nuggoo - unifi...explanation of bromobenzene-mediated hepatoxicity 0 50 100 150 200 250 cvc bb 262:...
TRANSCRIPT
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOUn Oslo
Un Munich
Un Florence
Un Balearic Illes
Un Cork
Trinity
Un. Ulster
Rowett
Un Newcastle
Un Reading
IFR DiFE
Un Krakow
Inserm Marseille
TNO
Un Wageningen
Un Maastricht
EBI
NuNuGOGO
Un Lund
RikiltRivm
Examples of pathway analysis in nutrigenomics and toxicogenomics studies
Rob Stierum, [email protected]
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
Opening up a few nutrition and
toxicology-related journals for pathway
content…..
Opening up a few nutrition and
toxicology-related journals for pathway
content…..
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Pathways in nutrition. Glycolysis
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOBritish Journal of Nutrition, (1947) 1, 245-253
still… life seemed quite simple in the pre-genomics era…
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO…mechanism of azo-dye induced liver tumours…
British Journal of Nutrition, (1947) 1, 245-253
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO …worldwide…
Japanese Journal of Nutrition (1950-1952)
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO ..more complex…Mannose metabolism: Am. J. Clin. Nut. (1971), 24, 488-498
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO …modeling CYP gene expression
Toxicol. Appl. Pharmacol. (1998)
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Now, examples from the omics era
Establishing pathways in nutrition and toxicology from transcriptome, proteome and metabolomedatasets
Several approaches:•‘manual’ : use your brain; current knowledge•tools: pathway mapping and biological network tools•statistical networks biological inference
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOCase I ‘use your brain’ + tool
Discovery of functional genomics-derivednew biomarkers to assess the hepatotoxiceffects at the mechanistic level ofcombined exposure to food additives.
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
thiabendazole
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOAdministration of additives through the diet for 28 days, male Sprague-Dawley rats (n=6)
thiabendazole (TB)(102-5077 ppm)
control
Clinical signs, clinical chemistry and liver histopathology
Total RNA isolation from liver
Data analysis•Criteria for selecting genes:
•maximum of one missing value per dose response curve, •maximum CV of difference between the dye swap of 0.5 (dye swaps should behave in similar direction•at least 1.5 fold difference between any of the doses/compound
Confirmation of microarray data findings for selected genes at the biochemical level: •BHT and TB: RT PCR for CYP2B1 and CYP2B1/2; 7-ethoxyresorufin O-deethylase (CYP1A) and 7-
pentoxyresorufin O-depentylase (CYP2B) activity.•CC: palmitoyl-CoA oxidation as marker for peroxisome proliferation•TB: cellular tumor antigen p53 ELISA•All additives: western blotting for CYP1A2 and CYP2B1/2
cDNA array
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Gene expression profiles
3 x Control3 x Low
3 x Mid3 x High
3 x Control3 x Low
3 x Mid3 x High
5 x Control5 x Low
5 x Mid5 x High
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Applied bioinformatics versus manual
‘Manual’ network creation, concordance with a bioinformatics approach performed two years after
Network building around p53 protein, using MetacoreTM software from GeneGO, comparison with manually created network from array results from thiabendazoleexperiment
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOSystems ReconstructionTM Technology
GeneGoGeneGo, Inc, Inc
Systems Biology for Drug Discovery
www.genego.com
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
Agilent Affymetrix Proteomic SAGE
Concurrent visualization of different data types and experiments
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Build new network from MetacoreTM
• Around p53 protein• Making us of biological dB• Filtered to reduce complexity:
– for ‘rat ortholog’– for ‘transcriptional regulation’– for ‘liver’
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
Filtering needed to reduce complexity
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Case II ‘use your brain’
• Transcriptome, proteome and metabolome changes in rats exposed to bromobenzene
• Based on work by Wilbert Heijne:– Wilbert H.M. Heijne, Rob H. Stierum, Monique Slijper, Peter J. van Bladeren and Ben van Ommen (2003)
Toxicogenomics of bromobenzene hepatotoxicity: a combined transcriptomics and proteomics approach. Biochemical Pharmacology, 2003 Mar 1;65(5):857-75.
– Heijne WH, Slitt AL, Van Bladeren PJ, Groten JP, Klaassen CD, Stierum RH, Van Ommen B. (2004) Bromobenzene-induced hepatotoxicity at the transcriptome level. Toxicol Sci. 2004 Jun;79(2):411-422.
– Wilbert H.M. Heijne et al. Metabolome changes in bromobenzene treated rat liver, accepted
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOBr Br
O
BrOH
Br
O
Br
OH
Br
OH
BB BB 2,3 oxide2-Bromophenol
P450
P450
BB 3,4 oxide 3-Bromophenol
4-Bromophenol
P450
3,4 Dihydroxybromobenzene
+ GSSG
other bromophenolsand bromoquinones
Br
OHOH
Br
OHOH
GS
Br
OHOH
Br
OO
P450
4-Bromo-ortho-benzoquinone
GSH + GST
2 GSH
CYP450
CYP450
CYP450
CYP450 GST
Epoxide Hydrolase
Biotransformation of bromobenzene in rat liver
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOBromobenzene Rat studies:
II. Oral dosing of rats n=3 • Low (0.5 mmol/kg) • Mid (2 mmol/kg) • High (5 mmol/kg)• Untreated and solvent controls• Liver isolation at 6, 24 and 48 hrs.• Transcriptomics and metabolomics
I. Intraperitoneal dosing of rats • High (5 mmol/kg), untreated and solvent controls• Liver isolation at 24 hrs.• Transcriptomics and proteomics
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOBB upregulated (oral)
6 hours 24 hours 48 hoursDose (mmol/kg BW)
0.01.02.03.04.05.06.0
Electrophile response element (EpRe; ARE)
GST AGamma-GCSFerritinHeme oxygenasePeroxiredoxin1……
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO4 4.5 5 5.5 6 6.5 7
94
67
43
30
20.1
14.4
pI
Mass
364: 4-hydroxyphenyl pyruvic
acid dioxygenase (N)
3436
112: ATP synthase β subunit (M)
157: HSP60 (M)
308: carbamoyl-phosphate synthase(N)
358: 4-hydroxyphenyl pyruvicacid
dioxygenase (M)and HMG-CoA synthase (N)
277 287
420422434
611
687: aldehydedehydrogenase 2, mitochondrial (M)
697: carbamoyl-phosphate synthase(N)
698
821
846883
925 926 934: D-Dopachrome tautomerase(M; N)
981: aryl sulfotransferase (M; N)
991
Proteomics results IP
Change in the expression of individual rat liver proteins upon exposure to bromobenzene--> additional explanation of bromobenzene-mediated hepatoxicity
0
50
100
150
200
250
C VC BB
262:glutathione synthase
NS
0100200300400500600700800900
100 0
UT CO BB
687:aldehyde dehydrogenase 2,
mitochondrialb***
112:ATP synthase beta subunit.
0200400
600800
100 0
120 0140 0160 0
180 0
UT CO BB
b***
157:HSP60 protein
0
100
200
300
400
500
600
700
800
UT CO BB
b***
bromobenzene
Lipid peroxidation
4-hydroxy-2-nonenal
Detoxified by aldehydedehydrogenases
The mitochondrial, class 2,aldehyde dehydrogenase
has the highest affinity for aldehydes
Reactive intermediates, e.g. quinones
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
-0.2 -0.1 0 0.1 0.2PC1(22.67%)
-0.3
-0.2
-0.1
0
0.1
0.2
PC2(12.93%)
PCA Score plot PC1 vs PC2 urine 6 hours
bbua02abbua02bbbua02c
bbua04a
bbua04bbbua04c
bbua06a
bbua06bbbua06c
bbub20a
bbub20b
bbub20c
bbub22abbub22b
bbub22cbbub24a
bbub24b
bbub24c
bbuc38abbuc38bbbuc38c
bbuc40a
bbuc40bbbuc40c
bbuc42a
bbuc42bbbuc42c
bbud56a
bbud56b
bbud56c
bbud58abbud58b
bbud58c
bbud60abbud60b
bbud60c
bbue74abbue74b
bbue74c
bbue76abbue76bbbue76c
bbue78abbue78b
bbue78c
Winlin V1.11
Control
Vehicle
Low Dose
Medium Dose
High Dose
Metabolomics (oral)
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO
: mRNA
: protein
: (Putative)metabolite
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Case 3 ‘mapping tool’
Time- and dose-dependent effects of curcumin on gene expression in human colon cancer cells.
Van Erk MJ, Teuling E, Staal YC, Huybers S, Van Bladeren PJ, Aarts JM, Van Ommen . Time- and dose-dependent effects of curcumin on gene expression in human colon cancer cells. J Carcinog. 2004 May 12;3(1):8.
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOEffects of curcumin on gene expression in human colon cancer cells
• Curcumin incubation of colorectal cancer cell line HT29• Time course experiment: 3h, 6h, 12h, 24h, 48h• Two datasets: low dose and high dose• Select genes from expression data set present in at least 3 or
more time points• Use GenBank accession numbers to find up-to-date Unigene
Clusters• Final dataset: ~2600 genes for both low-dose and high-dose
dataset
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOEffect of curcumin in colon cancer cells:effect on cell cycle parameters
Changes in cell cycle distribution in response to curcumin:G2/M phase arrest• decrease in % of cells in G1 phase, • increase in % of cells in G2/M phase
24h
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
sub-G1 G1 S G2/M
% o
f tot
al c
ell p
opul
atio
n
*
*
**
* *
3h
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
sub-G1 G1 S G2/M
% o
f tot
al c
ell p
opul
atio
n
*
*
*
*
controllower concentrationhigher concentration
6h
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
sub-G1 G1 S G2/M
% o
f tot
al c
ell p
opul
atio
n
*
**
*
Van Erk et al., 2004
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOMicroarray data: Effect on cell cycle genes
p16-INK4 (CDKN2A)
-0.20
0.00
0.20
0.40
0.60
0.80
1.00
3h 6h 12h 24h 48h
expr
essi
on ra
tio (2
log)
low high
PCNA
-0.80
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
3h 6h 12h 24h 48h
expr
essi
on ra
tio (2
log)
low high
polo-like kinase
-2.00
-1.50
-1.00
-0.50
0.00
0.50
3h 6h 12h 24h 48h
expr
essi
on ra
tio (2
log)
low high
histone deacetylase 1
-0.80-0.70-0.60-0.50-0.40-0.30-0.20-0.100.000.100.20
3h 6h 12h 24h 48h
expr
essi
on ra
tio (2
log)
low high
Van Erk et al., 2004
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO 24h high
early low
24h high
early high
early low+high12h+24h high
12h low
12h high
48h high
12h low
early low+high
24h high
24h high
12h low
early low
early low+high
12h high
12h high
24h high
early low
24h high
Visualize changes in ‘cell cycle genes’ in pathway using GenMAPP
Lipoprotein Metabolism
Atherosclerotic ChangesIn Mature Mice (25 wks)
Lipoprotein
ApoE3 Transgenic Mouse Model
• Over-expressed human mutant apolipoprotein gene• Apolipoprotein changes prevent binding to liver receptors• Clearance of plasma lipoproteins inhibited• Mice are pre-atherosclerotic at 9 weeks…
Case 4: statistical networks biologicalinference: Nutritional systems biology on ApoE3 Leiden Model
Methodology• Specimens: 10 male ApoE3 Leiden /10 male wild type controls
(WT)• Normal chow diets/environmentally similar• Sacrificed at 9 weeks• Samples analyzed:
CompleteNANAUrine
CompleteCompleteNAPlasma
CompleteCompleteCompleteLiver
MetabolomicAnalysis
ProteomicAnalysis
GenomicAnalysis
E+067.518
8:00 16:00 24:00 32:00 40:000
20
40
60
80
100
m/z:450>950
E+067.518
8:00 16:00 24:00 32:00 40:000
20
40
60
80
100
m/z:450>950
time (min)
Inte
nsity
• Mass Chromatograms of Plasma Lipids
Wildtype
APOE3
m/z:450>950
E+066.221
8:00 16:00 24:00 32:00 40:000
20
40
60
80
100 E+066.221
8:00 16:00 24:00 32:00 40:000
20
40
60
80
100
Inte
nsity
Metabolite Analysis- ApoE3 vs. WT: LC-MS Plasma Lipid Profiles
• Mouse Plasma• Metabolite (lipid)• LC-MS• IMPRESSTM algorithm
• Mouse Plasma• Metabolite (lipid)• LC-MS• IMPRESSTM algorithm
ApoE3
Wildtype
Lyso-PC’s
triglycerides
0 200 400 600 800 1000 1200 1400 1600
-1
-0.5
0
0.5
1
369.00
569.00
623.00
823.00847.00848.00
849.00850.00
851.00855.00
871.00872.00
873.00874.00
875.00876.00 877.00
878.00879.00
880.00900.00
901.00902.00
903.00904.00
905.00906.00
918.00919.00
926.00928.00
929.00931.00
932.00945.00946.00
948.00950.00
951.00952.00
953.00967.00972.00
975.00995.00
996.00
1040.001041.00
1062.001063.00
1064.001065.00 1086.00
1087.001088.00
1536.001537.00
1668.00
1706.001707.00 1711.00
1732.001733.00
1734.001735.00
1736.001737.00
1738.001739.00
m/z
Mor
eab
unda
ntIn
Apo
E3Le
ssab
unda
ntIn
Apo
E3
triglycerides
Lyso-phosphatidylcholine
Metabolite Analysis- ApoE3 vs. WT: Plasma Lipid
Difference Factor Spectrum
• Mouse Plasma• Metabolite (lipid)• LC-MS• IMPRESSTM algorithm• PARCTM pattern recognition
• Mouse Plasma• Metabolite (lipid)• LC-MS• IMPRESSTM algorithm• PARCTM pattern recognition
• Unexpected Metabolite Information
Relative Variance• Unsupervised clustering reveals segregation• Hundreds of components contribute to observed differences
can be used as powerful, selective and specific biomarkers
Rel
ativ
e Va
rianc
e
Proteomic Analysis- ApoE3 vs. WT: Principal Component and
Discriminant Analysis
• Mouse Plasma• Protein• ESI-Ion Trap• IMPRESSTM algorithm• PARCTM pattern recognition
• Mouse Plasma• Protein• ESI-Ion Trap• IMPRESSTM algorithm• PARCTM pattern recognition
500 1000 1500 2000
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Regression
Factor Spectrum
465.00485.00
535.00
622.00652.00
662.00
750.00
811.00
911.00926.00
992.00
1041.001050.00
1208.001216.00
1217.001243.00
1267.001268.00
1269.001298.00
1299.00
1340.001342.00
1583.00
m/z
Fraction 2
HigherIn
ApoE3
Higherin
WT
• Mouse Plasma• Protein• ESI-Ion Trap• IMPRESSTM algorithm• PARCTM pattern recognition
• Mouse Plasma• Protein• ESI-Ion Trap• IMPRESSTM algorithm• PARCTM pattern recognition
Proteomic Analysis- ApoE3 vs. WT: Difference Factor Spectrum
Peptides Exhibiting Significant Differences
1366.00
Identifyusing
MS-MS
Beyond Nutrition 40
Mor
eab
unda
ntIn
Apo
E3Le
ssab
unda
ntIn
Apo
E3
• Mouse Plasma and Liver• Protein + Metabolite + mRNA• IMPRESSTM algorithm• PARCTM/WinlinTM pattern recognition
• Mouse Plasma and Liver• Protein + Metabolite + mRNA• IMPRESSTM algorithm• PARCTM/WinlinTM pattern recognition
peptidesmetabolites
genes
Each bar represents a measured componentand up/down-regulation
Proteomic, Genomic, Metabolomic Data: Identifying / Mining Differences- ApoE3 vs. WT: Difference Spectrum
M-7: Diacylglycerol (DAG)
G-693:Fatty Acid Binding Gene
P-1059: Fatty Acid Binding Protein
G-8147: Apolipoprotein A1
M-9: LysoPC
BioSystematics™- Mapping Known Relations onto Non-Linear PCA Correlations
Down-regulatedUp-regulated
Non-Linear Kernal PCAFeature Correlation (ρ > 0.8)
Protein Kinase Cepsilon
Cell Signaling
PCA Correlation
Text / PubMed MiningRegulatory Factors
Fatty Acid Metabolism
Apoptosis Inhibitory 6 Gene
Apoptosis
RNP H1 GeneTranscription Splicing
Ubiquitin Conj E3A GeneProteosome
Translation Init FactorProtein Synthesis
PLA2 VII
PPARα
Phosphatidylcholine Lipid Metabolism Pathway Model Suggested by Results
• New insights in targets/biomarkers in the ApoE3 system
transmembraneprotein
small molecule
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGO Acknowledgements• TNO, NL:
– Dr. Marjan van Erk: curcumin in vitro studies– Ing. Marinus Dansen: array quality control– Prof. Dr. Louis Havekes and Prof. Dr. Jan van de Greef + colleagues from Beyond
Genomics Medicine, Waltham, MA, USA: APOeE3 systems biology– Prof. Dr. John Groten: food additives project– Dr. Wilbert Heijne: bromobenzene studies – Dr. Robert Jan Lamers and Dr. Joop van Nesselrooij: metabolomics of bromobenzene– Dr. Ben van Ommen: coordinator of The European Nutrigenomics Organisation, group
leader Genomics
• BIBRA INTERNATIONAL LTD, UK– Prof. Dr. Brian Lake, Dr. Clive Meredith + colleagues thiabendazole project
• EUROPEAN BIOINFORMATICS INSTITUTE, HINXTON, UK– Dr. Susanna Assunta Sansone and Dr. Phillipe Rocca-Serra: support with
ArrayExpress
• UNIVERSITY OF UTRECHT, NL– Dr. Frank Holstege group: microarrays fabrication– Prof Dr. Albert Heck group: BB proteomics mass spectrometry
• FOOD STANDARDS AGENCY, UK financial support (project number T01021)
the European Nutrigenomics Organisation
NuNuGOGONuNuGOGOPathways and networks assisted me in getting here!
NederlandseSpoorwegen(Static, delay!)
Meridiana Airlines(Dynamic)
Taxi in the ancient city of Florence(Conserved)