Building innovative
drug discovery alliances
Discovery of Selective OX2 Antagonists and
Study of Orexin 1 and 2 GPCR Receptors
GPCRs in Medicinal Chemistry, 18th September 2012
Agenda
 Introduction
 Medicinal Chemistry
 Computational Chemistry
PAGE
1
Agenda
 Introduction
 Medicinal Chemistry
 HTS Campaign
 Indole-sulfone Profile
 Indole-sulfone Optimisation
 Conclusions
 Computational Chemistry
PAGE
2
Orexin Receptor Antagonists
Insomnia
 Clinically validated new mechanism for insomnia
 Almorexant, SB-649868, Suvorexant (MK-4305)
 All 3 compounds are dual OX1/2 Antagonists
 An OX2 selective compound has been reported
to have hypnotic activity*
Goal: discover a selective OX2 antagonist for insomnia
PAGE
3
*Dugovic et al, JPET 2009, 330(1), 142-151
Orexin Antagonist Programme
Screening Campaigns
250K Evotec
Library
 Primary screen:
Ca2+ flux FLIPR in
CHO cells
transfected with
hOX1 or hOX2
70 confirmed,
specific hits
 Synthetic
tractability
5 series, 7
singletons
1 ‘series’
selected
 Preliminary SAR in
first 10-20 cpds?
 Med Chem
assessment
O
N
 514 Primary hits
 Counter-screen:
activity against
CHO-endogenous
ATP receptor
HN
EP-109-0092
N
O
O
OX1 HTS also produced hits (EP2161266):
PAGE
4
EP-009-0049
HN
N
Hit Profiles
Indole-3-sulfones
O
O
S
S
O
O
N
O
O
N
N
Potent, OX2 selective
N
EP-009-0236
EP-009-0237
3989 / 103
3704 / 50
Aq Sol (M)
<20
<20
MW / tPSA
433 / 59
447 / 59
cLogP / LipE
3.9 / 3.1
4.2 / 3.1
High / High
High / High
OX1 / OX2 (nM)
FLIPR
ChemAxon
Microsomal Clearance
Human / Rat
Metabolically unstable
PAGE
5
Indole-3-Sulfones
Optimisation
Goals:
 Establish SAR
 Increase stability
 Increase aqueous solubility
Strategy:
 Block (or remove) potentially labile positions
 Most on benzylamine fragment
 Globally reduce lipophilicity
PAGE
6
Indole-3-sulfones
Synthetic Route
Br
O
1) I 2 , thiourea
2) KOH
NH
S
SH
NaH
Dioxane
NH
MeOH / water
(55%)
NH
O
Oxone
NaHCO 3
NH
Acetone / Water
(87%)
(55%)
HN
O
O
O
S
1) NaH, ethyl bromoacetate, DMF
2) NaOH, MeOH / water
O
O
O
(90%)
 Late-stage variation of benzylamine
PAGE
7
WO2011138265
DMF
N
O
(68%)
OH
 Core and Sulfone introduced early
S
HOBt, HBTU
N
NH
O
S
S
N
Indole-3-sulfone Optimisation
Screening Cascade
hOX1 and hOX2 (FLIPR)
P2Y1 (endogenous CHO receptor Gq), rOX1 and rOX2
Non-radioactive GTPS assay
Schild plot / MOA
Human and rat microsomes, aq solubility
P450, PPB, Caco-2
IV / PO PK, human and rat hepatocytes
 No OX1 or OX2 species differences noted
 Good correlation between FLIPR and GTPS assays
PAGE
8
Indole-Sulfone SAR
Which changes are tolerated?
Benzylsulfone
Substitution
O
O
Electron-withdrawing
indole substituents
S
N
O
N
N-methyl deletion
 Fluoro and chloro best
PAGE
9
Benzylamine
Substitution
Indole-Sulfone SAR
Which changes aren’t beneficial?
O
O
O
S
Early SAR
established
N
N
N
O
O
N
N
O
O
S
N
Cl
Amide
replacement
Sulfone
replacement
N
N
O
O
Conformational
restriction
N
Shortened
linkers
N
O
O
S
O
N
F
Electron-donating
indole substituents
10
S
(+ indole inversion)
O
PAGE
O
O
Indole-3-sulfone Optimisation
Blockade of labile positions
O
O
S
O
S
O
N
O
N
N
N
O
O
N
N
F
(R)
F
Cl
EP-009-0236
EP-009-0249
EP-009-0312
EP-009-0372
EP-009-0339
3989 / 103
1221 / 23
971 / 46
>20 M / 33
1576 / 179
Aq Sol (M)
<20
-
33
<20
<20
MW / tPSA
433 / 59
451 / 59
477 / 59
485 / 59
465 / 59
cLogP / LipE
3.9 / 3.1
4.0 / 3.6
4.5 / 2.9
4.6 / 2.9
4.4 / 2.3
High / High
High / High
High / High
High / High
High / High
OX1 / OX2 (nM)
FLIPR
ChemAxon
Microsomal
Clearance
Human / Rat
PAGE
N
N
O
F
O
O
F
N
O
S
S
O
N
O
O
S
O
11
Blocking Strategy
Summary
 Microsomal clearance remains high
 Blocking strategy unsuccessful
 Would lipophilicity reduction fare any better?
PAGE
12
Reduced Lipophilicity
Benzylamine Modification
O
O
O
O
S
S
S
O
O
N
N
O
N
O
O
N
O
N
N
S
O
O
N
N
N
N
O
EP-009-0236
EP-009-0393
EP-009-0314
EP-009-0273
3989 / 103
>20 M / >20 M
>20 M / 4.1 M
>20 M / 2.9 M
Aq Sol (M)
<20
-
-
-
MW / tPSA
433 / 59
438 / 85
434 / 72
434 / 72
cLogP / LipE
3.9 / 3.1
2.5 / -
2.7 / 2.7
2.6 / 2.9
OX1 / OX2 (nM)
FLIPR
ChemAxon
 Loss of potency on changing the benzylamine group
 Revised synthetic routes needed to target different cores
PAGE
N
13
Indolizine Synthetic Route
O
O
O
Br
Cl
N
O
1) DBU
2) Oxone
O
O
S
SH
N
N
(40%)
(1) 76%
2) 61%)
O
O
S
O
O
O
O
S
1) HCl
2) K 2 CO3
O
O
1) NaBH 4
2) PPh 3 / I2
S
Cl
S
O
N
N
N
O
O
 Core and sulfone properties fixed early on
 2-Methyl group introduced for synthetic reasons
PAGE
14
WO2011138266
(1) 77%
2) 78%)
(61%)
O
O
O
Imidazopyridine Synthetic Route
O
O
Cl
O
Et 3N
N
NH2
N
H
N
THF
N
O
N
POCl 3
(48%)
O
O
(67%)
O
O
SH
O
Pd 2(dba) 3
Xantphos
i
Pr 2 NEt
I
NIS
Et 3N
N
MeCN
N
Dioxane
O
S
Oxone
N
N
N
DCM
(77%)
O
 Core properties fixed early on
 Sulfone introduced later
PAGE
15
WO2011138266
N
O
O
O
(66%)
S
(70%)
O
O
Indole-3-sulfone Optimisation
Reduction of Lipophilicity – core changes
N
N
O
N
O
N
N
O
O
O
O
N
N
N
N
N
N
N
S
O
O
O
O
O
S
S
S
S
O
O
O
O
N
F
EP-009-0236
EP-009-0456
EP-009-0403
EP-009-0495
EP-009-0466
3989 / 103
1399 / 538
998 / 14
>20 M / 194
3784 / 1.0
Aq Sol (M)
<20
85
37
177
22
MW / tPSA
433 / 59
434 / 72
434 / 72
452 / 72
447 / 59
cLogP / LipE
3.9 / 3.1
3.0 / 3.3
3.6 / 4.3
3.2 / 3.5
3.8 / 5.2
High / High
High / High
High / High
High / High
High / High
OX1 / OX2 (nM)
FLIPR
ChemAxon
Microsomal
Clearance
Human / Rat
Two orders of magnitude potency increase for 2-methylindolizine!
PAGE
16
Core Replacement Follow up
Indazole and 2-Methylindole
O
O
O
S
N
N
O
N
S
O
O
N
O
S
S
O
O
N
N
N
O
O
N
N
HN
O
HN
N
O
EP-009-0403
EP-009-0483
EP-009-0404
998 / 14
>20 M / 123
278 / 1.1
>20 M / >20 M
Aq Sol (M)
37
-
<20
-
MW / tPSA
434 / 72
445 / 105
465 / 59
428 / 90
cLogP / LipE
3.6 / 4.3
3.2 / 3.7
4.2 / 4.8
2.0 / -
High / High
High / High
High / High
High / High
OX1 / OX2 (nM)
FLIPR
ChemAxon
Microsomal Clearance
Human / Rat
PAGE
F
17
EP-009-0482
LipE Progression Summary
From Ep-009-0236
O
S
O
O
N
S
O
N
O
N
HN
O
N
N
O
S
O
O
S
N
O
N
O
N
N
PAGE
18
O
High Probability
Space
N
Electrostatic Complementarity
Useful tool for structure-base drug discovery
 We developed a method1 to predict regions of the ligand that are electrostatically attracted to or
repelled by the protein, and to map / visualize it on the ligand surface
 Modifications in these regions usually have a significant effect on binding affinity
 We called these important regions of the ligand “hot spots”
OX2 IC50 = 195 nM
OX2 IC50 = 1 nM
Attraction
Repulsion
PAGE
19
1 Davenport
and Heifetz et al ,Assay Drug Dev
Technol. 2010 Dec;8(6):781
Agenda
 Introduction
 Medicinal Chemistry
 Computational Chemistry
 Receptors selectivity study
 Conclusions
PAGE
20
hOX1 vs. hOX2 Selectivity
Study Motivation
 Precise determinants of antagonist binding and
selectivity were neither fully known nor rationalised
 Site-directed mutagenesis and domain exchange (chimera) studies
have provided important insight on key features of the OX1/2 binding
sites1,2 & 3
 11 mutations were performed for OX1 and 18 for OX2
 To explain the role of each mutated residue for binding
and selectivity of a set of OX1/2 antagonists
 Support discovery of novel OX1/2 antagonists
PAGE
21
1 Malherbe,
et al (2010) Mol. Pharmacol. 78, 81-93
et al (2011) Eur. J. Pharmacol. 667, 120-128
3 Heifetz et al Biochemistry. 2012 Apr 17;51(15):3178
2 Tran,
hOX1 vs. hOX2
What we can learn from sequence alignment?
 TMD:
 Identity:
69.3%
 Similarity:
80.5
Mutations by Evotec
Mutations by other
Mutations by Evotec and other
Not tested
PAGE
22
Intriguing SDM data
Small differences in sequence can lead to large changes in IC50
Position
hOX1
Almorexant
3.33
7.43
A127T
Y348A
↓↓
↓↓
?
?
Position
hOX2
Almorexant
3.33
7.43
T135A
Y354A
=
=
?
?
 Why does the A1273.33T mutation significantly reduce binding of
Almorexant to OX1, while the T1353.33A mutation in OX2 has no effect?
 Why does mutating the conserved tyrosine located in the 7.43 position
give opposite effects in OX1 vs OX2?
 mutation of Y3487.43 in OX1 eliminated Almorexant binding, the same
mutation Y3547.43 in OX2 had no effect
PAGE
23
Exploration of OX1/2 Receptors
Method validation
PAGE
24
GPCR Modeling Protocol
The optimized similarity routines
GPCR specific seq. alignment matrix
Helical rotation alignment
Homology Modeling based on single template
Hybridized Homology Modeling
Addition of SDM data
MD kink detection & formation
Optimized Similarity Modeling
Molecular Dynamics Simulations
Ab-initio Decoy Template Modeling
Loop remodeling
MC side-chain rotamer library sim.
Protein packing & complementary score
Fit to sequence alignment
Individual optimization routine scoring
Fit to SAR and SDM data
PAGE
25
Ligand Pocket Optimization
● Each tier of the modeling process assesses the need for the next stage
MD Simulations
Established Protocol for GPCR Modelling
Aims
• Optimise homology models
• Explore structural behaviour  OX1/OX2 and mutants/wild-types
System size
GROMACS
OPLS-AA forcefield
Performance = ~1ns/day 8 cores
PAGE
26
~117k atoms
~31k waters
~ 327 residues
~ 5200 protein atoms
~350 lipids
Focus on TM3
Focus on TM3 – important for selectivity of OX antagonists
B: OX1
A: OX1
TM7
TM5
TM2
T1363.42
TM3
W1694.50
Y3487.43
Y2155.38
Q1263.32
N872.45
A1273.33
TM3
TM2
TM4
Q1794.60
C: OX2
TM7
TM5
TM2
TM3
TM4
T1443.42
Y3547.43
T1353.33
TM2
T1062.56
27
D: OX2
Y2235.38
Q1343.32
PAGE
F1393.45
F842.42
TM4
I982.56
Q1874.60
W1774.50
TM3
TM4
N952.45
Focus on TM3
 Different inter-helical interactions force the core of TM3OX1 and TM3OX2 to adopt slightly different
conformations
OX2:TM3
A
OX1:TM3
B
Almorexant
Almorexant
Q1874.60
T1353.33
T135
of T M
Y3487.43
Q1343.32
T1062.56
3
Y2235.38
Y3547.43
C
OX2:TM3
OX1:TM3
Q1263.32
st heli
x axis
 In OX2, there appeared
to be a tilting of the
TM3OX2 core away from
its original position in
OX1
OX1
Longe
W1694.50
N872.45
T1363.42
OX2
F842.42
F1393.45
TM3OX1 vs. TM3OX2

PAGE
28
Almorexant Complex with OX1/2
A: hOX1R
WT
B: hOX2R with Almorexant
with Almorexant
W20645.54
W21445.54
H3507.39
H3447.39
C: hOX1R
WT
,
Y2155.38
Y2235.38
TM3OX1, TM3OX2 with Almorexant
5.42
F227
Y3547.43
7.43
Y348
Y3176.48
Y2325.47
Y3116.48
Q1263.32
A1273.33 F2195.42
Y2245.47
T1062.56
Y3487.43
 The residue Y3487.43 does not form a
direct interaction with Almorexant but it
still plays a key role by stabilising
Q1263.32 in a conformation in which it
interacts with Almorexant.
PAGE
29
Q1343.32
4.60
T1353.33 Q187
Q1263.32
Q1343.32
T1062.56
TM3OX1
OX2
TM3OX2
T1353.33
 The mutation of A1273.33 into the
larger residue threonine limits the
approach of the antagonists into
the OX1 sub-pocket between
TM3, 4 and 5
Intriguing SDM data
Solving the mystery
Position
hOX1
Almorexant
3.33
7.43
A127T
Y348A
↓↓
↓↓
?
?
Position
hOX2
Almorexant
3.33
7.43
T135A
Y354A
=
=
?
?
 We showed that different inter-helical interactions force the conformations of TM3OX1 and
TM3OX2 to be different
 A127 and T135 are located in the same position in the sequences but in different locations
in the structures
 The mutation of A1273.33 into the larger residue threonine limits the approach of antagonists
into the OX1 sub-pocket between TM3, 4 and 5, resulting in a significant decrease in the
binding of certain antagonists
 The OX1 residue Y3487.43 does not form a direct interaction but it still plays a key role by stabilising Q1263.32 in a
conformation in which it interacts with antagonists – so mutating it into Alanine will affect antagonist binding
 The OX2 residue Y3547.43 (in contrast to Y3487.43 of OX1) is “unemployed”, it neither interacts directly with antagonists
nor does it stabilise Q1343.32 – so mutating it to Alanine will not have any effect
PAGE
30
GPCRs Modelling Conclusions
 Minor differences in sequences can lead to significant differences in the tertiary structure of GPCRs, their
ability to bind ligands, and their selectivity
 GPCR models based solely on homology modelling might not be sufficient to rationalise potency and
selectivity
 MD simulations allow refinement of GPCR models to a degree that is not possible with static homology
modelling alone
 The structural insights gained from this process are critical for rationalising the SDM data, and for the design of
new GPCR ligands
PAGE
31
Discovery of Novel OX1, -2 Selective Antagonists
B: hOX2R with EP-009-0513 (OX2 Selective)
A: hOX1R with EP-009-0049 (OX1 selective)
a
W206
H3447.39
H3507.39
Y2155.38
Y3487.43
W21445.54
a
45.54
Y3547.43
Q1263.32
N324
Y2235.38
6.55
V3477.42
b
A1273.33
Y3116.48
V1303.36
b
PAGE
32
Y2245.47
I3206.51
F2195.42
c
F2275.42
Q1343.3
Y3176.48
2
T1353.33
Y2325.47
c
Acknowledgements
Evotec (UK)
Evotec (Hamburg)
University of Oxford
Tara Fryatt
Dominique Manikowski
G. Benjamin Morris
Oliver Barker
Rita Reifegerste
Philip C. Biggin
Gurubaran Raju
Mark Slack
David Hallett
Mark Whittaker
Natacha Prevost
Julia Vile
Sandeep Pal
Richard Law
 In vitro DMPK Group (UK)
 Evotec (India)
 Royal Society, UK for Industry award
PAGE
33
Building innovative
drug discovery alliances
Appendix
 TMD C atom RMSDs fluctuate within the same narrow range of 1.7 to 2.4 Å, which is comparable to the
values typically obtained for MD simulations of GPCR crystal structures
 TM3 C atom RMSDs fluctuate within a narrow range between 1.0 and 1.5 Å for the OX1WT, OX1A127T and
OX2WT indicates that the TM3 conformation of receptors is almost “frozen” AND is not affected by the
A127T3.33 mutation.
TMD
PAGE
35
TM3
OX1 vs. OX2 Receptor
Final Models
 TMD conformational
differences observed in
structures and binding site
topology of the OX1vs. OX2
structures
 OX1 is orange and OX2 is
green
PAGE
36
Electrostatic Complementarity
Useful tool for structure-base drug discovery
 We developed a method1 to predict regions of ligand that are electrostatically attracted to or
repelled by the protein, and to map / visualise it on the ligand surface
 Modifications in these regions usually have a significant effect on binding affinity
 We called these important regions of the ligand “hot spots”
OX2 IC50 = 195 nM
OX2 IC50 = 1 nM
Attraction
Repulsion
PAGE
37
1 Davenport
and Heifetz et al ,Assay Drug Dev
Technol. 2010 Dec;8(6):781
Building innovative
drug discovery alliances
Your contacts:
Jonathan Bentley and Alex Heifetz
+44 (0)1235 861561
[email protected]
[email protected]