Novel Triazolopyrimidine-Derived Cannabinoid Receptor 2 Agonists as Potential Treatment for Inflammatory Kidney Diseases

Article abstract of DOI:10.1002/cmdc.201500218

The cannabinoid receptor 2 (CB2) system is described to modulate various pathological conditions, including inflammation and fibrosis. A series of new heterocyclic small-molecule CB2 receptor agonists were identified from a high-throughput screen. Lead optimization gave access to novel, highly potent, and selective (over CB1) triazolopyrimidine derivatives. A preliminary structure-activity relationship was established, and physicochemical properties in this compound class were significantly improved toward better solubility, lipophilicity, and microsomal stability. An optimized triazolopyrimidine derivative, (3S)-1-[5-tert-butyl-3-[(1-cyclopropyltetrazol-5-yl)methyl]triazolo[4,5-d]pyrimidin-7-yl]pyrrolidin-3-ol (39), was tested in a kidney ischemia-reperfusion model, in which it showed efficacy at a dose of 10 mg kg^-1 (p.o.). A significant depletion of the three measured kidney markers indicated a protective role of CB2 receptor activation toward inflammatory kidney damage. Compound 39 was also protective in a model of renal fibrosis. Oral treatment with 39 at 3 mg kg^-1 per day significantly decreased the amount of fibrosis by ～40 % which was induced by unilateral ureter obstruction.

Full text of DOI:10.1002/cmdc.201500218

DOI: 10.1002/cmdc.201500218
Communications
Novel Triazolopyrimidine-Derived Cannabinoid Receptor 2
Agonists as Potential Treatment for Inflammatory Kidney
Diseases
Matthias Nettekoven,*[a] Jean-Michel Adam,[a] Stefanie Bendels,[a] Catarina Bissantz,[a]
Jürgen Fingerle,[b] Uwe Grether,[a] Sabine Grüner,[b] Wolfgang Guba,[a] Atsushi Kimbara,[a]
Giorgio Ottaviani,[c] Bernd Püllmann,[a] Mark Rogers-Evans,[a] Stephan Rçver,[a]
Benno Rothenhäusler,[c] Sebastien Schmitt,[a] Franz Schuler,[c] Tanja Schulz-Gasch,[a] and
Christoph Ullmer[b]
Dedicated to Prof. Steven V. Ley on the occasion of his 70th birthday
The cannabinoid receptor 2 (CB2) system is described to mod-
ulate various pathological conditions, including inflammation
and fibrosis. A series of new heterocyclic small-molecule CB2
receptor agonists were identified from a high-throughput
screen. Lead optimization gave access to novel, highly potent,
and selective (over CB1) triazolopyrimidine derivatives. A pre-
liminary structure–activity relationship was established, and
physicochemical properties in this compound class were signif-
icantly improved toward better solubility, lipophilicity, and mi-
crosomal stability. An optimized triazolopyrimidine derivative,
(3S)-1-[5-tert-butyl-3-[(1-cyclopropyltetrazol-5-yl)methyl]triazo-
lo[4,5-d]pyrimidin-7-yl]pyrrolidin-3-ol (39), was tested in
a kidney ischemia–reperfusion model, in which it showed effi-
cacy at a dose of 10 mgkgÀ1 (p.o.). A significant depletion of
the three measured kidney markers indicated a protective role
of CB2 receptor activation toward inflammatory kidney
damage. Compound 39 was also protective in a model of
renal fibrosis. Oral treatment with 39 at 3 mgkgÀ1 per day sig-
nificantly decreased the amount of fibrosis by ~40% which
was induced by unilateral ureter obstruction.
this protective mechanism is in lipid signaling through activa-
tion of the cannabinoid 2 (CB2) receptor-mediated pathway.[1]
During the insult a rapid increase in local endocannabinoid
levels causes a sudden modulation of various signaling path-
ways in immune and other cell types. Endocannabinoids, endo-
cannabinoid-like, and synthetic CB2 receptor agonists positive-
ly affect a large number of pathological conditions, including
cardiovascular,[2] gastrointestinal,[3] liver,[4] kidney,[5] lung,[6] neu-
rodegenerative[7] and psychiatric[8] disorders, as well as pain,[9]
cancer,[10] bone,[11] reproductive system,[12] and skin patholo-
gies.[13] Typical endogenous CB2 receptor agonists are com-
pounds such as anandamide (AEA)[14] and 2-arachidonoylgly-
cerol (2-AG),[15] as well as plant-derived D9-tetrahydrocannabi-
nol ((À)-D9-THC)[14] and exogenous cannabinoid-type agonists
such as HU-308,[14] JWH-133,[16] and HU-910.[4] Non-cannabi-
noid-type agonists have been described in several recent over-
views.[17] All of these CB2 receptor agonists possess various de-
grees of potency and selectivity versus the cannabinoid 1
(CB1) receptor.[18]
With the goal to identify new selective CB2 receptor agonist
hit structures, a high-throughput screen was performed. Vari-
ous hit clusters were identified by an automated analysis of
the hit list with a combination of fingerprint and maximum
common substructure methods.[19] The analysis was based on
agonistic potency at the CB2 receptor (cAMP assay) and selec-
tivity versus the CB1 receptor (cAMP assay) within a cluster of
similar compounds. A cluster with seven members of heterocy-
clic compounds was identified for hit evaluation and lead opti-
mization. Four typical derivatives with their preliminary profiles
are shown in Table 1.
Several protective mechanisms in mammalian tissues have de-
veloped over the course of evolution to prevent and limit inju-
ries caused by various types of insults. An important part of
[a] Dr. M. Nettekoven, Dr. J.-M. Adam, Dr. S. Bendels, Dr. C. Bissantz,
Dr. U. Grether, Dr. W. Guba, Dr. A. Kimbara, B. Püllmann,
Dr. M. Rogers-Evans, Dr. S. Rçver, S. Schmitt, Dr. T. Schulz-Gasch
Roche Pharmaceutical Research and Early Development
Small-Molecule Research, Roche Innovation Center Basel
Grenzacher Str. 124, 4070 Basel (Switzerland)
The most potent compound from the seven members in
this cluster was triazolopyrimidine 1 (IUPAC atom numbering),
with an EC50 value of 9 nm at the CB2 receptor as measured by
cAMP assay,[20] a full agonist in this assay with high selectivity
versus CB1 (EC50 >10 mm) in the respective cAMP assay. CB2 b-
arrestin data were also determined for selected compounds.[21]
The ability of CB2 agonists to recruit b-arrestin ultimately leads
to the steric inhibition of G-protein coupling and termination
of signaling-independent of G-protein classes. Partial agonists
might constitutively stimulate receptor activation without
causing b-arrestin recruitment. This might be considered a cru-
E-mail: matthias.nettekoven@roche.com
[b] Dr. J. Fingerle, Dr. S. Grüner, Dr. C. Ullmer
Roche Pharmaceutical Research and Early Development
Discovery Biology, Roche Innovation Center Basel
Grenzacher Str. 124, 4070 Basel (Switzerland)
[c] Dr. G. Ottaviani, Dr. B. Rothenhäusler, Dr. F. Schuler
Roche Pharmaceutical Research and Early Development
DMPK, Roche Innovation Center Basel
Grenzacher Str. 124, 4070 Basel (Switzerland)
This article is part of a Special Issue on Drug Discovery in the Field of
Autoimmune and Inflammatory Disorders. The complete issue can be
browsed via Wiley Online Library.
ChemMedChem 2016, 11, 179 – 189
179
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
tion of physicochemical proper-
ties as well as enhancement of
binding affinity. As described
below, this pocket is also instru-
mental for optimizing selectivity
Table 1. Four representative derivatives of a cluster identified through HTS.
toward
the
CB1
receptor
(Figure 1).
This analysis of computational
properties in combination with
fitting of the compounds of this
cluster into the CB2 homology
model supports our confidence
in this hit cluster in conjunction
with CB2 functional activity, and
allowed the start of a chemical
optimization program. Starting
with triazolopyrimidine 1, the
most active compound of this
Compd:
1
2
3
4
hCB2 cAMP EC50 [mm]
Efficacy [%]
hCB1 cAMP EC50 [mm]
hCB2 b-arrestin EC50 [mm]
Efficacy [%]
0.009
100
>10
0.08
100
3.6
100
0.8
73
>10
0.64
49
0.26
77
>10
ND[a]
49
>10
ND[a]
ND[a]
3.85
310.4
AlogP
Mr [Da]
4.2
386.9
4.12
281.4
3.5
315.7
[a] Not determined.
cial parameter for the translation of in vitro potency into in
vivo efficacy.[22] Triazolopyrimidine 1 was active with EC50 =
80 nm in the human CB2 b-arrestin assay with 100% efficacy.
Calculated lipophilicity (AlogP[23]) and molecular weight values
(Mr) served to determine the potential for optimization of mo-
lecular properties by diversification. Triazolopyrimidine
1 showed an AlogP value of 4.2, with Mr <400 Da, and at least
three vectors available for diversification. Close analogues 2–4
were of even lower molecular weight with similar AlogP
values.[24]
A 3D homology model of CB2 was built in MOE using the
published X-ray crystal structure of activated bovine rhodopsin
as a template.[25] Compounds 1–4 were manually docked into
the binding site and superimposed to rationalize the SAR
within the series and to guide optimization efforts. As shown
in Figure 1, substituents at positions 3 and 5 are in contact
with the external loops region. The proposed binding mode in
the homology models suggests that the R5 pocket should be
able to accommodate small and branched linear as well as
cyclic alkyl substituents. The R5 pocket is defined by aromatic
and aliphatic amino acid side chains; therefore, hydrogen
bond donors or acceptors are not expected to contribute ben-
eficially to CB2 activity. The R3 pocket is predominantly hydro-
phobic as well; however, the pocket size is significantly larger
than R5. The orthogonal orientation of delineating residues
forms an aromatic box, which is optimally suited for accommo-
dating aromatic substituents. For an optimal placement within
the R3 pocket, linkers such as CH2 are required to project aro-
matic cores with an exit vector of 908 in the back of the
pocket. The pocket size also allows further small substituents
at the R5 aromatic rings. In contrast to R3 and R5, substituents
at position 7 extend downward into the transmembrane
region. The volume of the pocket is sufficiently large for up to
six-membered aliphatic rings. Besides the orientation of the R7
pocket deep in the transmembrane core region, this pocket
also differs from R3 and R5 by two additional polar amino
acids: serine and threonine. Thus, the proper placement of hy-
drogen bond donors or acceptors should enable the modula-
Figure 1. Manual docking of compound 1 into the binding site of the CB2
homology model. Protein residues and nonpolar hydrogen atoms of the
ligand are omitted for clarity. The N-terminal loop region is located at the
top.
cluster, three vectors of diversification were investigated. The
nature and role of the substitution at position 5 was studied
regarding its potential to influence the activation of the CB2
receptor, selectivity versus the CB1 receptor, and molecular
properties. In the early optimization phase, lipophilicity (calcu-
lated with AlogP[23]), aqueous solubility in phosphate buffer
pH 6.5 (LYSA[26]), and liver microsomal clearance (human and
mouse)[27] of tested compounds were systematically assessed
to improve drug-likeness and pharmacokinetics (PK) via early
ADME optimization.[27] To decrease the lipophilicity, the aroma-
ticity load of the final compounds was kept low by synthesiz-
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
180
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
only triazolopyrimidine
8 was
found to retain the excellent se-
lectivity against CB1 in the cAMP
assay. The activity in the human
CB2 b-arrestin assay remained
very similar. Molecular properties
(AlogP, LYSA) for all derivatives
5–8 were very similar, and only
the human and mouse microso-
mal stabilities differentiated this
compound set. Clearance in
human and mouse microsomes
increased for 6, 7, and 8 relative
to 5 (particularly in mouse mi-
crosomes); however, compound
5 is more stable than the origi-
nal triazolopyrimidine 1, as an-
Scheme 1. Synthesis of triazolopyrimidine derivatives 5–8. Reagents and conditions: a) for 11: 10+tert-butylcar-
bonyl chloride, pyridine, 808C, 3 h; for 12: 10+isopropanecarbonyl chloride, pyridine, 808C, 3 h; for 13: 9+cyclo-
propanecarbonyl chloride, pyridine, 808C, 3 h; for 14: 10+cyclobutanecarbonyl chloride, pyridine, 808C, 5 h;
b) for 15: 11 +KHCO3(aq), H2O, reflux; for 16: 12+KHCO3(aq), H2O, reflux; for 17: 13+KHCO3(aq), H2O, reflux; for 18:
14+KHCO3(aq), H2O, reflux; c) for 5: 1. 15+POCl3, reflux, 2. difluoropyrrolidine hydrochloride, DIPEA; for 6:
1. 16+POCl3, reflux, 2. difluoropyrrolidine hydrochloride, DIPEA; for 7: 17+POCl3, reflux, 2. difluoropyrrolidine hy-
drochloride, 3. TFA, reflux, 4. 1-(bromomethyl)-2-chlorobenzene, DBU; for 8: 1. 18+POCl3, reflux, 2. difluoropyrroli-
dine hydrochloride.
ing analogues with alkyl and
Table 2. Triazolopyrimidine derivatives 5–8 variously substituted at position 5.
(substituted) cycloalkyl residues
only. Initial profiling of triazolo-
pyrimidine 1 in terms of micro-
somal stability revealed medium
stability in human (h) and
mouse
(m)
microsomes
(CLint [mLminÀ1 kgÀ1] (h/m): 34/
62). Therefore, for initial studies
of the role of the substituent at
the 5-position, a generally more
stable amine at the 7-position
was chosen, that is, 3,3-difluoro-
pyrrolidine. The synthesis of rep-
resentative examples is shown in
Scheme 1. The synthetic strategy
focused on the versatility of the
approach, and two pathways are
exemplified. Triazoles 9 and 10
Compd:
5
6
7
8
hCB2 cAMP EC50 [mm]
Efficacy [%]
hCB1 cAMP EC50 [mm]
hCB2 b-arrestin EC50 [mm]
Efficacy [%]
0.0002
99
0.165
0.023
88
5.08
<1
10/43
0.005
100
0.774
0.011
66
4.66
<1
18/189
0.001
97
0.749
0.012
60
4.29
<1
42/330
0.001
99
>10
0.015
51
4.74
<1
29/164
AlogP
LYSA [mgmLÀ1
]
CLint [mLminÀ1 kgÀ1] (h/m)
are commercially available and can react conveniently with de-
sired acyl chlorides under basic conditions to access intermedi-
ates 11–14 (compounds were typically used crude). Cyclization
of 11–14 occurred under basic conditions to give compounds
15–18 (compounds either used crude or yields recorded be-
tween 14 and 41%). Derivatization of intermediates 15, 16,
and 18 was carried out by chlorination with POCl3 and subse-
quent nucleophilic substitution to access triazolopyrimidines 5,
6, and 8, respectively. For the synthesis of triazolopyrimidine
derivative 7, intermediate 17 was first chlorinated and substi-
tuted with 3,3-difluoropyrrolidine, the 4-methoxybenzyl group
of the intermediate was cleaved with trifluoroacetic acid (TFA),
and subsequent derivatization with 2-chlorobenzyl bromide
gave access to derivative 7 (Scheme 1).
The compounds were tested in functional CB2 and CB1
assays, and their preliminary molecular property profiles were
assessed and compared with those of the parent, tert-butyltria-
zolopyrimidine derivative 5 (Table 2). Derivatives 5–8 all had
nanomolar activity in the CB2 cAMP assay, similar to (5) or
even better than (6, 7, 8) that of triazolopyrimidine 1, whereas
ticipated. This demonstrated that isopropyl, cyclopropyl, and
cyclobutyl (amongst other variations at position 5; data not
shown) are not viable alternatives to the originally identified
tert-butyl group. Therefore, optimization of activities and prop-
erties of the triazolopyrimidine class was studied with the tert-
butyl group fixed at the 5-position. The primary focus was
then turned to substitution at position 7 with ethers and
amines. Substitution of intermediate 15 (obtained through re-
action with POCl3) with various alcohols under basic conditions
yielded a set of substituted triazolopyrimidines, of which a se-
lection (19–22) of the most potent derivatives is listed in
Table 3.
The compounds were tested in functional CB2 and CB1
assays, and their preliminary molecular property profiles were
assessed. In general, triazolopyrimidines with an ether substitu-
tion at the 7-position were all active in the nanomolar range,
displaying full efficacy toward the hCB2 receptor coupled with
good selectivity versus the hCB1 receptor (EC50 >10 mm). EC50
values for b-arrestin for 19–22 were found <1 mm, with partial
efficacy of ~70%. For these most potent derivatives, AlogP
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
181
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
properties to those of the ali-
phatic primary amine-substitut-
ed derivatives: that is, highly
potent at the CB2 receptor with
very good selectivity versus CB1,
AlogP >5, low solubility, and
high clearance in microsomes.
The ring size, however, in triazo-
lopyrimidine derivatives substi-
tuted with cycloaliphatic amines
was found to have a significant
impact on potency, with a five-
membered ring (compound 27)
being active at CB2 in the sub-
nanomolar range, whereas com-
pounds with four- (26) and six-
membered rings (28) were gen-
Table 3. Triazolopyrimidine derivatives 19–22 substituted with ethers at position 7.
Compd:
19
20
21
22
hCB2 cAMP EC50 [mm]
Efficacy [%]
0.017
97
0.01
98
0.008
97
0.022
98
>10
0.61
72
5.74
<1
60/303
hCB1 cAMP EC50 [mm]
hCB2 b-arrestin EC50 [mm]
Efficacy [%]
>10
0.31
61
>10
0.349
61
>10
0.746
72
AlogP
5.15
<11
15/703
5.06
<1
140/1000
5.72
<1
14/149
LYSA [mgmLÀ1
]
CLint [mLminÀ1 kgÀ1] (h/m)
values were not improved and were calculated to be >5. They
subsequently showed low aqueous solubility (LYSA:
<1 mgmLÀ1). The microsomal clearance in human and mouse
microsomes was higher than the nitrogen-substituted triazolo-
pyrimidine 5 (cf. Table 2), and therefore optimization of the
ether derivatives was not pursued further.
erally 10-fold less active. Activity at the CB1 receptor for these
derivatives was also affected. Derivatives 26–28 showed activi-
ty in the high-nanomolar/low-micromolar range. The selectivity
against the CB1 receptor remains very high (e.g., the ratio
[hCB1 cAMP EC50]/[hCB2 cAMP EC50] for 27 is >2100).
b-Arrestin values were measured only for derivatives 26 and
27, and for both compounds activities of ~100 nm with partial
efficacy (~76%) were observed. The molecular properties of
these compounds were not significantly improved, with one
main issue being high clearance in human and mouse micro-
somes. Nevertheless, the significant improvement in CB2 activi-
ty with the pyrrolidine-substituted triazolopyrimidine 27 (as
also previously observed with triazolopyrimidine 5 with a 3,3-
difluoropyrrolidine group at the 7-position) led to an investiga-
tion of the influence of substituted pyrrolidine derivatives on
activity/selectivity and molecular properties. Synthetic access
to such derivatives was identical to that described above: that
is, substitution of core intermediate 18 with various substitut-
ed pyrrolidines. A selection of these derivatives is shown in
Table 5 to illustrate the general effects.
Optimization of activities and properties of the triazolopyri-
midines was studied with a wide selection of amines at the 7-
position. Substitution of intermediate 15 (as before) with vari-
ous amines under basic conditions yielded a set of substituted
triazolopyrimidines, of which a selection is shown in Table 4
which serve to illustrate the emerging SAR. Triazolopyrimidines
substituted at the 7-position with primary amines (i.e., 23 and
24) generally yielded compounds active at the CB2 receptor in
the nanomolar range (23, EC50 =0.011 mm); however, depend-
ing on the size of the substituent, significantly less active (24,
EC50 =0.647 mm) with very good selectivity versus the CB1 re-
ceptor (EC50 >10 mm). Molecular properties of such derivatives
were very similar, with AlogP values >5, mainly with low solu-
bility and generally high microsomal clearance values. Triazolo-
pyrimidines substituted with aliphatic secondary amines (i.e.,
25) displayed a very similar profile in terms of activities and
2-Substituted pyrrolidine triazoloyprimidines (i.e., 29) were
generally weaker in terms of potency at the CB2 receptor and
metabolically weaker than their
isomeric 3-substituted pyrroli-
dine triazolopyrimidine counter-
parts (i.e., 30), although still
active in the low-nanomolar
Table 4. Selection of triazolopyrimidine derivatives 23–28 substituted with amines at position 7.
range.
Although
selectivity
versus the CB1 receptor for the
2-substituted pyrrolidine triazo-
lopyrimidine derivatives could
be influenced, no general trend
was deduced. Alkyl substitution
(at the 2- or 3-positions) led to
Compd:
23
24
25
26
27
28
hCB2 cAMP EC50 [mm]
Efficacy [%]
hCB1 cAMP EC50 [mm]
hCB2: b-arrestin EC50 [mm]
Efficacy [%]
0.011
98
>10
ND[a]
ND[a]
4.51
0.647
90
0.017
98
>10
ND[a]
ND[a]
5.21
0.003
100
1.0
0.09
75
4.39
<1
0.0003
100
0.631
0.132
78
4.97
<1
76/256
0.0016
100
0.555
ND[a]
ND[a]
5.43
>10
ND[a]
ND[a]
6.02
ND[a]
ND[a]
compounds
with
b-arrestin
values in the higher nanomolar
range with partial or full efficacy.
Substitution of the pyrrolidine
ring with one fluorine atom at
the 3-position led to triazolopyr-
AlogP
LYSA [mgmLÀ1
]
<1
103/915
<1
37/287
<1
45/154
CLint [mLminÀ1 kgÀ1] (h/m)
46/324
[a] Not determined.
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
182
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
of 23 nm and full efficacy. Lipo-
philicity remained above 4 and
solubility low. Microsomal clear-
ance was also influenced unfav-
orably. Disubstituted pyrrolidine
34 was also active toward the
CB2 receptor in the sub-nano-
molar range, selective versus the
CB1 receptor, and potent and ef-
ficacious in the CB2 b-arrestin
assay. The lipophilicity of 34 was
decreased to AlogP 3.37, and
the solubility was enhanced
slightly to 7 mgmLÀ1; however,
the compound was even less
stable in human microsomes. In
Table 5. Selection of triazolopyrimidine derivatives 29–34 carrying substituted pyrrolidines at position 7.
Compd:
29
30
31
32
33
34
hCB2 cAMP EC50 [mm]
Efficacy [%]
0.026
100
0.013
101
0.0005
101
0.0006
101
0.0003
101
0.003
101
hCB1 cAMP EC50 [mm]
hCB2 b-arrestin EC50 [mm]
Efficacy [%]
>100
0.93
65
1.7
0.321
86
0.432
0.071
85
>10
0.031
114
0.491
0.023
106
>10
0.102
106
AlogP
5.35
<1
44/263
5.29
<1
25/63
4.82
<1
16/59
3.88
<1
10/43
4.46
<1
25/166
3.37
7
10/27
LYSA [mgmLÀ1
]
CLint [mLminÀ1 kgÀ1] (h/m)
imidine 31, which is active at the CB2 receptor in the sub-
nanomolar range and active versus the CB1 receptor in the
sub-micromolar range. b-Arrestin values remained in the
higher nanomolar range with partial efficacy. Substitution of
the pyrrolidine ring at the 3-position generally led to com-
pounds (i.e., 30, 31, and 32) with medium to low clearance
values in human and mouse microsomes. This was also already
anticipated and observed with compound 5. Solubility re-
mained low for the three compounds; lipophilicity, however,
could be decreased from AlogP=5.08 (5) to AlogP=3.88 for
compound 32.
A comparative analysis of the CB1 and CB2 binding pockets
of the homology models revealed that the substituent at posi-
tion 7 is deeply buried and is in close proximity to Phe199 in
CB1. The corresponding residue in CB2 is Ser193. This differ-
ence in the sub-pocket for the 7-position substituent suggests
the attachment of a hydroxy group to the pyrrolidine ring,
which would be able to form a hydrogen bond to the Ser193
side chain in CB2 (Figure 2). The expectation was to introduce
more polarity into the molecule and to improve its selectivity
for CB1. The Phe199 side chain in CB1 is sterically more de-
manding than the corresponding Ser193 residue in CB2; more-
over, it cannot form a hydrogen bond with the hydroxy group
of the ligand.
Figure 2. Compound 32 with a hydroxylated pyrrolidine at position 7 within
the CB2 homology model. The OH group is postulated to form a hydrogen
bond with Ser193. The corresponding residue in CB1 is Phe199 (not shown),
which disfavors the binding of the hydroxylated substituents at position 7.
Indeed, derivative 32, with an unprotected hydroxy group
on the pyrrolidine ring, was found to have activity at the CB2
receptor in the sub-nanomolar range, and remarkably, this
compound has greater than 10000-fold selectivity versus the
CB1 receptor and a b-arrestin value of 32 nm with full efficacy.
Furthermore, the lipophilicity could be decreased (AlogP:
3.88). The solubility, however, remained low. Stability in micro-
somes was observed to be in the same range as for com-
pounds 30 and 31, and an initial concern that the free hydroxy
group might give rise to an alternate clearance mechanism
through hepatocytes was not fully substantiated, as the com-
pound was cleared in human and mouse hepatocytes as well
with medium clearance values (data not shown). 2-(Hydroxy-
methyl)pyrrolidine derivative 33 was active as well at the CB2
receptor in the sub-nanomolar range, and active at the CB1 re-
ceptor with an EC50 value of 0.491 mm, with a b-arrestin value
summary, appropriate substitution of the pyrrolidine ring yield-
ed very potent compounds at the CB2 receptor, some with re-
markably high selectivity versus the CB1 receptor. The molecu-
lar properties could be influenced to yield compounds with
some solubility, lipophilicity at ~3, with medium to low micro-
somal clearance.
With these encouraging results we turned our attention to
optimization of the substituent at the 3-position with the goal
of maintaining potency toward CB2 and selectivity versus CB1
and improving properties of lipophilicity, solubility, and stabili-
ty even further. Synthetic access followed the same principle
as detailed previously; this is shown in Scheme 2. 5-Amino-1-
(4-methoxybenzyl)-1H-1,2,3-triazole-4-carboxamide 9 was al-
lowed to react with pivaloyl chloride in pyridine to access the
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
183
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Scheme 2. Synthesis of triazolopyrimidine derivatives 38–41. Reagents and conditions: 39: a) 1. 9+pivaloyl chloride, pyridine, 808C, 3 h, 2. KHCO3(aq), H2O,
reflux (35% over two steps); b) 1. 35+POCl3, reflux, 2. (S)-pyrrolidin-3-ol, DIPEA, CH3CN, RT (88% over two steps); c) 36+Et3SiH, TFA, 708C, 22 h, 2. 5-(chloro-
methyl)-1-cyclopropyl-1H-tetrazole, DBU, DMF, 3. CH3OH (40% over three steps).
respective pivalamide, which cyclized under aqueous basic
conditions to give triazolopyrimidine 35 in 35% yield over two
steps. Subsequent chlorination of 35 with POCl3 yielded the
building block for derivatization with various amines at the 7-
position. To access precursors for the final (S)-3-hydroxypyrroli-
dine-substituted triazolopyrimidines 38 and 39, the chlorinated
intermediate was reacted with 3-(S)-hydroxypyrrolidine to yield
intermediate 36. Cleavage of the 4-methoxybenzyl protecting
group at position 3 from 36 was accomplished with triethylsi-
lane and TFA, during which the free hydroxy group was tri-
fluoroacetylated. This approach was deemed convenient, as
derivatization at the 3-position was achieved by reaction with
electrophiles under strong basic conditions without the free
hydroxy group leading to undesired byproducts. Final cleavage
of the trifluoroacetyl ester was carried out with methanol
during the workup procedure to obtain compounds 38 and
39. Similarly, the chlorinated precursor was reacted with 3,3-di-
fluoropyrrolidine to access intermediate 37, the PMB group
again was removed with triethylsilane/TFA, and derivatization
was performed with electrophiles under basic conditions to
access triazolopyrimidine derivatives 40 and 41. The com-
pounds were tested in CB2 and CB1 receptor assays, and their
initial profiles regarding lipophilicity, solubility, and stability
were determined. The compounds are shown in Table 6.
versus CB1 of >1000; however, the compound was still active
at the CB1 receptor at 1.14 mm. The compound was active in
the hCB2 b-arrestin test, with 4.7 nm and full efficacy. Remarka-
bly, the AlogP was 2.74 with a solubility of 21 mgmLÀ1 and low
metabolic clearance (CLint [mLminÀ1 kgÀ1] (h/m): 10/36) in
human and mouse microsomes, respectively. There is a 10-fold
decrease in potency for the CB2 receptor in combining the (S)-
3-pyrrolidine substituent at the 7-position with a benzylic tet-
razole at the 3-position of triazolopyrimidine 39 relative to 38;
however, the selectivity versus CB1 is enhanced, with EC50 >
10 mm (b-arrestin EC50 =22 nm (119% efficacy)). Tetrazole 39
has a remarkably decreased AlogP of 1.21 combined with a sol-
ubility of 162 mgmLÀ1 and low human and mouse microsomal
clearance. Substituting the triazolopyrimidine core with 3,3-di-
fluoropyrrolidine at the 7-position generally yielded very
potent compounds (i.e., 40 and 41) toward the CB2 receptor,
with excellent selectivities versus the CB1 receptor, very good
potency on b-arrestin with full efficacy and decent lipophilicity
in terms of AlogP (4.41 and 2.24, respectively). The solubility
for 40 was poor, as anticipated with the AlogP value of 4.14 in
comparison with (S)-3-hydroxypyrrolidine-derived triazolopyri-
midines; however, human and mouse microsomal stability
could be retained (40 CLint [mLminÀ1 kgÀ1] (h/m): 10/41; 41
CLint [mLminÀ1 kgÀ1] (h/m): 10/42). As triazolopyrimidine 39
showed one of the most balanced profiles in terms of potency,
selectivity, and in vitro ADME, it was further characterized in
vivo in mouse PK studies, and the data are listed in Table 7.
Substitution at the triazolopyrimidine 3-position was carried
out with (substituted) alkyl and cycloalkyl groups, which gener-
ally yielded compounds inactive at the CB2 receptor. The po-
tencies of derivatives with meta- or para-substituted benzylic
groups were usually in the mi-
cromolar range. Benzylic groups
with appropriate ortho substitu-
tion (generally electron-with-
drawing groups), however, were
found to be very potent. Deriva-
tization with five-membered aro-
matic heterocycles with various
substitution patterns yielded the
most potent and interesting
compounds. The combination of
(S)-3-pyrrolidine substitution at
the 7-position with a 2-methyl-
sulfonylbenzyl substituent at the
3-position yielded triazolopyrimi-
dine 38. This compound was
found to be active in the sub-
nanomolar range with selectivity
Table 6. Selection of triazolopyrimidine derivatives 38–41 substituted at position 3.
Compd:
38
39
40
41
hCB2 cAMP EC50 [mm]
Efficacy [%]
0.0001
100
0.0007
102
0.00007
103
0.0005
100
>10
0.021
122
2.24
32
10/42
hCB1 cAMP EC50 [mm]
hCB2 b-arrestin EC50 [mm]
Efficacy [%]
1.41
0.0047
110
>10
0.022
119
0.098
0.0041
86
AlogP
2.74
21
10/63
1.21
162
10/71
4.14
<1
10/41
LYSA [mgmLÀ1
]
CLint [mLminÀ1 kgÀ1] (h/m)
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
184
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
bution (Vss: 1 LkgÀ1), and terminal half-life of 0.3 h. Data from
the mouse PK study showed triazolopyrimidine 39 to reach
a concentration of 49 ngmLÀ1 1 h after i.v. application in
plasma, which corresponds to a free concentration 10-fold
above the measured mouse cAMP EC50 value. This overall pro-
file was found suitable for in vivo potency assessment. The
mouse kidney ischemia–reperfusion model is used widely to
access acute kidney injury (AKI).[36] In this protocol a test com-
pound is given to a cohort of mice at a certain concentration
via gavage 30 min prior to surgery, in which the blood flow to
the kidneys is stopped by firm clamping of the kidney vessels
for 25 min. The resulting ischemia and subsequent reperfusion
caused kidney damage, which could be ameliorated upon ad-
ministration of the CB2 agonist. The level of damage and its
prevention by CB2 agonists was assessed by using several
readouts 24 h after ischemia–reperfusion (I/R). As direct makers
of damage, plasma creatinine and urea levels were deter-
mined. Additionally, the plasma levels of three kidney biomark-
ers (NGAL, KIM1, and osteopontin) were analyzed at the same
time point.[37] The effect on these kidney markers upon admin-
istration of triazolopyrimidine 39 at 10 mgkgÀ1 given orally are
shown in Figure 3.
Table 7. Physicochemical and early safety profile of triazolopyrimidine
39.
CB2 cAMP EC50 [mm] (h/m)
Efficacy [%] (h/m)
Mr [Da]
0.007/0.004
103/101
384
3 (8/30/62)[a]
10/71/143
37
PAMPA Peff [10À6 cmsÀ1
]
Microsomal CLint [mLminÀ1 kgÀ1] (h/m/rat)
In vivo mouse CL [mLminÀ1 kgÀ1
]
PPB(fu) [%][b] (h/m)
13.2/3.1
GSH (human liver microsomes) adducts
CYP reversible inhibition [mm] (3A4/2C9/2D6)
CB2 Ki [mm] (h/m)
none detected
>50/16/>50
0.008/0.021
>1190
Ki ratio hCB1/hCB2
clogP
0.8 (logD: 2.8)
56/80/237
15/141
1.0 / 0.3 (i.v.)
6.1/19.1
H2O solubility/FaSSIF/FeSSIF [mgmLÀ1
]
Hepatocytes CL [mLminÀ1 (106 cells)À1] (h/m)
mouse Vss [LkgÀ1] / t1/2 [h]
P-gp ER[c] (h/m)
The two main readouts, creatinine and urea levels, measured
24 h after I/R showed significant decreases of ~50 and ~40%,
respectively, at a dose of 10 mgkgÀ1 after oral administration.
The three biomarkers NGAL, KIM1, and osteopontin, indicative
of kidney health,[37] were also measured 24 h post-I/R. NGAL
and osteopontin concentrations were significantly decreased
by ~75 and ~85%, respectively. KIM1, although visibly re-
duced, failed to show significance due to high variability.
These data clearly demonstrate that the CB2 agonist 39 has
protective function, as shown by the results of the I/R experi-
ment. Owing to the severely diseased conditions of the ani-
mals, plasma samples could not be taken during the 24 h
course of the experiment. From other experiments, however
(UUO; see below), in vivo PK as well as hepatocyte clearance
rates could be obtained. These data suggest that high receptor
occupancies are likely, especially during the critical initial phase
of the experiment.
hERG IC50 [mm]
>20
Stability in aq. media at pH 1–10
stable
[a] Acceptor
compartment/membrane/donor
compartment
(%).
[b] Plasma protein binding, fraction unbound. [c] P-glycoprotein efflux
ratio.
Triazolopyrimidine 39 was equally potent toward the human
and mouse CB2 receptor isoforms in the sub-nanomolar range.
In a binding assay[28] the compound was also active in the
nanomolar range on the human and mouse CB2 isoforms.
Here as well, the selectivity in binding versus the CB1 receptor
was very high (>1190-fold). Compound 39 has a molecular
weight of 384 Da and a clogP of 0.8 with a measured logD of
2.8. The clogP value is in agreement with the AlogP value
(1.21), which was used as a reference value throughout the op-
timization process.
Test compound was subjected to the kidney fibrosis unilater-
al ureter obstruction (UUO) model, in which the ureter from
one kidney is surgically ligated to cause pressure injury due to
the blocked urine outflow over the course of one week.[38] The
backpressure of urine in the left kidney results in atrophy and
fibrosis, as the kidney size is limited by a rigid capsule. After
UUO surgery, triazolopyrimidine 39 was given orally for seven
days via food admix at a dose of 3 mgkgÀ1 per day. At the end
of the experiment the kidney was perfusion fixed for immuno-
histochemical determination of collagen-I deposition in longi-
tudinal paraffin sections. The relative content of tissue colla-
gen-I was determined morphometrically. The CB2 agonist in-
duced a decrease in collagen-I deposition relative to placebo-
treated animals (Figure 4). Interestingly this inhibition is in the
range of inhibition of fibrosis described for angiotensin-con-
verting enzyme (ACE) inhibitors.[39] In our own UUO experi-
ments, enalapril (32 mgkgÀ1 per day in drinking water) inhibit-
ed fibrosis by 20, 43, and 54% versus placebo (UUO kidneys in
Table 7 shows that passive permeability measured by
PAMPA[29] is good, as is the solubility in aqueous buffer[30] and
biologically relevant media (FaSSIF and FeSSIF).[31] Therefore,
we could expect that overall, compound 39 has a good per-
meability and solubility profile, and thus we anticipate good
oral bioavailability. Compound 39 is not extensively bound to
plasma protein, with a fraction unbound (fu) of 13.2% in
human and 3.1% in mouse.[32] It also has a high efflux ratio
(ER) as measured in the P-glycoprotein transporter assay,[33]
which makes the compound potentially suitable for peripheral
disease indications. Furthermore, triazolopyrimidine 39 does
not inhibit the hERG channel and has a low potential to form
reactive metabolites.[34] Compound 39 also does not inhibit cy-
tochrome P450 (CYP),[35] and is chemically stable in the pH
range of 1–10. After single i.v. dosing at 1 mgkgÀ1 in mouse,
39 also demonstrated a good PK profile, with a moderate
clearance rate (37 mLminÀ1 kgÀ1), a moderate volume of distri-
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
185
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Figure 3. Results from experiments with triazolopyrimidine 39 (10 mgkgÀ1 p.o.) to show decreases in the extent of kidney injury 24 h post-I/R using readouts
for: a) plasma creatinine, b) plasma urea, c) NGAL, d) KIM1, and e) osteopontin. Data are the mean ÆSD for n=6 independent experiments; Student’s t-test:
*p<0.05, ***p>0.001; KIM1 failed to reach statistical significance at p=0.06. Black horizontal lines indicate sham operated animals without ligating the
ureter.
gen-I). From 10 microscopic fields of each kidney, the collagen-
I-positive pixel counts were determined in the cortex of each
kidney. As the total area analyzed was identical for all kidneys
under investigation, the absolute number of pixels was used
as the relevant readout. The untreated (but ligated) vehicle
kidneys show a large increase in collagen-I deposition (n=5,
red bar) over unligated contralateral kidneys (n=5, orange
bar). Oral treatment with triazolopyrimidine 39 at 3 mgkgÀ1
per day decreased the amount of fibrosis by ~40% (n=6, blue
bar) significantly (one-way ANOVA, p<0.05).
In summary, starting from an HTS campaign for the identifi-
cation of potent and selective small-molecule CB2 agonists,
several interesting hit clusters and single molecules were iden-
tified. The triazolopyrimidine cluster attracted our attention, as
potent derivatives with SAR were identified. An extensive lead-
optimization campaign in which three vectors were very sys-
tematically balanced led to the identification of preliminarily
optimal substitution patterns. Thus, molecular design resulted
in various highly potent CB2 agonists with excellent selectivi-
ties versus the CB1 receptor. The molecular properties could
be modulated to achieve reasonable to very good solubility,
lipophilicity, and stability parameters in human and mouse mi-
crosomes. The most promising triazolopyrimidines were further
characterized, and compound 39 was identified as having one
of the most balanced property profiles of all the compounds
examined. Subsequent submission to the ischemia–reperfusion
mouse model showed efficacy at a dose of 10 mgkgÀ1 p.o. in
addition to the three kidney markers, which measured a signifi-
cant depletion effect, thus indicating a protective potential
toward inflammatory kidney damage. Triazolopyrimidine 39
was also efficacious in a model of renal fibrosis (unilateral
Figure 4. Results from triazolopyrimidine 39 in the UUO model.
three independent experiments with n=6, p<0.05, Student’s
t-test).
To monitor exposure, plasma samples were collected at
three time points (170 h, day 7: 21 ngmLÀ1; 176 h, day 7:
17 ngmLÀ1; 192 h, day 8: 14 ngmLÀ1). Similar plasma concen-
trations were achieved at days 7 and 8, indicating that steady-
state conditions were likely established. The average free
plasma concentration was found to be 0.54 ngmLÀ1, which
corresponds to high receptor occupancy (e.g., 74% based on
mouse cAMP data), thereby suggesting that the observed in
vivo efficacy is mediated by CB2.
Immunohistochemical (IHC) analysis of collagen-I deposition
in the UUO model seven days after ureter ligation of the
kidney was used to determine the amount of fibrosis (colla-
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
186
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
ureter obstruction model) at 3 mgkgÀ1 per day. Further investi-
gations into the triazolopyrimidine class to arrive at com-
pounds with even higher potency and improved molecular
properties that would permit reduced dosing in any given and
relevant animal model are currently being pursued.
zyl)-3H-[1,2,3]triazolo[4,5-d]pyrimidine. The residue was used for
the next reaction step without further purification. A mixture of 5-
tert-butyl-7-chloro-3-(2-chlorobenzyl)-3H-[1,2,3]triazolo[4,5-d]pyrimi-
dine (crude, 47.2 mmol), 3,3-difluoropyrrolidine hydrochloride
(13.6 mg, 94.4 mmol), and N,N-diisopropylethylamine (DIPEA;
12.2 mg, 94.4 mmol) in CH3CN (250 mL) was stirred at room temper-
ature overnight. The reaction mixture was directly purified by prep-
arative HPLC [column: Gemini 5 mm C18 110A 75”30 mm; mobile
phase: (0.05% HCO2H in H2O)/CH3CN 40:60!5:95 v/v gradient
over 8 min; l=280 nm; flow rate: 30 mLminÀ1] to afford 5-tert-
butyl-3-(2-chlorobenzyl)-7-(3,3-difluoropyrrolidin-1-yl)-3H-[1,2,3]tria-
zolo[4,5-d]pyrimidine (5) as a colorless gum (13.3 mg, 69% for two
steps). MS (m/z): 407.7 [M+H]+; 1H NMR (300 MHz, CDCl3): d=
7.34–7.46 (m, 1H), 7.16–7.26 (m, 3H), 5.88 (s, 2H), 4.40–4.78 (m,
2H), 3.91–4.29 (m, 2H), 2.33–2.74 (m, 2H), 1.34 ppm (s, 9H).
Experimental Section
Chemicals were purchased from commercial sources and were
used without further purification. Reactions were magnetically and
mechanically stirred and monitored by thin-layer chromatography
(TLC) on Merck silica gel 60 F254 TLC glass plates (visualized by UV
fluorescence at l=254 nm) or analytical HPLC–MS on an Agi-
lent 1100 instrument, Finnigan single-quadrupole ESI, autosampler
Gilson 215, Chromeleon 6.7, rapid resolution cartridge Zorbax XDB
3.5 mm (21”30 mm). The purity of new compounds was >95%, as
determined by 1H NMR spectroscopy after purification. Flash
column chromatography was performed with silica cartridges from
Silicycle, eluting with distilled technical-grade solvents on a Combi-
Flash RF column. Yields refer to purified compounds. Preparative
HPLC purification was carried out on a reversed-phase Phenomen-
ex Gemini 5 mm NX-C18 column, 75”30 mm, AXIA, eluting with
a gradient composed of CH3CN, H2O, and HCO2H. NMR data were
recorded on a Bruker spectrometer operating at 300, 400, and
600 MHz. Chemical shifts (d) are reported in ppm with TMS as in-
ternal standard. The data are reported as: s=singlet, d=doublet,
t=triplet, m=multiplet, br=broad signal, and coupling con-
stant(s) J in Hz. Service measurements were performed by the
NMR or MS service teams at F. Hoffmann–La Roche, Basel.
(3S)-1-[5-tert-Butyl-3-[(1-cyclopropyltetrazol-5-yl)methyl]tria-
zolo[4,5-d]pyrimidin-7-yl]pyrrolidin-3-ol (39)
a) Synthesis of 35: A mixture of 5-amino-1-(4-methoxybenzyl)-1H-
1,2,3-triazole-4-carboxamide 9 (10.0 g, 10 mmol) and pivaloyl chlo-
ride (7.47 mL, 60.7 mmol) in pyridine (20.2 mL) was stirred at 808C
for 2 h under N2 atmosphere. Aqueous NaOH (8m, 15.2 mL,
121 mmol) and CH3OH (20.2 mL) were then added to the reaction
mixture. After stirring at 808C for 1 h, the reaction mixture was
poured into aqueous 1m HCl, extracted with Et2O, washed with
aqueous 2m HCl, H2O and brine, dried over Na2SO4, and concen-
trated in vacuo to afford the mixture of crude 1-(4-methoxyben-
zyl)-5-pivalamido-1H-1,2,3-triazole-4-carboxamide and N-(4-cyano-
1-(4-methoxybenzyl)-1H-1,2,3-triazol-5-yl)pivalamide. The residue
was used for the next reaction without further purification. A mix-
ture of the above crude residue and KHCO3 (12.1 g, 121 mmol) in
H2O (242 mL) was held at reflux for 18 h. The reaction mixture was
poured into aqueous 1m HCl, extracted with EtOAc, washed with
brine, dried over Na2SO4, and concentrated in vacuo. The crude
residue was purified by flash chromatography (silica gel, 10!70%
EtOAc in heptane) to afford 5-tert-butyl-3-(4-methoxybenzyl)-3H-
[1,2,3]triazolo[4,5-d]pyrimidin-7(4H)-one (35) (4.44 g, 35% for two
steps). MS (m/z): 314.2 [M+H]+.
5-tert-Butyl-3-(2-chlorobenzyl)-7-(3,3-difluoropyrrolidin-1-yl)-
3H-[1,2,3]triazolo[4,5-d]pyrimidine (5)
a) Synthesis of 15: A mixture of 5-amino-1-(2-chlorobenzyl)-1H-
1,2,3-triazole-4-carboxamide (2 g, 7.95 mmol, commercially avail-
able) and pivaloyl chloride (1.47 mL, 11.9 mmol) in pyridine
(3.98 mL) was stirred at 808C for 2 h under N2 atmosphere. Aque-
ous NaOH (8m, 2.98 mL, 23.8 mmol) and CH3OH (3.98 mL) were
then added to the reaction mixture. After stirring at 808C for 2 h,
the reaction mixture was poured into aqueous 1m HCl, extracted
with Et2O, washed with 2m HCl, H2O, and brine, dried over Na2SO4,
and concentrated in vacuo to afford the mixture of crude 1-(2-
chlorobenzyl)-5-pivalamido-1H-1,2,3-triazole-4-carboxamide and N-
(1-(2-chlorobenzyl)-4-cyano-1H-1,2,3-triazol-5-yl)pivalamide. The res-
idue was used for the next reaction without further purification. A
mixture of the above crude residue and KHCO3 (3.00 g, 30.0 mmol)
in H2O (60.0 mL) was held at reflux for 18 h. The reaction mixture
was poured into aqueous 1m HCl, extracted with EtOAc, washed
with brine, dried over Na2SO4, and concentrated in vacuo. The
crude residue was purified by flash chromatography (silica gel,
10!70% EtOAc in heptane) to afford 5-tert-butyl-3-(2-chloroben-
b) Synthesis of 36: 5-tert-Butyl-3-(4-methoxybenzyl)-3H-[1,2,3]tria-
zolo[4,5-d]pyrimidin-7(4H)-one (35) (255 mg, 814 mmol), POCl3
(5.03 g, 3.0 mL, 32.8 mmol), and N,N-diethylaniline (243 mg, 259 mL,
1.63 mmol) were combined in a 5-mL vessel to give a light-yellow
solution. The reaction mixture was heated at 1208C and stirred for
4 h. The crude reaction mixture was concentrated in vacuo and
poured into 1 mL H2O and extracted with EtOAc (2”2 mL). The or-
ganic layers were combined, washed with saturated aqueous NaCl
(1”1 mL), dried over Na2SO4, and concentrated in vacuo to yield 5-
tert-butyl-7-chloro-3-(4-methoxybenzyl)-3H-[1,2,3]triazolo[4,5-d]pyri-
midine, which was used crude in the subsequent reaction. 5-tert-
Butyl-7-chloro-3-(4-methoxybenzyl)-3H-[1,2,3]triazolo[4,5-d]pyrimi-
dine (159 mg, 478 mmol, crude), (S)-pyrrolidin-3-ol (62.5 mg,
717 mmol), and DIPEA (124 mg, 167 mL, 956 mmol) were combined
with CH3CN (500 mL) in a 5-mL vessel to give a dark-green solution.
The reaction mixture was stirred overnight and purified directly by
preparative reversed-phase HPLC (column: Gemini 5 mm C18 110A
75”30 mm; l=300 nm; flow rate: 30 mLminÀ1), eluting with a gra-
dient formed from 0.05% Et3N in H2O and CH3CN (80:20!5:95
over 8 min). The combined product-containing fractions were
evaporated to yield 161 mg (88%) of (S)-1-(5-tert-butyl-3-(4-me-
thoxybenzyl)-3H-[1,2,3]triazolo[4,5-d]pyrimidin-7-yl)pyrrolidin-3-ol
(36) as a light-yellow solid. MS (m/z): 383.3 [M+H]+; 1H NMR
(400 MHz, CDCl3): d=7.45–7.47 (d, 2H, J=8 Hz), 6.82–6.84 (d, 2H,
zyl)-3H-[1,2,3]triazolo[4,5-d]pyrimidin-7(4H)-one (15) as
a white
1
solid (1.03 g, 41% for two steps). MS (m/z): 318.2 [M+H]+; H NMR
(300 MHz, [D6]DMSO): d=12.20 (brs, 1H), 7.08–7.61 (m, 4H), 5.73
(s, 2H), 1.29 ppm (s, 9H).
b) Synthesis of 5: A mixture of 5-tert-butyl-3-(2-chlorobenzyl)-3H-
[1,2,3]triazolo[4,5-d]pyrimidin-7(4H)-one (15) (12.3 mg, 38.7 mmol)
and, N,N-diethylaniline (12.3 mL, 77.4 mmol) in POCl3 (250 mL,
2.73 mmol) was held at reflux for 3 h under N2 atmosphere. The re-
action mixture was concentrated in vacuo, diluted with EtOAc,
washed with cold H2O and brine, dried over Na2SO4, and concen-
trated in vacuo to afford crude 5-tert-butyl-7-chloro-3-(2-chloroben-
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
187
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
[5] P. Mukhopadhyay, M. Rajesh, H. Pan, V. Patel, B. Mukhopadhyay, S.
Batkai, B. Gao, P. Pacher, Free Radic Biol. Med. 2010, 48, 457–467.
[6] P. Pacher, S. Batkai, G. Kunos, Pharmacol. Rev. 2006, 58, 389–462.
[7] a) D. Centonze, A. Finazzi-Agro, G. Bernardi, M. Maccarrone, Trends Phar-
macol. Sci. 2007, 28, 180–187; b) J. Fernµndez-Ruiz, Br. J. Pharmacol.
2009, 156, 1029–1040.
[8] B. Hu, H. Doods, R. D. Treede, A. Ceci, Pain 2009, 143, 206–212.
[9] P. Anand, G. Whiteside, C. J. Fowler, A. G. Hohmann, Brain Res. Rev.
2009, 60, 255–266.
[10] C. J. Fowler, S. B. Gustafsson, S. C. Chung, E. Persson, S. O. Jacobsson, A.
Bergh, Curr. Top. Med. Chem. 2010, 10, 814–827.
[11] I. Bab, A. Zimmer, E. Melamed, Ann. Med. 2009, 41, 560–567.
[12] M. Maccarrone, Prog. Lipid Res. 2009, 48, 344–354.
J=8 Hz), 5.64 (s, 2H), 4.60–4.76 (m, 1H), 4.15–4.54 (m, 2H), 3.82–
4.09 (m, 2H), 3.76 (s, 3H), 2.03–2.30 (m, 2H), 1.39 ppm (s, 9H).
c) Synthesis of 39: (S)-1-(5-tert-Butyl-3-(4-methoxybenzyl)-3H-
[1,2,3]triazolo[4,5-d]pyrimidin-7-yl)pyrrolidin-3-ol
(36)
(156 mg,
408 mmol) and triethylsilane (474 mg, 651 mL, 4.08 mmol) were
combined with TFA (15.0 mL) in a 200-mL flask to give a colorless
solution. The reaction mixture was heated at 708C and stirred for
22 h. The reaction mixture was concentrated in vacuo to yield (S)-
1-(5-tert-butyl-3H-[1,2,3]triazolo[4,5-d]pyrimidin-7-yl)pyrrolidin-3-yl
2,2,2-trifluoroacetate, which was used crude in the subsequent
step. (S)-1-(5-tert-Butyl-3H-[1,2,3]triazolo[4,5-d]pyrimidin-7-yl)pyrroli-
din-3-yl 2,2,2-trifluoroacetate (crude, 52.8 mmol), 5-(chloromethyl)-
1-cyclopropyl-1H-tetrazole (16.8 mg, 106 mmol), and 1,8-diazabicy-
clo[5.4.0]undec-7-ene (DBU; 106 mmol) were combined with N,N-di-
methylformamide (DMF; 250 mL) in a 5-mL vessel at room tempera-
ture and stirred overnight. CH3OH (500 mL) was added to the reac-
tion mixture and stirred at room temperature for 1 h. The reaction
mixture was purified directly by preparative reversed-phase HPLC
(column: Gemini 5 mm C18 110A 75”30 mm; l=300 nm; flow rate:
30 mLminÀ1) eluting with a gradient formed from 0.05% Et3N in
H2O and CH3CN (80:20!5:95 over 8 min). The combined product-
containing fractions were evaporated to yield 8.2 mg (40%) of (3S)-
1-[5-tert-butyl-3-[(1-cyclopropyltetrazol-5-yl)methyl]triazolo[4,5-
d]pyrimidin-7-yl]pyrrolidin-3-ol (39) as a light-yellow gum. MS (m/
z): 385.3 [M+H]+; 1H NMR (600 MHz, [D6]DMSO): d=6.22 (s, 2H),
5.03–5.17 (d, 1H, J=36 Hz), 4.38–4.57 (d, 1H, J=36 Hz), 3.94–4.31
(m, 3H), 3.62–3.90 (m, 3H), 3.32 (s, 3H), 1.87–2.21 (m, 2H), 1.28 (s,
9H), 1.21–1.26 (m, 2H), 1.06–1.16 ppm (m, 2H); HRMS: (m/z) [M+
H]+ calcd for C17H24N10O: 384.2134, found: 385.2215.
[13] T. Bíró, B. I. Tóth, G. Haskó, R. Paus, P. Pacher, Trends Pharmacol. Sci.
2009, 30, 411 –420.
[14] R. G. Pertwee, A. C. Howlett, M. E. Abood, S. P. Alexander, V. Di Marzo,
M. R. Elphick, P. J. Greasley, H. S. Hansen, G. Kunos, K. Mackie, R. Me-
choulam, R. A. Ross, Pharmacol. Rev. 2010, 62, 588–631.
[15] R. G. Pertwee, Handb. Exp. Pharmacol. 2005, 168, 1–51.
[16] J. W. Huffman, Mini Rev. Med. Chem. 2005, 5, 641–649.
[17] a) S. Brogi, F. Corelli, V. Di Marzo, A. Ligresti, C. Mugnaini, S. Pasquini, A.
Tafi, Eur. J. Med. Chem. 2011, 46, 547–555; b) E. J. Gilbert, G. Zhou, M. K.
Wong, L. Tong, B. B. Shankar, C. Huang, J. Kelly, B. J. Lavey, S. W. McCom-
bie, L. Chen, R. Rizvi, Y. Dong, Y. Shu, J. A. Kozlowski, N. Y. Shih, R. W.
Hipkin, W. Gonsiorek, A. Malikzay, C. A. Lunn, L. Favreau, D. J. Lundell,
Bioorg. Med. Chem. Lett. 2010, 20, 608–611; c) L. Tong, B. B. Shankar, L.
Chen, R. Rizvi, J. Kelly, E. Gilbert, C. Huang, D. Y. Yang, J. A. Kozlowski,
N. Y. Shih, W. Gonsiorek, R. W. Hipkin, A. Malikzay, C. A. Lunn, D. J. Lun-
dell, Bioorg. Med. Chem. Lett. 2010, 20, 6785–6789; d) G. A. Thakur, R.
Tichkule, S. Bajaj, A. Makriyannis, Expert Opin. Ther. Pat. 2009, 19, 1647–
1673; e) J. H. Lange, M. A. van der Neut, A. J. Borst, M. Yildirim, H. H. van
Stuivenberg, B. J. van Vliet, C. G. Kruse, Bioorg. Med. Chem. Lett. 2010,
20, 2770–2775; f) C. Mugnaini, S. Nocerino, V. Pedani, S. Pasquini, A.
Tafi, M. De Chiaro, L. Bellucci, M. Valoti, F. Guida, L. Luongo, S. Dragoni,
A. Ligresti, A. Rosenberg, D. Bolognini, M. G. Cascio, R. G. Pertwee, R.
Moaddel, S. Maione, V. Di Marzo, F. Corelli, Chem. Med. Chem. 2012, 7,
920–934.
Animals in research
[18] L. A. Matsuda, S. J. Lolait, M. J. Brownstein, A. C. Young, T. I. Bonner,
Nature 1990, 346, 561–564.
Animal work was performed according to the Swiss federal law for
animal protection and was approved by the Veterinary Office Basel.
Animals were provided an acclimation period of >5 days before
use, conventional hygienic conditions for housing (temperature:
20–248C, minimum relative humidity: 40%, light/dark cycle: 12 h),
and standard diet; access to food and drinking water ad libitum.
Specific details on permissions can be provided upon request.
[19] M. Stahl, H. Mauser, J. Chem. Inf. Mod. 2005, 45, 542–548.
[20] cAMP assays were performed with CHO cells expressing human CB1
and CB2 receptor variants as described in: C. Ullmer, S. Zoffmann, B.
Bohrmann, H. Matile, L. Lindemann, P. J. Flor, P. Malherbe, Br. J. Pharma-
col. 2012, 167, 1448–1466. Efficacies are expressed as percentages rela-
tive to 1 mm CP55940. EC50 values are the average of determinations
(n=1) performed in triplicate. Binding data (Ki) generally correlated
with the cAMP data for CB1 and CB2.
[21] b-Arrestin recruitment assays were performed according to the instruc-
tions of the manufacturer (DiscoveRx, Fremont, CA, USA). Efficacies are
expressed as percentages relative to 1 mm CP55940. Compounds with
>80% efficacy were considered full agonists.
[22] Q. Liu, A. Makriyannis (University of Connecticut), Int. PCT Pub. No.
WO 2003035005 A3, 2003.
[23] AlogP: A. K. Ghose, V. N. Viswanadhan, J. J. Wendoloski, J. Phys. Chem. A
1998, 102, 3762–377224.
[24] Ligand efficiency: a) C. Abad-Zapatero, J. T. Metz, Drug Disc. Today 2005,
10, 464–469; b) I. D. Kuntz, K. Chen, K. A. Sharp, P. A. Kollman, Proc.
Natl. Acad. Sci. USA 1999, 96, 9997–10002.
Acknowledgements
It is with pleasure that we thank all our collaborators whose con-
tributions made this work possible and so enjoyable, especially
Michel Dietz, Olivier Gavelle, Jørgen Nielsen, Didier Pomeranc,
Didier Rombach, Astride Schnçbelen, Tanja Minz, Anja Osterwald,
Camille Perret, and Elisabeth Zirwes.
Keywords: CB2 receptor agonists · fibrosis · inflammation ·
lead optimization · triazolopyrimidines
[25] a) Molecular Operating Environment (MOE) ver. 2014.09, Chemical Com-
puting Group Inc., 1010 Sherbooke Street West, Suite #910, Montreal,
QC H3A 2R7 (Canada), 2015; b) J. Standfuss, P. C. Edwards, A. D’Antona,
M. Fransen, G. Xie, D. D. Oprian, G. F. Schertler, Nature 2011, 471, 656–
660.
[1] P. Pacher, R. Mechoulam, Prog. Lipid Res. 2011, 50, 193–211.
[2] P. Pacher, P. Mukhopadhyay, R. Mohanraj, G. Godlewski, S. Batkai, G.
Kunos, Hypertension 2008, 52, 601–607.
[3] A. A. Izzo, K. A. Sharkey, Pharmacol. Ther. 2010, 126, 21–38.
[4] a) S. Lotersztajn, F. Teixeira-Clerc, B. Julien, V. Deveaux, Y. Ichigotani, S.
Manin, J. Tran-Van-Nhieu, M. Karsak, A. Zimmer, A. Mallat, Br. J. Pharma-
col. 2008, 153, 286–289; b) B. Horvµth, L. Magid, P. Mukhopadhyay, S.
Bµtkai, M. Rajesh, O. Park, G. Tanchian, R.-Y. Gao, E. C. Goodfellow, M.
Glass, R. Mechoulam, P. Pacher, Br. J. Pharmacol. 2012, 165, 2462–2478.
[26] Lyophilization solubility assay (LYSA):
a high-throughput solubility
screening assay that assesses the kinetic solubility of a test compound
in phosphate buffer at pH 6.5 from evaporated DMSO compound stock
solution.
[27] The microsomal stability assay measures the rate of disappearance
(clearance: CLint [mLminÀ1 kgÀ1]) of a test compound from incubations
with liver microsomes from human, rat, or mouse supplemented with
metabolic cofactors (typically NADPH). The assay is used primarily for
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
188
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
early assessment of metabolic stability by ranking relative CYP-mediat-
ed metabolism propensities of compounds within chemical series and
guiding lead optimization toward sufficiently stable compounds to
attain the required pharmacokinetics. L. Di, C. Keefer, D. O. Scott, T. J.
Strelevitz, G. Chang, Y.-A. Bi, Y. Lai, J. Duckworth, K. Fenner, M. D. Trout-
man, R. S. Obach, Eur. J. Med. Chem. 2012, 57, 441–448.
[33] P-glycoprotein (permeability glycoprotein, abbreviated as P-gp) is the
most studied and best characterized drug transporter. J. D. Wessler, L. T.
Grip, J. Mendell, R. P. Giugliano, J. Am. Coll. Cardiol. 2013, 61, 2495–
2502.
[34] The GSH assay is the primary tool to assess the potential of reactive
metabolite formation of new compounds. Review: a) I. A. Blair, Curr.
Drug Metab. 2006, 7, 853–872. The hERG assay serves to access the po-
tential of a test compound to interfere with the channel and in turn
result in undesired side effects at the cardiac system: b) J. W. Warmke,
B. Ganetzky, Proc. Natl. Acad. Sci. 1994, 91, 3438–3442; c) M. C. Sangui-
netti, M. Tristani-Firouzi, Nature 2006, 440, 463–469.
[28] Radioligand binding assays were performed with membranes prepared
from cells expressing human or mouse CB2 receptors using
[3H]CP55940 (PerkinElmer) as radioligand. EC50 or Ki values were calcu-
lated from a single experiment using triplicates of ten different com-
pound concentrations.
[35] Cytochrome P450 (CYP) enzymes play a central role in drug metabo-
lism. The potential of a test compound to be metabolized and subse-
quently excreted is assessed systematically: a) C. Barau, B. Ghaleh, A.
Berdeaux, D. Morin, Fund. Clin. Pharm. 2015, 29, 1–9; b) J. M. McKim in
Encyclopedia of Drug Metabolism and Interactions, Vol. 4 (Ed.: A. V. Lyubi-
mov), Wiley, Chichester, 2012, pp. 223–250.
[36] Q. Wei, Z. Dong, Am. J. Physiol.-Renal. Physiol. 2012, 303, F1487–F1494.
[37] M. G. Madsen, Dan. Med. J. 2013, 60, B4582.
[38] A. C. Ucero, A. Benito-Martin, M. C. Izquierdo, M. D. Sanchez-Ninð, A. B.
Sanz, A. M. Ramos, S. Berzal, M. Ruiz-Ortega, J. Egido, A. Ortiz, Int. Urol.
Nephrol. 2014, 46, 765–776.
[29] The parallel artificial membrane permeability assay (PAMPA) is
a method that determines the permeability of substances from a donor
compartment through a lipid-infused artificial membrane into an ac-
ceptor compartment. M. Kansy, F. Senner, K. Gubernator, J. Med. Chem.
1998, 41, 1007–1010.
[30] The thermodynamic solubility assay (THESA) determines the equilibrium
solubility of a compound. THESA is a medium-throughput assay that
differentiates from other high-throughput solubility tests in that the
assay is conducted in customized buffers, and the starting material is
a characterized solid sample. Review: C. Saal, A. C. Petereit, Eur. J.
Pharm. Sci. 2012, 47, 589–595.
[39] K. Wojcikowski, H. Wohlmuth, D. W. Johnson, G. Gobe, Phytotherapy Res.
2010, 24, 875–884.
[31] FaSSIF/FeSSIF: equilibrium solubility determined for powder dissolved
in fasted (FaSSIF) and fed (FeSSIF) simulated intestinal fluids buffers at
pH 6.5. Review: P. Augustijns, B. Wuyts, B. Hens, P. Annaert, J. Butler, J.
Brouwers, Eur. J. Pharm. Sci. 2014, 57, 322–332.
Received: May 18, 2015
[32] M. L. Howard, J. J. Hill, G. R. Galluppi, M. A. McLean, Comb. Chem. High
Throughput Scr. 2010, 13, 170–187.
Revised: June 23, 2015
Published online on July 21, 2015
ChemMedChem 2016, 11, 179 – 189
www.chemmedchem.org
189
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim