Download Fragment-Based Chemogenomics

Fragment-Based Chemogenomics
Chris de Graaf1, Gerdien de Kloe1, Henry Vischer1, Mark Verheij1, Saskia Nijmeijer1, Azra Delic1,
David Maussang1, Ken Chow1, Anitha Shanmugham2, Paul England2, Rogier Smits1, Rob Leurs1, Iwan de Esch1,2
1Leiden/Amsterdam
Center for Drug Research, Division of Medicinal Chemistry, VU University Amsterdam,
De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands. E-mail: [email protected]
2Iota
Pharmaceuticals Ltd. St Johns Innovation Centre, Cowley Road, Cambridge CB4 0WS, Cambridge, UK.
Fragment-based chemogenomics
Introduction
We have created a proprietary and structurally diverse
fragment library (Fig. 1-2) and screened it for a variety of Gprotein coupled receptors (GPCRs) and a number of other drug
targets (Fig. 3). The resulting data allows for a fragment-based
chemogenomics study (Fig. 4-5) to interrogate the interactions
of GPCRs and their ligands (Fig. 6).
Fig. 4: Fragment bio-affinity profiles illustrate binding site differences and
similarities, some of which are anticipated while others are surprising:
Anticipated differences: GPCRs vs. kinase (PKA-c) + phosphodiesterase (PDE4)
Anticipated similarities: histamine receptors H1R, H3R, and H4R
Surprising differences: bioaminergic GPCRs ADRB2 vs. H1R, H3R, and H4R
Surprising similarities: H4R (GPCR) and 5-HT3R (ion channel)
•
•
•
•
1
Diverse Fragment Library
N
H1R
14
5
6
H3R
1010
Fragments
Ls-AChBP
50
H4R
8
43
41
0.8
12
17
5
 Heavy atoms ≤22
 Log P, H-bond acceptors
& H-bond donors ≤3
 Rotatable bonds ≤5
 No reactive functional groups
Proprietary
database
5-HT3
39
2
Cyclicity
Fragment rules
Fig. 1
N
N
H4R
33
7
IOTA library
H1R specific hits
H4R specific hits
H1R/H4R dual hits
CXCR7 hits
ADRB2
0.6
0.4
0.35
Fragment diversity analysis
Fig. 2
H
N
HO
H
N
O
Structure:
Ring bonds
Side chain bonds
Linker bonds
Double bonds at
linker/ring bonds
N
%
100
H1R
ADRB2
100
80
H4R
CXCR7
80
60
%
0
0
19
20
11
59
0
40
20
20
0
0
0
0
1
42
3
272
0
48
55
1
47
4
2
3
4
0
5
# Rotatable Bonds
1
2
3
# H-Bond Donors
clogP
O
Fragment-protein interactions
21
0
N
12
21
N
57
3
3
1
24
N
16
49
0
2
0
# Rings
NH
0
%
ADRB2
CXCR7
24
0 H
N
H
N
%
H1R
H4R
60
N
N
58
HN
N
80
H
N
HN
50
NH
100
CXCR7
40
26
9
N
0
94
ADRB2
H4R
60
20
68
124
H
N
0
H1R
80
40
N
H
0
N
52
100
CXCR7
71
N
1
ADRB2
H4R
20
26
S
53
0.95
N
0
O
H
N
H1R
60
40
N
455
0.85
NH
First 12 frameworks from
CMC database
(5120 compounds) 5 H
232
0.75
NH2
O
First 12 frameworks from
CNC-active drugs database
433
232
(17.000 compounds)6
0.65
Complexity
Fig. 5: Similarities/differences between target binding sites are reflected by
similarities/differences in physical chemical properties of fragment hits.
scaffold
diversity plots1
Scaffold
N
0.55
N
O
Prune
side chains
N
0.45
N
N
N
Fig. 6: Fragment bio-affinity profiles (Fig. 3-5) are used to optimize structural
models and define protein-ligand interaction fingerprints2 that aid targetselective structure-based virtual screening3 and ligand design4 methods.
Almost 700 different
scaffolds in our
fragment library
Fragment binding modes derived from chemogenomics studies (Fig. 3-5) match with H1R/H3R/H4R cavities
Diverse & Novel Fragment Hits
H 1R
Fig. 3: Screening of 1010 fragments against a variety of
proteins, incl. GPCRs (H1R, H3R, H4R, ADRB2, CXCR4, CXCR7)
and other targets (AChBP, 5-HT3R, PKA-c, PDE4B) yielded
diverse sets of novel and target-specific hits (color coded).
H 3R
F6.52
D3.32
D3.32
H1R
0.45
0.55
0.75
0.85
0.4
0.35
0.95
0.45
0.55
0.4
0.35
0.95
0.45
0.55
0.65
0.85
CXCR4
0.45
0.55
Complexity
0.65
Cyclicity
CXCR7
CXCR4 Hits
0.75
0.85
0.4
0.35
0.95
0.55
0.65
0.75
0.85
0.95
Cyclicity
0.6
PDE4
IOTA library
PDE4 Hits
0.65
0.55
0.75
0.85
0.65
0.75
0.85
0.95
0.4
0.35
0.45
0.55
0.65
Complexity
8 hits
0.6
PKA-c
34% specific
IOTA library
5-HT3 Hits
0.75
0.85
0.95
0.4
0.35
0.45
0.55
0.65
75% specific
0.95
C3.36
IOTA library
PKA-c
0.75
0.85
0.95
Conclusions and Perspectives
• We have created a proprietary and diverse fragment library that has
been screened for a variety of GPCRs and other protein targets.
• All screens have resulted in unique and novel fragment hits.
• The use of fragments in chemogenomics approaches results in higher
resolution interaction fingerprints and leads to improved structural
understanding and novel insights in ligand binding characteristics.
Complexity
Hits (227): >50% radioligand displacement at 10 M
(H1R, H3R, H4R, PKA-c) or 30 M (ADRB2, CXCR4,
CXCR47) or >60% displacement at 100 M (AChBP,
PDE4)
0.8
Complexity
5-HT3R
48% specific
IOTA library
61%
novel
AChBP Hits
1
0.55
70 hits
0.6
Complexity
0.45
0.45
36% specific
IOTA library
100%
novel
CXCR7
0.8
Cyclicity
Cyclicity
AChBP
0.45
0.95
Complexity
0.8
99 hits
0.6
0.85
1
1
0.8
0.75
11 hits
0.6
,
IOTA library
Complexity
1
0.65
0.8
21 hits
0.4
0.35
0.95
0.55
Complexity
0.6
67% specific
IOTA library
75%
ADR-B2 novel
0.75
0.45
9% specific
IOTA library
2%
novel
H4R Hits
1
Cyclicity
Cyclicity
ADRB2
0.4
0.35
0.85
0.8
12 hits
0.6
0.4
0.35
0.75
1
0.8
0.4
0.35
0.65
H4R
Complexity
Complexity
1
E5.46
1) contact; 2) face-to-face stacking; 3) edge-to-face stacking; 4) HB acc.-HB don.; 5) HB don.-HB acc.; 6) neg.-pos.; 7) pos.-neg.
56 hits
0.6
25% specific
IOTA library
20%
novel
H3R
H3R
39% specific
IOTA library
65%
novel
H1R
0.65
Cyclicity
91 hits
0.6
Cyclicity
0.4
0.35
T6.52
0.8
Cyclicity
Cyclicity
23 hits
0.6
C3.36
1
0.8
0.8
H 4R
D3.32
N5.46
S3.36
1
1
M6.52
Novel hits (164): ECFP-4 Tanimito similarity < 0.3 to
any known ligand in chEMBLdb
• These studies can support the design of novel ligands with specified
activity profiles.
References
1) J.Xu et al., J Chem Inf Comput Sci (2000) 40:1177; 2) C. de Graaf et al., J Med Chem (2008) 51:4978; 3) C. de Graaf et al. Curr Pharm Des
(2009) 15:4026; 4) Smits et al. J Med Chem (2008) 51:2457