Azepinone as a conformational constraint in the design of κ-opioid receptor agonists

Article abstract of DOI:10.1016/j.bmcl.2004.08.041

A new class of κ-opioid receptor agonists is described. The design of these agents was based upon energy minimization and structural overlay studies of the generic azepin-2-one structure 3 with the crystal structure of arylacetamide κ agonist 1, ICI 19944

Full text of DOI:10.1016/j.bmcl.2004.08.041

0
Bioorganic & Medicinal Chemistry Letters 14 (2004) 5693–5697
Azepinone as a conformational constraint in the design
of j-opioid receptor agonists
Paul A. Tuthill,a Pamela R. Seida,a William Barker,a Joel A. Cassel,b Serge Belanger,b
Robert N. DeHaven,b Michael Koblish,b Susan L. Gottshall,b Patrick J. Little,b
Diane L. DeHaven-Hudkinsb and Roland E. Dollea,*
aDepartment of Chemistry, Adolor Corporation, 700 Pennsylvania Drive, Exton, PA 19341, USA
bDepartment of Pharmacology, Adolor Corporation, 700 Pennsylvania Drive, Exton, PA 19341, USA
Received 13 April 2004; accepted 17 August 2004
Available online 17 September 2004
Abstract—A new class of j-opioid receptor agonists is described. The design of these agents was based upon energy minimization
and structural overlay studies of the generic azepin-2-one structure 3 with the crystal structure of arylacetamide j agonist 1, ICI
199441. The most active compound identified was ligand 4a (Ki = 0.34nM), which demonstrated potent antinociceptive activity after
oral administration in rodents.
ꢀ 2004 Elsevier Ltd. All rights reserved.
j-Opioid receptors are one of three well-characterized
opioid receptor types.1 Centrally-acting j-receptor
agonists, for example, arylacetamides 1 (ICI 199441)
and 2 (U50,488) display potent antinociceptive activity
in vivo.2 j Agonists initially held great promise as anal-
gesics putatively free of respiratory depression, constipa-
tion, and other undesirable side effects observed with
clinically administered l-opioid receptor agonists. How-
ever, early clinical trial data revealed that j-receptor
agonists were accompanied with their own set of central
nervous system (CNS) liabilities, namely dysphoria and
diuresis.3 These side effects prevented their continued
clinical development and commercialization as analge-
sics. Recently, j-opioid receptors were identified in tis-
sues outside the CNS.4 It has been hypothesized that
peripherally acting j agonists may retain significant
antinociceptive activity without their associated classical
CNS side effects.5 This hypothesis was supported in a
recent clinical investigation of chronic pancreatitis
patients that were treated with ADL 10-0101, a periph-
erally acting j agonist and experienced pain relief with-
out dysphoria.6
As part of an ongoing program aimed at discovering
agonists that target peripheral j receptors,7 a novel ser-
ies of constrained analogs 3 of ICI 199441 (1)8 was syn-
thesized and evaluated as potential j agonists. The
rationale for the selection of the azepinone (seven-mem-
bered ring) as the cyclic constraint was based upon
energy minimization and structural overlay studies of
4a,b with the crystal structure of 1.9 As illustrated in
Figure 1, diastereomers 4a (syn-isomer) and 4b (anti-iso-
mer) reveal a high degree of overlap with the backbone
conformation of 1, suggesting such compounds may be
biologically active. This analysis presupposes that the
crystal structure reflects, at least to some extent, the
receptor bound bioactive conformation of 1. It was
not known a priori, which diastereomer (if either) would
be the more active conformer. The details of the synthe-
sis of 4a,b and related analogs and their in vitro binding
affinity and selectivity against the human j-opioid
receptor is the subject of this letter.10
The synthesis of the azepin-2-ones 4a,b is presented in
Scheme 1. Fukuyama sulfonylation (standard Schot-
ten–Baumann conditions) of commercially available
S-(+)-phenylglycine 5 followed by DCC coupling with
pyrrolidine afforded sulfonamide 6 in 40% yield. This
material was alkylated with a slight excess of allyl
bromide using potassium carbonate as base in DMF
*
Corresponding author. Tel.: +1 484 595 1024; fax: +1 484 595
1550; e-mail: rdolle@adolor.com
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.08.041

5694
P. A. Tuthill et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5693–5697
Figure 1. Overlays of 4a and 4b with 1 showing nearly coincident backbone conformations.
NO2
O
S
O
a, b
c, d
e
N
N
N
N
H
HN
H2N
CO2H
HN
O
O
5: (S)-phenylglycine
6
7
8
Cl
Cl
Cl
O
O
N
Cl
OH
N
Cl
Cl
O
9a
4a (syn isomer)
syn:anti = 3:1
N
f (8)
g
N
or
Cl
Cl
Cl
Cl
10
O
O
N
diastereomeric mixture formed
regardless of the stereochemistry of the
carboxylic acid or coupling cocktail used
N
OH
9b
4b (anti isomer)
Scheme 1. Preparation of azepinone-based j-opioid receptor agonists 4a,b. Reagents and conditions: (a) 1M aq NaOH (1.1equiv), 5 (1.0equiv),
25ꢁC, 30min; then 6M aq NaOH, 1.05equiv(2-NO 2)PhSO2Cl, THF, 0–25ꢁC, 20h (85%); (b) sulfonamide intermediate (1equiv), HOBT (1.1equiv),
pyrrolidine (1.1equiv), THF; then 1M solution DCC in DCM (1.1equiv), 0–25ꢁC, 16h (91%); (c) 6 (1equiv), K2CO3 (2equiv), allyl-Br (1.3equiv),
DMF, 25ꢁC, 16h (96%); (d) allyl sulfonamide intermediate (1equiv), K2CO3 (3equiv), PhSH (1.4equiv), DMF, 25ꢁC, 16h (85%); (e) 7 (1equiv), 1M
solution LAH in THF (1equiv), 0–25ꢁC, 16h (86%); (f) 8 (1equiv), 9a (or 9b or racemate; 1equiv), 2-Cl-1-methylpyridinium iodide (2equiv), TEA (3
equiv), DCM, 25ꢁC, 44h (93%); (g) 10 (1equiv), DCM, 1M HCl in Et2O (4equiv), 25ꢁC, 30min; solvents removed in vacuo and HCl salt dried in
vacuo, 60ꢁC, 3h; then GrubbÕs catalyst (0.05equiv), DCM, 40ꢁC, 16h (4a: 30%; 4b: 13%).

P. A. Tuthill et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5693–5697
5695
Table 1. Azepinone in vitro binding and selectivity data17
Compound Structure Ki (nM)
its hydrochloride salt. Most surprisingly, a 3:1 mixture
of ring closed products 4a and 4b (vida infra) was ob-
tained in 60% yield.12a These products were separated
j
l
d
1
by silica gel chromatography and gave very distinct H
NMR spectra.12b The chemistry was repeated using anti-
pode 9b. However, this also gave the identical 3:1 ratio
of cyclized products 4a,b. Apparently, racemization of
9a and 9b occurred before coupling with the hindered
chiral diamine 8.13 This effect is reminiscent of that seen
in kinetic resolutions.14 A wide range of coupling rea-
gents and reaction conditions were investigated includ-
ing DCC, oxalyl chloride, MukaiyamaÕs reagent,
TBTU, PyBrOP, and GoodmanÕs DEPBT;15 however,
in every case a 3:1 ratio of RCM products was obtained.
Since racemization could not be circumvented, all subse-
quent couplings were carried out using the requisite
racemic 2-aryl-pent-4-enoic acids or 2-aryloxy-pent-4-
enoic acid16 and MukaiyamaÕs reagent to generate a sep-
arable 3:1 mixture of the azepin-2-ones 11–15 (Table 1).
With respect to assigning relative stereochemistry, the
X-ray crystallographic analysis of a major RCM prod-
uct was secured (compound 14; Fig. 2), confirming the
syn-stereochemistry.12 In conjunction with 1H NMR
spectral analysis that clearly distinguished between the
syn- and anti-diastereomers, the syn-stereochemistry
was confidently assigned to the major product isolated
from all the RCM reactions.
1
0.043
53
24
Cl
Cl
O
N
N
N
N
Cl
Cl
O
4a
0.34 1000 370
Cl
Cl
O
a
a
a
a
a
a
a
a
a
a
a
a
a
4b
17
N
N
Cl
Cl
O
O
11a
11b
12a
12b
13
2.0
N
N
N
N
35
22
O
O
N
N
N
The results of the in vitro binding assay for compounds
4a,b, 11–15 and reference compound 1 are presented in
Table 1. Azepin-2-one 4a possessed subnanomolar affin-
ity for the j-opioid receptor (Ki = 0.3nM), comparing
O
O
800
N
favorably to the acyclic arylacetamide
1
(Ki =
01.04nM).17 This result supports the hypothesis put
forth by the modeling overlay studies. The l/j selectivity
for 4a and 1 are likewise comparable with each being on
the order of ca. 1000-fold. The anti-diastereomer 4b was
50-fold less potent than the syn-diastereomer 4a. This
trend, syn- more active than the anti-diastereomer, was
consistently observed throughout the azepin-2-one series
(compare 11a to 11b and 12a to 12b) and may arise from
an unfavorable non-bonded interaction between the
F
F
F
O
0.72
1000
N
N
O
a
14
1900
N
O
N
S
N
O
O
a
15
2.6
Cl
Cl
O
N
N
For the assay description, see Ref. 19. Ki, values are the geometric
means computed from at least three separate determinations.
a Inhibition (<60%) at 10lM screening concentration; estimated
Ki > 5000nM.
at room temperature and subsequently deprotected with
benzenethiol to produce amide 7. Treatment of 7 with
LAH in THF gave the key chiral diamine 8 (ꢀ65%
overall yield from 6).11 Coupling of diamine 8 using
MukaiyamaÕs reagent to the optically pure 2-(30,40-
dichlorophenyl)pent-4-enoic acid 9a (Pharmacore, Inc.)
gave the penultimate diene intermediate 10. Ring closing
metathesis (RCM) of 10 with either GrubbsÕ first or sec-
ond generation catalysts proved unsuccessful until the
tertiary pyrrolidine amine nitrogen was converted to
Figure 2. Crystal structure of 14 showing syn-stereochemistry.

5696
P. A. Tuthill et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5693–5697
hydrocarbon ring atoms in the anti- versus the syn-lig-
ands in the active site of the receptor. Another discern-
ible SAR trend was the preference for lipophilic
substituents present in the aromatic group attached to
the azepinone ring. For example, 4a (3,4-dichlorophen-
yl), 11a (4-chlorophenyl), and 13 (4-trifluoromethylphe-
nyl) possessed potent j binding ranging from 0.34 to
2nM. This is in contrast to the more hydrophilic aryl
ring substituents, for example, 12a (4-methoxyphenyl)
and 14 (3-pyrollidinylsulfonylphenyl) that were 10- to
1000-fold less active. This same preference toward lipo-
philic aryl ring substitution was also observed in the acy-
clic series.7 Lastly, the 3,4-dichlorophenoxy-substituted
azepin-2-one 15 also demonstrated significant potency
and high selectivity for the j receptor (Ki = 2.6nM,
>10,000 selectivity versus l and d). This again under-
scores the preference for a lipophilic aryl substituent
and indicates that the receptorÕs binding pocket is able
to accommodate an extended aryl ring unit.18
preference for lipophilic aromatic substituents. j Agon-
ist 4a (Ki = 0.3nM) was orally active in rodent models of
antinociception and may be regarded as possessing some
measure of peripheral selectivity as evidenced by the ca.
30-fold separation observed between antinociception
and sedation and ataxia.
References and notes
1. (a) Scopes, D. I. C. Exp. Opin. Invest. Drugs 1994, 3, 369–
376; (b) Dhawan, B. N.; Cesselin, F.; Raghubir, R.;
Reisine, T.; Bradley, P. B.; Portoghese, P. S.; Hamon, M.
Pharmacol. Rev. 1996, 48, 567–592.
2. (a) Barber, A.; Gottschlich, R. Exp. Opin. Invest. Drugs
1997, 6, 1351–1368; (b) Costello, G. F.; Main, B. G.;
Barlow, J. J.; Carroll, J. A.; Shaw, J. S. Eur. J. Pharmacol.
1988, 151, 475–478; (c) Von Voigtlander, P. F.; Lahti,
R. A.; Ludens, J. H. J. Pharmacol. Exp. Ther. 1983, 224,
7–12.
3. (a) Dionne, R. A.; Dobbins, K. R.; Hargreaves, K. M.
Clin. Pharmacol. Ther. 1991, 49, 183; (b) Pande, A. C.;
Pyke, R. E.; Greiner, M.; Cooper, S. A.; Benjamin, R.;
Pierce, M. W. Clin. Neuropharmacol. 1996, 19, 92–97; (c)
Pande, A. C.; Pyke, R. E.; Greiner, M.; Wideman, G. L.;
Benjamin, R.; Pierce, M. W. Clin. Neuropharmacol. 1996,
19, 451–456.
In further studies, compound 4a was evaluated in sev-
eral in vivo animal models of antinociception (Table
2). Compound 4a displayed potent antinociceptive activ-
ity in the mouse and rat after oral administration. The
ED50 values of inhibition of acetic acid-induced writhing
in mice and inhibition of formalin-induced flinching in
rats after oral administration were 1.6 and 2.7mg/kg,
respectively. There was a clear separation between anti-
nociceptive activity and sedation and ataxia in both the
mice and rats following oral administration of 4a, with a
30-fold separation between the ED50 values for sedation
and the ED50 values for antinociception. Separation of
these in vivo activities is one indicator of peripheral
selectivity, suggesting that the azepinone class of j agon-
ists may represent a good starting point for the further
design and optimization of peripherally acting agents.
4. Zhang, Q.; Schafer, M.; Elde, R.; Stein, C. Neuroscience
1998, 85, 281–291.
5. For a review regarding peripherally acting opioid agonists
as analgesics, see: DeHaven-Hudkins, D. L.; Dolle, R. E.
Curr. Pharm. Des. 2004, 10, 743–757.
6. Eisenach, J. C.; Carpenter, R.; Curry, R. Pain 2003, 101,
89–95.
7. Kumar, V.; Marella, M. A.; Cortes-Burgos, L.;
Chang, A.-C.; Cassel, J. A.; Daubert, J. D.; DeHaven,
R. N.; DeHaven-Hudkins, D. L.; Gottshall, S. L.;
Mansson, E.; Maycock, A. L. Bioorg. Med. Chem. Lett.
2000, 10, 2567–2570.
8. Costello, G. F.; James, R.; Shaw, J. S.; Slater, A. M.;
Stutchbury, N. C. J. J. Med. Chem. 1991, 34, 181–189.
9. For the crystal structure of 1, see: Subramanian, G.;
Paterlini, M. G.; Larson, D. L.; Portoghese, P. S.;
Ferguson, D. M. J. Med. Chem. 1998, 41, 4777–4789.
10. Tuthill, P. A., Sr.; Seida, P.; Dolle, R. E.; Cassel, J. A.;
DeHaven, R. N. Abstracts of Papers, 226th ACS National
Meeting, New York, NY, September 7–11, 2003, MEDI
151.
11. To ensure chiral integrity, diamine 8 was acylated with 2-
naphthoyl chloride. The corresponding amide was then
subjected to chiral HPLC analysis (Chiralpak AD) and
was determined to be 96% ee, indicating that no significant
racemization had occurred at the carbon atom bearing the
phenyl ring.
In summary, the azepin-2-one was hypothesized as a
conformational constraint for j-opioid ligands based
on near coincident backbone overlay of 4a,b with the
crystal structure of the archetypal arylacetamide j agon-
ist 1. The hypothesis was confirmed upon the synthesis
and evaluation of the series 4a,b, and 11–15, demon-
strating potent, selective j-receptor binding affinity.
Structure–activity relationships in vitro clearly indicated
a predilection for syn- versus anti-stereochemistry and a
Table 2. Antinociceptive and sedative effects of 4a upon oral
administration
Mouse
Rat
12. (a) All new compounds showed physical and spectroscopic
properties consistent with their structures; (b) For
4a: 1H NMR CDCl3 (400MHz) d 11.95–11.89 (1H, br
s), 7.71–7.70 (1H, d, J = 1.8Hz), 7.61–7.59 (1H, dd,
J = 1.8, 8.3Hz), 7.42–7.40 (1H, d, J = 8.3Hz), 7.39–7.34
(3H, m), 7.27–7.23 (2H, m), 6.42–6.38 (1H, dd, J = 3.0,
12.5Hz), 5.60–5.55 (1H, m), 5.28–5.20 (2H, m), 4.96–4.85
(1H, dd, J = 3.3, 13.6Hz), 4.13–4.05 (1H, m), 3.96–3.88
(2H, m), 3.24–3.14 (2H, m), 2.95–2.68 (3H, m), 2.41–2.36
(1H, m), 2.35–2.29 (1H, m), 2.24–2.14 (1H, m), 2.31–1.95
(2H, m). For 4b: 1H NMR CDCl3 (400MHz) d 12.40–
12.34 (1H, s), 7.48–7.46 (2H, m), 7.42–7.35 (5H, m), 7.19–
7.16 (1H, dd, J = 1.9, 8.3Hz), 5.77–5.67 (2H, m), 5.39–5.36
(1H, t, J = 6.4Hz), 4.82–4.78 (1H, d, J = 8.1Hz), 4.42–4.38
(1H, dd, J = 3.4, 12.8Hz), 4.28–4.23 (1H, m), 3.80–3.59
Acetic
Sedation
(ED50
(mg/kg))
Formalin-
induced
flinching
Rotarod
(% inhibition)c
acid-induced
writhing
(ED50 (mg/kg))
(ED50 (mg/kg))
1.6
(1.1–2.4)a,b
48 2.7
(25.3–90.7) (1.2–4.5)
54%
@ 100mg/kg
For the assay descriptions, see Ref. 20.
a Values in parentheses are the 95% confidence intervals.
b The oral ED50 of 1 in writhing = 0.17mg/kg (0.05–0.52) and the
platform sedation ED50 = 7.5mg/kg (3.6–18.3).
c Value represents the percent decrease in rotarod performance after
drug treatment based on the comparison of baseline and post-treat-
ment rotarod latencies.

P. A. Tuthill et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5693–5697
5697
(4H, m), 2.91–2.81 (1H, m), 2.70–2.58 (2H, m), 2.45–2.39
(1H, dd, J = 3.3, 18.4Hz), 2.24–2.12 (2H, m), 2.04–1.92
(2H, m). For 14: tan solid; mp 97–99ꢁC (methanol); H
NMR CDCl3 (400MHz) d 7.75 (1H, br s), 7.73 (1H, d, J =
7.8Hz), 7.64 (1H, d, J = 7.6Hz), 7.52 (1H, t, J = 7.8Hz),
7.35–7.25 (5H, m), 6.04 (1H, br t, J = 7.6Hz), 5.68 (1H,
m), 5.43 (1H, m), 4.49 (1H, d, J = 11.7Hz), 4.25 (1H, d,
J = 17.4Hz), 3.36 (1H, dd, J = 7.6, 17.40Hz), 3.29–3.23
(4H, m), 3.03 (1H, t, J = 10.7Hz), 2.84 (2H, m), 2.63 (2H,
s), 2.50 (3H, m), 1.77–1.71 (8H, m); LCMS m/z 494.3
(M+1).
1994, 4, 677–682; (b) The RCM reaction proved to be a
versatile reaction and was used to prepare higher order
lactams, for example, 19
1
Cl
Cl
Cl
Cl
O
O
OH
N
N
N
N
18: K (κ) = 0.2 nM
19: K (κ) = 12 nM
i
i
13. The 1H NMRs of the intermediate dienes were rather
complex, the presence of rotomers notwithstanding, indi-
cating a mixture of diastereomers. In one case, the mixture
of dienes was separated and the major diene diastereomer
so obtained was subjected to the ring closing metathesis
reaction, affording exclusively the syn-product (unpub-
lished observation); hence, the metathesis reaction was
racemization free.
14. For a review on kinetic resolution strategies using non-
enzymatic catalysts, see: Robinson, D. E. J. E.; Bull, S. D.
Tetrahedron: Asymmetry 2003, 14, 1407–1446.
15. Li, H.; Jiang, X.; Ye, Y.-H.; Fan, C.; Romoff, T.;
Goodman, M. Org. Lett. 1999, 1, 91–93.
(>100x selective versus µ and δ)
(>1000x selective versus µ and δ)
19. DeHaven, R. N.; DeHaven-Hudkins, D. L. In Current
Protocols in Pharmacology; Enna, S. J., Williams, M.,
Ferkany, J. W., Kenakin, T., Porsolt, R. D., Sullivan, J.
P., Eds.; John Wiley and Sons: New York, 1998, pp 1.4.1–
1.4.12.
20. In vivo test methods: (a) Acetic acid-induced writhing: Mice
were treated with vehicle or compound 4a and 15min later
injected with 0.6% acetic acid intraperitoneally. Five
minutes after the administration of acetic acid, the number
of writhes was counted for a 10min period; (b) Horan, P.;
de Costa, B. R.; Rice, K. C.; Porreca, F. J. J. Pharmacol.
Exp. Ther. 1991, 257, 1154–1161. Briefly, mice were placed
on a raised plastic platform (11 · 7.5 · 3cm) and the
latency to completely step off the platform was determined
before (baseline sedation latency) and 30min after drug
treatment. Mice with baseline sedation latencies of 15s or
16. Acid 16 required for 15 was prepared from allyl 3,4-
dichlorophenoxyacetate 17 via the enolate Claisen rear-
rangement using the method of Burke: Burke, S. D.;
Fobare, W. F.; Pacofsky, G. J. J. Org. Chem. 1993, 48,
5221–5228.
1) LDA (1.1 equiv), THF, -100 oC,
1 h; then TEA/TMSCl (1 equiv),
-100 oC to 25 oC, 1 h
2) 1 M aq HCl
O
O
Cl
Cl
O
Cl
Cl
O
OH
O
17
16
17. The azepin-2-ones were full agonists as determined
in
[35S]GTPcS functional binding assay (see
Ref. 19). For 4a: EC50 = 10nM; for 4b: EC50 = 50nM.
18. (a) The 3-hydroxy analog 18 of 4a was prepared. The
hydroxy group is known to enhance receptor affinity for j:
Gottschlich, R.; Ackermann, K. A.; Barber, A.; Bart-
oszyk, G. D.; Greiner, H. E. Bioorg. Med. Chem. Lett.
greater were not used. After the drug treatment, a 30s cut-
off was used for the latency to step off the platform; (c)
Formalin-induced flinching: Late phase formalin-induced
flinching was determined using the method of DeHaven-
Hudkins, D. L.; Cortes Burgos, L.; Cassel, J. A.; Daubert,
J. D.; DeHaven, R. N.; Mansson, E.; Nagasaka, H.; Yu,
G.; Yaksh, T. J. Pharmacol. Exp. Ther. 1999, 289, 494–502.
a