Decahydroquinoline amides as highly selective CB2 agonists: Role of selectivity on in vivo efficacy in a rodent model of analgesia

Article abstract of DOI:10.1016/j.bmcl.2011.02.078

A novel series of decahydroquinoline CB2 agonists is described. Optimization of the amide substituent led to improvements in CB2/CB1 selectivity as well as physical properties. Two key compounds were examined in the rat CFA model of acute inflammatory pain. A moderately selective CB2 agonist was active in this model. A CB2 agonist lacking functional CB1 activity was inactive in this model despite high in vivo exposure both peripherally and centrally.

Full text of DOI:10.1016/j.bmcl.2011.02.078

Bioorganic & Medicinal Chemistry Letters 21 (2011) 2359–2364
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Decahydroquinoline amides as highly selective CB2 agonists: Role
of selectivity on in vivo efficacy in a rodent model of analgesia
Peter J. Manley a, , Amy Zartman , Daniel V. Paone , Christopher S. Burgey , Darrell A. Henze ,
⇑
a
a
a
b
b
b
b
b
c
c
Kimberly Della Penna , Reshma Desai , Michael D. Leitl , Wei Lemaire , Rebecca B. White , Suzie Yeh ,
Mark O. Urban , Stefanie A. Kane , George D. Hartman , Mark T. Bilodeau , B. Wesley Trotter
b
b
a
a
d
a
Department of Medicinal Chemistry, Merck Research Laboratories, West Point PA, United States
Department of Pain Research, Merck Research Laboratories, West Point PA, United States
Department of Drug Metabolism & Pharmacokinetics, Merck Research Laboratories, West Point PA, United States
Department of Medicinal Chemistry, Merck Research Laboratories, Boston, MA, USA
b
c
d
a r t i c l e i n f o
a b s t r a c t
Article history:
A novel series of decahydroquinoline CB2 agonists is described. Optimization of the amide substituent led
to improvements in CB2/CB1 selectivity as well as physical properties. Two key compounds were exam-
ined in the rat CFA model of acute inflammatory pain. A moderately selective CB2 agonist was active in
this model. A CB2 agonist lacking functional CB1 activity was inactive in this model despite high in vivo
exposure both peripherally and centrally.
Received 10 November 2010
Revised 16 February 2011
Accepted 22 February 2011
Available online 26 February 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
CB2 agonist
CB1 agonist
Selectivity
Pain
CFA model
The cannabinoid receptors, CB1 and CB2, are well characterized
members of the G-protein coupled receptor (GPCR) family of
receptors and are a part of the endocannabinoid system. Non-
selective ligands for these receptors exhibit a number of biological
properties, one of which is analgesia. In addition to this desirable
effect these molecules also exhibit undesired neurological effects
linked to central CB1 agonist activity1 which, in turn, could limit
their usefulness in the treatment of pain. Many labs in both acade-
mia and industry have postulated that a selective CB2 agonist
might be therapeutically useful in the treatment of pain without
the undesired neurological effects of the non-selective
cannabinoids.
ical properties. Compounds which were selective for CB2 but
retained some CB1 agonist activity in vitro elicited a positive re-
sponse in the rat CFA hyperalgesia model similar to other reports
in the literature. However, compounds which lacked measurable
in vitro CB1 agonist activity failed to affect pain responses in the
same rat model despite high peripheral and central exposures.
This result was rationalized by previous reports of the existence
of a large CB1 receptor reserve in the CNS and the ability of com-
pounds with weak CB1 binding to elicit substantial agonist re-
,2
5
sponses. The authors concluded that CB2 agonist activity alone
was not sufficient to alleviate pain in a rat model of acute inflam-
matory pain.
The literature contains a number of reports describing the de-
sign, synthesis, and in vivo evaluation of selective CB2 agonists
in the pursuit of new treatments for pain.3 The vast majority of
these studies, however, involve molecules which retain some mea-
In order to interrogate this hypothesis further our group sought
out compounds with unique structures and similar in vitro profiles
to those of the imidazopyridines. Compounds were required to
have a range of selectivity profiles for CB2/CB1 and physical prop-
erties appropriate for in vivo evaluation. High-throughput screen-
ing of the Merck sample collection resulted in the identification of
lead 1 having undetermined stereochemistry (Fig. 1).
A two-pronged approach was taken to investigate the structural
requirements around the decahydroquinoline core. In order to
determine the optimal stereochemical arrangement about carbons
1, 2, and 3, two large, diverse libraries of amides derived from race-
mic amines 2 and 3 were synthesized. Analysis of in vitro CB2 and
4
surable in vitro CB1 agonist activity. In the previous Letter Trotter
et al. described a novel class of imidazopyridine CB2 agonists. The
authors synthesized and evaluated a collection of potent imidazo-
pyridine CB2 agonists with a range of selectivity profiles and phys-
⇑
Corresponding author. Tel.: +1 215 652 9466; fax: +1 215 652 3971.
E-mail address: peter_manley@merck.com (P.J. Manley).
0
960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.02.078

2
360
P. J. Manley et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2359–2364
deletion or truncation
deletion
>
100x loss
little effect
H 1
2
OH
OH
H
H
H
H
OH
OH
3
N
N
H
O
N
H
N
H
reduction
O
or replacement
not tolerated
1
2
3
Figure 2.
Figure 1.
6
Table 2
CB1 agonist activity revealed a clear correlation between decahy-
droquinoline ring stereochemistry and CB2 agonist potency. The cis
stereochemistry of amine 3, in which the two ring protons and the
aryl group are all on the same face of the molecule, was crucial for
CB2 agonist activity over a wide range of structures (Table 1).
Simultaneous work on the decahydroquinoline core itself re-
vealed that deletion or truncation of the unsubstituted cyclohex-
ane ring resulted in a 100-fold or greater loss of CB2 agonist
activity. Deletion of the tertiary alcohol in the trans-ring system
Aryl substituents: hCB2 cAMP IC50 nM (% Emax)a
O
H
R
H
OH
RMgX or RLi
4
N
N
H
O
H
O
THF, -78 °C
2-1
2-4
2-5
2-6
2-7
2-8
2-9
SO Me
Cl
Cl
(
4
amine 2) had little effect on CB2 agonist activity (83 nM vs
5 nM); this hydroxyl group was therefore retained in the cis-ring
2
SO Me
2
system for ease of synthesis and improved physical properties.
Altering the connectivity about the amide nitrogen, either by
reducing the carbonyl to the corresponding tertiary amine or
replacing the carbonyl with either a sulfonamide or urea linkage,
was generally not tolerated (Fig. 2).
106 (56)
OMe
155 (88)
493 (69)
338 (52)
2-2
CF
3
4
In analogy to the previous Letter we sought to increase polarity
3
58 (66)
810 (-16)
in our series in order to lower the log P and protein binding of our
lead structures. Initially this was attempted by varying the aryl
ring at carbon 1. Ketone 4 was synthesized according to the meth-
ods of Prost. Grignards or organolithiums were added to 4 in THF
at low temperature to give the desired alcohols (Table 2) in most
cases. When the addition did not occur in sufficient yield the use
2-3
N
CN
1045 (85)
N
7
9
38 (86)
1357 (65)
a
See Ref. 6.
Table 1
Representative structures from diverse amide libraries
H
H
N
O
R
Trans-fused amine 2
Cis-fused amine 3
HO
Entry
1
R
hCB2 cAMP IC50 nM (% Emax)a
hCB1 AMP IC50 nM (% Emax)a
hCB2 cAMP IC50 nM (% Emax)a
hCB1 cAMP IC50 nM (% Emax)a
45 (99)
>17,000 (14)
4.2 (100)
4543 (76)
2
3
24 (90)
41 (98)
963 (72)
0.14 (90)
0.82 (98)
312 (79)
188 (98)
Cl
N
2022 (84)
NH
4
5
22 (94)
>17,000 (31)
0.62 (100)
21 (99)
>17,000 (64)
>17,000 (46)
N
NH
F
1286 (81)
>17,000 (ꢀ56)
6
>17,000 (34)
>17,000 (34)
5.9 (102)
>17,000 (6)
N
Me
a
See Ref. 6.

P. J. Manley et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2359–2364
2361
8
of CeCl
3
was employed to give the desired products. While some
achieved in a similar manner to that of the fused heterocycles. In
general both of these classes of compounds had no measurable
CB1 agonist activity in the human cAMP assay (CB1
IC50 >17,000 nM) across a broad range of structures and CB2 ago-
nist activities.
substitution on the phenyl ring was moderately tolerated (2-1, 2-
) polar heterocycles (2-3, 2-9) were not. Ultimately the unsubsti-
4
tuted phenyl ring was determined to be optimal.
With the pharmacophore around the decahydroquinoline core
well defined, efforts were focused on optimizing the amide
substituent for agonist activity, selectivity, and physical properties.
This was achieved using an iterative library approach based on the
acylation of racemic amine 3. The synthesis of amine 3 was initi-
ated by modifying the procedures of Prost.7 First, elaboration of
Substituted pyridines emerged as promising candidates for fur-
ther evaluation from the aromatic heterocycle class of compounds.
The 6-methyl-3-pyridyl analog 4-13 exhibited good overall physi-
cal properties (log P 2.26) and high plasma free fraction (15%) in
9
rats. The single enantiomer 4-13a maintained CB2 agonist activity
10
1
-acetylcyclohexene (I) with paraformaldehyde and dimethyl-
and was not a substrate for rat P-glycoprotein efflux (BÀA/
À6
amine hydrochloride under Mannich conditions gave aminoketone
II in 60% yield. II was treated with concentrated aqueous ammonia
in 1,4-dioxane at 120 °C in a stainless steel pressure vessel to give
decahydroquinolinone III. After work-up crude III was acylated
with benzyloxycarbonyl chloride to give a primarily trans mixture
of IV. Isomerization of the center adjacent to the ketone was
in MeOH to give cis-V in
0% isolated yield over three steps. Addition of PhMgBr to V at
low temperature gave alcohol VI in 74% yield. Deprotection of VI
by hydrogenolysis gave racemic 3 in quantitative yield. Acylation
of 3 was achieved using standard peptide coupling conditions
AÀB = 1.3, Papp 36 Â 10e cm/s). 4-13a did exhibit some erosion
in selectivity over CB1 in both the human and rat cAMP assays (Ta-
ble 6). N-Methyl-morpholine 5-2 stood out in the saturated hetero-
cycles class of compounds, displaying good overall physical
properties (log P 2.28) and very high free fraction (34%) in rats. Like
9
4-13a, the single enantiomer 5-2a was not a substrate for rat P-
1
0
À6
achieved by treatment of IV with K
3
2
CO
3
glycoprotein efflux (BÀA/AÀB = 1.3, Papp 39 10e cm/s). 5-2a
maintained its in vitro selectivity profile and exhibited only a slight
loss in CB2 agonist activity in the rat cAMP assay relative to the hu-
man data (Table 6).
4-13a and 5-2a were selected for evaluation in the rat CFA
1
1
(Scheme 1).
hyperalgesia model. Compound levels were measured in the
plasma, brain homogenate (brain), and cerebrospinal fluid (CSF)12
in all experiments to ensure adequate exposure for CB2 activity.
When dosed subcutaneously (10, 30, and 100 mpk) 4-13a exhib-
ited a dose dependent reversal of paw withdrawal threshold
(28%, 53%, and 76% respectively) at the 60 min timepoint (Fig. 3).
At the minimum dose needed to achieve naproxen-like efficacy
(30 mpk) 4-13a reached levels in the plasma, brain, and CSF of
4200, 3200, and 223 nM respectively. In contrast, dosing 5-2a sub-
cutaneously at 30 mpk failed to elicit a response in the rat CFA
hyperalgesia model (Fig. 4) despite exposure in plasma, brain,
and CSF (6600, 9800, and 1500 nM respectively) well above its
Fused heterocycles, derived from the coupling of 6,5- and 6,6-
fused heterocyclic carboxylic acids and racemic amine 3, were
generally potent CB2 agonists (Table 3). While only moderate
selectivity was achieved for most heterocycles tested, a few excep-
tions were observed. Indazole 3-6 was a potent CB2 agonist with
no measurable functional CB1 activity in the human cAMP assay.
3
-6 exhibited a reasonable plasma free fraction (4.8%) in rats but
displayed poor pharmacokinetic behavior (2 mpk iv; Cl = 70 mL/
min/kg, t1/2 = 0.43 h) and was thus unsuitable for evaluation in
in vivo pain models.
Acylation of racemic amine 3 with either aromatic (Table 4) or
saturated (Table 5) monocyclic heterocyclic carboxylic acids was
CB2 IC50
.
Table 3
6
,5- and 6,6-fused heterocyclic amides
H
H
N
O
R
hCB2 cAMP IC50 nM (% Emax)a
hCB1 cAMP IC50 nM (% Emax)a
Selectivity hCB2/hCB1
HO
Entry
R
3
-1
3.38 (96)
1528 (69)
452
N
N
3
3
-2
-3
0.53 (87)
5.5 (99)
261 (95)
492
626
N
3441 (69)
N
3
3
-4
-5
1.4 (104)
35 (101)
1002 (91)
716
N
H
N
NH
>17,000 (ꢀ53)
>486
N
NH
3
-6
4.0 (102)
>17,000 (15)
>4000
a
See Ref. 6.

2
362
P. J. Manley et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2359–2364
Table 4
Monocyclic aromatic heterocyclic amides
H
H
N
O
R
hCB2 cAMP IC50 nM (% Emax)a,b
hCB2 cAMP IC50 nM (% Emax)a,b
HO
Entry
R
Entry
R
N
N
4
-1
-2
O
11 (102)
6.3 (99)
4-8
41 (101)
29 (101)
S
N
4
4-9
N
N
F
N
N
N
N
4
4
4
-3
-4
-5
124 (100)
6.1 (99)
4-10
4-11
4-12
12 (96)
N
CF3
Me
N
Me
O
N
140 (103)
156 (98)c
O
N
N
16 (100)
CF
3
N
4
-6
-7
N
282 (101)
12 (96)
4-13
4-14
14 (99)
HN
Me
N
N
N
4
26 (104)
CF3
a
b
c
See Ref. 6.
Except where noted, compounds exhibited hCB1 IC50 >17,000 nM.
hCB1 IC50 = 3445 nM (50%).
Table 5
Monocyclic saturated heterocyclic amides
H
H
N
O
R
hCB2 cAMP IC50 nM (% Emax)a,b
hCB2 cAMP IC50 nM (% Emax)a,b
HO
Entry
R
Entry
R
5
-1
-2
56 (100)
30 (100)
5-6
135 (99)
N
Me
O
N
Me
O
O
5
5-7
5-8
N
H
3471 (89)
N
Me
H
N
O
N
H
5
-3
62 (91)
193 (102)
30 (99)c
N
H
O
N
N
Me
OH
5
-4
-5
395 (98)
5-9
CF3
5
2175 (87)
HN
a
b
c
See Ref. 6.
Except where noted, compounds exhibited hCB1 IC50 >17,000 nM.
hCB1 IC50 = 4044 nM (96%).

P. J. Manley et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2359–2364
2363
Table 6
Properties of 4-13a and 5-2a
O
N
O
O
H
a
b
Me
N
Me
O
60%
Me
N
H H
H
H
O
R
N
Me
Me
N
I
II
III
HO
4-13a
5-2a
c
a
a
hCB2 cAMP IC50 nM (% Emax
hCB1 cAMP IC50 nM (% Emax
rCB2 cAMP IC50 nM (% Emax
rCB1 cAMP IC50 nM (% Emax
)
)
)
7.3 (103)
5020 (78)
6.3 (119)
1887 (94)
15
16 (101)
>17,000 (11)
81 (98)
>17,000 (8)
34
a
O
N
O
a
e
H
H
)
d
H
H
Rat plasma free fraction (%)
OH
VI
a
See Ref. 6.
74%
30% over
N
Cbz
H
H
N
Cbz
Cbz 3 steps
V
IV
4
In this and the previous Letter we have demonstrated that two
f
1
00%
structurally novel and moderately selective CB2 agonists show a
dose dependent effect on analgesia in a rat model of acute inflam-
matory pain. The in vivo results of 4-13a and imidazopyridne 6a
4
are similar to other moderately selective CB2 agonists in the
1
3,14
The observation of Gifford et al.5 that a large CB1
H
OH
H
\literature.
g
OH
R
receptor reserve exists in the CNS and that compounds with weak
CB1 binding exhibit substantial agonist responses led us to search
for structurally similar compounds which were functionally CB2
specific in vitro. Improving the functional selectivity of our com-
N
N
H H
H
O
4
racemic 3
pounds for CB2 over CB1, as in 5-2a and imidazopyridine 17b, re-
sulted in abolishment of the previously observed antinociceptive
effects in rats despite high central and peripheral exposure. These
results are consistent with Sain et al.’s15 recent demonstration
using transgenic mice and a non-selective CB2/CB1 agonist that
Scheme 1. Conditions: (a) dimethylamine hydrochloride, paraformaldehyde, HCl,
EtOH, reflux; (b) concentrated NH OH, 1,4-dioxane, 120 °C; (c) CBz-Cl, TEA, CH Cl ,
4
2
2
rt; (d) K
2
CO
3
, MeOH, rt; (e) phenylmagnesium bromide, THF, À78 °C; (f) H
2
, 10% Pd/
C, EtOH; (g) carboxylic acid, EDC, HOAt, TEA, DMF.
Vehicle n=6
2
1
3
1
0 mg/kg naproxen
4
4
3
3
2
2
1
1
50
00
50
00
50
00
50
00
76.4%
0 mg/kg Cmpd 4-13a
0 mg/kg Cmpd 4-13a
00 mg/kg Cmpd 4-13a
100
5
3.6%
75
50
25
0
*
*
42.4%
*
*
28.2%
*
4.8%
5
0
0
BL
CFA
30 min 60 min
60 min post-dose
Figure 3. Effects of 4-13a on paw withdrawal threshold in rat CFA model.
1
00
4
4
3
3
2
2
1
1
50
00
50
00
50
00
50
00
Vehicle
2
3
0 mg/kg Naproxen
0 mg/kg Cmpd 5-2a
75
5
2
0
5
0
2
3.9%
*
*
11.0%
3
.3%
5
0
0
BL
CFA
60 min
60 min post-dose
Figure 4. Effects of 5-2a on paw withdrawal threshold in rat CFA model.

2
364
P. J. Manley et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2359–2364
CB1 receptor activation is necessary for antinociception in animal
models.
H
H
O
R
Enantiomer 3a
Enantiomer 3b
N
HO
hCB2
cAMP
hCB1
cAMP
hCB2
cAMP
hCB1
cAMP
Acknowledgments
IC50 nM IC50 nM
IC50 nM
IC50 nM
(Emax
)
We would like to thank Joan S. Murphy and Dr. Charles W. Ross
III for HRMS analysis. We would also like to thank Janine N. Brouil-
lette and Dr. Steven M. Pitzenberger for NMR support.
a
a
a
a
R
(Emax
)
(Emax
)
(Emax
)
N
NH
1.6
>17,000
(28)
388
(97)
>17,000
(51)
(
100)
N
References and notes
4.9
4247
(95)
5221
(77)
>17,000
(10)
(
100)
1.
2.
3.
Graham, E. S.; Ashton, J. C.; Glass, M. Front. Biosci. 2009, 14, 944.
Beltramo, M. Mini-Rev. Med. Chem. 2009, 9, 11.
Thakur, G. A.; Tichkule, R.; Bajaj, S.; Makriyannis, A. Expert Opin. Ther. Pat. 2009,
N
S
0.55
(100)
>17,000
(12)
597
(89)
>17,000
(3)
N
19, 1647.
H
4
.
Trotter, B. W.; Nanda, K. K.; Burgey, C. B.; Potteiger, C. M.; Deng, J. Z.; Green, A.
I.; Hartnett, J. C.; Kett, N. R.; Wu, Z.; Henze, D. A.; Della Penna, K.; Desai, R.; Leitl,
M. D.; Lemaire, W.; White, R. B.; Yeh, S.; Urban, M. O.; Kane, S. A.; Hartman, G.
D.; Bilodeau, M. T. Bioorg. Med. Chem. Lett. 2011, in press.
N
N
7.9
>17,000
(18)
>17,000
(19)
930
(46)
F
(
98)
5
.
.
Gifford, A. N.; Bruneus, M.; Gatley, S. J.; Lan, R. X.; Makriyannis, A.; Volkow, N.
D. J. Pharmacol. Exp. Ther. 1999, 288, 478.
6
Compounds were added to a Greiner black 384 well low volume assay
plate. 1000 CHO-K1 cells (expressing human or rat CB1 or CB2) were
added to each well of the assay plate containing compound, then
incubated at room temperature for 15 min. Then, forskolin at EC70 was
added and incubated at room temperature for an additional 30 min.
Detection of cAMP was performed using Cisbio’s HRTF cAMP dynamic 2 kit
following the manufacturer’s protocol for cAMP detection. After adding d2-
cAMP and anti-cAMP-cryptate to all the wells, there was a final incubation
at room temperature for 1 h. Plates were then read on an EnVision plate
a
See Ref. 6.
1
1
0. Miller, D. S.; Bauer, B.; Hartz, A. M. S. Pharmacol. Rev. 2008, 60, 196.
1. Nagakura, Y.; Okada, M.; Kohara, A.; Kiso, T.; Toya, T.; Iwai, A.; Wanibuchi, F.;
Yamaguchi, T. J. Pharmacol. Exp. Ther. 2003, 306, 490.
2. Liu, X. R.; Smith, B. J.; Chen, C.; Callegari, E.; Becker, S. L.; Chen, X.; Cianfrogna,
J.; Doran, A. C.; Doran, S. D.; Gibbs, J. P.; Hosea, N.; Liu, J. H.; Nelson, F. R.; Szewc,
M. A.; Van Deusen, J. Drug Metab. Dispos. 2006, 34(9), 1443.
3. GSK CB2 agonist clinical candidate GW-842166X is unique in that it is reported
to produce analgesia in preclinical models despite exceedingly weak CB1
agonist activity. However, investigators at Abbott (Ref. 13) did observe
measurable rat CB1 agonist activity for this compound. Giblin, G. M. P.;
O’Shaughnessy, C. T.; Naylor, A.; Mitchell, W. L.; Eatherton, A. J.; Slingsby, B. P.;
Rawlings, D. A.; Goldsmith, P.; Brown, A. J.; Haslam, C. P.; Clayton, N. M.; Wilson,
A. W.; Chessell, I. P.; Wittington, A. R.; Green, R. J. Med. Chem. 2007, 50, 2597.
4. Yao, B. B.; Hsieh, G. C.; Frost, J. M.; Fan, Y.; Garrison, T. R.; Daza, A. V.; Grayson,
G. K.; Zhu, C. Z.; Pai, M.; Chandran, P.; Salyers, A. K.; Wensink, E. J.; Honore, P.;
Sullivan, J. P.; Dart, M. J.; Meyer, M. D. Br. J. Pharmacol. 2008, 153, 390.
5. Sain, N. M. H.; Liang, A.; Kane, S. A.; Urban, M. O. Neuropharmacology 2009, 57,
1
reader (Perkin Elmer). CV values calculated for
compound (>250 replicates) were 59% and 70%, respectively, for CB1 and
CB2 IP values.
a
positive control
1
7.
8.
9.
Prost, M.; Urbain, M.; Schumer, A.; Houben, C.; Van Meerbeeck, C. Helv. Chim.
Acta 1975, 58, 40.
Dimitrov, V.; Kostova, K.; Genov, M. Tetrahedron Lett. 1996, 37,
6
787.
Racemic amine
3
was resolved into its enantiomers (3a and 3b) by
preparative chiral HPLC (Chiracel-OD, 20% IPA in hexanes containing 0.1%
DEA). 3a and 3b were coupled with a number of different carboxylic acids to
probe the effects of absolute stereochemistry on CB2 agonist activity. In every
example compounds derived from 3a were significantly more potent than
those derived from 3b.
1
1
2
35.