Alpha-helical Topology and Tertiary Structure Prediction of Globular Proteins Scott R. McAllister Christodoulos A. Floudas Princeton University

Transcription

1 Alpha-helical Topology and Tertiary Structure Prediction of Globular Proteins Scott R. McAllister Christodoulos A. Floudas Princeton University Department of Chemical Engineering Program of Applied and Computational Mathematics Department of Operations Research and Financial Engineering Center for Quantitative Biology

2 Outline Protein structure prediction overview Predicting α-helical contacts Probability development Model Results Predicting α-helical contacts in α/β proteins Distance bounding Model Results Structure prediction of α-helical proteins Framework Results

3 Protein Structure Prediction Problem Given an amino acid sequence, identify the three-dimensional protein structure Approaches Homology modeling Fold recognition/threading Fragment assembly First Principles - Optimization Statistical potentials Physics-based potentials TLQAETDQLEDEKSALQ??? Floudas, Biotechnology & Bioengineering, Floudas. AIChE Journal. 2005, 51: Floudas, et al. Chemical Engineering Science. 2006, 61:

4 ASTRO-FOLD Helix Prediction -Detailed atomistic modeling -Simulations of local interactions (Free Energy Calculations) β-sheet Prediction -Novel hydrophobic modeling -Predict list of optimal topologies (Combinatorial Optimization, MILP) Loop Structure Prediction -Dihedral angle sampling -Discard conformers by clustering (Novel Clustering Methodology) Derivation of Restraints -Dihedral angle restrictions -C α distance constraints (Reduced Search Space) Overall 3D Structure Prediction -Structural data from previous stages -Prediction via novel solution approach (Global Optimization and Molecular Dynamics) Klepeis, JL and Floudas, CA. Biophys J. (2003)

5 Outline Protein structure prediction overview Predicting α-helical contacts Probability development Model Results Predicting α-helical contacts in α/β proteins Distance bounding Model Results Structure prediction of α-helical proteins Framework Results

6 Overview Problem Topology prediction of globular α-helical proteins Approach Thesis: Topology is based on certain Inter-helical Hydrophobic to Hydrophobic Contacts Create a dataset of helical proteins Develop inter-helical contact probabilities Apply two novel mixed-integer optimization models (MILP) Level 1 - PRIMARY contacts Level 2 - WHEEL contacts McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

7 Dataset Selection Protein Sources 229 PDBSelect25 1 database 62 CATH 2 database 20 Zhang et al. 3 7 Huang et al. 4 Restrictions No β-sheets, at least 2 α-helices No highly similar sequences Dataset 318 proteins in the database set 1 Hobohm, U. and C.Sander. Prot Sci 3 (1994) Orengo, C.A. et al. Structure 5 (1997) Zhang, C. et al. PNAS 99 (2002) Huang, E.S. et al. J Mol Biol 290 (1999) 267. McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

8 Probability Development Contact Types PRIMARY contact Minimum distance hydrophobic contact between 4.0 Å and 10.0 Å WHEEL contact Only WHEEL position hydrophobic contacts between 4.0 Å and 12.0 Å Classified as parallel or antiparallel contacts McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

9 Model Overview Formulation: Maximize inter-helical residue-residue contact probabilities Binary variable helical contact Binary variable A yh m,n m n w, i, j indicates antiparallel indicates residue contact Goal: Produce a rank-ordered list of the most likely helical contacts Contacts used to restrict conformational space explored during protein tertiary structure prediction McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

10 Pairwise Model Objective Level 1 Objective Maximize probability of pairwise residueresidue contacts McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

11 Pairwise Model Constraints Level 1 Constraints At most one contact per position Helix-helix interaction direction Linking interaction variables McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

12 Pairwise Model Constraints Level 1 Constraints Restrict number of contacts between a given helix pair (MAX_CONTACT) Vary the number of helix-helix interactions (SUBTRACT) McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

13 Pairwise Model Constraints Level 1 Constraints Allow for and Limit helical kinks McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

14 Pairwise Model Constraints Level 1 Constraints Consistent numbering i j k l McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

15 Pairwise Model Constraints Feasible topologies 1 1 m n p

16 Pairwise Model Objective Level 2 Objective Maximize the sum of predicted wheel probabilities McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

17 Pairwise Model Constraints Level 2 Constraints Require at most one wheel contact for a specified primary contact Level 2 Aim Distinguish between equally likely Level 1 predictions Increase the total number of contact predictions McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

18 Results 2-3 helix bundles PDB:1mbh in PyMol PDB:1nre in PyMol McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

19 Results 1nre Contact Predictions subtract 0, max_contact 2 PRIMARY Contact PRIMARY Distance WHEEL Contact WHEEL Distance Helix-Helix Interaction 25L-49L L-45L A 28L-83V P 45L-85L L-81L A 51I-77L A

20 Results 1hta Contact Predictions subtract 0, max_contact 1 PRIMARY Contact PRIMARY Distance Helix-Helix Interaction 5I-28L A 46L-62L A

21 Results Contact Prediction Summary McAllister, Mickus, Klepeis, Floudas. Proteins. 2006, 65:

22 Summary Thesis: Topology of alpha helical globular proteins is based on inter-helical hydrophobic to hydrophobic contacts Validated on alpha helical globular proteins

23 Outline Protein structure prediction overview Predicting α-helical contacts Probability development Model Results Predicting α-helical contacts in α/β proteins Distance bounding Model Results Structure prediction of α-helical proteins Framework Results

24 Overview Problem α-helical topology prediction of globular α/β proteins Approach Predict/determine the secondary structure and β-sheet topology Establish bounds on inter-residue distances Apply novel optimization model (MILP) to maximize hydropobocity of interhelical interactions Helix Prediction -Detailed atomistic modeling -Simulations of local interactions (Free Energy Calculations) β-sheet Prediction -Novel hydrophobic modeling -Predict list of optimal topologies (Combinatorial Optimization) McAllister and Floudas. 2008, In preparation.

25 Establishing Distance Bounds Approach Use secondary structure location and β-sheet topology Develop local and non-local bounds (PDBSelect25) Local bounds based on residue separation and secondary structure

26 Establishing Distance Bounds Non-local Extended β-contacts Cross β-contacts

27 Tightening Distance Bounds Use of triangle inequality relationships Model is iteratively applied to determine tightest distance bounds

28 Objective Function Maximize number of hydrophobic interactions between α-helices and hydrophobicity PRIFT scale * Number Hydrophobicity α values are weight factors * Cornette et al. J Mol Biol. 1987, 195:

29 Constraints Residue contact constraints Residue i forms at most one contact with residue in helix n Residue i forms at most two contact Additional constraints limiting the size of allowed helix kinks Similar to constraints for α-helical topology prediction of α-helical proteins

30 Constraints Residue contact constraints Disallow (i,i+2), (i,i+5), and (i,i+6) residue pairs from both having contacts with helix n These residues exist on opposite faces of a helix i i+2 i+5 i+6

31 Constraints Helix contact constraints Maximum of 2 helix-helix contacts for a helix Only 1 helix-helix interaction direction Ensure feasible topologies Similar to constraints for α-helical topology prediction of α-helical proteins

32 Constraints Relating residue contacts to helix contacts Ensure consistent numbering i j k l

33 Constraints Relating distances to residue contacts If residue pair (i,j) forms an inter-helical contact, then d ij falls within contact distance If residue pair (i,j) does not form an inter-helical contact, d ij falls beyond contact upper bound

34 Constraints Distance constraints Satisfies initial bounds Satisfies triangle inequality constraints

35 Constraints Distance constraints Restrict distances based on right angle interaction assumption If residue pair (i,k) is an interhelical contact, line segment (i,k) is perpendicular to line segment (i,j) Relationship is used to bound the distance d jk

36 Results 1dcjA 1o2fB

37 Results 1bm8

38 Results - Summary Best average contact distance for 11 of 12 proteins in the test set was less than 11.0 Angstroms

39 Results CASP7 T350 Prediction with the optimal topology is shown

40 Outline Protein structure prediction overview Predicting α-helical contacts Probability development Model Results Predicting α-helical contacts in α/β proteins Distance bounding Model Results Structure prediction of α-helical proteins Framework Results

41 ASTRO-FOLD for α-helical Bundles Helix Prediction -Detailed atomistic modeling -Simulations of local interactions (Free Energy Calculations) Interhelical Contacts -Maximize common residue pairs -Rank-order list of topologies (MILP Optimization Model) Loop Structure Prediction -Dihedral angle sampling -Discard conformers by clustering (Novel Clustering Methodology) Derivation of Restraints -Dihedral angle restrictions -C α distance constraints (Reduced Search Space) Overall 3D Structure Prediction -Structural data from previous stages -Prediction via novel solution approach (Global Optimization and Molecular Dynamics) McAllister, Floudas. Proceedings, BIOMAT 2005.

42 Derivation of Restraints Dihedral angle restraints For residues with α-helix or β- sheet classification For loop residues using the best identified conformer from loop modeling efforts Distance restraints Helical hydrogen bond network (i,i+4) α-helical topology predictions β-sheet topology predictions Klepeis, JL and Floudas, CA. Journal of Global Optimization. (2003)

43 Constrained optimization Problem definition Atomistic level force field (ECEPP/3) Distance constraints

44 Tertiary Structure Prediction Hybrid global optimization approach αbb deterministic global optimization Conformational Space Annealing (CSA) Modifications/Enhancements Improved initial point selection using a torsion angle dynamics based annealing procedure from CYANA Inclusion of a rotamer optimization stage for quick energetic improvements Streamlined parallel implementation

45 αbb Global Optimization Based on a branch-and-bound framework Upper bound on the global solution is obtained by solving the full nonconvex problem to local optimality Lower bound is determined by solving a valid convex underestimation of the original problem Convergence is obtained by successive subdivision of the region at each level in the brand & bound tree Guaranteed ε-convergence for C 2 NLPs Floudas, CA and co-workers, Adjiman, CS, et al. Computers and Chemical Engineering. (1998a,b)

46 Conformational Space Annealing Induce variations Mutations Crossovers Subject to local energy minimization Anneal through the gradual reduction of space Scheraga and co-workers, Lee, JH, et al. Journal of Computational Chemistry. (1997)

47 Rotamer Side Chain Optimization Side chain packing is crucial to the stability and specificity of the native state Rotamer optimization is a quick way to alleviate steric clashes Better starting point for constrained nonlinear minimization

48 Torsion Angle Dynamics Why? Difficult to identify low energy feasible points Fast evaluation of steric based force field Unconstrained formulation with penalty functions Implemented by solving equations of motion as preprocessing for each constrained minimization Guntert, P, et al. Journal of Molecular Biology. (1997) Klepeis, JL, et al. Journal of Computational Chemistry. (1999) Klepeis, JL and Floudas, CA. Computers and Chemical Engineering. (2000)

49 Hybrid Global Optimization Algorithm All secondary nodes begin performing αbb iterations Once the CSA bank is full, CSA takes control of a subset of secondary nodes αbb CSA work Torsion Rotamerangle optimization dynamics αbb Control CSA Control Idle Work Rotamer Minimization optimization of CSA trial conformation Minimization of lower bounding function Minimization of upper bounding function CSA αbb Idle work control Maintains Performs shear CSA list of bank movements lower bounding and subregions perturbations on CSA Maintains Tracks structures overall queue upper of αbb and minima lower bounds for bank increases Handles Only Defines executed branching bank updates during directions idle time of primary processor Sends and receives work to and from CSA αbb work nodes Primary processor Secondary processors McAllister and Floudas. 2007, Submitted for publication.

50 Results Tertiary Structure Prediction PDB: 1nre Energy RMSD 6.63 Energy RMSD 3.52 Lowest energy predicted structure of 1nre (color) versus native 1nre (gray) Lowest RMSD predicted structure of 1nre (color) versus native 1nre (gray)

51 Results Tertiary Structure Prediction PDB: 1hta Energy RMSD 6.70 Energy RMSD 2.58 Lowest energy predicted structure of 1hta (color) versus native 1hta (gray) Lowest RMSD predicted structure of 1hta (color) versus native 1hta (gray)

52 Results Blind Tertiary Structure Prediction (Collaboration with Michael Hecht) S836 Energy RMSD 2.84 Energy RMSD 2.39 Lowest energy predicted structure of s836 (color) versus native s836 (gray) Lowest RMSD predicted structure of s836 (color) versus native s836 (gray)

53 Conclusions Two novel mixed-integer linear programming models were developed for α-helical topology prediction in α-helical proteins PRIMARY and WHEEL contacts For all 26 test α-helical proteins, best average contact distance predictions fell well below 11.0 Å A novel mixed-integer linear programming model was aslo developed for α-helical topology prediction in α/β proteins For 11 of 12 test α/β proteins, best average contact distance predictions fell below 11.0 Å Topology predictions were useful for restraining the tertiary structures during global optimization and obtaining a near-native predictions in a blind study

54 Acknowledgements Funding sources National Institutes of Health (R01 GM52032) US EPA (GAD R )* *Disclaimer: This work has not been reviewed by and does not represent the opinions of the funding agency.

55 Questions