Discovery of 3,9-diazabicyclo[4.2.1]nonanes as potent dual orexin receptor antagonists with sleep-promoting activity in the rat

Article abstract of DOI:10.1016/j.bmcl.2010.05.047

Orexins are excitatory neuropeptides that regulate arousal and sleep. Orexin receptor antagonists promote sleep and offer potential as a new therapy for the treatment of insomnia. In this Letter, we describe the synthesis of constrained diazepanes having

Full text of DOI:10.1016/j.bmcl.2010.05.047

Bioorganic & Medicinal Chemistry Letters 20 (2010) 4201–4205
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of 3,9-diazabicyclo[4.2.1]nonanes as potent dual orexin receptor
antagonists with sleep-promoting activity in the rat
a
a
a
b
Paul J. Coleman a, , John D. Schreier , Anthony J. Roecker , Swati P. Mercer , Georgia B. McGaughey ,
Christopher D. Cox a, George D. Hartman a, C. Meacham Harrell c, Duane R. Reiss c, Scott M. Doran c,
Susan L. Garson c, Wayne B. Anderson d, Cuyue Tang d, Thomayant Prueksaritanont d,
Christopher J. Winrow c, John J. Renger c
*
a Department of Medicinal Chemistry, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, United States
b Chemistry Modeling and Informatics, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, United States
c Department of Depression and Circadian Disorders, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, United States
d Department of Drug Metabolism, Merck Research Laboratories, PO Box 4, Sumneytown Pike, West Point, PA 19486, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Orexins are excitatory neuropeptides that regulate arousal and sleep. Orexin receptor antagonists pro-
mote sleep and offer potential as a new therapy for the treatment of insomnia. In this Letter, we describe
the synthesis of constrained diazepanes having a 3,9 diazabicyclo[4.2.1]nonane bicyclic core with good
oral bioavailability and sleep-promoting activity in a rat EEG model.
Received 16 April 2010
Revised 10 May 2010
Accepted 12 May 2010
Available online 25 May 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
Sleep
Insomnia
Diazepane
GPCR antagonist
Brain penetration
Orexin
Receptor
Orexins are excitatory neuropeptides that bind to two G-protein
coupled receptors (GPCRs), orexin-1 (OX1R) and orexin 2 (OX2R).1
Neurons that secrete orexins project from the lateral hypothala-
mus and innervate diverse regions of the central nervous system
that govern sleep, behavioral arousal, and stability of the wake
state. Based on genetic studies and a clear link to the pathophysi-
ology of narcolepsy, orexin antagonists have been explored as a po-
tential novel treatment for sleep disorders.2 Indeed, a dual orexin
antagonist has recently been reported to be sleep-promoting in ro-
dents, dogs, and in human subjects.3
In previous publications from this laboratory, we described the
discovery of a novel series of diazepane dual orexin receptor antag-
onists including 2-{4-[5-methyl-2-(2H-1,2,3-triazolyl-2-yl)ben-
zoyl]-1,4-diazepan-1-yl}quinazoline, 1.4 Compound 1 (Fig. 1) has
high affinity for both orexin receptors but has low bioavailability
(<10%) following oral administration to rats. Spectroscopic, X-ray,
and molecular modelling studies indicate that N,N-1,4-diazepane
antagonists such as 1 adopt a low energy conformation that is char-
boat conformation.5 Based on this knowledge we designed con-
strained diazepanes such as 2 that enforce the aryl–aryl interaction
present in compound 1. These bridged diazepanes are constrained
by inclusion of a one-carbon bridge.6 Despite maintaining excellent
potency against OX1R and OX2R, these constrained analogs also
had poor preclinical pharmacokinetics with high plasma clearance
and low bioavailability. In this Letter, we describe the synthesis of
constrained diazepanes with two-carbon bridges. These efforts
were rewarded by the discovery of an orexin antagonist with a
3,9-diazabicyclo[4.2.1]nonane core 8a that has improved oral bio-
availability and sleep-promoting activity in a rat EEG model.
We sequentially varied the linkage of a two-carbon bridge on the
seven-membered diazepane ring. We synthesized these analogs by
reaction sequences detailed in Schemes 1 and 2. In preparing these
orexin antagonists, we maintained the 6-fluoroquinazoline and
5-methyl-2-triazolylbenzamide as preferred, potency-enhancing
features. The 3,9-diazabicyclo[4.2.1]nonane core was synthesized
as shown in Scheme 1. Schmidt reaction with the N-benzyl-3-nor-
tropinone 3 afforded the bicyclic lactam 4 in excellent yield. Lactam
reduction followed by manipulation of amine protecting groups
provided bicyclic diamine 5. Amide formation, BOC-removal, and
nucleophilic aromatic substitution reaction provided receptor
acterized by an intramolecular p-stacking interaction and a twist-
* Corresponding author. Tel.: +1 215 652 4618.
E-mail address: paul_coleman@merck.com (P.J. Coleman).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.05.047

4202
P. J. Coleman et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4201–4205
N
F
N
N
N
N
O
N
N
N
O
N
N
N
N
N
N
Cl
Me
2
OX1R Ki =0.4 nM
OX2R Ki = 0.6 nM
1 OX1R Ki = 1.2 nM
OX2R Ki = 0.6 nM
X-Ray Overlap of 1 and 2.
Figure 1. Structure and X-ray superimposition of orexin receptor antagonists 1 and 2.
antagonist 8. Reordering the heteroarylation and amide formation
steps afforded the isomeric diazabicyclic[4.2.1]nonane derivative
9. Compound rac-8 could be resolved by HPLC on a chiral stationary
support to afford enantiomers 8a and 8b.7
pounds 8a, 9, and 16a (Table 1). Nearly all of these bridged
analogs maintained balanced potency on OX1R and OX2R. Interest-
ingly, constrained diazepane 14b represents an exception and
gained selectivity for OX1R over OX2R. Although selected diaze-
panes with two-carbon bridges tended to retain potency in the
OX1R and OX2R binding assays, FLIPR potencies were more highly
shifted than unconstrained diazepanes (FLIPR IC50/binding
Ki = 160–340Â for 8a vs 24–45Â shift for 1). We attributed this
shift to greater lipophilicity and higher nonspecific binding in the
FLIPR assay. The measured Log P for 8a is higher than for 1 (>3.4
vs 2.9, respectively) while polar surface area was comparable
(71 Å2 for 8a and 74 Å2 for 1). Passive cellular permeability re-
mained high for these analogs (Papp >25 Â 10À6 cm/s; LLC-PK1
cells).
For compounds shown in Scheme 2, we began the synthesis with
a catalytic, asymmetric aza-Diels–Alder reaction to produce 11 in
high yield and enantioselectivity.8 Bicyclic piperidone 11 was then
treated sodium azide under Schmidt conditions to provide a sepa-
rable mixture of the two bicyclic lactams 12 and 13. Lactam 12
was reduced with lithium aluminum hydride and the protected dia-
zepane treated with 2-chloro-6-fluoroquinazoline. Removal of the
PMP group with ceric ammonium nitrate (CAN) followed by amide
coupling with 6 afforded 14a. Reordering the heteroarylation and
amide coupling reactions provides the bridged compound 15a.
The sequence shown in Scheme 2 was then repeated using
as an amino acid catalyst and provided stereoisomers 14b and 15b.
The (1R,5R)-2,6 diazabicyclo[3.2.2]nonane 16a was derived from a
D
-proline
We were unable to generate diffraction quality crystals from
any of the 2-carbon bridged compounds and 1H NMR spectra for
these molecules were highly complex with multiple conformers
in equilibrium.10 However, modeling studies with 8a provided a
minimized structure that has a favorable aryl–aryl interaction
and superimposes on the X-ray of compound 1.11 These results
D
-proline catalyzed Diels–Alder reaction affording ent-11 (scheme
not shown). Subsequent transformations afforded 16a.9 This se-
quence was repeated with -proline as the asymmetric catalyst
L
and provided (1S,5S)-2,6-diazabicyclo[3.2.2]nonane 16b.
Constraining the diazepanes by installation of various two-car-
bon bridges provided several potent compounds including com-
agree with earlier studies that suggest enforcing the p-stacking ar-
ray in the diazepane series leads to a highly favorable binding
interaction with the orexin receptors.5,6
Ph
N
Ph
N
º
NaN3, H2SO4, 5 C
HN
4
95 %
O
3
O
1. LiAlH4, THF,98%
2. BOC2O, DCM, 99%
3. Pd(OH)2, H2(g) 45 PSI
98%
N
N
PyBOP, Et3N,
DMF, 96%
O
N
NH
N
N
BOC
N
BOC
N
N
N
5
CO2H
7
Me
6
Me
1. HCl , dioxane
2. 2-chloro-6-fluoro
quinazoline,
N
N
O
N
N
N
Et3N, DMF
8
F
N
N
3. Chiral HPLC,
22% overall
Me
Scheme 1. Synthesis of 3,9-diazabicyclo[4.2.1]nonanes as orexin receptor antagonists.

P. J. Coleman et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4201–4205
4203
O
O
MeO
NH2
11
N
CH2O, L-proline, DMSO
10
OMe
NaN3
H2SO4
HN
HN
N
OMe
N
OMe
O
O
12 (19%)
13 (27%)
1. LiAlH4 , 93%
1. LiAlH4 , 98%
2. EDC, HOBt, acid 6
2. 2-chloro-6-fluoroquinazoline, 30%
3. CAN, MeCN/H2O, 0 C, then
EDC, HOBt, acid 6
3. CAN, MeCN/H2O, 0 C, then
2-chloro-6-fluoroquinazoline
(27% over 2 steps)
(17% over 2 steps)
N
N
N
N
O
N
O
N
N
N
N
N
F
F
N
N
N
N
Me
Me
15a (1S, 5S)
15b (1R, 5R)
14a (1R, 5S)
14b (1S, 5R)
Scheme 2. Synthesis of 3,6-diazabicyclo[3.2.2]nonanes and 2,6-diazabicyclo[3.2.2]nonanes as orexin receptor antagonists.
The most promising compounds identified from these studies
were further characterized in terms of the pharmacokinetic prop-
erties, brain penetration, and in vivo evaluation. In vitro micro-
somal studies of related diazepane analogs suggested that human
and dog hepatic stability would be similar (and dissimilar from
rat). Therefore we emphasized low plasma clearance in the dog
as a criterion for selecting compounds for additional studies.
The plasma clearance of 8a was low in the dog following intra-
venous administration (Cl = 3.6 mL/min/kg) and more favorable
than reference compound 1 (Cl = 12 mL/min/kg) or other bridged
compounds 9 (Cl = 11 mL/min/kg) and 16 (Cl = 14 mL/min/kg). In-
deed when dosed orally to dogs, compound 8a had excellent bio-
bound plasma concentration of 0.4% in the rat. Based on these pro-
spective rat PK experiments, a dose of 60 mpk was chosen for
in vivo evaluation providing plasma and CSF exposures of
14.6 lM and 0.17 lM at 1 h after oral administration. At this dose
the CSF concentration of compound 8a is similar to the FLIPR IC50
concentrations in the OX1R and OX2R assays.
Based on favorable rat brain penetration and oral bioavailability
in both rats and dogs, 8a was evaluated in a rat sleep model15 using
telemetric recording of electricorticogram (ECoG) and electromyo-
gram (EMG) signals to assess time spent in various states of arou-
sal, including active wake, light sleep REM sleep, and delta or slow
wave sleep (SWS). Compound 8a was dosed orally at 60 mg/kg/day
to telemetry-implanted rats in their active phase (lights off) and
the amount of time spent in each wake/sleep state was measured
and binned into 30 min intervals and compared to vehicle treated
rats. In this assay, compound 8a decreased wakefulness in the trea-
ted rats for greater than 4 h and sustained increases in slow wave
and REM sleep for 3.5 and 6 h, respectively. A lower dose (30 mpk)
of 8a was not efficacious in promoting rat sleep.16
availability (F = 100%), with a Cmax of 3.4 lM (Table 2).
Steady state plasma, brain, and cerebrospinal fluid (CSF) con-
centrations of compound 8a were assessed in anesthetized Spra-
gue-Dawley rats.12 Compound 8a was administered to rats by
intravenous infusion in HPCD at 2 mg/kg, the animals were sacri-
ficed at 30 min; brain, plasma, and CSF levels were measured. Plas-
ma and brain levels reached 1455 nM and 584 nM, respectively
(brain/plasma = 0.40). At this dose, CSF levels of 8a were below
the limits of quantitation, consistent with the very high levels of
plasma protein binding measured in the rat (99.6% bound). Com-
pound 8a was not a substrate of the P-glycoprotein transporter.13
CSF concentrations of 8a were quantified as a surrogate mea-
sure of receptor-available free compound in the brain.14 In order
to select an appropriate dose for in vivo pharmacodynamic evalu-
ation in the rat, we explored a range of oral doses in the rat and
directly measured rat plasma and CSF levels at single time points
(1–2 h; Table 3). After oral administration of 8a, plasma exposures
in the rat increased nonlinearly from 10 to 60 mg/kg while the CSF/
plasma ratio ranged from 0.004 to 0.011 consistent with an un-
In conclusion, we have continued to drive the design of our 1,4-
diazepane orexin receptor antagonists based on an understanding
of the solution and solid state conformations of these systems. Ear-
lier efforts to enforce an energetically favorable p-stacking interac-
tion with one-carbon bridged diazepanes provided analogs with
comparable potency to 1 but lacking acceptable pharmacokinetics.
The current study demonstrates that the introduction of a two-
carbon constraint can significantly improve pharmacokinetics in
both rats and dogs while maintaining excellent dual orexin recep-
tor potency. 3,9-Diazabicyclo[4.2.1]nonane 8a promotes sleep in
EEG-telemetered rats at 60 mg/kg when dosed orally; dose selec-
tion for this experiment was informed by direct measurements of

4204
P. J. Coleman et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4201–4205
Table 1
2-Carbon bridged diazepane orexin antagonists
F
N
N
O
N
N
R1
N
Me
Compound
R1
OX1R (Ki, nM)
OX2R (Ki, nM)
FLIPR
OX1R (IC50, nM)
OX2R (IC50, nM)
N
N
8a (ent A)
8b (ent B)
0.6
3.7
0.4
6.7
98
99
136
150
N
N
Het
Het
9 (rac)
0.9
0.8
72
175
N
N
N
14a (1R,5S)
14b (1S,5R)
10
2.0
2.9
62
104
46
370
790
N
N
N
Het
Het
Het
15a (1S,5S)
15b (1R,5R)
160
345
105
590
2300
1600
1450
>10,000
16a (1R,5R)
16b (1S,5S)
0.6
7.6
2.2
14
125
800
140
740
Table 2
Dog and rat pharmacokinetics for bridged compound 8a versus 1
Compd
Species
%F
Cl (mL/min/kg)
T1/2 (h)
AUC0–24
Cmax
1
Doga
Ratb
16
2
12
53
1.3
0.3
0.44
0.19
l
l
M-h
M-h
0.23
0.30
l
l
M
M
8a
Dog
Rat
100
14
3.6
23
1.8
1.0
14.11
1.57
l
M-h
3.4 lM
0.57 lM
lM-h
a
Dogs dosed orally at 3 mpk po in PEG200 and at 0.125 mpk iv in DMSO.
Rats dosed orally at 10 mpk, po in PEG200 and dosed iv at 2 mpk in DMSO.
b
nants of in vivo activity. Further studies to reduce plasma protein
binding, enhance unbound brain concentrations, and refine phar-
macokinetic-pharmacodynamic relationships for orexin receptor
antagonists are ongoing.
Table 3
Rat plasma and CSF concentrations for 8a
Dose
Plasma concn at 2 h (
l
M) CSF concn at 2 h (
l
M) CSF/plasma
10 mpk
30 mpk
60 mpk
1.1
0.003
0.023
0.170
ND
0.004
0.007
0.011
ND
3.1
14.6a
A. Supplementary data
100 mpk 12.5b
Rats dosed orally in VitE-TPGS at doses specified.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2010.05.047.
a
Plasma/CSF levels measured at 1 h.
b
Dosed in PEG 400.
References and notes
CSF levels of 8a after oral administration. Despite an improved PK
profile for 8a, the introduction of the two-carbon constraint is
associated with an increase in lipophilicity and corresponding de-
creases in unbound plasma and CSF concentrations. The high plas-
ma exposures (14 lM at 1 h) required for 8a to achieve efficacy in a
rat sleep model suggest that achieving appropriate CSF/plasma
1. (a) de Lecea, L.; Kilduff, T. S.; Peyron, C.; Gao, X.; Foye, P. E.; Danielson, P. E.;
Fukuhara, C.; Battenberg, E. L.; Gautvik, V. T.; Bartlett, F. S.; Frankel, W. N.; van
den Pol, A. N.; Bloom, F. E.; Gautvik, K. M.; Sutcliffe, J. G. Proc. Natl. Acad. Sci.
1998, 95, 322; (b) Sakurai, T.; Amemiya, A.; Ishii, M.; Matsuzaki, I.; Chemelli, R.
M.; 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.-S.; Terrett, J. A.; Elshourbagy, N. A.; Bergsma, D. J.; Yanagisawa, M.
Cell 1998, 92, 573.
concentrations of orexin antagonists may be important determi-

P. J. Coleman et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4201–4205
2. (a) Boss, C.; Brisbare-Roch, C.; Jenck, F. J. Med. Chem. 2009, 52, 891; (b) Roecker,
4205
A. J.; Coleman, P. J. Curr. Top. Med. Chem. 2008, 8, 977.
3. Brisbare-Roch, C.; Dingemanse, J.; Koberstein, R.; Hoever, P.; Aissaoui, H.; Flores,
S.; Mueller, C.; Nayler, O.; van Gerven, J.; de Haas, S.; Hess, P.; Qiu, C.; Buchmann,
S.; Scherz, M.; Weller, T.; Fischli, W.; Clozel, M.; Jenck, F. Nat. Med. 2007, 13, 150.
4. 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. ChemMedChem
2009, 4, 1069.
5. Cox, C. D.; McGaughey, G. B.; Bogusky, M. J.; Whitman, D. B.; Ball, R. G.;
Winrow, C. J.; Renger, J. J.; Coleman, P. J. Bioorg. Med. Chem. Lett. 2009, 19, 2997.
6. 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. Bioorg. Med. Chem. Lett. 2010, 20, 2311.
7. Racemic mixture 8 was resolved by chromatography using a Chiracel AD
column eluting with 40% hexanes/ 60% isopropanol with 0.1% diethylamine.
Enantiomer 8a eluted first and 8b second.
8. Sunden, H.; Ibrahem, I.; Eriksson, L.; Cordova, A. Angew. Chem., Int. Ed. 2005, 44,
4877.
9. The 2,6-diazabicyclo[3.2.2]nonane derivatives were synthesized from ent-11
by (1) Schmidt reaction (30%); (2) LiAlH4 reduction, heteroarylation with 2-
chloro-6-fluoroquinazoline (60% for two steps); (3) CAN deprotection followed
by amide coupling with acid 6 (41% over two steps). The (1R,5S) isomer was
12. Seven to 10 weeks old Sprague-Dawley rats were anesthetized with isoflurane
(5% with oxygen at 2 l/min). Once anesthetized, the rat was shaved from the
ears to the shoulders and then placed on a circulating water warming blanket
set at 37 °C. Anesthesia was maintained using 2% isoflurane by means of a nose
cone. The tail vein was cannulated with
a 25G winged infusion needle
connected via tubing to a syringe containing the test compound. The syringe
derived from
derived from same reaction using
L
-proline catalyzed Diels–Alder and the (1S,5S) antipode was
-proline.
D
N
N
O
N
Steps
N
N
F
N
MeO
N
O
N
Me
ent-11
was placed on a Harvard Apparatus infusion pump set to deliver 2 mg/kg over
30 min. At the end of the infusion, CSF was collected from the cisterna magna
by needle puncture with slow aspiration using a 25G butterfly connected by
tubing to a 1 cc syringe. The CSF sample was dispensed into a vial and
immediately frozen on dry ice. A 1 ml blood sample was then obtained via
cardiac puncture, centrifuged, plasma collected and frozen on dry ice. The brain
was rapidly harvested and frozen on dry ice. CSF, plasma, and brain samples
were then analyzed to determine compound levels.
16a (1R, 5R)
16b (1S, 5S)
10. The absolute stereochemistry of 8a has not established by X-ray
crystallography. The depiction of 8a as R,R is based on analogy to the X-ray
of compound 2 (see Ref.6).
11. Conformational searching using the mixed torsion/low-mode sampling
algorithm as implemented in Schrödinger, v9.0 (Schrödinger, LLC, v9.0) was
13. The MDR1 B-A/A-B ratio for 8b = 0.7 with an apparent permeability of
29 Â 10À6 cm/s in LLC-PK1 cells. For a description of this method see: (a)
Hochman, J. H.; Yamazaki, M.; Ohe, T.; Lin, J. H. Curr. Drug Metab. 2002, 3, 257;
(b) Lin, J. H.; Yamazaki, M. Drug Metab. Rev. 2003, 35, 417.
14. Liu, X.; Smith, B. J.; Chen, C.; Callegari, E.; Becker, S. L.; Chen, X.; Cianfrogna, J.;
Doran, A. C.; Doran, S. D.; Gibs, J. P.; Hosea, N.; Liu, J.; Nelson, F. R.; Szewc, M. A.;
Deusen, J. V. Drug Metab. Dispos. 2006, 34, 1443.
performed using the OPLS force field in
conformers within 5 kcal/mol of the global minimum were saved and
visually inspected for evidence of -stacking. See: Kolossváry, I.; Guida,W. C.
a constant dielectric of 1. All
p
15. See Supplementary data for a description of the rat EEG assay.
16. Compound 8a had no effect on rat sleep when dosed at 30 mpk ip in PEG 200.
J. Am. Chem. Soc. 1996, 118, 5011. Minimization of orexin receptor antagonists
8a and superimposition with X-ray of 1 are shown below.