COMPUTATIONAL MODELLING OF CANCER

Download Report

Transcript COMPUTATIONAL MODELLING OF CANCER

SYSTEMS MODELLING OF CANCER
Leto Kyritsi
Talk Outline
1.
2.
3.
4.
5.
6.
Cancer.
Systems approach to biology.
Cancer from a systems perspective.
Modelling biological systems.
Modelling cancer.
References.
May 2005
ECCO Group (VUB)
2
Cancer 1
Uncontrolled cellular proliferation.
Cancer cell arises out of a possible 1014 cellular targets.
“At least 1 in 3 people will develop cancer - 1 in 4 men & 1
in 5 women will die from it” (Franks, 1998).
Visible tumour: end result of complex series of events.
Gradual process: 5-6 (up to 12) control mechanisms
deregulated.
Accumulation of mutations in 3 types of genes:
1. Oncogenes (↑ cell division).
2. Tumour suppressor genes (↓ cell division).
3. DNA repair genes.
May 2005
ECCO Group (VUB)
3
Cancer 2
Hereditary & environmental component:
-Spontaneous mutations, physical mutagens, chemical
mutagens, viruses…
-Mutagenesis: direct or indirect (DNA repair errors).
-Or, mutations can be inherited (familial cancers).
Changes are fixed through replication.
Initiation: initial change may persist in latent form.
“Promoting agents” induce initiated cells to divide.
Outcome = balance between growth-inhibiting factors
and extent of changes in initiated cells.
May 2005
ECCO Group (VUB)
4
The Eukaryotic Cell Cycle
Cancer: a cell cycle disease.
Passage through different
phases determined at
“checkpoints”, where
integration of inputs of
intrinsic data or growth
signals reaching the exterior
of the cell, determines what
happens next.
http://www.bioteach.ubc.ca/CellBiology
May 2005
ECCO Group (VUB)
Mediated by sequential
activation of key proteins.
5
Cell Signalling Pathways
“Cell signalling or ‘signal transduction’ is the study of the
mechanisms by which biological information is
transferred between and within cells”
(Systems biologist Olaf Wolkenhauer).
May 2005
ECCO Group (VUB)
6
Cancer Signalling Pathways 1
1. Tumour suppressor pathways.
RB (retinoblastoma):
-Regulated by CyclinD-CDK4/6 & CyclinE-CDK2 complexes.
-Represses transcriptional activation of genes controlled by E2Fs
(transcription factors that regulate cell cycle progression genes).
P53:
-A transcription factor that activates apoptosis.
-Blocks G0 to G1 transition by activating p21 transcription (p21
inhibits a number of cyclin-CDK complexes).
-Monitors genomic instability.
TGF- β (Transforming Growth Factor - β):
-Activate SMAD transcription factors who regulate large number of
cell cycle regulators.
-Dual role: can promote invasion and metastasis.
May 2005
ECCO Group (VUB)
7
Cancer Signalling Pathways 2
2. Oncogenic pathways.
RAS.
MYC (c-, N-, L-, B-).
WNT- Frizzled.
3. Anti-apoptotic pathways.
PI3K.
4. Angiogenesis pathways.
FGF-2 (fibroblast growth factor 2).
VEGF (vascular endothelial growth factor).
5. Metastasis pathways
TGF.
Hepatocyte Growth Factor.
Loss of adhesion molecules.
May 2005
ECCO Group (VUB)
8
Cancer Signalling Pathways 3
6. Telomere end maintenance.
TERT (protein component of telomerase enzyme).
ALT (alternative lengthening in absence of telomerase).
7. Mobilisation of resources.
-Activation of metabolic programmes that confer specific
advantages to the cancer cell.
-E.g. differentiation-associated antigens, enzymes involved in
nutrient metabolism and enzymes that regulate oxidative potential.
8. Tissue-specific pathways.
9. Immune surveillance.
TNF-α, TNF-β from macrophages: dual role.
May 2005
ECCO Group (VUB)
9
Cancer Signalling “Subway Map”
http://www.nature.com/nrc/journal/v2/n5/weinberg_poster/index.html
May 2005
ECCO Group (VUB)
10
The Cancer Challenge
Very complex disease to address.
Metastasis via circulatory system is the main problem in
cancer treatment  the “macroscopic” component.
Phenotypic heterogeneity (both inter- and intra-tumour,
Struikmans 1999, Less 2002) esp. with respect to
treatment response  the “sub-cellular”/”cellular”
component.
Amazing genetic diversity among cancer cells, so much so
that some cancer cells "might fairly be called new species"
(Gibbs, 2003).
May 2005
ECCO Group (VUB)
11
Systems Approach To Biology 1
“…If we break up a living organism by isolating its different
parts, it is only for the sake of ease in analysis and by no
means in order to conceive them separately. Indeed
when we wish to ascribe to a physiological quality its
value and true significance, we must always refer it to
this whole and draw our final conclusions only in relation
to its effects on the whole.”
(Physiologist Claude Bernard, 1865)
May 2005
ECCO Group (VUB)
12
Systems Approach To Biology 2
Integration of data (informatics perspective) or dynamic
interactions?
SB is about methodologies rather than tools and
technologies: “modelling process itself more important
than the model” (Wolkenhauer).
Hypothesis- or Data-driven?
Distinction between “-omics” family approaches for data
integration and fusion…
…and data-based modelling and simulation (system
identification).
May 2005
ECCO Group (VUB)
13
Systems Approach To Biology 3
Human genes: initially believed ~100,000, now known to be
20,000.
Mechanisms of gene regulation:
1. Transcriptional
2. Post-transcriptional
3. Translational
4. Post-translational
Context-specific: “genetic information turns out not to be
physical at all, and to be subject to contextual modulations”
(G. Kampis).
Genome plays data-like, software-like, hardware-like roles
(Hofstadter).
May 2005
ECCO Group (VUB)
14
Life’s Complexity Pyramid
Oltvai et.al. (2002).
Large-scale structures emerge from low-level interactions.
•Koestler (1967): “holon”
•Jacob (1974): “integron”
..
Organism
Specificity
Principle
Universality
FUNCTIONAL
MODULES
REGULATORY MOTIFS
MOLECULES
May 2005
ECCO Group (VUB)
15
Cancer From A Systems Perspective 1
Acquired tumour “robustness” (Kitano, 2004).
Enabled by:
-Heterogeneous functional redundancy of the tumour mass.
-Intracellular (mutation-associated) feedback loops, inherent cell
cycle robustness, environment-associated feedback loops.
Many anticancer drugs increase heterogeneity by causing drugresistant mutations!
Emergence: “no deductive causality”.
Bhalla et al (1999): signalling networks exhibit emergent
properties (e.g. self-sustaining feedback loops that result in
bistable behavior with discrete steady states).
Cancer, which results from signalling network deregulation, can
be studied as an emergent system.
May 2005
ECCO Group (VUB)
16
Cancer From A Systems Perspective 2
Homeostasis: the automatic maintenance of constant
conditions in open systems by counteracting influences
tending toward disequilibrium.
“… an examination of the self-righting methods employed in
the more complex living beings may offer hints for improving
and perfecting the methods which still operate inefficiently
and unsatisfactorily” (Walter B. Cannon, ”Wisdom of the
Body”, 1932).
Complexity pyramid: “lower” levels often get “sacrificed” for
sake of “higher” levels (e.g. apoptosis).
May 2005
ECCO Group (VUB)
17
“Wisdom Of The Body”
G. Zajicek (Professor of Experimental Medicine and Cancer
Research, Hebrew University of Jerusalem ).
WOB metaphor: attribute of all living systems, “movable
equilibrium point”.
WOB decides on the balance between disease-driving and
homeostatic, health-maintaining mechanisms.
Consequences: biomedical, conceptual/philosophical.
May 2005
ECCO Group (VUB)
18
Cancer From A Systems Perspective 3
Tumour cachexia: cause rather than effect?
“Systemic disease with local manifestations” (Deighton
1975).
Normal development / injury: precise mechanisms allow
organs to reach fixed size.
Contact inhibition (homeostatic mechanism) but…
Cells are acting as a system, in parallel.
Single instruction multiple data stream.
Distinction between what happens (e.g. loss of
homeostasis) and how it happens (pathway deregulation
etc).
“Egg or chicken”?
May 2005
ECCO Group (VUB)
19
ORGANISMAL
CELLULAR
MOLECULAR
May 2005
ECCO Group (VUB)
20
Key Features of Biological Systems 1
Openness
Complexity
Non-linearity
Order
Integration
Hierarchy
Adaptiveness
May 2005
ECCO Group (VUB)
21
Key Features of Biological Systems 2
Self-organisation
Dissipativeness
Information-richness
Autopoiesis
Emergence
Organisation
Teleonomy
May 2005
ECCO Group (VUB)
22
Modelling Biological Systems
Model: idealised representation of one thing in terms of
something else.
Target object always of lower complexity than source object.
R. Rosen: “A measure of the complexity of a system is the
number of models required to understand its behaviour”.
R. Levins: “No model can be simultaneously optimised for
generality, precision and realism”.
N. Dioguardi:
-“Every system in nature can be expressed in a formal
language”.
- Hepatone: the liver abstracted as a fractal object.
May 2005
ECCO Group (VUB)
23
The “Six Challenges” (Wolkenhauer, 2002)
How to:
1. Simultaneously capture dynamic regulation and spatial
organisation.
2. Capture intra- / inter-cellular actions and interactions.
3. Cross organisational levels.
4. Integrate experimental levels (genome, transcriptome,
proteome, metabolome, physiome).
5. Combine data analysis and data management.
6. Relate formal representations, provide conceptual
frameworks / theoretical foundations to the previous five
points.
May 2005
ECCO Group (VUB)
24
Modelling Cancer
Flow of information / decision making.
Integration of different hierarchical levels:
-Timing.
-Topology.
-Precedence.
Cancer cell as self-organising system.
Cancer dynamics:
-Benign  In Situ  Malignant
-Discrete cancer states as system attractors.
Flow rather than states.
-Critical transition thresholds.
The cancer knowledge domain (ontology).
May 2005
ECCO Group (VUB)
25
Modelling: Subcellular / Cellular Level
Control/regulatory systems within cancer cell:
-Cancer signalling pathways.
-Cancer signalling networks.
Action of mutagens on specific sites.
Immune response.
Endocrine response.
Angiogenesis.
Adhesion mechanisms & metastasis.
Action of therapeutic agents at molecular level.
May 2005
ECCO Group (VUB)
26
Modelling: Macroscopic Level
Cell-to-cell interactions and behaviour.
Tumour dynamics:
-Tumour growth.
-Metastasis.
-Angiogenesis.
-Response to various treatments.
Immune response to cancer:
- Effect on metastasis: “double-edged sword”.
Tumour is a heterogeneous system  distributed
parameters.
May 2005
ECCO Group (VUB)
27
Questions / Issues
Is the “Computer Metaphor” adequate?
Oyama: [A computer program] has features […] for deciding
outcomes just because a computer lacks the biological
structure and dynamics of an organism.
G. Kampis:
-Living systems able to evolve & re-define their state space.
-Representation problem.
-External vs. internal programming: mimicking vs. evolution.
-Not simply manipulating interfaces to deal with complexity (“if
you can define it, we can model it”).
May 2005
ECCO Group (VUB)
28
How To Model Cancer In Silico?
Choosing the scale: cancer cell / tumour / organism.
Set of functions describing the relations between
different variables and adjacent levels.
Predictive model: deterministic or stochastic?
Signalling pathways as networks of multivariable
Boolean switches.
Cellular Automata (CA) – type machine models.
Agent-based models (Individual-Based Models).
Weighted inputs into Neural Network  training.
Bi-directional data flow  Recurrent Network.
May 2005
ECCO Group (VUB)
29
Creating A Cancer Model - Benefits
“Interactome”.
Turning data into animated system behaviour.
Different models can be built & tested against
experimental / clinical data.
Predictive models and novel therapeutics: e.g. cell cycle
phase-specific chemotherapy (Gardner, 2002).
Introduction of personalised treatments.
Mapping lower-level states to higher-level states.
Understanding control & decision mechanisms in cancer
and living systems in general.
Integration of research domains.
May 2005
ECCO Group (VUB)
30
Trends
Since 1960’s:
-Models of solid tumours (collections of cells feeding from
nutrient supply).
Since 1980’s:
-Models of angiogenesis/metastasis.
Since 1990’s:
-Specific, data-oriented models.
-Interconnected cell cycle signal transduction pathways
across spatial & temporal scales of organisation.
-Fast increasing!
May 2005
ECCO Group (VUB)
31
Macroscopic Models
Predictive epidemiological models from screening & clinical data.
Phenomenological modelling of tissue & tumour growth aspects.
Regression prediction based on tissue architecture:
http://www.bccrc.ca/ci/tm01_results2.html
3D simulation of tumour response to treatment (radiotherapy
targeting).
Fractal morphometry applied to tumours
-E.g. fractal tumour boundary a prognostic tool:
http://www.europhysicsnews.com/full/09/article1/
Greece :
-”In Silico Oncology” Group: www.in-silico-oncology.iccs.ntua.gr/
-Simulation models of the evolution of malignant tumours in vitro
& in vivo and their response to various therapeutic modalities &
mapping to genetic status.
May 2005
ECCO Group (VUB)
32
Cellular Models
Continuous dynamics:
-Kinetic models of cell population growth (differential equations).
-Topological models of cell population growth emphasising chaotic
behaviour.
CA-based models of cell population growth.
Zajicek: http://www.what-is-cancer.com/papers/contents/
Extensive simulation of homeostatic mechanisms in stem cells.
CA model of stem cell division: the “Proliferon” (Starlogo).
CA model of immune system & hypersensitivity to chemotherapy:
www.imbm.org/PDF/chap12.pdf
Cancer detection via determination of fractal cell dimension:
www.pa.msu.edu/~bauer/cancer/cancer.pdf
May 2005
ECCO Group (VUB)
33
Subcellular Models
Time-course experiments on mRNA abundance & protein activity.
Mapping mutational profiles to therapeutic combinations.
Tyson & Novak (1996):
http://www.euchromatin.org/Ciliberto01.htm
Eukaryotic cell cycle checkpoint controls modelled in yeast using
partial differential equations (XML).
Tyson & Novak (2002):
http://www.euchromatin.org/Ciliberto01.htm
Irreversible transitions in & out of cell cycle controlled by
hysteresis in control mechanisms – verified experimentally.
Increasing number of databases, modelling projects on genetic
regulation, biochemical pathways, signal transductions.
http://www.brc.dcs.gla.ac.uk/projects/bps/links.html
May 2005
ECCO Group (VUB)
34
Proteomic Co-expression of Genes
Warenius et.al. (2004).
Human cell lines (normal keratinocytes vs. 19 cancer lines)
May 2005
ECCO Group (VUB)
35
“Neostasis”
Warenius et.al. (2004).
Following immortalisation, neostasis is the stabilisation necessary
for continued cancer cell growth and survival.
May 2005
ECCO Group (VUB)
36
Random Graph Models Of Protein
Interactions
Warenius, Zito, Kyritsi (to be submitted).
Based on A.L. Barabasi’s work on Scale-Free (SF) Networks.
Plotted protein expression data as Connection or No
Connection (cut-off points: r>0.5, p<0.05).
Results:
-Cancer cells: P(k) decays following a power law (SF).
-Normal cells: no SF properties.
Redundancy and attack survivability.
“Hubs” may be “Critical Normal Gene” products suitable for
therapeutic targeting?
More work to be done.
May 2005
ECCO Group (VUB)
37
Modelling The Knowledge Domain
R. Paton:
-Metaphors in scientific thinking.
-Modelling: representation of one thing in terms of another.
-So, language of models is metaphorical by definition.
-Pluralistic approach  richness of knowledge.
-Displacement of ideas across knowledge domains.
-Recursive relations between systemic metaphors in the
biosciences:
E.g. machine-as-text ↔ text-as-machine
organism-as-machine ↔ machine-as-organism.
Category Theory: formal tool / language / conceptual
framework for study of structures & systems of structures.
May 2005
ECCO Group (VUB)
38
Modelling The Knowledge Domain
CELL CYCLE
CHECKPOINT
PHYSIOLOGICAL
PROCESS
is_a
is_a
is_a
is_a
CELL CYCLE
REGULATION
CELLULAR
PROCESS
CELL PHYSIOLOGICAL
PROCESS
BIOLOGICAL
PROCESS
is_a
is_a
part_of
is_a
CELL CYCLE
part_of
CELL PROLIFERATION
May 2005
is_a
CELL GROWTH /
MAINTENANCE
ECCO Group (VUB)
ALL ALL
Gene
Ontology
Consortium
39
Systems Biology Modelling Tools
CellML: http://www.cellml.org/
OSS XML-based standard for storage & exchange of computerbased mathematical models.
BioPAX: http://www.biopax.org/
OSS OWL-based ontology for metabolic pathway & molecular data.
SBML (Systems Biology Markup Language): http://sbml.org/
OSS XML-based ontology for models of biochemical reaction
networks.
SBW (Systems Biology Workbench): http://64.17.162.114/research/
Free C++ framework for communication & cooperation of application
components (written in diverse languages).
SCIpath: http://www.ucl.ac.uk/oncology/MicroCore/
OSS Java-based suite of programs for microarray data analysis.
May 2005
ECCO Group (VUB)
40
Summary
Main focus: flow of information & control.
Integration of different hierarchical levels.
Right assumptions.
Well-defined parameters.
Hypothesis- or Data-driven approach?
Which side (realism, precision, generality) to “sacrifice” in
modelling?
-E.g. may need to narrow down to one type of cancer to
avoid “idealised” descriptions.
Pluralistic models: an interdisciplinary approach.
May 2005
ECCO Group (VUB)
41
References 1
-Albert R. et.al. (2000). Error and attack tolerance of complex networks.
Nature 406: 378 – 382.
-Barabasi A.L., Oltvai Z.N. (2004). Network biology: understanding the cell’s
functional organization. Nature Reviews Genetics 5: 101-113.
-Bhalla et.al. (1999). Emergent properties of networks of biological signalling
pathways. Science 15: 283(5400): 381-387.
-Camphausen K. et.al. (2001). Radiation therapy to a primary tumour
accelerates metastatic growth in mice. Cancer Res 61(5): 2207-11.
-Deighton K.J. (1975). Cancer – a systemic disease with local manifestations.
Med Hypotheses 1(2): 37-41.
-Dioguardi N. (1992). Fegato a piu dimensioni. Etaslibri-RCS Medicina,
Milano.
-Kitano H. (2004). Cancer as a robust system. Nature Reviews Cancer 4:
227-235.
-Oltvai Z.N., Barabasi A.L. (2002). Systems biology: life’s complexity pyramid.
Science 298(5594): 763-4.
May 2005
ECCO Group (VUB)
42
References 2
-Paton R. et.al.(1998). Information processing in computational tissues. In:
Information processing in cells and tissues, Plenum Press, New York.
-Struikmans et.al. (1999). Regional heterogeneity and intra-observer
variability of DNA-content and cell proliferation markers determined by
flow cytometry in head and neck tumours. Oral Oncol. 35: 217-213.
-Tyson J.J. et.al. (1996). Chemical kinetic theory: understanding cell-cycle
regulation. Tr Biochem Sci 21(3): 89-96.
-Warenius H. et.al. (2002). Are Critical Gene Products in cancer cells the
real therapeutic targets? Anticancer Res. 22(5): 2651-5.
-Warenius H. et.al. (to be submitted). Heterogeneity of therapeutic response
in human cancer cell lines: relationship to RNA and protein profiles.
-Wolkenhauer O. (2002). Simulating what cannot be simulated. Dagstuhl
Position Statement.
-Zajicek G. (1994). Wisdom of the Body. Cancer J. 7: 212-213.
-Zimmerman L.E. et.al. (1978). Does enucleation of the eye containing a
malignant melanoma prevent or accelerate the dissemination of tumour
cells? Br J Ophthalmol 62(6): 420-5.
May 2005
ECCO Group (VUB)
43