Discovery of the dual orexin receptor antagonist [(7 R)-4-(5-chloro-1,3- benzoxazol-2-yl)-7-methyl-1,4-diazepan-1-yl][5-methyl-2-(2 H -1,2,3-triazol-2-yl)phenyl]methanone (MK-4305) for the treatment of insomnia

Article abstract of DOI:10.1021/jm100541c

Despite increased understanding of the biological basis for sleep control in the brain, few novel mechanisms for the treatment of insomnia have been identified in recent years. One notable exception is inhibition of the excitatory neuropeptides orexins A and B by design of orexin receptor antagonists. Herein, we describe how efforts to understand the origin of poor oral pharmacokinetics in a leading HTS-derived diazepane orexin receptor antagonist led to the identification of compound 10 with a 7-methyl substitution on the diazepane core. Though 10 displayed good potency, improved pharmacokinetics, and excellent in vivo efficacy, it formed reactive metabolites in microsomal incubations. A mechanistic hypothesis coupled with an in vitro assay to assess bioactivation led to replacement of the fluoroquinazoline ring of 10 with a chlorobenzoxazole to provide 3 (MK-4305), a potent dual orexin receptor antagonist that is currently being tested in phase III clinical trials for the treatment of primary insomnia.

Full text of DOI:10.1021/jm100541c

5320 J. Med. Chem. 2010, 53, 5320–5332
DOI: 10.1021/jm100541c
Discovery of the Dual Orexin Receptor Antagonist [(7R)-4-(5-Chloro-1,3-benzoxazol-2-yl)-
7-methyl-1,4-diazepan-1-yl][5-methyl-2-(2H-1,2,3-triazol-2-yl)phenyl]methanone (MK-4305) for the
Treatment of Insomnia
Christopher D. Cox,*,† Michael J. Breslin,† David B. Whitman,† John D. Schreier,† Georgia B. McGaughey,‡
Michael J. Bogusky,† Anthony J. Roecker,† Swati P. Mercer,† Rodney A. Bednar,§ Wei Lemaire,§ Joseph G. Bruno,§
Duane R. Reiss, C. Meacham Harrell, Kathy L. Murphy, Susan L. Garson, Scott M. Doran, Thomayant Prueksaritanont,^
Wayne B. Anderson,^ Cuyue Tang,^ Shane Roller,^ Tamara D. Cabalu,^ Donghui Cui,^ George D. Hartman,†
Steven D. Young,† Ken S. Koblan, Christopher J. Winrow, John J. Renger, and Paul J. Coleman†
†Departments of Medicinal Chemistry, ‡Chemistry Modeling and Informatics, §In Vitro Sciences, Depression and Circadian Disorders, and
^Drug Metabolism, Merck Research Laboratories, P.O. Box 4, Sumneytown Pike, West Point, Pennsylvania 19486
Received May 5, 2010
Despite increased understanding of the biological basis for sleep control in the brain, few novel
mechanisms for the treatment of insomnia have been identified in recent years. One notable exception
is inhibition of the excitatory neuropeptides orexins A and B by design of orexin receptor antagonists.
Herein, we describe how efforts to understand the origin of poor oral pharmacokinetics in a leading HTS-
derived diazepane orexin receptor antagonist led to the identification of compound 10 with a 7-methyl
substitution on the diazepane core. Though 10 displayed good potency, improved pharmacokinetics,
and excellent in vivo efficacy, it formed reactive metabolites in microsomal incubations. A mechanistic
hypothesis coupled with an in vitro assay to assess bioactivation led to replacement of the fluoroquina-
zoline ring of 10 with a chlorobenzoxazole to provide 3 (MK-4305), a potent dual orexin receptor
antagonist that is currently being tested in phase III clinical trials for the treatment of primary insomnia.
Introduction
dependence, rebound insomnia, or safety concerns remain
elusive.4 The most widely prescribed agents for treatment of
insomnia are central nervous system depressants that enhance
signaling of the inhibitory neurotransmitter γ-aminobutyric
acid (GABAa). Developed beginning in the early 1960s, benzo-
diazepines decrease sleep onset time by binding to GABAA
receptors on postsynaptic neurons in the central nervous
system but are associated with issues such as daytime sleepi-
ness, dizziness, impaired memory, and risk of dependence.1 The
subtype selective benzodiazepine and nonbenzodiazepine
GABAA modulators developed more recently, such as zolpi-
dem, zaleplon, and eszopiclone, address some of the issues
associated with earlier agents but questions remain about the
overall safety and efficacy of these agents.3 The FDA recently
mandated black-box warnings related to the safety profile of all
sedative hypnotics, underscoring the agency’s concern with the
currently available treatments.5 In 2005, the FDA approved
the melatonin receptor antagonist ramelteon for the treat-
ment of insomnia, representing the first novel mechanism to
be approved in 35 years and the only hypnotic not classified as a
Insomnia is characterized by difficulty initiating and/or
maintaining sleep, waking early, or finding sleep nonrestora-
tive. In a recent poll of American adults, over 50% reported
having at least one symptom of insomnia during the previous
year, and a third reported symptoms almost every night.1
Insomnia is considered primary if no known underlying
mental or physical condition leads to its development, and
has been directly linked to significant daytime consequences
such as irritability, inability to concentrate, reduced energy
levels, poor job performance, work absenteeism, and a greater
risk of traffic and work-related accidents as well as a higher
incidence of health problems.1,2 Insomnia is also commonly
experienced as a secondary symptom of comorbid disorders.
The direct and indirect costs to the US economy related to
insomnia have been estimated to be well in excess of $100
billion per year.3 It is believed that a significant number of
cases go undiagnosed and/or untreated due to an opinion by
many sufferers that sleep problems are inconsequential and a
pervasive belief by physicians that currently available treat-
ments have questionable safety and efficacy.1 Notwithstand-
ing, one-quarter of American adults take sleep medications
during the course of a year, and the U.S. market for sleep
drugs is estimated to reach $5 billion per year by 2010.3a
Despite the physical, psychological, and economic toll of
insomnia, treatments that improve total sleep time and sub-
jective sleep quality without risk of next-day impairment,
a Abbreviations: BDC, bile-duct cannulated; Boc, tert-butoxycarbo-
nyl; BSA, bovine serum albumin; CSF, cerebrospinal fluid; DORA, dual
orexin receptor antagonist; ECoG, electrocorticogram; EMG, electro-
myogram; FLIPR, fluorometric imaging plate reader; GABA, γ-ami-
nobutyric acid; GSH, glutathione; ICV, intracerebroventricular; LPS,
latency to persistent sleep; NSB, nonspecific binding; OTC, over-the-
counter; OX1R, orexin receptor 1; OX2R, orexin receptor 2; PEG,
polyethylene glycol; REM, rapid-eye movement; SORA, selective orexin
receptor antagonist; SWS, slow-wave sleep; TB, total binding; TPGS,
D-R-tocopheryl polyethylene glycol-100-succinate; WASO, wake after
sleep onset.
*To whom correspondence should be addressed. Phone: þ 1-215-652-
2411. Fax: þ 1-215-652-7310. E-mail: chris_cox@merck.com.
r
pubs.acs.org/jmc
Published on Web 06/21/2010
2010 American Chemical Society

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5321
Figure 1. DORAs studied to date in clinical trials.
controlled substance.6 However, questions remain about the
effectiveness of ramelteon relative to GABAA modulators.7
Clearly, novel mechanisms for the treatment of insomnia that
simultaneously address the need for improved safety and
efficacy are of great interest, and an increased understanding
of the neuronal circuitry involved in the sleep/wake cycle has
spurred a rational search for such targets.
One of the most promising discoveries in this regard is the
orexin (or hypocretin) system that is highly conserved across
species and intricately linked to the regulation of sleep and
wake states.8 Orexins A and B are neuropeptides derived from
the same prepro-orexin peptide that is produced in neurons
located in the lateral hypothalamus. These neurons project
widely to regions of the brain involved in regulation of
arousal. Discovered in 1998 by independent research groups,
the orexin peptides were found to bind to the previously
identified orphan G protein-coupled receptors, orexin recep-
tor 1 (OX1R) and orexin receptor 2 (OX2R).9 Whereas OX1R
binds orexin A with greater affinity than orexin B, OX2R
binds to both peptides with equal affinity.
A number of genetic and pharmacological studies have
confirmed the central role that orexin signaling plays in
regulation of the sleep/wake cycle. Mice engineered to be
deficient in orexin peptides demonstrate a robust phenotype
consistent with human narcolepsy, a sleep disorder character-
ized by excessive daytime sleepiness, sleep fragmentation, and
abnormal transitions to REM sleep.10 Selective knockout of
OX1R in mice revealed a mild phenotype characterized by
sleep fragmentation, whereas knockout of OX2R produced a
narcoleptic phenotype but one that is less severe than the
peptide knockout. Simultaneous knockout of both OX1R and
OX2R produced an acute narcoleptic phenotype similar to the
peptide-deficient animals.11 Additionally, a colony of dogs
that naturallysuffer fromnarcolepsy possessa mutationin the
gene coding for the OX2R,12 whereas the disorder in humans
has been linked to a reduction of orexin-producing neurons,
resulting in low or undetectable levels of orexin A in the
cerebrospinal fluid (CSF) of narcoleptics.13 Consistent with
these observations, direct intracerebroventricular (ICV) infu-
sion of the orexin peptides into rat CSF leads to an increase in
arousal and a decrease in REM and non-REM sleep.14
Because hypofunction of the orexin system leads to de-
creased wakefulness, orexin receptor antagonists have emerged
as a potential target for the treatment of insomnia as well as
conditions resulting from disruption of the natural circadian
cycle such as shift work or jet lag.15 However, on the basis of
genetic studies, the potential for orexin receptor blockade
to cause cataplexy, a condition closely linked to a subset
of narcolepsy sufferers and characterized by sudden loss of
muscle tone, was identified as a potential concern. A number of
laboratories therefore embarked on a search for selective
(OX1R or OX2R) as well as dual (OX1R/OX2R) orexin
receptor antagonists to better characterize their thera-
peutic potential. These efforts have resulted in identification
of potent dual orexin receptor antagonists (DORAs) as
well as selective orexin receptor antagonists (SORAs) that
have enabled pharmacological studies of their in vivo
activity.15c,d,16 Most work to date has focused on DORAs,
likely due to the fact that genetic studies suggested a more
robust effect on the sleep/wake cycle when both receptors are
affected;11 however, a recent publication revealed that an
OX2R-selective SORA promotes sleep in rats.17
The most fully characterized DORA studied to date is (2R)-
2-[(1S)-6,7-dimethoxy-1-{2-[4-(trifluoromethyl)phenyl]ethyl}-
3,4-dihydroisoquinolin-2(1H)-yl]-N-methyl-2-phenylethan-
amide(1, Almorexant), a compound currently in phase III clinical
trials for the treatment of primary insomnia (Figure 1).18
Developed by Actelion Pharmaceuticals, it has been shown
to significantly decrease wakefulness in rats, dogs, and humans,
with no evidence of cataplexy.19 In a phase II trial, Almorexant
provided clinical proof of concept for the orexin antagonist
mechanism in the treatment of insomnia, improving sleep
efficiency in primary insomniacs, and showing no significant
next-day effects on motor function or reaction time.20 The
second entry into the clinic was the DORA N-{[(2S)-1-{[5-(4-
fluorophenyl)-2-methyl-1,3-thiazol-4-yl]carbonyl}piperidin-
2-yl]methyl}-1-benzofuran-4-carboxamide (2, SB-649868)
from GlaxoSmithKline. It showed positive results in terms
of efficacy without residual next day effects; however, pre-
clinical toxicology findings have delayed advancement of
this compound.21 Importantly, both compounds have been
shown to induce a natural sleep architecture characterized by
increases in the time spent in REM and non-REM sleep,
differentiating them from GABAA modulators such as zolpi-
dem that reduce time spent in these sleep stages believed to be
important for consolidation of memory and the restorative
functions of sleep.15c In December 2008, Merck presented
preliminary phase I results indicating that DORA [(7R)-4-(5-
chloro-1,3-benzoxazol-2-yl)-7-methyl-1,4-diazepan-1-yl]-
[5-methyl-2-(2H-1,2,3-triazol-2-yl)phenyl]methanone (3, MK-
4305) decreased latency to persistent sleep (LPS) and wake
after sleep onset (WASO), with no disruption of sleep archi-
tecture, in healthy volunteers.22
The favorable clinical results obtained with three structurally
diverse DORAs underscores the promise that orexin receptor
antagonism holds for the treatment of sleep disorders. In this
article, we describe the discovery and preclinical profile of 3,
a potent brain-penetrant DORA that recently entered phase III
clinical trials for the treatment of primary insomnia.23
Results and Discussion
A recent publication from our group described the dis-
covery of 4, a novel diazepane-based DORA identified by hit-
to-lead efforts following high throughput screening of the

5322 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Cox et al.
Merck sample collection (Figure 2).24 Compound 4 demon-
strates low nanomolar affinity for the human OX1R and
OX2R as measured by a radioligand-displacement binding
assay (expressed as Ki), as well as excellent potency in a
fluorometric imaging plate reader (FLIPR) assay that pro-
vides a functional readout of orexin receptor antagonism in
CHO cells engineered to overexpress the human receptors
(expressed as IC50). Itwas found to promote sleepinrats when
dosed orally at 100 mg/kg by increasing the amount of time
spent in both REM and delta sleep.25 This observation is
consistent with the profile of clinically studied DORAs 1 and
215c and opposite to the effect seen following ICV adminis-
tration of the native peptide.14
Though 4 has excellent brain penetration, good physical
properties, a clean ancillary profile, and produces desired
effects on sleep architecture, it suffers from high clearance
and low bioavailability in the rat (Table 1). In the dog, 4
displays moderate plasma clearance (11.6 mL/min/kg) but
low bioavailability (16%). In light of the observed clearance
(calculated blood clearance accounts for ∼50% of hepatic
blood flow), maximal bioavailability is anticipated to be
∼50% in dog assuming complete absorption and primary
elimination via the liver. Thus, the observed low bioavail-
ability was unexpected and suggested the possibility of poor
absorption in this species. However, screening a number of
vehicles did not result in any improvement over polyethylene
glycol (PEG) 200 used in the initial study.26
Bile-duct cannulated (BDC) dog and portal/cephalic vein
infusion studies were therefore carried out to better under-
stand the reason for poor bioavailability. These studies
revealed that 4 is well absorbed when dosed in PEG 200,
with 78% of the total dose recovered in bile and urine
as products of oxidative metabolism. Hepatic extraction,
determined by comparing plasma AUCs during infusion
via the portal or cephalic vein, was 80%. Therefore, despite
the moderate clearance of 4 in dog, hepatic first-pass
metabolism was the major factor limiting its bioavailability.
As the dog was predicted to be most like human in phar-
macokinetic profile based on the similar intrinsic clearance
in these two species, improved dog pharmacokinetics
(reduced clearance, increased bioavailability) was required
to advance a compound into development. We therefore
began a rational approach to identify molecules with this
potential.
In vitro metabolism studies of 4 in microsomal prepara-
tions indicated several sites of metabolism that we sought
to eliminate or modify by synthetic modifications. A major
metabolic pathway identified was oxidation of the tolyl
methyl group on the 5-methyl-2-(triazolyl)-benzamide por-
tion of the molecule to form the benzyl alcohol and further
oxidation to the benzoic acid. Additionally, complex meta-
bolic patterns were observed on the remainder of the molecule
that indicated oxidation was occurring on both the core and
the quinazoline heterocycle. Interestingly, when 4 was incu-
bated in microsomal preparations fortified with semicarb-
azide, mass spectral adducts consistent with structure 4b were
identified (Scheme 1). Mechanistically, this is suggestive of
oxidation at one of the four positions on the core adjacent to a
nitrogen atom to form intermediate 4a, followed by ring-
opening to reveal an aldehyde, which is subsequently trapped
by semicarbazide.27 Taken as a whole, the metabolism data in
hand suggested that removal of the benzylic methyl group,
alteration of the heterocycle, and chemical modification of the
core, especially blocking positions adjacent to the nitrogen
atoms, were logical tactics to employ in an effort to improve
pharmacokinetics in the diazepane series.
With this knowledge, we began an intensive synthetic
investigation focused on identification of novel diazepane
cores that had substitution at one or more carbon atoms.
During this effort, we identified over 15 mono- or disubsti-
tuted diazepane cores that provided potent DORAs (Ki <
5 nM on both OX1R and OX2R). In general, substitution at
positions3 and5 ofthe diazepanering was nottoleratedbythe
receptors, whereas substitution at positions 2, 6, and 7 was not
only tolerated but often potency enhancing (Figure 3).28
With a large number of potent diazepane cores in hand, we
combined these with the optimal triazolyl-benzamide substi-
tuent and various heterocycles to assemble a large number of
potent DORAs for pharmacokinetic evaluation. Our expec-
tation was that reduced IV clearance would predict improved
bioavailability as a result of reduced metabolism, so we
studied many compounds in parallel by dosing IV in the
dog. Promising analogues were then followed up by PO
dosing to determine bioavailability.29 On the basis of the
overall potency, ancillary, and pharmacokinetic profiles of
analogues made with a methyl substituent at the 7-position of
the diazepane core, these analogues were targeted for further
investigation.
Surprisingly, a search of the literature revealed a dearth of
practical synthetic strategies to access a simple orthogonally
protected 7-methyldiazepane.30 The most promising route
Figure 3. General scaffold for examining mono- and disubstituted
diazepane cores.
Figure 2. Previously reported diazepane-based DORA 4.
Table 1. Pharmacokinetics of 4 in Rat and Dog
IVa
POb
species
dose (mg/kg)
Cl (mL/min/kg)
Vdss (L/kg)
T1/2 (h)
dose (mg/kg)
AUC ( μM h)
Cmax ( μM)
F (%)
3
rat
dog
2
0.5
52.6
11.6
1.08
0.86
0.35
1.28
100
3
1.50
1.81
0.65
1.44
2
16
a Vehicle = DMSO; n = 2 in rat, n = 3 in dog. b Rat: dosed as the HCl salt in 20% TPGS; n = 3; dog: dosed as free base in PEG 200, n = 2.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5323
Scheme 1. Proposed Mechanism of Semicarbazide Trapping When 4 Is Incubated in Liver Microsomes Fortified with Semicarbazide
Boc deprotection and subsequent intramolecular reductive
amination with NaBH(OAc)3 forms the diazepane ring. A
Boc group is reinstalled to simplify purification, resulting in a
38% overall yield of 6 for the four steps from methyl vinyl
ketone. Importantly, this chemistry can be readily carried out
on large scale to provide access to quantities of 6 for analogue
synthesis and in vivo evaluation of leading compounds.
Subsequent studies indicated that the (R)-antipode of
analogues in this series had superior potency for the orexin
receptors.32 A chiral stationary phase HPLC resolution was
therefore carried out on racemic 6 to provide the desired
(R)-enantiomer as the first eluting isomer. To complete the
synthesis of analogues in this series, Boc deprotection fol-
lowed by amide coupling to 2-(2H-1,2,3-triazol-2-yl)-5-
methylbenzoic acid provided 7 in excellent yield. Removal
of the benzyl carbamate followed by reaction with 2-chloro-
quinazoline provided 8, the 7-(R)-methyldiazepane analogue
of 4 in 75% yield from 7.
Scheme 2. Previously Reported Synthesis of the 7-Methyldi-
azepane Core30ba
a Reagents and conditions: (a) Et2O (70%); (b) PPh3, Et2O (72%);
(c), LAH, Et2O (91%).
Scheme 3. Synthesis of 7-Methyldiazepane DORA Analoguesa
As illustrated in Table 2, 8 displayed excellent potency for
OX1R and OX2R receptors in both binding and FLIPR
assays, but a reduction in dog clearance did not lead to
improved oral bioavailability relative to 4. Synthesis of 9
which lacks the benzylic methyl group of 8 provided an
analogue with similar IV clearance but with diminished
potency on OX1R as measured in the binding assay.33
Installation of a fluorine atom at the 6-position of the
quinazoline, a substitution we found to impart improved
pharmacokinetics to a number of analogues in the diazepane
series,25b provided 10 with improved potency, a further
reduction in IV clearance, and favorable bioavailability of
37% in dog.
Similar to compound 4, 10 is not a substrate for Pgp efflux
(BA/AB = 0.6), has excellent passive permeability (38 ꢀ 10-6
cm/s), a plasma free fraction of ∼3% in rat, and possesses a
clean ancillary profile as determined by an MDS Pharma off-
target screen. In vivo analysis in rat indicates 10 demonstrates
reasonable rat pharmacokinetics with a clearance of 33 mL/
min/kg and oral bioavailability of 47% when dosed at 10 mg/
kg in 20% D-R-tocopheryl polyethylene glycol-100-succinate
(TPGS)34 and maintains a favorable brain/plasma ratio of
0.4-0.6 and a CSF/plasma ratio of 0.02-0.03.35 Compound
10 possesses the overall profile we were seeking in a develop-
ment compound, and it was therefore advanced into a rat
sleep assay to assess its effects on sleep architecture.
The preclinical sleep laboratory at Merck makes use of
radiotelemetric recording of electrocorticogram (ECoG) and
electromyogram (EMG) signals to assess the time spent by
animals in various states of arousal, including active wake,
light sleep, REM sleep, and delta or slow wave sleep (SWS).
Upon dosing 10 orally at 10 mg/kg/day to telemetry-im-
planted rats during their active phase (lights off), the amount
oftime spentin eachstate ofarousalwas measured andbinned
into 30 min intervals and compared to vehicle treated rats
(Figure 4). Similar to the effects seen with compound 4, 10
a Reagents and conditions: (a) Et2O, then TEA and benzyl chloro-
formate; (b) HCl(g), EtOAc; (c) NaBH(OAc)3, HOAc, CH2Cl2; (d)
Boc2O, TEA, CH2Cl2 (38% for 4 steps); (e) Chiralpak AD, 60% EtOH
in hexanes (0.1% DEA as modifier), first isomer to elute is the desired
(R)-isomer; (f ) HCl(g), EtOAc; (g) 2-(2H-1,2,3-triazol-2-yl)-5-methyl-
benzoic acid, EDC, HOAT, NMM, DMF (87% for 2 steps);
(h) H2, Pd(OH)2, MeOH/EtOAc; (i) 2-chloroquinazoline, K2CO3, DMF
(75% for 2 steps).
involved conjugate addition of a 1,2-amino azide to methyl
vinyl ketone, followed by an intramolecular aza-Wittig cycli-
zation and subsequent reduction (Scheme 2).30b Although
short and convergent, this route is unattractive for large scale
synthesis because it necessitates the synthesis and use of a
potentially explosive alkyl azide.31
As illustrated in Scheme 3, this route did, however, provide
inspiration for a related strategy utilizing the commercially
available N-Boc-1,2-diaminoethane as a surrogate for the
azide. In our modified route, conjugate addition of the
monoprotected diamine to methyl vinyl ketone followed by
in situ trapping with benzyl chloroformate provides ketone 5.

5324 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Table 2. Profile of 7-(R)-Methyldiazepane Analogues 8-10
Cox et al.
compd
R1
R2
R3
OX1R Ki, FLIPRa (nM)
OX2R Ki, FLIPRa (nM)
dog Clb (mL/min/kg)
dog F c (%)
4
H
Me
Me
H
H
H
H
F
1.2, 29
0.39, 30
6.2, 26
1.8, 27
0.59, 27
0.36, 30
0.47, 18
0.17, 27
11.6
6.0
6.4
5.2
16
8
Me
Me
Me
<5
ND
37
9
10
H
a n g 3. b n = 2 in cassette (8, 9) or single (4, 10) mode. c n = 2 as a single in PEG 200.
can be trapped with glutathione (GSH) when incubated in
human or rat liver microsomes. It is generally accepted that
reducing the potential for bioactivation of drug candidates
during the lead optimization process is a prudent strategy to
minimize potentialriskfordrug-inducedidiosyncratictoxicity
in humans.36 We therefore next focused attention on reducing
bioactivation of leading diazepanes.
To assess bioactivation potential, we employed an assay
wherein test compounds were incubated in human liver
microsomes fortified with GSH. Upon quenching the reac-
tion, GSH-drug adducts were detected by ultra performance
liquid chromatography-high resolution mass spectrometry
(UPLC/HRMS).37 The area under the peaks for all GSH-
adducts was measured relative to an internal standard (IS),
and the GSH-adduct/IS ratio was used as a semiquantitative
measure of bioactivation potential.
In this assay, compound 10 had a GSH-adduct/IS
ratio of 3.3. On the basis of mass spectral analysis of the
GSH-adducts, we believed that the fluoroquinazoline ring
was largely responsible for the bioactivation observed
in 10.38 We therefore decided to hold constant the optimized
7-methyldiazepane core and triazolyl-benzamide while focus-
ing on heterocycle replacements to maintain good pharmaco-
kinetics and reduce bioactivation.
As illustrated in Table 3, replacing the fluoroquinazoline
ring of 10 with a difluoroquinoxaline to provide 11 main-
tained favorable potency and IV clearance, but had little effect
on the GSH-adduct/IS ratio. Believing a change from the
6,6-fused saturated ring system was required to reduce bio-
activation, an analogue incorporating a 5,6,7,8-tetrahydro-
quinazoline ring was tested (12). Though potency remained
excellent, dog clearance was adversely affected by this struc-
tural modification. Importantly, a significant improvement
in the GSH-adduct/IS ratio was observed, indicating that
progress toward reduced bioactivation was being made by
alteration of the heterocycle.
Analysis of the historical dog IV data collected on the
program to this point identified benzoxazole-containing di-
azepanes as generally having reduced clearance values relative
to other heterocycles tested. Installation of an unsubstituted
benzoxazole to provide 13 resulted in an analogue with the
lowest plasma clearance in dog observed to date for a diaze-
pane DORA (2.4 mL/min/kg). However, despite inclusion of
the potency-enhancing methyl group on the benzamide,33 13
suffered a nearly 20-fold loss in OX2R binding potency relative
to 10. From previous SAR knowledge dating back to the HTS
hit that led to the discovery of the diazepane series, it was
known that addition of lipophilic groups on benzoxazoles and
Figure 4. Effects of a 10 mg/kg dose of 10 on rat sleep. The duration
of time spent in active wake, light sleep, delta sleep, and REM sleep
in vehicle (closed circles) vs drug (open circles) treated animals
following a 10 mg/kg po dose of 10 in 20% TPGS administered at
09:00. The open area in the bar above the abscissa designates lights
on and the filled area designates lights off. Values are mean ( SEM.
Vertical gray lines represent statistically significant differences from
vehicle treatment as determined using mixed-model ANOVAs at
each time point. Significance levels are indicated by length of tic
marks (short <0.5, medium <0.01, long <0.001).
produced a significant decrease in active wake with a con-
current increase in REM and delta sleep during the three to
four hours following dosing. At approximately one hour post
dose when the greatest effects on sleep are observed, the
plasma exposure of 10 was ∼1.0 μM, which leads to an
estimated 20-30 nM in the CSF. Thus, compounds 4 and
10 induce favorable sleep effects in rats at comparable plasma
exposures, as predicted based on their in vitro profile, but
compound 10works at a 10-fold lower dose (ona mg/kg basis)
due to its improved bioavailability.
Excitement over the favorable attributes of 10 was tem-
pered by the discovery that it undergoes bioactivation to form
reactive electrophilic intermediates (reactive metabolites) that

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5325
Table 3. Profile of 7-(R)-Methyldiazepane Analogues
a n g 3. b n = 2 in cassette (12, 13) or single (10, 11, 3) mode. c n = 2 as a single in PEG 200. d See Experimental Section for a description of how this
number is calculated.
benzothiazoles provided significant potency increases. Indeed,
installation of a chlorine at the 5-position of the benzoxazole
provided 3 with excellent potency and pharmacokinetics as
well as minimal GSH-adduct formation (GSH-adduct/IS
ratio=0.1). We therefore decided to further characterize 3 for
its potential as a development candidate.39
receptor expression, we developed a transgenic line of rats
engineered to overexpress the human OX2R ubiquitously in
the brain at relatively high density (12-14 nM).41 This assay
allowed us to determine receptor occupancy in the rat brain
followingperipheraladministrationof antagonists. Following
a 30 min IV infusion of 3 over a concentration range of
0.1-2.0 mpk, we found that a 1.34 mg/kg dose was required
to occupy 90% of the orexin receptors (Occ90), corresponding
to exposures of 1070 nM in plasma, 1180 nM in brain, and
13 nM in the CSF (Figure 6). This suggests that in the rat sleep
study at 30 mg/kg/day, 3 achieved greater than 90% receptor
occupancy.
As illustrated in Table 4, 3 has high clearance in rat but low
to moderate clearance in dog, with a short terminal half-life of
0.6 and 3.3 h, respectively. Compound 3 also has good
bioavailability in dog with a rapid Tmax but lower bioavail-
ability in rat. Investigation of 3 in BDC rat and dog studies
following IV administration of 14C-labeled drug not unex-
pectedly identified oxidation of the tolyl methyl group to the
corresponding benzyl alcohol (14) as a major route of meta-
bolism(Scheme4). Further oxidation tothecarboxylicacid15
was also observed, as was oxidation of the methyl group on
the core (16) and oxidation on the left-hand portion of the
molecule (17).42 Further conversion to the corresponding
glucuronides was also observed. In rat, dog, and human liver
microsomes, these same oxidative metabolites were observed,
with rat microsomes displaying the highest rate of metabolism
relative to dog and human. On the basis of the in vitro and in
vivo data, and an assumption that oxidative metabolism will
be the primary mode of elimination, 3 was predicted to be a
Like its predecessors, 3 possesses a favorable profile for
a CNS-penetrant drug. It is not a substrate for Pgp efflux
(BA/AB = 0.8), has excellent passive permeability (37 ꢀ
10-6 cm/s), and possesses a clean ancillary profile (>10000-
fold selectivity for OX2R) as determined by an MDS Pharma
off-target screen of 170 enzymes, receptors, and ion channels.
Because of the added lipophilicity introduced by the chloro-
benzoxazole relative to the fluoroquinazoline (the LogP
increased from 3.0 in 10to 3.6for 3), protein binding increased
to∼99% inbothrat andhuman plasma. DORA3 hasa brain/
plasma ratio of ∼1 in the rat, and its CSF concentration is well
reflected by unbound plasma concentration, suggesting a lack
of concentration gradient between plasma and brain.40
When studied in telemetry-implanted rats, 3 induced
significant changes in sleep behavior at 30 mg/kg where, as
expected, significant increases in REM and delta sleep are
observed with a corresponding decrease in active wake
(Figure 5).41 In this study, the plasma exposure of 3 reached
a Cmax of 2.1 μM, which leads to an estimated 21 nM in the
CSF.
Along with the rat sleep assay, another powerful tool that
aided lead optimization efforts was an ex vivo occupancy
assay designed to measure orexin receptor binding directly in
the rat brain. Because of low levels of endogenous orexin

5326 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Cox et al.
low to moderate clearance compound with good bioavail-
ability and a rapid onset of action in humans.
Compound 3 is a semicrystalline solid with good thermal-
and pH-dependent stability and modest aqueous solubility.
It was not genotoxic and was well-tolerated in preclinical
pharmacology and toxicity studies. There were no cardiovas-
cular effectsindogs, and neurobehavioralchanges inmice and
rats were generally limited to transient decreases in locomotor
activity that were consistent with the intended mechanism of
action. Results from one-month general toxicity studies inrats
and dogs supported the safety and tolerability of 3 for short-
term testing in humans. On the basis of the overall profile of 3,
it was approved for development and entered phase I clinical
trials in 2007. In late 2009, a phase III clinical trial was
initiated to study the long-term safety and efficacy of 3 for
the treatment of primary insomnia.23 Additional preclinical as
well as clinical results will be reported in due course.
In conclusion, we have described how a diazepane lead
compound was advanced to provide a potent, brain-penetrant
dual orexin receptor antagonist for the treatment of insomnia.
Efforts to understand the cause of the poor pharmacokinetics
of 4 led to a rational search for chemical modifications that
reduced metabolism. Identification of the favorable 7-methyl
substituent on the diazepane core resulted in compound 10
withgoodpotency, improvedpharmacokinetics, andexcellent
in vivo efficacy. An in vitro assay to assess bioactivation, in
concert with a mechanistic hypothesis as to the site of
bioactivation, led to replacement of the fluoroquinazoline
ring of 10 with a chlorobenzoxazole to provide 3, a potent
dual orexin receptor antagonist that is currently being tested
in phase III studies for the treatment of primary insomnia.
Experimental Section
A. Biology. All animal studies were performed according to
the NIH Guide for the Care and Use of Laboratory Animals,
and experimental protocols were reviewed by the Merck Animal
Care and Use Committee.
Figure 5. Effects of a 30 mg/kg dose of 3 on rat sleep. The duration
of time spent in active wake, light sleep, delta sleep, and REM sleep
in vehicle (closed circles) vs drug (open circles) treated animals
following a 30 mg/kg po dose of 3 in 20% TPGS administered at
09:00. The open area in the bar above the abscissa designates lights
on and the filled area designates lights off. Values are mean ( SEM.
Vertical gray lines represent statistically significant differences from
vehicle treatment as determined using mixed-model ANOVAs at
each time point. Significance levels are indicated by length of tic
marks (short < 0.5, medium < 0.01, long < 0.001).
Radioligand Binding Assay. Membranes were prepared from
expressed human orexin 2 receptor (hOX2R) and the Ile408-Val
variant of orexin 1 receptor (hOX1R) in CHO cells according to
the method described by Kunapuli et al.43 Confluent CHO/
OX2R and CHO/OX1R cells were dissociated from flasks with
PBS/1 mM EDTA and centrifuged at 1000g for 10 min. The cell
pellets were homogenized with a Polytron in ice-cold 20 mM
Hepes, 1 mM EDTA, at pH 7.4, and centrifuged at 20000g for 20
min at 4 ꢀC. This process was repeated twice. The final mem-
brane pellet was resuspended at 5 mg of membrane protein/mL
in assay buffer (20 mM Hepes, 125 mM NaCl, 5 mM KCl, pH
7.4). Bovine serum albumin was added to achieve a final
concentration of 1% and aliquots stored at -80 ꢀC. Radioli-
gand binding assays were performed utilizing an automated
Tecan Liquid handling system and Packard Unifilter-96 as
described by Mosser et al.44 Assays were performed at room
temperature in 96-well microtiter plates with a final assay
volume of 1.0 mL in 20 mM Hepes buffer (pH 7.4) containing
125 nM NaCl and 5 mM KCl. Solutions of test compounds were
prepared in DMSO and serially diluted with DMSO to yield 20
μL of each of 10 solutions differing by 3-fold in concentration.
Nonspecific binding (NSB) is determined using a high-affinity
ligand (1 μM final concentration) and total binding (TB) is
determined by using DMSO (2% final concentrations). A solu-
tion of receptor (30 pM final, typically 2-10 μg membranes),
Figure 6. Plot of occupancy versus dose for compound 3 in trans-
genic rats. Similar plots of occupancy versus concentration of 3 in
plasma, brain, or CSF can be generated.
Table 4. Pharmacokinetic Parameters of 3 in Rat and Dog
IVa
POb
species
dose (mg/kg)
Cl (mL/min/kg)
Vdss (L/kg)
T1/2 (h)
dose (mg/kg)
AUC ( μM h)
Cmax ( μM)
Tmax (h)
F (%)
3
rat
dog
2
0.5
44
2.6
0.8
0.6
3.3
10
3
1.7
16.9
0.48
3.2
0.5
0.5
19
56
4.0
a Vehicle = DMSO; n = 3. b PEG 200, n = 3.

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5327
Scheme 4. Oxidative Metabolites of 3 Identified from a BDC Rat Study
and tritiated ligands (∼80 Ci/mmole) were added to the test
compounds. For the OX2R receptor, 0.15 nM of compound 18
(KD =0.3 nM) was used. For the OX1R receptor, 0.7 nM of
compound 19 (KD = 3 nM) was used. The OX1R assay was also
performed with equivalent results using compound 20 at a
concentration of 0.03 nM (KD = 0.03 nM); however, in this
case 920 μL of membranes was added first to the compounds
followed by the addition of 60 μL of the hot ligand. After
3 h of incubation at room temperature (20 h for compound
20), samples were filtered through Packard GF/B filters
(presoaked in 0.2% PEI, polyethyleninine Sigma P-3143) and
washed five times with 1 mL of cold 20 mM Hepes buffer
(pH 7.4) per wash. After vacuum drying of the filter plates,
50 μL of Packard Microscint-20 was added and bound radio-
activity (CPM bound) determined in a Packard TopCount.
constants in eq 1 that are determined from the controls located
in columns 1 and 12 of each assay plate. The concentration of
radioligand in eq 1 ([Radiolabel]) is calculated for each run based
on the ligand specific radioactivity and the measured CPM of
an aliquot of the assay solution. The apparent dissociation
constant of the radioligand for its receptor (KD) in eq 1, and the
receptor concentration (Bmax) is determined by a hot saturation
experiment.
FLIPR Assay. For intracellular calcium measurements,
Chinese hamster ovary (CHO) cells expressing the Ile408-Val
variant of the orexin 1 receptor or the human orexin 2 receptor,
were grown in Iscove’s modified DMEM containing 2 mM
L-glutamine, 0.5 g/mL G418, 1% hypoxanthine-thymidine sup-
plement, 100 U/mL penicillin, 100 ug/mL streptomycin, and
10% heat-inactivated fetal calf serum. The cells were seeded at
20000 cells/well into Becton-Dickinson black 384-well clear
bottom sterile plates coated with poly-D-lysine. All reagents
were from GIBCO-Invitrogen Corp. The seeded plates were
incubated overnight at 37 ꢀC and 6% CO2. Ala-6,12 human
orexin-A as the agonist was prepared as a 0.5 mM stock solution
in 1% bovine serum albumin (BSA) and diluted in assay buffer
(HBSS containing 20 mM HEPES and 2.5 mM probenecid, pH
7.4) for use in the assay at a final concentration of 0.3-2 nM.
Test compounds were prepared as 10 mM stock solution in
DMSO, then diluted and pipetted in 384-well plates, first in
DMSO, then assay buffer. On the day of the assay, cells were
washed three times with 100 μL assay buffer and then incubated
for 60 min (37 ꢀC, 6% CO2) in 60 μL of assay buffer containing
1 μM Fluo-4AM ester, 0.02% pluronic acid, and 1% BSA. The
dye loading solution was then aspirated and cells washed three
times with 100 μL of assay buffer. Then 30 μL of that same
buffer is left in each well. Within the fluorescent imaging plate
reader (Molecular Devices), test compounds were added to the
plate in a volume of 15 μL, incubated for 5 min, and finally 15 μL
of agonist was added. Fluorescence was measured for each well
at 1 s intervals for 1 min and at 6 s intervals for 4 min, and the
height of each fluorescence peak was compared to the height of
the fluorescence peak induced by 0.3-2 nM Ala-6,12 orexin-A
with buffer in place of antagonist. For each antagonist, the IC50
value (the concentration of compound needed to inhibit 50% of
the agonist response) was determined.
Dose-inhibition profiles for each compound were character-
ized by fitting the data to a four-parameter equation using an
Excel based fitting program as previously described by Mosser
et al.44 The apparent inhibition constant (KI, KI = 10-pKI), the
maximum inhibition at the low plateau relative to “100% inhibi-
tion control” (Imax; e.g. 1 g same as this control), the minimum
inhibition at the high plateau relative to the “0% inhibition
control” (Imin, e.g. 0 g same as the no drug control) and the Hill
slope (nH) are determined by a weighted nonlinear least-squares
fitting of the raw experimental observations (CPM Bound) as
a function of dose ([Drug]) according to the equation below:
ðTB - NSBÞðImax - Imin
Þ
CPMBound ¼
2
3
nH
Rat Sleep Assay. Adult male Sprague-Dawley rats (450-
600 g; Taconic Farms, Germantown, NY) were subcutaneously
implanted with telemetric physiologic monitors (model F50-
EEE or 4ET SI; Data Sciences International, Arden Hills, MN)
that were used to simultaneously record both the electrocorti-
cogram (ECoG) and electromyogram (EMG) activities of the
rat. For placement of the 4ET SI, animals were anesthetized
with isoflurane and electrodes for recording ECoG signals and
EMG signals were placed. Position of the wires are based on the
following coordinates. Channel 1 wire. From Lambda AP þ2,
6
6
7
7
½Drugꢁ
10- pK 1 þ
6
6
6
4
7
7
7
5
!
1 þ
ꢀ
ꢁ
½Radiolabelꢁ
I
KD
þ NSB þ ðTB - NSBÞð1 - Imax
Þ
ð1Þ
The median signal of the “0% inhibition controls” (TB) and
the median signal of the “100% inhibition controls” (NSB) are

5328 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Cox et al.
ML þ2 -2. Channel 2 wires From BREGMA AP þ1.5 ML
þ3.2 (hole 1) AP -10.5 (hole2). Channel 3 wires From BREG-
MA AP -3.0 ML þ1.5, -3.5. EMG lead placement was in neck
muscle. An incision was made ∼3-5 cm in length midline on
the dorsal thorax to form a pocket on the left and right side
of midline, and the telemetry module was placed with a
saddlebag placement method. The animals were given a single
dose of antibiotic (gentomycin, 5.8 mg/kg) and an analgesic
(buprenorphine, 0.1 mL) within 3 h following surgery. The
animals were allowed to recover from surgery for at least two
weeks prior to recording. Throughout these experiments, ani-
mals were housed individually in plastic cages (19 in. ꢀ 101/2 in.
ꢀ 8 in.; Lab Products, Seaford, DE) and were provided water
and food ad libitum. Lights were on a 12 h light: 12 h dark cycle
with lights off at 4:00 a.m. and on at 4:00 p.m. ECoG and EMG
signals were collected simultaneously from all animals using
Dataquest ART software system (Data Sciences International,
Arden Hills, MN), digitally sampled at 500 Hz, and stored on a
PC for off-line analysis.
nights and the results were statistically compared based upon a
mixed ANOVA analysis.
Bioactivation Assay. Human liver microsomes (pooled, BD
Gentest) were preincubated at 1 mg/mL protein in 100 mM
potassium phosphate buffer (pH 7.4) with 10 μM test com-
pound, 1 mM MgCl2, 1 mM EDTA, 5 mM glutathione, and
1 mM NADPH for 60 min at 37 ꢀC. The reactions were
terminated with 25% acetonitrile containing 0.15 μM labetalol
(internal standard). The samples were vortex-mixed and centri-
fuged at 14000 rpm for 10 min. The supernatant from each
sample was transferred to an HPLC vial for HRMS analysis.
UPLC-high resolution mass spectrometry (HRMS) was used
to identify the GSH-derived adducts. The system consis-
ted of a Waters Acquity sample manager and two Waters
Acquity UPLC pumps. HRMS was performed using a Waters
Q-TOF Xevo mass spectrometer. Separation was achieved using
˚
a Phenomenex Synergi 2.5 μm MAX-RP 100 A column
(50 mm ꢀ 2 mm) heated to 60 ꢀC. The mobile phase consisted
of water with 0.1% formic acid (solvent A) and acetonitrile
with 0.1% formic acid (solvent B) at a flow rate of 0.5 mL/min.
The gradient began with 5% solvent B for the first minute,
followed by a linear increase to 15% solvent B over the next
0.5 min. Solvent B was then increased to 50% over the next
11.5 min followed by a further increase to 90% over 2 min. The
column was then washed with 95% solvent B for 1.5 min. At the
end of each run, the column was re-equilibrated for 5 min at the
initial conditions. Mass spectral analyses were carried out
using electrospray ionization in positive ion mode. The ESI
capillary voltage was set at 1.5 kV, and the source and
desolvation temperatures were set at 100 and 600 ꢀC, res-
pectively. The mass scan range was 150 to 1000 amu with
0.25 s/scan. The lock mass of 588.8691 amu was used with a
frequency of one scan per 5 scans. The relative amount of GSH
adducts formed with each test compound was estimated using
peak area ratios. The areas obtained from the mass spectral
peaks associated with GSH-derived adducts were divided by
the area of the internal standard, labetalol.
Ex Vivo Occupancy Assay. Transgenic rats expressing human
OX2R were dosed intravenously by infusion over a 30 min
period or orally with 3 at doses of 0.1-2.0 mg/kg in 25%
hydroxypropyl-β-cyclodextrin and then sacrificed. Samples of
brain were quickly removed and frozen for use in the ex vivo
occupancy assay, while a second set of tissue samples, a plasma
sample, and CSF were frozen for LCMS determination of drug
levels. For the ex vivo assay, approximately 60 mg of cord or
brain was homogenized in 67 volumes of ice-cold assay buffer
(20 mM HEPES, 120 mM NaCl, 5 mM KCl, pH7.4) and
centrifuged at 21000g for 1 min. The pellets were resuspended
in ice-cold buffer at a concentration of 10 mg tissue/mL and
100 μL aliquots were rapidly distributed to tubes with 0.5 mL rof
oom temperature buffer containing 200 pM compound A. At 2,
4, 6, 8, 10, 12, and 15 min following membrane addition,
incubations were terminated by filtration of three tubes over
glass fiber filters. A parallel set of incubations performed in the
presence of 1 μM of an unlabeled, potent DORA (OX2R Ki =
1.0 nM) was used to determine nonspecific radioligand binding
at each time point. Radioactivity on the filters was determined
by liquid scintillation counting and compound A rates of
association were determined by linear regression. Receptor
occupancy in a drug treated animal is calculated as: % occu-
pancy = (1 - (slopedrug/slopevehicle)) ꢀ 100. The concentrations
of drug required to achieve 90% receptor occupancy were
derived by nonlinear curve fitting using Prism software.
The hydrochloride salt of compound 10 (458 mg) was dis-
solved in 70.2 mL of a 20% aqueous solution of TPGS (obtained
from Eastman Chemical Company) and administered by oral
gavage at 10 mpk of the free-base equivalent to four rats, 5 h into
their active period (09:00 or ZT 17:00). For 3, the free-base
(1.27 g) was suspended in 70.2 mL of a 20% aqueous solution of
TPGS and dosed as above. Recordings were started just prior to
compound administration and were collected for 23 h. The
experiments were based on a standard crossover design with
two animals receiving compound for one week and the com-
plementary group receiving vehicle, followed by a week of
reversed administration. All animals were exposed to two days
administration of orally gavaged vehicle prior to initiation of
experimental drug administration to allow for habituation. For
baseline sleep measurements, continuous recordings were col-
lected for two days to get average sleep behaviors for each
animal over contiguous days prior to drug and vehicle admin-
istration. During the drug administration studies, recordings
were collected each day prior to, during, and following drug
administration. Recordings were begun prior to compound
administration so that the exact time of administration was
recorded within the raw data file as artifactual noise which was
caused by removing the implanted transmitter from the recep-
tive field of the receiver during administration. This information
allowed a direct measure of drug/vehicle administration time
during offline analysis and was not included in the data analysis.
Following the completion of data collection, all data were
scored with automated sleep stage analysis software, Somno-
logica (Embla, Broomfield, CO). Assignment of sleep stages was
made in general accord with those described by J. M. Monti’s
group.45 Sleep/wake stages were assigned based upon a combi-
nation of level of movement within the field of the radio
frequency receiver over which individually housed rats were
caged, EMG activity, and ECoG frequencies over 10 s epochs.
Active wake was assigned to the epoch when movement of the
animal was detected over the receiver or when there was an
active EMG signal over the epoch and the ECoG frequencies
consisted of low-voltage high frequency activity. An epoch was
scored as light sleep when there was no movement activity, the
EMG was moderately activ,e and the ECoG consisted of either
theta or theta activity mixed with less than 50% of the epoch
showing delta activity. Delta sleep was scored when there was no
gross movement, reduced EMG activity, and the ECoG con-
sisted of more than 50% delta wave activity (i.e., 0.5 to 4 Hz).
Rapid eye movement (REM) sleep was scored when there was
no movement or EMG activity and the ECoG consisted of
primarily theta activity. Results of staging were grouped into
30 min periods following drug administration and the number of
entries into each stage and the duration of minutes spent in
each stage were calculated. The results for all four animals were
averaged by treatment, or vehicle, over seven administration
B. Chemistry. All solvents used were commercially available
“anhydrous” grade, and reagents were utilized without purifica-
tion unless otherwise noted. Nonaqueous reactions were carried
out in oven- or heat gun-dried glassware under a nitrogen
atmosphere. Magnetic stirring was used to agitate the reactions
and they were monitored for completion by either TLC (silica
gel 60, Merck KGaA) or LC/MS. A SmithCreator microwave

Article
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5329
from Personal Chemistry was used for microwave heating, and a
CombiFlash system utilizing RediSep cartridges by Teledyne
Isco was utilized for silica gel chromatography with fraction
collection at 254 nm. Reverse phase HPLC purification was
carried out on a Waters HPLC (XBridge Prep C-18 5 μM 19 mm ꢀ
150 mm column) utilizing a gradient of 0.1% TFA in water/
acetonitrile with sample collection triggered by photodiode
array detection. The reported yields are for isolated compounds
of g95% purity, unless otherwise noted, and were not exten-
sively optimized. Purity analysis was carried out by HPLC with
a Waters 2690 separations module equipped with a YMC Pro
50 mm ꢀ 3 mm I.D. C-18 column interfaced with a Waters
Micromass ZMD spectrometer utilizing a gradient of 0.05%
TFA in water/acetonitrile with UV detection at 215 and 254 nm.
1H and 13C NMR spectra were recorded on a Varian INOVA
400, 500, or 600 MHz spectrometer, and all chemical shifts
are referenced to an internal standard of tetramethylsilane
or the CDCl3 solvent peak, respectively. High resolving power
accurate mass measurement electrospray (ES) and atmo-
spheric pressure chemical ionization (APCI) mass spectral data
were acquired by use of a Bruker Daltonics 7T Fourier trans-
form ion cyclotron resonance mass spectrometer (FT-ICR MS).
Optical rotations were determined at 23 ꢀC on a Perkin-Elmer
241 spectrometer in the solvents and at the concentrations
specified.
DMSO-d6 at 25 ꢀC. See the Supporting Information for repro-
ductions of the 1H and 13C NMR spectra at 25 ꢀC in CDCl3.
Rotamer 1: (48% of total) 1H NMR (599.458 MHz, DMSO-
d6, 25 ꢀC) δ 7.90 (2H), 7.82 (1H), 7.44 (2H), 7.35 (1H), 7.32 (1H),
7.04 (1H), 4.23 (1H), 4.06 (1H), 3.98 (1H), 3.77 (1H), 3.62 (1H),
3.53 (1H), 3.32 (1H), 2.41 (3H), 2.04 (1H), 1.64 (1H), 1.15 (3H).
13C NMR (150.742 MHz, DMSO-d6, 25 ꢀC) δ 167.9, 162.9,
147.4, 147.1, 144.9, 137.9, 135.7, 133.4, 130.3, 128.0, 127.7,
122.1, 119.6, 114.9, 109.6 51.6, 46.3, 43.0, 39.8, 34.8, 20.2, 18.9.
Rotamer 2: (24% of total) 1H NMR (599.458 MHz, DMSO-
d6, 25 ꢀC). 8.04 (2H), 7.79 (1H), 7.42 (1H), 7.38 (1H), 7.31 (1H),
7.24 (1H), 7.02 (1H), 4.64 (1H), 4.25 (1H), 4.00 (1H), 3.94 (1H),
3.73 (1H), 3.58 (1H), 3.22 (1H), 2.34 (3H), 2.20 (1H), 1.77 (1H)
1.13 (3H). 13C NMR (150.742 MHz, DMSO-d6, 25 ꢀC) δ 168.3,
162.6, 147.3, 147.1, 144.7, 138.2, 135.8, 133.0, 130.2, 128.1,
127.9, 121.9, 119, 114.9, 109.6, 47.7, 46.5, 44.4, 38.4, 32.7,
20.1, 16.1.
tert-Butyl {2-[(3-Oxobutyl)amino]ethyl}carbamate, Carbo-
benzyloxy-Protected (5). To a solution of tert-butyl N-(2-ami-
noethyl)carbamate (39.4 mL, 250 mmol) in 500 mL of Et2O was
added methyl vinyl ketone (20.5 mL, 250 mmol) dropwise. After
stirring 18 h, the reaction was cooled to 0 ꢀC and triethylamine
(45.2 mL, 325 mmol) was added, followed by dropwise addition
of benzyl chloroformate (39.2 mL, 250 mmol). The reaction was
allowed to warm slowly to room temperature and then stirred
48 h before being partitioned between EtOAc and 10% aqueous
citric acid in a separatory funnel. After separation, the aqueous
layer was extracted again with EtOAc, the combined organic
layers were washed with saturated aqueous NaHCO3, then
brine, dried over Na2SO4, and concentrated by rotary evapora-
tion to provide 71.3 g of crude 5 as pale yellowish-brown oil
that was carried into the next step without further purification.
1-Benzyl 4-tert-Butyl 5-Methyl-1,4-diazepane-1,4-dicarboxy-
late (6). HCl gas was bubbled through a solution of 71.3 g crude
6 in 500 mL of EtOAc until the reaction was warm to the touch,
the flask was capped, and stirred for 2 h at room temperature.
The volatiles were removed by rotary evaporation, the residue
was dissolved in 1 M HCl and washed twice with Et2O. The
aqueous layer was carefully basified with NaOH and extracted
three times with 2:1 CHCl3/EtOH. The combined organic
phases were washed with a minimum amount of brine and
concentrated by rotary evaporation. The residue was dissolved
in CH2Cl2, filtered, and concentrated to provide 32.9 g of a
brown oil. This material was dissolved in 400 mL of CH2Cl2 and
treated with 10 mL of acetic acid. After stirring 2 h at room
temperature, sodium triacetoxyborohydride (39.6 g, 187 mmol)
was added and the reaction was stirred 48 h at room tempera-
ture. Most of the solvent was removed by rotary evaporation,
and the residue was dumped into saturated aqueous NaHCO3
and extracted three times with 2:1 CHCl3/EtOH. The combined
organic phases were washed with a minimum amount of brine
and concentrated by rotary evaporation. The residue was dis-
solved in CH2Cl2, filtered, and concentrated to provide 32.3 g of a
brown oil. This material was dissolved in 200 mL of CH2Cl2 and
treated with triethylamine (26.3 g, 260 mmol) and Boc2O (34.1 g,
156 mmol). After stirring overnight at room temperature, the
reaction was diluted with additional CH2Cl2 and washed sequen-
tially with 10% aqueous citric acid, saturated aqueous NaHCO3,
water, dried over Na2SO4, and concentrated by rotary evapora-
tion. The residue was purified by column chromatography with a
gradient of 0 to 30% EtOAc in hexanes to provide 33.1 g (38%
from methyl vinyl ketone) of 6 as a colorless oil. 1H NMR (500
MHz, CDCl3) δ 7.4-7.3 (m, 5H), 5.1 (s, 2H), 4.25 (m, 0.5H),
4.0-3.6 (m, 4H), 3.3-2.9 (m, 3.5H), 2.1 (m, 1H), 2.55 (s, 9H), 1.1
(d, J = 6.6 Hz, 3H) ppm. LRMS = 249.1 (M þ H - Boc).
1-Benzyl 4-tert-Butyl (5R)-5-Methyl-1,4-diazepane-1,4-dicar-
boxylate [(R)-6]. Resolution of the enatiomers of 6 was carried
out chromatographically using a Chiralpak AD 10 cm ꢀ 50 cm
column with 60% EtOH in hexanes (containing 0.1% diethyl-
amine as a modifier) at 175 mL/min. Resolution of 17.4 g of 6 in
As we have recently described, 1,4-disubstituted diazepane
orexin receptor antagonists exhibit unusual conformations in
solution and exist in multiple conformations as a result of
hindered rotations that are slow on the NMR time-scale.25a In
the case of compound 4, we attributed the four conformations to
cis-trans amide isomerization and hindered rotation about the
exocyclic C-N bond between the core and the quinazoline. In a
similar vein, the proton spectra of compounds 3 and 7-13
consist of broad, overlapping multiplets precluding a detailed
coupling constant analysis, and we feel that pictorial reproduc-
tions of the NMR spectra are more informative for comparative
purposes; thus, NMR resonances of final compounds will not be
listed in numerical format. Instead, we include reproductions of
the 1H and 13C NMR spectra at 25 ꢀC of all final compounds in
the Supporting Information, as well as the synthetic procedures
1
for preparation of 8-13. For compound 3 only, H and 13C
chemical shifts assigned from the analysis of COSY, HSQC, and
HMBC spectra will be presented herein.
[(7R)-4-(5-Chloro-1,3-benzoxazol-2-yl)-7-methyl-1,4-diaze-
pan-1-yl][5-methyl-2-(2H-1,2,3-triazol-2-yl)phenyl]methanone
(3). To a solution of 7 (29.5 g, 68.3 mmol) in 300 mL of EtOAc
and 200 mL of MeOH under nitrogen was added 2.4 g of 20%
Pd(OH)2 on carbon. The reaction was stirred for 48 h at room
temperature under a balloon of H2, then filtered through a pad
of celite with the aid of additional MeOH and concentrated to
provide 21.0 g (100%) of the secondary amine as a white foam.
This material was dissolved in 250 mL of DMF, and to the
solution was added triethylamine (29.3 mL, 210 mmol) and 2,5-
dichloro-1,3-benzoxazole46 (13.2 g, 70.1 mmol). After stirring
for 2 h at 75 ꢀC, the reaction was partitioned between EtOAc and
saturated aqueous NaHCO3, the layers were separated, and the
organic layer was washed with water, brine, dried over MgSO4,
filtered, and concentrated by rotary evaporation. The residue
was purified by column chromatography with a gradient of
0-100% EtOAc in hexanes to provide a colorless foam. This
material was stirred in a solution of 150 mL of EtOAc and 300
€
mL of hexanes overnight, filtered through a Buchner funnel, and
the solids were collected to provide 29.1 g (92% from 7) of 3 as a
white solid. HRMS calcd for C23H23ClN6O2 (M þ H): 451.1644;
found 451.1631. [R]D = -24.8 (c = 1.0, chloroform).
Compound 3 exists as a mixture of four rotameric compo-
nents in chemical exchange in DMSO-d6 at room temperature
(48:24:18:10). 1H and 13C chemical shifts of the two most
populated rotamers were assigned from the analysis of COSY,
HSQC, and HMBC spectra acquired on a sample dissolved in

5330 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Cox et al.
three equal injections provided 8.2 g (47%, 95% of theoretical)
of (R)-6 as the first eluting enatiomer. Analytical HPLC analysis
carried out on a 46 mm ꢀ 250 mm Chiralpak AD column with
the same eluent as above at a flow rate of 1 mL/min indicated
conventional pharmacological approaches in the treatment of sleep and
wake disorders. Curr. Top. Med. Chem. 2008, 8, 937–953.
(4) Wafford, K. A.; Ebert, B. Emerging anti-insomnia drugs: tackling
sleeplessness and the quality of wake time. Nature Rev. Drug
Discovery 2008, 7, 530–540.
(5) FDA Requests Label Change for All Sleep Disorder Drug Pro-
ducts. FDA News Release P07045, March 14, 2007; http://www.fda.
gov/NewsEvents/Newsroom/PressAnnouncements/2007/ucm108868.
htm.
1
that (R)-6 was of >98% ee. H NMR (500 MHz, CDCl3) δ
7.4-7.3 (m, 5H), 5.1 (s, 2H), 4.25 (m, 0.5H), 4.0-3.6 (m, 4H),
3.3-2.9 (m, 3.5H), 2.1 (m, 1H), 2.55 (s, 9H), 1.1 (d, J = 6.6 Hz,
3H) ppm. LRMS = 249.1 (M þ H - Boc). [R]D = -24.3 (c =
1.0, chloroform).
(6) Palomer, A.; Prinep, M.; Guglietta, A. Recent advances in the
treatment of insomnia. Annu. Rep. Med. Chem. 2007, 42, 63–80.
(7) (a) Borja, N. L.; Daniel, K. L. Ramelteon for the treatment of
insomnia. Clin. Ther. 2006, 28, 1540–1555. (b) Sateia, M. J.; Kirby-
Long, P.; Taylor, J. L. Efficacy and clinical safety of ramelteon: an
evidence-based review. Sleep Med. Rev. 2008, 12, 319–332.
(8) For a compilation of investigations carried out to identify the
orexin system and define its physiological roles, see: Hypocretins:
Integrators of Physiological Functions; de Lecea, L., Sutcliffe, J. G.,
Eds.; Springer: New York, 2005.
(9) (a) De Lecea, L.; Kilduff, T. S.; Peyron, C.; Gao, X.-B.; Foye, P. E.;
Danielson, P. E.; Fukuhara, C.; Battenberg, E. L. F.; Gautvik,
V. T.; Bartlett, F. S., II; Frankel, W. N.; Van Den Pol, A. N.;
Bloom, F. E.; Gautvik, K. M.; Sutcliffe, J. G. The hypocretins:
hypothalamus-specific peptides with neuroexcitatory activity.
Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 322–327. (b) Sakurai, T.;
Amemiya, A.; Ishii, M.; Matsuzaki, I.; Chemelli, R.; Tanaka, H.;
Williams, S. C.; Richardson, J. A.; Kozlowski, G. P.; Wilson, S.; Arch,
J. R. S.; Buckingham, R. E.; Haynes, A. C.; Carr, S. A.; Annan, R. S.;
McNulty, D. E.; Liu, W.; Terrett, J. A.; Elshourbagy, N. A.; Bergsma,
D. J.; Yanagisawa, M. Orexins and orexin receptors: A family of
hypothalamic neuropeptides and G-protein receptors that regulate
feeding behavior. Cell 1998, 92, 573–585.
2-(2H-1,2,3-Triazol-2-yl)-5-methylbenzoic Acid. A solution
of 2-iodo-5-methylbenzoic acid (4.0 g, 15.3 mmol) in DMF
(10mL) wastreatedwith1,2,3-triazole (2.1g, 30.5 mmol), Cs2CO3
(9.95 g, 30.5 mmol), CuI (0.145 g, 0.76 mmol), and trans-N,N0-
dimethylcyclohexane-1,2-diamine (0.43 g, 3.05 mmol). The mix-
ture was heated at 120 ꢀC for 10 min in a microwave reactor. The
reaction was cooled to room temperature, diluted with water,
and washed with EtOAc. The aqueous phase was acidified with
1N HCl and extracted with EtOAc. The organic layer was
dried over Na2SO4, filtered, and concentrated by rotary evapora-
tion. The residue was purified by silica gel chromatography
(0-10% MeOH in DCM with 0.1% AcOH) to provide the
desired regioisomer as a white solid (1.93 g, 62%) followed by
the undesired N1-isomer. Data for 2-(2H-1,2,3-triazol-2-yl)-
1
5-methylbenzoic acid: H NMR (500 MHz, DMSO-d6) δ 12.98
(bs, 1H), 8.04 (s, 2H), 7.72-7.45 (m, 3H), 2.41 (s, 3H) ppm.
HRMS-ESI m/z [M þ H]þ calcd for C10H10N3O2, 204.0768;
found, 204.0769.
Benzyl (5R)-5-Methyl-4-[5-methyl-2-(2H-1,2,3-triazol-2-yl)-
benzoyl]-1,4-diazepane-1-carboxylate (7). HCl gas was bubbled
through a solution of 27 g (77 mmol) of (R)-6 in 500 mL of
EtOAc until the reaction reached approximately 50 ꢀC, and the
flask was capped and stirred for 15 min at room temperature.
HCl gas was again bubbled through the solution, and the
flask was capped and stirred 30 min at room temperature. The
volatiles were removed by rotary evaporation to provide 22.3 g
of a pale-yellow gum. This material was dissolved in 300 mL of
DMF and treated with 2-(2H-1,2,3-triazol-2-yl)-5-methylben-
zoic acid (15.9 g, 78 mmol), EDC (22.5 g, 118 mmol), 1-hydroxy-
7-azabenzotriazole (12.8 g, 94 mmol), and N-methylmorpholine
(43.1 mL, 392 mmol). After stirring at room temperature over-
night, the reaction was partitioned between EtOAc and satu-
rated aqueous NaHCO3, the layers were separated, and the
organic layer was washed with water, brine, dried over MgSO4,
filtered, and concentrated by rotary evaporation. The residue
was purified by column chromatography with a gradient of
0-100% EtOAc in hexanes to provide 29.5 g [87% from (R)-6]
of 7 as a colorless gum. LRMS = 434.5 (M þ H). [R]D = -34.4
(c = 1.0, chloroform). See the Supporting Information for a
reproduction of the 1H NMR spectrum at 25 ꢀC in CDCl3.
(10) (a) Chemelli, R. M.; Willie, J. T.; Sinton, C. M.; Elmquist, J. K.;
Scammell, T.; Lee, C.; Richardson, J. A.; Williams, S. C.; Xiong,
Y.; Kisanuki, Y.; Fitch, T. E.; Nakazato, M.; Hammer, R. E.;
Saper, C. B.; Yanagisawa, M. Narcolepsy in orexin knockout mice:
molecular genetics of sleep regulation. Cell 1999, 98, 437–451.
(11) Willie, J. T.; Chemelli, R. M.; Sinton, C. M.; Tokita, S.; Williams,
S. C.; Kisanuki, Y. Y.; Marcus, J. N.; Lee, C.; Elmquist, J. K.;
Kohlmeier, K. A.; Leonard, C. S.; Richardson, J. A.; Hammer,
R. E.; Yanagisawa, M. Distinct narcolepsy syndromes in orexin
receptor 2 and orexin null mice: molecular genetic dissection of
non-REM and REM sleep regulatory processes. Neuron 2003, 38,
715–730.
(12) Lin, L.; Faraco, F.; Li, R.; Kadotani, H.; Rogers, W.; Lin, X.; Qiu,
X.; de Jong, P. J.; Nishino, S.; Mignot, E. The sleep disorder canine
narcolepsy is caused by a mutation in the hypocretin receptor 2
gene. Cell 1999, 98, 365–376.
(13) (a) Thannickal, T. C.; Moore, R. Y.; Nienhuis, R.; Ramanathan,
L.; Gulyani, S.; Aldrich, M.; Cornford, M.; Siegel, J. M. Reduced
number of hypocretin neurons in human narcolepsy. Neuron 2000,
27, 469–474. (b) Crocker, A.; Espana, R. A.; Papadopoulon, M.; Saper,
C. B.; Faraco, J.; Sakurai, T.; Honda, M.; Mignot, E.; Scammell, T. E.
Concomitant loss of dynorphin, Narp and orexin in narcolepsy. Neu-
rology 2005, 65, 1184–1188. (c) Blouin, A. M.; Thannickal, T. E.;
Worley, P. F.; Baraban, J. M.; Reti, I. M.; Siegel, J. M. Narp immuno-
staining of human hypocretin (orexin) neurons: loss in narcolepsy.
Neurology 2005, 65, 1189–1192.
(14) Akanmu, M. A.; Honda, K. Selective stimulation of orexin recep-
tor type 2 promotes wakefulness in freely behaving rats. Brain Res.
2005, 1048, 138–145.
Acknowledgment. We thank Dr. Chuck Ross and Joan
Murphy for HRMS analysis, Kim Michel for animal dosing,
and Kristi Hoffman, Ken Anderson, and Mary Beth Young
for pharmacokinetic analysis.
(15) (a) Selbach, O.; Eriksson, K. S.; Haas, H. L. Drugs to interfere with
orexins (hypocretins). Drug News Perspect. 2003, 16, 669–681.
(b) Baumann, C. R.; Bassetti, C. L. Hypocretins (orexins): clinical
impact of the discovery of a neurotransmitter. Sleep Med. Rev. 2005, 9,
253–268. (c) Boss, C.; Brisbare-Roch, C.; Jenck, F.; Aissaoui, H.;
Koberstein, R.; Sifferlen, T.; Weller, T. Orexin receptor antagonism: A
new principle in neuroscience. Chemia 2008, 62, 974–979. (d) Boss,
C.; Brisbare-Roch, C.; Jenck, F. Biomedical application of orexin/
hypocretin receptor ligands in neuroscience. J. Med. Chem. 2009,
52, 891–903.
Supporting Information Available: Experimental procedures
1
for compounds 8-13, as well as reproductions of the H and
13C NMR spectra for key compounds. This material is available
free of charge via the Internet at http://pubs.acs.org.
(16) (a) McAtee, L. C.; Sutton, S. W.; Rudolph, D. A.; Li, X.; Aluisio,
L. E.; Phoung, V. K.; Dvorak, C. A.; Lovenberg, T. W.;
Carruthers, N. I.; Jones, T. K. Novel substituted 4-phenyl-
[1,3]dioxanes: potent and selective orexin 2 (OX2R) antagonists.
Bioorg. Med. Chem. Lett. 2004, 14, 4225–4229. (b) Bingham, M. J.;
Cai, J.; Deehan, M. R. Eating, sleeping and rewarding: orexin receptors
and their antagonists. Curr. Opin. Drug Discovery Dev. 2006, 9, 551–
559. (c) Cai, J.; Cooke, F. E.; Sherborne, B. S. Antagonists of the orexin
receptors. Expert Opin. Ther. Patents 2006, 16, 631–646. (d) Roecker,
A. J.; Coleman, P. J. Orexin receptor antagonists: Medicinal chemistry
and therapeutic potential. Curr. Top. Med. Chem. 2008, 8, 977–987.
References
(1) Passarella, S.; Duong, M.-T. Diagnosis and treatment of insomnia.
Am. J. Health-Syst. Pharm. 2008, 65, 927–934.
(2) Daley, M.; Morin, C. M.; LeBlanc, M.; Gregoire, J. P.; Savard, J.;
ꢀ
Baillargeon, L. Insomnia and its relationship to health-care utili-
zation, work absenteeism, productivity and accidents. Sleep Med.
2009, 10, 427–438.
(3) (a) Sullivan, S. S.; Guilleminault, C. Emerging drugs for insomnia:
new frontiers for old and new targets. Expert Opin. Emerging Drugs
2009, 14, 411–422. (b) Renger, J. J. Overview of experimental and

Article
(e) Bergman, J. M.; Roecker, A. J.; Mercer, S. P.; Bednar, R. A.; Reiss,
Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14 5331
(27) The position of oxidation is not known. For simplicity in visualiza-
tion, oxidation at the 7-position of the core is shown.
D. R.; Ransom, R. W.; Harrell, C. M.; Pettibone, D. J.; Lemaire, W.;
Murphy, K. L.; Li, C.; Preuksaritanont, T.; Winrow, C. J.; Renger, J. J.;
Koblan, K. S.; Hartman, G. D.; Coleman, P. J. Proline bis-amides as
potent dual orexin receptor antagonists. Bioorg. Med. Chem. Lett.
2008, 18, 1425–1430. (f ) Cole, A. G.; Stroke, I. L.; Qin, L.-Q.; Hussain,
Z.; Simhadri, S.; Brescia, M.-R.; Waksmunski, F. S.; Strohl, B.; Tellew,
J. E.; Williams, J. P.; Saunders, J.; Appell, K. C.; Henderson, I.;
Webb, M. L. Synthesis of (3,4-dimethoxyphenoxy)alkylamino aceta-
mides as orexin-2 receptor antagonists. Bioorg. Med. Chem. Lett.
2008, 18, 5420–5423. (g) Aissaoui, H.; Koberstain, R.; Zumbrunn, C.;
Gatfield, J.; Brisbane-Roch, C.; Jenck, F.; Treiber, A.; Boss, C.
N-Glycine-sulfonamides as potent dual orexin 1/orexin 2 receptor
antagonists. Bioorg. Med. Chem. Lett. 2008, 18, 5729–5733.
(h) Sifferlen, T.; Boss, C.; Cottreel, E.; Koberstain, R.; Gude, M.;
Aissaoui, H.; Weller, T.; Gatfield, J.; Brisbane-Roch, C.; Jenck, F. Novel
pyrazolo-tetrahydropyridines as potent orexin receptor antagonists.
Bioorg. Med. Chem. Lett. 2010, 20, 1539–1542. (i) Coleman, P. J.;
Renger, J. J. Orexin receptor antagonists: a review of promising
compounds patented since 2006. Expert Opin. Ther. Patents 2010,
20, 307–324.
(28) A large number of the cores made during this effort are described
in Coleman, P. J.; Cox, C. D.; Roecker, A. J. Discovery of Dual
Orexin Receptor Antagonists (DORAs) for the Treatment of
Insomnia. Curr. Top. Med. Chem. in press.
(29) Dog pharmacokinetic “cassettes” where utilized wherein 5-6 test
compounds plus a standard were dosed IV in DMSO, with the goal
being to identify compounds with reduced IV clearance. Promising
analogues were then repeated as both IV and PO singles. In general,
the data obtained in cassette format was reproduced well in the IV
singles. Though some exceptions were noted, this strategy led to a
number of diazepanes with improved bioavailability.
(30) (a) Poppelsdorf, F.; Myerly, R. C. A novel synthesis of homo-
piperazine and its monomethyl derivatives. J. Org. Chem. 1961, 26,
131–134. (b) Benalil, A.; Guerin, A.; Carboni, B.; Vaultier, M. Synth-
esis of 2,3,6,7-tetrahydro- and 2,3,4,5,6,7-hexahydro-1H-1,4-diaze-
pines via a tandem Michael-type addition-intramolecular aza-Wittig
sequence. J. Chem. Soc., Perkin Trans 1 1993, 1061–1064. (c)
Subsequent to our work reported herein, an efficient synthesis of the
mono-protected 5-methyldiazepane core was reported, see: Wlodarczyk,
ꢀ
(17) Dugovic, C.; Shelton, J. E.; Aluisio, L. E.; Fraser, I. C.; Jiang, X.;
Sutton, S. W.; Bonaventure, P.; Yun, S.; Li, X.; Lord, B.; Dvorak,
C. A.; Carruthers, N. I.; Lovenberg, T. W. Blockade of orexin-1
receptors attenuates orexin-2 receptor antagonism-induced sleep
promotion in rat. J. Pharmacol. Exp. Ther. 2009, 330, 141–152.
N.; Gilleron, P.; Millet, R.; Houssin, R.; Henichart, J.-P. Synthesis of
1,4-diazepin-5-ones under microwave irradiation and their reduction
products. Tetrahedron Lett. 2007, 48, 2583–2586.
(31) Leonard, J.; Lygo, B.; Procter, G. Advanced Practical Organic
Chemistry; Blackie: New York, 1995; p 6.
(32) For instance, the (S)-antipode of compound 9 has OX1R Ki =
54 nM and OX2R Ki = 8.5 nM, corresponding to 8- and 18-fold
losses in potency, respectively, relative to the (R)-antipode. We
have confirmed the stereochemistry in the active series is (R) by
both X-ray crystallography using anomalous dispersion, as well as
an asymmetric synthesis starting from the chiral pool.
(33) In general, installation of the tolyl methyl group on the benzamide
increases potency on both receptors, but has a greater effect on
OX1R potency.
(34) We have found that a vehicle of 20% TPGS in water to be an
excellent vehicle for improving exposure of BCS class II molecules
such as the diazepanes described herein and is also highly compa-
tible with our sleep studies. For a study of the effects that TPGS
has on oral absorption, see: Varma, M. V. S.; Panchagnula, R.
Enhanced oral paclitaxel absorption with vitamin E-TPGS: effect
on solubility and permeability in vitro, in situ and in vivo. Eur. J.
Pharm. Sci. 2005, 25, 445–453.
(35) Brain, plasma, and CSF concentrations were measured following a
30 min continuous IV infusion of 10 at 0.25, 0.75, and 1.5 mg/kg in
25% hydroxypropyl-β-cyclodextrin.
(36) (a) Evans, D. C.; Watt, A. P.; Nicoll-Griffith, D. A.; Baillie, T. A.
Drug-protein adducts: an industry perspective on minimizing the
potential for drug bioactivation in drug discovery and develop-
ment. Chem. Res. Toxicol. 2004, 17, 3–16.(b) Kumar, S.; Baillie, T. A.
Minimizing metabolic activation in drug discovery in Handbook of
Drug Metabolism, 2nd ed.; Pearson, P. G., Wienkers, L. C., Eds.;
Informa Healthcare: New York, 2009; Chapter 23.
~
ꢀ
(18) Owen, R. T.; Castener, R.; Bolos, J.; Estivill, C. Almorexant. Drugs
Future 2009, 34, 5–10.
(19) Brisbare-Roch, C.; Dingemanse, J.; Koberstein, R.; Hoever, P.;
Aissaoui, H.; Flores, S.; Mueller, C.; Nayler, O.; van Gerven, J.; de
Haas, S. L.; Hess, P.; Qiu, C.; Buchmann, S.; Scherz, M.; Weller, T.;
Fischli, W.; Clozel, M.; Jenck, F. Promotion of sleep by targeting
the orexin system in rats, dogs and humans. Nature Med. 2007, 13,
150–155.
(20) New Data On Orexin Receptor Antagonist Almorexant Shows
Therapeutic Potential To Restore Normal Physiological Sleep In
Insomnia Patients. Medical News Today Sep 04, 2007; http://www.
medicalnewstoday.com/articles/81354.php.
(21) Following the observation of preclinical toxicity, GSK entered into
a deal with Actelion to jointly develop and market Almorexant, see:
http://www.bioworld.com/servlet/com.accumedia.web.Dispatcher?
next=bioWorldHeadlines_article&forceid=48183
(22) Merck 2008 Annual Business Briefing;Final. Goliath Business
News Dec 9, 2008; http://goliath.ecnext.com/coms2/gi_0199-9773483/
Merck-2008-Annual-Business-Briefing.html.
(23) A Long Term Safety Study of MK4305 in Patients With Primary
Insomnia.ClinicalTrials.gov 2009; http://clinicaltrials.gov/ct2/show/
NCT01021813
(24) Whitman, D. B.; Cox, C. D.; Breslin, M. J.; Brashear, K. M.;
Schreier, J. D.; Bogusky, M. J.; Bednar, R. A.; Lemaire, W.; Bruno,
J. G.; Hartman, G. D.; Reiss, D. R.; Harrell, C. M.; Kraus, R. L.;
Li, Y.; Garson, S. L.; Doran, S. M.; Prueksaritanont, T.; Li, C.;
Winrow, C. J.; Koblan, K. S.; Renger, J. J.; Coleman, P. J.
Discovery of a potent, CNS-penetrant orexin receptor antagonist
based on an N,N-disubstituted-1,4-diazepane scaffold that pro-
motes sleep in rats. ChemMedChem 2009, 4, 1069–1074.
(25) Further investigations by NMR spectroscopy, X-ray crystallogra-
phy and molecular modeling revealed that 4 exists both in the solid
state and solution in an unusual horseshoe- or U-shaped confor-
mation characterized by a π-stacking interaction between the
quinazoline and phenyl rings. (a) Cox, C. D.; McGaughey, G. B.;
Bogusky, M. J.; Whitman, D. B.; Ball, R. G.; Winrow, C. J.;
Renger, J. J.; Coleman, P. J. Conformational analysis of N,N-
disubstituted-1,4-diazepane orexin receptor antagonists and im-
plications for receptor binding. Bioorg. Med. Chem. Lett. 2009, 19,
2997–3001. We recently reported that bridged diazepane analogues,
whose design and synthesis was inspired by the low energy conforma-
tion of 4, are also potent DORAs that promote sleep in rats, see: (b)
Coleman, P. J.; Schreier, J. D.; McGaughey, G. B.; Bogusky, M. J.; Cox,
C. D.; Hartman, G. D.; Ball, R. G.; Harrell, C. M.; Reiss, D. R.;
Prueksaritanont, T.; Winrow, C. J.; Renger, J. J. Design and synthesis of
conformationally constrained N,N-disubstituted-1,4-diazepanes as po-
tent orexin receptor antagonists. Bioorg. Med. Chem. Lett. 2010, 20,
2311–2315. (c) Coleman, P. J.; Schreier, J. D.; Roecker, A. J.; Mercer,
S. P.; McGaughey, G. B.; Cox, C. D ; Hartman, G. D.; Harrell, C. M.;
Reiss, D. R.; Garson, S. L.; Anderson, W. B.; Prueksaritanont, T.;
Winrow, C. J.; Renger, J. J. Discovery of 3,9-diazabicyclo-
[4.2.1]nonanes as potent orexin receptor antagonists with sleep-promo-
ting activity in the rat. Bioorg. Med. Chem. Lett. 2010, DOI: 10.1016/j.
bmcl.2010.05.047
(37) See the Experimental Section for details.
(38) GSH-adducts were identified that had a mass spectral signal
corresponding to loss of the fluorine atom, suggesting the fluoro-
quinazoline as a site of bioactivation. The GSH-adduct/IS ratio of
compound 4 is 5.0, suggesting that the nonfluorinated quinazoline
also likely undergoes bioactivation and subsequent trapping with
GSH.
(39) As with diazepane DORA 4, solution NMR studies of 3 indicate a
very complex situation with several low energy conformations in
equilibrium. Studies are underway to characterize the low energy
conformations of 3 and how they fit with our proposed bioactive
conformation model.25
(40) Brain, plasma, and CSF concentrations were measured following a
30 min continuous IV infusion of 3 at 0.25, 0.75, and 1.5, and
2.0 mg/kg in 25% hydroxypropyl-β-cyclodextrin. Brain/plasma
ratios ranged from 0.6-1.2 and the CSF/plasma ratios ranged
from 0.007-0.017. When dosed at 10 mg/kg PO in 20% TPGS, at
1 h post dose the brain/plasma ratio was 0.7 and the CSF/plasma
ratio was 0.007.
(41) A manuscript in preparation will describe the pharmacology of 3 in
more detail, including analysis of its dose-dependent effects on
sleep in rats, dogs, and monkeys. Winrow, C. J. unpublished
results.
(42) The identities of 14-16 were confirmed by independent synthesis.
(43) Kunapuli, P.; Ransom, R.; Murphy, K. L.; Pettibone, D. J.; Kerby,
J.; Grimwood, S.; Zuck, P.; Hodder, P.; Lacson, R.; Hoffman, I.;
Inglese, J.; Strulovici, B. Development of an intact cell reporter
gene β-lactamase assay for G-protein coupled receptors for high-
throughput screening. Anal. Biochem. 2003, 314, 16–29.
(26) Other vehicles studied included 0.5% aqueous methylcellulose,
10% Tween 80 in water, and 0.1 M citric acid.

5332 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 14
Cox et al.
(44) Mosser, S. D.; Stanley, L.; Gaul, S. L.; Bednar, B.; Koblan, K. S.;
Bednar, R. A. Automation of in vitro dose-inhibition assays
using the Tecan Genesis and an integrated software package to
support the drug discovery process. J. Assoc. Lab. Autom. 2003, 8,
54–62.
(45) (a) Alvarino, F.; Monti, J. M.; Jantos, H.; Monti, D. Effect of SX-
3228, a selective ligand for the BZ1 receptor, on sleep and waking
during the light-dark cycle in the rat. Braz. J. Med. Biol. Res. 1999,
32, 1007–1014. (b) Monti, J. M.; Jantos, H.; Monti, D. Increase in
waking and reduction of NREM and REM sleep after nitric oxide
synthase inhibition: prevention with GABAA or adenosine A1 receptor
agonists. Behav. Brain Res. 2001, 123, 23–35.
(46) Lazer, E. S.; Miao, C. K.; Wong, H.-C.; Sorcek, R.; Spero, D. M.;
Gilman, A.; Pal, K.; Behnke, M.; Graham, A. G.; Watrous, J. M.;
Homon, C. A.; Nagel, J.; Shah, A.; Guindon, Y.; Farina, P. R.;
Adams, J. Benzoxazolamines and benzothiazolamines: Potent,
enantioselective inhibitors of leukotriene biosynthesis with a novel
mechanism of action. J. Med. Chem. 1994, 37, 913–923.