1. RAW264.7 Cell Ligand Screen Summary Progress Report and Perspectives AfCS 5/24/04

2. RAW Cell Ligand Screens System complexity / richness / interest How many inputs are there? How many primary signal outputs are there? How many outputs/inputs show up only in the two-ligand screen? How many interactions (complexity)? How many are macrophage specific?

3. RAW Cell Ligand Screens System complexity / richness / interest How many inputs are there? How many primary signal outputs are there (weak inference)? How many show up only in the two-ligand screen? How many interactions (complexity)? How many are macrophage specific? Pathway topology and structure – double ligand spinoff What are the interactions? What are the likely nodes? What are the likely mediators of interaction? Where do we put the meters?

4. RAW Cell Ligand Screens System complexity / richness / interest How many inputs are there? How many primary signal outputs are there (weak inference)? How many show up only in the two-ligand screen? How many interactions (complexity)? How many are macrophage specific? Pathway topology and structure – double ligand spinoff What are the interactions? What are the likely nodes? What are the likely mediators of interaction? Where do we put the meters? Mechanism Inference of causality from order of events and identity of proteins

5. RAW Cell Ligand Screens Progress Report Summary Ligand list chosen (literature, Affy, Macrophage Committee) Assays adapted, cytokine assay added, lipid assays operational Cyclic AMP, Ca 2+, cytokines: single AND double almost done Phosphoproteins: interpretable informative now, done in 2004 Lipids and transcripts: in progress for select ligands and pairs Analytical approaches: in place and largely automated

6. RAW264.7 Cells System complexity / richness / interest How many inputs are there? ~ 60 compounds tested of possible ~90 ~25 positives in single ligand screen Do “negatives” alter responses in the double ligand screen ~10 ligands alter activity in assays where they alone have no effect Probability of effect for these ligands lower No assays yet for these mechanisms Perhaps 40 active regulatory ligands Proposal: screen all the negatives as doubles with positives in highest throughput, most global assays Conclusion: We have a big toolbox Much of it is NOT macrophage specific

7. How many primary signal outputs are there ? It’s an OUTPUT screen too !! Are patterns of outputs discrete? Probably not. They are clusterable ! ~10 identifiable groups cAMP, Ca2+, cytokines, phosphoproteins, transcripts Clusters imply shared mechanisms -- Madhu Separation of clusters traceable to specific outputs RAW264.7 Cells System complexity / richness / interest RRRR EEE EEE E

8. B Cell Merged Unified Experiment Space Madhu Natarajan Move the bar left or right according to statistical significance of groupings. Then look at the data to see what distinguishes the groups.

9. -20 20 M. Natarajan

10. How many primary signal outputs are there ? Are patterns of outputs discrete? Probably not. They are clusterable ! ~10 identifiable groups cAMP, Ca2+, cytokines, phosphoproteins, transcripts Clusters imply shared mechanisms -- Madhu Separation of clusters traceable to specific outputs Lipids and phosphoproteins: infer enzymes from metabolism RAW264.7 Cells System complexity / richness / interest RRRR EEE EEE E

11. RAW264.7 Cells The Double Ligand Screen for System Complexity AKA Interactive Output Screen ABC Y ABC X ABC X Additive Non-additive - Cross-talk Three-way

12. How many interactions ? Phosphoproteins: 39 from a 15x15 matrix 17% Cyclic AMP and Ca 2+ : 35 from 181 19% Cytokines: ~50 so far, pending follow-up Lipids, transcripts yet to come Transcript interactions apparently abundant Can we identify, classify and understand them? RAW264.7 Cells The Double Ligand Screen for System Complexity

13. RAW cells: Ligand pair interactions - P.C.S. / Cell Lab

14. Ca ++ & cAMP combined Stimulating ligand pairs (either assay) Stimulating vs. Non- stim pairs No stim- ulation (either assay) Totals Pairs28136 300 Non-additive2212NA34 Additive / Not called6124NA130 Not tested00136 % Non- additive79 %9 %11 % % Non- additive of tested 21 % RAW 264.7 Cells: Density of Interactions - P.C.S. / Cell Lab So: Test the others (as pools, perhaps) Novel inputs observable in phosphoprotein, cytokine assays

15. How many interactions ? Phosphoproteins: 39 from a 15x15 matrix 17% Cyclic AMP and Ca 2+ : 34 from 164 21% Cytokines, lipids, transcripts yet to come Can we identify, classify and understand them? Classification by OUTPUT is more meaningful than classification by LIGANDS RAW264.7 Cells The Double Ligand Screen for System Complexity Mechanistic Information from the Ligand Screen

16. Pathway Interactions -- Identify: Non-additivity in N-dimensional parameter space Rama, Madhu Classify: Direct analysis of interactions Paul PCA of difference vectors -- classification by output interaction Yet to come Can they be clustered to give new mechanisms? Understand : These experiments will point to the nodes in the RAW cell signaling network Regulation, downstream amplifiers, inhibitors Mechanistic Information from the Ligand Screen

17. Ca cAMP PGEISOUDPPAFC5AI04IFGTGF848P2CS1P2MALPAUTP Paul Sternweis

18. Pathway Interactions -- Identify: Non-additivity in N-dimensional parameter space Rama, Madhu Classify: Direct analysis of interactions Paul PCA of difference vectors -- classification by output interaction Yet to come Can they be clustered to give new mechanisms? Understand : Clusters of outputs expand / reinforce identification of pathways Use RNAi, inhibitors, to identify the shared components Non-additivity = interactions = linkages among pathways Use RNAi, inhibitors, to identify the likely nodes These nodes are where the biosensors should go Mechanistic Information from the Ligand Screen

19. Time Domains of the Ligand Screen Assays Ca 2+ Cyclic AMP Lipids Phosphoproteins mRNAs Cytokines Time (s) 1 10 100 1000 10,000 Mechanistic Information from the Ligand Screen

20. RAW: Stimulation of Ca 2+ --Cell Lab

21. Time Domains of the Ligand Screen Assays Ca 2+ Cyclic AMP Lipids Phosphoproteins mRNAs Cytokines Time (s) 1 10 100 1000 10,000 Mechanistic Information from the Ligand Screen Both responses and interactions can be ordered in time. Post hoc ergo propter hoc??

22. Mechanistic Information from the RAW Cell Ligand Screen Classes of responses that the RAW cell can generate Responses that group -- common mechanisms Responses that are ordered in time -- hints at causation Modulation of responses -- intersection of pathways 0 + 1 = 7: Directionality of modulation Modulations coupled with primary responses -- hints at mechanism Inhibitors, amplifiers, regulators The ligand screens provide the benchmarks for mechanistic experimtents and modeling the RAW signaling network

23. RAW264.7 Cell Ligand Screen Summary Progress Report and Perspectives AfCS 5/24/04