S Y S T E M S B I O L O G Y

ver the last decade, the areas ofbioinformatics and genomics haveidentified the nuts and bolts of themachinery that make up a living cell.Scientists are now embarking on anendevour to discover how the manycomponents interact to create thecomplex behaviour that underlies development and disease.

Systems Biology

A principal challenge for the life sciencesis to understand the “organisation” and“dynamics” of those components thatmake up a living system, i. e., to investi-gate the spatio-temporal relationshipsbetween the elements that give rise tocause and effect in living systems. A ma-jor problem is that networks of cellularprocesses are regulated through com-plex interactions among a large numberof genes, proteins, and other molecules.From these considerations, the funda-mental goal of systems biology is to un-derstand the nature of this regulation in

orderto

gain greater insight into the mechanismthat determine the functions of genesand cells, and ultimately their conse-quences at physiological or phenotypiclevel. In systems biology, this is achievedthrough mathematical modelling andsimulation. The emergence of systemsbiology signals a shift of focus away frommolecular characterisation of the com-ponents in the cell to an understandingof functional activity through temporalchanges and dynamic interactions incells. Pathways (i. e., regulated networksof biochemical reactions) are the concep-tual framework of molecular and cell bi-ology, employed to describe inter- and in-tra-cellular dynamics.

Cell Signalling

Fig. 1 provides a (dramatic) simplifi-cation of the dynamic processes involvedin cells signalling. Signal transductionpathways commonly consist of a largenumber cascaded modules between re-ceptor and genome. There can be nu-merous intermediate steps before the

O

Systems BiologyTowards an Understanding of the Dynamics of Life

Fig. 1: Cell signaling (signal transduction): The expression of information in the genome can be af-fected by extracellular signals. Receptors spanning the cell membrane receive signals (by binding with ligands) and transmit the information to activate intracellular proteins (the response regulator),which can lead to changes in the gene expression programme. In the figure, the response regulatorbinds to the operator region of a gene and prevents the RNA polymerase from reading the adjacentgene. Disruptions or errors of the signal transfer can have catastrophic consequences and may lead to diseases, including cancer.

Bioforum Europe 1/2004, S. 50-52, GIT VERLAG GmbH & Co. KG, Darmstadt, Germany, www.gitverlag.com/bioint

S Y S T E M S B I O L O G Y

signal transduction process ends, oftenwith a change in the gene expressionprogram of the cell.

In addition to crosstalk between pathways, negative feedback loops are akey regulatory mechanism. Mathemati-cal modeling and simulation in this fieldhas the purpose to help and guide thebiologist in detecting feedback loops,nonlinear (and thus counter-intuitive) in-teractions. It further helps in designingexperiments, suggesting which variablesshould be measured and why. Appli-cations of dynamic pathway modeling(i.e. the simulation of dynamic interac-tions among proteins) are in drug targetidentification and validation.

Dynamic Pathway Modelling and Simulation

We argued that it is dynamic interactionsof gene products and other molecules,not the genome in itself, that gives rise to biological function, regulation andcontrol. For our experimental designsthis means that instead of trying to iden-tify genes as causal agents for somechange, function or phenotype, weshould relate these observations to se-quences of events.

Conducting time course experimentsthat investigate signal transfer and tran-sient behaviour form the basis for dy-namic pathway modelling. The aim isthen to identify the direction andstrength of relationships between vari-ables in a pathway. The fact that most ofthese relationships are nonlinear, andthat feedback connections exist, makesthis is a non-trivial problem. Nonlinear-ity implies a breakdown of the superposi-tion principle (i. e. the whole is morethan the sum of its parts). As a conse-quence, observations often do not matchan intuitive or expected response (seeFig. 3). Feedback loops provide a differ-ent challenge in that their existence is often a key aspect of the biological in-vestigation but mathematical formalismsto detect and quantify them are in shortsupply.

As an example, we consider theRas/Raf-1/MEK/ERK module that conveyssignals from the cell membrane to thenucleus. This kinase cascade is spatiallyorganized in a signalling complex nucle-ated by Ras proteins. The small G proteinRas is activated by many growth factorreceptors and binds to the Raf-1 kinasewith high affinity when activated. Thisinduces the recruitment of Raf-1 fromthe cytosol to the cell membrane. Acti-vated Raf-1 then phosphorylates and activates MAPK/ERK Kinase (MEK), a kinase that in turn phosphorylates and

activates Extracellular signal RegulatedKinase (ERK), the prototypic Mitogen-Ac-tivated Protein Kinase (MAPK). ActivatedERKs can translocate to the nucleus andregulate gene expression by the phos-phorylation of transcription factors. This kinase cascade controls the prolifer-ation and differentiation of different celltypes. The specific biological effects arecrucially dependent on the amplitudeand kinetics of ERK activity – hence theneed for mathematical modelling of thedynamics.

Walter Kolch from the Cancer Re-search UK, Beatson Laboratories in Glas-gow, identified a novel protein (RKIP)that affects the Ras/Raf-1/MEK/ERKmodule (see Fig. 2). Western blot timecourse experiments were conducted anda mathematical model constructed. Theentire pathway is thereby composed of

template modules that allow a graphrepresentation, and each of which is associated with a set of four nonlinearordinary differential equations. Once amodel structure is established and themodel parameters are determined, simu-lations of the mathematical model fordifferent initial concentrations of RKIPcan be conducted very easily as shown inFig. 3.

While, at present, we usually cannotgenerate complete and accurate data setsfor parameter estimation, the value ofsuch simulations leis largely in the gener-ation of hypotheses and in supporting ex-perimental design. Simulating differentmodel structures or conducting a sensi-tivity analysis to identify key variables ofa system can be done within minutes on acomputer, helping us to decide whichvariables to measure and why.

Fig. 2: Graphical representation of a mathematical model that consists of sixteen nonlinear ordinarydifferential equations with 17 parameters. The model investigates the influence RKIP has on theRas/Raf-1/MEK/ERK pathway. The figure is read from left to right: from growth factor stimulation ofthe receptors, recruitment of Raf-1 and binding to Ras, followed by the activation of Raf-1 and MEK,to the activation and translocation of ERK. These graphical representations are translated into mathe-matical equations. Subsequent simulation studies of this model (see Figure 3) help the experimentalscientist in formulating hyptheses and design his experiments.

Fig. 3: The two figures show the dynamic response of phosphorylated Ras (left) and double phosphory-lated ERK (right) as a function of increasing initial concentrations of RKIP. The plot on the left illus-trates a nonlinear relationship: as the initial concentration of RKIP increases, the concentration pro-files of Ras* appears to be lower than expected. However, as the initial concentration of RKIPincreases further, the level of Ras* increases again. The plot on the right confirms the experimentalfinding that increasing levels of RKIP determine the Raf-1 available and the subsequent strong sur-pression of ERK through high levels of RKIP.

References

For further information on the researchreferred to in this article, publications,detailed references, and activities of theresearch groups, see www.sbi.uni-rostock.de.

Olaf Wolkenhauer He holds the chair inSystems Biology &Bioinformatics at Rostock University sinceJuly 2003. Before moving to Rostock, heworked for over nineyears at UMIST in theU.K. The research groupcovers a wide spectrumof bioinformatics tech-niques that are basedon statistical dataanalysis and mathematical modelling; including sequence-, mass-spec-, microarray-, and proteomic-data analysis. Dynamic pathway modelling and sys-tems biology, as described in this article, are emergentareas of research that are a primary focus of thegroup.

Chair in Bioinformatics & Systems BiologyUniversity of RostockAlbert Einstein Str. 2118051 Rostock, Germanywolkenhauer@informatik.uni-rostock.dewww.sbi.uni-rostock.de

S Y S T E M S B I O L O G Y

Kwang-Hyun ChoHe received his B.S., M.S.,and Ph.D. in electrical en-gineering from the KoreaAdvanced Institute of Science and Technology(KAIST). He joined theSchool of Electrical Engi-neering at the Universityof Ulsan, Korea in 1999as an Assistant Professor.From 2002 to 2003, hewas a research fellow at the Control SystemsCentre, UMIST, U.K. His research interests cover the areasof systems science and control engineering including applications to complex biological systems. The focus has been on mathematical modeling and simulation ofgenetic- and signal transduction pathways, and DNA mi-croarray data analysis. The research has been supportedby a grant from the Korean Ministry of Science and Technology (Korean Systems Biology Research Grant,M10309000006-03B5000-00211).

School of Electrical EngineeringUniversity of Ulsan · Ulsan, 680-749, Koreackh@mail.ulsan.ac.kr · http://sys.ulsan.ac.kr

Walter KolchHe is a Senior GroupLeader at the BeatsonInstitute for Cancer Re-search and Professor forMolecular Cell Biologyand Director of the SirHenry Wellcome Func-tional Genomics Facilityat the University ofGlasgow. He is inter-ested in the molecularmechanisms of signaltransduction with a re-search focus on MAPK and the role of protein inter-actions in the regulation of signalling pathways. Acurrent aim is to understand signalling networks byusing proteomics, real time imaging and mathe-matical modeling.

Cancer Research UK Beatson Laboratories Garscube Estate, Switchback RoadBearsden, Glasgow G61 1BD, United Kingdomwww.beatson.gla.ac.uk

Box 1: Cell Signalling

Biological cells are not running a program, but rather continually sensing their environment and making

decisions on the basis of that information. To determine how cells act and interact within the context of the

organism, we need to understand how information is transferred between and within cells. Cell signaling, or

“signal transduction”, is the study of mechanisms by which this transfer of biological information comes about.

Signaling impinges on all aspects of biology, from development to disease. Many diseases, such as cancer,

involve malfunction of signal transduction pathways.

© GIT VERLAG

300

400

500

600

700

800

900

2004 2005 2010

The world market for RNAi reagentsmio. US $

Source: Research & Markets© GIT VERLAG

300

400

500

600

700

800

900

2004 2005 2010

The world market for RNAi-based drugsmio. US $

Source: Research & Markets

1 bio.

RNAi Technology