Transcript CBS2005_AMJ - Center for Biological Sequence Analysis

Apr 05/AMJ
Computational decision support for drug
design
Profiling of small molecule compound
libraries
Anne Marie Munk Jørgensen
Lundbeck
Apr 05/AMJ
Lundbeck’s Vision is to become the world leader in psychiatry and
neurology
Focus solely on treatment of diseases in the central nervous system
(CNS)
•depression
•Psychoses
•Migraine
•Alzheimer
•Sleep disorders
5000 people worldwide – app 800 in R & D
Outline
Apr 05/AMJ
o What is a small molecule drug?
o How can computational methods help during the
drug discovery phase?
•
Library profiling: overall characterisation
of a large pool of structures.
•
Prediction of more specific
characteristics like biological activity and ADME
properties
•
Privileged structures….
A small molecule drug
Apr 05/AMJ
… is a compound (ligand) which binds to a protein, often a receptor
and in this way either initiates a process (agonists) or inhibits the
natural signal transmitters in binding (antagonists)
The structure/conformation of the ligand is complementary to the
space defined by the proteins active site
The binding is caused by favourable interactions between the ligand
and the side chains of the amino acids in the active site.
(Electrostatic interactions, hydrogen bonds, hydrophobic
contacts…)
The ligand binds in a low energy conformation < 3 kcal/mol
Binding site complementarity
Apr 05/AMJ
HIV-Portease inhibitor
JACS,V.16,pp847 (1994)
H-bond donating
H-bond accepting
Hydrophobic
Flo98, Colin McMartin.
J.Comp-Aided Mol. Design,
V.11, pp 333-44 (1997)
Example of ligand binding
Apr 05/AMJ
1UVT,
Trombin
Inhibitor
No vacancy!
Apr 05/AMJ
Molecular factors
Apr 05/AMJ
Conformatio
n
Intramolecular
interactions
Ionization
Intermolecula
r forces
Electronic
distribution
Solubility,
Partitioning
Carrupt P-A., Testa B., Gailard P.
Boyd D.B., Lipkowitz K.B., Reviews in Computational Chemistry, Vol. 11, 1997, pp. 241-304.
Compound library profiling
Apr 05/AMJ
Analyze a pool of structures to find out how
attractive they are to us…..
• 10 years ago: Diversity + HTS
• Now: very high focus on how biologically
relevant the screening collection is.
• Computational methods to predict drug
likeness, CNS likeness….
High throughput is not enough … to get high output…..
Compound analysis
Apr 05/AMJ
Ideal
50.000
Structures:
Chemical intuition
Choosing the right descriptors is
difficult
Wolfgang Sauer, SMI 2004
Apr 05/AMJ
How we describe the structures in the
computer
Apr 05/AMJ
o Calculate a number of phys chem descriptors, like
molecular weight, nhba, nhbd, logP, SASA…..
o Describe the structures by keys….
Lipinski statistics
Apr 05/AMJ
Drug Like
1
CNS Like,
present work,
90% limit.
MW
< 500
149.4 – 446.6
# hydrogen acceptors
< 10
1-5
# hydrogen donors
<5
0-3
logP
<5
-0.3 – 4.9
# rotatable bonds
NR
0 – 8.4
Rule of 5
References
(1)
Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P. J. Experimental and computational approaches
to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev JID - 8710523 1997, 23, 3-25.
Diversity and "Chemical Space"
Apr 05/AMJ
PCA
Chemical space navigator
Apr 05/AMJ
Global Positioning System (GPS)
Chem GPS (Oprea & Gottfries,
J. Comb. Chem 2001)
We want to define the CNS ”world” – the space which
is biologically relevant when considering CNS drugs
CNS model
Apr 05/AMJ
PCA
12 descriptors 
3 components,
R2X=0.71
Blue dots define::
CNS drug space
CNS ”World”
CNS ”world” sub classes
Apr 05/AMJ
O
O
O
O
N
O
Chiral
O
O
Br
N
N
N
H
N
O
O
N
N
O
O
O
H
N
O
Model used to predict CNS-likeness
Apr 05/AMJ
N
O
N
O
F
O
I
O
O
N
O
O
N
I
I
N
O
O
O
O
Chiral
N
H
N
O
N
N
N
N
N
O
H
O
O
O
N
N
N
N
H
H
O
N
N
S
N
O
O
O
O
O
O
O
O
S
N
O
O
O
N
O
O
O
N
Structural clustering based on keys
Apr 05/AMJ
C-N
1
0.349
3
6
13
19
26
31
38
1
O
N
…01000100110001….
C=O
O
N
Cl
Cl
C=C
O
N
O
N
Cl
Similarity by Tanimoto:
O
N
Cl
Tc= Bc/(B1 + B2 – Bc)
clust_benzo (order)
O
N
O
Structural analysis
Apr 05/AMJ
o Clustering
o Virtual screening – looking for structural similar
compounds in a large pool of structures…..
Apr 05/AMJ
I have talked about overall profiling of a large
number of compounds…… in terms of CNS-likeness
… now I will turn to talk about prediction
of more specific characteristics like biological activity
and ADME properties…..
Quantitative Structure Activity Relationship
or
Quantitative Structure Property Relationship
In house QSAR study
Apr 05/AMJ
2,5
SigmaP/pi
2
O
1,5
1
pi
0,5
N
N
S
0
-0,5
O
sigmaP
0
1000
2000
3000
4000
IC50
Correlation between Glyt-1 inhibitor activity and pi
(lipophilicity) and SigmaP (electronic characteristics)
for the R substituent
R
ADME property predictions
Apr 05/AMJ
Oral absorption …depends
heavily on permeability and
Solubility… high interest in
predicting these things in silico…
Other things: Blood-brain
Barrier penetration,
clearance, Metabolism, tox…..
Aqueous Solubility
Apr 05/AMJ
QSRP model
n=775,R2=0.84, Q2=0.83
8 2D descriptors, Cerius2
Most important descriptors:
logP, hba*hbd, hba, hbd
Drugs: –6 < logS < 0;
If error of 1 log unit is OK 
model predicts 60-80% of the
compounds correctly
Journal of Medicinal Chemistry, 2003, Vol. 46, No. 17
Permeability
Apr 05/AMJ
QSRP
N= 13
R2=0.93 Q2= 0.83
Key descriptors:
PSA> Odbl >N-H >
..NPSA >SA
Polar descriptors important and
…. size matters….
Simple Rule: PSA < 120 Å2
Journal of Medicinal Chemistry, 2003, Vol. 46, No. 4
Pharmacophore modelling
Apr 05/AMJ
….. Another method of biological activity prediction…
Observations that modification of some parts of a ligand
results in minor changes of activity, whereas modifications of
other parts of the ligand result in large change of activity.
Pharmacophore element: Atom or functional group essential for
biological activity
3D Pharmacophore mode: Collection of pharmacophore elements
including their relative position in space
From TCAs to SSRIs and Beyond
Selective Serotonin Reuptake Inhibitors
(SSRIs)
CH3
N
Apr 05/AMJ
NHCH3
CN
CH3
O
N
N
CH3
NHCH3
CH3
O
F3C
F
Br
Cl
zimelidine
citalopram
cipramil/celexa
fluoxetine
prozac/fontex
28.04.1971
14.1.1976
10.1.1974
First synt. Aug 1972
First synt. May 1972
Cl
sertraline
zoloft
1.11.1979
F3C
F
NH
NH
N
O
NH2
O
N
H
indalpine
12.12.1975
O
paroxetine
paxil/seroxat
30.1.1973
O
O
fluvoxamine
fevarin
20.3.1975
The mechanism of SSRI’s
Apr 05/AMJ
Pharmacophore modelling example
Apr 05/AMJ
Fluoxetine
Paroxetine
Citalopram
Sertraline
Chapter 13. Pharmacophore Modeling by Automated Methods: Possibilities and Limitations M.Langgård, B.Bjørnholm, K.Gundertofte
In "Pharmacophore Perception, Development, and use in Drug Design". Edited by Osman F. Güne
International University Line (2000)
Privileged structures
Apr 05/AMJ
……. are ligand substructures that can provide
high-affinity ligands for more than one target…..
Privileged structures
Apr 05/AMJ
”A single ring system, the 5-phenyl-1,4-benzodiazepine
ring, provides ligands for a surprisingly diverse
collection of receptors…..”
Evans et al., J. Med.Chem 1988, 31, 2235-2246
G-protein coupled receptors
Apr 05/AMJ
•7 TM
•Example:dopamine, serotonine,
muscarinic, histamine, neurokinin
•Family A, B, C, A = Rhodopsin like
•In general low sequence homology even
within each family, but highly conserved
residues in the TM regions
•Small molecule ligands bind wholly or
partly within the transmembrane region
mainly in the region flanked by helix 3,5,6
and 7
•From site-directed mutagenesis studies,
side chains involved in binding has been
characterised
ChemBioChem 2002, 3, 928-944
GPCR Privileged structures type
of receptor
Apr 05/AMJ
J. Med. Chem., 47 (4), 888 -899, 2004
Amino acid ”hot spots”
Apr 05/AMJ
Priviledged
sub structure
for target
T1 and T2
Align
T1 and T2
Look for these
in other GPCR’s
Examine which
amino acids are
conserved in binding
pocket for T1 and T2
Amino acid ”HOT SPOTS”
Linking target and ligand side…..
Didier Rognan at the 5ht international workshop in New Approaches
In drug design & discovery, Marburg 21-24 marts 2005
Fluoxetine scaffold common for SERT
and GLYT-1
N
O
Apr 05/AMJ
COOH
Gibson et al, Biorg. Med. Chem Letters
2001 (11), 2007-2009
CF3
N
O
F
COOH
Atkinson et al, Mol. Pharm. 2001 (60),
1414-1420
Comparison between SERT and GLYT-1
Apr 05/AMJ
Y102
F288
Y310
GLYT1 sequence; RED: conserved residues
GREY: conservative mutations
SERT model From Na+/H+
antiporter, J. Pharmacol &
Exp Therapeutics, 307, 34-41
Resume
Apr 05/AMJ
Computational methods for
o Compound library profiling, Chem GPS
o activity QSAR prediction and pharmacophore
modelling
o Solubility and permeability QSPR prediction
o Privileged structures of GPCR’s
”Hit finding”
Apr 05/AMJ
Drug discovery ~ Looking for
a needle in a haystack
Filtering of compounds ~
remove some of the hay
Serendipity
Apr 05/AMJ
“To look for the needle in the
haystack and coming out with the
farmer’s daughter”
Arvid Carlsson