Discovery of Highly Potent Dual Orexin Receptor Antagonists via a Scaffold-Hopping Approach

Article abstract of DOI:10.1002/cmdc.201600175

Starting from suvorexant (trade name Belsomra), we successfully identified interesting templates leading to potent dual orexin receptor antagonists (DORAs) via a scaffold-hopping approach. Structure–activity relationship optimization allowed us not only to improve the antagonistic potency on both orexin 1 and orexin 2 receptors (Ox1 and Ox2, respectively), but also to increase metabolic stability in human liver microsomes (HLM), decrease time-dependent inhibition of cytochrome P450 (CYP) 3A4, and decrease P-glycoprotein (Pgp)-mediated efflux. Compound 80 c [{(1S,6R)-3-(6,7-difluoroquinoxalin-2-yl)-3,8-diazabicyclo[4.2.0]octan-8-yl}(4-methyl-[1,1′-biphenyl]-2-yl)methanone] is a potent and selective DORA that inhibits the stimulating effects of orexin peptides OXA and OXB at both Ox1 and Ox2. In calcium-release assays, 80 c was found to exhibit an insurmountable antagonistic profile at both Ox1 and Ox2, while displaying a sleep-promoting effect in rat and dog models, similar to that of the benchmark compound suvorexant.

Full text of DOI:10.1002/cmdc.201600175

DOI: 10.1002/cmdc.201600175
Full Papers
Discovery of Highly Potent Dual Orexin Receptor
Antagonists via a Scaffold-Hopping Approach
Bibia Heidmann,* John Gatfield, Catherine Roch, Alexander Treiber, Simone Tortoioli,
Christine Brotschi, Jodi T. Williams, Martin H. Bolli, Stefan Abele, Thierry Sifferlen,
FranÅois Jenck, and Christoph Boss[a]
Starting from suvorexant (trade name Belsomra), we success-
fully identified interesting templates leading to potent dual
orexin receptor antagonists (DORAs) via a scaffold-hopping ap-
proach. Structure–activity relationship optimization allowed us
not only to improve the antagonistic potency on both orexin 1
and orexin 2 receptors (Ox1 and Ox2, respectively), but also to
increase metabolic stability in human liver microsomes (HLM),
decrease time-dependent inhibition of cytochrome P450 (CYP)
3A4, and decrease P-glycoprotein (Pgp)-mediated efflux. Com-
pound 80c [{(1S,6R)-3-(6,7-difluoroquinoxalin-2-yl)-3,8-diazabi-
cyclo[4.2.0]octan-8-yl}(4-methyl-[1,1’-biphenyl]-2-yl)methanone]
is a potent and selective DORA that inhibits the stimulating ef-
fects of orexin peptides OXA and OXB at both Ox1 and Ox2. In
calcium-release assays, 80c was found to exhibit an insur-
mountable antagonistic profile at both Ox1 and Ox2, while dis-
playing a sleep-promoting effect in rat and dog models, similar
to that of the benchmark compound suvorexant.
Introduction
Insomnia is a medical condition characterized by difficulty to
initiate and/or to maintain restorative sleep. It can lead to irri-
tability, concentration problems, excessive fatigue, poor work
performance, driving impairments, and other health prob-
lems.[1,2] The direct and indirect costs to the global economy
are considered enormously high; for the US economy alone,
the costs related to insomnia have been estimated to exceed
$100 billion per year.[3] The physical and psychological distress
of people suffering from insomnia as well as the high econom-
ic toll and the lack of treatments that do not only improve re-
storative sleep but that are also free of side effects such as de-
pendency, next day impairment, and other safety concerns led
to intensive research in the field of insomnia over the last de-
cades.[4]
The orexin peptides were originally identified as stimulating
food consumption (orexinergic) when directly injected into the
central nervous system of rodents.[8] Further studies showed
that the orexin producing neurons project to widespread and
diverse regions of the brain including areas that, under circadi-
an influence, modulate the sleep wake cycle.[9,10] Both genetic
and pharmacological research confirmed that the orexin
system plays an important role in maintaining arousal and
wakefulness in rodents, dogs and humans.[9,11–14]
It was therefore expected that pharmacological blockade of
brain orexin receptors would decrease wakefulness and pro-
mote sleep by a mechanism of action very different from avail-
able drugs (i.e., GABA receptor modulators) known to suppress
overall central nervous system activation through potentiation
of inhibitory neurotransmission.[9] It was demonstrated that
pharmacological modulation of orexin signaling with small-
molecule antagonists decreases wake and induces sleep in pre-
clinical species (rodent and canine models) and in patients suf-
fering from insomnia.[9,15,16]
In 1998 two independent research groups discovered the
neuropeptides orexin A and orexin B also known as hypocre-
tin-1 and hypocretin-2.[5] Both neuropeptides are secreted by
a small population of excitatory neurons located in the lateral
hypothalamus where they play essential roles in hypothalamic
functions.[6,7] Orexin A and orexin B were also found to bind to
Interest in the concept of orexin receptor antagonism has
been boosted in 2007 with the disclosure of almorexant 1, an
orally active dual orexin receptor antagonist (DORA) that pro-
motes sleep in rats, dogs and in humans without evidence of
cataplexy and therefore provided proof-of-concept of orexin
antagonism for the treatment of primary insomnia.[16—17] More
recently, Merck’s DORA suvorexant 3, which dose dependently
increases total sleep time by decreasing latency to persistent
sleep and wake after sleep onset, was granted US Food and
Drug Administration (FDA) approval as first-in-class drug for
the treatment of insomnia (Belsomra).[12b,18,19–28]
previously identified orphan
G protein-coupled receptors,
Ox1R and Ox2R (Hcrt1 and Hcrt2 receptors).[5]
[a] Dr. B. Heidmann, Dr. J. Gatfield, Dr. C. Roch, Dr. A. Treiber, Dr. S. Tortoioli,
Dr. C. Brotschi, Dr. J. T. Williams, Dr. M. H. Bolli, Dr. S. Abele, Dr. T. Sifferlen,
Dr. F. Jenck, Dr. C. Boss
Actelion Pharmaceuticals Ltd.,
Drug Discovery and Preclinical Research & Development,
Gewerbestrasse 16, 4123 Allschwil (Switzerland)
E-mail: heidmab@actelion.com
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cmdc.201600175: Descriptions of all biological assays,
in vivo pharmacology experiments, and experimental details for the syn-
thesis of the reported compounds.
Over the past years, several high throughput screening cam-
paigns using Ca2+ release assays (FLIPR) for the search of new
ChemMedChem 2016, 11, 1 – 16
1
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
DORAs were performed at Actelion. They provided interesting
hits that were further optimized into potent and brain pene-
trant DORAs.[9b,29–35] Lately, we reported on the sulfonamide
ACT-462206 as preclinical candidate that emerged from those
efforts.[35] With the increasing knowledge on DORA binding
modes and the availability of several potent DORAs in clinical
trials for primary insomnia (Figure 1), we decided to tackle the
search for new lead structures via a scaffold-hopping approach
using 3 as a template.[9b,26,36–41]
active conformation.[41] The final goal being to replace the pe-
ripheral substituents of 3 with proprietary substituents.
Analysis of potential templates (from commercial sources
and/or described in the literature) was performed using
“Chem3DPro” and “Actelion 3D Modeling” tools allowing for
manual alignments of ligands of interest with selected tem-
plates. Our studies suggested that several constrained systems
enforce the low energy conformation observed with diazepane
3 leading to highly potent compounds on both Ox1 and Ox2
receptors.[41] The templates’ selection was based on three crite-
ria: conformational similarity to compound 3, freedom to oper-
ate based on published patents, and commercial availability/
synthetic accessibility. In Table 1, we report the data generated
Table 1. In-house data for benchmark compound 3.
Diazepane
R1
R2
Compd
IC50 [nm][a]
hOx1
hOx2
A
B
5
1.8
1.0
5.9
A
A
B
B
3
2.8
58
Figure 1. DORAs reported in clinical trials.
3a
13
Herein we report the identification of novel starting points
as dual orexin receptor antagonists following a scaffold-hop-
ping approach. We also describe herein the structural optimi-
zation of one scaffold into an orally bioavailable and CNS pen-
etrant DORA.
[a] Values are the geometric mean of at least two to three independent
experiments as determined by FLIPR assay (see the Supporting Informa-
tion for details).
in house for the benchmark compound 3. In Tables 2–4, the
identified templates fulfilling these requirements are depicted
with their antagonistic potency on Ox1 and Ox2 receptors de-
termined using FLIPR assay technology. Compounds listed in
Tables 2–4 were synthesized by following the synthetic route
depicted in Scheme 1 using the cis-3,8-diazabicyclo[4.2.0]oc-
tane template as example.[42]
Results and Discussion
With the active conformation reported by Merck for their N,N-
disubstituted-1,4-diazepane derived series, we decided to use
3 as a template for our scaffold-hopping approach to identify
new starting points for the design of new DORAs.[41] Our strat-
egy was to first retain the 5-chlorobenzo[d]oxazole and the (5-
methyl-2-(2H-1,2,3-triazol-2-yl)phenyl)methanone substituents
and to replace the central 1,4-diazepane ring with new bicyclic
“diamine” templates (Figure 2) that allow for the same intramo-
lecular p-stacking interaction between the substituents and
the adoption of the suggested low-energy twist-boat ring bio-
tert-Butyl 3,8-diazabicyclo[4.2.0]octane-8-carboxylate 56 can
be purchased or synthesized according to known proce-
dures.[42] Amide coupling with 5-methyl-2-(2H-1,2,3-triazol-2-yl)-
benzoic acid 57, followed by Boc-deprotection and nucleophil-
ic aromatic substitution using 2,5-dichlorobenzo[d]oxazole 60
Figure 2. Scaffold hopping proof-of-concept (PoC) validation.
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Scheme 1. Synthesis of 3,8-diazabicyclo[4.2.0]octane 22 and 23 as orexin receptor antagonists (details given in the Supporting Information). Reagents and con-
ditions: a) TBTU, DIPEA, CH3CN, RT, 17 h, 92%; b) HCl 4n in dioxane, RT, 4 h, quant.; c) K2CO3, DMF, 608C, 17 h, 82%; d) K2CO3, DMF, 608C, 17 h, 95%; e) HCl
4n in dioxane, RT, 4 h, quant.; f) TBTU, DIPEA, CH3CN, RT, 17 h, 77%.
afforded the DORA 22 in Table 2. Rac-22 could be resolved by
chiral HPLC to afford enantiomers 22a and 22b.[44] The se-
quence was reversed in order to reach compound 23 in
Table 2: 56 was treated with 2,5-dichlorobenzo[d]oxazole 60
followed by Boc-deprotection and amide coupling with acid
57. As for the regioisomer rac-22, rac-23 could be resolved by
HPLC using a chiral stationary phase to afford enantiomers
23a and 23b.[45]
The fused bicyclic templates reported in Table 2 delivered
several interesting scaffolds with reference compounds (7, 8,
16, 22, and 23) reaching potencies below 100 nm on both
orexin receptors. Interestingly, the spiro scaffolds (Table 3),
when active, resulted in compounds preferentially active at the
Ox1 receptor (e.g., 28, 37, and 39). The same was true for the
bridged templates depicted in Table 4. The three most attrac-
tive scaffolds led to preferential Ox1R antagonists (53, 54, and
55).[43] Following our project’s goal, we favored the scaffolds
with the highest potential to deliver DORAs, with reasonable
molecular weight and with a modular structure easily amena-
ble to parallel synthesis for SAR optimization. This was the
case for 7, 8, 16, 22/23 (Table 2). Among these, we decided to
concentrate our efforts on the 3,8-diazabicyclo[4.2.0]octane
template (compounds 22 and 23) that not only re-
Ox1R and by greater than 190-fold on Ox2R in favor of the
single enantiomer 23a. Notably, the active enantiomer 22a
(amide on the piperidine ring) exhibits (1R,6S)-chirality, while
the active enantiomer 23a (amide on the azetidine ring) exhib-
its the opposite (1S,6R)-chirality (Figure 3).
To identify potential issues to be addressed prior to embark-
ing on the SAR optimization, we examined the in vitro and the
in vivo properties of 22a and 23a. For the in vivo experiments,
all procedures were approved by the local Veterinary Office
and strictly adhered to Swiss federal and international regula-
tions on animal experimentation.
Both compounds exhibited identical physicochemical prop-
erties (Table 5). They were moderately lipophilic with LogD7.4 of
3.2 (cLogP=4.0) with a polar surface area of 73.1 ꢁ2 indicating
a high probability for brain penetration. Their profiles on cyto-
chrome P450 (CYP) inhibition were similar except for the
strong time-dependent CYP3A4 inhibition with a shift in IC50
values after pre-incubation (thereafter called “CYP3A4 shift”) of
30-fold measured for 22a as opposed to the moderate
CYP3A4 shift for 23b (4.7-fold). Both compounds were similarly
bound to human plasma proteins (98%) with high passive per-
meability in human MDR1-MDCK cells of 86.10ꢀ6 and
sulted in reference compound 22 being equally
potent to compound 3, but which can interestingly
also accommodate its regioisomer 23 keeping high
affinity for both orexin receptors.[42] Enantiomers 22a
and 22b differed in receptor affinity by greater than
400-fold on both receptors in favor of the single
enantiomer 22a while enantiomers 23a and 23b dif-
fered in receptor affinity by greater than 50-fold on
Figure 3. Stereochemistry for 3,8-diazabicyclo[4.2.0]octane active enantiomers.
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
3
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 2. Potency obtained for selected bicyclic diamine templates: fused bicyclic templates.
Bicyclic template
R1
R2
Compd
IC50 [nm][a]
Stereochemistry
hOx1
hOx2
A
B
B
A
6
7
1742
13
3100
25
Racemic
B
B
A
A
7a
7b
5.1
17
Optically pure
Optically pure
1880
2935
A
B
8
71
64
Meso
A
B
B
A
9
10
1030
118
544
119
Unknown chirality
A
B
11
1556
2375
Unknown chirality
Racemic
A
B
B
A
12
13
1548
2249
528
1118
A
B
B
A
14
15
386
125
746
137
Racemic
Racemic
A
B
B
A
16
17
26
204
45
290
A
B
B
A
18
19
669
1925
1048
1315
Racemic
Racemic
A
B
B
A
20
21
1343
2242
915
912
A
A
B
B
22
1.7
0.9
7.5
2.9
Racemic
22a
This enantiomer
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
4
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 2. (Continued)
Bicyclic template
R1
A
B
R2
B
Compd
22b
IC50 [nm][a]
Stereochemistry
hOx1
474
hOx2
1251
This enantiomer
Racemic
A
23
22
31
B
B
A
A
23a
23b
12
17
This enantiomer
This enantiomer
Racemic
668
3240
A
B
B
A
24
25
96
840
288
369
A
B
B
A
26
27
163
359
2080
995
Racemic
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails).
76.10ꢀ6 cmsꢀ1, respectively. Yet 22a was a strong substrate for
human P-glycoprotein (Pgp) with an efflux ratio of 11 as op-
posed to the moderate efflux of 4.7 measured for 23a. Finally,
23a was also more stable than 22a in human and dog liver
microsomes (HLM and DLM) while both compounds were
rats to assess their brain penetration potential. The dose used
was 100 mgkgꢀ1 to overcome the high first-pass metabolism
in rats. As shown in Table 6, both compounds showed excel-
lent total brain concentrations of [B]=2741 and 3394 nggꢀ1
for 22a and 23a, respectively. Based on the high total brain
concentrations, both compounds were evaluated for their
sleep-promoting effects in male Wistar rats equipped with radi-
otelemetry probes continuously recording electroencephalo-
gram/electromyography (EEG/EMG) and locomotor activity. The
highly
unstable
in
rat
liver
microsomes
(RLM
>1250 mLminꢀ1 mgꢀ1). Consistent with their relatively lipophilic
nature and their high intrinsic clearance in RLM, both com-
pounds showed moderate clearance in Wistar rats (17 and
23 mLminꢀ1 kgꢀ1, Table 6) and low bioavailability (F=8 and
11%). Apparent terminal t1/2 was shorter for 22a (1.1 h) as
compared to 23a (21 h) but both compounds were quickly ab-
sorbed after oral dosing as peak drug levels were reached
within the first 30 min (tmax =0.5 h). The pharmacokinetic pa-
rameters evaluated in Beagle dogs (Table 6) were similar for
both compounds showing lower clearance than the rat data
(9.6 and 12 mLminꢀ1 kgꢀ1), similar volumes of distribution
(Vss =0.8 and 1.3 Lkgꢀ1), and short t1/2 (~1 h) even for a sleeping
drug that should be cleared from the body within 6–8 h. In
Beagle dogs, compound 23a was more bioavailable than 22a
(77 vs. 58%) and was absorbed three times faster (tmax =0.38
vs. 1 h). Compounds 22a and 23a were further profiled in
in vivo blood–brain barrier (BBB) experiments in male Wistar
animals were administered with
a single oral dose of
100 mgkgꢀ1 at the beginning of the nocturnal active phase,
when endogenous orexin levels increase. All experiments
where done in a crossover design. Over the 12 h night period
following administration, both 22a and 23a significantly de-
creased the home cage activity by 31 and 32% relative to their
respective vehicle-treated rats (p<0.001 for both, paired t-
test). Electrophysiologically 22a slightly but not significantly
decreased the time spent in active wake and quiet wake (ꢀ9
and ꢀ5%, respectively; p>0.05, paired t-test; Figure 4). The
time spent in non-REM (rapid eye movement) sleep increased
significantly by 10% (p<0.05, paired t-test) and the time spent
in REM sleep increased not significantly by 20% relative to ve-
hicle-treated rats (p<0.05, paired t-test; Figure 4). The total
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 3. Potency obtained for selected bicyclic diamine templates: spiro templates.
Bicyclic Template
R1
R2
Compd
IC50 [nm][a]
Stereochemistry
hOx1
hOx2
A
B
B
A
28
29
127
1403
1253
1008
Racemic
Racemic
Racemic
A
B
B
A
30
31
1643
464
1415
722
A
B
B
A
32
33
281
291
117
528
A
B
B
A
34
35
1270
892
392
895
Racemic
Racemic
Racemic
A
B
36
494
955
A
B
B
A
37
38
1.4
292
47
1200
A
B
B
A
39
40
8.6
212
514
1039
Racemic
Racemic
A
B
B
A
41
42
1270
1190
2350
960
A
B
B
A
43
44
2223
1087
1858
1451
Racemic
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails).
sleeping time increased by 12% relative to the vehicle-treated
rats and the relative proportion of the time spent in non-REM
and REM sleep was stable (82 and 18% of non-REM and REM
sleep, respectively, for the 22a-treated rats compared with the
83 and 17% non-REM and REM sleep for the vehicle-treated
rats (p>0.05, paired t-test). Based on its high potency on both
orexin receptors and its unbound brain concentration ([B]u =
67 nm ~56ꢂIC50 rOx2/35ꢂIC50 rOx1), 22a was expected to be
more active and to show a similar sleep-promoting effect as
our benchmark suvorexant 3, which under the same experi-
mental conditions, increased the total sleep time by 22% over
the vehicle-treated rats. Contrasting effects were observed
Figure 4. All experiments were done in a crossover design. Effect of 22a,
23a, and suvorexant 3 on the time spent in sleep and wake stages (percent
of total time) during the 12 h night active period following administration in
male Wistar rats. Rats were administered a single oral dose of vehicle or
compound (100 mgkgꢀ1). Data are presented as mean ꢁSEM; *p<0.05,
***p<0.001 (n=7 for 22a, n=8 for 23a and n=14 for suvorexant 3).
with 23a which showed good efficacy despite lower potency
on both orexin receptors and lower unbound brain concentra-
tion, when reported to the potency on both receptors ([B]u =
129 nm ~15ꢂIC50 rOx2/4ꢂIC50 rOx1), than 22a. It decreased slight-
ly and not significantly the time spent in active wake and
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
6
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 4. Potency obtained for selected bicyclic diamine templates: monocyclic and bridged templates.
Template
R1
R2
Compd
IC50 [nm][a]
Stereochemistry
hOx1
1651
hOx2
1831
A
A
A
B
B
B
45
46
47
None
1473
277
883
650
Racemic
Racemic
A
B
B
A
48
49
464
2017
153
2009
Racemic
A
B
B
A
50
51
2032
2403
753
1670
Unknown chirality
Unknown chirality
A
B
B
A
52
53
583
17
3367
223
A
A
B
B
54
55
6.5
542
Meso
Meso
163
>6460
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails).
quiet wake relative to vehicle-treated rats (ꢀ10 and ꢀ13%, re-
spectively; p>0.05, paired t-test) but increased significantly
both the time spent in non-REM sleep (+18% compared with
vehicle-treated rats; p<0.001, paired t-test) and the time spent
in REM sleep (+37% compared with vehicle-treated rats; p<
0.05, paired t-test; Figure 4). The total sleeping time increased
by 22% relative to the vehicle-treated rats. These effects were
similar to those observed with benchmark compound 3.
The data compiled in Tables 5 and 6 were clearly in favor of
compound 23a, which showed a lower shift in the time-de-
pendent CYP3A4 inhibition assay and a much lower Pgp efflux
than 22a. Moreover 22a, despite its high unbound brain con-
centration and high potency on both orexin receptors, was
surprisingly poorly active in the EEG/EMG in vivo experiment
contrasting with 23a that showed effects similar to those ob-
served with compound 3.
We therefore decided to concentrate our SAR efforts on the
template (1S,6R)-3,8-diazabicyclo[4.2.0]octane with the amide
bound to the azetidine ring and the aromatic substituent at
the piperidine ring.[46] In addition to improving the potency on
both orexin receptors, our main goal was to decrease the Pgp
efflux ratio to values below three. In view of the targeted clini-
cal indication we also aimed at having moderately stable com-
pounds in HLM in order to achieve an appropriate clearance
yielding to a sleep effect of 6 to 8 h in humans to avoid next-
day drowsiness.
Final compounds in the (1S,6R)-3,8-diazabicyclo[4.2.0]octane
series were prepared according to the route depicted in
Scheme 1, and the core building block was synthesized follow-
ing published procedures.[42]
Our SAR work started with the goal of finding a suitable iso-
steric replacement for the 5-chlorobenzo[d]oxazole western
part of 23a. Previously reported SAR analysis on the Merck dia-
zepam amide DORA series demonstrated that the lipophilic in-
teractions in the 5-chlorobenzo[d]oxazole region are of key im-
portance for antagonist potency.[18,39] Removal of the distal
fused phenyl in the western side heterocycle leads to inactive
compounds on both receptors. Therefore, we favored fused
heterocycles that allow for the same level of lipophilicity in
this region. We also aimed at keeping the planar topology of
the fused benzo[d]oxazole ring. Table 7 summarizes our efforts
to replace the 5-chlorobenzo[d]oxazole moiety.[47] All reported
derivatives were assayed as racemic mixtures, except if other-
wise stated. Interesting racemic mixtures with potencies below
20 nm on one receptor and at least 40 nm on the other recep-
tor were separated by chiral HPLC for further profiling. In our
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
7
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 5. Comparative in vitro data for 22a and 23a.[a]
Table 6. Comparative in vivo data for 22a and 23a.[a]
Properties
Compound 22a
Compound 23a
Properties
Compound 22a
Compound 23a
PhysChem properties
Pharmacokinetics: Wistar rat, 1 mgkgꢀ1 i.v.
Mr [gmolꢀ1
LogD7.4
]
448.91
3.2
73.1
448.91
3.2
73.1
AUC [nghmLꢀ1
CL [mLminꢀ1 kgꢀ1
Vss [Lkgꢀ1
Terminal t1/2 [h]
]
965
17
0.6
1.1
732
23
1
]
PSA [ꢁ2]
]
21
Human and rat orexin receptors
IC50 (hOx1/rOx1) [nm]
IC50 (hOx2/rOx2) [nm]
0.9/1.9
2.9/1.2
12/35
17/8.6
Pharmacokinetics: Wistar rat,[a] 10 mgkgꢀ1 p.o.
AUC [nghmLꢀ1
]
770
372
0.5
8
822
818
0.5
11
Cmax [ngmLꢀ1
tmax [h]
]
CYP inhibition
IC50 (CYP3A4T) [mm]
IC50 (CYP3A4M) [mm]
CYP3A4 shift
IC50 (CYP2C9) [mm]
CYP2C9 shift
12
12
30-fold
32
1
26
F [%]
>50
4.2-fold
28
Pharmacokinetics: Beagle dog[b]
AUC [nghmLꢀ1
CL [mLminꢀ1 kgꢀ1
Vss [Lkgꢀ1
Terminal t1/2 [h]
1 mgkgꢀ1 i.v.
1740
9.6
0.7
0.8
0.3 mgkgꢀ1 i.v.
416
12
1.3
1.2
]
1
]
IC50 (CYP2D6) [mm]
CYP2D6 shift
>50
1
>50
1
]
Metabolic stability in liver microsomes
Pharmacokinetics: Beagle dog[b]
5 mgkgꢀ1 p.o.
5050
1400
1
3 mgkgꢀ1 p.o.
3190
936
0.4
HLM [mLminꢀ1 mgꢀ1
]
219
59
>1250
44
AUC [nghmLꢀ1
]
RLM [mLminꢀ1 mgꢀ1
DLM [mLminꢀ1 mgꢀ1
]
]
>1250
165
Cmax [ngmLꢀ1
tmax [h]
]
F [%]
58
77
Plasma protein binding
rat PPB [%]
Fu_rat brain
98.0
0.011
98.5
0.017
Blood–brain barrier (BBB) experiments:[c] total values
[B] [nggꢀ1
[P] [ngmLꢀ1
[CSF] [ngmLꢀ1
[B]/[P] ratio [%]
]
2741
4973
191
55
3394
8488
247
40
]
MDR1 assay
]
Papp A!B [10ꢀ6 cmsꢀ1
Papp B!A [10ꢀ6 cmsꢀ1
Efflux ratio
]
]
7.5
85
11
16
76
4.7
[a] Data are the geometric mean; the median is given for tmax. [b] For for-
mulation details see the experimental section in the Supporting Informa-
tion. [c] Administered at 100 mgkgꢀ1 p.o. in male Wistar rats (see experi-
mental details in the Supporting Information). Sampling at t=3 h post-
dosing; formulation in PEG400.
[a] Orexin activity data are the geometric mean of at least two to three
independent experiments and were determined by FLIPR assay (see the
Supporting Information for details). CYP3A4T: testosterone as marker sub-
strate; CYP3A4M: midazolam as marker substrate; HLM, RLM, and DLM
values represent a normalized rate constant.
tative modifications reported in Table 7 confirmed the strong
affinity of the quinoxalines for the orexin receptors with poten-
cies reached below 10 nm on both receptors [(1S,6R)-67,
(1S,6R)-71, and (1S,6R)-72]. Unfortunately, the compounds
were all strong Pgp substrates. At this stage, we started to sus-
pect the 2H-1,2,3-triazole moiety to be part of the pharmaco-
phore recognized by the Pgp resulting in high Pgp efflux
values. To verify this hypothesis, we replaced the 1,2,3-triazole
ring by 1H-pyrazole, 2-pyridine, or by a simple phenyl ring to
study the impact on Pgp efflux ratio. In Table 8, we report
some representative combinations.[47]
screening cascade, metabolic stability in HLM was assessed
first, followed by the time-dependent CYP3A4 inhibition and
eventually permeability and active transport in a human
MDR1-MDCK assay. Indeed, these key properties for a sleeping
drug were identified as potential liabilities in the lead com-
pound 23a. Thus for HLM values below 500 mLminꢀ1 mgꢀ1, the
compounds were further tested in the time-dependent
CYP3A4 inhibition assay and only those with a shift lower than
5 were tested in the in vitro Pgp assay.
Benzo[d]oxazoles led to poorly or moderately active com-
pounds as illustrated by the 5-fluoro-benzo[d]oxazole present
in 63. More active compounds were obtained when the ben-
zo[d]oxazoles were replaced by 6-chloro or 6-fluoro-benzo[d]-
thiazoles [(1S,6R)-64 and (1S,6R)-65]. Unfortunately, these com-
pounds were still moderate Pgp substrates with increased
shifts in the time-dependent CYP3A4 inhibition as compared
with 23a. The most potent derivatives were obtained when
the benzo[d]oxazole ring was replaced by the quinoxaline
moiety [(1S,6R)-66]. We therefore investigated different substi-
tution patterns on the quinoxaline skeleton in order to keep
the strong potency on both receptors found for (1S,6R)-66 yet
minimizing the CYP3A4 shift and the affinity for Pgp. Represen-
Interestingly for all compounds, we observed an orexin an-
tagonist potency increase for R1 =A<B<C antiparallel to the
affinity for Pgp. Compounds bearing a pyrazole or a 2-pyridine
ring as a replacement for the 1,2,3-triazole were similarly
strong Pgp substrates, while compounds bearing a phenyl ring
were poor Pgp substrates. This left us with five compounds ful-
filling our criteria: (1S,6R)-76c, (1S,6R)-79c, (1S,6R)-80c, (1S,6R)-
81c, and (1S,6R)-82c.
Figure 5 summarizes the BBB results of the five most promis-
ing compounds of Table 8. Based on these results we selected
(1S,6R)-80c which was a highly brain penetrant compound
([B]=2999 nggꢀ1) with an excellent [B]/[P] ratio of 186% for
further characterization.
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
8
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 7. Representative isosteric replacements of 5-chlorobenzo[d]oxazole.
Fused heterocycle
Compd
IC50 [nm][a]
hOx1
HLM
CYP3A4 inhibition shift
Pgp efflux ratio
hOx2
63
26
35
40
n.m.
n.m.
n.m.
4.8
(1S,6R)-64
(1S,6R)-65
12
92
>4
6.0
7.5
125
17
3.8
(1S,6R)-66
(1S,6R)-67
5.6
9.3
8
707
61
>19.6
n.m.
4.3
>2.4
29
68
90
79
13
n.m.
n.m.
n.m.
n.m.
(1S,6R)-69
7.0
8.5
>1250
n.m.
3.1
(1S,6R)-70
19
427
30
(1S,6R)-71
(1S,6R)-72
73
6.4
1.9
2.3
0.8
13
323
92
>2.3
9
>2.8
16
22
n.m.
n.m.
n.m.
74
21
42
n.m.
n.m.
n.m.
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails); n.m.=not measured.
For comparative purposes we prepared the corresponding
regioisomers (1R,6S)-83, 84, 85, and 86 of compound 80c. The
data on HLM, shift in the time-dependent CYP3A4 inhibition
and MDR1, reported in Table 9, show the same trend in poten-
cy and Pgp efflux ratios but indicate that compounds 83, 84,
85, and 86 are highly unstable in HLM. This confirmed our ini-
tial choice to favor the (1S,6R)-3,8-diazabicyclo[4.2.0]octane re-
gioisomer.
To provide large enough amounts of (1S,6R)-80c for full in
vitro and in vivo characterization, an easily scalable synthesis
was developed as illustrated in Scheme 2. The quinoxaline was
synthesized in two steps starting from 4,5-difluorobenzene-1,2-
diamine 87. Condensation with glyoxylic acid followed by
chlorination using POCl3 afforded the 2-chloro-6,7-difluoroqui-
noxaline 89 in 81% over two steps. The 4-methyl-[1,1’-biphen-
yl]-2-carboxylic acid 94 was obtained in three steps starting
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
9
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 8. Bioisosteric replacements for the 5-chlorobenzo[d]oxazole and the 2H-1,2,3-triazole.
Fused heterocycle
R1
Compd
IC50 [nm][a]
HLM
CYP3A4 inhibition shift
Pgp efflux ratio
hOx1
hOx2
A
B
C
75a
75b
75c
53
25
9.2
132
154
26
n.m.
n.m.
n.m.
n.m.
n.m.
n.m.
n.m.
n.m.
n.m.
A
B
C
76a
(1S,6R)-76b
(1S,6R)-76c
31
17
4.0
152
35
6.8
n.m.
158
276
n.m.
>3.8
3.2
n.m.
12
1.6
A
B
C
(1S,6R)-77a
(1S,6R)-77b
(1S,6R)-77c
20
18
4.4
15
29
4.8
242
148
181
23
15
9.3
n.m.
n.m.
n.m.
A
B
C
(1S,6R)-78a
(1S,6R)-78b
(1S,6R)-78c
9.4
n.m.
2.3
26
n.m.
2.8
606
n.m.
>1250
>46
n.m.
23
n.m.
n.m.
n.m.
A
B
C
(1S,6R)-79a
(1S,6R)-79b
(1S,6R)-79c
36
20
3.9
12
8.8
1.4
139
198
196
>4
>4.5
4.6
10
17
3
A
B
C
(1S,6R)-80a
(1S,6R)-80b
(1S,6R)-80c
17
14
4.5
4.0
6.5
1.3
363
489
336
5.1
4.5
2.9
8.5
8.5
1.7
A
B
C
(1S,6R)-81a
(1S,6R)-81b
(1S,6R)-81c
11
11
2.0
4.8
6.6
1.3
193
225
375
>2.8
>2.2
2.6
8.7
14
2.9
A
B
C
82a
82b
(1S,6R)-82c
114
126
12
173
240
17
n.m.
n.m.
383
n.m.
n.m.
3.8
n.m.
n.m.
1.8
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails); n.m.=not measured.
Table 9. (1R,6S)-Regioisomers of (1S,6R)-71, (1R,6S)-80a, (1R,6S)-80b, and (1R,6S)-80c.
Fused heterocycle
R1
Compd
IC50 [nm][a]
HLM
CYP3A4 inhibition shift
Pgp efflux ratio
hOx1
hOx2
I
(1R,6S)-83
(1R,6S)-84
(1R,6S)-85
(1R,6S)-86
1.5
2.0
3.6
1.1
2.1
7.5
4.5
1.4
913
605
694
893
4.5
12.3
9.2
12
8.6
12
A
B
C
4.9
1.2
[a] Values are the geometric mean of at least two to three independent experiments as determined by FLIPR assay (see the Supporting Information for de-
tails).
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
10
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Figure 5. Structure–brain penetration analysis: BBB experiments performed in male Wistar rats after oral administration (100 mgkgꢀ1 formulated in PEG 400;
sampling at 3 h).
Scheme 2. Synthesis of DORA ((1S,6R)-3-(6,7-difluoroquinoxalin-2-yl)-3,8-diazabicyclo[4.2.0]octan-8-yl)(4-methyl-[1,1’-biphenyl]-2-yl)methanone 80c (details
given in the Supporting Information). Reagents and conditions: a) glyoxylic acid monohydrate, EtOH, reflux, 1 h, 86%; b) POCl3, reflux, 30 min, 94%; c) H2SO4
conc. MeOH, reflux, 40 h, 99%; d) Pd(PPh3)4, Na2CO3 2m, toluene, EtOH, 808C, 16 h, 86%; e) NaOH 32%, MeOH, 658C, 24 h, 88%; f) K2CO3, DMF, 608C, 2 h,
84%; g) HCl 4n in dioxane, RT, 5 h, quant.; h) TBTU, DIPEA, CH3CN, RT, 3 h, 80%.
from the commercially available 2-iodo-5-methylbenzoic acid
90. Esterification followed by Suzuki coupling using phenylbor-
onic acid 92 afforded the methyl 4-methyl-[1,1’-biphenyl]-2-car-
boxylate 93 that was further saponified yielding the desired
acid 94 in 75% over three steps. tert-Butyl (1S,6R)-3,8-diazabi-
cyclo[4.2.0]octane-8-carboxylate 95 was synthesized according
to known procedures.[42] Nucleophilic aromatic substitution
using 2-chloro-6,7-difluoroquinoxaline 89 followed by Boc-de-
protection and amide coupling with 4-methyl-[1,1’-biphenyl]-2-
carboxylic acid 94 yielded ((1S,6R)-3-(6,7-difluoroquinoxalin-2-
yl)-3,8-diazabicyclo[4.2.0]octan-8-yl)(4-methyl-[1,1’-biphenyl]-2-
yl)methanone 80c.
The in vitro and in vivo data generated for (1S,6R)-80c are
reported in Tables 10 and 11. Compound (1S,6R)-80c is
a potent DORA not only on human but also on rat and dog re-
ceptors, with an excellent PSA (48.25 ꢁ2) and high passive per-
meability (Papp =34.10ꢀ6 cmsꢀ1). In line with its high lipophilici-
ty, low free fractions in human, dog and rat plasma (0.2–0.4%),
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
11
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Table 10. In vitro data generated with compound (1S,6R)-80c.[a]
Table 11. In vivo data generated with compound (1S,6R)-80c.[a]
Properties
(1S,6R)-80c
Properties
(1S,6R)-80c
PhysChem properties
Pharmacokinetics: Wistar rat,[a,b] 1 mgkgꢀ1 i.v.
Mr [gmolꢀ1
LogD7.4
]
470.52
4.7
48.25
AUC [nghmLꢀ1
CL [mLminꢀ1 kgꢀ1
Vss [Lkgꢀ1
Terminal t1/2 [h]
]
266
63
3
]
PSA [ꢁ2]
]
1.6
Human, rat, and dog orexin receptors
IC50 (hOx1/rOx1/dOx1) [nm]
IC50 (hOx2/rOx2/dOx2) [nm]
4.5/8.8/13
1.3/0.8/7.5
Pharmacokinetics: rat,[a,b] 100 mgkgꢀ1 p.o.
AUC [nghmLꢀ1
]
6160
966
5
Cmax [ngmLꢀ1
tmax [h]
]
CYP inhibition
IC50 (CYP3A4T) [mm]
IC50 (CYP3A4M) [mm]
CYP3A4 shift
IC50 (CYP2C9) [mm]
CYP2C9 shift
11
18
2.9
8.5
1
F [%]
23
Pharmacokinetics: Beagle dog,[a,b] 1 mgkgꢀ1 i.v.
AUC [nghmLꢀ1
CL [mLminꢀ1 kgꢀ1
Vss [Lkgꢀ1
Terminal t1/2 [h]
]
1420
12
3.9
]
IC50 (CYP2D6) [mm]
CYP2D6 shift
37
1
]
6.2
Metabolic stability in liver microsomes
Pharmacokinetics: Beagle dog[a,b]
3 mgkgꢀ1 p.o.
6.5 mgkgꢀ1 p.o.
1040
237
1.3
HLM [mLminꢀ1 mgꢀ1
]
336
>1250
296
AUC [nghmLꢀ1
]
611
108
1.1
11
RLM [mLminꢀ1 mgꢀ1
DLM [mLminꢀ1 mgꢀ1
]
]
Cmax [ngmLꢀ1
tmax [h]
]
F [%]
16
Plasma protein binding
PPB [%] (human/rat/dog)
Fu_rat brain
[a] Data are the geometric mean; the median is given for tmax. [b] For for-
mulation details, see the experimental section in the Supporting Informa-
tion.
99.8/99.6/99.8
0.001
MDR1 assay
Papp A!B [10ꢀ6 cmsꢀ1
Papp B!A [10ꢀ6 cmsꢀ1
Efflux ratio
]
]
20
34
1.7
high intrinsic clearance in RLM (>1250 mLminꢀ1 mgꢀ1), (1S,6R)-
80c showed a high clearance in Wistar rats (63 mLminꢀ1 kgꢀ1).
The volume of distribution at steady-state (Vss) as a general
measure of tissue distribution, was 3.0 Lkgꢀ1 exceeding total
body water volume and indicating significant distribution in
tissues. In vivo pharmacokinetics were determined in male
Wistar rats and in male Beagle dogs. To overcome the high
clearance and allow sufficient exposure for the in vivo pharma-
cology in Wistar rats, the oral PK was performed at
100 mgkgꢀ1. The AUC (6160 nghmLꢀ1, rat) and oral bioavaila-
bility (23%, rat) obtained under these experimental conditions
[a] Orexin activity data represent the geometric mean of at least two to
three independent experiments and were determined by FLIPR assay (see
the Supporting Information for details). CYP3A4T: testosterone as marker
substrate; CYP3A4M: midazolam as marker substrate; HLM, RLM, and
DLM values represent a normalized rate constant.
and in rat brain (0.1%) were measured. No liability on CYP in-
hibition could be identified and the compound was not a Pgp
substrate. It showed moderate stability in HLM and in DLM,
which is favorable for a sleeping drug. Also consistent with its
Figure 6. Effect of (1S,6R)-80c on OXA-induced calcium release. CHO cells expressing either the human Ox1 or human Ox2 receptor were pre-incubated with
dilution series of 80c for 120 min followed by the addition of a dilution series of OxA. Calcium release was recorded, peak fluorescence values were exported,
and concentration–response curves were generated. IC50 values at 3.9 nm OxA were determined and used as a basis to calculate the apparent Kb via the gen-
eralized Cheng–Prusoff equation. Shown is one out of n=2 identical experiments. Values represent arithmetic mean of duplicates ꢁSD.
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
12
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
were acceptable for a broad in vivo pharmacological character-
ization in rats as well as in dogs (see Table 11).
The mode of antagonism of (1S,6R)-80c at Ox1R and Ox2R
was assessed in more detail using a Ca2+ release assay with
stably transfected Chinese hamster ovary (CHO) cells recombi-
nantly expressing human Ox1 or Ox2 receptors. Orexin A con-
centration–response curves (CRC) were generated in the pres-
ence of increasing concentrations of (1S,6R)-80c (Figure 6). The
compound induced rightward shifts of the OXA CRCs and
a suppression of the maximal OXA response demonstrating in-
surmountable antagonism.[48] Apparent Kb values (an approxi-
mation of the inhibitory constant if IC50 values generated at
low agonist concentrations are used) were thus calculated
with the help of the generalized Cheng–Prusoff equation as
described in the Supporting Information. Apparent Kb values
for 120 min antagonist pre-incubation were calculated to be
Kb =1 nm (human Ox1) and 0.32 nm (human Ox2).
Figure 8. Effect of (1S,6R)-80c on the time spent in sleep and wake stages
(percent of total time) during 9 h of active period following administration
in male Beagle dogs (n=7; experiment done in a crossover design). Dogs
were administered a single oral dose of vehicle or (1S,6R)-80c at 30, 100 or
300 mg in the daytime. Data are presented as mean ꢁSEM; Dunnett post-
hoc analysis: **p<0.01, ***p<0.001 (n=7).
With the reasonable rat pharmacokinetics at 100 mgkgꢀ1 in
PEG400, the high brain exposure reached under the same con-
ditions ([B]=2999 nggꢀ1) and the favorable insurmountable
antagonistic profile on both receptors, the effect on sleep pro-
motion of (1S,6R)-80c was assessed as previously described for
22a and 23b. At a single oral dose of 100 mgkgꢀ1, (1S,6R)-80c
significantly decreased by 32% the home cage activity over
the 12 h night period following administration relative to vehi-
cle-treated rats (p<0.01, paired t-test; Figure 7). Based on elec-
dependently increased the time spent in non-REM sleep (one-
way ANOVA; p<0.001) and in quiet wake (one-way ANOVA;
p<0.01). The first significant effects were already observed at
the dose of 30 mg/dog. The time spent in REM sleep was not
significantly increased (one-way ANOVA, p>0.05). Vehicle-
treated dogs slept in total 28% of the 9-hour daylight period
following administration, divided into 84% non-REM sleep and
16% REM sleep. At the highest dose tested (300 mg/dog), the
relative proportion of time spent in non-REM and REM sleep
stayed similar with dogs sleeping in total 44% of the time, di-
vided in 86% non-REM sleep and 14% REM sleep (one-way
ANOVA, p>0.05).
With those results, we further profiled the compound to pre-
pare it for preclinical candidate selection. It was shown not to
be mutagenic in the AMES test but our excitement over the fa-
vorable attributes of (1S,6R)-80c ended with the discovery that
it was highly phototoxic with a photo-irritancy factor (PIF) of
44 and an ED50 of 190 ngmLꢀ1 (0.4 mm), which is in the range
of plasma concentrations achieved in vivo.[49] These data and
the fact that two patent applications that cover our own series
appeared in literature prompted us to suspend further studies
of this compound class.[46,47,50]
Figure 7. Effect of (1S,6R)-80c on the time spent in sleep and wake stages
(percent of total time) during the 12 h night active period following adminis-
tration in male Wistar rats. Rats were administered a single oral dose of vehi-
cle or 100 mgkgꢀ1 of compound. Data are presented as mean ꢁSEM;
*p<0.05, **p<0.01 (n=8).
trophysiological parameters, it significantly decreased (16%)
the time spent in active wake compared with vehicle-treated
rats (p<0.01, paired t-test). The time spent in quiet wake was
nearly unchanged (ꢀ3% relative to vehicle-treated rats, p>
0.05, paired t-test). By compensation, the time spent in non-
REM and REM sleep was significantly increased by 14 and 42%
respectively compared with vehicle-treated rats (p<0.01 for
both, paired t-test). The total sleeping time was increased by
18% compared with vehicle-treated rats. The effects observed
were similar to benchmark compound 3 (Figure 4).
Based on these results and the encouraging PK in Beagle
dogs, we further tested the compound in male Beagle dogs
with single oral dose of 30, 100 and 300 mg total dose formu-
lated in mannitol capsules (n=7; experiment done in a cross-
over design). The animals were treated in the morning during
their active phase. Over the 9-hour daylight period following
administration, the compound dose-dependently decreased
electrophysiological signs of wakefulness (time spent in active
wake; one-way ANOVA, p<0.001; Figure 8). (1S,6R)-80c dose
Conclusions
In summary, we have identified new DORAs following a scaf-
fold-hopping approach using 3 as a template. The replacement
of the initial ring with “in principle equally suited” scaffolds led
to marked differences in activity, and even within one structur-
al class, fine tuning was necessary to obtain optimal overall
properties. Among the different possible starting points, we
decided to pursue with the 3,8-diazabicyclo[4.2.0]octane tem-
plate for SAR optimization. The insurmountable highly brain
penetrant DORA 80c was broadly characterized for physico-
chemical and pharmacokinetic properties as well as in in vivo
pharmacological models. The sleep-promoting effect of 80c
was similar to the effect observed with suvorexant 3. In vitro
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
13
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
S. Buchmann, M. Scherz, T. Weller, W. Fischli, M. Clozel, F. Jenck, Nat.
Med. 2007, 13, 150–155.
phototoxicity prevented us from further pursuing with this
compound.
[17] D. N. Neubauer, Curr. Opin. Invest. Drugs 2010, 11, 101–110.
[18] C. D. Cox, M. J. Breslin, D. B. Whitman, J. D. Schreier, G. B. McGaughey,
M. J. Bogusky, A. J. Roecker, S. P. Mercer, R. A. Bednar, W. Lemaire, J. G.
Bruno, D. R. Reiss, C. M. Harrell, K. L. Murphy, S. L. Garson, S. M. Doran, T.
Prueksaritanont, W. B. Anderson, C. Tang, S. Roller, T. D. Cabalu, D. Cui,
G. D. Hartman, S. D. Young, K. S. Koblan, C. J. Winrow, J. J. Renger, P. J.
Coleman, J. Med. Chem. 2010, 53, 5320–5332.
Acknowledgements
The authors thank Daniel Hafner, Elise Salomꢀ, Daniel Marchal,
Katalin Menyhart, Celia Mꢁller-Grandjean, Thomas Sasse, Cꢀdric
Fischer, Hꢀlꢂne Roellinger, Hꢀlꢂne Massinet, Wenyu Wang, Chang-
bin Qui, Stephane Delahaye, Isabelle Weber, Fabienne Drouet,
Rolf Wuest, Julia Friedrich, Stephanie Hertzog, Pꢃivi Aꢃnismaa,
Markus van Raumer, Bruno Capeleto, Patrick Bouis, Elise Jacob,
Pascal Fullhardt, Christine Metzger, Sandrine Chopinet, Amandine
Fraichard, Kristina Kamin, and Mourad Nabate. Special thanks to
Thomas Weller for his support, his enthusiasm, and for the stimu-
lating discussions.
[19] P. J. Coleman, C. D. Cox, A. J. Roecker, Curr. Top. Med. Chem. 2011, 11,
696–725.
[20] P. Hoever, P. G. Dorffner, H. Benes, T. Penzel, H. Danker-Hopfe, M. J. Bar-
banoj, G. Pillar, B. Saletu, O. Polo, D. Kunz, J. Zeitlhofer, S. Berg, M. Parti-
nen, C. L. Bassetti, B. HÅgl, I. O. Ebrahim, E. Holsboer-Trachsler, H.
Bengtsson, Y. Peker, U. M. Hemmeter, E. Chiossi, G. Hajak, J. Dinge-
manse, Clin. Pharmacol. Ther. 2012, 91, 975–985.
[21] D. Michelson, E. Snyder, E. Paradis, M. Chengan-Liu, D. B. Snavely, J. Hut-
zelmann, J. K. Walsh, A. D. Krystal, R. M. Benca, M. Cohn, C. Lines, T.
Roth, W. J. Herring, Lancet Neurol. 2014, 13, 461–471.
[22] C. J. Winrow, J. J. Renger, Br. J. Pharmacol. 2014, 171, 283–293.
[23] Y. Zhu, Y. Miwa, A. Yamanaka, T. Yada, M. Shibahara, Y. Abe, T. Sakurai, K.
Goto, J. Pharmacol. Sci. 2003, 92, 259–266.
[24] L. Lin, J. Faraco, R. Li, H. Kadotani, W. Rogers, X. Lin, X. Qiu, P. J. de Jong,
S. Nishino, E. Mignot, Cell 1999, 98, 365–376.
Keywords: drug design · orexin receptor antagonists · scaffold
hopping · sleep disorders · structure–activity relationships
[25] S. Nishino, B. Ripley, S. Overeem, G. J. Lammers, E. Mignot, Lancet 2000,
355, 39–40.
[26] P. Bettica, L. Squassante, S. Zamuner, G. Nucci, H. Danker-Hopfe, E. Ratti,
Sleep 2012, 35, 1097–1104.
[1] S. Passarella, M.-T. Duong, Am. J. Health-Syst. Pharm. 2008, 65, 927–934.
[2] M. Daley, C. M. Morin, M. LeBlanc, J. P. Gregoire, J. Savard, L. Baillargeo,
Sleep Med. 2009, 10, 427–438.
[27] L. Citrome, Int. J. Clin. Pract. 2014, 68, 1429–1441.
[3] a) S. S. Sullivan, C. Guilleminault, Expert Opin. Emerging Drugs 2009, 14,
411–422; b) J. J. Renger, Curr. Top. Med. Chem. 2008, 8, 937–953.
[4] K. A. Wafford, B. Ebert, Nat. Rev. Drug Discovery 2008, 7, 530–540.
[5] a) T. Sakurai, A. Amemiya, M. Ishii, I. Matsuzaki, R. M. Chemelli, H.
Tanaka, S. C. Williams, J. A. Richardson, G. P. Kozlowski, S. Wilson, J. R. S.
Arch, R. E. Buckingham, A. C. Haynes, S. A. Carr, R. S. Annan, D. E. McNul-
ty, W. Liu, J. A. Terrett, N. A. Elshourbagy, D. J. Bergsma, M. Yanagisawa,
Cell 1998, 92, 573–585; b) L. de Lecea, T. S. Kilduff, C. Peyron, X.-B. Gao,
P. E. Foye, P. E. Danielson, C. Fukuhara, E. L. Battenberg, V. T. Gautvik,
F. S. Bartlett II, W. N. Frankel, A. N. van den Pol, F. E. Bloom, K. M. Gautvik,
J. G. Sutcliffe, Proc. Natl. Acad. Sci. USA 1998, 95, 322–327.
[28] Belsomra US Package Insert, United States Food and Drug Administra-
tion product information. Available from: www.accessdata.fda.gov/
scripts/cder/drugsatfda/index.cfm?fuseaction=Search.Label_Approval-
History#labelinfo.
[29] T. Sifferlen, C. Boss, E. Cottreel, R. Koberstein, M. Gude, H. Aissaoui, T.
Weller, J. Gatfield, C. Brisbare-Roch, F. Jenck, Bioorg. Med. Chem. Lett.
2010, 20, 1539–1542.
[30] H. Aissaoui, R. Koberstein, C. Zumbrunn, J. Gatfield, C. Brisbare-Roch, F.
Jenck, A. Treiber, C. Boss, Bioorg. Med. Chem. Lett. 2008, 18, 5729–5733.
[31] T. Sifferlen, R. Koberstein, E. Cottreel, A. Boller, T. Weller, J. Gatfield, C.
Brisbare-Roch, F. Jenck, C. Boss, Bioorg. Med. Chem. Lett. 2013, 23,
3857–3863.
[6] C. Peyron, D. K. Tighe, A. N. van den Pol, L. de Lecea, H. C. Heller, J. G.
Sutcliffe, T. S. Kilduff, J. Neurosci. 1998, 18, 9996–10015.
[32] J. T. Williams, J. Gatfield, C. Roch, A. Treiber, F. Jenck, M. H. Bolli, C. Brot-
schi, T. Sifferlen, B. Heidmann, C. Boss, MedChemComm 2015, 6, 1054–
1064.
[33] T. Sifferlen, A. Boller, A. Chardonneau, E. Cottreel, J. Hoecker, H. Aissaoui,
J. T. Williams, C. Brotschi, B. Heidmann, R. Siegrist, J. Gatfield, A. Treiber,
C. Brisbare-Roch, F. Jenck, C. Boss, Bioorg. Med. Chem. Lett. 2014, 24,
1201–1208.
[34] T. Sifferlen, A. Boller, A. Chardonneau, E. Cottreel, J. Gatfield, A. Treiber,
C. Roch, F. Jenck, H. Aissaoui, J. T. Williams, C. Brotschi, B. Heidmann, R.
Siegrist, C. Boss, Bioorg. Med. Chem. Lett. 2015, 25, 1884–1891.
[35] C. Boss, C. Roch-Brisbare, M. A. Steiner, A. Treiber, H. Dietrich, F. Jenck,
M. von Raumer, T. Sifferlen, C. Brotschi, B. Heidmann, J. T. Williams, H.
Aissaoui, R. Siegrist, J. Gatfield, ChemMedChem 2014, 9, 2486–2496.
[36] P. Malherbe, O. Roche, A. Marcuz, C. Kratzeisen, J. G. Wettstein, C. Bis-
santz, Mol. Pharmacol. 2010, 78, 81–93.
[37] d.-T. Tran, P. Bonaventure, M. Hack, T. Mirzadegan, C. Dvorak, M. Letavic,
N. Carruthers, T. Lovenberg, S. W. Sutton, Eur. J. Pharmacol. 2011, 667,
120–128.
[38] A. Heifetz, G. B. Morris, P. C. Biggin, O. Barker, T. Fryatt, J. Bentley, D. Hal-
lett, D. Manikowski, S. Pal, R. Reifegerste, M. Slack, R. Law, Biochemistry
2012, 51, 3178–3197.
[39] D. B. Whitman, C. D. Cox, M. J. Breslin, K. M. Brashear, J. D. Schreier, M. J.
Bogusky, R. A. Bednar, W. Lemaire, J. G. Bruno, G. D. Hartman, D. R.
Reiss, C. M. Harrell, R. L. Kraus, Y. Li, S. L. Garson, S. M. Doran, T. Prueksar-
itanont, C. Li, C. J. Winrow, K. S. Koblan, J. J. Renger, P. J. Coleman,
ChemMedChem 2009, 4, 1069–1074.
[40] P. J. Coleman, J. D. Schreier, C. D. Cox, M. J. Breslin, D. B. Whitman, M. J.
Bogusky, G. B. McGaughey, R. A. Bednar, W. Lemaire, S. M. Doran, S. V.
Fox, S. L. Garson, A. L. Gotter, C. M. Harell, D. R. Reiss, T. D. Cabalu, D.
Cui, T. Prueksaritanont, J. Stevens, P. L. Tannenbaum, R. G. Ball, J. Stella-
[7] For a compilation of the extensive investigations carried out to identify
the orexin system and define its physiological roles, see: Hypocretins:
Integrators of Physiological Functions (Eds.: L. de Lecea, J. G. Sutcliffe),
Springer, New York, 2005.
[8] S. Nishino, Sleep Med. 2007, 8, 373–399.
[9] a) C. Boss, C. Brisbare-Roch, F. Jenck, J. Med. Chem. 2009, 52, 891–902;
b) C. Boss, C. Brisbare-Roch, F. Jenck, H. Aissaoui, R. Koberstein, T. Siffer-
len, T. Weller, Chimia 2008, 62, 974–979; c) A. J. Roecker, P. J. Coleman,
Curr. Top. Med. Chem. 2008, 8, 977–987; d) J. Gatfield, C. Brisbare-Roch,
F. Jenck, C. Boss, ChemMedChem 2010, 5, 1197–1214.
[10] a) D. Smart, J. C. Jerman, Pharmacol. Ther. 2002, 94, 51; b) C. R. Bau-
mann, C. L. Bassetti, Lancet Neurol. 2005, 4, 673–682.
[11] M. J. Bingham, J. Cai, M. R. Deehan, Curr. Opin. Drug Discovery Dev.
2006, 9, 551–559.
[12] a) T. Mochizuki, A. Crocker, S. McCormack, M. Yanagisawa, T. Sakurai,
T. E. Scammell, J. Neurosci. 2004, 24, 6291–6300; b) R. M. Chemelli, J. T.
Willie, C. M. Sinton, J. K. Elmquist, T. Scammell, C. Lee, J. A. Richardson,
S. C. Williams, Y. Xiong, Y. Kisanuki, T. E. Fitch, M. Nakazato, R. E.
Hammer, C. B. Saper, M. Yanagisawa, Cell 1999, 98, 437–451.
[13] a) A. S. Foutz, M. M. Mitler, L. L. Cavalli-Sforza, W. C. Dement, Sleep 1979,
1, 413–421; b) J. Riehl, S. Nishino, R. Cederberg, W. C. Dement, E.
Mignot, Exp. Neurol. 1998, 152, 292–302.
[14] C. Peyron, J. Faraco, W. Rogers, B. Ripley, S. Overeem, Y. Charnay, S. Nev-
simalova, M. Aldrich, D. Reynolds, R. Albin, R. Li, M. Hungs, M. Pedrazzo-
li, M. Padigaru, M. Kucherlapati, J. Fan, R. Maki, G. J. Lammers, C. Bouras,
R. Kucherlapati, S. Nishino, E. A. Mignot, Nat. Med. 2000, 6, 991–997.
[15] J. Cai, F. E. Cooke, B. S. Sherborne, Expert Opin. Ther. Pat. 2006, 16, 631–
646.
[16] C. Brisbare-Roch, J. Dingemanse, R. Koberstein, P. Hoever, H. Aissaoui, S.
Flores, C. Mueller, O. Nayler, J. van Gerven, S. L. de Haas, P. Hess, C. Qiu,
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
14
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
bott, S. D. Young, G. D. Hartman, C. J. Winrow, J. J. Renger, ChemMed-
Chem 2012, 7, 415–424.
and aromatic systems at the azetidine ring. As feared, all interesting
compounds showed high CYP3A4 time-dependent shift and/or were
prone to be strong Pgp substrates. The same observations were made
with the 3,6-diazabicyclo[3.2.0]heptane core template: C. Boss, C. Brot-
schi, B. Heidmann, T. Sifferlen, J. T. Williams (Actelion Pharmaceuticals
Ltd.), Int. PCT Pub. No. WO2012085857 A1, 2012.
[41] a) C. D. Cox, G. B. McGaughey, M. J. Bogusky, D. B. Whitman, R. G. Ball,
C. J. Winrow, J. J. Renger, P. J. Coleman, Bioorg. Med. Chem. Lett. 2009,
19, 2997–3001; b) P. J. Coleman, J. D. Schreier, G. B. McGaughey, M. J.
Bogusky, C. D. Cox, G. D. Hartman, R. G. Ball, C. M. Harrell, D. R. Reiss, T.
Prueksaritonont, C. J. Winrow, J. J. Renger, Bioorg. Med. Chem. Lett.
2010, 20, 2311–2315.
[42] J. M. Frost nꢃe Pac, W. H. Bunnelle, K. R. Tietje, D. J. Anderson, L. E.
Rueter, P. Curzon, C. S. Surowy, J. Ji, J. F. Daanen, K. L. Kohlhaas, M. J.
Buckley, R. F. Henry, T. Dyhring, P. K. Ahring, M. D. Meyer, J. Med. Chem.
2006, 49, 7843–7853.
[43] For SAR around 3,7-diazabicyclo[3.3.1]nonane and 9-oxa-3,7-diazabicy-
clo[3.3.1]nonane, see: C. Boss, C. Brotschi, B. Heidmann, T. Sifferlen, J. T.
Williams (Actelion Pharmaceuticals Ltd.), Int. PCT Pub. No.
WO2013050938 A1, 2013.
[44] Racemic 22 was resolved on a chiral Daicel ChiralPak IC HPLC column
(5 mm; 20ꢂ250 mm) eluting with heptane/EtOH+0.1% DIEA 50:50 at
a flow rate of 16 mLminꢀ1. The first eluting enantiomer (22a) has a re-
tention time of tR =10.73 min, and the second eluting enantiomer
(22b) has a retention time of tR =14.18 min.
[47] For a more exhaustive overview of the SAR using the core template
(1S,6R)-3,8-diazabicyclo[4.2.0]octane with amide at the azetidine ring
and aromatic systems at the piperidine ring, see: C. Boss, C. Brotschi, B.
Heidmann, T. Sifferlen, J. T. Williams (Actelion Pharmaceuticals Ltd>),
Int. PCT Pub. No. WO2012085852 A1, 2012.
[48] A. L. Gotter, C. J. Winrow, J. Brunner, S. L. Garson, S. V. Fox, J. Binns,
C. M. Harrell, D. Cui, K. L. Yee, M. Stiteler, J. Stevens, A. Savitz, P. L. Tan-
nenbaum, S. J. Tye, T. McDonald, L. Yao, S. D. Kuduk, J. Uslaner, P. J. Co-
leman, J. J. Rengen, BMC Neurosci. 2013, 14, 90.
[49] The compound was tested in the NRU3T3 in vitro assay using chlorpro-
mazine as positive control. A compound in this assay is flagged as pho-
totoxic for PIF values exceeding 5. The PIF value for chlorpromazine in
this assay is 23.
[50] B. J. Branstetter, M. A. Letavic, K. S. Ly, D. A. Rudolph, B. M. Savall, C. R.
Shah, B. T. Shireman (Janssen Pharmaceutica NV), Int. PCT Pub. Nos.
WO2011050200 A1 and WO2011050202 A1.
[45] Racemic 23 was resolved on a chiral Daicel ChiralPak IA HPLC column
(5 mm; 30ꢂ250 mm) eluting with CH3CN/MeOH+0.1% DIEA 95:5 at
a flow rate of 34 mLminꢀ1. The first eluting enantiomer (23a) has a re-
tention time of tR =8.62 min, and the second eluting enantiomer (23b)
has a retention time of tR =11.90 min.
Received: March 29, 2016
Revised: May 18, 2016
[46] Some SAR work was also later performed using the core template
(1R,6S)-3,8-diazabicyclo[4.2.0]octane with amide at the piperidine ring
Published online on && &&, 0000
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
15
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
B. Heidmann,* J. Gatfield, C. Roch,
A. Treiber, S. Tortoioli, C. Brotschi,
J. T. Williams, M. H. Bolli, S. Abele,
T. Sifferlen, F. Jenck, C. Boss
Scaffold hopping for hits: Dual orexin
receptor antagonism has been shown
to be a new mechanism for the treat-
ment of primary insomnia and other
sleep-related disorders. We report a scaf-
fold-hopping approach as a powerful
tool to identify high-quality hits as well
as the structural optimization of one
scaffold into 80c, a potent and orally
bioavailable brain-penetrant dual orexin
receptor antagonist.
&& – &&
Discovery of Highly Potent Dual
Orexin Receptor Antagonists via
a Scaffold-Hopping Approach
&
ChemMedChem 2016, 11, 1 – 16
www.chemmedchem.org
16
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!