From cell biology to Petri nets. Rainer Breitling, Groningen, NL David Gilbert, London, UK Monika Heiner, Cottbus, DE

Transcription

1 From cell biology to Petri nets Rainer Breitling, Groningen, NL David Gilbert, London, UK Monika Heiner, Cottbus, DE

2 Biology = Concentrations Breitling / 2

3 The simplest chemical reaction A B irreversible, one-molecule reaction examples: all sorts of decay processes, e.g. radioactive, fluorescence, activated receptor returning to inactive state any metabolic pathway can be described by a combination of processes of this type (including reversible reactions and, in some respects, multi-molecule reactions) Breitling / 3

4 The simplest chemical reaction A B various levels of description: homogeneous system, large numbers of molecules = ordinary differential equations, kinetics small numbers of molecules = probabilistic equations, stochastics spatial heterogeneity = partial differential equations, diffusion small number of heterogeneously distributed molecules = single-molecule tracking (e.g. cytoskeleton modelling) Breitling / 4

5 Kinetics Description Main idea: Molecules don t talk Imagine a box containing N molecules. How many will decay during time t? k*n Imagine two boxes containing N/2 molecules each. How many decay? k*n Imagine two boxes containing N molecules each. How many decay? 2k*N In general: dn( t) dt = λ * n( t) differential equation (ordinary, linear, first-order) n( t) = N e 0 λt exact solution (in more complex cases replaced by a numerical approximation) Breitling / 5

6 Kinetics Description If you know the concentration at one time, you can calculate it for any other time! (and this really works) Breitling / 6

7 Probabilistic Description Main idea: Molecules are isolated entities without memory Probability of decay of a single molecule in some small time interval: p = λδt 1 Probability of survival in Δt: p 2 = p = 1 λδt 1 1 Probability of survival for some time t: p t = lim(1 λ ) x x x = e λt Transition to large number of molecules: n( t) = N 0 e λt or dn( t) λt = λn0e = λn( t) dt Breitling / 7

8 Probabilistic Description 2 Probability of survival of a single molecule for some time t: p t = lim(1 λ ) x x x = e λt Probability that exactly x molecules survive for some time t: p x = ( e λt ) x (1 e λt ) N 0 x N x 0 Most likely number to survive to time t: max( x p x ) = N e 0 λt Breitling / 8

9 Probabilistic Description 3 Markov Model (pure death!) Decay rate: Probability of decay: Probability distribution of n surviving molecules at time t: Description: Time: t -> wait dt -> t+dt Molecules: n -> no decay -> n n+1 -> one decay -> n Λ ( n, t) = nλ p = Λ( n, t) dt, ) P( n t P( n, t P( n + + dt) = + 1, t) Λ( n + 1, t) dt P( n, t)[1 Λ( n, t) dt] Final Result (after some calculating): The same as in the previous probabilistic description Breitling / 9

10 Petri Net representation? Breitling / 10

11 Some (Bio)Chemical Conventions Concentration of Molecule A = [A], usually in units mol/litre (molar) Rate constant = k, with indices indicating constants for various reactions (k 1, k 2...) Therefore: A B d[ A] dt = d[ B] dt = k [ A] 1 Breitling / 11

12 Reversible, Single-Molecule Reaction A B, or A B B A, or Differential equations: d[ A] forward reverse dt = k 1 [ A] + k 2 [ B] d[ B] dt = k 1 [ A] k 2 [ B] Main principle: Partial reactions are independent! Breitling / 12

13 Breitling / 13 Reversible, single-molecule reaction 2 Differential Equation: Equilibrium (=steady-state): equi equi equi equi equi equi equi K k k B A B k A k dt B d dt A d = = = + = = ] [ ] [ 0 ] [ ] [ 0 ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ B k A k dt B d B k A k dt A d = + =

14 Irreversible, two-molecule reaction A+B C Differential equations: The last piece of the puzzle d[ A] dt d[ A] dt d[ B] d[ C] = = dt dt = k[ A][ B] Non-linear! Underlying idea: Reaction probability = Combined probability that both [A] and [B] are in a reactive mood : * * p ( AB) = p( A) p( B) = k1 [ A] k2[ B] = k[ A][ B] Breitling / 14

15 A simple metabolic pathway A B C+D Differential equations: d/dt decay forward reverse [A]= -k1[a] [B]= +k1[a] -k2[b] +k3[c][d] [C]= +k2[b] -k3[c][d] [D]= +k2[b] -k3[c][d] Breitling / 15

16 Metabolic Networks as Bigraphs A B C+D A B C k1 k2 k3 D k1 k2 k3 d/dt decay forward reverse A B [A] [B] -k1[a] +k1[a] -k2[b] +k3[c][d] C [C] +k2[b] -k3[c][d] D [D] +k2[b] -k3[c][d] Breitling / 16

17 Petri nets Breitling / 17

18 Petri nets Breitling / 18

19 Qualitative Petri-Net Modelling & Analysis Graphical representation - Snoopy Qualitative analysis Charlie Unbounded, live & reversible Covered by T invariants P invariants Breitling / 19

20 Biological description bigraph differential equations KEGG Breitling / 20

21 Biological description bigraph ODEs substance A substance B EC A k1 B Breitling / 21

22 Biological description bigraph ODEs substance A EC substance B E A k B k1 k2 EA k* EB Breitling / 22

23 A special case: enzyme reactions In a quasi steady state, we can assume that [ES] is constant. Then: If we now define a new constant K m (Michaelis constant), we get: Breitling / 23

24 A special case: enzyme reactions Substituting [E] (free enzyme) by the total enzyme concentration we get: Hence, the reaction rate is: Breitling / 24

25 A special case: enzyme reactions Underlying assumptions of the Michaelis-Menten approximation: Free diffusion, random collisions of infinite number of molecules Irreversible reactions Quasi steady state In cell signaling pathways, all three assumptions will be frequently violated: Reactions of rather rare molecules happen at membranes and on scaffold structures Reactions happen close to equilibrium and both reactions have non-zero fluxes Enzymes are themselves substrates for other enzymes, concentrations change rapidly, d[es]/dt d[p]/dt Breitling / 25

26 Cell signaling pathways Fig. courtesy of W. Kolch Breitling / 26

27 Metabolic pathways vs. Signaling Pathways (can you give the mass-action equations?) Metabolic (initial substrate) S E1 Input Signal X Signaling cascade S S1 P1 E2 S S2 P2 E3 S (final product) Classical enzyme-product pathway S3 P3 Output Product become enzyme at next stage Breitling / 27

28 Metabolic pathways vs. Signalling Pathways Breitling / 28

29 Cell signaling pathways Breitling / 29

30 Cell signaling pathways Breitling / 30

31 Cell signaling pathways Breitling / 31

32 Cell signaling pathways Common components: Receptors binding to ligands R(inactive) + L RL(active) Proteins forming complexes P1 + P2 P1P2-complex Proteins acting as enzymes on other proteins (e.g., phosphorylation by kinases) P1 + K P1* + K Breitling / 32

33 MA1: Mass action for enzymatic reaction E+A k1 k 2 A: substrate B: product E: enzyme E A substrate-enzyme complex E A k 3 A E + B E B Breitling / 33

34 MA2 model A+ E k1 k2 A E k ' 3 B E k 1' k ' 2 B + E Breitling / 34

35 MA3 model A+ E k1 k2 A E k ' 3 k ' 4 B E k 1' k ' 2 B + E Breitling / 35

36 Cell signaling pathways feedback loops Breitling / 36

37 Cell signaling pathways feedback loops Fig. courtesy of W. Kolch Breitling / 37

38 Feedback loops in Petri Nets (a) S 1 (b) S 1 R R p R R p P 1 P 1 RR RR p RR RR p P 2 P 2 (c) S 1 (d) S 1 R R p R R p P 1 P 1 RR RR p RR RR p P 2 P 2 Breitling / 38

39 Feedback loops in Petri Nets Breitling / 39

40 Feedback loops in Petri Nets Breitling / 40

41 and added inhibitor Breitling / 41

42 Many PN modelling challengings remain Lack of parameters Qualitative vs. Continuous PN Small molecule numbers Deterministics vs. Stochastic models Spatial heterogeneity??? Breitling / 42

43 Cell signaling pathways Fig. courtesy of W. Kolch Breitling / 43

44 Stochastic vs. Continuous 5 5 X X Breitling / 44

45 Stochastic model checking Two Reaction Model First a simple model of two reactions: 0.01 A B C D 0.1 Assess property: P=?[ A = $X { A = D } ] What is the probability that, when A and D first equal each other, they both have $X number of molecules? Breitling / 45

46 Two Reaction Model Property: P=?[ A = $X { A = D } ] A Concentration * D Time Breitling / 46

47 Two Reaction Model Set reactants to 10 molecules (model bound to 10 molecules) Simulate with Gillespie 1,000 times and model check each output Number of simulations which are true over total number of simulations is the probability. Also checked the continuous model and the answer is the solid line. Breitling / 47

48 Two Reaction Model Breitling / 48

49 Spatial heterogeneity concentrations are different in different places, n = f(t,x,y,z) diffusion superimposed on chemical reactions: n( t) t xyz = λ n( t) xyz ± diffusion partial differential equation Breitling / 49

50 Spatial heterogeneity one-dimensional case (diffusion only, and conservation of mass) Δx n( t, x) Δx = inflow outflow t n( t, x + Δx) outflow = K x n(t,x) inflow = K x inflow outflow Breitling / 50

51 Breitling / 51 Spatial heterogeneity 2 ) ( ) ( ) ( reaction : Combination with chemical ),,, ( ),,, ( Shorthand for three dimensions : ), ( ), ( equation to get diffusion equation : to differential Transition ), ( ), ( ), ( t n K t n t t n z y x t n K t z y x t n x x t n K t x t n x x t n K x x x t n K x t x t n + = = = + Δ = Δ λ

52 Acknowledgements David Gilbert, Brunel University, London Monika Heiner, Cottbus University, Germany Robin Donaldson, Glasgow University Breitling / 52

53 The Groningen Bioinformatics Centre (Netherlands) is expanding its young and successful team. Several PhD and Postdoc positions are available for creative bioinformaticians with an interest in Systems Biology, Metabolomics, Proteomics, Quantitative Genetics, Network Reconstruction, Dynamic Modelling For more information and to apply visit talk to Rainer Breitling at Petri Nets 2009 Recent GBiC papers: Breitling R et al. New surveyor tools for charting microbial metabolic maps Nature Reviews Microbiology (2008). Fu J et al. MetaNetwork: a computational protocol for the genetic study of metabolic networks Nature Protocols (2007). Swertz MA et al. Beyond standardization: dynamic software infrastructures for systems biology Nature Reviews Genetics (2007). Keurentjes JJB et al. The genetics of plant metabolism Nature Genetics (2006). Hoeller D et al. Regulation of ubiquitin-binding proteins by monoubiquitylation Nature Cell Biology (2006). Bystrykh L et al. Uncovering regulatory pathways that effect hematopoietic stem cell function using genetical genomics Nature Genetics (2005).