Design, Synthesis, Structure-Activity Relationship Studies, and Three-Dimensional Quantitative Structure-Activity Relationship (3D-QSAR) Modeling of a Series of O-Biphenyl Carbamates as Dual Modulators of Dopamine D3 Receptor and Fatty Acid Amide Hydrolase

Article abstract of DOI:10.1021/acs.jmedchem.6b01578

We recently reported molecules designed according to the multitarget-directed ligand paradigm to exert combined activity at human fatty acid amide hydrolase (FAAH) and dopamine receptor subtype D3 (D3R). Both targets are relevant for tackling several types of addiction (most notably nicotine addiction) and other compulsive behaviors. Here, we report an SAR exploration of a series of biphenyl-N-[4-[4-(2,3-substituted-phenyl)piperazine-1-yl]alkyl]carbamates, a novel class of molecules that had shown promising activities at the FAAH-D3R target combination in preliminary studies. We have rationalized the structural features conducive to activities at the main targets and investigated activities at two off-targets: dopamine receptor subtype D2 and endocannabinoid receptor CB_1. To understand the unexpected affinity for the CB₁ receptor, we devised a 3D-QSAR model, which we then prospectively validated. Compound 33 was selected for PK studies because it displayed balanced affinities for the main targets and clear selectivity over the two off-targets. 33 has good stability and oral bioavailability and can cross the blood-brain barrier.

Full text of DOI:10.1021/acs.jmedchem.6b01578

Subscriber access provided by University of Newcastle, Australia
Article
Design, synthesis, structure-activity relationship studies, and
three-dimensional quantitative structure-activity relationship (3D-
QSAR) modeling of a series of O-biphenyl carbamates as dual
modulators of dopamine D3 receptor and fatty acid amide hydrolase
Alessio De Simone, Debora Russo, Gian Filippo Ruda, Alessandra Micoli,
Mariarosaria Ferraro, Rita Maria Concetta Di Martino, Giuliana Ottonello, Maria
Summa, Andrea Armirotti, Tiziano Bandiera, Andrea Cavalli, and Giovanni Bottegoni
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01578 • Publication Date (Web): 09 Feb 2017
Downloaded from http://pubs.acs.org on February 9, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
eryDesign, synthesis, structureꢀactivity relationship
studies, and threeꢀdimensional quantitative structureꢀ
activity relationship (3DꢀQSAR) modeling of a
series of Oꢀbiphenyl carbamates as dual modulators
of dopamine D3 receptor and fatty acid amide
hydrolase
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
†
2
2†
1†
Alessio De Simone, Debora Russo, Gian Filippo Ruda, Alessandra Micoli, Mariarosaria
1
1
2
2
Ferraro, Rita Maria Concetta Di Martino, Giuliana Ottonello, Maria Summa, Andrea
2
2
1,3
1,4*†
Armirotti, Tiziano Bandiera, Andrea Cavalli, Giovanni Bottegoni.
1
2
CompuNet, Istituto Italiano di Tecnologia, 16163 Genova (GE), Italy; PharmaChemistry,
3
Istituto Italiano di Tecnologia, 16163 Genova (GE), Italy; FaBit, Università di Bologna, 40127
4
Bologna (BO), Italy; BiKi Technologies, Genova (GE), Italy.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
We recently reported molecules designed according to the multiꢀtargetꢀdirected ligand paradigm
to exert combined activity at human fatty acid amide hydrolase (FAAH) and dopamine receptor
subtype D3 (D3R). Both targets are relevant for tackling several types of addiction (most notably
nicotine addiction) and other compulsive behaviors. Here, we report an SAR exploration of a
series of biphenylꢀNꢀ[4ꢀ[4ꢀ(2,3ꢀsubstitutedꢀphenyl)piperazineꢀ1ꢀyl]alkyl]carbamates, a novel
class of molecules that had shown promising activities at the FAAH – D3R target combination in
preliminary studies. We have rationalized the structural features conducive to activities at the
main targets, and investigated activities at two offꢀtargets: dopamine receptor subtype D2 and
endocannabinoid receptor CB To understand the unexpected affinity for the CB receptor, we
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
.
1
devised a 3DꢀQSAR model, which we then prospectively validated. Compound 33 was selected
for PK studies because it displayed balanced affinities for the main targets and clear selectivity
over the two offꢀtargets. 33 has good stability and oral bioavailability, and can cross the bloodꢀ
brain barrier.
ACS Paragon Plus Environment

Page 3 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Multiꢀtargetꢀdirected ligands (MTDLs) are small organic molecules purposely designed to
1
interact with multiple biological targets. Developed within the framework of polypharmacology,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
the basic concept behind this relatively novel but already consolidated approach is quite simple:
when a single molecule modulates multiple selected targets, it can have synergistic efficacy,
milder sideꢀeffects, and a lower probability of triggering resistance or drugꢀdrug interactions.
2
ꢀ3
MTDLs may thus overcome the limitations of the traditional “one drug, one target” approach.
MTDLs can be seen as a straightforward way to put into practice the insights of network
medicine, which approaches the etiology and, possibly, the treatment of diseases as perturbations
4
of complex and interconnected signaling pathways. Thus, research on MTDLs has traditionally
5
ꢀ6
7ꢀ8
focused on treating complex and multifaceted conditions such as neurodegeneration, cancer,
9
and infectious diseases.
We recently began investigating the possibility of developing MTDLs to treat tobacco addiction.
Responsible for over five million deaths every year, tobacco smoking is the single most common
10
cause of avoidable death in Western countries. Tobacco contains harmful chemicals that are
directly connected to about 33% of all cancer deaths, including about 90% of all cases of lung
1
1
cancer, the most deadly form of cancer. However, the actual addiction to tobacco is caused by
12ꢀ13
the alkaloid nicotine.
Nicotine elevates the dopaminergic tone in specific areas of the brain,
including the ventral tegmental area (VTA) and the nucleus accumbens (NAc). These areas
belong to the mesocorticolimbic system, a brain circuit that is critical in natural and drugꢀ
induced reward and motivated behavior. This is nicotine’s best understood effect, and it is
mediated by interaction with nicotinic acetylcholine receptors (nAChRs) in the brain.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Nicotine’s effects have also been associated with alterations in most of the main neurotransmitter
systems, including biogenic amines, amino acid transmitters such as glutamate and gammaꢀ
1
4
aminobutyric acid (GABA), opioids, and endocannabinoids. For example, the activity of
dopamine neurons in the VTA is regulated by nAChRꢀexpressing glutamatergic and GABAergic
neurons. In this circuit, the balance between the stimulating and repressing actions of the two
neurotransmitters is crucial for modulating dopamine levels in the NAc. Moreover, presynaptic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
CB receptors located on GABAergic fibers or on glutamatergic interneurons have been shown
1
to abolish GABAꢀdependent tonic inhibition of dopamine neurons, facilitating nicotine
15ꢀ16
actions.
This profound and widespread modification of multiple neurotransmitter levels is
conducive to a complex behavioral condition dominated by withdrawal symptoms, craving,
drugꢀseeking, and reinforcing effects. This complexity may explain why none of the currently
17
proposed pharmacological treatments for nicotine addiction have longꢀlasting efficacy.
Varenicline, a nicotine receptor partial agonist, and Bupropion, a norepinephrineꢀdopamine
reuptake inhibitor and nicotine receptor antagonist, are marketed drugs for the treatment of
18
nicotine addiction.
These molecules have shown promising results in counteracting
nicotine cravings and withdrawal symptoms. However, as mentioned, their effectiveness in
promoting longꢀterm abstinence avoiding relapse is quite limited. Furthermore, their mechanism
of action, based on directly interfering with the cholinergic transmission in various brain areas
connected to reward and gratification, is likely responsible for the mild to intense severity of the
side effects (e.g., eating disorders, suicidal thoughts) associated with treatments based on these
1
9
molecules. For these reasons, nicotine addiction remains a largely unmet medical need. Hence,
our idea of tackling nicotine addiction by means of MTDLs that can concurrently modulate the
dopamine receptor subtype D3 (D3R) and the human fatty acid amide hydrolase (FAAH), likely
ACS Paragon Plus Environment

Page 5 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
eliciting enduring effects. D3R is a Gꢀproteinꢀcoupled receptor that is predominantly expressed
in the mesocorticolimbic system. For many years, D3R has been the focus of synthetic chemistry
2
0ꢀ21
campaigns to develop drugs to treat addiction and compulsive behaviors.
Specific
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
overexpression of D3R has been observed in nicotineꢀexposed brains and is a common feature of
2
2
drug addictions. Earlier studies focused on developing antagonists, while more recent
2
3
investigations have sought to develop potent and selective partial agonists. Pharmacologically,
partial agonists are more appealing because they provide a functional response, which depends
on dopamine levels in the mesolimbic area. In the absence of dopamine (after withdrawal), a
partial agonist would act as a functional agonist with lower intrinsic efficacy. In the presence of
the natural substrate, it would functionally antagonize dopamine binding to D3R, preventing the
reinstatement of nicotineꢀinduced signaling circuits. In contrast, pure D3R antagonists are of
limited use in alleviating withdrawal symptoms, but they are more effective in reducing craving,
2
4ꢀ25
drugꢀseeking, relapse, and reinforcement effects.
FAAH inhibition has only recently been
2
6
considered as a potential strategy for treating addiction. FAAH’s physiological role is to
terminate the endocannabinoid (EC) signaling by cleaving anandamide (AEA), one of the
2
7
agonists of the endogenous CB receptor (CB R). Interestingly, while considerable evidence
1
1
has emerged that directly links the endocannabinoid system to nicotine addiction, FAAH’s role is
2
8
likely to be mediated by nonꢀcannabinoid signaling pathways. FAAH is also responsible for
cleaving two more substrates: oleoylethanolamide (OEA) and palmitoylethanolamide (PEA),
which are endogenous agonists of the nuclear receptor PPARꢀα. Activating PPARꢀα stimulates
the activation of tyrosine kinases, which in turn leads to the phosphorylation and negative
regulation of neuronal nAChR. Inhibiting FAAH in order to maintain artificially high levels of
OEA and PEA thus appears to be a reasonable strategy for pharmacologically counteracting the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
29
effects of nicotine. Furthermore, regionꢀspecific FAAH inhibition leads to increased levels of
EC, with the potential of modulating in a more effective way nicotineꢀrelated behaviors as
compared to effects observed after ubiquitous activation of CB Rs by direct agonists, suggesting
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
that selective alterations of FAAH activity in limbic regions of the brain could have a key role in
2
6
the development of addictive behaviors. Increased AEA levels due to FAAH inhibition can
likely activate CB receptors on glutamatergic neurons that, in turn, negatively modulate
1
dopamine release in NAc, interfering with the creation of the addiction cycle. Last, the activation
of CB R has been associated to anxiolytic effects, which can be useful in the treatment of
1
30
negative emotional states during withdrawal. Inhibition by FAAH inhibitor URB597 was
recently reported to prevent selfꢀadministration and cueꢀ and substanceꢀinduced reinstatement in
2
6
animal models of nicotine addiction. In 2014, we reported the first set of rationally designed
compounds endowed with potent in vitro activities against these two targets. Our compounds
3
1ꢀ32
turned out to be potent FAAH inhibitors and D3R partial agonists.
While the Oꢀnaphthyl
carbamate series of compounds, exemplified by 1 (see Figure 1), was fairly selective with respect
to the investigated offꢀtargets, the biphenylꢀbearing molecules 2 and 3 displayed unexpected and
potent activity at CB R (EC 0.3 nM and 0.9 nM, respectively). Here, we report on an extensive
1
50
SAR exploration of a series of biphenylꢀNꢀ[4ꢀ[4ꢀ(2,3ꢀsubstitutedꢀ phenyl)piperazineꢀ1ꢀ
yl]alkyl]carbamates, which helped identify the structural drivers for the desired activities at the
two targets. This study also helped rationalize activities at the investigated offꢀtargets, pointing
to potential strategies for designing more selective compounds.
ACS Paragon Plus Environment

Page 7 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Previously reported Oꢀbiphenyl carbamates with dual D3R and FAAH activity
CHEMISTRY
The syntheses of the compounds are largely based on wellꢀdescribed reactions. Scheme 1
reports the synthetic pathway used to prepare the final carbamate derivatives. Amines with
general structure 7 were obtained by alkylation of arylpiperazine derivatives 5 with different
commercially available bromoalkylphthalimides 6 (Scheme 1, see Supporting Information for
details). The obtained phthalimides were deprotected by treatment with hydrazine. In the final
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
step, the obtained amines were activated as isocyanates by reaction with a slight excess of Boc
anhydride and 4ꢀdimethylamino pyridine (DMAP) in acetonitrile. These were then reacted with
the appropriate phenol derivatives to produce the various final carbamates. An alternative
procedure involved activating the amine derivative with pꢀnitrophenylchloroformate and then
reacting with the phenolic derivatives.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Synthetic route to compounds 16-29, 32-37, and 40-42
Reagents and conditions: a) CH CN, K CO , reflux, 6 h. Yields: 83ꢀ99%; b) Hydrazine monohydrate, MeOH,
3
2
3
8
0 °C, 2 h, then 2N HCl, 1 h. Yields: 53ꢀ88%; c) ArOH, (Boc) O, DMAP, CH CN, rt, 23 h. Yields: 7ꢀ50% d)
2 3
ArOH, pꢀnitrophenylchloroformate, DIPEA, DMA:DCM 1:1, rt, overnight. Yields: 10ꢀ26%.
For the synthesis of the nonꢀcommercially available phenolic derivatives, different
procedures were applied and are outlined in Scheme 2. Biphenyl alcohols of general formula 10
3
3
were prepared with a modified Suzuki crossꢀcoupling reaction between commercially available
aryl halides and boronic acid derivatives 8 and 9, in the presence of a catalytic amount of
palladium acetate in a medium of water and ethylene glycol monomethyl ether (EGME). This
procedure provided the desired products in high yield and with good reproducibility. When R is
3
a methoxy group, this procedure is ineffective for the steric hindrance of this group on the
ACS Paragon Plus Environment

Page 9 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
halogen atom. In these attempts, PdCl (dppf) was used as a catalyst (Scheme 2, see Supporting
2
Information for details).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Preparation of noncommercially available phenolic derivatives
Reagents and conditions: a) Pd(OAc) , K CO , EGME/H O, rt, 5 min Yields: 58ꢀ97%. b) PdCl (dppf), K CO ,
2
2
3
2
2
2
3
1
,4ꢀDioxane: Water 3:1, 90 °C, 3 hours. Yield: 48% (average of two attempts).
To synthesize analogues bearing an alkene in the linker (30-31), the synthesis started from
transꢀ1,4ꢀdichloroalkene 12 and potassium phthalimide 11 to yield intermediate 13 (Scheme 3).
Then, reaction with arylpiperazine 5 and deprotection with hydrazine afforded the desired amine
15 in good yields. The amine was used in a reaction with biphenolic derivatives 10b-c (prepared
using the same conditions reported in Scheme 2) in the presence of Boc O and DMAP to obtain
2
the carbamate derivative 30-31, as shown in Scheme 3.
Scheme 3. Synthetic route to alkenyl derivatives 30 and 31
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cl
O
Cl
N
b
a
N
Cl
Cl
Cl
O
O
-
O
N
K
+
O
N
+
N
13
1
1
12
O
14
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
HO
O
NH2
c
1
0b
NH2
Cl
N
O
Cl
N
O
Cl
N
O
Cl
N
H
3
0
d
HO
O
N
NH2
NH
2
F
15
NH
2
Cl
N
O
1
0c
O
Cl
N
N
O
F
H
3
1
Reagents and conditions: a) DMF, rt, under nitrogen atmosphere, overnight. Yield: 67%. b) 1ꢀ(2,3ꢀ
.
dichlorophenyl)piperazine, K CO , MeCN, reflux, 6 hours. Yield: 83%. c) NH NH H O, MeOH, then 2N HCl,
2
3
2
2
2
reflux, 3 hours. Yield: 88%. d) Boc O, DMAP, MeCN, rt, 20ꢀ25 hours. Yields: 30: 16%; 31: 22%.
2
RESULTS AND DISCUSSION
The newly synthesized compounds fit into the general structure reported in Figure 2.
Figure 2. General structure of the newly synthesized dual D3R FAAH modulators
ACS Paragon Plus Environment

Page 11 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
We sought to characterize the SAR for this class of compounds by introducing different
substituents on the aromatic ring of the arylꢀpiperazine and investigating both meta (m-)
biphenylꢀbearing and para (p-) biphenylꢀbearing compounds. We also considered the role of the
carboxamide in the distal ring of the O-biphenyl moiety. Finally, we studied a limited set of
variants by exploring substituents of the proximal ring in the biphenyl group as well as the role
of the alkyl spacer. We evaluated the biological activities of the reported compounds as FAAH
inhibitors and D3R modulators by means of a fluorescent assay and a cellular functional assay,
respectively. To manage our resources, we only assessed compounds for selectivity for D2R and
CB R if they were also characterized by reasonable activities at the main targets (FAAH IC <
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
50
6
0 nM, D3R EC < 100 nM, and no more than two orders of magnitude less potent than 4; D3R
50
efficacy < 75% as compared to 300 nM dopamine, as we were interested in developing D3R
partial agonists). Considering our starting point in the biphenyl series, we sought an initial
selectivity ratio with respect to the investigated offꢀtargets of at least one order of magnitude.
The results are reported in Table 1.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 1. Results of the SAR studies.
Structure
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
D3R
%
Human
FAAH
IC50 (nM)
0.6
CB
1
D3R
D2R
EC50 (nM)
13.0
D2R%
EC50
(nM)
14.0
0.9
HCl
salt
EC50
(nM)
Effica
a
b
b
Compound
R
1
R
2
X
R
3
R
4
cy
Efficacy
c
2
Cl
Cl
H
Cl
Cl
CH
2
H
H
H
H
H
H
H
CONH
CONH
CONH
CONH
CONH
CONH
H
2
2
2
2
2
2
no
no
no
no
no
no
yes
1.0
6.5
56
52
22
39
46
92
c
3
CH
2
CH
2
1.6
45
1
1
6
7
OMe
OMe
H
CH
2
1.8±0.4
2.3±0.3
1.3±0.1
1.1±0.2
4.1±0.7
19.0
14.0
2.1
60
54.2
1.4
H
CH
2
CH
2
34
56.4
2.3
18
H
CH
2
>90
77
1
2
9
0
H
H
CH
2
2
CH
2
2
3.7
Cl
Cl
CH
CH
240.0
>90
c
4
Cl
Cl
CH
2
CH
2
H
CONH
2
no
0.6
1.0
56
13.0
22
14.0
ACS Paragon Plus Environment

Page 13 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
21
Cl
H
Cl
CH
2
2
H
H
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
CONH
H
2
2
2
2
2
2
2
2
2
2
2
no
no
yes
no
no
no
no
no
no
no
no
no
yes
yes
yes
no
no
0.5±0.2
2.8±0.3
10.7±1.2
3.0±0.6
2.7±0.4
3.5±0.4
1.0±0.2
2.7±0.1
63.0±20.0
1.2±0.1
0.6±0.2
3.3±0.6
0.9±0.1
2.6±0.5
39±7.4
0.9
7.4
77
64
57
74
69
88
74
83
52
39.9
21.5
53.0
49.3
56
46
48
18
6.1
2.6
320
52
2
2
2
2
3
4
OMe
OMe
Me
CH
H
CH
2
CH
2
CH
2
CH
2
CH
2
CH
2
CH
2
CH
CH
CH
CH
CH
CH
CH
2
2
2
2
2
2
2
H
0.9
H
H
4.0
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
H
CF
3
H
11.0
3.6
2
2
6
7
H
H
Cl
H
Cl
Cl
H
F
18.6
4.4
13.7
40.4
15
28
Cl
OMe
OMe
H
29
30
31
OMe
Cl
3.9
10.4
147.0
66.2
66
22
51
65
48
110
20
d
Cl
Cl
Cl
Cl
H
CHCH
CHCH
37.8
6.9
A
d
Cl
F
A
9.2
32
Cl
CH
2
H
17.0
18.0
14.0
330.0
130.0
1200.0
43
65
14.7
70
33
34
35
Cl
CH
2
CH
2
H
H
135.0
1500
OMe
OMe
Me
CH
2
H
H
82
H
CH
2
CH
2
CH
2
CH
CH
CH
2
H
H
85
36
H
2
H
H
7.6±1.5
20.7±7.4
>90
>90
37
H
CF
3
2
H
H
a
b
c
d
n=3; as of 300 nM of dopamine; activity (FAAH, D3R, CB
1
) and efficacy (D3R) data as reported in reference 28; Antagonist.
SAR Studies. We further investigated the properties of the mꢀbiphenyl series by introducing
the classic 2ꢀmethoxy substituent in the arylꢀpiperazine in the presence of threeꢀ (16) and fourꢀ
2
0
methyleneꢀunit (17) linkers. With respect to 2 and 3, their 2,3ꢀdiꢀClꢀsubstituted counterparts,
both derivatives maintained potent activities at the main targets (Table 1), with the 2ꢀmethoxy
substituent being only slightly detrimental for D3R activity. In line with results reported for 2
and 3, 16 and 17 were also potent CB R agonists with 1.4 nM and 2.3 nM EC50 values,
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
respectively. Derivatives 18 and 19, displaying a naked pendant aromatic ring, maintained potent
and balanced activities at both targets, but their D3R activation profile switched significantly
toward a full agonist effect. Eliminating the carboxamide in the distal mꢀbiphenyl ring in 20
severely reduced the activity against D3R.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The results for 4 suggested that a potent dual modulator profile devoid of selectivity problems
could be achieved by introducing a pꢀbiphenyl group at the Oꢀside of the carbamate. We
therefore turned our attention to derivatives bearing this moiety. Reducing the length of the
linker of 4 to three methylene units (21) was not detrimental for activities at the main targets
(Table 1). However, while over 50ꢀfold selectivity over D2R could be obtained, 21 displayed 6.1
nM EC at CB R. Studying the effect of substituents in the aromatic ring of the arylꢀpiperazine,
50
1
we observed that, when we introduced a 2ꢀmethoxy group in combination with a threeꢀ
methyleneꢀunit linker, compound 22 maintained potent activities at FAAH and D3R but did not
achieve selectivity toward both the offꢀtargets. When the same substitution was coupled with a
longer linker (23), activity at D3R was maintained and activity at FAAH was only slightly
affected. Notably, activity at CB R decreased by up to 320 nM. Introducing smaller substituents
1
at position 2, as with the 2ꢀmethylꢀbearing (24) and 2ꢀtrifluoromethylꢀbearing (25) derivatives,
did not affect activity at the two targets but did lead to increased efficacy at D3R, shifting
activity closer to a full agonist profile. Compound 24 had moderate 18ꢀfold selectivity toward
D2R and acceptable selectivity with respect to CB R (4.0 nM EC at D3R and 52 nM EC at
1
50
50
CB R). Consistently, and in line with the results obtained with the mꢀbiphenyl derivatives,
1
compound 26, bearing a naked phenyl ring, behaved almost like a full agonist (88% efficacy).
We also introduced some substituents in the proximal ring of the biphenyl moiety. To address
possible solubility issues, we introduced a fluorine atom, which endowed 27 with subꢀnanomolar
ACS Paragon Plus Environment

Page 15 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibitory potency against FAAH. However, activity at D3R (18 nM EC ) was almost 20 times
50
less than that observed for 4. Most importantly, this derivative displayed no selectivity with
respect to the offꢀtargets. As previously reported by MorenoꢀSanz and coworkers, a methoxyl
group introduced in the proximal ring of the biphenyl should further improve inhibitory potency
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3
4
at FAAH without compromising brain penetration. Indeed, 28 turned out to be a potent FAAH
inhibitor (2.7 nM IC ) but its efficacy at D3R shifted toward pure agonism. Conversely, when
5
0
this substitution was coupled with a 2ꢀmethoxyl substituent in the arylꢀpiperazine (29), the partial
D3R agonist profile was restored but inhibitory activity at FAAH decreased quite substantially
(
63 nM IC ). Introducing an unsaturation in the linker (30) decreased activity at D3R (EC 37.8
5
0
5
0
nM) while maintaining activity at FAAH (IC50 1.2 nM). Introducing a fluorine group in the
proximal biphenyl ring (31) restored activity at D3R in the single digit nanomolar range.
Interestingly, 30 and 31 behaved as D3R antagonists. As expected in light of classic studies of
3
5
D3R modulators, the presence of a central trans doubleꢀbond in the linker lowered the affinity
for D2R, which decreased with respect to the saturated analogues from 23 nM (4) to 147 nM (30)
and from 13.7 nM (27) to 66.2 nM (31). However, neither 30 nor 31, with EC values of 20 nM
5
0
and 9.2 nM, respectively, displayed selectivity toward CB R.
1
To assess the effects of the carboxamide in the distal ring of the biphenyl region, we
synthesized and tested several compounds lacking this substituent. Activity data are reported in
Table 1 and systematically compared to those of the carboxamideꢀbearing analogues. 2,3ꢀdiꢀClꢀ
substituted arylꢀpiperazine compounds 32 and 33 were characterized by potent activities at
FAAH in line with those of 21 and 4. In 32, activity at D3R decreased with respect to 21 (see
Table 1); consistently, in 33, D3R EC increased from 1.0 nM to 18 nM. In terms of selectivity,
50
an acceptable ratio could be achieved for D2R. Remarkably, activity at CB R dropped
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
substantially in 32 (EC 70 nM) with respect to 21 and decreased in 33 by over two orders of
5
0
magnitude (EC 1500 nM) with respect to 4. The 2ꢀmethoxy substituted compound 34 preserved
50
potent and balanced activities at both main targets, but its efficacy shifted toward an agonist
profile (D3R efficacy 82%). In the fourꢀmethyleneꢀunit linker analog 35, activities decreased at
both targets, with FAAH IC equal to 39 nM and D3R EC to 330 nM (efficacy 85%). In
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
50
agreement with the behavior of 35, introducing a 2ꢀmethylꢀ (36) or a 2ꢀtrifluoromethylꢀ (37)
group in the pendant aromatic ring of the carboxamideꢀlacking derivatives led to a significant
drop in D3R activity (Table 1).
In light of these results, we propose a first rationale for the chemical features required for a
balanced and potent activity at FAAH and D3R. Substitutions on the pendant aromatic ring of
the arylꢀpiperazine were required for a partial agonist profile against D3R. In the presence of a
saturated linker, bulkier substituents shifted the efficacy profile toward antagonism. Regarding
the linker, the fourꢀmethyleneꢀunit linker resulted in lower activation of D3R than the threeꢀ
methyleneꢀunit linker, and was thus preferred in order to have partial agonist derivatives. By
introducing a central trans doubleꢀbond to increase the system’s rigidity, we obtained D3R
antagonists. These results are in qualitative agreement with previous studies by Newman and
colleagues on how substitutions at the terminal aromatic ring, in combination with different
3
6
linkers, can affect efficacy. Regarding FAAH, a longer linker, especially when combined with
a bulkier substituent in position 2, led to a drop in activity. This may be due to a more
constrained fit in the binding pocket region known as the acyl chain binding channel. At the Oꢀ
side of the carbamate, the distal phenyl ring could be placed in both the m- and p-positions of the
proximal ring without affecting the potencies against the two main targets. For the p-biphenyl
derivatives, a substituent in the meta position of the proximal phenyl ring could be inserted
ACS Paragon Plus Environment

Page 17 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
without dramatically altering activities. The carboxamide substituent on the distal ring revealed a
beneficial role in the affinity for D3R, possibly because of an Hꢀbond interaction with Thr92,
31
which was consistently suggested by our previous docking experiments. However, while
consistent across the series, the loss in D3R affinity due to the deletion of the carboxamide was
less severe for compounds with the 2,3ꢀdiꢀClꢀsubstitution in the arylpiperazine. In FAAH, the
activity of our derivatives was only marginally affected by the presence of the carboxamide,
although the reported activities of known inhibitors URB524 (naked biphenyl group in the O-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
37
side of the carbamate, FAAH IC 63 nM) and URB597 (carboxamide substituted biphenyl
5
0
38
ring, FAAH IC 4.6 nM) had suggested this would not be the case. This was probably due to
5
0
the presence of an elongated substituent at the Nꢀside of the carbamate, which is here the key
3
9
structural driver for the FAAH inhibitory effect. These SAR considerations are summarized in
Figure 3.
Figure 3. Summary of the results of SAR studies performed on different regions of the
synthesized derivatives.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Selectivity at the off-targets. Twelve derivatives that showed a promising activity profile for
the main targets were assessed for selectivity for D2R and CB R. For D2R, some general
1
considerations can be outlined. With the sole exception of 27, in which affinities for D3R and
D2R were almost equivalent, our compounds had greater affinity for D3R than for D2R. This is
not surprising, given that the original concept for these compounds was to introduce a carbamate
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
0
as a conservative variation of a classic D3Rꢀselective pharmacophore. The aryl carbamate
protrudes outside the orthosteric binding site in a region defined by the extracellular loop 3 and
transmembrane helices I, II, and VII (secondary binding site). Selectivity can be obtained by
40
exploiting subtle differences in both side chain and backbone rearrangements in this region.
Selectivity over D2R decreased: i) in pꢀbiphenyl derivatives with respect to their mꢀbiphenyl
counterparts, ii) in molecules bearing a shorter linker, and iii) in 2ꢀmethoxy substituted
molecules (Table 1). Again, this is consistent with previously reported studies indicating that
selectivity for D3R over D2R is affected by the nature and (most importantly) orientation of the
3
6
portion of the molecule in contact with the secondary binding site. In turn, this orientation
varies in the two receptor subtypes depending on the substituents in the arylꢀpiperazine and the
linker length. While the present study did not match the remarkable D2R/D3R selectivity ratio
that characterized our previously described derivatives 2 and 3, as well as classic arylamideꢀ
4
1
based D3R modulators reported in the literature, several compounds obtained did display
sufficient selectivity.
From a structural perspective, the CB R activity data was more puzzling. A threeꢀunit linker
1
usually led to increased activity at CB R so long as it was coupled to the carboxamide substituent
1
in the distal phenyl ring. The presence of this substituent was the structural feature exerting the
more evident effect, as suggested by activity data on 32 and 33 (Table 1). We hypothesized that
ACS Paragon Plus Environment

Page 19 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
42
these results could be interpreted within the framework of the classic SAR of CB R agonists.
1
9
Figure 4 reports the structures of four prototypical CB R agonists, namely ꢁ ꢀTHC (38, CB R
1
1
EC 10.2 nM), HUꢀ210 (39, CB R EC 0.73 nM), JWHꢀ051 (40, CB R EC 1.2 nM), and
50
1
50
1
50
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
JWHꢀ133 (41, CB R EC 667 nM).
1
50
Figure 4. Structures of four CB R agonists
1
No crystal structure of an agonistꢀbound CB R was available when we conducted our study;
1
however, several reports have suggested that the aliphatic chain of classic CB R agonists lays
1
43ꢀ44
perpendicular to the plane of the tricyclic system.
This is also consistent with the pose
9
adopted by ꢁ ꢀTHC docking studies carried out at the recently reported crystal structure of
45
antagonistꢀbound CB R. Using as a template a conformation of the subꢀnanomolar agonist 39
1
that reflected this arrangement, we performed an APFꢀbased structural alignment (see
4
6
Experimental Section) and generated an overlapping conformation of 4 (Figure 5A). The
resulting structural alignment, with the biphenyl group matching the tricyclic system, is
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
particularly evocative, as the classic SAR of THC analogs could be applied to our molecules.
First, the presence of a hydroxyl group in position 1 (38) is very important for eliciting potent
activity. Another hydrogen bond donor/acceptor group in position 11 can further increase
activity (39) or can maintain activity if the hydroxyl group in 1 is removed (40). The removal of
both hydrogen bond donor groups is detrimental for activity, as exemplified by 41. This would
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
explain the strong effect of the carboxamide substituent. Second, the aliphatic chain of CB R
1
agonists can be modified quite substantially without compromising activity, even if elongated or
decorated with bulky pendants. Hence, it stands to reason that linker and arylꢀpiperazine could be
lodged in the corresponding region without compromising activity. Finally, it has been reported
that CB R activity is modulated by the planarity of the tricyclic system, with more planar
1
systems being less active and more susceptible to the effect of the substitutions in positions 1 and
1
1, respectively. To further test the hypothesis that CB R activity can be explained by the classic
1
SAR of THC analogs, we synthesized three novel compounds (42ꢀ44), introducing a flat
aromatic group at the Oꢀside of the carbamate. Potent and selective D3R modulators displaying
47
rigid tricyclic systems have already been reported.
Table 2. Activity data of the carbazoleꢀbased analogs
Structure
Human
CB1
FAAH
D3R
D3R%
D2R
D2R%
a
b
b
Compound
R
1
R
2
X
IC50 (nM)
EC50 (nM)
Efficacy
EC50 (nM)
Efficacy
EC50
ACS Paragon Plus Environment

Page 21 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(nM)
380
4
4
2 (15197)
3 (15387)
Cl
Cl
H
Cl
Cl
CH
2
4.8±2.0
8.0±0.8
150.0
120.0
12.0
>90
>90
66
CH
2
2
CH
2
2
370
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
44 (15959)
OMe
CH
CH
37.0±10.0
57.9±18.8
44
2300
a
b
n=3; as of 300 nM of dopamine;
As expected, CB R activity was significantly reduced in all three derivatives: 42, EC 380
1
50
nM; 43, EC 370 nM; and 44, EC 2300 nM. However, 42 and 43 displayed reduced activity at
5
0
50
D3R (D3R EC 150 nM and 120 nM, respectively), and 44 coupled an only mild selectivity over
5
0
D2R (D2R/D3R selectivity ratio equal 4.8) with a reduced activity against FAAH (FAAH IC50
3
7 nM). Therefore, we did not investigate further carbazoleꢀbearing derivatives.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 23 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 5. Computational studies to rationalize activity at CB R. A) APFꢀbased superimposition
1
of 4 and 39, using the putative receptorꢀbound conformation of the latter as a rigid template; B)
3D alignment of the 18 molecules included in the training set; C) Equipotential contours of the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Hꢀbond acceptor (red) and Hꢀbond donor (blue) APF components; D) Equipotential contours of
the sp2 carbon atom hybridization APF component (white); E) Equipotential contours of the
lipophilicity APF component (green); F) Correlation plot between pIC50exp and pIC50pred. The
structures of the predicted compounds (red dots) are reported explicitly.
3D-QSAR Studies. At this point, we attempted a 3DꢀQSAR analysis to systematically
describe, at the structural level, the CB R activity in our series and to prospectively test our
1
hypothesis. This 3DꢀQSAR analysis was carried out by means of the Atom Property Fields
46
(
APF) methods originally developed by Totrov (see Experimental Section for details). A set of
18 molecules, encompassing 2ꢀ4, 16ꢀ17, 21-24, 27, 29ꢀ33, and 42ꢀ44, namely all molecules
which had been tested for CB R activity, were automatically aligned to the previously generated
1
conformation of 4 (Figure 5B). For each molecule, we generated a set of seven threeꢀdimensional
continuous potentials, representative of seven molecular properties (namely, presence of a
2
hydrogen bond donor, presence of a hydrogen bond acceptor, sp carbon atom hybridization,
charge, lipophilicity, size and electropositivity/negativity), approximated on a set of regularly
spaced grids. Each molecule could thus be expressed by 126 descriptors: the pseudoꢀenergy
values generated by the molecule’s fit in each of the 7 property fields of the 18 compounds in the
training set. Each descriptor was assigned a weight by means of a partial least square (PLS)
fitting procedure in order to reproduce experimental activity data. Eventually, we selected 4
latent vectors (see Table 3).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
Table 3. Squared correlation coefficients (R ) and leaveꢀoneꢀout squared correlation coefficients
2
(
Q ) according to the number of latent vectors.
2
a
2
b
b
No. of latent
vectors
R
RMSE
Q
RMSELOO
LOO
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0.10
0.82
0.87
0.92
0.94
0.95
0.97
0.98
0.99
0.99
1.04
0.46
0.39
0.30
0.26
0.24
0.19
0.15
0.10
0.09
0.05
0.74
0.74
0.79
0.82
0.82
0.82
0.83
0.81
0.81
1.1
0.54
0.57
0.51
0.46
0.46
0.47
0.46
0.48
0.48\
10
a
b
Root Mean Squared Error; Leaveꢀoneꢀout;
According to best practice, selecting roughly one latent vector for every 5 compounds in the
training set is a good heuristic ratio for building the model in order to avoid overfitting and to
4
8
achieve external validity. Furthermore, in this specific case, including more latent vectors
would not have substantially increased our model’s predictive power within the set (Table 3). By
ACS Paragon Plus Environment

Page 25 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
combining the activityꢀweighted maps from individual compounds, we can identify two key
features conducive to CB R activity: i) the presence of a hydrogen bond donor/acceptor in the
1
distal phenyl ring of the biphenyl group (Figure 5C) and ii) an extended and not completely flat
ring system (Figure 5D). This is consistent with our previously reported findings that compounds
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
bearing a naphthyl group at the Oꢀside of the carbamate display much lower CB R activities.
1
Given the structural homogeneity of the training set, other features (such as the hydrogen bond
donor/acceptor signal in carbamate region or the strong lipophilic signal generated by the arylꢀ
piperazine tail (Figure 5E)) are consistent with the data but not very informative. Table 4 reports
the experimental (pEC50ex), predicted (pEC50pred), and leaveꢀoneꢀout crossꢀvalidated (pEC50l.o.o.
pEC values for the training set.
)
50
Table 4. Experimental, Predicted, and LeaveꢀOneꢀOut CrossꢀValidated pEC Values for the
50
Training Set Compounds
Compound
pEC50exp
pEC50pred
8.90
pEC50LOO
8.72
2
3
4
9.04
9.52
7.85
8.85
8.64
8.21
9.50
9.36
7.73
7.71
1
6
7
9.06
9.18
1
8.40
8.23
21
8.51
8.68
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
2
2
2
3
4
8.58
7.17
7.28
7.82
6.96
7.70
8.04
7.15
5.82
6.42
6.43
5.63
7.98
7.30
7.63
7.76
7.19
7.74
7.74
7.12
6.48
6.73
6.02
5.33
7.50
7.35
7.72
7.76
7.36
7.67
7.32
6.91
6.70
7.01
5.75
5.10
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
27
29
30
31
32
33
42
43
44
2
The model was internally very consistent with a R equal to 0.92 (Figure 5F) and a root mean
squared error (RMSE) of 0.30. No compound emerged as a clear outlier. The leaveꢀoneꢀout
2
(
LOO) procedure for crossꢀvalidation returned a still predictive Q of 0.79 (RMSE 0.51, Table
ACS Paragon Plus Environment

Page 27 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
). Again, no predicted activity departed by more than one order of magnitude from the
experimental value.
Prospective validation of the 3D-QSAR model. Finally, in order to prospectively challenge
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
our hypothesis and confirm that the SAR for classic CB R agonists applies to our molecules, we
1
designed three novel compounds and used the APF–based 3DꢀQSAR model to predict their
activities at the CB R receptor. This is particularly relevant because our model was generated
1
from a limited number of molecules. Our design concept for compounds 45ꢀ47 was to introduce
small structural variations of 33 in order to retain its selectivity against CB R, avoiding hydrogen
1
bond donor/acceptor groups in the distal biphenyl ring while introducing substitutions in the
proximal one (see Table 5).
Table 5. Biological data of 3DꢀQSAR predicted and synthesized molecules
Structure
Human
FAAH
D3R
%
Predicted
CB
Experimental
CB
Experimental
CB
D3R
1
1
1
EC50
(nM)
a
b
Compound
R
1
R
2
X
R
3
R
4
pEC50
6.64
6.55
6.50
EC50 (nM)
220
pEC50
6.66
6.47
7.03
IC50 (nM)
4.3±1.3
6.0±0.8
efficacy
4
4
4
5 (19679)
Cl
Cl
H
Cl
Cl
CH
2
CH
2
CH
2
CH
2
2
2
OMe
F
H
H
H
1.66
3.7
37
32
6 (19659)
CH
CH
340
40
7 (16050)
OMe
OMe
93
43.0±12.0
80
a
b
n=3; as of 300 nM of dopamine;
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 75
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Substitutions in the proximal ring: i) are generally wellꢀtolerated in FAAH and help improve
3
4
the solubility profile and ii) are detrimental for CB R activity when coupled with a lack of
1
hydrogen bond donor/acceptor groups (Table 1). Figure 5F reports the fitting of 45ꢀ47 in the
global APF maps generated by the molecules of the training set. We then synthesized and tested
these compounds. Predictions were generally in good agreement with the experimental data (red
circles in Figure 5F). For 45 and 46, the experimental activities matched the predictions
remarkably well. Furthermore, these two molecules were endowed with potent and balanced
activities at the main targets with a clear partial agonist profile at D3R. Although their affinity
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
for CB R increased with respect to 33, the selectivity profile remained acceptable thanks to these
1
potency values, and actually improved in 46. In 47, while still rather accurate, the model
suggested a slightly less potent molecule (pEC50pred equal 6.50) than the experimental result
(
pEC50pred equal 7.03, Table 5). Here, the selectivity with respect to CB R was also compromised
1
by decreased potency at the main targets.
Preliminary pharmacokinetics characterization and selectivity profiling of 33. Given its
promising activity at the main targets, its physicoꢀchemical profile (molecular weight: 498; polar
surface area: 46.1; calculated logP: 5.48), and its selectivity profile at off targets known from
previous studies, we selected 33 as the new lead compound for further pharmacokinetics (PK)
studies and selectivity profile characterization. First, we assessed how 33 remained stable in rat
plasma for over an hour with an average t of 97±13 min (n=3). The compound was also stable
½
in rat liver microsomes (t > 60 min, n=3). PK analyses in rats showed that 33 possessed good
½
oral bioavailability and bloodꢀbrain barrier (BBB) penetration. After dosing 33 intravenously and
ꢀ1
orally, we monitored plasma concentrations for 8 hours. After intravenous dosing (3 mg kg ),
ꢀ1
we measured a maximal plasma concentration of 500 ng ml (5 minutes time point). The halfꢀ
ACS Paragon Plus Environment

Page 29 of 75
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
life of 33 for the elimination phase was 145 minutes. Compound 33 showed a volume of
ꢀ1
distribution of 23.7 l kg and disappeared from the systemic circulation with a clearance of 0.11
ꢀ1
ꢀ1
ꢀ1
l min kg . After oral administration (10 mg kg ), the maximal plasma concentration of 33 (54
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ꢀ1
ng ml ) was observed 120 minutes after dosing. The oral bioavailability of 33 was 21%. BBB
ꢀ1
penetration of this compound was also measured for the per os arm of the study (10 mg ml ): 33
ꢀ1
reached a maximum concentration of 37 ng ml of brain homogenate, 4 hours after oral
ꢀ1
administration; the total brain exposure (AUC) was quantified in 84.5 ng min ml (see also
*
Figure S1 in the Supporting Information). It can be calculated that 33 reaches a total brain
concentration of approximately 391 nM, more than adequate to effectively engage the target (see
SI for details).
In terms of selectivity, we reasoned that CB R and D2R were off targets selected based on the
1
compound’s alleged mechanism of action. Next, we decided to begin expanding the
polypharmacological profile of 33 investigating its activity at off targets selected based on the
compound structure. Displaying an arylꢀpiperazine, which is a privileged moiety for interacting
49
with aminergic receptors, we tested the activity of 33 at this class of GPCRs starting from beta
adrenergic receptors subtype β and β and serotonergic receptor subtypes 5ꢀHT and 5ꢀHT . In
1
2
1A
1b
binding experiments, 33 turned out to be selective with respect to both beta adrenergic receptor
subtypes and 5ꢀHT . Interestingly a moderate affinity for 5ꢀHT1A could be detected (60.4%
1b
inhibition of the binding of a labelled radioligand at 100 nM, see Supporting Information for
details). On the same line, as 33 bears an aryl carbamate, which is a warhead exploited by
several compounds to interact with hydrolases, we tested 33 for activity at monoacylglycerol
lipase (MAGL). We selected this classic serine hydrolase since it is responsible for over 80% of
the degradation of 2ꢀarachidonoylglycerol (2ꢀAG), another key mediator together with AEA in
ACS Paragon Plus Environment