Complex Systems Biology @ Case Status and a Vision for the Future

ComplexComplex SystemsSystems BiologyBiology SessionSession atat ICSBICSB 20062006 YokohomaYokohoma JapanJapan

Kwang-Hyun Cho Sree N. Sreenath Director, Systems Biology Lab Director, Case Complex Systems Biology Center College of Medicine & Bio-MAX Case Western Reserve University, Institute, Seoul National University, Cleveland OHIO, USA Korea

Case Complex ICSBICSBICSB 200620062006 Systems Biology YokohomaYokohomaYokohoma,,, JapanJapanJapan Center ComplexityComplexity andand SystemsSystems BiologyBiology

“What is the difference between a live cat and a dead one? One scientific answer is systems biology. A live cat is the emergent behavior of the system incorporating those parts.”

- Editorial, Nature, May 2005. What is Complex Systems Biology?

DEFINITION: Applying “Complex Systems” knowledge to understanding biological problems. ComplexComplexSystems Systems Systems Biology

Discovery of organizing principles Hierarchical systems, Bounded autonomy of levels, (Biological emergence, dynamics Uncertainty Principle) (chaos, equilibria), etc. Interaction Balance Principle: coordination of multi-level behavior Focus on “vertical” interaction (between levels) Goes beyond emergence Integration of data-driven modeling (e.g., Laufenberger's) with rigorous mathematical modeling (e.g., Cho, Doyle, Klipp, Tyson ,Wolkenhauer). Biological Processes Variance

A. In Time Illustration of how a signal is relayed Process time constants example(adapted from through sequential activation of proteins Fell D., Understanding the Control of in the Ras/Raf/MEK/REK pathway. Metabolism, Portland Press, 1997).

B. In Space Hierarchical, multi-level representation example

Tumor Tissue Faster Spatial Reaction size Cellular Time increase

Nucleus Other Intra-Cellular (R)Evolutionary View of Systems Biology

MODULARIZATION

Step 1 Step 2

* * * * * ** * * * * * * * ** * *

Biological Network Systems Biology System Dynamics Data and Connectivity, clustering, Dynamics - evolution in topology time, feedback, Knowledge Search for drug targets in homeostasis the entire network using (resistance to change) Complex Systems ad hoc methods such as /adaptation, etc. Biology deletion experiments and Search for therapeutics Paradigm of Organized randomized searches. (devices, drug targets, Complexity Bioinformatics etc.) guided by Search for organizing biological principles principles Chemical Engineering Search for therapeutics and Bioengineering (devices, drug targets, etc.) guided by organizing principles HistoryHistory ofof SystemsSystems BiologyBiology - According to Wolkenhauer, Kitano & Cho, IEEE Control Sys. Mag., pp 38-48,Aug 2003.

“Systems Biology” - Origin of Concept in Proc. Intl. Symposium at Case in 1968.

- Mihajlo Mesarovic Complex Systems Biology - ICSB, October 11, 2006 (9:15-10:30)

Session organizers: Peter Wellstead, Kwang-Hyun Cho and Sree N. Sreenath

Chairs: Kwang-Hyun Cho (Seoul National Univ.), Sree N. Sreenath (Case Western Univ.) Mihajlo D. Mesarovic (Case Western Reserve University) "Interaction Balance Coord as Org. Principle in Complex Systems Biology" Jack Donald Keene (Duke University Medical Center) "Coordination of Gene Expression by RNA Operons" Kenneth Alan Loparo (Case Western Reserve University) "Applications of Complex Systems Biology to the Study of Neural Systems " Interaction Balance Coordination as Organizing Principle in Complex Systems Biology

Mihajlo Mesarovic, Sree N. Sreenath and Girish Balakrishnan Complex Systems Biology Center Case Western Reserve University Cleveland, Ohio, U.S.A

Case Complex Systems Biology Center AGENDA

o Complex Systems Biology o Biological Complexity o Conceptual Framework oCoordination o Higher Level Regulatory Function o Motifs in Complex Systems Biology o Mathematical General Systems Theory Proof of Universality of IBCP Systems Biologies 1. Mathematical Systems Biology Dynamic Systems and Control Theory Motivation Homeostasis: Cannon Cybernetics: Control and Information in Human /Animal, Nobert Wiener 2. Computational Biology Data based modeling using computer algorithms “Systems biology is a scientific discipline that endeavors to quantify all of the molecular elements of a biological system to assess their interaction into graphical network models that serve as predictive hypothesis to explain emergent behavior.” [1] - Leroy Hood, 2005 Systems Biologies (contd.) Complex Systems Biology Multilevel, Hierarchical, Systems Theory (Mesarovic and Takahara, 1976 [2] ) Motivation “Connecting different levels-from molecules through module to organisms is essential for an understanding of biology.” [3] - Leyland Hartwell, 1999. “… epigenetic cell-cell and extracellular influences are pivotal in tumor progression beyond genetic mutation.” [4] -Judah Folkman, 2006. “Progress in biology benefits greatly from technology but ultimately the more fundamental changes will be conceptual.” [5] - Douglas Hanahan and Robert A. Weinberg, 2000 On Complex Systems Biology Complex Systems Paradigm • Complex system is a relation on systems (i.e., a system of systems) •Minimum two levels - Individual levels follow their own laws •Organizing principles govern System structure and function Complex Systems Biology •Objective: To advance understanding of biological complexity. •Through search for organizing principles (our research) •Organizing principles in biology are akin to the natural laws in physics. • Complexity - function of regulatory multilevel, hierarchical, organization. •Coordination • Higher level regulatory/decision function. • Implemented by organizing principles •Such as Interaction Balance Coordination Principle. Biological Complexity

Paradigm of control – feedback or otherwise – is insufficient to address the problem of complexity in biology as demonstrated in eukaryotes •99% of the human proteins have recognizable equivalent in mice. • Relation between the amount of nonprotein-coding DNA sequences and organism complexity is more consistent [6]. • Protein coding genes are 2%< of human genome. •Noncoding genes (introns and transposons) • Perform regulation function governing organized complexity of eukaryotes [6]. UNICELLULAR LIFE, primarily prokaryotes, ruled the earth for billions of years. When multicellular life appeared, however, its complexity rose with dizzying speed. The evolution of an additional genetic regulatory system might explain both the jump to multicellularity and the rapid diversification into complexity. [6] Coordination: Higher Level Regulatory Function Coordination •An organizing principle in a complex system. •A higher level decision function. •Influence subsystems to achieve goal (desired behavior) of the overall system • Coordination is distinct from control. •Control: A “directive/order” to achieve certain objectives/goals in spite of variability. •Objective of coordination is not to control but to harmonize •Influence/motivate interacting subsystems (with distinct functions) to advances overall system objective •Modules have to be coordinated to produces required cell functions. Coordination (contd.)

Coordination is a generic concept •Similar to feedback (an organizing principle for homeostasis) •Feedback - all levels of the human body’s hierarchy, from genome to brain functions and the whole body. Perform. regulatory/control functions. •Implemented in different circumstances by specific coordination principles. •Coordination Principles • ”Laws" (logic) of coordination as implemented in specific situations. •Interaction Balance Coordination Coordination Principle (IBCP) •Has been proved to have the broadest application in: •Environmental studies, Management, Global issues, Engineering, etc Appears to be promising for biology. METHODOLOGY Hierarchical Systems Theory Interaction Balance Coordination Principle (IBCP) In a complex system, S(Δ) functioning of a S0(Δ0) subsystem Coordination depends on the interactions with ε1 ε2 other subsystems. c1 c2

u1d + + u2d IBCP then states: The overall system is S1(Δ) coordinated/- u21a S2(Δ) harmonized if the Control Control actual and desired u12a interactions for each and every subsystem y y1 2 are balanced, i.e., Process Process uija -uijd ≤ εj x1 x2

x1, x2: inputs u12a, u21a: actual interaction y1, y2: outputs u12d, u21d: desired interaction ε1, ε2: imbalance c1, c2: coordination action Interaction Balance Coordination Principle, IBCP

Objective of coordination •Influence subsystems In a perturbed state for overall system state to be near original quiescent state but within a harmonious domain of the overall systems behavior. •In quiescent state: interactions are considered as balanced. •In perturbed state: objective of coordination by IBCP is to influence the subsystems so that interaction balance is restored as required under newly perturbed conditions; •i.e. “desired” and actual interactions are balanced. •IBCP can be implemented by either • endogenous adaptation of the subsystems •a distributed process. METHODOLOGY Hierarchical Systems Theory

Coordination In a multilevel, hierarchical system the task of the higher-level is not to control but to coordinate/harmonize the functioning of the lower subsystems which are pursuing their own objectives so as to meet the requirement of the overall system. Robustness Resources The overall system objective Blue dots indicate states has been violated and a shift Coordination of subsystems; the overall is implemented by the coordination system objective (blue circle) by shifting blue circle to red circle. is being satisfied. Coordination The subsystem has received a new is indicated by blue lines. coordination intervention (red lines). Performance Fraility

Control Control

Dynamic Process Dynamic Process Coordination Motifs in Complex Systems Biology IBP-based Coordination Motifs: Three Categories

Coordination by (a) Subsystems Endogenous Actions e.g. Angiogenesis

Coordination by a (b) Distributed Process e.g. in Immune System ……..

(c) Coordination by an Identifiable Coordination Unit Subsystem …….. e.g. on Organ Level Angiogenesis: Instantiation of Interaction Balance Coordination Principle

Is Angiogenesis An Organizing Principle? - Judah Folkman [4]

For a set of angiogenesis-dependent diseases : 1) Oncogene-dependence • The control of metastasis by a primary tumor. 2) A new area of platelet biology • Relation of hemangiomas to placenta. Coordination of E.Coli Metabolic Network for Cell’s Growth

(A) (B)

(A) Change in E.Coli metabolic network ,e.g. by knockout of some reactions triggers coordination process resulting in a new network topology. Topologies within the Bounded Autonomy of levels result in the cell’s growth to remain within 5% of the optimal magnitude.

(B) Re-rounting of metabolic fluxes in a metabolic mutant. Growth-maximizing flux for a wild-type metabolic genotype and in the zwf mutant. This mutation (indicated by the crossed arrow) eliminates the glucose-6-phosphate dehydrogenase reaction that leads into the pentose phosphate shunt. Predicted biomass yield in this mutant is only one percent lower than that in the wild type. [8] Facilitated Variation Theory of Evolution Marc W. Krischner and John C.Gerhart (A) (B) (C)

PhR – Physiological Range M.Genome – Mutated Gene Conserved core processes are constituents in the organism’s evolution. Their function is to generate the phenotype from the genotype. In normal state, core processes delineate physiological range and adaptive potential to environmental changes, which are not sufficient to initiate evolutionary response and adaptive potential (A). When environmental changes reach sufficient intensity for an evolutionary change, corresponding physiological range and adaptive potential do not coincide with the normal, which triggers coordination of core processes to a corresponding new range. Mutation ensues to stabilize new organization of core processes, (B), which in turn generate new phenotype (C). COORDINATION MECHANISM Signaling Pathway Crosstalk

Interaction Balance Coordination Principle

Coordinating Process

ε ε1 2 c1 c2

+ + i1d i2d Subsystem 1 i1a Subsystem 2

Pathway 1 Control Control Pathway 2 i2a

y y1 2 Process Process u u1 2

Subsystem interaction (pathway crosstalk)

u1, u2 inputs i1a, i2a actual interaction y1, y2 outputs u1d, u2d desired interaction ε1, ε2 imbalance c1, c2 coordination action IL3IL3--inducedinduced Jak2Jak2--STAT5STAT5 CoordinationCoordination (joint work with Bunting/Qu lab) Objective: To understand the constitutive activation of IL3 - induced JAK2/STAT5 mechanism implicated in hematopoiesis.

Interleukin-3 (IL3) • A growth factor acting on committed hematopoietic progenitor cells • IL3 depletion causes apoptosis. • Activates a cascade of signaling pathways: PI3K/Akt, Erk and STAT5. • Aberrant activation of these pathways causes abnormal hematopoiesis leading to various blood disorders including leukemia. • STAT5 remains latent in cytoplasm in the absence of IL3 stimulus. • In normal cells, STAT5* is transient, and indicates a tight negative pathway regulation (i.e. SHP2, SOCS). • Constitutively active STAT5 is observed in erythroleukemia, and acute and chronic myelocytic leukemia [3]. PathwayPathway Model:Model: ModularizationModularization • Biochemical reactions based model (Yamada, 2003) – Modifiction: Direct interaction between SHP2 and STAT5

• Nonlinear Ordinary Differential Equations (ODE) – x ∈ ℜ49, y ∈ ℜ2, k ∈ ℜ121 BlockBlock DiagramDiagram [SOCS1-(IL3J)2], [SOCS1-(IL3J)2-SHP2], [SOCS1-(IL3RJ)2*-STAT5c-SHP2], [SOCS1], [SOCS1- (IL3RJ)2*-SHP2], [SOCS1-(IL3RJ2)-STAT5c], [SOCS1-(IL3RJ)2*-STAT5c], [SOCS1-(IL3RJ2)*], [SOCS1-(IL3RJ)2], [SOCS1-(IL3RJ)2-STAT5c-SHP2], [SOCS1-(IL3RJ)2-SHP2]

Output 2 Total STAT5c* [STAT5n] [STAT5c], [STAT5c*], [IL3J2*-STAT5c*], [IL3RJ2*-STAT5c], [IL3RJ2-STAT5c] Output 1 STAT [(STAT5c*)2] Nucleus [(STAT5n*)2]

[(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*2], [mRNAn] [(IL3RJ)2*-STAT5c] [IL3RJ2*-STAT5c], Input [(IL3RJ)2], [IL3RJ2-STAT5c], SOCS [STAT5c] Receptor [(IL3RJ]2*] IL3 JAK

[SHP2], [IL3RJ2*-STAT5c-SHP2], [IL3RJ2*-SHP2], [IL3RJ2-SHP2], [IL3RJ2-STAT5c-SHP2] [(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*], [STAT5c*], [STAT5c], [(IL3RJ)2-STAT5c], [(IL3RJ)2*-STAT5c]

SHP2

[SOCS1-(IL3RJ)2*-STAT5c], [SOCS1-(IL3RJ)2*-STAT5c-SHP2], [SOCS1- (IL3RJ)2-STAT5c], [SOCS-(IL3RJ)2*-SHP2], [SOCS1-(IL3RJ)2*], [SOCS1- (IL3RJ)2], [SOCS1-IL3RJ2-STAT5c-SHP2], [SOCS1-(IL3RJ)2-SHP2]

[SHP2], [(IL3RJ)2*-SHP2], [(IL3RJ)2-SHP2]

[SOCS1], [SOCS1-(IL3RJ)2*], [SOCS1-(IL3RJ)2] ModelModel CalibrationCalibration

• Estimate parameters using semi-quantitative data (Chen, 2004)

•• NominalNominal parameterparameter simulationsimulation ––STAT5 STAT5 activationactivation isis transienttransient SimulationSimulation -- SHP2SHP2 KnockdownKnockdown

[SOCS1-(IL3J)2], [SOCS1-(IL3J)2-SHP2], [SOCS1-(IL3RJ)2*-STAT5c-SHP2], [SOCS1], [SOCS1- (IL3RJ)2*-SHP2], [SOCS1-(IL3RJ2)-STAT5c], [SOCS1-(IL3RJ)2*-STAT5c], [SOCS1-(IL3RJ2)*], [SOCS1-(IL3RJ)2], [SOCS1-(IL3RJ)2-STAT5c-SHP2], [SOCS1-(IL3RJ)2-SHP2]

Output 2 Total STAT5c* [STAT5n] [STAT5c], [STAT5c*], [IL3J2*-STAT5c*], [IL3RJ2*-STAT5c], [IL3RJ2-STAT5c] Output 1 STAT [(STAT5c*)2] Nucleus [(STAT5n*)2]

[(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*2], [mRNAn] [(IL3RJ)2*-STAT5c] [IL3RJ2*-STAT5c], [ IL3RJ2- Input [(IL3RJ)2], STAT5c], [STAT5c] SOCS Receptor [(IL3RJ]2*] IL3 JAK

[SHP2], [ IL3RJ2*-STAT5c-SHP2], [IL3RJ2*-SHP2], [ IL3RJ2-SHP2], [IL3RJ2-STAT5c-SHP2] [(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*], [STAT5c], [(IL3RJ)2-STAT5c], [(IL3RJ)2*-STAT5c]

SHP2

[SOCS1-(IL3RJ)2*-STAT5c], [SOCS1-( IL3RJ)2*-STAT5c-SHP2], [SOCS1- (IL3RJ)2-STAT5c], [SOCS-( IL3RJ)2*-SHP2], [SOCS1-(IL3RJ)2*], [SOCS1- (IL3RJ)2], [SOCS1- IL3RJ2-STAT5c-SHP2], [SOCS1-( IL3RJ)2-SHP2] •• SHP2SHP2 knockdownknockdown causescauses [SHP2], [( IL3RJ)2*-SHP2 ], [(IL3RJ)2-SHP2] constitutiveconstitutive activationactivation

[SOCS1], [SOCS1-( IL3RJ)2*], [SOCS1-( IL3RJ)2] SimulationSimulation -- SOCSSOCS KnockdownKnockdown

Output 2 Total STAT5c* [STAT5n] [STAT5c], [STAT5c*], [IL3J2*-STAT5c*], [IL3RJ2*-STAT5c], [IL3RJ2-STAT5c] Output 1 STAT [(STAT5c*)2] Nucleus [(STAT5n*)2]

[(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*2], [mRNAn] [(IL3RJ)2*-STAT5c] [IL3RJ2*-STAT5c], [ IL3RJ2- Input [(IL3RJ)2], STAT5c], [STAT5c] SOCS Receptor [(IL3RJ]2*] IL3 JAK

[SHP2], [ IL3RJ2*-STAT5c-SHP2], [IL3RJ2*-SHP2], [ IL3RJ2-SHP2], [IL3RJ2-STAT5c-SHP2] [(IL3RJ)2-STAT5c], [STAT5c], [STAT5c*], [STAT5c], [(IL3RJ)2-STAT5c], [(IL3RJ)2*-STAT5c]

SHP2 [SOCS1-(IL3RJ)2*-STAT5c], [SOCS1-( IL3RJ)2*-STAT5c-SHP2], [SOCS1- •• STAT5STAT5 activationactivation isis (IL3RJ)2-STAT5c], [SOCS-( IL3RJ)2*-SHP2], [SOCS1-(IL3RJ)2*], [SOCS1- (IL3RJ)2], [SOCS1- IL3RJ2-STAT5c-SHP2], [SOCS1-( IL3RJ)2-SHP2] retainedretained atat itsits maximummaximum [SHP2], [( IL3RJ)2*-SHP2 ], [(IL3RJ)2-SHP2] forfor thethe durationduration ofof stimulus.stimulus. [SOCS1], [SOCS1-( IL3RJ)2*], [SOCS1-( IL3RJ)2] KnockdownKnockdown SimulationSimulation SummarySummary

•• Knocking down either SHP2 or SOCS – New category of behavior in STA5 activation. – SHP2 and SOCS subsystem can change the behavior of the other subsystems (Candidate coordinator ?) •• If SHP2 is a coordinator: – STAT5 and SOCS subsystem are the coordinated subsystems (lower level) – But they are dependent on each other – Does not satisfy coordination requirements. •• SOCS as a coordinator: – STAT5 and SHP2 subsystem are in the lower level – Independent subsystems – A good candidate for a coordinator. SOCSSOCS AsAs AA CoordinatorCoordinator

• A coordinator can change system behavior to advance overall system goal. – The system is in the perturbed state (red line) – Change of parameters of SOCS subsystem (green line) • Low activation • Transient • SOCS subsystem is capable of carrying the system back to its quiescent state. • SOCS is a candidate coordinator

•• HierarchicalHierarchical representationrepresentation ofof •• BoundedBounded AutonomyAutonomy ofof LevelLevel -- SOCSSOCS asas aa coordinator.coordinator. surfacesurface forfor SOCSSOCS subsystem.subsystem. BiomarkersBiomarkers forfor ProstateProstate CancerCancer PI3K/AktPI3K/Akt andand NFNFκκBB PathwayPathway (joint work with Gupta lab)

Hierarchical Multi-level System Modeling Process Representation

Experiments designed Prostate Cancer for parameter estimation

Model Parameter Model Tissues Structure Estimation Validation

Spatial Faster size Reaction Cellular increase Time

Nucleus Other Intra-Cellular BiomarkersBiomarkers forfor ProstateProstate CancerCancer PI3K/AktPI3K/Akt andand NFNFκκBB PathwayPathway (joint work with Gupta lab)

LEVEL 2 PHENOTYPE Manifestation of disease Loss of androgen dependence Increased cancer aggressiveness

FEATURE EXTRACTION FMQAS/Neural networks Additional LEVEL 2 data [Shukla, 2005]. Immunostaining for NF-κB/p50 and IκBα in representative samples of benign prostate tissue and prostate cancer specimens of various Gleason grade.

LEVEL 1 SIGNALING R LR (ODE model) Receptor L LR LR2 LR2* R-PI3K PI3K PI3K R-PI3K*

PI3K*

Akt PI PIP3 PDK Akt Input IKKi Akt.PI** Akt.PI* Akt.PIP3

Akt.PI**.IKKi PP2A IKK* NFκB IKK.IkBα.NFkB IKK.IkBα NFkB

IkBα IkBα.NFkB

IC Nucleus IkBαn IkBαn.NFkBn

Source IkBαt NFkBn

Figure 5. Biochemical reactions of the pathway and its modularization. Modularization is based on the Figure 6. System arranged according to hierarchical Here levels. IC refers to Initial Conditions. biological functionality and contiguity HIERARCHICAL SYSTEMS THEORY Mesarovic, Macko, and Takahara, 1976

General Systems Formalization

Overall System Image P : D → Y D = D1⊗… ⊗ Dn = Decisions Y = Y1⊗… ⊗ Yn = Outputs Goal, Δ G : Y ⊗ D → Vn = Goal Value Set Subsystems Image Pi : Di ⊗ Ui → Yi Ui = Ki(D) = Uncertainty for subsystem i (due to interactions) U = U1⊗… ⊗ Un Goal Giβ : Yi ⊗ Di ⊗ {β} → Vi Coordination Process Find β ∈ B such that IBP is valid, i.e.,

satΔ &K & satΔ & satΔ ⇒ satΔ 1iβ nβ ∈ 0 References [1] Leroy Hood, Institute for Systems Biology

[2] Mesarovic M and Takahara Y, Mathematical Theory of General Systems, Academic Press, 1976.

[3] Leland H. Hartwell, John J.Hopfield, Stanislas Leibler and Andrew W.Murray, From molecular to modular cell biology, , Nature Vol 402, Supp, 2 December 1999.

[4] Judah Folkman, “Is Angiogenesis An Organizing Principle?” , The Miami Nature Biotechnology Winter Symposium, 2006.

[5] Douglas Hanahan and Robert A. Weinberg, The Hallmarks of Cancer, Cell, Vol. 100, January 7, 2000.

[6] John S.Mattick, Program of Complex Organisms, Scientific American, October 2004.

[7] Sreenath SN, Soebiyanto RP, Mesarovic MD, Wolkenhauer O. Coordination of Crosstalk between MAPK-PKC pathways. IEE Systems Biology, 2006.

[8] Andreas Wagner, Robustness and Evolvability in Living Systems, Princeton Studies in Complexity. ThankThank UU

Questions?Questions?