Discovery and optimization of novel purines as potent and selective CB2 agonists

Article abstract of DOI:10.1016/j.bmcl.2012.06.035

A focused screening strategy identified thienopyrimidine 1 as a hCB2 cannabinoid receptor agonist with moderate selectivity over the hCB1 receptor. This initial hit suffered from poor in vitro metabolic stability and high in vivo clearance. Structure-activity relationships describe the optimization and modification to a less lipophilic purine core. Examples from this novel series were found to be highly potent and fully efficacious agonists of the human CB2 receptor with excellent selectivity against CB1. Compound 10 possesses good biopharmaceutical properties, is highly water soluble and demonstrates robust oral activity in rodent models of joint pain.

Full text of DOI:10.1016/j.bmcl.2012.06.035

Bioorganic & Medicinal Chemistry Letters 22 (2012) 4962–4966
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and optimization of novel purines as potent and selective
CB2 agonists
Sean P. Hollinshead a, , Peter C. Astles b, Mark G. Chambers c, Michael P. Johnson d, John Palmer a,
⇑
Michael W. Tidwell a,
a Discovery Chemistry, Lilly Research Laboratories, A Division of Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA
b Discovery Chemistry, Erl Wood Manor, Lilly Research Laboratories, A Division of Eli Lilly and Company, Sunningdale Road, Windlesham, Surrey GU20 6PH, UK
c Musculoskeletal Research, Lilly Research Laboratories, A Division of Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA
d Discovery Biology, Lilly Research Laboratories, A Division of Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
A focused screening strategy identified thienopyrimidine 1 as a hCB2 cannabinoid receptor agonist with
moderate selectivity over the hCB1 receptor. This initial hit suffered from poor in vitro metabolic stability
and high in vivo clearance. Structure–activity relationships describe the optimization and modification to
a less lipophilic purine core. Examples from this novel series were found to be highly potent and fully
efficacious agonists of the human CB2 receptor with excellent selectivity against CB1. Compound 10 pos-
sesses good biopharmaceutical properties, is highly water soluble and demonstrates robust oral activity
in rodent models of joint pain.
Received 16 May 2012
Revised 7 June 2012
Accepted 11 June 2012
Available online 16 June 2012
Keywords:
CB2 agonist
CB1 selective
Pain
Ó 2012 Elsevier Ltd. All rights reserved.
Purine
Osteoarthritis
The cannabinoid receptors CB1 and CB2 are members of the
superfamily of G protein-coupled receptors (GPCR).1 Cannabis has
been well known for its analgesic effects since ancient times, and
effects of some historical CB2 preferring compounds may actually be
a consequence of residual CB1 agonism (especially given the large
CB1 receptor reserve in the central nervous system and the fact that
weak CB1 binding is able to produce significant agonism5). The lack
of efficacy observed for recent highly selective CB2 agonists (de-
spite high in vivo exposure both centrally and peripherally)6 might
support this hypothesis. In this present Letter we disclose our own
results in this area and disclose the discovery of novel purine com-
pounds which are potent CB2 agonists with no measurable in vitro
CB1 agonist activity and which also demonstrate robust oral activ-
ity in rodent models of joint pain.
Osteoarthritis (OA) is highly prevalent and a leading cause of dis-
ability. Patients face poor quality of life, functional impairment, in-
creased economic burden and reduced independence. Existing drug
therapies reduce painful symptoms but are only moderately effec-
tive over time and have variable risks with respect to GI and cardio-
vascular side effects.7 The CB2 receptor is expressed in nociceptive
nerve fibers on both peripheral nerve endings in the synovium of
skeletal joints and in the dorsal horn of the spinal cord beyond the
blood brain barrier.8 The CB2 receptor is up-regulated at these loca-
tions in painful or inflammatory conditions and it has been shown
that CB2 agonists block the activation of these nerve fibres.9 These
facts along with reports that selective CB2 agonists (e.g.,
GW842166X)3c are efficacious in animal models of inflammatory
pain make this a compelling target for the treatment of OA pain.
the action of the pharmacologically active component (D
9-tetrahy-
drocannabinol) of cannabis is mediated through these cannabinoid
receptors. Non-selective ligands for these receptors exhibit various
biological properties including analgesia, however clinical use has
been limited by severe side effects and the potential for abuse.
The CB1 receptor is expressed mainly in the central nervous system.
The contribution of CB1 and its role in analgesic effects are well doc-
umented.2 However, activation of CB1 is also linked to undesirable
psychotropic effects, sedation, and catalepsy3 which are observed
in non-selective agonists. It has been hypothesized therefore that
a selective CB2 agonist would possess the potential for pain allevi-
ation without the undesired neurological effects of non-selective
endocannabinoid ligands. Indeed, there has been a growing number
of recent reports from various academic and industrial groups on
the development and in vivo evaluation of selective CB2 agonists
and this subject has recently been reviewed.4 However this remains
a complex area and there have been suggestions that the analgesic
⇑
Corresponding author. Tel.: +1 317 277 9146; fax: +1 317 276 6545.
E-mail address: hollinshead_sean_p@lilly.com (S.P. Hollinshead).
Present address: Southwest Research Institute, 6220 Culebra Road, San Antonio,
TX 78238 5166, USA.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.06.035

S. P. Hollinshead et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4962–4966
4963
hCB2 EC50 = 25 nM (85%)
hCB1 EC50 = 2600 nM (64%)
ratio CB1/2 = 104
rat revealed a bioavailability of 91%, however with a high rate of
clearance (CL) at hepatic blood flow (90 ml/min/kg). In vitro
metabolite ID studies revealed extensive oxidative metabolism as
the likely cause. The volume of distribution (Vd) was 10 L/kg as
might be expected from a basic molecule. Encouragingly 1 demon-
strated in vivo efficacy in a rodent monoiodoacetic acid (MIA)
model of osteoarthritic (OA) pain, and significantly reduced pain
compared to vehicle at 10 mg/kg po.16 In addition, this analgesia
was reversed upon treatment of a selective CB2 antagonist SR-
14452817 (dosed at 3 mg/kg ip 30 min prior to 1) indicating that
the effect is mediated by CB2 receptors (data not shown).
Despite the issues described compound 1 was deemed a prom-
ising starting point for optimization with the potential for improv-
ing in vitro CB2 potency, selectivity over CB1 and in vivo clearance
characteristics.
Preliminary SAR observations were made upon modification of
the piperazine (see Table 1). It was established that the corre-
sponding N-methyl analog 2 was about twofold less potent at
CB2 and at CB1 (CB1/2 = 133-fold). Although a basic amine was
not necessary for CB2 activity the 4-methyl piperidine 3 and
N-acetyl piperazine 4 were ꢀ7- to 8-fold less potent than 1. None-
theless, these compounds had improved selectivity over CB1 (>502
and >543-fold, respectively) although the metabolic turnover for 3
remained high (see Table 1). In contrast the N-acetyl compound 4
displayed improved metabolic stability across rat, dog and human
species. However, amine 1 conferred high solubility as a desirable
N
N
hCB2 Ki = 82 nM
rCB2 Ki = 16 nM
PSA = 32
logP (calc/measured) 4.3 / 3.7
microsomal stability: rat 94%; human 73%;
dog 97%
N
Clearance (rat, 1mpk, i.v.) = 90 ml/min/kg
Bioavailability in rat = 91%
S
N
1
Figure 1. Structure and select properties of 1.
As a part of our own research program directed towards the
identification of selective CB2 agonists we performed a focused
medium throughput screen of a GPCR library of 130 K compounds.
This compound set was tested in a panel of GPCR receptor ligand
binding assays including hCB2 and hCB1. We selected actives on
the basis of affinity for hCB2 (Ki<1 lM), >10 X selectivity over
hCB1, a lack of promiscuity against the panel of GPCRs and finally
CNS drug-like properties.10 The actives from this screen were then
confirmed in a functional [35S]-GTP
cS hCB2 binding assay in CHO
or SF9 cell homogenates expressing hCB2 (Perkin Elmer, Boston,
MA) using a scintillation proximity method.11 Selectivity over
CB1 was subsequently determined in a corresponding [35S]-GTP
cS
hCB1 functional assay. These two assays were used to drive the
structure activity relationship determinations described in this
communication. Given the high lipophilicity and low solubility
associated with cannabinoid ligands12 potential hits were also tri-
aged on the basis of possessing good aqueous solubility.
The hits identified included the thienopyrimidine 1 (Fig. 1) pos-
sessing a weakly basic center (logP = 3.7, pKa = 7.5). This displayed
high human and rat affinity as determined from displacement of
[3H]-CP-55,940 from CHO cell homogenates transiently expressing
rat or human CB2 receptors (h and r CB2 Ki = 82 and 16 nM, respec-
tively)13 and potent agonist activity (CB2 EC50 = 24.8 nM).
Compound 1 had moderate selectivity over CB1 (104-fold) and
attribute when compared with N-acetyl compound 4 (>200
and 41 M, respectively, in simulated GI fluid at pH 6.8).
lM
l
The main SAR focus became to add polarity and lower the
lipophilicity in an early attempt to reduce the susceptibility to
microsomal metabolism19 and to improve in vivo clearance. The
initial medicinal chemistry strategy at this point was to identify
other potential core scaffolds as a replacement to the thienopyrim-
idine. This tactic eventually elaborated the hit series to the purine
core, as exemplified initially by 5 (see Table 2). This molecule (5)
had a lower measured logP (3.1) when compared directly with thi-
enopyridine 1 (3.7). The introduction of the ortho-Cl substitution
on the phenyl ring was found to be preferred for maintaining
in vitro CB2 potency. Compound 5 was fivefold more potent (CB2
EC50 = 4.6 nM) than 1 and also had a twofold improvement in
selectivity over CB1 (CB1/2 = 206-fold). Gratifyingly this change
transferred into an improved metabolic profile (7% and 30% metab-
olized in human and rat hepatic microsomes respectively). The
purine 5 exhibited excellent aqueous solubility characteristics
very good solubility (>200 lM, simulated GI fluid pH 6.8). Lipo-
philic basic drugs often possess liabilities14 so it was encouraging
that 1 did not have an affinity for the hERG ion channel (IC50
>10 l
M, as measured in a [3H]-dofetilide displacement assay using
hERG channels stably expressed in HEK293 cells.15
Compound 1 however did inhibit CYP2D6 activity (IC50 = 3 lM)
and showed some binding to 5HT receptors particularly 5HT1
(>2 mg/ml at pH7) and an IC50 >10
l
M against a panel of CYP en-
subtypes (Ki ꢀ1
lM). An assessment of oral pharmacokinetics in
Table 1
Functional [35S]-GTP
c
S assay resultsa and in vitro metabolic stability for compounds 1–4
R
X
N
N
S
N
X
R
Compound
CB2 and CB1 GTP
c
S assays11
Microsomal stability18
hCB2 EC50 (nM)
hCB2 Eff (%)
hCB1 EC50 (nM)
hCB1 Eff (%)
Ratio CB1/2 Human % met. Rat % met. Dog % met.
N
N
C
Et
1
2
3
4
24.8 (9.50, n = 4) 84.5 (14.2, n = 4) 2600 (279, n = 3)
44.4 (15.4, n = 4) 90.6 (4.32, n = 4) 5910 (2160, n = 3)
199 (83.6, n = 4)
184 (87.0, n = 4)
73.6 (3.02, n = 3)
70.9 (0.985, n = 3) 133
104
73
56
49
25
94
95
93
28
97
97
75
62
Me
Me
Ac
85.3 (8.93, n = 4) >100,000 (0, n = 2)b
95.4 (4.17, n = 4) >100,000 (0, n = 3)
>502
>543
N
a
[
35S]-GTP
cS results represented are the mean values ( SEM, and the number of runs n).
b
Note—in one run (n = 3) hCB1 EC50 = 2530 nM (Efficacy = 17.7%).

4964
S. P. Hollinshead et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4962–4966
Table 2
Functional GTPc
S assay resultsa and in vitro metabolic stability for compounds 5–12
R2
N
N
N
Cl
N
N
N
R1
R Groups
Compound
CB2 and CB1 GTP
hCB2 EC50 (nM) hCB2 Eff (%)
4.6 (1.86, n = 5) 90.1 (6.58, n = 5) 948 (120, n = 3)
4.7 (1.00, n = 4) 97.6 (6.79, n = 4) 1430 (700, n = 3) 95.4 (3.44, n = 3) 304
c
S assays11
Microsomal stability18
R1
R2
hCB1 EC50 (nM)
hCB1 Eff (%)
Ratio CB1/2 Human % met. Rat % met. Dog % met.
Me
H
Et
Et
5
6
89.8 (9.01, n = 3) 206
7
—
30
—
88
—
Me
Me
7
4.1 (0.7, n = 3)
93 (4.2, n = 3)
>100,000 (0, n = 3)
>24,390
31
44
98
CH2CH2OH
8
9
8.1 (2.90, n = 4) 96.9 (7.32, n = 4) 6160 (2560, n = 4) 66.2 (6.52, n = 4) 760
13
14
6
25
99
62
CH2CH2NHCO2Me Me
47.5 (2.62, n = 2) 79.5 (0.849, n = 2) >100,000 (0, n = 2)
8.8 (1.98, n = 5) 85.2 (3.30, n = 5) >100,000 (0, n = 3)
>2110
Me 10
>11,360
>4385
31
32
61
58
55
80
O
O
Me 11
22.8 (7.6, n = 4) 83 (10, n = 4)
>100,000 (0, n = 3)
O
Me 12
62.0 (13.3, n = 3) 92.0 (6.28, n = 3) >100,000 (0, n = 3)
>1612
13
28
60
N
a
[
35S]-GTP
cS results represented are the mean values ( SEM, and the number of runs n).
zymes. However some low micromolar cross reactivity at 5HT1
receptors remained. Despite the polarity of the molecule being in-
creased the PSA (50 Å2) was acceptable and compound 5 pene-
trated the CNS after an oral dose with a brain/plasma ratio of 1.
When the purine nitrogen was unsubstituted (R1 = H, 6) a com-
parable CB2 potency and CB1 selectivity was observed (Table 2)
along however with low micromolar cross reactivity at 5HT2
receptors. Having so far been unable to significantly improve the
separation between CB2 and CB1 agonist activity we continued
the investigation of the R1 SAR. The critical and distinguishing
SAR finding was that the in vitro selectivity for CB1 as measured
and 12-fold and the selectivity over CB1 was 760- and >2110-fold,
respectively. With these new R1 substitutions the metabolic pro-
files were largely lowered across species (comparing with cyclo-
propyl analog 7).
We next synthesized the (R)-tetrahydrofuran 10 and (S)-tetra-
hydrofuran 11. Compound 10 was more potent than the
corresponding (S)-isomer 11 (CB2 EC50 = 8.8 and 22.8 nM, respec-
tively). Selectivity against CB1 for both compounds was excellent
(CB1 EC50 >100,000 nM for 10 and 11: CB1/2>11,360 and >4385,
respectively). Binding affinity13 Ki for 10 at hCB2 and rCB2 was
determined to be 20 and 24 nM, respectively (compared with corre-
sponding Ki of 48 and 84 nM for enantiomer 11). It should also be
noted that these molecules were not inverse agonists at CB1 (EC50
>100,000 nM). Compound 10 did not inhibit astemizole binding to
in the functional GTPcS assays could be significantly improved
with a judicious choice of R1 substitution. Indeed some R1 modifi-
cations (see 7, 9, 10, 11 and 12 in Table 2) had no measurable activ-
ity at CB1 (EC50 >100,000 nM) and the ratio of CB1/2 for these
compounds ranged from >1600 to >24,300.
the hERG channel (7% inhibition @ 10
lM) and measured IC50’s at
a panel of CYP enzymes and 5HT receptors were >10
lM. Although
Specifically, when R1 is cyclopropyl (7) excellent CB2 potency
(EC50 = 4.1 nM) and selectivity (CB1 EC50 >100,000 nM; CB1/2
>24,390) was achieved. In addition, the in vitro metabolism for this
analog was determined to be low for human and rat (31 and 44%,
respectively). However, in dog hepatic microsomes the metabolic
turnover remained very high (98%). Other examples of allowed N
substitution include alcohol 8 and carbamate 9, although the
in vitro CB2 potencies (as compared with 5) were reduced by 2-
the measured solubility of 10 was somewhat reduced (0.06 mg/ml
at pH7.4) the compound demonstrated acceptable in vitro micro-
somal stability (Table 2) and oral PK (see Table 3).
Rat PK (1 mg/kg iv) of 10 demonstrated moderate clearance (CL)
of 30 ml/min/kg and oral dosing (1 mg/kg) showed acceptable bio-
availability (27%) and half-life (2.8 h). The brain/plasma ratio was
determined to be 0.8 and the compound is not a substrate for
PgP.20 Dog PK (1 mg/kg iv) of 10 displayed acceptable clearance
Table 3
Pharmacokinetic (PK) properties for compound 10
species
CL (ml/min/kg)
Vd (L/kg)
T1/2 po (h)
AUC (ng h/ml)
Cmax (ng/ml)
Bioavailability (%)
Rat PKa
30
21
5
6.8
2.8
4.2
158
1480
46
368
27
55
Dog PKb
a
Rat (Male Sprague–Dawley) 1 mg/kg iv, 1 mg/kg po.
Dog (Beagle) 1 mg/kg iv, 3 mg/kg po.
b

S. P. Hollinshead et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4962–4966
4965
Cl
N
Cl
Cl
R2
i
ii, iii
O
N
N
N
N
N
N
N
Ar
H
+
N
Cl
N
N
R1
R1
13
14
15
16
R3
N
R3
N
R3
N
.HCl
v
iv
N
N
N
R2
R2
+
N
H
N
N
N
R1
18
N
N
N
R1
19
N
17
Scheme 1. Reagents and conditions: (i) R1-NH2, IPA, heat; (ii) 15% FeCl3 on SiO2, dioxane, heat (Ref. 22); (iii) DDQ, DCM; (iv) amine, EtOH–THF (4:1 v/v), heat; (v) acetone,
1 M aqueous HCl.
Figure 2. Dose dependent effect of CB2 agonist 10 in the standard MIA model.16 All three doses of compound 10 significantly reduced weight bearing deficits (⁄p <0.05 by
Dunnett’s test vs vehicle), and the 0.3 and 1 mg/kg doses were significantly more efficacious than the 0.1 mg/kg (#p <0.05, Tukey–Kramer HSD). Insert: Effect of 3 mg/kg dose
of diclofenac from a separate but highly representative experiment (⁄p <0.05 by Dunnett’s test vs vehicle). Note the similar degree of efficacy seen with 10 and diclofenac.
(CL = 21 ml/min/kg). Oral dosing (3 mg/kg) achieved moderate bio-
availability (55%) and with a half life of 4.2 h.
The general synthetic route for the compounds described in this
communication is shown in Scheme 1.21 Typically the starting

4966
S. P. Hollinshead et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4962–4966
78, 549; (d) Lambert, D. M.; Fowler, C. J. J. Med. Chem. 2005, 48, 1; (e) Pertwee, R.
G. Prog. Neurobiol. 2001, 63, 569.
material 5-amino-4,6-dichloro-2-methyl pyrimidine 13 was re-
acted with an amine (R1-NH2) to afford the mono-amino substi-
tuted intermediates 14. Subsequent reaction with aldehydes
ArCHO 15 was followed by cyclization promoted by 15% FeCl3–
SiO2 according to the procedure of Dang et al.22 This was typically
followed by DDQ oxidation to insure complete conversion to the
desired purines 16. Subsequent reaction with amines 17 gave the
desired products 18. Those analogs containing basic amine func-
tionality could then be converted to the corresponding HCl salt
19 if required.
3. (a) Malan, T. P., Jr.; Ibrahim, M. M.; Lai, J.; Vanderah, T. W.; Makriyannis, A.;
Porreca, F. Curr. Opin. Pharmacol. 2003, 3, 62; (b) Ermann, M.; Riether, D.;
Walker, E. R.; Mushi, I. F.; Jenkins, J. E.; Noya-Marino, B.; Brewer, M. L.; Taylor,
M. G.; Kahrs, A. F.; Ebneth, A.; Lobbe, S.; O’Shea, K.; Shih, D.-T.; Thomson, D.
Bioorg. Med. Chem. Lett. 2008, 18, 1725; (c) 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.; Chessel,
I. P.; Wittington, A. R.; Green, R. J. Med. Chem. 2007, 50, 2597.
4. Riether, D. Expert Opin. Ther. Pat. 2012, 22, 495.
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. (a) Trotter, B. W.; Nanda, K. K.; Burgey, C. S.; 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, 21, 2354; (b)
Manley, P. J.; Zartman, A.; Paone, D. V.; Burgey, C. S.; 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.; Trotter, B. W. Bioorg. Med. Chem. Lett.
2011, 21, 2359.
7. Laufer, S. Rheumatology 2004, 43, i9. and Refs. 3–9 cited therein.
8. Anand, P.; Whiteside, G.; Fowler, C. J.; Hohmann, A. G. Brain Res. Rev. 2009, 60,
255.
9. Anand, U.; Otto, W. R.; Sanchez-Herrera, D.; Facer, P.; Yiangou, Y.; Korchev, Y.;
Birch, R.; Benham, C.; Bountra, C.; Chessell, I. P.; Anand, P. Pain 2008, 138, 667.
10. (a) Gleeson, M. P. J. Med. Chem. 2008, 51, 817; (b) Lipinski, C. A.; Lombardo, F.;
Dominy, B. W.; Feeney, P. J. Adv. Drug Deliv. Rev. 2001, 46, 3.
11. Delapp, N. W.; McKinze, J. H.; Sawyer, B. D.; Vandergriff, A.; Falcone, J.;
McClure, D.; Felder, C. C. J. Pharmacol. Exp. Ther. 1999, 289, 946. Note: testing in
CHO or SF9 membranes expressing the hCB2 receptor was found to be similar
(data not shown).
12. (a) Worm, K.; Dolle, R. E. Curr. Pharm. Des. 2009, 15, 3345; (b) Di Marzo, V. Rev.
Physiol. Biochem. Pharmacol. 2006, 160, 1.
13. Felder, C. C.; Joyce, K. E.; Briley, E. M.; Mansouri, J.; Mackie, K.; Blond, O.; Lai, Y.;
Ma, A. L.; Mitchell, R. L. Mol. Pharmacol. 1995, 48, 443.
14. Leeson, P. D. Nat. Rev. Drug Disc. 2007, 6, 881.
15. Compounds which display a high affinity for hERG ion channels are generally
considered to have a higher risk of cardiotoxicity Snyders, D. J.; Chaudhary, A.
Mol. Pharmacol. 1996, 49, 949.
16. (a) Guingamp, C.; Gegout-Pottie, P.; Philippe, L.; Terlain, B.; Netter, P.; Gillet, P.
Arthritis Rheum. 1997, 40, 1670; (b) Kalbhen, D. A. J. Rheumatology 1987, 14,
130.
17. Rinaldi-Carmona, M.; Barth, F.; Millan, J.; Derocq, J.-M.; Casellas, P.; Congy, C.;
Oustric, D.; Sarran, M.; Bouaboula, M.; Calandra, B.; Portier, M.; Shire, D.;
Brelière, J.-C.; Le Fur, G. L. J. Pharmacol. Exp. Ther. 1998, 284, 644.
18. The method utilized fast gradient elution LC-ESI/MS with column switching to
estimate the percent loss of Phase I metabolism in hepatic microsomes over a
30-min incubation period at 37 °C. Incubations with and without NADPH
(2 mM) were performed in a 96-well format. The reaction was initiated with
addition of substrate and terminated by protein precipitation. All incubations
The analgesic effect of 10 was evaluated in
a rodent
monoiodoacetic acid (MIA) induced model of osteoarthritic (OA)
knee joint pain.16 Oral administration of compound 10 was found
to significantly and dose dependently reduce pain compared to
vehicle and was comparable to efficacy seen with the NSAID dic-
lofenac a standard of care for osteoarthritic pain (see Fig. 2). To as-
sess the in vivo selectivity of 10 we used an established rodent
assay that measures hypothermia induced by central CB1 receptor
activation.23 Oral administration of 10 did not induce hypothermia
in rats up to 30 mg/kg po In contrast the non-selective cannabinoid
agonist CP-55,940 (0.5 mg/kg ip) tested as a positive control,
caused a significant hypothermia thereby confirming the high
24
selectivity for CB1 in vivo.
In summary, from the initial medium throughput screening hit
thienopyrimidine 1 we have discovered a novel purine series of
CB2 receptor agonists with very high selectivity over CB1 recep-
tors. This scaffold was identified from the initial medicinal chemis-
try effort to improve the overall clearance by a reduction of
lipophilicity (logP) of the core. The selectivity for CB2 and meta-
bolic profile could be further optimized by a suitable choice of sub-
stituent at R1. This work culminated in a number of potent CB2
agonists with very high selectivity over CB1. In addition these mol-
ecules have attractive rat and dog PK profiles and have demon-
strated efficacy in rodent models of OA pain. The use of these
highly selective agonists to evaluate the potential role of the CB2
receptor in OA pain and the identification of a clinical candidate
will be reported in due course in future publications.
Acknowledgements
were performed using
a final concentration of 4 lM in 50 mM sodium
phosphate buffer, pH 7.4. The amount of enzyme present was fixed at
1.11 mg/ml protein irrespective of the species of microsomes used. During LC-
ESI/MS only the signal for the protonated molecule (M+H)+ of the substrate was
monitored. The peak area of the NADPH(+) to the NADPH(À) incubations
enabled the percent remaining to be calculated. The reported values for the
percent metabolism is 100 minus the percent remaining.
The authors would like to thank Arunendra Pathak and Somesh
Sharma and their chemistry team at Jubilant Chemsys. The authors
also acknowledge the support of Wendy Gough, Sherri Andis, and
Laura Martin (Eli Lilly and Company) for providing in vitro charac-
terization of the molecules at CB1 and CB2.
19. Obach, R. S.; Lombardo, F.; Waters, N. J. Drug Metab. Dispos. 2008, 36, 1385.
20. Transport of 10 across MDCK-MDR1 cells at 5
l
M: A to B flux = 42 cm/s  10À6
,
B to A flux = 56 cm/s  10À6. Efflux ratio = 1.3. In presence of inhibitor: A to B
flux = 47 cm/s  10À6, B to A flux = 52 cm/s  10À6. Efflux ratio = 1.1.
21. All compounds described gave satisfactory 1H NMR and LC–MS analytical data.
22. Dang, Q.; Brown, B. S.; Erion, M. D. Tetrahedron Lett. 2000, 41, 6559.
23. Rawls, S. M.; Cabassa, J.; Geller, E. B.; Adler, M. W. J. Pharmacol. Exp. Ther. 2002,
301, 963.
References and notes
1. (a) Matsuda, L. A.; Lolait, S. J.; Brownstein, M. J.; Young, A. C.; Bonner, T. I.
Nature 1990, 346, 561; (b) Munro, S.; Thoma, K. L.; Abu-Shaar, M. Nature 1993,
365, 61.
2. (a) Muccioli, G. Chem. Biodivers. 1805, 2007, 4; (b) Mackie, K. Annu. Rev.
Pharmacol. Toxicol. 2006, 46, 101; (c) Demuth, D. G.; Molleman, A. Life Sci. 2006,
24. McGregor, I. S.; Issakidis, C. N.; Prior, G. Pharmacol. Biochem. Behav. 1996, 53,
657.