Transcript binding site

Computers in Medicinal
Chemistry
Chapter 17
De novo design
The design of novel lead compounds based on the structure of the binding site
Procedure
•Crystallise protein + ligand
•Download for molecular modelling studies
•Identify the binding site
•Identify binding interactions
•Identify other binding regions in the binding site
•Remove ligand in silico
•Design ligands to fit and bind to the binding site in silico
•Calculate strength of binding
•Synthesise and test promising structures
•Optimise by structure-based drug design
Molecular and Quantum Mechanics
Molecular Mechanics
Laws of Classical Physics
dependant on Force Fields
Quantum Mechanics
Follows Quantum Physics
ab inito
Semi- empirical
Choosing a method
Drawing Structures
ChemDraw
3D viewers
Chem 3D
Alchemy
Sybyl
Manual vs Automated Studies
Manual
•Operator directs study
•Allows input of designer’s ideas
•Useful for identification of a single lead compound
•Slow and limited to designer’s originality
Automated
•Program is automated
•No bias introduced by operator
•Useful for generating large number of possible lead compounds
•Produces novel structures
•May generate impractical structures for synthesis
•Scoring structures for binding strengths is unreliable
Points to consider
•Avoid designing structures that fill the binding site
- experimental error in crystal structure
- different binding modes from predicted
- space needed for drug optimisation
•Flexible molecules better than rigid
- allows alternative binding modes
•Synthetic feasibility
•Stable conformations
•Energy of desolvation
•Structural differences in targets from different species
Docking
•The fitting of a drug into a binding site
•Considerations - the size, shape and binding interactions of the drug
•Carried out using molecular modelling programs
•Carried out manually or automatically
•Latter is preferable for docking experiments on large numbers of compounds
•Latter avoids any ‘bias’ on part of operator
•Useful for virtual screening of compound libraries
•Identifies structures likely to bind to target binding site
•Docking algorithms are required to dock molecules and ‘score’ how effectively they bind
•Depends on how well they fit and how many binding interactions are formed
•Compromise between speed and accuracy
Methods of Docking
Rigid target and rigid ligand
•Simplest and quickest method
•Acceptable if active conformation of ligand is known
Rigid target and flexible ligand
•More complex but feasible
Flexible target and flexible ligand
•Very complex and expensive on computer time
•Not practical
The target binding site
•Identify the structure of the protein by X-ray crystallography
•Identify the binding site
•Identify the amino acids lining the binding site
•Define the molecular surface of the binding site
•Molecular surface defined by van der Waals radius of atoms
• extensive surface area
• includes areas inaccessible to ligand
Inaccessible surfaces to ligands
Binding site
Protein
van der Waals radii of atoms lining
the binding site
Molecular surface based on van der
Waals area of atoms
The target binding site
Connolly Surface
•Defined as the van der Waals surface accessible to a solvent molecule
•Determined by ‘rolling’ a sphere over the van der Waals surface
•Sphere represents a water molecule
•Surface defined by convex and concave regions
•Convex regions - areas where probe makes contact with van der Waals surface
•Concave regions - re-entrants representing how far the probe accesses space
between atoms lining the binding site
•Surface represented by regularly distributed dots
A
Binding site
Probe
atom
B
Binding site
convex
surface
Protein
re-entrant
Protein
C
Binding site
convex
surface
re-entrant
Protein
Connolly Surface
Rigid docking by shape
Dock Program
•Rigid docking by shape complementarity
•Shape of ligand matched to available space in binding site
•Connolly surface of binding site defined by dots
•Space available defined by spheres
Creation of spheres
•Spheres are created that touch two dots of the Connolly surface
•For n dots, there are n-1 spheres in contact with each specific dot
•The sphere of smallest radius is chosen for each dot
•The sphere of largest radius is chosen for the surface associated with a specific
atom
•The number of spheres remaining = number of atoms lining the binding site
•Each sphere is now treated as a pseudoatom, allowing an overlay between
ligand atoms and the pseudoatoms.
Rigid docking by shape
Dock Program
Ligand
N
Me
N
Me
Atoms
of ligand overlaid with
Pseudoatoms
pseudoatoms
Rigid docking by shape
Dock Program
•Overlay carried out by systematic matching operation
•Distance matching or clique searching
•Distances between ligand atoms are measured
•Distances between centres of pseudoatoms are measured
•Identify sets of ligand atoms which can match up with sets of pseudoatoms
•Carry out docking process
8
8
9
9
3
2
6
4
7
3
5
B
2
10
1
1A
D
4
6F
5
C
10E
LIGAND
BINDING SITE
Compatible atom/sphere pairs = 1A, 6F, 7G, 10E
7G
Rigid docking by shape
Dock Program
•Docking procedure carried out after the matching operation is completed
•Docking procedure repeated for other possible matching pairs of atoms and
spheres
•Several solutions possible - require filtering process
Rigid docking by shape
Dock Program
Filtering process removes any solutions with steric clashes between unmatched
ligand atoms and the binding site surface
Steric clash
8
9
11
3
2
1
A
B
6F
D
4
C
7G
5
10E
Unacceptable binding mode
Rigid docking by shape
Dock Program
•Acceptable solutions are optimised to fine tune the position of the ligand
•Minimises bad steric interactions and optimises binding interactions
•Binding energy is measured
•Grids and ‘look-up’ tables are used to speed up calculations
•Docking by shape is ideal if the ligand fills up most of the space available
•Less satisfactory for ligands that are small with respect to space available
Rigid docking by matching H-bonding groups
Notes
•Docking is carried out by ‘clique technique’
•Hydrogen bonding groups in ligand are matched up with hydrogen bonding groups in
binding site
•Complementary hydrogen bonding groups must be correctly separated from each other
•Complementary hydrogen bonding groups must have correct orientation with respect to
each other
•Interaction points are positions in space within the binding site where ligand atoms can
be positioned to satisfy above criteria
Rigid docking by matching H-bonding groups
Definition of interaction points
Identify
H-bonding
group
O
Binding site
Construct
surrounding
sphere
Define
interaction
sites
Binding site
.
. .
. .
. .
. ..
.
. .
. .
. . .
. .
. .. .
. .
Binding site
.
. .
.
. .
. .
. .
. . .
. .
a
Filter
points
Binding site
Notes
•Sphere surface represents the optimum distance for H-bonding
•Sphere is covered with uniformly separated points that define the surface
•Points are filtered to include only points at the allowed bonding angle
•Remaining interaction points are used for matching to ligand atoms
Rigid docking of flexible ligands
FLOG program
Notes
•Flexible Ligands Orientated on Grid
•Generates conformational libraries of ligands called Flexibases
•10-20 Conformations generated per ligand
•Each conformation is tested in the docking process
•No guarantee that active conformation is included in Flexibase
Docking of flexible ligands - anchor and grow
Introduction
•Conformations are generated as part of the docking process
•The ligand is fragmented
•A rigid anchor is identified and docked
•The ligand is reconstructed or grown
Docking of flexible ligands - anchor and grow
Directed Dock and Dock 4.0
•Algorithm identifies rotatable bonds in ligand
•Defines flexible and rigid regions
H3C
H3C
H3C
CH3
O
N
O
N
Docking of flexible ligands - anchor and grow
Directed Dock and Dock 4.0
•Molecule is split into fragments
•Identify an anchor - the most rigid fragment
•Dock anchor by shape complementarity
H3C
CH3
N
H3C
H3C
N
CH3
Fragment
C
O
C
C
O
H3C
H3C
Dock
C
C
N
C
N
Anchor
O
O
N
Anchor
Docking of flexible ligands - anchor and grow
Directed Dock and Dock 4.0
•Grow the ligand
•Fragments are added sequentially to the anchor
•Torsion angles are varied systematically after each fragment is added
•Energy minimisation carried out
•Binding interactions measured for each construct
•Pruning process
•Fragments are added in layers
Layer 2
H3C
CH3
Layer 1
N
Anchor
C
C
O
H3C
C
O
N
Docking of flexible ligands - anchor and grow
FlexX
•Anchor is docked by possible binding interactions
•Interaction points are built round each binding group in the binding site
•Match anchor atoms to interaction points
•Clique technique
•Anchor atom and interaction point must have binding compatibility
•Three matched pairs of binding interactions are required for docking
•Complementary pharmacophore triangles for anchor and binding site are
required for docking
Docking of flexible ligands - anchor and grow
FlexX
HBA
HBA
Anchor
N
H
HBD
HBD
H
H O
HBD
Hydrophobic
group
Binding site
HBA
HBA
Pharmacophore
triangle
for the anchor
Hydrophobic
group
Hydrophobic
interaction
Hydrophobic
interaction
Docking of flexible ligands - anchor and grow
FlexX
N
H
HBD
HBD
H
H O
HBD
Binding site
Anchor
Hydrophobic
interaction
Hydrophobic
interaction
Docking of flexible ligands - anchor and grow
FlexX
Docking of flexible ligands - anchor and grow
FlexX
Anchor
Anchor
Construct
interaction
surfaces
Identify
interaction
centres
..
.. . .. . ..
Interaction
Anchor
. . . .. . .
..
Binding site
Binding site
Binding site
Notes
•Program also builds interaction points round anchor points
•Checks angle requirements for H-bonding with respect to anchor atoms
•Anchor atom is placed on an interaction point for the binding site
•Binding site atom is placed on an interaction point for the ligand
Docking of flexible ligands - anchor and grow
FlexX
•Several docking solutions are possible for the anchor
•Cluster the solutions and choose a representative structure for each cluster
•Attach fragments to anchor at discrete torsion angles
•Anchor is chosen manually - unsuitable for automatic docking
•Vast number of pharmacophoric triangles are possible for binding site
Docking of flexible ligands - anchor and grow
Hammerhead program
•Designed for automatic docking studies on large databases
•Probes are used to identify optimum locations for binding interactions
•H atoms are used to probe for hydrophobic interactions
•C=O and N-H fragments are used as probes for H-bonds
•Probes act as targets for docking
•Matching operation matches atoms of molecular fragment with probes
•Steric and chemical complementarity are both used
•Docking follows the matching process
Docking of flexible ligands - anchor and grow
Hammerhead program
BINDING SITE
C=O probe
N-H probe
H probe
Docking of flexible ligands - anchor and grow
Hammerhead program
•Ligand is split into fragments
•Fragments contain an atom or bond shared with another fragment
R
Shared
bond
Fragment
R
R
R
Shared
atom
R
R
Fragment
R
R
Docking of flexible ligands - anchor and grow
Hammerhead program
•Fragments are docked and scored
•High scoring fragments are defined as ‘heads’
•Heads are used as anchors
•Remaining fragments are defined as ‘tails’
•Reconstruction is carried out on each potential anchor
•Tail fragment is docked into the area round the anchor with the shared atom or
bond aligned
•Merging of fragments involves overlaying shared atoms or bonds
•Anchor remains fixed
•Optimisation carried out at each stage
•No bias involved in choosing anchor
•Different anchors are possible
•Highest binding anchor does not necessarily produce the best docking mode
for the final ligand
Docking of flexible ligands - anchor and grow
Hammerhead program
Shared
bond
R
R
Anchor
R
Fragment
Translate fragment
to align atoms
R
Rotate fragment
to orientate bond.
Merge
R
R
Docking of flexible ligands - anchor and grow
Hammerhead program
Anchor
Shared atom
R
R
R
Translate fragment
to align shared atoms
R
Fragment
Adjust bond
angle
R
R
Docking of Flexible Ligands
Metropolis method (simulated annealing)
•Incorporates a conformational search of flexible ligands
•Slower process
•Computationally expensive
•Ligand is placed randomly in the binding site
•Monte Carlo algorithms generate different conformations
•Molecule is ‘tumbled’ in the binding site
•Binding energy of each structure is measured
•Simulated annealing identifies the best structures
•Process is speeded up by ‘look up’ tables
•Quality of results may depend on the initial placement of ligand
•Can use DOCK to identify initial binding modes then apply Metropolis method
Docking of Flexible Ligands
Genetic and evolutionary algorithms
•Incorporate a conformational search of flexible ligands
•Slower process
•Computationally expensive
•Chromosomes determine conformation, position and orientation of ligand binding site
•Mutations and cross-overs change conformations and orientation of the ligand
•Selection of best docking modes is based on interactions with the binding site
LUDI
•Automated program
•Identifies possible interaction sites in the binding site
•Fits molecular fragments to different regions of the binding site
•Links fragments
LUDI
H
O
N
X
Y
H X
O
Interaction sites
H
H
N
Bridging
N
H X
H
Y
Fragment fitting
O
O
H
O
X
N
N
LUDI
Interaction points for van der Waals interactions
Identify aliphatic
carbon
4Å
CH 3
Binding site
Construct
surrounding
sphere
CH 3
Binding site
. .
.
. . . .
.
. . CH
.
.
. 3
.
Define
interaction
sites
Binding site
.
Trim
the sphere
.
. .
.
. . .
.
CH 3
Binding site
Notes
•Non-directional interaction
•Radius of the sphere is the optimum distance for interaction
•Interaction sites are evenly distributed points on the surface of the sphere
•Trimming the sphere removes points that are too close to binding site surface
LUDI
Interaction points for hydrogen bonds
X
1.0 Å
X
H 1.0 Å
H
1.9 Å
X
180 o
180 o
H
X
H
1.9 Å
O
120 o
O
120 o
Binding site
Binding site
Required distances
Required bond angles
Notes
•Directional interactions involving an optimum distance and angle
•Identify HBAs and HBDs
•Define interaction points as a vector involving two atoms (H-X)
LUDI
Interaction points for hydrogen bonds
1.90 Å
H
O
o
180
A
1.23 Å
Y
120o
Binding site
Notes
•Directional interactions involving an optimum distance and angle
•Identify HBAs and HBDs
•Define interaction points as a vector involving two atoms (H-X)
LUDI
Molecular fragments
O
NH2
O
C
H3C
OH
H
O
N
OH
N
H
O
NO2
H
N
N
H
H
N
H
N
N
N
H
Notes
•Program has a library of fragments
•Fragments are usually rigid structures containing 5-30 atoms
•Flexible fragments are present as several conformations
•Atoms of fragments are defined for the fitting process
•Aliphatic atoms are fitted onto van der Waals interaction points
•HBAs and HBDs are fitted onto H-bond interaction points
LUDI
Fitting molecular fragments
Better
fit
O HBA
H
X
O H
X
Y
Y
N
H3C
C
CH3
O
O H
Aliphatic
O
C
CH3
CH3
H3C
Fitting process
Interaction sites
Notes
Identify best fit with maximum number of interactions
Fitting fragments
C
H
N
CH3
LUDI
Fragment bridging
H
N
H
N
H
H
Identify closest fragments
H
Identify closest hydrogens
H
N
Link points
N
Fit bridge
LUDI
Fragment bridging
Examples of bridges
O
O
H
N
O
O
S
O
O
O
O
C
O
S
N
H
N
H
O
O
SPROUT
•Automated program
•Fits fragments to interaction sites
•Interaction sites are atom-sized spheres
•Sphere represents a volume of space into which a ligand atom should fit to interact with
the binding site
SPROUT
Fragments
•Fragment templates represent molecular fragments
•Template is defined by vertices and edges
•Vertex represents a generalised sp, sp2 or sp3 hybridised atom
•Edge represents a single, double or triple bond
•Fragment template can represent several molecular fragments
•Reduces number of stored fragments
•Increases efficiency of search
sp3
sp
sp
3
sp
3
sp
O
2
2
H
N
O
=
O
sp3
Fragment template
H
N
Molecular fragments
etc
SPROUT
Generation of structures
•Fragment templates are selected randomly
•Vertices are chosen randomly and fitted to the target sphere
•No consideration of binding interactions at this stage
Aromatic
HBA
H
N
H
HBA
O
O
H
H
Hydrophobic
Interaction sites
Fitting fragments
N
SPROUT
Generation of structures
•Fragment templates are grown towards each other and linked
•Capable of bridging interaction sites that are far apart
•Molecular template is converted to a molecule
•Atoms are added to allow required interactions
•Large number of molecules are possible for each template
H
O
N
O
O
H
Growth and linkage
H
O
Molecular structure
H
N
SPROUT
Generation of structures
Ability to modify unrealistic features
Interaction
point
for HBD
3
sp3
sp3
2
sp
sp
sp3
sp2
Add atoms
OH
Modify
OH
O
Enol
Molecular template
O
Ketone
Carboxylic acid
SPROUT
Generation of structures
Ability to modify structures for synthetic feasibility
Modify
O
Easier to synthesise
Synthesis
Na
+
I
O
O
+ NaI
SPROUT
CAESA
•Program used to evaluate synthetic feasibility of structures
•Indicates likely starting materials
•Retrosynthetic analysis
SPROUT
Synthetic feasibility of partial structures
•Partial structures are analysed during construction process
•Allows pruning of structures
•Increases number of synthetically feasible structures generated
•Quicker method of analysis required than CAESA
•Identify molecular features within the partial structure
•Identify frequency of occurrence in known structures
•Corresponds to synthetic feasibility
•Substitution patterns vary in frequency
3748
486
3288
1608
459
403
782
397
504
362
LEGEND
•Automated program
•Grid set up within binding site
•Identifies steric and electrostatic interactions
•Tabulated for different atoms at each grid point
•Allows estimation of van der Waals interactions during building process
•Single heteroatom is placed in the binding site to form a H-bond - root atom
•Atoms added one at a time - growth stage
•Random choices for atoms, connections and torsion angles
•Atoms changed to take account of grid point interactions
•Generates greater diversity of skeletons than fragment-based methods
•Evaluation of partial structures and pruning required
•Slower process than fragment-based methods
GROW
Notes
•Automated program
•Amino acids used as molecular fragments
•Structures generated limited to peptides
SYNOPSIS
Notes
•Automated program
•Fragment-based process
•Generates synthetically feasibly structures
•Synthetic rules incorporated into structure building process
•Fragments must be commercially available
•Fragments only linked if synthetically feasible
•Program provides possible synthetic route