Design, synthesis and structure-activity relationships of (indo-3-yl) heterocyclic derivatives as agonists of the CB1 receptor. Discovery of a clinical candidate

Article abstract of DOI:10.1016/j.bmcl.2011.02.023

We report an expansion of the structure-activity relationship (SAR) of a novel series of indole-3-heterocyclic CB1 receptor agonists. Starting from the potent but poorly soluble lead, 1, a rational approach was taken in order to balance solubility, hERG a

Full text of DOI:10.1016/j.bmcl.2011.02.023

Bioorganic & Medicinal Chemistry Letters 21 (2011) 2541–2546
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design, synthesis and structure–activity relationships of (indo-3-yl)
heterocyclic derivatives as agonists of the CB1 receptor. Discovery of a clinical
candidate
Paul Ratcliffe a, , Julia M. Adam a, , James Baker b, Roberta Bursi c, Robert Campbell b, John K. Clark a,
Jean E. Cottney b, Maureen Deehan b, Anna-Marie Easson a, Daniel Ecker e, Darren Edwards a, Ola Epemolu b,
Louise Evans b, Ruth Fields a, Stuart Francis a, Paul Harradine a, Fiona Jeremiah a, Takao Kiyoi a,
Duncan McArthur a, Angus Morrison a, Paul Passier c, Jack Pick a, Peter G. Schnabel e, Jurgen Schulz a,
Heinz Steinbrede e, Glenn Walker b, Paul Westwood b, Grant Wishart a, Joanna Udo de Haes d
⇑
⇑
a Department of Chemistry, Merck Research Laboratories, MSD Ltd, Newhouse, Lanarkshire, ML1 5SH, UK
b Department of Pharmacology, Merck Research Laboratories, MSD Ltd, Newhouse, Lanarkshire, ML1 5SH, UK
c Department of Pharmacokinetics, Pharmacodynamics and Pharmacometrics, MSD, The Netherlands
d Department of Clinical Pharmacology, MSD, The Netherlands
e (CMO) Profil Institut für Stoffwechselforschung GmbH, Hellersbergstr. 9, 41460 Neuss, FOCUS, Clinical Drug Development GmbH, Stresemannallee 6, 41610 Neuss, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
We report an expansion of the structure–activity relationship (SAR) of a novel series of indole-3-hetero-
cyclic CB1 receptor agonists. Starting from the potent but poorly soluble lead, 1, a rational approach was
taken in order to balance solubility, hERG activity and potency while retaining the desired long duration
of action within the mouse tail flick test. This led to the discovery of compound 38 which successfully
progressed into clinical development.
Received 2 December 2010
Revised 4 February 2011
Accepted 5 February 2011
Available online 12 February 2011
Ó 2011 Elsevier Ltd. All rights reserved.
Keywords:
Cannabinoid
CB1 receptor agonist
GPCR
Clinical candidate
The CB1 cannabinoid receptor is a member of the G-protein
coupled receptor (GPCR) superfamily, which is characterized by se-
ven-transmembrane receptors. The CB1 receptor is located primar-
ily in the central nervous system but is also expressed on
peripheral neurones. Activation of the CB1 receptor has been sug-
gested as a potential strategy for the treatment of pain and several
other diseases including glaucoma, traumatic brain injury and
multiple sclerosis,1 while inhibition of CB1 receptors has been ex-
plored as a strategy for treatment of obesity and addiction.2 Several
lines of evidence have been reported regarding the analgesic ef-
fects of CB1 agonists in both experimental animal models and clin-
dole-3-carboxamide derivatives as water soluble CB1 agonists,
suitable for intravenous administration as potential post-operative
analgesics, with a fast onset and short duration of action.3 Recently,
we followed up these compounds with a series of indole 3-hetero-
cyclic derivatives culminating in the discovery of compound 1, Ta-
ble 1, found to have an improved duration of action.4
Unfortunately, 1 suffered from poor aqueous solubility an attribute
that appears almost ubiquitous within the CB1 agonist pharmaco-
phore. In this Letter, we report the further optimisation of the in-
dole 3-heterocyclic series and, in particular, our approaches
taken to improve the physicochemical profile.
In an effort to introduce solubility we focussed initially on the
structure–activity relationships established around the N-1 substi-
tuent. The compounds were synthesised according to Scheme 1, or
in the case of compound 1 as previously described.3,4 Starting from
the 7-substituted indole 4 the carboxylic acid was introduced using
trifluoroacetic anhydride followed by aqueous sodium hydroxide.
Treatment of the commercially available methyl tetrahydro-2H-
pyran-4-carboxylate 7 with lithium aluminium hydride furnished
the requisite alcohol 8, which was converted to the tosylate 9 using
ical studies. Moreover, a couple of CB1 agonists including
D
9-THC,
one of the major bioactive components of cannabis, are used clin-
ically as antiemetics in cancer chemotherapy or appetite stimu-
lants in AIDS patients. However, the classical cannabinoids
represented by
D
9-THC are highly lipophilic and the administra-
tion methods are still limited. We have previously described in-
⇑
Corresponding authors.
E-mail address: Paul.Ratcliffe@grunenthal.com (P. Ratcliffe).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.02.023

2542
P. Ratcliffe et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2541–2546
Table 1
Table 2
CB1 activities and Solkin solubility comparisons for THP substituted indole-3-
heterocycles
SAR around indole 3-thiazole
R2
N
H
b
Entry Compound
pEC50
Coeff. Var. (%) Solkin7 (mg LÀ1
)
S
N
S
N
N
N
N
1a
8.4
1.9
<1
R1
N
O
Entry R1 Amine pEC50
Coeff. Var. (%) Solkin (mg LÀ1
)
hERG pKi
b
S
N
N
3a
Et
Et
F
7.1
7.8
6.2c
7.1c
6.4
7.6
4.1
4.0
76
71
68
72
72
66
6.0
N
N
N
N
N
N
N
14a
15
5.9
5.8
6.1
6.1
6.2
2b
8.3
2.3
<1
N
N/A
N/A
6.0
7.5
O
16
F
N
S
17a
18a
Cl
Cl
N
3c
7.1
4.1
76
N
a
Compound was tested as the HCl salt.
pEC50 values are shown as mean and coefficient of variation from n = 2–5
b
separate experiments performed in triplicate (except where indicated with c where
O
n = 1).
a
Compound was tested as the HCl salt.
pEC50 values are shown as mean and coefficient of variation from n = 2–5
b
separate experiments performed in triplicate.
O
O
CF
OH
3
a
b
N
H
N
H
N
H
O
OH
4
5
6
e
N
O
S
O
O
O
OH
O
c
O
d
10
O
O
O
9
7
8
f
N
Cl
S
S
O
S
N
N
NH2
NH2
i
h
g
N
N
N
N
O
O
O
O
3
13
12
11
Scheme 1. Preparation of indole analogs. Reagents and conditions: (a) Trifluoroacetic anhydride, DMF, 0 °C to rt, 1 h; (b) NaOH (4 M, aq), reflux, 1 h, (60%) [over steps a and
b]; (c) lithium aluminium hydride, THF, 0 °C to 45 °C, 1 h, (81%); (d) para toluene sulfonylchloride, pyridine, 0 °C to rt, 1.5 h, (92%); (e) NaH, DMF, 50 °C, o/n, (82%); (f)
oxalylchloride, 0 °C to rt, NH3, K2CO3, DCM, o/n, (87%); (g) P2S5, THF, 55 °C, 3 h; (h) 1,3-dichloroacetone, EtOH, 60 °C, 1.5 h, (48%) [over steps g and h]; (i) pyrrolidine, K2CO3,
MeCN, microwave 150 °C for 600s, (64%).

P. Ratcliffe et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2541–2546
2543
1
1
R
R
CO Et
2
OH
S
S
S
N
N
NH2
a
b
N
N
N
Cl
Cl
O
Cl
Cl
O
O
1
R
CO Et
2
O
12
20; R1 = Cl
23; R1 = Et
24; R1 = Cl
25; R1 = Et
26; R1 = Cl
27; R1 = Et
c
1
1
R
R
3
R
N
Cl
S
S
2
R
N
N
d
N
N
Cl
Cl
O
O
30 to 33
28; R1 = Cl
29; R1 = Et
Scheme 2. Preparation of indole analogs. Reagents and conditions: (a) DMF, microwave 140 °C, 30 min; (b) lithium borohydride, THF, 0 °C for 1 h then 70 °C for 1 h; (c)
methanesulfonylchloride, triethylamine, DCM, 1 h, (26–38%) [over steps a–c]; (d) amine, K2CO3, NaI, acetonitrile, microwave 160 °C, 5 min, (32–93%).
standard conditions of para toluenesulfonyl chloride in pyridine.
Subsequent N-alkylation was achieved via double deprotonation
of the 7-substituted indole carboxylic acid 6 using sodium hydride,
and reaction with (tertahydro-2H-pyran-4-yl)methyl4-methylben-
zenesulfonate 9. Conversion to the primary amide 11 was achieved
via the acid chloride, and subsequent treatment with Lawesson’s
reagent furnished the thioamide 12. Reaction with 1,3-dichloro-
acetone gave the desired chloromethyl thiazole 13,5 which then
left only a simple displacement with the desired amine, for exam-
ple, pyrrolidine, to deliver the final compound.
The prepared compounds were tested for CB1 agonist activity
using CHO cells doubly transfected with human CB1 and luciferase
reporter gene.6 As shown in Table 1, replacement of the cyclohexyl
group with the tetrahydropyranyl (THP) moiety lead to compounds
with good CB1 agonist potency. However, it was clear that the role
of the heterocycle in combination with this more polar group did
have an impact on the intrinsic physical chemical properties of
the molecule, as while the thiadiazole containing compound 2
exhibited <1 mg/L solubility, compound 3 demonstrated excellent
levels of solubility while maintaining good potency. It was for this
reason that we focussed our attention on the thiazole moiety, and
pursued exploration of the structure–activity relationships while
keeping this group constant.
served binding affinity at the hERG channel,8 as exemplified by
compound 15, which demonstrated similar levels of activity at
both receptors.
It is well documented that lipophilic compounds containing a
basic amine are prime candidates for hERG binding.9 A number
of strategies for improving selectivity over hERG have been re-
ported in the literature, ranging from discrete structural modifica-
tions such as; altering the electronics of aryl groups by
incorporation of electron withdrawing groups or removal of elec-
tron donating groups, control of log P, and manipulation of the
pKa of the basic nitrogen.9 In this case, we felt we could target both
improved solubility and selectivity over hERG, through lowering
log P by the introduction of increased polarity in the amine region
of the molecule, hence utilising an analogous route to that de-
scribed in Scheme 1 compounds 34–42 were synthesised.10 In par-
allel with this we also felt that substitution on the thiazole ring,
which had to date remained untouched, should also become an
area of focus. Compounds 30–33 were synthesised from the thio-
amide 12 according to Scheme 2. Ethyl 3,3-dichloropyruvate 20
and ethyl 3-chloro-2-oxopentanoate 23 were prepared as de-
scribed by Garcia-Alles11 and Tsuboi.12 Reaction with the thioam-
ide 12 provided the 5-substituted thiazole 4-carboxylate units,
which were reduced with lithium borohydride to give the primary
alcohols. These were converted to the chloride, using metha-
nesulfonylchloride and triethylamine, which was subsequently
displaced with a range of amines to produce the final compounds
30–33.
With this in mind, we sought to probe changes to the 7-position
on the indole core while also making subtle changes to the amine
moiety, Table 2. In all cases the THP substituted indole 3-thiazoles
retained good levels of solubility. Overall, the 7-substituent ap-
peared to have only a minor effect on CB1 agonist potency, while
replacement of the pyrrolidine with a diethylamine moiety consis-
tently led to improved activity. However, of concern was the ob-
Pleasingly, the introduction of a polar moiety did not lead to a
loss of CB1 activity. Furthermore, open chain amines bearing both
alkoxy and hydroxy substituents were preferred, compounds 35–

2544
P. Ratcliffe et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2541–2546
Table 3
SAR of amine unit and thiazole substitution
R1
R2
N
H
S
N
N
Cl
X
b
Entry
34a
35a
36
X
C
R1
H
Amine
pEC50
7.2c
8.2
Coeff. Var. (%)
Solkin (mg LÀ1
)
hERG pKi
RLM CLint
22
(l
l minÀ1 mgÀ1
)
HLM CLint
(
l
l minÀ1 mgÀ1
)
O
O
N
N/A
3.9
8
<1
65
55
6.6
6
36
N
O
O
H
111
95
OH
HN
H
7.4
5.7
167
<16
H
N
37
38
O
O
H
H
7.9
8.0
4.3
2.9
ND
1
6.1
ND
ND
OH
OH
HN
HN
6.0a
>160
161
39
O
H
7.2
3.8
ND
5.6
16
12
OH
HN
40
O
O
O
H
H
H
6.6
7.3
7.3
5.8
2.3
2.6
1
5.8
5.4
5.6
>270
>270
>270
80.4
97.9
111
O
N
41
3
O
OH
HN
42a
61
OH
OH
30a
31a
O
O
Et
Et
8.2
8.7
0.7
2.2
25
<1
6.2
6.5
ND
79
HN
HN
>270
114
OH
HN
HN
32a
33a
O
O
Et
Cl
8.6
8.0
1.5
1.3
<1
<1
6.6
6.2
ND
ND
OH
OH
>270
>270
a
Compound was tested as the HCl salt.
b
c
pEC50 values are shown as mean and coefficient of variation from n = 2–5 separate experiments performed in triplicate (except where indicated with where n = 1).
These compounds were progressed into the mouse tail flick test
and assayed for both in vivo potency and duration of action (DofA).
All compounds tested demonstrated improved levels of in vivo po-
tency as compared to compound 1 however, of the thiazole con-
taining compounds tested, compound 38 maintained the longest
duration of analgesic efficacy in this test, Table 4.
Table 4
Potency and duration of action of compounds in the mouse tail flick test following iv
administration
Compound
ED50
(
l
mol kgÀ1
)
DofA (min)
1
30
32
35
38
0.8
180
>60
<60
120
150
0.02
0.04
0.08
0.2
The in vitro and in vivo DMPK profile of 38 is summarised below.
As demonstrated in Table 3, compound 38 is very rapidly metabo-
lised in vitro by rat and human hepatic microsomes, which is also
observed in mouse and dog hepatic microsomes (CLint values of
176 and 131 l
l minÀ1 mgÀ1, respectively). The high in vitro intrin-
38, to alkoxy and hydroxyl substituted cyclic amines compounds
39–41, Table 3. Substitution on the 5-position of the thiazole again
led to compounds with good levels of CB1 potency, however, it also
resulted in a trend towards reduced solubility. Importantly, com-
pounds 30, 32, 35 and 38 all showed good CB1 agonist activity
(pEC50 P 8.0), a tolerable level of selectivity over hERG binding
and a measurable level of aqueous solubility.
sic clearance in mouse, rat and dog is in agreement with the ob-
served in vivo clearance, Table 5. However, despite the high
degree of metabolism compound 38 achieves a relatively long plas-
ma half-life, perhaps as a consequence of the extensive volume of
distribution. Importantly, compound 38 demonstrates increased
levels of brain penetration as compared to the lead compound 1
with a Brain/Plasma (B:P) ratio at Cmax of 3.9 and 0.05, respectively.

P. Ratcliffe et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2541–2546
2545
Table 5
Comparison of in vivo pharmacokinetics of compounds 1 and 38
Compound
Species
CL (ml minÀ1 kgÀ1
)
Vss (L/Kg)
T1/2 (min)
B:P ratio (Cmax
)
%ppb
1
38
38
38
Mouse ICR
Mouse ICR
Rat Wistar
Dog (Beagle)
28
159
127
28
1.4
11.3
15
72
83
96
0.05
3.9
ND
ND
99.9
98.9
98.5
99.2
5
180
Table 6
Cytochrome P450 inhibition profile of compound 38
CYP
IC50
1A1
1.4
1A2
2C8
3.3
2C9
2.8
2C19
7.2
2D6
97.8
2E1
3A4 _BQa
>100
3A4_DBb
>100
(lM)
>100
>100
a
BQ—7-Benzyloxyquinoline used as substrate.
DB—Dibenzylfluorescein used as substrate.
b
Table 7
Micro-shake-flask solubility data for compounds 1 and 38 compared with CP 55,94013
Compound
CP 55,940
Buffer
Initial pH
Final pH
Solubility (mg LÀ1
)
Comment
Water
Citrate buffer pH 5.0
Water
Citrate buffer pH 5.0
Water
Citrate buffer pH 5.0
5.3
5.0
5.7
5.0
5.3
5.0
6.2
5.1
2.2
4.7
2.3
4.7
<1
<1
25
<1
>4000
3611
1
38
Over solubility limit Solubility >4 mg mlÀ1
Further investigation of the CYP enzymes involved in the
metabolism of 38 indicates that hepatic metabolism of 38 is med-
iated by CYP3A4. Additionally there is rapid metabolism in vitro by
CYP1A1, an enzyme found in tissues such as the lung and kidney,
but not expressed in the human liver (data not shown). Compound
38 was found to be an inhibitor of the major human P450 enzymes
CYP1A1, 2C8, 2C9, 2C19 and 2D6, Table 6.
Acknowledgements
We would also like to acknowledge the analytical chemistry
group based at Merck Research Laboratories in Newhouse, UK for
their effort during and after the completion of the project.
References and notes
Despite having low solubility at neutral pH, compound 38 had a
solubility of 3611 mg LÀ1 using a pH 5 citrate buffer as the vehicle,
a vast improvement upon 1 and known CB1 agonist CP 55,940 both
of which exhibited <1 mg LÀ1 solubility under the same conditions,
Table 7.
1. Recent reviews for cannabinoid receptor agonists: (a) Adam, J.; Cowley, P.
M.; Kiyoi, T.; Morrison, A. J.; Mort, C. J. W. In Progress in Medicinal
Chemistry; King, F. D., Lawton, G., Eds.; Elsevier, 2006; Vol. 44, pp 207–
329; (b)Handbook of Experimental Pharmacology; Pertwee, R. G., Ed.;
Springer, 2005; Vol. 168, (c) Huffman, J. W.; Padgett, L. W. Curr. Med.
Chem. 2005, 12, 1395.
2. (a) Jagerovic, N.; Fernandez-Fernandez, C.; Goya, P. Curr. Top. Med. Chem. 2008,
8, 205; (b) Bermudez-Silva, F. J.; Viveros, M. P.; McPartland, J. M.; Rodriguez de
Fonseca, F. Pharmacol. Biochem. Behav. 2010, 95, 375.
3. (a) Adam, J.; Cairns, J.; Caulfield, W.; Cowley, P.; Cumming, I.; Easson, M.;
Edwards, D.; Ferguson, M.; Goodwin, R.; Jeremiah, F.; Kiyoi, T.; Mistry, A.; Moir,
E.; Morphy, R.; Tierney, J.; York, M.; Baker, J.; Cottney, J.; Houghton, A.;
Westwood, P.; Walker, G. Med. Chem. Commun. 2010, 1, 54; (b) Kiyoi, T.; et al.
Bioorg. Med. Chem. Lett. 2010, doi:10.1016/j.bmcl.2010.06.067.; (c) Moir, E. M.;
Yoshiizumi, K.; Cairns, J.; Cowley, P.; Ferguson, M.; Jeremiah, F.; Kiyoi, T.;
Morphy, R.; Tierney, J.; Wishart, G.; York, M.; Baker, J.; Cottney, J. E.; Houghton,
A. K.; McPhail, P.; Osprey, A.; Walker, G.; Adam, J. A. Bioorg. Med. Chem. Lett.
2010, 24, 7327.
4. Morrison, A. J.; Adam, J. A.; Baker, J.; Campbell, R.; Clark, J. K.; Cottney, J.;
Deehan, M.; Easson, A. M.; Fields, R.; Francis, S.; Jeremiah, F.; Keddie, N.; Kiyoi,
T.; McArthur, D. R.; Meyer, K.; Ratcliffe, P. D.; Schulz, J.; Wishart, G.; Yoshiizumi,
K. Bioorg. Med. Chem. Lett. 2011, 21, 506.
Furthermore, compound 38 was shown to be potent in the rat
tail flick test with an ED50 of 0.55
analgesic efficacy for 140 min, following an intravenous bolus dose
of 1.1
mol kgÀ1. As a consequence of both the sustained duration
l
mol kgÀ1 and maintained its
l
of action within the rat tail flick test and the excellent solubility at
pH 5, compound 38 was chosen in preference to compound 1 for
subsequent clinical development and pharmacokinetic evaluation
in a Phase 1 clinical trial. The predicted human clearance, volume
of distribution and terminal half-life were calculated14 and re-
sulted in anticipated values of 18 ml minÀ1 kgÀ1, 9.7 L kgÀ1 and
6.1 h, respectively. The human PK prediction worked well, with
the observed figures from the Phase
1 clinical trial being
14 ml minÀ1 kgÀ1, 6.9 L kgÀ1 and 9.6 h, respectively, for a 6
intravenous bolus dose.
l
g kgÀ1
5. Einsiedel, J.; Thomas, C.; Hübner, H.; Gmeiner, P. Bioorg. Med. Chem. Lett. 2000,
10, 2041.
In summary, a series of indole-3-thiazole derivatives was opti-
mised through a systematic SAR study of the indole N-1 position,
thiazole substitution and the amine moiety. These efforts resulted
in novel ligands with good potency, selectivity and in vivo efficacy
and ultimately a clinical candidate. Compound 38 is a potent CB1
receptor agonist, which retained the desired long duration of ac-
tion after intravenous administration in the mouse and rat tail flick
tests, while also demonstrating improved aqueous solubility. Fur-
thermore, the observed human pharmacokinetic parameters for
compound 38 were well predicted from the pre-clinical data using
simple scaling models.
6. Price, M. R.; Baillie, G. L.; Thomas, A.; Stevenson, L. A.; Easson, M.; Goodwin, R.;
McLean, A.; McIntosh, L.; Goodwin, G.; Walker, G.; Westwood, P.; Marrs, J.;
Thomson, F.; Cowley, P.; Christopoulos, A.; Pertwee, R. G.; Ross, R. A. Mol.
Pharmacol. 2005, 68, 1484.
7. Solkin, aqueous solubilities were determined using
a medium-throughput
adaptation of shake-flask methodology. 10 mM solution of the test
a
A
compound in DMSO was added to 0.05 M phosphate buffered saline pH 7.4
such that the final concentration of DMSO was 2%. The resultant mixture
was then vortex mixed (1500 rpm) for 24 0.5 h at 21 2 °C. After mixing,
the resultant solution/suspension was filtered under vacuum using a filter
plate (Millipore Multiscreen HTS, 0.4 lM). The concentration of the
compound in the filtrate was determined by High Performance Liquid
Chromatography (HPLC) running a generic acid gradient method with UV
detection at 230 nm. Peak areas from analysis of the diluted filtrates were

2546
P. Ratcliffe et al. / Bioorg. Med. Chem. Lett. 21 (2011) 2541–2546
quantified by comparison to a calibration line prepared by injecting onto
the HPLC three different volumes of a 50 M solution of the test compound
11. Garcia-Alles, L. F.; Erni, B. Eur. J. Biochem. 2002, 269, 3226.
l
12. Tsuboi, S.; Furutani, H.; Takeda, A. Synthesis 1987, 3, 292.
13. Micro-shake-flask solubility was measured as described in Ref. 3a using Milli-Q
water and 25 mM Citrate buffer, pH 5.0.
14. The clearance in man was estimated using in vitro microsomal stability data
and standard scaling techniques (non-restricted well-stirred model) which
predicted well in pre-clinical species. The volume of distribution in man was
estimated from the mean unbound volume in mouse, rat and dog. Correction
for human plasma protein binding gave the estimate for volume of
distribution.
in DMSO. Solubilities were determined in duplicate for each test compound
and average values reported.
8. Inhibition of [3H]-dofetilide binding to hERG K+ channels stably expressed in
HEK293 cells following a 10 lM dose of test compound. Data represents means
of two experiments performed in duplicate.
9. Jamieson, C. et al J. Med. Chem. 2006, 49, 5029.
10. Compound 38 was synthesised, starting from 7-Cl indole, in an analogous way
to that described in Scheme 1 with an overall yield of 25%.