Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
Computational Molecular Biology
Protein - Ligand
And Protein - Protein Docking Methods
Prof. Alejandro Giorge1 Dr. Francesco Musiani
Part 1: Protein - Ligand Docking Methods
What is the protein-ligand docking (molecular docking)
Goal: Given a protein structure, predict its ligands and where they bind
Applications:
v Function prediction v Drug design v Mechanisms
Protein-ligand docking: QUESTIONS
v Where will the ligand bind?
v Which ligand will bind?
v How will the ligand bind?
v When?
v Where?
v Why?
In other words: given a protein and a ligand, determine the poses and conformations
of the ligand minimizing the total energy of the protein-ligand complex
In practice…
Challenges…
Predicting ligand-binding energies by searching in the space
of possible poses and conformations
Challenges…
Predicting ligand-binding energies by searching in the space
of possible poses and conformations
Relative position (3 degrees of freedom) Relative orientation (3 degrees of freedom)
Rotatable bonds in ligand (N degrees of freedom)
Challenges…
Predicting ligand-binding energies by searching in the space
of possible poses and conformations
Relative position (3 degrees of freedom) Relative orientation (3 degrees of freedom)
Rotatable bonds in ligand (N degrees of freedom) Rotatable bonds in protein (M degrees of freedom)
Challenges…
Searching poses & conformations v Systematic search v Molecular dynamics v Simulated annealing v Genetic algorithms
v Incremental construction v Rotamer libraries
Scoring functions
v Molecular mechanics v Empirical functions v Knowledge-based
Results & Discussion
v Clustering
Intra- and Inter-molecular forces
Intramolecular Forces (covalent) v Bond lengths v Bond angles
v Dihedral angles
Intermolecular Forces (non covalent) v Electrostatics
v Dipolar interactions v Hydrogen bonding v Hydrophobicity v van der Waals v Pi-stacking
Intra- and Inter-molecular forces
Coulombic interactions…
Arg
Ligand
Hydrogen bonds
Trypsin and substrate Mannitol Dehydrogenase and NAD
Salt bridges
Salt bridges and ligand binding
Binding of napsagatran to thrombin
Pi-Stacking Interactions: end to face
Pi-Stacking interactions: face to face
Cation-pi interactions…
Interactions with metal ions
Hydrophobicity
Binding pocket becames «interior» upon compexa6on with ligand
Big penality: charged or polar groups buried but umpaired
Energe6c contribu6on is propor6onal to the size of the surface buried upon ligand binding (e.g. –CH3 group (25 Å2): 3 to 6 kcal/mol)
Solvation and desolvation
ΔG (binding, vacuo)
ΔG (binding, soluFon)
ΔG (soluFon (E+I))
ΔG (soluFon (EI))
Solvation and desolvation
Solvation and desolvation
v Rupture of H-‐bonds within water matrix v Reform H-‐bonds
v Reorganize water molecules at surface v Bury a hydrophobic pocket surface v Loose degrees of freedom v Some water molecules released
ΔG (binding, vacuo)
ΔG (binding, soluFon)
ΔG (soluFon (E+I))
ΔG (soluFon (EI))
Scoring functions
Molecular mechanics force fields: • CHARMM [Brooks83] • AMBER [Cornell95]
Empirical methods:
• ChemScore [Eldridge97] • GlideScore [Friesner04]
• AutoDock [Morris98]
Knowledge-based methods • PMF [Muegge99]
• Bleep [Mitchell99] • DrugScore [Gohlke00]
Empirical scoring functions
Empirical scoring functions
Macromolecular docking: empirical scoring function
Van der Waals
H-‐bond
ElectrostaFcs
SolvaFon
Torsional angles
VdW H-‐bond Elec
Energy
Distance Distance Distance
Computing scoring functions
Computing scoring functions
v Systematic search
v Molecular dynamics
v Simulated annealing
v Genetic algorithms
v Incremental construction
v Rotamer libraries
Searching poses & conformations
Systematic search
Uniform sampling of search space: Relative position (3 Degrees of Freedom (DoF))
Relative orientation (3 DoF) Rotable bonds in ligand (m DoF) Rotable bonds in protein (n DoF)
Search space dimensions: 3 + 3 + m + n
Systematic search
Uniform sampling of search space: Exhaustive, deterministic
Quality dependent on granularity of sampling Feasible only for low-dimensional problems
Simulated annealing
Monte Carlo search of parameter space: v Start from a random or specific state (position, orientation, conformation)
v Make a random state changes, accepting up-hill moves with probability dictated by “temperature”
v Reduce temperature after each move
v Stop after temperature gets very small
Genetic algorithm
Genetic search of parameter space: v Start with a random population of states
v Perform random crossovers and mutations to make children
v Select children with highest scores to populate the next generation
v Repeat for a number of iterations
Part 2: Protein - Protein Docking Methods
Just to fix some ideas…
Basis of protein –protein complex formation
- Shape of the interacFon surfaces
-‐ ElectrostaFcs charaterisFcs of surface residues -‐ FuncFonal residues
Macromolecular docking
The term macromolecular docking includes several computational techniques which have the aim of calculate models of the complexes between two or more macromolecules (protein-protein, protein-DNA, protein-RNA, etc.) Objective: prediction of the tridimensional structure of a complex between two marcomolecules. Techniques: - Rigid docking - Flexible docking
Macromolecular docking
Atomic coordinates of protein A
Atomic coordinates of protein B
DOCKING
Protein complex model structure
Macromolecular docking: methods
Macromolecular docking algorithms Are characterized by four steps: 1-‐ calcula6on of an appropriate representaFon of the macromolecules together with the defini6on of the degrees of freedom of the calcula6on; 2-‐ an algorithm able to explore the space of conforma6ons with the highest possible completeness and efficiency; 3-‐ a scoring funcFon able to evaluate the quality of the predic6ons 4-‐ a clustering algorithm
Macromolecular docking: methods
Systematic search: the two macromolecules are calculated in as many orientations as possibile. Guided search: complex formation is guided by an appropriate scoring function.
Knowledge-based search: similar to guided search, but the calculation makes use of external information (i.e. experimental data, bioinformatic predictions, etc.) to guide the calculation.
Macromolecular docking: general scheme
Protein representaFon using the
molecular surface
Different probes produce Different surfaces
Coordinates of macromolecules A and B
Representations of A and B
Exploration of conformational space
Candidate complexes
AB model complex
Refining
Macromolecular docking: general scheme
Case 1: sistemaFc search
The conformaFonal space is divided into segments.
This can be achieved with a grid representaFon of the space
Coordinates of macromolecules A and B
Representations of A and B
Exploration of conformational space
Candidate complexes
AB model complex
Refining
Macromolecular docking: scoring function
Macromolecule A (ρ << 0)
Macromolecule B (0 > δ > 1)
SCORING FUNCTION
a=1
a<<0
b=1
0>b>1
c > 0
Macromolecular docking: scoring function
a=1
a<<0
b=1
0>b>1
c << 0
Macromolecular docking: general scheme
Coordinates of macromolecules A and B
Representations of A and B
Exploration of conformational space
Candidate complexes
AB model complex
Refining
Case 2: guided search
The search for the minima of the scoring funcFon is made inducing a
perturbaFon on the iniFal orientaFon.
This ‘move’ is accepted or refused on the basis of the employed
algorithm.
Macromolecular docking: empirical scoring function
Van der Waals
H-‐bond
ElectrostaFcs
SolvaFon
Torsional angles
VdW H-‐bond Elec
Energy
Distance Distance Distance
Macromolecular docking: general scheme
Both in the exploraFon of the conformaFonal space and in the refining step it is possibile to
include some external informaFon (knowledge-‐based informaFon)
Coordinates of macromolecules A and B
Representations of A and B
Exploration of conformational space
Candidate complexes
AB model complex
Refining
Macromolecular docking: Haddock
Dominguez, C.; Boelens, R.; Bonvin A.M.J.J. (2003) J. Am. Chem. Soc. 125, 1731-‐1737. de Vries, S.J. et al.(2007) Proteins: Struc. Funct. & Bioinforma;c 69, 726-‐733 (2007).
Mem
bran
e
Periplasm
Cytoplasm
Photosynthe6c reac6on center from T. tepidum (1EYS)
HP1
HP2 LP1
LP2
Nogi, T. et al. (2000) Proc. Nat. Acad. Sci. USA 97:13561.
Puta6ve HiPIP interac6on site
Macromolecular docking: examples (THC – HiPIP)
Tetra-‐heme (THC)
Macromolecular docking: examples (THC – HiPIP)
Venturoli, G. et al. (2004) Biochemistry 43:437-‐445 Ciurli, S.; Musiani, F. (2005) Photosynth. Res. 85:115-‐131
THC HiPIP
Macromolecular docking: examples (THC – HiPIP)
THC
HiPIP
Venturoli, G. et al. (2004) Biochemistry 43:437-‐445 Ciurli, S.; Musiani, F. (2005) Photosynth. Res. 85:115-‐131
THC HiPIP
Macromolecular docking: examples (THC – HiPIP)
THC
HiPIP
Venturoli, G. et al. (2004) Biochemistry 43:437-‐445 Ciurli, S.; Musiani, F. (2005) Photosynth. Res. 85:115-‐131
Hope, H.P. (2000) Biochim. Biophis. Acta 1456:5-‐26
Macromolecular docking: examples (Cytochrome f - plastocyanin)
S(Met)
Cu S(Cys)(His)N(His)N
PC
NN
NN
OH OOH O
SS
Fe
Cys 24
Cys 21
FeN
N
N
N
NH2
NH
N
His 25
Tyr 1
Cyt f
Musiani, F.; Dikiy, A.; Semenov, A.Y.; Ciurli, S. (2005) J. Biol. Chem. 280:18833-‐18841.
Macromolecular docking: examples (Cytochrome f - plastocyanin)
Bioinformatic predictions: multi-domain protein conformations
Del Campo, C.; Agries6, F.; Danielli, A.; Roncara6, D.; Musiani, F.; Ciurli, S. ; Scalrato, V. (2012) in prepara6on
Bioinformatic predictions: protein – DNA docking
Del Campo, C.; Agries6, F.; Danielli, A.; Roncara6, D.; Musiani, F.; Ciurli, S. ; Scalrato, V. (2012) in prepara6on