carcinoma associated with Wilson's disease. Intern Med 43, 1042-1045.
JAIN, S.K., SUZUKI, K., JAIN, S., SUGIYAMA, T., and TANIGUCHI, N. (1992). Hypomethilation-associated expression of cytochrome P-450 and γ-glutamyl transpeptidase during hereditary hepatocarcinogenesis in LEC rats. In The LEC rat. M. Mori, M.C. Yoshida, H. Takahashi, and N. Taniguchi, eds. (Springer-Verlag, New York) pp. 162-168.
JIA, G., TOHYAMA, C., and SONE, H. (2002). DNA damage triggers imbalance of proliferation and apoptosis during development of preneoplastic foci in the liver of Long-Evans Cinnamon rats. Int J Oncol 21, 755-761.
JONG-HON, K., TOGASHI, Y., KASAI, H., HOSOKAWA, M., and TAKEICHI, N. (1993). Prevention of spontaneous hepatocellular carcinoma in Long-Evans cinnamon rats with hereditary hepatitis by the administration of D-penicillamine. Hepatology 18, 614-620.
JOYCE, A.R., and PALSSON, B.O. (2006). The model organism as a system: integrating 'omics' data sets. Nat Rev Mol Cell Biol 7, 198-210.
KAMADA, K., GOTO, S., OKUNAGA, T., IHARA, Y., TSUJI, K., KAWAI, Y., et al. (2004). Nuclear glutathione S-transferase pi prevents apoptosis by reducing the oxidative stress-induced formation of exocyclic DNA products. Free Radic Biol Med 37, 1875-1884.
KASAI, N., OSANAI, T., MIYOSHI, I., KAMIMURA, E., YOSHIDA, M.C., and DEMPO, K. (1990). Clinico-pathological studies of LEC rats with hereditary hepatitis and hepatoma in the acute phase of hepatitis. Lab Anim Sci 40, 502-505.
145
KERAI, M.D., WATERFIELD, C.J., KENYON, S.H., ASKER, D.S., and TIMBRELL, J.A. (2001). The effect of taurine depletion by beta-alanine treatment on the susceptibility to ethanol-induced hepatic dysfunction in rats. Alcohol Alcohol 36, 29-38.
KITA, Y., MASAKI, T., FUNAKOSHI, F., YOSHIDA, S., TANAKA, M., KUROKOHCHI, K., et al. (2004). Expression of G1 phase-related cell cycle molecules in naturally developing hepatocellular carcinoma of Long-Evans Cinnamon rats. Int J Oncol 24, 1205-1211.
KITAMURA, Y., NISHIKAWA, A., NAKAMURA, H., FURUKAWA, F., IMAZAWA, T., UMEMURA, T., et al. (2005). Effects of N-acetylcysteine, quercetin, and phytic acid on spontaneous hepatic and renal lesions in LEC rats. Toxicol Pathol 33, 584-592.
KLAUNIG, J.E., and KAMENDULIS, L.M. (2004). The role of oxidative stress in carcinogenesis. Annu Rev Pharmacol Toxicol 44, 239-267.
KLEIN, D., LICHTMANNEGGER, J., FINCKH, M., and SUMMER, K.H. (2003). Gene expression in the liver of Long-Evans cinnamon rats during the development of hepatitis. Arch Toxicol 77, 568-575.
KLEIN, D., LICHTMANNEGGER, J., HEINZMANN, U., MULLER-HOCKER, J., MICHAELSEN, S., and SUMMER, K.H. (1998). Association of copper to metallothionein in hepatic lysosomes of Long-Evans cinnamon (LEC) rats during the development of hepatitis. Eur J Clin Invest 28, 302-310.
KLEIN, D., LICHTMANNEGGER, J., HEINZMANN, U., and SUMMER, K.H. (2000). Dissolution of copper-rich granules in hepatic lysosomes by D-penicillamine prevents the development of fulminant hepatitis in Long-Evans cinnamon rats. J Hepatol 32, 193-201.
KLENO, T.G., KIEHR, B., BAUNSGAARD, D., and SIDELMANN, U.G. (2004). Combination of 'omics' data to investigate the mechanism(s) of hydrazine-induced hepatotoxicity in rats and to identify potential biomarkers. Biomarkers 9, 116-138.
KOIKE, K. (2007). Hepatitis C virus contributes to hepatocarcinogenesis by modulating metabolic and intracellular signaling pathways. J Gastroenterol Hepatol 22 Suppl 1, S108-111.
KONDOH, H., LLEONART, M.E., BERNARD, D., and GIL, J. (2007). Protection from oxidative stress by enhanced glycolysis; a possible mechanism of cellular immortalization. Histol Histopathol 22, 85-90.
KRANC, K.R., PYNE, G.J., TAO, L., CLARIDGE, T.D., HARRIS, D.A., CADOUX-HUDSON, T.A., et al. (2000). Oxidative degradation of bilirubin produces vasoactive compounds. Eur J Biochem 267, 7094-7101.
LACONI, E., PANI, P., and FARBER, E. (2000). The resistance phenotype in the development and treatment of cancer. Lancet Oncol 1, 235-241.
LEE, J.S., CHU, I.S., HEO, J., CALVISI, D.F., SUN, Z., ROSKAMS, T., et al. (2004). Classification and prediction of survival in hepatocellular carcinoma by gene expression profiling. Hepatology 40, 667-676.
LEONARD, S.S., HARRIS, G.K., and SHI, X. (2004). Metal-induced oxidative stress and signal transduction. Free Radic Biol Med 37, 1921-1942.
LEVY, E., BRUNET, S., ALVAREZ, F., SEIDMAN, E., BOUCHARD, G., ESCOBAR, E., et al. (2007). Abnormal hepatobiliary and circulating lipid metabolism in the Long-Evans Cinnamon rat model of Wilson's disease. Life Sci 80, 1472-1483.
146
LI, Y., TOGASHI, Y., SATO, S., EMOTO, T., KANG, J.H., TAKEICHI, N., et al. (1991). Spontaneous hepatic copper accumulation in Long-Evans Cinnamon rats with hereditary hepatitis. A model of Wilson's disease. J Clin Invest 87, 1858-1861.
LINDER, M.C., and HAZEGH-AZAM, M. (1996). Copper biochemistry and molecular biology. Am J Clin Nutr 63, 797S-811S.
LINDON, J.C., KEUN, H.C., EBBELS, T.M., PEARCE, J.M., HOLMES, E., and NICHOLSON, J.K. (2005). The Consortium for Metabonomic Toxicology (COMET): aims, activities and achievements. Pharmacogenomics 6, 691-699.
LOMAX, J. (2005). Get ready to GO! A biologist's guide to the Gene Ontology. Brief Bioinform 6, 298-304.
LONDON, M.K., KECK, B.I., RAMOS, P.C., and DOHMEN, R.J. (2004). Regulatory mechanisms controlling biogenesis of ubiquitin and the proteasome. FEBS Lett 567, 259-264.
MARSIN, S., VIDAL, A.E., SOSSOU, M., MENISSIER-DE MURCIA, J., LE PAGE, F., BOITEUX, S., et al. (2003). Role of XRCC1 in the coordination and stimulation of oxidative DNA damage repair initiated by the DNA glycosylase hOGG1. J Biol Chem 278, 44068-44074.
MASUDA, R., YOSHIDA, M.C., and SASAKI, M. (1989). Gene expression of placental glutathione S-transferase in hereditary hepatitis and spontaneous hepatocarcinogenesis of LEC strain rats. Jpn J Cancer Res 80, 1024-1027.
MASUDA, R., YOSHIDA, M.C., SASAKI, M., DEMPO, K., and MORI, M. (1988). High susceptibility to hepatocellular carcinoma development in LEC rats with hereditary hepatitis. Jpn J Cancer Res 79, 828-835.
MCGLYNN, K.A., and LONDON, W.T. (2005). Epidemiology and natural history of hepatocellular carcinoma. Best Pract Res Clin Gastroenterol 19, 3-23.
MEINERS, S., HEYKEN, D., WELLER, A., LUDWIG, A., STANGL, K., KLOETZEL, P.M., et al. (2003). Inhibition of proteasome activity induces concerted expression of proteasome genes and de novo formation of Mammalian proteasomes. J Biol Chem 278, 21517-21525.
MENG, Y., MIYOSHI, I., HIRABAYASHI, M., SU, M., MOTOTANI, Y., OKAMURA, T., et al. (2004). Restoration of copper metabolism and rescue of hepatic abnormalities in LEC rats, an animal model of Wilson disease, by expression of human ATP7B gene. Biochim Biophys Acta 1690, 208-219.
MORGAN, E.T. (2001). Regulation of cytochrome p450 by inflammatory mediators: why and how? Drug Metab Dispos 29, 207-212.
MORI, M., HATTORI, A., SAWAKI, M., TSUZUKI, N., SAWADA, N., OYAMADA, M., et al. (1994). The LEC rat: a model for human hepatitis, liver cancer, and much more. Am J Pathol 144, 200-204.
MORTISHIRE-SMITH, R.J., SKILES, G.L., LAWRENCE, J.W., SPENCE, S., NICHOLLS, A.W., JOHNSON, B.A., et al. (2004). Use of metabonomics to identify impaired fatty acid metabolism as the mechanism of a drug-induced toxicity. Chem Res Toxicol 17, 165-173.
NAGASAKA, H., INOUE, I., INUI, A., KOMATSU, H., SOGO, T., MURAYAMA, K., et al. (2006). Relationship between oxidative stress and antioxidant systems in the liver of patients with Wilson disease: hepatic manifestation in Wilson disease as a consequence of augmented oxidative stress. Pediatr Res 60, 472-477.
NAIR, J., SONE, H., NAGAO, M., BARBIN, A., and BARTSCH, H. (1996). Copper-dependent formation of miscoding etheno-DNA adducts in the liver of Long
147
Evans cinnamon (LEC) rats developing hereditary hepatitis and hepatocellular carcinoma. Cancer Res 56, 1267-1271.
NAIR, J., STRAND, S., FRANK, N., KNAUFT, J., WESCH, H., GALLE, P.R., et al. (2005). Apoptosis and age-dependant induction of nuclear and mitochondrial etheno-DNA adducts in Long-Evans Cinnamon (LEC) rats: enhanced DNA damage by dietary curcumin upon copper accumulation. Carcinogenesis 26, 1307-1315.
NAKATSURA, T., YOSHITAKE, Y., SENJU, S., MONJI, M., KOMORI, H., MOTOMURA, Y., et al. (2003). Glypican-3, overexpressed specifically in human hepatocellular carcinoma, is a novel tumor marker. Biochem Biophys Res Commun 306, 16-25.
NICHOLSON, J.K., CONNELLY, J., LINDON, J.C., and HOLMES, E. (2002). Metabonomics: a platform for studying drug toxicity and gene function. Nat Rev Drug Discov 1, 153-161.
OKADA, K., WANGPOENGTRAKUL, C., OSAWA, T., TOYOKUNI, S., TANAKA, K., and UCHIDA, K. (1999). 4-Hydroxy-2-nonenal-mediated impairment of intracellular proteolysis during oxidative stress. Identification of proteasomes as target molecules. J Biol Chem 274, 23787-23793.
OLINSKI, R., GACKOWSKI, D., ROZALSKI, R., FOKSINSKI, M., and BIALKOWSKI, K. (2003). Oxidative DNA damage in cancer patients: a cause or a consequence of the disease development? Mutat Res 531, 177-190.
PAOLICCHI, A., MARCHI, S., PETRUCCELLI, S., CIANCIA, E., MALVALDI, G., and POMPELLA, A. (2005). Gamma-glutamyltransferase in fine-needle liver biopsies of subjects with chronic hepatitis C. J Viral Hepat 12, 269-273.
PEIRO, G., ALARY, J., CRAVEDI, J.P., RATHAHAO, E., STEGHENS, J.P., and GUERAUD, F. (2005). Dihydroxynonene mercapturic acid, a urinary metabolite of 4-hydroxynonenal, as a biomarker of lipid peroxidation. Biofactors 24, 89-96.
PEREZ-CARREON, J.I., LOPEZ-GARCIA, C., FATTEL-FAZENDA, S., ARCE-POPOCA, E., ALEMAN-LAZARINI, L., HERNANDEZ-GARCIA, S., et al. (2006). Gene expression profile related to the progression of preneoplastic nodules toward hepatocellular carcinoma in rats. Neoplasia 8, 373-383.
PEREZ-ENCISO, M., and TENENHAUS, M. (2003). Prediction of clinical outcome with microarray data: a partial least squares discriminant analysis (PLS-DA) approach. Hum Genet 112, 581-592.
PFEIL, S.A., and LYNN, D.J. (1999). Wilson's disease: copper unfettered. J Clin Gastroenterol 29, 22-31.
PITOT, H.C., DRAGAN, Y.P., TEEGUARDEN, J., HSIA, S., and CAMPBELL, H. (1996). Quantitation of multistage carcinogenesis in rat liver. Toxicol Pathol 24, 119-128.
PLASTARAS, J.P., RIGGINS, J.N., OTTENEDER, M., and MARNETT, L.J. (2000). Reactivity and mutagenicity of endogenous DNA oxopropenylating agents: base propenals, malondialdehyde, and N(epsilon)-oxopropenyllysine. Chem Res Toxicol 13, 1235-1242.
POON, T.C., and JOHNSON, P.J. (2001). Proteome analysis and its impact on the discovery of serological tumor markers. Clin Chim Acta 313, 231-239.
PRADELLES, P., GRASSI, J., and MACLOUF, J. (1985). Enzyme immunoassays of eicosanoids using acetylcholine esterase as label: an alternative to radioimmunoassay. Anal Chem 57, 1170-1173.
148
PRATICO, D., ROKACH, J., LAWSON, J., and FITZGERALD, G.A. (2004). F2-isoprostanes as indices of lipid peroxidation in inflammatory diseases. Chem Phys Lipids 128, 165-171.
REED, S.H., and GILLETTE, T.G. (2007). Nucleotide excision repair and the ubiquitin proteasome pathway--do all roads lead to Rome? DNA Repair (Amst) 6, 149-156.
RUTENBURG, A.M., KIM, H., FISCHBEIN, J.W., HANKER, J.S., WASSERKRUG, H.L., and SELIGMAN, A.M. (1969). Histochemical and ultrastructural demonstration of gamma-glutamyl transpeptidase activity. J Histochem Cytochem 17, 517-526.
SAMUELE, A., MANGIAGALLI, A., ARMENTERO, M.T., FANCELLU, R., BAZZINI, E., VAIRETTI, M., et al. (2005). Oxidative stress and pro-apoptotic conditions in a rodent model of Wilson's disease. Biochim Biophys Acta 1741, 325-330.
SANINS, S.M., NICHOLSON, J.K., ELCOMBE, C., and TIMBRELL, J.A. (1990). Hepatotoxin-induced hypertaurinuria: a proton NMR study. Arch Toxicol 64, 407-411.
SAWAKI, M., ENOMOTO, K., TAKAHASHI, H., NAKAJIMA, Y., and MORI, M. (1990). Phenotype of preneoplastic and neoplastic liver lesions during spontaneous liver carcinogenesis of LEC rats. Carcinogenesis 11, 1857-1861.
SCANDALIOS, J.G. (2005). Oxidative stress: molecular perception and transduction of signals triggering antioxidant gene defenses. Braz J Med Biol Res 38, 995-1014.
SCHAFER, F.Q., and BUETTNER, G.R. (2001). Redox environment of the cell as viewed through the redox state of the glutathione disulfide/glutathione couple. Free Radic Biol Med 30, 1191-1212.
SEIDEGARD, J., and EKSTROM, G. (1997). The role of human glutathione transferases and epoxide hydrolases in the metabolism of xenobiotics. Environ Health Perspect 105 Suppl 4, 791-799.
SEITZ, H.K., and STICKEL, F. (2006). Risk factors and mechanisms of hepatocarcinogenesis with special emphasis on alcohol and oxidative stress. Biol Chem 387, 349-360.
SETCHELL, K.D., HEUBI, J.E., BOVE, K.E., O'CONNELL, N.C., BREWSAUGH, T., STEINBERG, S.J., et al. (2003). Liver disease caused by failure to racemize trihydroxycholestanoic acid: gene mutation and effect of bile acid therapy. Gastroenterology 124, 217-232.
SHELINE, C.T., and CHOI, D.W. (2004). Cu2+ toxicity inhibition of mitochondrial dehydrogenases in vitro and in vivo. Ann Neurol 55, 645-653.
SHIMADA, H., TAKAHASHI, M., SHIMADA, A., OKAWARA, T., YASUTAKE, A., IMAMURA, Y., et al. (2005). Protection from spontaneous hepatocellular damage by N-benzyl-d-glucamine dithiocarbamate in Long-Evans Cinnamon rats, an animal model of Wilson's disease. Toxicol Appl Pharmacol 202, 59-67.
SPAHN, C.M., LEHRACH, H., and JUNGBLUT, P.R. (2004). A European focus on proteomics. Genome Biol 5, 322.
STRANGE, K. (2005). The end of "naive reductionism": rise of systems biology or renaissance of physiology? Am J Physiol Cell Physiol 288, C968-974.
SUGIYAMA, T., TAKEICHI, N., KOBAYASHI, H., YOSHIDA, M.C., SASAKI, M., and TANIGUCHI, N. (1988). Metabolic predisposition of a novel mutant (LEC rats) to hereditary hepatitis and hepatoma: alterations of the drug metabolizing enzymes. Carcinogenesis 9, 1569-1572.
149
TANITO, M., AGBAGA, M.P., and ANDERSON, R.E. (2007). Upregulation of thioredoxin system via Nrf2-antioxidant responsive element pathway in adaptive-retinal neuroprotection in vivo and in vitro. Free Radic Biol Med 42, 1838-1850.
TANZI, R.E., PETRUKHIN, K., CHERNOV, I., PELLEQUER, J.L., WASCO, W., ROSS, B., et al. (1993). The Wilson disease gene is a copper transporting ATPase with homology to the Menkes disease gene. Nat Genet 5, 344-350.
TEAHAN, O., GAMBLE, S., HOLMES, E., WAXMAN, J., NICHOLSON, J.K., BEVAN, C., et al. (2006). Impact of analytical bias in metabonomic studies of human blood serum and plasma. Anal Chem 78, 4307-4318.
TEW, K.D. (1994). Glutathione-associated enzymes in anticancer drug resistance. Cancer Res 54, 4313-4320.
TEW, K.D. (2007). Redox in redux: Emergent roles for glutathione S-transferase P (GSTP) in regulation of cell signaling and S-glutathionylation. Biochem Pharmacol 73, 1257-1269.
TOGASHI, Y., LI, Y., KANG, J.H., TAKEICHI, N., FUJIOKA, Y., NAGASHIMA, K., et al. (1992). D-penicillamine prevents the development of hepatitis in Long-Evans Cinnamon rats with abnormal copper metabolism. Hepatology 15, 82-87.
TRUEBA, G.P., SANCHEZ, G.M., and GIULIANI, A. (2004). Oxygen free radical and antioxidant defense mechanism in cancer. Front Biosci 9, 2029-2044.
TRYGG, J. (2002). Everything you need to know about Orthogonal Signal Correction (OSC) filters- and how they can improve interpretation of your data. In (homepage of chemometrics http://www.acc.umu.se/~tnkjtg/chemometrics/editorial/march2002.pdf).
TUDEK, B. (2007). Base excision repair modulation as a risk factor for human cancers. Mol Aspects Med 28, 258-275.
TUDEK, B., SWOBODA, M., KOWALCZYK, P., and OLINSKI, R. (2006). Modulation of oxidative DNA damage repair by the diet, inflammation and neoplastic transformation. J Physiol Pharmacol 57 Suppl 7, 33-49.
TUTTLE, S.W., MAITY, A., OPRYSKO, P.R., KACHUR, A.V., AYENE, I.S., BIAGLOW, J.E., et al. (2007). Detection of reactive oxygen species via endogenous oxidative pentose phosphate cycle activity in response to oxygen concentration: Implications for HIF-1alpha stability under moderate hypoxia. J Biol Chem.
USAMI, H., KUSANO, Y., KUMAGAI, T., OSADA, S., ITOH, K., KOBAYASHI, A., et al. (2005). Selective induction of the tumor marker glutathione S-transferase P1 by proteasome inhibitors. J Biol Chem 280, 25267-25276.
VALAFAR, F. (2002). Pattern recognition techniques in microarray data analysis: a survey. Ann N Y Acad Sci 980, 41-64.
VALKO, M., RHODES, C.J., MONCOL, J., IZAKOVIC, M., and MAZUR, M. (2006). Free radicals, metals and antioxidants in oxidative stress-induced cancer. Chem Biol Interact 160, 1-40.
VENDEMIALE, G., GRATTAGLIANO, I., PORTINCASA, P., SERVIDDIO, G., PALASCIAMO, G., and ALTOMARE, E. (2001). Oxidative stress in symptom-free HCV carriers: relation with ALT flare-up. Eur J Clin Invest 31, 54-63.
WANG, T., and WEINMAN, S.A. (2006). Causes and consequences of mitochondrial reactive oxygen species generation in hepatitis C. J Gastroenterol Hepatol 21 Suppl 3, S34-37.
150
WANG, Z., CIABATTONI, G., CREMINON, C., LAWSON, J., FITZGERALD, G.A., PATRONO, C., et al. (1995). Immunological characterization of urinary 8-epi-prostaglandin F2 alpha excretion in man. J Pharmacol Exp Ther 275, 94-100.
WERNER, T. (2007). Regulatory networks: linking microarray data to systems biology. Mech Ageing Dev 128, 168-172.
WU, G., FANG, Y.Z., YANG, S., LUPTON, J.R., and TURNER, N.D. (2004). Glutathione metabolism and its implications for health. J Nutr 134, 489-492.
WU, J., FORBES, J.R., CHEN, H.S., and COX, D.W. (1994). The LEC rat has a deletion in the copper transporting ATPase gene homologous to the Wilson disease gene. Nat Genet 7, 541-545.
XIE, Y., and VARSHAVSKY, A. (2001). RPN4 is a ligand, substrate, and transcriptional regulator of the 26S proteasome: a negative feedback circuit. Proc Natl Acad Sci U S A 98, 3056-3061.
XU, D., MAKKINJE, A., and KYRIAKIS, J.M. (2005). Gene 33 is an endogenous inhibitor of epidermal growth factor (EGF) receptor signaling and mediates dexamethasone-induced suppression of EGF function. J Biol Chem 280, 2924-2933.
YAMADA, J., TOMIYAMA, H., YAMBE, M., KOJI, Y., MOTOBE, K., SHIINA, K., et al. (2006). Elevated serum levels of alanine aminotransferase and gamma glutamyltransferase are markers of inflammation and oxidative stress independent of the metabolic syndrome. Atherosclerosis 189, 198-205.
YAMAGUCHI, Y., HEINY, M.E., SHIMIZU, N., AOKI, T., and GITLIN, J.D. (1994). Expression of the Wilson disease gene is deficient in the Long-Evans Cinnamon rat. Biochem J 301 ( Pt 1), 1-4.
YAMAMOTO, H., WATANABE, T., MIZUNO, H., ENDO, K., HOSOKAWA, T., KAZUSAKA, A., et al. (2001). In vivo evidence for accelerated generation of hydroxyl radicals in liver of Long-Evans Cinnamon (LEC) rats with acute hepatitis. Free Radic Biol Med 30, 547-554.
YANG, Y., SHARMA, R., SHARMA, A., AWASTHI, S., and AWASTHI, Y.C. (2003). Lipid peroxidation and cell cycle signaling: 4-hydroxynonenal, a key molecule in stress mediated signaling. Acta Biochim Pol 50, 319-336.
YASUDA, J., EGUCHI, H., FUJIWARA, N., OOKAWARA, T., KOJIMA, S., YAMAGUCHI, Y., et al. (2006). Reactive oxygen species modify oligosaccharides of glycoproteins in vivo: a study of a spontaneous acute hepatitis model rat (LEC rat). Biochem Biophys Res Commun 342, 127-134.
YOUNG, I.S., and MCENENY, J. (2001). Lipoprotein oxidation and atherosclerosis. Biochem Soc Trans 29, 358-362.
ZARKOVIC, N. (2003). 4-hydroxynonenal as a bioactive marker of pathophysiological processes. Mol Aspects Med 24, 281-291.
ZARKOVIC, N., ZARKOVIC, K., SCHAUR, R.J., STOLC, S., SCHLAG, G., REDL, H., et al. (1999). 4-Hydroxynonenal as a second messenger of free radicals and growth modifying factor. Life Sci 65, 1901-1904.
ZHANG, B., SCHMOYER, D., KIROV, S., and SNODDY, J. (2004). GOTree Machine (GOTM): a web-based platform for interpreting sets of interesting genes using Gene Ontology hierarchies. BMC Bioinformatics 5, 16.
151
G. ANNEXES