Lilia Alberghina Dept. of Biotechnology and Biosciences University of Milano-Bicocca Milan, Italy

1. SYSTEM-LEVEL PROPERTIES OF CELL CYCLE NETWORKS Lilia Alberghina Dept. of Biotechnology and Biosciences University of Milano-Bicocca Milan, Italy Seminar at CNR-IASI Rome, Dec 18, 2008

2. Systems Biology THE GREAT CHALLENGE FOR 21st CENTURY BIOLOGY • Only rarely a cellularfunction is determined by an individual gene product, but more often it is determined by the dynamic interactionof hundreds or thousands of gene products making it difficult to fully understand biological functions at a molecular level. • As a first step, it is necessary to identify the structure and dynamics of networks that execute and control basic complex cellular functions (metabolism, growth, cycle, differentiation, death, senescence, transformation). 2

3. SYSTEM-LEVEL PROPERTIES SYSTEMS BIOLOGY • Systems Biology concerns the mechanisms by which macromolecules interact dynamically to produce the functional properties of living cells. • It integrates molecular analysis with mathematical modeling and simulations • Cellular processes can be dissected into modules: subsystems of interacting molecules (DNA, RNA, proteins, small molecules) that perform a given task in a way that is largely independent from the context. • Modularity is organized by “global connectors” among modules and by “party hubs” that connect partners of each module. • The function of each system derives as an emergent property from interactions of the various elements of its network. • Biological networks are robust, since they are mostly able to maintain their function despite external and internal perturbations. 3

4. SYSTEMS BIOLOGY OF THE CELL CYCLE Essential functions of cell cycle: • coordination between growth and cycle progression • fidelity in nuclear genome replication and transmission A homeostasis of cell size setting the critical cell size required to enter S phase (Ps) B trigger acoherent, synchronous onset of DNA replication Our task has been, using a modular approach, to analyse the role of the G1/S network in determining these functions 4

5. G1 S G2 M Ps GROWTH, CYCLE AND Ps A • In the budding yeast Saccharomyces cerevisiae • During evolution the sequences of many cell cycle components are conserved from yeast to humans T = ln 2/λ 2 –TP/T +2 –TD/T =1 5

6. FROM A TOP-DOWN MODEL OF CELL CYCLE TO NETWORK IDENTIFICATION OF THE G1 TO S TRANSITION • Alberghina et al, Oncogene 20, 1128-1134, 2001 • Alberghina et al, Current Genomics 5, 615-627, 2004 • A top-down mathematical model of cell cycle The G1 to S transition is controlled by a cell sizer that involves Cki and is modulated by growth rate • Involvement of the Cki Far1 in mitotic cell cycle • Alberghina et al, J. Cell. Biol. 167, 433-443, 2004 • Role of nucleo/cytoplasmic localization of Sic1 for G1 to S transition • Rossi et al, Cell Cycle 4, 1798-1807, 2005 6

7. G1 to S transition Ps A Far1 amount endowed at the previous mitotic exit Budding Cln1.2. CdK1 Far1 Whi5 SBF/MBF Cln3.CdK1 THRESHOLD Clb5.6. CdK1/Sic1 Sic1 degradation Cln3 made in G1 proportional to cell mass Cell sizer DNA replication timer MATHEMATICAL MODEL OF THE G1 TO S TRANSITIONSTARTING FROM SMALL DAUGHTER CELLS (S-Cdk) 7 Barberis M, Klipp E. Vanoni M. and Alberghina L., PLoS Comput. Biol., 3, e64, 2007

8. THE MODEL: EQUATIONS AND SIMULATIONS • The model has been implemented by a set of ordinary differential equations (ODEs), that describe the temporal changes of the concentrations of the involved proteins and complexes. • The model considers the localization of components in different cell compartments (cytoplasm or nucleus) as well as the cell size growth during the G1 phase. • Parameter identification has been done by text mining for kinetic constants, by mathematical fitting of simulated versus experimental time series, by utilization of available experimental data as input quantities, and by parameter values utilized in literature models. • The model predicts the dynamics of key cycle players and allows to estimate Ps • It accounts for a variety of genetic and nutritional growth conditions 8

9. THE BREAKTHROUGH: PS IS AN EMERGENT PROPERTY OF THE G1/S NETWORK Growth rate Far1 initial concentration Cln3 initial concentration Binding value of Sic1 to Cdk1-Clb5,6cyt The value of Ps increases with growth rate 9

10. WHY GROWTH RATE MODULATES Ps S phase T2 S phase T1 = sizer T2 – T1 = timer (experimentally determined to be about 40 min for daughter cells growing in glucose T = 104 min) This model allows us to set in a unified framework all previously proposed regulatory events for the setting of Ps 10

11. INDEPENDENT CONFIRMATION OF THE SIZER-TIMER STRUCTURE OF THE G1 TO S TRANSITION single-cell imaging Average T2 ~ 20 min Average T1 D ~ 20 min Average T1 M ~ 1 minS. Di Talia et al, Nature 448, 947-951, 2007 40 min 11

12. HOW TO CONNECT S-Cdk ACTIVITY WITH INITIATION OF DNA REPLICATION? B ? • the amount of S-Cdk activity varies with the growth rate (Rossi et al, Cell Cycle, 2005) • the rate of degradation of Sic 1 may be modulated by Ck2 phosphorylation of Cdc34 (Coccetti et al, Cell Cycle, 2008) Tanaka et al, Cell Division, 2007 12

13. FOR A FAITHFUL DNA REPLICATION IT HAS TO START SYNCHRONOUSLY FROM ALL INVOLVED ORC • How does the availability of Clb5,6.Cdk1 control the onset of DNA replication in budding yeast? ORIGINS OF DNA REPLICATION 13

14. MODELING THE NETWORK CONTROLLING THE ONSET OF DNA REPLICATION R. HEINRICH 14 A. Brummer, V. Zinzalla, C. Salazar, L. Alberghina and T. Hoefer, 2008, submitted

15. EQUATIONS AND PARAMETERS OF THE MODEL • The mathematical model, 57 equations and 44 parameters,gives the probability that a particular origin is, after a certain time t in one of the states, S0 through S7. From the complete ensemble of each replication origin, this probability translates into the fraction of origins in a given state. • The parameters are grouped in three categories: protein concentrations (taken from Ghaemmaghani et al, 2003); binding/dissociation constants (estimated following Gabdoulline and Wade, 2001); protein phosphorylation/dephosphorylation rates (estimated following Shaw et al, 1995; Okamura et al, 2004). • This formulation allows us to study the coherence of origin firing and the molecular parameters that influence it. 15

16. STANDARD SIMULATED DYNAMICS OF INITIATION OF DNA REPLICATION AND EFFECTS OF S-Cdk AVAILABILITY ON FIRING COHERENCE coherent firing sharp synchrony coherent firing sharp synchrony less coherent firing loose synchrony 16

17. MULTI-SITE PHOSPHORYLATION OF Sld2 by S-Cdk IS THE MOLECULAR DEVICE RESPONSIBLE FOR THE KINETICS OF DNA FIRING Distributive phosphorylation of Sld2 by S-Cdk Sld2(6Ser/Thr // Thr84) total 17

18. MULTI-SITE PHOSPHORYLATION OF Sld2 by S-Cdk IS A DECOUPLING MOLECULAR DEVICE RESPONSIBLE FOR THE ROBUST KINETICS OF DNA FIRING • Multi-site phosphorylation of Sld2 works as a decoupling device, a robustness mechanism that isolates system’s functionality from variations of the input • Towards retarded activationfull reproducibility of standard performance • Towards reduced activation 92% of origin license and fire at low (20% of standard) S-Cdk availability longer S phase 18

19. KINETIC AND STRUCTURAL DETERMINANTS OF NETWORK ROBUSTNESS • Statistical ensemble of model parameterization by randomly changing all rate constants in a 2 order of magnitude interval around the reference value • Selection of those that satisfy > 185 origins fired within 45 min 200 admissible parameter sets The vast majority of functional network design kinetics behave as the reference case • prepare and fire kinetics 19

20. EMERGENT PROPERTIES AND ROBUSTNESS IN CELL CYCLE CONTROL • In conclusion: • the setting of the critical cell size at the onset of S phase is an emergent property of the G1 to S network modulated by growth rate; • the robustness of the coherent synchronous onset of DNA replication relies on molecular design principles (chiefly the multi-site phosphorylation of Sld2) of the molecular network executing and controlling DNA firing. 20

21. TAKE HOME LESSON • To identify system-level properties (emergence, robustness) a well-defined molecular structure of the network is needed • Integrated molecular/computational analysis is required to identify system-level properties • Only in this way can we understand the link between molecular networks and biological functions 21

22. 7 FP Project WHAT NEXT? MIUR-FIRB ITALBIONET • Systems biology concerns the mechanisms by which macromolecules interactdynamically to produce the functional properties of living cells Sic1 Whi5 Far1 Clb5.CdK1 SBF-MBF Cln3.CdK1 Modulation of level (synthesis/degradation) Perturbations by NCE Quantitative Proteomics/ Phosphoproteomics Modulation of nucleo/cytoplasmic localization Modulation of binding activity (phosphorylation?) How does cell signalling (TOR, Ck2, Snf1/AMPK, PKA, etc.) affect the strength of binding of different interactors? 22

23. Project ongoing CHANGING THE GENE DOSAGE OF A KEY CYCLE PLAYER Genome-wide changes in response to FAR1 gene dosage PCA analysis: done at IASI/RM (Paola Bertolazzi, Giovanni Felici) 23

24. Dfar1 FAR1OE WT Dfar1 FAR1OE WT GROWTH PARAMETERS OF YEAST STRAINS WITH ALTERED FAR1 DOSAGES 24

25. Expression + - FAR1OE FAR1OE wt wt far1D far1D Glucose Ethanol FAR1 gene dosage: analysis of the transcriptome 25

26. PCA analysis separates transcriptional profiles as a function of the FAR1 gene dosage ethanol glucose In glucose-grown cells PC1 (explaining over 70% of variability) separates well the three samples, while PC2 (explaining about 15% of variability) only distinguishes far1Dfrom wild type and Far1-overexpressing cells. In ethanol-growing cells, PC1 (explaining ca. 63% of variability) does not distinguish wild type and far1D mutant cells, that are instead well separated on the PC2 axis (explaining about 20% of variability). 26

27. Ethanol-grown cells 440 ORFs 295 ORFs Glucose-grown cells 816 ORFs 917 ORFs Different statistical tools identify some superimposable but also distinct FAR1-modulated genes The biological evaluation is ongoing 27

28. exponential growth in SCE pI 3 10 100 Wt Wt His4;Tup1 Leu1 Hsc82;Hsp82 Tkl1 Leu4 Hxk2 MW(Kda) Ddr48 Mls1 Ino1 Tef2 Arg1 Eno2 Sbp1;Pep4 Pgk1 Gdh3 Sbp1 Sgt2 Bat1 Fba1 Sec14 Tpm1 Snz1 Ura1 Hsp26 Rki1 Rib3 Rps7A 10 Ahp1 Hsp12 Effect of the FAR1 dosage on the proteome : growth in ethanol-supplemented media 28

29. Wt Cdc19 Eft1 Met6 Pdc1 Krs1 Hom2 Pgi1 Npl3 Hom6 Gua1 Hxk1 Wrs1 Eno1 Adh1 Ydl124w Dld3 Stm1 Tdh3 Rps2/Rps1A Rpl8/Rps4/Rpl2 Fur1 Rps7A Egd2 Rps18/Rps24/Rps17A Rib4 Rps12 Rpl26 Effect of the FAR1 dosage on the proteome : growth in glucose-supplemented media pI 3 10 100 exponential growth in SCD MW(Kda) 10 29

30. +10 ethanol +10 -10 glucose +10 Protein fold change -10 +10 -10 -10 mRNA fold change FAR1 gene dosage: transcriptome vs proteome 30

31. = no change = increase FAR1 overexpression stimulates rProt translation Ribosomal proteins are post-transcriptionally regulated in FAR1OE strains 31

32. ** Relative RNA content ** Wt far1D FAR1tet Wt far1D FAR1tet glucose ethanol FAR1 OVEREXPRESSION INCREASES CELLULAR RNA CONTENT FAR1tet cells Exponentially growing in glucose-supplemented media show a coordinate induction of some ribosomal proteins A loss of balance in ribosomal protein biogenesis could take place An imbalance in the synthesis of the two ribosome subunits 40S and 60S can induce ribosomal protein and rRNA synthesis by an autoregulation process (Zhao et al., 2003) FAR1tetmutantcells have morerRNA too? 32 32

33. PFK2 PFK2 PFK1 PFK1 TDH1 TDH1 TDH2 TDH2 FRU-6-P GLU-6-P FRU-6-P GLU-6-P GA-3-P FRU-1,6-P GA-3-P FRU-1,6-P PGI1 PGI1 FBA1 FBA1 TDH3 TDH3 TPI1 TPI1 FBP1 FBP1 HXK1 HXK1 PGK1 PGK1 GLK1 GLK1 Exponential growth in ethanol Exponential growth in glucose HXK2 HXK2 GPM1 GPM1 Glucose Glucose ENO1 ENO1 PDC1 PDC1 ENO2 ENO2 ADH2 ADH2 PYK2 PYK2 PDC5 PDC5 Ethanol Ethanol PEP Pyruvate Acetaldehyde PEP Pyruvate Acetaldehyde CDC19 CDC19 PDC6 PDC6 ADH1 ADH1 PDA1 PDA2 PDA1 PDA2 ALD2 ALD2 PCK1 ALD3 ALD3 PDB1 PDB1 PYC1,2 ALD5 ALD5 protein protein mRNA mRNA LPD1 LPD1 - - - - + + + + No change No change No change No change PDX1 PDX1 ACS1 ACS1 Acetate Acetate Acetyl-CoA Acetyl-CoA ACS2 ACS2 CIT1 Oxaloacetate CIT2 ACO1 CIT3 MDH1 IDH1,2 FUM1 a-ketoglutarate SDH3 SDH1,2,4 KGD1,2 Succinate Succinyl- CoA LSC2 FAR1 dosage modulates metabolism 33

34. FAR1 dosage affects expression of TOR-dependent Nitrogen Discrimination Pathway 34

35. FAR1 dosage modulates PKA and TOR pathways 35

36. FAR1 affects both growth and cell cycle Onset S phase SIZER TIMER Sic1 degradation Far1/Cln3 Whi5/SBF-MBF Sic1/Clb5 ● Critical Cell Size DOSAGE Metabolism building blocks energy SignalingPKA Tor protein synthesis Growth rate (?) Sfp1 ribosome biosynthesis 36 to be completed

37. Project ongoing NETWORKS AND CIRCUITS General properties of organization: positive and negative feedbacks, threshold, switch, error connection, cell sizer etc. Hartwell et al, Nature, 1999 in collaboration with L. Farina, P. Palumbo, G. Mavelli 37

38. FROM THE CONCEPT MODEL TO THE REAL THING 38

39. MODELS IN SYSTEMS BIOLOGY • A model is a symbolic representation of reality which is able to foster understanding and to support decision-making • A mathematical model is able to give a quantitative representation of a process and to make predictions • Models in systems biology • structural models • regulatory models • dynamic models 39

40. 40 Courtesy of A. Henney

41. START END Resetting Subsystem Master Control MITOSIS cell sizer (Ps) Pro Meta Ana Telo Kinesis Cki Cki C1 C2 C3 M G1 S G2 M fast growth cAMP hyperactivation fast growth STRESS GROWTH A TOP-DOWN MODEL OF CELL CYCLE (2001) • Two major areas of control • a cell sizer control (involving Cki and modulated by growth conditions) at the G1 to S transition • delays of mitosis execution, at metaphase/anaphase (End2) and at anaphase/telophase (End3), modulated by stress (DNA and spindle damages, conflicting metabolic signals, etc.) Alberghina et al – Oncogene 20, 1128-1134, 2001 Alberghina et al – Current Genomics 5, 615-627, 2004 41

42. ANALYSIS OF A SHIFT UP BY SIMULATION The model correctly predicts, for cell population, during transitory state, a continuous increase of Ps and an increase in duration of budded phase, followed by its decrease to the new steady state Alberghina et al., Oncogene 20, 1128-1134, 2001 Alberghina et al., J. Bacter. 180, 3864-3872 A proteomic analysis indicates that shift up cells undergo a stress response, conferming a connection between stress and delay of mitotic exit Querin et al., J. Biol Chem, 2008 42

43. THE SYSTEMS BIOLOGY APPROACH 43 Courtesy of A. Henney

44. STARTING NETWORK IDENTIFICATION WITH A MODULAR SYSTEMS BIOLOGY APPROACH 44 Alberghina L. et al, Curr. Genomics, 2004

45. G1 to S transition Ps A Far1 amount endowed at the previous mitotic exit Budding Cln1.2. CdK1 Far1 Whi5 SBF/MBF Cln3.CdK1 THRESHOLD Clb5.6. CdK1/Sic1 Sic1 degradation Cln3 made in G1 proportional to cell mass Cell sizer DNA replication timer MATHEMATICAL MODEL OF THE G1 TO S TRANSITIONSTARTING FROM SMALL DAUGHTER CELLS (S-Cdk) 45 Barberis M, Klipp E. Vanoni M. and Alberghina L., PLoS Comput. Biol., 3, e64, 2007

46. MODELING THE NETWORK CONTROLLING THE ONSET OF DNA REPLICATION R. HEINRICH 46 A. Brummer, V. Zinzalla, C. Salazar, L. Alberghina and T. Hoefer, 2008, submitted

47. Swi4 SBF Far1 Swi6 Mbp1 MBF Cln3 Whi5 Cln1 Cln2 Cdc28 Clb 5,6 ● Sic1 Cdc34 Sld3 Cdc4 ● Ck2 11-3-2 Sld2 Skp1 Cdc53 Dpb11 Cdc6 DNA polym ● Mcm2-7 Cdt1 ● Complex Cdc45 ● Enzymatic Complex Psf3 ● Reconstruction of Protein Interactome of G1 to S transition Psf2 Sld5 47 Psf1 GINS

48. cell Blue(print) cell Transcription Carbon metabolism Nuclear transport Translation Membrane traffic TCA cycle Mitochondria transport Energy metabolism

49. human Brain Model calibration, validation, comparison Eyes Heart Lungs Cartilage Kidney Intestine Liver Figure 1. Service Oriented computing Architecture integrating the Web Services (indicated by the monitors) representing the human organs, through its Service Broker (SB). Blue(print) organism

50. THE MILANO-BICOCCA TEAM AND COLLABORATIONS Max Planck Institute for Molecular Genetics, Berlin E. Klipp M. Barberis University of Milano-Bicocca L. Alberghina M. Vanoni R. Rossi V. Zinzalla L. Querin P. Coccetti A. Mastriani M. Graziola F. Tripodi F. Sternieri D. Porro A. Di Fonzo F. Magni S. Fantinato L. De Gioia P. Fantucci R. Sanvito V. Tsiarentsyeva C. Cirulli N. Campbell M. Marchegiano V. Zinzalla, University of Basel T. Höfer, A. Brummer, C. Salazar Dfkz-Heidelberg