Binding mode prediction via graph theory @ BioMolecular Science Gateway

A novel approach to protein-ligand binding mode prediction by rigidity analysis using graph theory

Transcript

1. BioMolecular Science Gateway Research Forum February 8, 2016 Sebastian Raschka A novel approach to protein-ligand binding mode prediction by rigidity analysis using graph theory
2. A little bit about myself …! 2! SiteInterlock A truly novel algorithm for protein- ligand docking based on graph theory A large-scale virtual screening framework for hypothesis-driven ligand-based protein-inhibitor discovery SeaScreen
3. A little bit about myself …! 3! SiteInterlock A truly novel algorithm for protein- ligand docking based on graph theory A large-scale virtual screening framework for hypothesis-driven ligand-based protein-inhibitor discovery SeaScreen
4. Protein Ligand Docking! When & Why?! Structure of Ibuprofen bound to cyclooxygenase-2 ! Orlando, B. J., Lucido, M. J., & Malkowski, M. G. (2015)! (PDB code: 4ph9)! 4!
5. Ranking and Discovery! Protein structure! Absolute or relative ! binding affinity! ! [scoring function]! 5! Ligand
6. Binding Mode Prediction! Protein structure! 6! Ligand! [scoring function]! ?! ?! ?!
7. Binding Mode Prediction! Ligand “Pose”! Orientation! Conformation! [ + flexible protein side chains ]! 7! deoxycytidylate hydroxymethylase cognate ligand 2'-deoxycytidine-5'-monophosphate ! (PDB code: 1b5e)! +
8. Evaluation Metric! 8!
9. Evaluate against hold-out data! Experimental structure! RMSD 1.0 Å! RMSD 2.8 Å! 9! Carboxypeptidase A + inhibitor L-benzylsuccinate (PDB code: 1cbx)!
10. Protein structure! Experimental structure! Representation as a! “docking problem”! Generating & ranking ! docking poses! Evaluating pose(s)! 10! [scoring function]! ?! ?! ?! Ligand!
11. Internal Scoring Metrics! 11! Statistical ! potentials! Molecular ! Mechanics ! (force fields)! Empirical! •  Accuracy! •  Computational efficiency! •  Apo-structures!
12. 12! Statistical ! potentials! Molecular ! Mechanics ! (force fields)! Empirical! SiteInterlock! •  Accuracy! •  Computational efficiency! •  Apo-structures!
13. 13! “bad” docking pose! “near -native” docking pose! more rigid! more! flexible! We can detect a local rigidity increase! upon protein-ligand complex formation! Hypothesis!
14. 14! Thermal Shift Assay! heat Kunfold (T) Kdye (T)
15. 15! M. D. Cummings, M. A. Farnum, and M. I. Nelen. Universal screening methods and applications of thermofluor. Journal of biomolecular screening, 11(7):854–863, 2006.! Thermal Shift Assay! Protein + Ligand Protein
16. Predicting Flexibility via ProFlex! 16! Penicillin-derived ! asymmetric inhibitor! Crystal structure! of HIV protease! Crystal structure! (after deleting the ligand)! (PDB code: 1htg)! D. J. Jacobs, A. J. Rader, L. A. Kuhn, and M. F. Thorpe. Protein flexibility predictions using graph theory. ! Proteins: Structure, Function, and Bioinformatics, 44(2):150–165, 2001.!
17. Apo Structure! 17! Apo Structure! (PDB code: 1rpi)!
18. 18! How ProFlex works! under-constrained! (flexible)! over-constrained! (rigid)! isostatic! (just rigid)!
19. 19!
20. 2D Pebble Game! 2 0 Jacobs and Thorpe. Generic rigidity percolation: The pebble game. Phys Rev Lett, 75(22):4051–4054, Nov 1995. minimally rigid graph with n nodes and m edges! m = 2n - 3 a c b (2,3 counting)!
21. 2D Pebble Game! a c b pebbles 21!
22. 2D Pebble Game! a c b 1) Draw an edge if 2 pebbles are present at both nodes. ! 22!
23. 2D Pebble Game! a c b 1) Draw an edge if 2 pebbles are present at both nodes. ! Next, consume 1 pebble from the starting node.! 23!
24. 2D Pebble Game! a c b 1) Draw an edge if 2 pebbles are present at both nodes. ! Next, consume 1 pebble from the starting node.! 2) Do a depth-first search to recover pebbles! 24!
25. 2D Pebble Game! a c b 1) Draw an edge if 2 pebbles are present at both nodes. ! Next, consume 1 pebble from the starting node.! 2) Do a depth-first search to recover pebbles! 3) Revert the edge and bring the pebble back to the node! 25!
26. 2D Pebble Game! a c b 1) Draw an edge if 2 pebbles are present at both nodes. ! Next, consume 1 pebble from the starting node.! 2) Do a depth-first search to recover pebbles! 3) Revert the edge and bring the pebble back to the node! 4) Go back to 1) and Insert a new edge! 26!
27. 2D Pebble Game! a c b 2) Do a depth-first search to recover pebbles! 3) Revert the edge and bring the pebble back to the node! 27!
28. 2D Pebble Game! a c b minimally rigid! 4) Go back to 1) and Insert a new edge! 28!
29. 2D Pebble Game! a c b flexible (2 d.o.f.)! d e 29!
30. 2D Pebble Game! a c b flexible (1 d.o.f.)! d 30!
31. 31! Crystal structure of the anti- bacterial sulfonamide drug target dihydropteroate synthase. 
 (PDB code: 1ajz)
32. 32! Selecting the best ligand binding pose with ProFlex
33. Take crystal pose! Generate low-energy conformations! Dock low-energy conformations! Score docked complexes! Compare best-scoring pose to crystal! 33!
34. 34! Protein Flexibility Changes in Docking Poses! Carboxypeptidase A + inhibitor L-benzylsuccinate (PDB code: 1cbx)! “most rigid”! protein! “least rigid”! protein!
35. 35! Protein avg. flexibility! Protein Flexibility Changes in Docking Poses!
36. 36! SiteInterlock-Score!
37. Generate low-energy conformations (OMEGA21)! Sample docking poses in flexible binding site (SLIDE2)! Determine parameters of stable ligand-free protein structure (HETHER)! Analyze rigidity of docked Poses (PROFLEX)! Extract binding pocket and rank poses (SiteInterlock-Score)! Extract ligand from crystal structure! [1] P. C. D. Hawkins, A. G. Skillman, G. L. Warren, B. A. Ellingson, and M. T. Stahl. Conformer generation with omega: algorithm and validation using high quality structures from the protein databank and cambridge structural database. J Chem Inf Model, 50(4):572–84, Apr 2010. [2] M. I. Zavodszky, P. C. Sanschagrin, L. A. Kuhn, and R. S. Korde. Distilling the essential features of a protein surface for improving protein-ligand docking, scoring, and virtual screening. Journal of computer-aided molecular design, 16(12):883–902, 2002. 37!
38. 19 x Holo! 11 x Apo! (holo ligand \ ! apo protein)! 1a9x ! 1amu 1b5e 1bgv 1bx4 1c96 1cbs! 1cbx! 1ccw 1chm 1com! 1coy! 1cps! 1did! 1hwr! 1rx1 7tim! 3ks9! 3odu! 10gs / 16gs! 1ahb / 1ahc 1aj2 / 1ajz! 1gmr / 1gmq ! 1kel / 1kem! 1nsc / 1nsb ! 1swd / 1swa! 3tmn / 1tli 1tmt / 1vr1 1ydb / 1ydc! 5sga / 2sga!
39. 39!
40. 40! [1] TroD, O., & Olson, A. J. (2010). AutoDock Vina: improving the speed and accuracy of docking with a new scoring funcOon, efficient opOmizaOon, and mulOthreading. Journal of computa.onal chemistry, 31(2), 455-461. [2] Fan, H., Schneidman-Duhovny, D., Irwin, J. J., Dong, G., Shoichet, B. K., & Sali, A. (2011). StaOsOcal potenOal for modeling and ranking of protein–ligand interacOons. Journal of chemical informa.on and modeling, 51(12), 3078-3092. [3] Neudert, G., & Klebe, G. (2011). DSX: a knowledge-based scoring funcOon for the assessment of protein–ligand complexes. Journal of chemical informa.on and modeling, 51(10), 2731-2745. [4] Wang, R., Lai, L., & Wang, S. (2002). Further development and validaOon of empirical scoring funcOons for structure-based binding affinity predicOon. Journal of computer-aided molecular design, 16(1), 11-26. [5] Allen, W. J., Balius, T. E., Mukherjee, S., Brozell, S. R., Moustakas, D. T., Lang, P. T., ... & Rizzo, R. C. (2015). DOCK 6: Impact of new features and current docking performance. Journal of computa.onal chemistry, 36(15), 1132-1156. [1] [2] [3] [4] [5]
41. 41!
42. 42!
43. 43! Spearman ρ! A “unique” signal!
44. Future Directions! 44!
45. Acknowledgements! Dr. Leslie A. Kuhn (Advisor)! Professor in the Department of Biochemistry and Molecular Biology! The Kuhn Lab! Joseph Buffington-Bemister! Undergraduate Researcher! 45! Alex Wolf! Undergraduate Researcher!