Advantages of Hiring UIUX Design Service Providers for Your Business
Can a combination of constrained-based and kinetic modeling bridge time scales in progressive diseases?
1. Lorentz workshop Multiscale Systems Biology of Cancer
Leiden, NL, Nov. 16, 2012
Natal van Riel
Dept. of Biomedical Engineering, n.a.w.v.riel@tue.nl
2. • Parameter Transition Analysis (PTA)
• Progressive adaptations in metabolism associated with
diseases (or therapeutic intervention)
• Linking phenotypes
• Integrate metabolome, proteome, transcriptome
• Indentifying and exploiting the structure of the parameter space
unidentifiability ↔ constraints paradox (?)
• Quantifying and analyzing uncertainty in model and data
/ biomedical engineering PAGE 219-8-2013
3. Systems biology and metabolic diseases
• Changes in metabolism associated with multi-factorial,
progressive diseases
/ biomedical engineering PAGE 319-8-2013
4. Kinetic modeling
• ATP metabolism in mitochondria
/ biomedical engineering PAGE 419-8-2013
Chance, Comput Biomed Res.1967;1:251-64.
Schmitz et al. PLoS ONE
2012, 7(3): e34118.
5. Kinetic modeling
/ biomedical engineering PAGE 519-8-2013
Chalhoub et al, 2007 AJP Endocrinol 93(6): E1676-
liver
skeletal muscle
Schmitz,et al. Am J Physiol
Cell Physiol 2012 Oct 31
7. • Inverse problems
• Numerical optimization
• Variability / Uncertainty analysis
/ biomedical engineering PAGE 719-8-2013
2
1
( )
( )
N
i i
d
i i
y d
X
σ=
−
∑
p
p
( )ˆ arg min ( )dX=
p
p p
( )
( ( ), , )
d t
t t
dt
=
s
Nv s p
=Nv 0subject to
i i ia v b< <
( )ˆ arg max ( )X=
v
v v
8. Metabolic Syndrome (MetS)
• The characteristics of plasma lipoprotein profiles codetermine
metabolic and cardiovascular disease risks
• Underlying molecular mechanisms are not fully understood
• Multi-factorial and progressive
/ biomedical engineering PAGE 819-8-2013
9. • Preclinical research
• Cohort studies: cross-sectional
e.g. BMI matched
• Patient-specific (VPH, ITFoM)
/ biomedical engineering PAGE 919-8-2013
10. / biomedical engineering PAGE 1019-8-2013
• Different diets
• Genetic manipulation
• Pharmacological compounds
…
experiments
phenotype A
experiments
phenotype B
Identify adaptations
11. Modulate lipoprotein metabolism
• Activate Liver X Receptor (nuclear receptor)
plays a central role in the control of cellular lipid
and sterol metabolism
• Metabolic profiling
/ biomedical engineering PAGE 1119-8-2013
0 10 20
0
100
200
Hepatic TG
Time [days]
[umol/g]
0 10 20
0
1
2
3
Hepatic CE
Time [days]
[umol/g]
0 10 20
0
2
4
6
Hepatic FC
Time [days]
[umol/g]
0 10 20
0
50
100
Hepatic TG
Time [days]
[umol]
0 10 20
0
0.5
1
1.5
Hepatic CE
Time [days]
[umol]
0 10 20
0
2
4
Hepatic FC
Time [days]
[umol]
0 10 20
0
1000
2000
3000
Plasma CE
Time [days]
[umol/L]
0 10 20
0
1000
2000
3000
HDL-CE
Time [days]
[umol/L]
0 10 20
0
500
1000
1500
Plasma TG
Time [days]
[umol/L]
0 10 20
6
8
10
12
VLDL clearance
Time [days]
[-]
0 10 20
100
200
300
400
ratio TG/CE
Time [days]
[-]
0 10 20
0
5
10
15
VLDL diameter
Time [days]
[nm]
0 10 20
0
1
2
3
VLDL-TG production
Time [days]
[umol/h]
0 10 20
1
2
3
Hepatic mass
Time [days]
[gram]
0 10 20
0
0.2
0.4
DNL
Time [days]
[-]
• Computational model
Grefhorst et al. Atherosclerosis, 2012, 222: 382– 389
Tiemann et al. BMC Systems Biology, 2011, 5:174
Control, 1, 2, 4, 7, 14, 21 days
12. Phenotype snapshots
• Observed:
• Unobserved:
• Metabolic network topology
is invariant
• Adaptations in metabolism:
- metabolic control
- interaction with proteome and transcriptome
Metabolic parameters (e.g. Vmax) can change
/ biomedical engineering PAGE 1219-8-2013
Metabolome
Proteome
Transcriptome
13. Parameter Trajectory Analysis (PTA)
• Algorithm: nesting of simulation and parameter estimation
/ biomedical engineering PAGE 1319-8-2013
Progressive disease /
Treatment intervention
Phenotype data at different
stages
Monte Carlo sampling of data
interpolants
Estimation of parameter and flux trajectories
Analysis
A priori information
Metabolic network topology
& Reaction kinetics
Differential Equation model
with time-dependent parameters
( )
( ( ), , )
d t
t t
dt
=
s
Nv s p
( )
( ( ) ,( ), )
d t
t t
dt
t=
s
Nv s p
2
1
( )
( )
N
i i
d
i i
y d
X
σ=
−
∑
p
p
( )
( )
ˆ( ) arg min ( ( )d
t
t X t=
p
p p
18. From adaptations in metabolome…
• … to predict changes in proteome / transcriptome
/ biomedical engineering PAGE 188/19/2013
T0901317
LXR
model development
predict changes
Metabolome
Proteome
Transcriptome
enzyme parameter gene/protein
HDL-CE synthesis ABCA1
HDL-CE uptake SR-B1
FC production ABCG5
… …
Fas, Abcg5, Abcg8, Cyp7a1, Lpl, Pltp, Cd36
19. Effects of LXR activation
• Increased cholesterol efflux from periphery to HDL particles
• Accumulation of hepatic TG (hepatic steatosis)
• Metabolic adaptation: decreased
hepatic capacity to clear cholesterol
• Predicts a decrease in SR-B1
/ biomedical engineering PAGE 1919-8-2013
• Experimental validation
Tiemann et al, submitted
20. The hepatic HDL-C uptake capacity is reduced
upon LXR activation
/ biomedical engineering PAGE 2019-8-2013
• The hepatic HDL-C uptake flux is increased (steatosis):
• increase in plasma HDL-C (metabolic control)
• transcriptional control (SR-B1) counteracts hepatic overloading
• Increase in plasma HDL-C is due to only a small imbalance in
uptake vs efflux in the beginning of the intervention
21. VLDL synthesis flux to plasma decreases
• VLDL-TG production is increased
• TG and CE content per VLDL particle increases 10 fold
/ biomedical engineering PAGE 2119-8-2013
22. Activation of LXR by pharmaceutical
compound T0901317
• Beneficial effect:
• increased excretion of cholesterol
from the body
• large, anti-atherogenic HDL
• Side effects:
• hepatic steatosis
• triglyceride-rich VLDL
• Model analysis predicts how side effects could be prevented
/ biomedical engineering PAGE 2219-8-2013
Liver section of mice
treated 4 days with LXR
agonist T0901317
Oil-Red-O staining for
neutral fat
hepatic steatosis
VLDL
HDL
T0901317
23. Is this possible with less data?
• A subset of the data (cross-sectional data)
• Only day 0 and day 4
/ biomedical engineering PAGE 2319-8-2013
…
24. Flux trajectories for acceptable parameter sets
/ biomedical engineering PAGE 2419-8-2013
[mM]
[mM/h]
4 days after LXR
activation
reference
25. Analysis of under-constrained trajectories
• Some show a clear pattern (positive correlation between HDL-CE
synthesis and HDL-CE uptake by the liver),
others just ‘clouds’ of solutions
• Can the ‘structure’ in one cross-section of the parameter space
be used to interpret other flux adaptations?
/ biomedical engineering PAGE 2519-8-2013
26. Predictions about changes in gene/ protein
expression
• Clustering of scenarios - testable hypotheses
• Also here SR-B1 is predicted to be decreased
• Measuring ABCG5, but especially ABCA1 is predicted to be
discriminative
/ biomedical engineering PAGE 2619-8-2013
fluxes
parameters
27. Trajectories can be used for many analyses
• Analysis of the cascade of induced adaptations
• Sensitivity and control analyses
• Optimal experiment design
which additional measurement (which metabolite, when) is most
effective in reducing the uncertainty in a prediction of interest and
tradeoff with the ‘cost’
• …
/ biomedical engineering PAGE 2719-8-2013
28. A theoretical study
• Insight in the approach
• Test and optimize computational methods
• Identify possibilities for further improvements
/ biomedical engineering PAGE 2819-8-2013
R1
u2
u1 1 S1
S3S2
S4
3
4 5
2
7
6
1 2 3 4 50
0.5
1
1.5
2
S1
1 2 3 4 50
0.1
0.2
0.3
0.4
0.5
S2
1 2 3 4 50
0.2
0.4
0.6
0.8
1
S3
1 2 3 4 50
0.2
0.4
0.6
0.8
S4
u2
u1 1 S1
S3S2
S4
3
4 5
2
Van Riel et al, submitted
29. Outlook: including other omics
• Transcriptomics
• Proteomics
• Trajectories correlating with
gene expression data are
more likely than parameter
changes that do not correlate
‘A transparent black box’
/ biomedical engineering PAGE 2919-8-2013
Metabolome
Proteome
Transcriptome2 2
1 1
( )
( )
N M
i i i
d
i ii i
y d d dt
dRNA dt
θ θ
χ θ λ
σ=
−
+
∑ ∑
30. Acknowledgement
Collaborators
• Computational Biology (TU/e)
• Ceylan Çölmekçi Öncü
• Christian Tiemann
• Joep Schmitz
• Joep Vanlier
• Huili Yuan
• Peter Hilbers
• Marijke Dermois
• Gijs Hendriks
• Fianne Sips
• Sandra van Tienhoven
• Robbin van den Eijnde
• Bram Wijnen
• Sjanneke Zwaan
Funding
• Netherlands Genomics Initiative
Netherlands Consortium
for Systems Biology
• AstraZeneca
• Univ. Medical Centre Groningen (NL)
• Aldo Grefhorst
• Maaike Oosterveer
• Jan Albert Kuivenhoven
• Barbara Bakker
• Bert Groen
• Biomedical NMR (TU/e)
• Klaas Nicolay
• Jeanine Prompers
• Ko Willems-van Dijk, Leiden University
Medical Center, Netherlands
• FP7-HEALTH.2012.2.1.2-2: Systems
medicine: Applying systems biology
approaches for understanding
multifactorial human diseases and their
co-morbidities, starting in 2013