Selective CB2 receptor agonists. Part 2: Structure-activity relationship studies and optimization of proline-based compounds

Article abstract of DOI:10.1016/j.bmcl.2014.12.019

Through a ligand-based pharmacophore model (S)-proline based compounds were identified as potent cannabinoid receptor 2 (CB2) agonists with high selectivity over the cannabinoid receptor 1 (CB1). Structure-activity relationship investigations for this compound class lead to oxo-proline compounds 21 and 22 which combine an impressive CB1 selectivity profile with good pharmacokinetic properties. In a streptozotocin induced diabetic neuropathy model, 22 demonstrated a dose-dependent reversal of mechanical hyperalgesia.

Full text of DOI:10.1016/j.bmcl.2014.12.019

Bioorganic & Medicinal Chemistry Letters xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Selective CB2 receptor agonists. Part 2: Structure–activity
relationship studies and optimization of proline-based compounds
a
a
a
c
c
Doris Riether b, , Renee Zindell , Lifen Wu , Raj Betageri , James E. Jenkins , Someina Khor ,
Angela K. Berry a, Eugene R. Hickey a, Monika Ermann c, Claudia Albrecht d, Angelo Ceci b,
Mark J. Gemkow d, Nelamangala V. Nagaraja a, Helmut Romig b, Achim Sauer b, David S. Thomson a
⇑
a Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridgebury Road, PO Box 368, Ridgefield, CT 06877-0368, USA
b Boehringer Ingelheim Pharma GmbH & Co. KG, Birkendorfer Strasse 65, 88397 Biberach an der Riß, Germany
c Evotec (UK) Ltd, 114 Innovation Drive, Milton Park, Abingdon, Oxfordshire OX14 4RZ, United Kingdom
d Evotec AG, Manfred Eigen Campus, Essener Bogen 7, 22419 Hamburg, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Through a ligand-based pharmacophore model (S)-proline based compounds were identified as potent
cannabinoid receptor 2 (CB2) agonists with high selectivity over the cannabinoid receptor 1 (CB1). Struc-
ture–activity relationship investigations for this compound class lead to oxo-proline compounds 21 and
22 which combine an impressive CB1 selectivity profile with good pharmacokinetic properties. In a strep-
tozotocin induced diabetic neuropathy model, 22 demonstrated a dose-dependent reversal of mechanical
hyperalgesia.
Received 7 October 2014
Revised 4 December 2014
Accepted 8 December 2014
Available online xxxx
Keywords:
Cannabinoid receptor 2 (CB2)
Proline
Ó 2014 Elsevier Ltd. All rights reserved.
Metabolic stability
Solubility
Pharmacokinetic properties
Since the ancient dynasties of China cannabis has been used for
its analgesic effects. Cannabinoids, such as
9-tetrahydrocannabi-
selectivity profile to ensure an appropriate safety and tolerability
profile in conjunction with good pharmacokinetic properties.
Therefore, we consider novel chemotypes with attractive selectiv-
ity profiles and/or pharmacokinetic properties of interest.
Based on literature-described and in-house CB2 selective scaf-
folds, computer-aided approaches using ligand-based techniques
were utilized to design novel scaffolds.11 This approach has led
to azetidine-based compound 1, proline-based compound 2 and
piperidine-based compound 312 (Fig. 1), all of which demonstrated
to be CB2 selective full agonists, that is, they demonstrate efficacies
greater 90% as compared to CP-55940.13 The initial evaluation of
D
nol, are the pharmacologically active ingredients of cannabis, the
dried leaves of the marijuana plant, and are known to mediate
some of their actions through the cannabinoid receptors 1 (CB1)
and 2 (CB2). These receptors belong to the G-protein coupled
receptor family and were cloned in the early 1990s.1 While the
CB1 receptor is largely associated with psychotropic effects due
to its abundant expression in the peripheral and central nervous
system,2 the CB2 receptor does not mediate psychotropic effects
and human CB2 receptor expression is detected most readily and
constitutively in many immune and inflammatory cell types.3,4 In
addition, constitutive or injury-induced expression of CB2 receptor
has been described for peripheral nerve tissue, including dorsal
root ganglion, and spinal and cerebral microglia.5,6 However,
despite around a decade of research and a high structural diversity
among reported selective small molecule CB2 agonists,7,8 only very
few have entered clinical development.9,10 This suggests that it is
challenging to identify CB2 agonists with the required CB1
H
N
H
N
H
N
O
N
N
N
N
O
O
O
N
N
O
O
N
N
N
CF3
CF3
CF3
1
2
3
CB2 EC50 = 1.15 nM
CB1 EC50 = 1,400 nM
CB2 EC50 = 0.070 nM
CB1 EC50 = 53 nM
CB2 EC50 = 0.093 nM
CB1 EC50 = 250 nM
⇑
Corresponding author. Tel.: +49 7351 54 94921; fax: +49 7351 54 5181.
E-mail address: doris.riether@boehringer-ingelheim.com (D. Riether).
Figure 1. Structures and data of azetidine 1, proline 2 and piperidine 3.
http://dx.doi.org/10.1016/j.bmcl.2014.12.019
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019

2
D. Riether et al. / Bioorg. Med. Chem. Lett. xxx (2014) xxx–xxx
structure–activity relationships (SAR) of these three structural
classes was performed in parallel. As the azetidine-based com-
pounds demonstrated steep SAR (data not shown) it was soon
deprioritized and we focused on the proline- and piperidine based
classes, for which we observed some subtle but also some signifi-
cant differences in the SAR leading us in different structural areas
within our optimization efforts. Therefore, in this manuscript the
proline SAR will be discussed and the piperidine SAR will be dis-
cussed in due course.12
human liver microsomes and a moderate to low stability of 11
and 2 min in rat liver microsomes.18 Based on the higher selectivity
over the CB1 receptor, 4b was also evaluated in an in vivo rat phar-
macokinetic study:19 The low in vitro metabolic stability in rat
microsomes showed to be consistent with the in vivo clearance,
which presumably also led to the low bioavailability of 5% after
oral dosing. The inhibition profile of 4a and b in Cytochrome
P450 (CYP) assessing various isoforms showed no relevant inhibi-
tion of CYP 3A4, 2C9 and 2D6, while inhibition of CYP 2C8 in the
In an assay measuring the inhibition of cyclic adenosine mono-
phosphate (cAMP) production in forskolin stimulated recombinant
CHO cells expressing CB2 and CB1,14 respectively, proline ana-
logues 4a and 4b showed to be two highly potent and fully effica-
cious CB2 agonists with a selectivity over the CB1 receptor of 1300
fold and 13,700 fold, respectively. The binding affinity to the
human CB2 receptor was confirmed by a filtration binding assay
using [3H]-CP-55,940 or [3H]-WIN-55,212-2 as radiolabeled
probes.15 As shown in Table 1, 4b demonstrated 10-fold different
Ki-values (3 and 30 nM, respectively) depending on the probe, sug-
gesting a better overlap of 4b with CP-55,940 within the binding
pocket. It was previously demonstrated by point mutations that
CP-55,940 and WIN-55,212-2 make different contacts in the recep-
tor binding pocket and from there it was concluded that binding
sites of ligands of different structural classes are, if overlapping,
not identical.16 In line with this and the recognized disparity
between the two properties of affinity and efficacy,17 we found
the correlation between the binding assay, using any of the two
probes, and the cAMP functional assay to be poor. In consequence
we decided to use the functional cAMP assay to establish SAR in
our CB2 program.
low lM range was observed. Further, a dramatic difference in
CYP 2C19 inhibition of 4a and b was demonstrated despite a very
subtle structural difference. This led to our decision to only moni-
tor the CYP inhibition profile rather than specifically optimizing for
it.
With the goal to identify highly selective CB2 agonists with
acceptable pharmacokinetic properties to enable proof-of-concept
studies with this structural class in a rat model of neuropathic pain,
we endeavored into our optimization effort with a focus on the two
key ADME issues microsomal stability and solubility. Unfortu-
nately, metabolite ID studies of 4b demonstrated that both aro-
matic substituents as well as the core are metabolically labile,
without identifying the exact metabolites. Therefore we decided
to systematically explore the impact of changes in these three
areas on the ADME parameters starting with the evaluation of ali-
phatic groups as replacements for the aromatic N-substituent.
Compounds were synthesized20 by converting commercially
available (S)-proline to the N-substituted acid by applying stan-
dard Cu-catalyzed N-arylation chemistry to introduce aromatic
R1 (as for the syntheses of 4a and b), or reductive aminations to
deliver the precursors of compounds 5–18 in Tables 2 and 3 con-
taining aliphatic R1. Classic peptide coupling conditions such as
PyCloP, PyBroP or POCl3 provided the final compounds (Scheme 1).
As shown in Table 2, the cyclic aliphatic groups (analogues 5–
10) in the presence of the 3-amino-5-tert-butyl isoxazole amide
demonstrated an apparent dependence of CB2 potency on the size
of the N-substituent, that is, cyclohexyl compound 5 was 20-fold
more potent than cyclopentyl analog 6 and 90-fold more potent
than cyclobutyl analog 7. While the microsomal stability of all
three compounds was very low, solubility was dramatically
The potency and selectivity profiles of 4a and b made them
attractive starting points for optimization and we profiled them
for ADME properties. As shown in Table 1, these two compounds
have low solubility, a moderate stability of 110 and 24 min in
Table 1
Profile of 4a and b
H
N
improved to >70 lg/mL. Introduction of a methylene linker as in
N
N
O
8 was tolerated for potency and selectivity, but did also not impact
the metabolic stability. Further tolerated was the introduction of a
heteroatom in the R1-ring as in 9 and 10. We therefore concluded
O
R1
Example
Table 2
Cl
4b
3
30
0.19
2600 (13,700
fold)
2.8
24/18
CF3
4a
nd
nd
0.23
307 (1300
fold)
4.4
SAR: Evaluation of the N-substituent R1
CB2 Ki ([3H]-WIN-55,212-2) (nM)
CB2 Ki ([3H]-CP-55,940) (nM)
CB2 EC50 (nM)
H
N
N
N
O
O
R1
CB1 EC50 (nM) (selectivity ratio)
Example
R1
CB2 EC50
(nM)
CB1 EC50 (nM)
(sel. ratio)
HLM
t1/2 (min)
Sol.
(lg/mL)
Solubility (
l
g/mL) at pH7.4
Caco-2 permeability A–B/B–A
nd
(10À6 cm/s)
390
(1950 fold)
CYP 3A4 IC50
CYP 2C8 IC50
CYP 2C9 IC50
CYP 2C19 IC50
CYP 2D6 IC50
Human liver microsomal stability
t1/2 (min)
Rat liver microsomal stability t1/2 (min)
Rat PK (iv) clearance (mL/min/kg)
Rat PK (iv) VSS (L/kg)
(
l
M)
M)
M)
>50
2.4
>50
28
18
110
>50
3.1
46
0.63
>50
24
5
0.20
4
>97
(l
l
2400
(570 fold)
9800
(510 fold)
(
6
7
8
4.2
19
4
4
7
71
(l
M)
(l
M)
>100
>110
10,100
(1300 fold)
7.7
11
nd
nd
nd
nd
2
127
7.9
1.0
5
41,000
(17,000 fold)
9
2.4
1.1
7
>100
93
O
Rat PK (iv) MRTdisp (h)
Rat PK (po) F (%)
4900
(4450 fold)
10
13
nd = not determined, rat pharmacokinetic profile: dosed at 1 mg/kg iv (solution)
and 5 mg/kg po (suspension).
O
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019

D. Riether et al. / Bioorg. Med. Chem. Lett. xxx (2014) xxx–xxx
3
Table 3
Table 4
SAR: Evaluation of R2
SAR: evaluation of the N-substituent R1
H
N
HO
R2
H
N
N
O
O
N
O
N
O
R1
Example
R1
Relative
CB2 EC50
(nM)
CB1 EC50 (nM)
(sel. ratio)
HLM t1/2
(min)
Example
R2
CB2 EC50 (nM)
CB1 EC50 (nM)
(sel. ratio)
configuration
>50,000
(>70 fold)
17
cis
700
nd
7200
(2600 fold)
O
O
11
12
13
14
2.8
59
O
N
N
N
N
>50,000
(>7100 fold)
18
19
trans
trans
7.0
4.7
48
88
S
>50,000
(>850 fold)
29,000
(6200 fold)
N
>20,000
(>200 fold)
100
4.7
S
N
28,000
(1600 fold)
20
trans
17
35
17,000
(3600 fold)
Cl
Cl
S
S
N
>20,000
(>150 fold)
15
16
130
170
Cl
Cl
Table 5
N
>50,000
(>300 fold)
Pharmacokinetic profile of 18 in Wistar rat
N
H
18
iv rat PK clearance (mL/min/kg)
iv rat PK VSS (L/kg)
iv rat PK MRT (h)
63
1.3
0.34
18
H
R2
N
(i)
(ii)
OH
OH
po rat PK (%F)
N
H
N
R1
N
R1
O
O
O
Rat pharmacokinetic profile: dosed at 1.1
(suspension).
l
mol/kg iv and 10 lmol/kg po
Scheme 1. Reagents and conditions: (i) aryl bromide, K2CO3, copper iodide, DMF;
or aldehyde, acetic acid, Na2SO4, dichloroethane, then NaBH(OAc)3; (ii) DIPEA,
dichloromethane, amine, PyCloP or PyBroP.
compounds with R1 being alkyl (e.g., cyclohexyl) a dramatic
improvement in microsomal stability was observed between
hydroxylated and non-hydroxylated compounds (19 vs 5). Also
compounds with aromatic (e.g., 20) or heterocyclic (e.g., 18) groups
as R1 demonstrated an improved stability in human liver micro-
somes compared to non-hydroxylated compounds (e.g., 4b and
10).
Compound 18 demonstrated a half-life of 9 min in rat liver
microsomes which was some improvement over compound 4b
(RLM t1/2 = 2 min) and therefore 18 was further evaluated in an
in vivo rat pharmacokinetic study (Table 5).
The improved rat pharmacokinetic profile in terms of both,
clearance and bioavailability) of 18 compared to 4b affirmed us
in our strategy to address the metabolic liability, and led us to look
for additional polar core modifications. Based on the hypothesis
that the position next to the nitrogen is susceptible to oxidative
metabolism, we tested blocking this position with a carbonyl
group. Oxoproline compounds 21–25 and 28–30 in Tables 6 and
7 are synthesized in a similar fashion as described above:20
that although reducing lipophilicity (clogP values of 9 and 10 were
1.7 and 2.4, respectively, compared to 4.5 in 4b or 3.5 in 5) signif-
icantly improves solubility, however exhibits limited impact on the
rate of metabolism. On the other hand, a dramatic impact on the
selectivity ratio was observed for 9 with the most polar substituent
(i.e., tetrahydropyrane) suggesting, that the ability of the CB2 and
CB1 receptors to tolerate polarity differs.
Next, we shifted our focus to the amide substituent (Table 3).
Various five-membered heteroaryl rings bearing a tert-butyl group
in the 3-position were evaluated. The structure–activity relation-
ship demonstrated to be very steep and that both potency and
selectivity are highly dependent on the nature of the heterocycle
(only some examples shown). The two isoxazole isomers 9 and
11 were clearly the most potent five-membered heteroaryl groups
as R2. We also explored a small series of 5,6-bicyclic heteroaryls as
R2 and found chloro-substituted benzothiazoles 14 and 15 and
benzimidazole 16 to be CB2 agonists, 14 being the only bicyclic
R2-substituent to result in a similar potency and selectivity range
as 9 and 11. While all compounds in Table 3 maintained compara-
ble solubility to 9, the half-lives of 11, 14–16 in human liver micro-
somes also remained low (13–16 min). Signs of improved
microsomal stability were however observed with less potent thia-
diazole compounds 12 and 13.
(L)-pyroglutamic acid and aryl boronic acids or esters are converted
in the presence of CuTMEDA to the N-aryl substituted intermedi-
ates. Classic peptide coupling conditions are applied to introduce
R2 to provide the final compounds. For the syntheses of compounds
26 and 27, (L)-pyroglutamic ester is used in an aromatic substitu-
tion using NaH and substituted 2-fluoropyridyls, followed by
hydrolysis and peptide coupling (Scheme 2).
The third area of exploration was the proline core. First, we
attempted to introduce more polarity in the core by substituting
it with a hydroxyl group (see Table 4). We observed a significant
potency difference between trans- and cis-hydroxyproline. While
the cis-hydroxyproline compound 17 lost around 500-fold in CB2
potency compared to the non-hydroxylated analog 10, trans-
hydroxyproline 18 was only around 5-fold less potent than 10. In
Our initial finding for the oxo-proline core was that unexpect-
edly aliphatic groups are not tolerated as R1, for example, the
oxo-version of compound 9 had no CB2 activity up to 1 lM.
Therefore mainly aromatic R1-substituents were explored despite
having a higher clogP. The data for some selected oxo-proline com-
pounds are displayed in Table 6. The para-substituted chloro- and
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019

4
D. Riether et al. / Bioorg. Med. Chem. Lett. xxx (2014) xxx–xxx
Table 6
Table 7
Oxo-proline SAR
SAR: Evaluation of the N-substituent R1
H
N
Example
CB2 EC50
(nM)
CB1 EC50 (nM)
(sel. ratio)
O
N
O
N
O
R1
H
N
S
O
N
N
N
N
Example
R1
CB2 EC50
(nM)
CB1 EC50
(nM)
(sel. ratio)
HLM t1/2
(min)
Sol. @pH7.4
(lg/mL)
O
O
>20,000
(>4 fold)
28
29
30
5000
Cl
N
H
N
>50,000
(>8200 fold)
O
21
22
23
6.1
1.4
14
52
63
O
O
O
Cl
3200
(270 fold)
12
>50,000
(35,700 fold)
170
99
36
CF3
N
CF3
H
N
OH
O
N
>50,000
(3570 fold)
O
970
(26 fold)
>96
38
F
CF3
H
>50,000
(1500 fold)
24
25
34
91
nd
75
nd
N
N
>20,000
(>5 fold)
O
N
31
32
33
4000
390
O
N
N
O
O
O
>50,000
(>40 fold)
O
1,150
N
S
N
>20,000
(>51 fold)
O
O
N
O
N
O
S
O
N
CF3
N
110
(580 fold)
26
27
0.19
1.2
nd
nd
nd
nd
O
N
N
>20,000
(>9 fold)
2200
O
N O
F
1100
(920 fold)
F
N
OH
O
(i)
O
O
N
H
O
H
CF3-phenyl compounds 21 and 2221 are both very potent at CB2
and displayed exceptionally high selectivity, that is, an EC50 greater
than 50 lM at the CB1 receptor (and a selectivity ratio of >35,700
R2
N
(ii)
O
OH
O
N
N
R1
R1
O
O
(iii, iv)
N
H
fold for 22). The p-fluoro- and p-cyano substituted phenyls 23 and
24 have a slightly attenuated CB2 potency while a polar methyl-
sulfone substituent in compound 25 is not tolerated. Compounds
with a p-trifluoromethyl- or p-cyano-2-pyridyl group as R1 such
as in 26 and 27 demonstrated to be the most potent at CB2 within
this array, however with an impressive loss in selectivity. Collec-
tively, the oxoproline analogues demonstrated an excellent selec-
tivity over CB1, an improved stability in human liver microsomes
and an improved solubility compared to their non-oxo compounds
(e.g., 4a and b).
Encouraged by the profiles of the oxo-proline compounds, the
SAR of amide substituent R2 was re-evaluated with this core (lim-
ited data shown in Table 7). This effort however led to the same
conclusion as for the non-oxo analogs, that is, the tert-butyl substi-
tuted isoxazoles are the most preferred R2. Even heterocycles with
moderate potency in the proline core (e.g., thiadiazole in com-
pound 12, EC50 = 59 nM) are much less potent in combination with
the oxo-proline core (compound 28). Therefore, additional substit-
uents at the isoxazole were evaluated, again with a focus on
decreasing lipophilicity in order to further improve solubility.
However, as shown for 29 and 30 such substituents dramatically
impact selectivity over CB1 again suggesting that polarity impacts
the CB1 and CB2 activity differently.
O
Scheme 2. Reagents and conditions: (i) aryl boronic acid or ester, acetonitrile,
CuTMEDA, DBU; (ii) amine, pyridine, POCl3; or DIPEA, dichloromethane, amine,
PyBroP; (iii) substituted 2-fluoro pyridine, DMF, sodium hydride; (iv) trifluoroacetic
acid.
In analogy to the exploration of piperidine cores12 we also
assessed compounds which bear a carbonyl group within the R1
substituent, that is, urea (compounds 31 and 32) or amide linkers
(compound 33) as. These compounds were prepared20 by starting
from Boc-protected (S)-proline, which is converted to the corre-
sponding amide using POCl3, followed by Boc-deprotection with
HCl. Subsequently, the amine is converted to ureas 31 and 32 using
the appropriate carbonylchloride, or to amide 33 using standard
EEDQ coupling conditions (Scheme 3).
In contrast to the piperidine core, these linkers are not well tol-
erated in the proline-based scaffold demonstrating divergent SAR
between the five- and six-membered core rings.12
The optimization effort of prolines has generated oxo-proline
compounds with an excellent balance of CB2 potency and selectiv-
ity over CB1 in addition to moderate to low clearance in human
liver microsomes and acceptable solubility. Further 21 and 22
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019

D. Riether et al. / Bioorg. Med. Chem. Lett. xxx (2014) xxx–xxx
5
H
N
H
(i, ii)
(iii)
Compound 22
R2
R2
OH
N
N
N
H
N
R1
O
O
O
O
O
120
110
*
**
**
Scheme 3. Reagents and conditions: (i) amine, pyridine, POCl3; (ii) hydrochloric
acid in dioxane; (iii) DMF, carbonylchloride, DIPEA; or EEDQ, toluene, RCOOH.
100
90
80
70
60
Table 8
Profile of compounds 21 and 22
21
22
Solubility (
l
g/mL) at pH7.4 16945
63
11/23
nd
36
Naive Vehicle
3
10
30
Caco-2 permeability A–B/B–A (10À6 cm sÀ1
)
14/7.5
>50
23
CYP 3A4 IC50
CYP 2C8 IC50
CYP 2C9 IC50
(l
M)
M)
M)
Mean SEM. N=7. * p<0.05, ** p<0.01 vs vehicle.
(
(
l
l
nd
17
Figure 2. Effect of compound 22 in the STZ induced diabetic neuropathy model: 22
was administered orally (0.5% Natrosol) at 3, 10 and 30 mg/kg in adult Wistar Han
rats (male, 300 20 g); paw withdrawal threshold was assessed in the Randall–
Selitto test 60 min after dosing.
CYP 2C19 IC50
CYP 2D6 IC50
Human liver microsomal stability t1/2 (min)
Rat liver microsomal stability t1/2 (min)
Rat PK (iv) clearance (mL/min/kg)
Rat PK (iv) VSS (L/kg)
(
l
l
M)
M)
4
(
>50
170
>60
11
3.8
6.0
88
52
15
56
4.5
1.3
44
Rat PK (iv) MRTdisp (h)
Rat PK (po) F (%)
with high selectivity over CB1, such as 22, lack the undesired
effects associated with CB1 activation.
Brain/plasma
Specificity profile: 31 receptors and channels
0.41
all <30%
0.90
nd
Acknowledgments
(% inh. @ 10
lM)
nd = not determined; rat pharmacokinetic profile: 21 dosed at 1 mg/kg iv and
The authors thank Carmen Hummel, Andreas Kremer, Tanja
Scheffold and Sabine Löbbe for their excellent technical assistance.
10 mg/kg po (suspension) and 1
pension) in Wister–Han rats.
lmol/kg iv, and 22 dosed at 10 lmol/kg po (sus-
Supplementary data
demonstrated good Caco-2 permeability and 22 had a CYP inhibi-
tion profile comparable to 4a and improved upon 4b in regards
to CYP 2C8. In a specificity screen, 21 demonstrated no cross-reac-
tivity with any of 31 receptors or channels implying very high
specificity for the CB2 receptor. Based on their superior profiles,
we evaluated 21 and 22 in in vivo rat pharmacokinetic studies
which demonstrated a good translation of the in vitro to the
in vivo clearance and a good correlation between clearance and
bioavailability: The pharmacokinetic profile of compound 22 is
characterized by a low clearance of 11 mL/min/kg, a long MRTdisp
of 6 h and a high bioavailability of 88% (Table 8).
Based on the favorable pharmacokinetic profile of 22, we eval-
uated the antinociceptive effect of this CB2 selective agonist in a
diabetic neuropathy model.22 In this model, diabetes is induced
in rats by an intraperitoneal injection of streptozotocin (STZ,
65 mg/kg), which is highly toxic to pancreatic b-cells that produce
insulin. This leads to hyperglycemia along with neuropathic pain
behavior. Compared to vehicle treated animals, 22 demonstrated
a dose-responsive reversal of mechanical hyperalgesia as mea-
sured by the paw withdrawal threshold in the paw pressure test
(Randall–Selitto method) (Fig. 2). The minimum effective dose of
22 is 3 mg/kg and a full reversal is observed at 30 mg/kg. This com-
pares to the effect of Duloxetine at 10 mg/kg (data not shown). Fur-
ther, at all doses no side effects were observed including CB1
mediated CNS effects such as sedation.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2014.12.
019. These data include MOL files and InChiKeys of the most
important compounds described in this article.
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.; Thomas, K. L.; Abu-Shaar, M. Nature 1993,
365, 61.
2. Mackie, K. Handb. Exp. Pharmacol. 2005, 16, 299.
3. Galiegue, S.; Mary, S.; Marchand, J.; Dussossoy, D.; Carriere, D.; Carayon, P.;
Bouaboula, M.; Shire, D.; Le, F. G.; Casellas, P. Eur. J. Biochem. 1995, 232, 54.
4. Schatz, A. R.; Lee, M.; Condie, R. B.; Pulaski, J. T.; Kaminski, N. E. Toxicol. Appl.
Pharmacol. 1997, 142, 278.
5. Anand, U.; Otto, W. R.; Sanchez-Herrera, D.; Facer, P.; Yiangou, Y.; Korchev, Y.;
Birch, R.; Benham, C.; Bountra, C.; Chessel, I.; Anand, P. Pain 2008, 138, 667.
6. Atwood, B. K.; Mackie, K. Br. J. Pharmacol. 2010, 160, 467.
7. Riether, D. Expert Opin. Ther. Patents 2012, 22, 495.
8. Han, S.; Thatte, J.; Buzard, D. J.; Jones, R. M. J. Med. Chem. 2013, 56, 8224.
9. Ostenfeld, T.; Price, J.; Albanese, M.; Bullman, J.; Guillard, F.; Meyer, I.; Leeson,
R.; Costantin, C.; Ziviani, L.; Nocini, P. F.; Milleri, S. Clin. J. Pain 2011, 27, 668.
10. Hollinshead, S. P.; Tidwell, M. W.; Palmer, J.; Guidetti, R.; Sanderson, A.;
Johnson, M. P.; Chambers, M. G.; Oskins, J.; Stratford, R.; Astles, P. J. Med. Chem.
2013, 56, 5722.
11. Hickey, E. R.; Zindell, R.; Cirillo, P. F.; Wu, L.; Ermann, M.; Berry, A. K.; Thomson,
D. S.; Albrecht, C.; Gemkow, M. J.; Riether, D. Bioorg. Med. Chem. Lett., in press.
12. Bartolozzi, A.; Cirillo, P. F.; Berry, A. K.; Hickey, E. R.; Thomson, D. S.; Wu, L.;
Zindell, R.; Albrecht, C.; Ceci, A.; Gemkow, M. J.; Nagaraja, N. V.; Romig, H.;
Sauer, A.; Riether, D. Bioorg. Med. Chem. Lett., in press.
13. Pertwee, R. G. Int. J. Obes. 2006, 30, S13.
In summary, we established structure–activity relationships
14. (a) CB2 and CB1 cAMP assays: Recombinant CHO cells expressing the human
CB2 or CB1receptor (Euroscreen) were plated at a density of 5000 cells per well
in 384-well plates and incubated overnight at 37 °C. After removing the media,
the cells were treated with test compounds diluted in stimulation buffer
around (S)-proline based compounds that originated from
a
ligand-based approach.11 Through modifications of R1, R2 as well
as the core we were able to identify compounds with good CB2
containing 1 mM IBMX, 0.25% BSA and 10 lM forskolin. The assay was
potency in conjunction with lacking CB1 activity up to 50 lM. At
incubated for 30 min at 37 °C. Cells were lysed and the cAMP concentration
was measured using DiscoveRx XS+ cAMP kit, following the manufacturer’s
protocol. The maximal amount of cAMP produced by forskolin compared to the
level of cAMP inhibited by 200 nM CP-55940 is defined as 100%. The EC50 value
of each test compound was determined as the concentration at which 50% of
the forskolin-stimulated cAMP synthesis was inhibited using a four-parameter
logistic model. (b) Reproducibility of the cAMP assays is assessed using the
the same time solubility, metabolic stability and the pharmacoki-
netic profile could be improved. In vivo efficacy was demonstrated
with compound 22 in a rat model of diabetic neuropathic pain. This
data further supports, that CB2 selective agonists may be effective
for the treatment of diabetic neuropathic pain23 and compounds
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019

6
D. Riether et al. / Bioorg. Med. Chem. Lett. xxx (2014) xxx–xxx
control compound CP-55940 which is tested twice on every assay plate to rule
ratio of area under the curve of reference to sample.
out any plate artifacts. Efficacy at CB1 and CB2 is expressed as a percentage
relative to the efficacy of CP-55940. Each compound is tested in triplicate at
least twice with individual dilutions from the stock. The results reported in the
table are the mean values of the measurements and individual values for the
compounds do not differ by more than a factor of three from the mean.
HLM assay: Test compounds were incubated in Duplicate Matrix MultiScreen
mintubes (Matrix Technologies, Hudson, NH) with liver microsomes
(Xenotech, Lenexa, KS). Each assay is performed in 50 mM potassium
phosphate buffer, pH 7.4, and 2.5 mM NADPH. Compounds were tested at a
final assay concentration of 1.0
mix was 1 mg/mL. Compounds were preincubated for 5 min at 37 °C and the
metabolic reactions were initiated by the addition of NADPH. Aliquots of 80
were removed from the incubation mix at 0, 5 and 30 min after the start of the
reaction for screening data. Each aliquot was added to 160 L acetonitrile for
lM. The protein concentration in the reaction
15. Human CB2 and CB1 receptor binding assays: CB2 and CB1 membranes were
isolated from HEK cells stably transfected with the respective receptors. The
membrane preparation was bound to scintillation beads (Ysi-Poly-Llysine SPA
beads, GE Healthcare) in assay buffer containing 50 mM Tris, pH 7.5, 2.5 mM
EDTA, 5 mM MgCl2, 0.8% fatty acid free Bovine Serum Albumin. Unbound
membrane was removed by washing in assay buffer. Membrane–bead mixture
lL
l
extraction by protein precipitation. These samples were mixed for 1 min by
vortexing, and a volume of the mixture was filtered through wells in 0.25 mm
glass fiber filter plates by centrifugation at 3000 rpm for 5 min. Sample extracts
were analyzed by LC–MS–MS to determine parent compound levels. Percent
loss of parent compound was calculated from the peak area at each time point
to determine the half-life for test compounds (T1/2, min).
was added to 96-well assay plates in the amounts of 2
lg membrane per well
(CB2) or 2.5 g per well (CB1) and 1 mg SPA bead per well. Compounds were
l
added to the membrane–bead mixture in dose-response concentrations
ranging from 1 Â 10À5
M
to 1 Â 10À10 M. The competition reaction was
initiated with the addition of [3H]-CP-55940 (Perkin–Elmer Life and
Analytical Sciences) at a final concentration.
19. Rat pharmacokinetic studies were performed in Wistar Han or Sprague Dawley
rats. Blood samples were taken at several time points post application of the
test compound, anticoagulated and centrifuged. The concentrations of analytes
were quantified in the plasma samples by means of HPLC-MS/MS. The
pharmacokinetic parameters were calculated by non-compartmental methods.
20. Berry, A.; Betageri, R.; Riether, D.; Hickey, E. R.; Wu, L.; Zindell, R. M.; Khor, S.
WO2010077836.
IC50 values were converted to inhibition constant (Ki) values using the Cheng–
Prusoff equation.
16. (a) Gouldson, P.; Calandra, B.; Legoux, P.; Kernéis, A.; Rinaldi-Carmona, M.;
Barth, F.; LeFur, G.; Ferrara, P.; Shire, D. Eur. J. Pharm. 2000, 401, 17; (b) Reggio,
P. H. Curr. Pharm. Des. 2003, 9, 1607; (c) Tuccinardi, T.; Ferrarini, P. L.; Manera,
C.; Ortore, G.; Saccomanni, G.; Martinelli, A. J. Med. Chem. 2006, 49, 984.
17. (a) Griffin, G.; Williams, S.; Aung, M. M.; Razdan, R. K.; Martin, B. R.; Abood, M.
E. Br. J. Pharmacol. 2001, 132, 525; (b) Kenakin, T.; Onaran, O. Trends Pharmacol.
Sci. 2002, 23, 275.
21. Analytical data for 21: ES-MS m/z = 362.6 (M+H+), 1H NMR (400 MHz, DMSO-
d6): 12.10 (s, 1H), 7.52 (d, J = 8 Hz, 2H), 7.43 (d, J = 8 Hz, 2H), 6.22 (s, 1H), 4.98–
5.00 (m, 1H), 2.46–2.66 (m, 3H), 2.09 (m, 1H), 1.22 (s, 9H); for 22: ES-MS m/
z = 396.2 (M+H+), 1H NMR (400 MHz, CDCl3): 8.75 (s, 1H), 7.55–7.70 (m, 4H),
6.3 (s, 1H), 4.85 (m, 1H), 2.75–2.90 (m, 1H), 2.60–2.75 (m, 2H), 2.25–2.40 (m,
1H), 1.3 (s, 9H).
18. Solubility assay; solubility of compounds is measured at pH 4.5 and pH 7.4
buffers. 6
lL of a 10 mM DMSO stock solution is spiked into pH 4.5 and 7.4
buffers targeting 50–200
lM final concentrations in buffers in duplicate deep
22. (a) Malcangio, M.; Tomlinson, D. R. Pain 1998, 76, 151.
well plates. The DMSO content is 0.5%. The samples are incubated for 16–18 h,
filtered and analyzed using spectrophotometer. The UV spectra of samples and
reference are scanned from 230–500 nm. Solubility is measured taking the
23. Ikeda, H.; Ikegami, M.; Kai, M.; Oshawa, M.; Kamei, J. Neuroscience 2013, 250,
446.
Please cite this article in press as: Riether, D.; et al. Bioorg. Med. Chem. Lett. (2014), http://dx.doi.org/10.1016/j.bmcl.2014.12.019