Joana Pereira!

Lamzin Group!
EMBL Hamburg, Germany

Small molecules
How to identify them
(with ARP/wARP)

YEARS I 19742014
Why ligands?
They are small molecules that form a complex with a biomolecule to serve a biological purpose:

Endogenous substrates and analogues thereof!

Known drugs and biological modulators!
Probe molecules!
Buffer constituents!
etc.

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Structure based drug design
Relies on knowledge of the three dimensional structure of the biological target:

During lead optimization, numerous structures

need to be determined representing the same
protein complex with a variety of different ligands
!
MX is used as the primary experimental technique
to determine protein-ligand complex structures
and provide detailed structural information!
requires rapid and automated ligand
identification and construction!
!
Elaboration of a lead to produce a drug depends
on accurate knowledge of ligand binding modes!
Purple: A lead for the inhibition of PDE4D; ! can also be aided by automated ligand
Yellow: the end result from SBDD. ! building procedures
Card et al. (2005) Nat. Biotechnol. 23: 201-207

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The task at hand
To find ligand density and build it!

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Fitting a ligand
We have:

electron density map (good

resolution)!
!
+ protein model !
shows excess density not
accounted for by protein!
!
difference density map
to fit the appropriate ligand into

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Fitting a ligand
Challenges:

different resolutions and data quality!

different ligand complexity /
topology!
partial disorder of a ligand!
different ligands at the same site?

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The protocol of a modelling task
What we know at the beginning will determine the protocol:

Binding site

Known Unknown
1 density cluster 1 ligand
Ligand identity

Known

X
Unknown

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The protocol of a modelling task
What we know at the beginning will determine the protocol:
N density clusters

Binding site

Known Unknown
1 ligand
Ligand identity

Known

X
Unknown

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The protocol of a modelling task
What we know at the beginning will determine the protocol:
N ligand candidates

Binding site

Known Unknown
1 density cluster
Ligand identity

Known
Unknown

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The protocol of a modelling task
What we know at the beginning will determine the protocol:
N density clusters N ligand candidates

Binding site

Known Unknown
Ligand identity

Known
Unknown

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Researchers needs #1

! I!am!trying!to!solve!a!structure!of!a!protein!with!
some!inhibitor.!I"want"to"know"how"I"can"put"in"
my"inhibitor"in"the"density"map"of"the"data"I"got.!I!
can!see!some!density!in!the!active!site!where!the!
!inhibitor!should!be!
! I!am!not!sure!of!how!to!do!it.

Which case is this?

1 density cluster 1 ligand

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand fitting with ARP/wARP
Organised as a pipeline of core modules for specific sub-tasks following an intuitive
approach

Prepare Identify ligand binding site and /or

ligand!
Sparse density map and generate
ligand topology

Construct ensemble of ligand

Construct
models in plausible conformation to
fit sparsed density

Refine ligand coordinates to satisfy

Refine geometric constraints and
maximise fit to density!
Choose best model

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand
To address the different ligand sizes and complexities encountered!
to increase the robustness of the software!
!
We use 2 separate construction methods:

Label swapping on the sparsed grid

Metropolis search in conformation space

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand: graph search
Uses the sparse grid representation of the electron density at the chosen
binding site & topology of known ligand

Knowledge about ligand Knowledge about grid

Connectivity! Connectivity!
Distances! Distances!
Angles! Density!
Chirality NO IDENTITIES

4
1
2 3

8 7 5
6
10
9

16 13
25
24 12
33 11
23
19 14
34 27 18
42 15
44 26 17
21
32
46 43 28 35 22
50 40
45 36 20
51
37 39 31
49
70 62 29
61 59 47 41 30
58
60 56 48 38
55
57 64
54 53
69 66
73 67 65 52
72
79 68
76 63
74 77
71
78
81
75
85
86 87 84
80
82
90 88 83
89
91 92

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand: graph search

Label swapping!
Ligand is expanded on the sparse density, preserving connectivities!
Every dummy atom of the sparse grid is tried as a start point!
An exhaustive graph search is performed!
Models are scored by their fit to density and expected
stereochemical features

4
1
2 3

8 7 5
6
10
9

16 13
25
24 12
33 11
23
19 14
34 27 18
42 15
44 26 17
21
32
46 43 28 35 22
50 40
45 36 20
51
37 39 31
49
70 62 29
61 59 47 41 30
58
60 56 48 38
55
57 64
54 53
69 66
73 67 65 52
72
79 68
76 63
74 77
71
78
81
75
85
86 87 84
80

Start Atom
82
90 88 83
89
91 92

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand: graph search
Index of starting grid point

No of ligand atoms placed

Dead!
branches

4
1
2 3

8 7 5
6
10
9

16 13
25
24 12
33 11
23
19 14
34 27 18
42 15
44 26 17
21
32
46 43 28 35 22
50 40
45 36 20
51
37 39 31
49
70 62 29
61 59 47 41 30
58
60 56 48 38
55
57 64
54 53
69 66
73 67 65 52
72
79 68
76 63
74 77
71
78
81
75
85
86 87 84
80

Start Atom
82
90 88 83
89
91 92

Surviving start points

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand: metropolis search
Perform a random walk in parameter space biased towards the optimum
of a score function!
Parameter space: position, orientation and conformation!
Score function: pseudo map correlation!
Advantage: less degrees of freedom
Rigid groups Rotatable bonds

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Constructing the ligand: metropolis search
Evolution of a model during a metropolis optimisation

A B C D

FMN to 1jqv at 2.1: 0.23 rmsd.

Final result: 9000 moves + refinement

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand building in the CCP4 interface
Ligand building GUI as a part of ARP/wARP in CCP4i

Protocol

Diffraction data

Protein without!
ligand

Template ligand

For batch jobs - use the command line...

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand building from the command line
For batch jobs - use the command line...

to launch job with

default scenario:!
only 3 files
needed

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand building within ArpNavigator
Load the density map and the protein. Load the ligand you want to build.!
Mark a density blob, detach and fit the ligand right there. Easy!
Right Click and
choose this!

Select Mark
Density Region
under Options

Detached Ligand Fit Ligand

Detach model
from crystal
frame

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand building within ArpNavigator

Fast and automated fitting of ligands into density not

accounted for by the protein structure!

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Researchers needs #2

! Is!there!a!simple!way!to!determine!whether!ligand!is!bound!or!not!by!
comparing!the!diraction!patterns!between!ligand;free!(structure!
known)!and!ligand;soaked!protein!crystals?!I!would!like!to!solve!the!
ligand!bound!protein!structure,!but!before!I!do!so,!I"have"to"nd"out"
if"the"ligand"is"actually"bound"and"if"so,"where.!
! Thank!you!very!much!!!
! ! !
Which case is this?

N density clusters 1 ligand

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Automatic binding site identification
The difference density map from low to high contour threshold

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Automatic binding site identification
Capturing the dependence of the difference density map on changing the
contour thresholds: the fragmentation tree
105

104 Deposited!
Selection
Ligand
Cluster volume [3]

103

102

101

100

10-1
1 2 3 4 5 6
Contour threshold []

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Shape features to measure similarity
An electron density map calculated from the ligand to be fit is compared to each
potential density cluster:
Find the matching cluster for this ligand!

123456

1) Surface to volume ratio !

123456 2) Bounding box limits!
3) Moments of inertia!
4) Rotation match score!
5) Eigenvalues!
6) Distance histogram!
123456
7) Geodesic distance histogram

Decision is based on feature vectors

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Locating the binding site in ArpNavigator

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Researchers needs #3

! We!are!working!on!protein/inhibitor!complex!structure.!However,!we!
did!nd!a!strange!density!at!the!active!site,!it!looks!really!like!GSH,!the!
natural!co;enzyme!of!this!protein.!We!tried!to!use!very!simple!solution!
to!get!crystal!then!exclude!the!possibility!of!buer!molecules,!but!that!
density!is!always!there.
I"want"to"identify"this"ligand."Are"there"any"methods"to"do"this"work?"

Which case is this?

1 density cluster N ligand candidates

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Its all in the shape, baby!

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Shape features to measure similarity
To assign a likelihood to a ligand knowing the binding site
Final ligand ranking

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
High throughput ligand identification
Projection/
Segmentation
normalisation

Density map (or otherwise

surface of protein pocket) Calculation of shape
descriptors

Ranking Comparison

Ligand 84 of the most

common ligands
in crystallography
Database

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand identification in ArpNavigator

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Researchers needs #4

! I!have!a!ligand!in!the!binding!pocket!and!mediocre!
data.!The"core"of"the"ligand"is"well"dened,"however"
there"are"chains/tails"of"the"ligand"which"are"
not.The!question!here!is!how!the!chains/tails!should!
be!modelled!(if!at!all)

Which case is this?

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Partial Disorder, Partially Occupied Ligands
Deposited in PDB
Built automatically!
with ARP/wARP when!
the full ligand is given

Compound chosen in!

SAM in 1v2x at 1.5
cocktail case
Artificial Cocktail
!34

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Partial ligand building in ArpNavigator

Complete partial ligand building in under 2 minutes !!!

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Success rate
Tested on > 20k PDB entries from EDS!
X-ray resolution limits: 1.0 to 3.0 !
Building the largest fully occupied ligand!
Correctness criterion: r.m.s.d. < 1.0

For ARP/wARP 7.2

56 atoms,! 7100 atoms,! 1040 atoms,! 2040 atoms,!

any map corr. any map corr. map corr. > 80% map corr. > 80%

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Success criteria: r.m.s.d < 1

Correct Correct

rmsd=0.4 rmsd=1.0

Incorrect Incorrect

rmsd=1.8 rmsd=1.5
In yellow: the deposited ligand

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
Ligand building methods

Sparse grids Conformational fit Fine skeletons

High-throughput ligand identification using shape

descriptors!

Building models which do NOT have repetitive

motifs like peptides or nucleotides

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014
The people - The power!

Collaborators!
EMBL Hamburg: Gleb Bourenkov,
Santosh Panjikar!
MRC Laboratory, Cambridge :
Garib Murshudovs group!
Rutherford Appleton
Laboratory: CCP4 team

Funding

TimJoana
Wiegels
Pereira 09.04.2014
30.03.12
YEARS I 19742014