Exploring the effectiveness of novel benzimidazoles as CB2 ligands: synthesis, biological evaluation, molecular docking studies and ADMET prediction?

Article abstract of DOI:10.1039/C8MD00461G

Herein we continued our previous work on the development of CB2 ligands, reporting the design and synthesis of a series of benzimidazole-containing derivatives that were explored as selective CB2 ligands with binding affinity towards both CB1 and CB2 receptors. Seven out of eighteen compounds exhibited preferential binding ability to CB2 over CB1 receptors with potencies in the sub-micromolar or low micromolar range. In particular, we identified two promising hit compounds, the agonist 1-[2-(N,N-diethylamino)ethyl]-2-(4-ethoxybenzyl)-5-trifluoromethylbenzimidazole (3) (CB2: K_i = 0.42 μM) and the inverse agonist/ antagonist 1-butyl-2-(3,4-dichlorobenzyl)-5-trifluoromethylbenzimidazole (11) (CB2: K_i = 0.37 μM). Docking studies also performed on other benzimidazoles reported in the literature supported the structure–activity relationship observed in this series of compounds and allowed the key contacts involved in the agonist and/or inverse agonist behaviour displayed by these derivatives to be determined. The in silico evaluation of ADMET properties suggested a favorable pharmacokinetic and safety profile, promoting the drug-likeness of these compounds towards a further optimization process.

Full text of DOI:10.1039/C8MD00461G

MedChemComm
View Article Online
View Journal | View Issue
RESEARCH ARTICLE
Exploring the effectiveness of novel
benzimidazoles as CB2 ligands: synthesis,
biological evaluation, molecular docking studies
and ADMET prediction†
Cite this: Med. Chem. Commun.,
2018, 9, 2045
a Elena Cichero,a Alì Mokhtar Mahmoud,b Alessandro Rabbito,b
*
Michele Tonelli,
Bruno Tasso,a Paola Fossa‡a and Alessia Ligresti‡b
Herein we continued our previous work on the development of CB2 ligands, reporting the design and syn-
thesis of a series of benzimidazole-containing derivatives that were explored as selective CB2 ligands with
binding affinity towards both CB1 and CB2 receptors. Seven out of eighteen compounds exhibited prefer-
ential binding ability to CB2 over CB1 receptors with potencies in the sub-micromolar or low micromolar
range. In particular, we identified two promising hit compounds, the agonist 1-[2-(N,N-diethylamino)ethyl]-
2-(4-ethoxybenzyl)-5-trifluoromethylbenzimidazole (3) (CB2: Ki = 0.42 μM) and the inverse agonist/
antagonist 1-butyl-2-(3,4-dichlorobenzyl)-5-trifluoromethylbenzimidazole (11) (CB2: Ki = 0.37 μM). Docking
studies also performed on other benzimidazoles reported in the literature supported the structure–activity
relationship observed in this series of compounds and allowed the key contacts involved in the agonist
and/or inverse agonist behaviour displayed by these derivatives to be determined. The in silico evaluation
of ADMET properties suggested a favorable pharmacokinetic and safety profile, promoting the drug-
likeness of these compounds towards a further optimization process.
Received 12th September 2018,
Accepted 7th October 2018
DOI: 10.1039/c8md00461g
rsc.li/medchemcomm
CB1 is widely expressed within the central nervous system
(CNS),4 where it mediates the psychotropic effects of
Introduction
Endocannabinoids and their receptors (CB1 and CB2) consti-
tute a large modulatory system that finely tunes the synaptic
neurotransmission and consequently a complex network of
physiological and pathological processes.1,2
(−)-trans-Δ9-tetrahydrocannabinol (THC).5
The failure of the CB1 receptor inverse agonist
rimonabant in 2009 (ref. 6) motivated researchers to explore
more deeply the presence and function of this receptor in pe-
ripheral non-neuronal tissues (adipose tissue, liver, gastroin-
testinal tract, pancreas, and skeletal muscles). Accordingly,
peripherally-restricted CB1 receptor antagonists/inverse ago-
nists have been shown to effectively reduce body weight, adi-
posity, insulin resistance and dyslipidaemia in obese animal
models.7
This widely distributed and differential expression of CB1
receptors both in the brain and in the periphery reflects the
complexity and can justify the variety of functions of
endocannabinoids.8
CB2 receptors are expressed at much lower levels in the
CNS compared to CB1 receptors. These receptors are predom-
inantly located in immune cells, although recent studies have
found the expression of genes that encode this receptor even
in the cerebellum9 and in microglia.10
Based on the ability of WIN-55,212-2 to activate unselec-
tively CB1 as well as CB2 subtypes, numerous studies
arose with the aim of designing a new series of analogues
with better selectivity and potency profiles towards the
The CB1 and CB2 cannabinoid receptors are G-protein
coupled receptors (GPCRs), which primarily couple with the
G proteins of the Gi and G0 classes.3 Receptor activation in-
hibits adenylyl cyclases and certain voltage-dependent cal-
cium channels as well as activates several mitogen-activated
protein kinases and inwardly rectifies potassium channels,
with some variation depending on the particular cell type.3
Activation of CB1 or CB2 receptors controls the cannabinoid
signal transduction pathways at different levels, exerting di-
verse consequences on cellular physiology, including synaptic
function, gene transcription, and cell motility.
a Dipartimento di Farmacia, Università degli Studi di Genova, V.le Benedetto XV,
3, 16132 Genova, Italy. E-mail: tonelli@difar.unige.it
b National Research Council of Italy, Institute of Biomolecular Chemistry,
Endocannabinoid Research Group, Via Campi Flegrei 34, 80078 Pozzuoli, (Na),
Italy
† Electronic supplementary information (ESI) available: Table 1S and 1H and 13
NMR spectra of compound 14 are reported. See DOI: 10.1039/c8md00461g
‡ Last authors.
C
This journal is © The Royal Society of Chemistry 2018
Med. Chem. Commun., 2018, 9, 2045–2054 | 2045

View Article Online
Research Article
MedChemComm
single CB2 receptor. Initially, research efforts were focused
on the indole scaffold, derived from the molecular simplifica-
tion of the reference compound WIN-55,212-2, leading to
aminoalkylindoles (AAIs), such as AM630 (Fig. 1, B). These
were considered as intriguing CB2 agonists confirming the
effectiveness of replacing the WIN-55,212-2 tricyclic system
with an indole core or, where appropriate, with other series
of heteroaromatic systems.
This motivated the researchers to discover many interest-
ing chemo-types, some of them still revolving around the in-
dole structure, while the other ones are characterized by dif-
ferent rings, allowing the evaluation of the reliability of the
bioisosteric approach.
Among the indole-based derivatives, a new series of com-
pounds were obtained by replacing the morpholine ring of
WIN-55,212-2 with a pyran one, and the 1-naphthyl residue
with a tetramethylcyclopropyl substituent (Fig. 2, C).
Fig. 2 Chemical structures and binding affinity values of known CB1/
CB2 indole-containing agonists.
These indolyl-ketones exhibited potent and selective ago-
nist activity at the CB2 receptor,11 even surpassing the bind-
ing affinity for the same receptor of WIN-55,212-2. The
subnanomolar Ki values of this pool of molecules 1a–13a
(Fig. 2, C) revealed the effectiveness of non-basic substituents
linked to position 1 of the indole core, as shown by the
methylpyran moiety of C versus the basic ethylmorpholine
chain of WIN-55,212-2. Notably, this proved to be true, inde-
pendent of the nature of substituents decorating the indole
ring. Concerning this issue, electron-donor groups examined
in positions 6 and 7 of the indole ring were able to enhance
the potency towards the CB2 receptor, (12a, R = 6-OCH3: CB2
Ki = 0.51 nM, CB1 Ki = 40 nM; 13a, R = 7-OCH3: CB2 Ki = 0.12
nM, CB1 Ki = 45 nM), leaving the low selectivity ratio
unaltered. In contrast, the substitutions with electron-
withdrawing groups (F, Cl and Br) in position 5 improved the
selectivity profile towards the CB2 receptor, ranging from 172
to 756. The same behavior was also maintained within those
analogues featuring in position 1 the ethylmorpholine chain
of the prototype (Fig. 2, D). Conversely, substitutions at posi-
tion 4 or 7 of the indole ring were detrimental when com-
pared with any variation involving position 2, which allowed
compounds with a better profile of both affinity and selectiv-
ity towards the CB2 receptor (19a, R = 2-CH3: CB2 Ki = 27 nM,
CB1 Ki > 10 000 nM) to be obtained. Starting from the afore-
mentioned indolyl-ketones, new different chains (Fig. 2, E)
were further explored, sometimes also varying the ketone
group in position 3 with other cycloaliphatic systems, leaving
the other ring positions unchanged.12 All the investigated al-
coholic or (thio)ethereal chains were characterized by suit-
able potency towards the CB2 receptor, in addition to a high
selectivity profile. The best requirement for an improved se-
lectivity index (SI) towards the CB2 receptor over CB1 was a
linker of two carbon atoms between the indole nitrogen and
an amino function (R = (CH2)2NIJCH3)2, SI > 5263); in con-
trast, the introduction of a longer alkyl spacer, moving from
the hydroxyethyl chain to the hydroxybutyl ones, was respon-
sible for a higher affinity for the CB2 receptor (R = (CH2)2OH,
SI > 179; (CH2)3OH, SI = 1047; (CH2)4OH, SI = 2545). It is
worth noting that the absence of a basic head group in R to
mimic the morpholine ring of WIN-55,212-2, in favor of a
thioethereal or a carbonyl function, had however led to pow-
erful CB2 agonists (R = (CH2)4SCH3, CB2: Ki 0.40 nM;
(CH2)4COCH3, CB2: Ki = 0.99 nM), even if with a reduced se-
lectivity versus the CB1 receptor.
While maintaining the indole core allowed so far the de-
sign of potent CB2 agonists, the search for other heterocyclic
rings allowed the discovery of a series of congeners featuring
agonist as well as inverse agonist activity, on the basis of spe-
cific substituents. This trend was previously observed in a se-
ries of isatin acylhydrazones discussed in the literature,13,14
and in benzimidazole skeletons proposed by AstraZeneca15,16
still exhibiting more potent and selective profiles towards the
human CB2 receptor in comparison with WIN-55,212-2.
In particular, among the benzimidazole derivatives, it is
possible to distinguish some structural features like an am-
ide function in position 5 and a bulky lipophilic chain in po-
sition 1 (Fig. 3), associated with a valuable CB2 agonist pro-
file, as shown by the most potent and selective compound
with R = cyclobutylmethyl (CB2 = 1.6 nM, SI = 1947).16 In
Fig. 1 Chemical structures and binding affinity values of the agonist
WIN-55,212-2 and the inverse agonist/antagonist AM630.
2046 | Med. Chem. Commun., 2018, 9, 2045–2054
This journal is © The Royal Society of Chemistry 2018

View Article Online
MedChemComm
Research Article
Fig.
3 Chemical structures and binding affinity values of CB2
benzimidazole-containing agonists and antagonists.
particular, the 4-ethoxybenzyl moiety was considered as the
main determinant for the agonist activity of this class of com-
pounds, independent of the nature of the chains in position
1 of the benzimidazole nucleus.
Moreover, Pagé et al.17 observed that the replacement of
the 4-ethoxyphenyl ring with a benzofuran one shifted the ag-
onist profile in favor of an antagonist/inverse agonist behav-
ior. In this set of antagonists, the authors observed that the
introduction of polar atoms (N, O) in the side chain (R)
matched negatively with the affinity towards the CB2 recep-
tor, while much better results were obtained with lipophilic
substitutions, particularly with alicyclic ones (R = cyclo-
pentylmethyl and cyclohexylmethyl).
During the last few years, our research group developed a
large variety of benzimidazole derivatives with different phar-
macological aims, in particular, exploring the analgesic,18
antiviral19–21 and anti-tumor22 activities. Our benzimidazoles,
studied in vivo as analgesic drugs, were designed on the basis
of previous work of CIBA (now Novartis), that led to etonitazene,
the most potent μ-opioid currently known.23 Besides the analge-
sic activity, some members of our benzimidazole sets displayed
CNS stimulant, anti-inflammatory and hypotensive activities
which are related to the involvement of other non-opioid recep-
tors, eventually also the cannabinoid ones.
Fig.
derivatives.
4
Chemical structure of the investigated benzimidazole
Based on the information coming from preliminary compu-
tational studies performed on the in-house available CB2 recep-
tor model,24 we in silico screened a series of benzimidazole-
containing derivatives in order to better investigate the effect of
unexplored substitutions at this core in terms of their selectiv-
ity and functionality to the CB2 receptor.
ethanol and dry HCl. Compound 13 was obtained by condens-
ing
N,N-diethylaminoethyl-5-trifluoromethyl-1,2-phenylene-
diamine with the aldehyde-sodium bisulphite adduct, which
was freshly prepared, as indicated by Shriner and Land.28 Fi-
nally, the 2-[(benzotriazol-1/2-yl)methyl]-benzimidazoles 15,29
16, 17 (ref. 30) and 18 (ref. 19) were synthesized by fusing at
Results and discussion
Chemistry
180 °C
a
mixture of the properly substituted 1,2-
phenylenediamine and (benzotriazol-1/2-yl)acetic acid.
The hydrazone derivative 14 was obtained by refluxing for 5
h a hydroalcoholic solution of 5-acetyl-1-(2-diethylaminoethyl)-
2-(4-ethoxybenzyl)benzimidazole (8) with a slight excess of hy-
drazine hydrate (Scheme 1).
Compounds 1–18 (Fig. 4) were re-synthesized, with the excep-
tion of the newly synthesized derivative 14, to be evaluated as
novel CB1 and CB2 ligands, bearing in positions 1, 2 and 5 of
the benzimidazole core some new unexplored substitutions.
For compounds 1–2, 5 and 6,25 3,26 4, 11 and 12,21 and
7–10,27 the benzimidazole ring was formed by heating, in
chloroform solution, the properly substituted 1,2-
phenylenediamine with the hydrochloride of the iminoester,
previously prepared from the corresponding nitrile, absolute
In vitro pharmacology and SAR
The investigated compounds (1–18) were evaluated using
radioligand binding assays for their ability to displace [3H]-
This journal is © The Royal Society of Chemistry 2018
Med. Chem. Commun., 2018, 9, 2045–2054 | 2047

View Article Online
Research Article
MedChemComm
The structure–activity relationship study showed that the
dimethylaminoethyl or diethylaminoethyl chains were best
suited for promoting the binding to the CB2 receptor (1–4,
8), while the longer diethylaminopropyl or the bulkier
quinolizidine moieties (6, 9 and 18) caused the loss of affin-
ity. Also, neutral chains proved to be favorable for substitu-
tions: the lipophilic n-butyl chain made compound 11 more
Scheme 1 Reagents and conditions: a) EtOH/H2O, NH2NH2·H2O (1.15
equiv.), 120 °C, 5 h.
effective than the analog 10, which bears
a dimethyl-
aminoethyl chain. Indeed, compound 11 was found to be the
most potent CB2 ligand, exhibiting the highest affinity value
for the target (Ki = 0.37 μM); also the polar methoxyethyl
chain, exhibited by 12, was proven effective, but with a 30-
fold lower affinity compared to 11.
With regard to the substitution in position 2, the more
flexible benzyl ring proved to be the only effective aromatic
substituent, whereas the planar benzene ring (13) and the
bulkier (benzotriazol-1/2-yl)methyl residue (15–18) yielded a
negative outcome, abolishing completely the activity. It is
worth noting that the topology of the benzotriazole moiety,
when combined with the favorable 2-benzyl and dialkyamino-
ethyl substitutions, also has a prevalent negative impact on
the activity, i.e. abolishing it.
The substituent in positions 3 and/or 4 of the benzyl ring
did not play a significant role in the interaction with the mo-
lecular target, since both the unsubstituted derivative (2) and
those functionalized with electron-withdrawing (4, 11) and
electron-donor groups (1, 3, 8) were tolerated. Indeed, the na-
ture of the substituents on the benzyl ring dictated the ago-
nist/antagonist behavior: in fact, the replacement of the
4-ethoxy group (3) with the 3,4-diCl substitution (11) on the
benzyl ring changed the agonist ability to the antagonist/in-
verse agonist one. Thus, the 4-ethoxy (or 4-methoxy) benzyl
moiety was confirmed to be the main feature for the agonist
activity, as well as of this class of benzimidazoles. Moreover,
the here presented full agonist activity of compound 3
matches with its analgesic properties (and those of analogs 1,
CP-55,940 from human recombinant CB1 and CB2 receptors
(Perkin Elmer, Italy). In Table 1, the inhibition constants (Ki)
of compounds 1–18 towards the CB1 and CB2 receptors are
reported. The selectivity index (SI) of CB2 versus CB1 is deter-
mined by dividing the CB1 Ki value by the CB2 Ki value.
The most interesting compounds showed selective binding
affinity towards the CB2 receptor with Ki values ranging from
0.37 to 10 μM. Only compound 11, the most potent CB2 li-
gand (Ki = 0.37 μM), was also able to bind the CB1 receptor
with lower affinity, although exhibiting a preferential selectiv-
ity (SI = 27.03) for CB2 over CB1.
In comparison with previously investigated benzimidazole
derivatives by Pagé et al. (Fig. 3),15–17 in some cases, our com-
pounds have the same pattern of substitutions (such as the
dimethylamino and methoxyethyl chains in position 1 and
the 4-ethoxyethylbenyl ring in position 2), although associat-
ing them with different decorations in positions 1, 2 and 5 of
the benzimidazole core. The synthesized structures were
designed by decorating the benzimidazole scaffold with sub-
stituents of different nature: a) an aliphatic chain in position
1 (basic or neutral); b) an aromatic (phenyl, benzyl) or hetero-
aromatic ring in position 2; c) an electron-withdrawing (CF3,
Cl, COCH3) or electron-donor (CH3, OCH3) group in position
5. The more recurring features responsible for CB2 receptor
affinity were a dialkylaminoethyl chain in position 1, a benzyl
ring in position 2 and a CF3 group in position 5 of the benz-
imidazole scaffold.
Table 1 Binding assaysa on CB1 and CB2 receptors of benzimidazoles 1–18
Cpd
CB1 Ki (μM)
Max tested conc. on CB1b (μM)
CB2 Ki (μM)
Max tested conc. on CB2b (μM)
SI (CB1 Ki/CB2 Ki)
1
2
3
4
5
6
7
8
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
∼10
>10
>10
>10
>10
>10
>10
>10
10 (3.23 1.49)
10 (19.64%)
10 (13.81%)
10 (10.47%)
10 (19.25%)
10 (5.50%)
10 (24.29%)
10 (25.52%)
10 (47.62%)
10 (37.86%)
10 (53.94%)
10 (41.40%)
10 (1.95%)
2.07 0.86
1.58 0.21
0.42 0.09
0.98 0.05
>10
>10
>10
0.96 0.16
>10
>10
50 (83.69 0.55)
50 (81.73%)
25 (94.11%)
25 (92.69%)
10 (16.79%)
10 (19.68%)
10 (38.47%)
25 (90.40%)
10 (15.56%)
10 (48.30%)
10 (94.26%)
10 (53.77%)
10 (46.94%)
>4.83
>6.33
>23.81
>10.20
—
—
—
>10.42
—
—
27.03
>1
—
—
—
—
—
—
9
10
11
12
13
14
15
16
17
18
0.37 0.08
∼10
>10
10 (5.58%)
>10
>10
>10
>10
10 (24.15%)
10 (19.03 13.06)
10 (14.07 0.72)
10 (15.00 13.01)
10 (17.63 8.01)
10 (17.29 6.41)
10 (24.57 3.13)
10 (17.49 10.97)
10 (10.13 12.08)
>10
a
Data are the means SEM of at least n = 3 experiments. b % of displacement.
2048 | Med. Chem. Commun., 2018, 9, 2045–2054
This journal is © The Royal Society of Chemistry 2018

View Article Online
MedChemComm
Research Article
8) we previously found in in vivo studies18a and with the rele-
vant role of CB2 agonists in controlling pain.31
Molecular docking studies
In this work, we deepened our study by performing molecular
docking calculations focused on the most promising benz-
imidazoles here investigated (1–4, 8 and 11, Fig. 4) and on
the reference compounds 2b, 3b and 5b (Fig. 3). In particular,
we wanted to explore the role played by some key decorations
on this chemotype at the CB2 receptor binding cavity, by
means of the in-house homology model of the human CB2
receptor.24 This model was built in the presence of the refer-
ence agonist WIN-55,212-2 and refined by molecular dynam-
ics simulations (MD), in order to better clarify the interacting
role played by the biological target when coupled with the re-
lated agonist. Briefly, the derived model displayed a CB2 ago-
nist recognition site delimited by TM3, TM5 and TM6, which
was in agreement with site-directed mutagenesis data and
other computational studies from the literature.32
Our previous MD results revealed a molecular portrait in
which the agonist WIN-55,212-2 is bound with a high confor-
mational stability at the receptor binding site, being
H-bonded to S112, N188 and S285. In addition, a number of
hydrophobic interactions keep the molecule strictly associ-
ated with the protein.24 According to our calculations, the ag-
onist prototypes by AstraZeneca namely 2b and 3b oriented
the carboxamide function towards the receptor cavity
delimited by L108, S112 and F117 (Fig. 6).
Concerning the influence of substituents in position 5 in
the active 2-benzylbenzimidazole set, the lipophilic electron-
withdrawing CF3 group (1–4), compared to the acetyl one (8),
allowed higher Ki values towards the CB2 receptor to be
reached, also generating a greater number of active com-
pounds (ratio 5 : 1). Finally, the chemical variation of the
H-bond acceptor acetyl group, as for the CB2 ligand 8, with
the H-bond donor hydrazone group (14) was however detri-
mental to the activity, leading to the loss of affinity for the
CB2 receptor.
Functional activity at CB2 receptors
Based on the results obtained, we further evaluated the capa-
bility of compounds 3 and 11 to activate CB2 receptors. We
used the cAMP Hunter™ assay enzyme fragment complemen-
tation chemiluminescence detection kit (Eurofins DiscoverX
Corporation, Fremont, CA) to measure whether the com-
pounds modulate intracellular cAMP levels in NKH-477-
stimulated CHO cells overexpressing human CB2 receptors.
As shown in Fig. 5, compound 3 displayed typical orthosteric
Gi agonist behavior by reducing cAMP levels induced by
NKH-477 (a water-soluble analog of forskolin). Compound 11
slightly increased the cAMP levels compared to that induced
by NKH-477 (Fig. 5A). However, when tested in the presence
of an EC80 CB2-ligand challenge (3 μM JWH-133), this com-
pound was able to displace the agonist, increasing the level
of cAMP up to the NKH-477 stimulus as expected for an
orthosteric antagonist/inverse agonist (Fig. 5B).
As we previously reported for WIN-55,212-2, this docking
mode supported for one the H-bond between the oxygen
atom of the two agonists, the carbonyl group and the S112
side-chain. The hydrophobic group linked at position 1 of the
benzimidazole core was projected towards S180 and F183
while the phenoxy moiety displayed π–π stacking and
cation–π contacts with H94 and F95.
The introduction of rigid and bulkier groups than the
ethoxybenzyl ring at position 2 of the benzimidazole, as
shown by the inverse agonist 5b, caused a different position-
ing of the ligand within the receptor crevice. As a conse-
quence, one nitrogen atom of the benzimidazole was
H-bonded to S112 while the aliphatic substituent in position
Fig.
5 Concentration–response curves of compounds 3 and 11
measured using the cAMP Hunter™ assay enzyme fragment
complementation chemiluminescence detection kit. (A) Effect of
increasing concentrations of compounds on NKH-477-induced cAMP
levels. (B) Effect of compound on the stimulus of NKH-477 in the pres-
ence of a challenge of known agonist (3 μM JWH-133).
Fig. 6 Docking mode of agonists 2b (C atom, green) and 3b (C atom,
yellow) within the homology model of the human CB2 receptor.
This journal is © The Royal Society of Chemistry 2018
Med. Chem. Commun., 2018, 9, 2045–2054 | 2049

View Article Online
Research Article
MedChemComm
1 was involved in van der Waals contacts with F95 (Fig. 7).
Along with this, the oxygen atom of the substituent placed in
position 2 of the benzimidazole displayed one H-bond with
the backbone of F183.
All these results suggested a key role determined by those
interactions involving S112 for CB2 targeting, while the po-
tency profile especially for the agonist compounds was
proved to be increased by means of additional contacts with
the aforementioned H94 and F95.
Our compounds 1–18 exhibited only a partially compara-
ble docking mode with agonist and inverse agonist proto-
types, sometimes displaying one H-bond between the benz-
imidazole scaffold and the S112 side-chain (Table 1S†).
Notably, this kind of positioning proved to be allowed only in
the presence of a small flexible substituent linked at position
1 of the heterocycle, such as the n-butyl chain of the inverse
agonist compound 11. These results are in agreement with
the lower affinity values featured by the bulkier analogues
15–18, which are decorated with 2-(benzotriazol-1/2-yl)methyl
moieties.
Fig. 8 Docking mode of agonist 2b (C atom, gray) and benzimidazole
3 (C atom, coral) within the homology model of the human CB2
receptor.
for some series of CB2 indole-containing ligands. Along with
this, compound 11 and the related basic analogue
4
As shown in Fig. 8, the benzimidazole core of 11 should
be arranged within the narrow lipophilic cavity delimited by
L107, L108 and L169, by detecting van der Waals contacts, as
well as moving the aliphatic chain towards F183. This kind of
behavior was highly comparable with that discussed for the
inverse agonist 5b, lacking in the H-bond with F183. Conceiv-
ably, this caused the lower affinity observed between the two
compounds.
displayed comparable Ki values. The quest for gaining more
H-bonds with the aforementioned residue was explored with
preliminary substitutions linked to N1, bearing ether moie-
ties such as the methyl ethyl ether function exhibited by 12.
This benzimidazole derivative was endowed with a modest
CB2 binding ability, supporting further elongation strategies
to be applied at this position of the scaffold, by means of
suitable H-bonding features.
On the other hand, the agonist compound
3 was
H-bonded to F183 and S285 by the benzimidazole nitrogen
atoms and the protonated basic chain, respectively. As
reported in Fig. 8, the phenoxy moiety of the agonist only
partially mimics the docking mode proposed for the refer-
ence compound 2b.
As a consequence, it should be noticed that the introduc-
tion of basic features tethering the N1 nitrogen atom of the
benzimidazole ring was not mandatory to afford ameliorated
binding affinity values towards CB2, as previously mentioned
Prediction of ADMET properties
In the search for novel and more druggable compounds, the
in silico prediction of descriptors related to absorption, distri-
bution, metabolism, excretion and toxicity (ADMET) proper-
ties allows efficient prioritization of the most promising
chemical entities to be further developed.33–35
In this work, for the most promising benzimidazoles pro-
posed as CB2 ligands (1–4, 8, 11), a series of ADMET proper-
ties were calculated. In detail, we took into account the loga-
rithmic ratio of the octanol–water partitioning coefficient
(clog P), extent of blood–brain barrier permeation (log BB),
rate of passive diffusion-permeability (log PS), human intesti-
nal absorption (HIA), volume of distribution (Vd), the role
played by plasmatic protein binding (% PPB) and by the com-
pound affinity toward human serum albumin (log KaHSA), and
an overall perspective of the molecule oral bioavailability (%
F). In addition, preliminary data concerning metabolism and
toxicity profiles of any compound were predicted, in terms of
the ability to behave as a cytochrome P450 3A4 inhibitor or a
substrate and calculating the median lethal dose (LD50) by
oral administration.
As shown in Table 2, all the compounds apart from 11
were predicted to weakly pass the central nervous system, be-
ing characterized by adequate lipophilicity values (calculated
cLog P around 5). Notably, these derivatives exhibited
favourable volume of distribution values (Vd) and suitable
Fig. 7 Docking mode of agonist 2b (C atom, gray) and inverse agonist
5b (C atom, green) within the homology model of the human CB2
receptor.
2050 | Med. Chem. Commun., 2018, 9, 2045–2054
This journal is © The Royal Society of Chemistry 2018

View Article Online
MedChemComm
Research Article
Table 2 Calculated ADMET descriptors related to absorption and distribution properties
Cpd
clog P
log BBa
log PSb
HIAc (%)
Vd (L kg−1
d
)
% PPB
log KaHSA
% F (oral)
1
2
3
4
8
11
12
5.58
5.31
5.96
5.65
4.53
6.86
5.99
0.47
0.40
0.48
0.41
0.59
−1.4
−1.3
−1.6
−1.4
−1.4
−1.3
−1.1
100
100
100
100
100
100
100
8.0
8.1
9.2
7.7
5.7
5.3
3.5
97.98
98.12
98.11
99.01
94.75
99.61
99.51
4.60
4.60
4.62
5.28
4.55
5.61
5.56
99.3
99.3
99.3
97.6
99.1
19.9
14.2
−0.11
−0.03
a
b
Extent of brain penetration based on the ratio of total drug concentrations in tissue and plasma under steady-state conditions. Rate of pas-
sive diffusion-permeability. PS represents the permeability–surface area product and is derived from the kinetic equation of capillary transport.
c
HIA represents the human intestinal absorption, expressed as the percentage of the molecule able to pass through the intestinal membrane.
Prediction of volume of distribution (Vd) of the compound in the body.
d
bioavailability profiles (% F). Conversely, 11 was ineffective in
brain permeability, featuring a too high lipophilicity profile
and low bioavailability values. All the analysed derivatives are
fully adsorbed at the human intestinal membrane (HIA).
None of the compounds here proposed should be involved in
cytochrome P450 3A4 inhibition events (Table 2), as all of
them are substrates for the same enzyme. Finally, all the de-
rivatives exhibited an acceptable toxicity profile, with the esti-
mated LD50 being in the range of 690–1100 mg kg−1 for
mouse, after oral administration. Accordingly, compounds 11
and 12, previously tested in antiviral assays (mean EC50 = 10
μM on CVB-5 and RSV viruses), didn't show any toxic effect
(CC50 > 100 μM) both against the human MT-4 and primate
Vero 76 cell lines.21
the 2-benzyl-5-trifluoromethylbenzimidazole derivative (12),
bearing an ethereal function in the 1-(2-methoxy)ethyl chain,
although being less effective than compounds 3 and 11,
could also represent an interesting prototype to explore new
side chains, more appropriately decorated by means of
H-bonding features, capable of gaining interactions with the
N188 residue of the CB2 receptor cavity.
The preliminary information concerning their pharmaco-
kinetic profile points to a prevalent peripherally-restricted ac-
tivity devoid of toxicity that, in combination with the previ-
ously found analgesic (not related to opioid receptors) and
anti-inflammatory activities, suggests the eventual involve-
ment of CB2 response for explaining their potential biologi-
cal application.
The results obtained in this study allowed the identifica-
tion of interesting benzimidazole-containing hit compounds,
worthy of further investigation towards even more potent and
selective CB2 ligands.
Conclusions
We developed a series of novel CB2 receptor ligands featuring
a benzimidazole core, which is again confirmed to be a valu-
able scaffold capable of efficient interaction with cannabi-
noid receptors. The influence on CB2 affinity of some struc-
tural features was analyzed, taking into account the
information derived from our previous computational studies
on the CB2 receptor model,24 which demonstrated the valid-
ity of incorporating small flexible substituents in position 1,
a benzyl moiety in position 2 and lipophilic groups (i.e. CF3)
in position 5 as suitable decorations of the benzimidazole
framework for promoting CB2 binding affinity. In fact,
1-(2-diethylaminoethyl)-2-(4-ethoxybenzyl)-5-trifluoromethyl-
Experimental
Chemistry
Chemicals, solvents and reagents used for the syntheses were
purchased from Sigma-Aldrich or Alfa Aesar, and were used
without any further purification. Column chromatography
(CC): neutral alumina (Al2O3), activity 1 (Merck). Mps: Büchi
apparatus, uncorrected. 1H NMR and 13C NMR spectra:
Varian Gemini-200 spectrometer; CDCl3; δ in ppm rel. to
Me4Si as the internal standard. J in Hz. Elemental analyses
were performed on a Carlo Erba EA-1110 CHNS-O instrument
in the Microanalysis Laboratory of the Department of Phar-
macy of Genoa University.
benzimidazole
(3)
and
1-butyl-2-(3,4-dichlorobenzyl)-5-
trifluoromethylbenzimidazole (11) proved to be the best prom-
ising ligands providing sub-micromolar Ki values and greater
preference towards the CB2 receptor. Interestingly, the pres-
ent study allowed the identification of the agonist ability of
compound 3, which might be correlated with the previously
described analgesic activity of this compound, and the in-
verse agonist/antagonist profile of compound 11, as a conse-
quence of a simple but significant chemical variation of sub-
stituents on the benzyl ring.
1-[2-(Diethylamino)ethyl]-2-(4-ethoxybenzyl)-5-(1-hydrazono
ethyl)-1H-benzimidazole (14)
To a solution of 5-acetyl-1-[2-(N′,N′-diethylamino)ethyl]-2-(4-
ethoxybenzyl)-1H-benzimidazole (compound 8, 0.38 mmol) in
EtOH, a solution of hydrazine hydrate (0.44 mmol) in 3 mL
of H2O was added and then refluxed for 5 h with stirring. At
r.t., 5 mL of H2O was added to the reaction mixture, which
was allowed to stand at cooling (0–5 °C) for 12 h. A pale
Docking studies were performed to study whether the ac-
tive compounds could target the CB2 receptor, revealing the
key contacts for further assessment. From this information,
This journal is © The Royal Society of Chemistry 2018
Med. Chem. Commun., 2018, 9, 2045–2054 | 2051

View Article Online
Research Article
MedChemComm
brown solid was collected by filtration and re-crystallized
with dry Et2O.
software, which include a consistent pool of molecules with
experimentally known pharmacokinetic and toxicity profiles.
Pale brown crystals; yield: 47%. mp: 80–82 °C (Et2O an.);
1H NMR (200 MHz, CDCl3): δ = 0.94 (t, J = 6.8, 6H,
NIJCH2CH3)2), 1.39 (t, J = 6.8, 3H, OCH2CH3), 2.23 (s, 3H,
CH3CN–NH2), 2.31–2.63 (m, 6H, CH2CH2NIJCH2CH3)2), 3.83
(m, 2H of CH2CH2NIJCH2CH3)2 and 2H of OCH2CH3); 4.31 (s,
2H, CH2–Ar), 5.05 (br. s, 2H, NH2, exchange with D2O), 6.83
(d, J = 7.4, 2 arom. H), 7.16 (d, J = 7.6, 2 arom. H), 7.27 (d, J =
8.6, 1 arom. H), 7.72 (d, J = 8.8, 1 arom. H), 7.93 (s, 1 arom.
H); 13C NMR (50 MHz, CDCl3): δ = 156.90, 153.21, 147.36,
141.51, 134.24, 132.90, 128.41, 127.19, 119.31, 115.88, 113.79,
108.1, 90.53, 62.41, 50.80, 46.47, 41.96, 32.59, 13.75, 11.07,
10.66; anal. calcd for C24H33N5O: C 70.73, H 8.16, N 17.18,
found: C 70.67, H 8.47, N 17.43.
Binding assays
Membranes from HEK-293 cells stably transfected with the
human recombinant CB1 receptor (Bmax = 2.5 pmol mg−1 pro-
tein) and human recombinant CB2 receptor (Bmax = 4.7 pmol
mg−1 protein) were incubated with [3H]-CP-55940 (0.14 nM/Kd
= 0.18 nM and 0.084 nM/Kd = 0.31 nM, respectively, for CB1
and CB2 receptors) as the high affinity ligand and displaced
with 10 μM WIN 55212-2 as the heterologous competitor for
non-specific binding (Ki values of 9.2 nM and 2.1 nM, respec-
tively, for CB1 and CB2 receptors). All the compounds were
tested following the procedure described by the manufacturer
(Perkin Elmer, Italy). Displacement curves were generated by
incubating drugs with [3H]-CP-55,940 for 90 minutes at 30 °C.
Ki values were calculated by applying the Cheng–Prusoff
equation to the IC50 values (obtained by GraphPad) for the
displacement of the bound radioligand by increasing the con-
centration of the test compound. Data are means SEM of at
least n = 3 independent experiments.
Molecular modeling
All the compound isomers were built, parameterised
(Gasteiger–Huckel method) and energy minimised within
MOE using the MMFF94 force field.36 Docking studies were
performed using the available in-house complex of the
homology CB2 receptor model in the presence of the refer-
ence agonist WIN-55,212-2, whose construction details were
reported in our previous publications.24 Molecular docking
calculations were conducted by means of the LeadIT 2.1.8
software suite (www.biosolveit.com). This tool includes the
FlexX scoring algorithm, which is based on the calculation of
the binding free energy by means of the Gibbs–Helmholtz
equation.37–39 The software detects the binding site defining
a radius of 10 Å far from the ligand place in the binding site,
in order to set up a spherical search space for the docking ap-
proach. The standard setting as the docking strategy was
followed, choosing the so-called hybrid approach (enthalpy
and entropy criteria); the related scoring function evaluation
is described in the literature.40 The ten derived docking poses
were prioritized taking into account the score values of the
lowest energy pose of the compounds docked to the protein
structure. All the ligands were further refined and rescored
by assessment with the algorithm HYDE, included in the
LeadIT 2.1.8 software. The HYDE module considers dehydra-
tion enthalpy and hydrogen bonding.41,42 Then, the stability
of the selected protein–ligand complexes was verified using a
short ∼1 ps run of molecular dynamics (MD) simulation at
constant temperature, followed by an all-atom energy mini-
mization (LowModeMD implemented in MOE software). In
this way, we performed an exhaustive conformational analy-
sis of the ligand receptor binding site subset, as we previ-
ously reported for other case studies.34,43,44
Functional activity at CB2 receptors
We used the cAMP Hunter™ assay enzyme fragment comple-
mentation chemiluminescence detection kit to characterize
the functional activity in a CB2 receptor-overexpressing cell
line. Gi-coupled cAMP modulation was measured following
the manufacturer's protocol (DiscoveRx, Fremont, CA) as pre-
viously reported.45 Briefly, the cells were incubated for 30
min at 37 °C with the samples prepared in the presence of
the cell assay buffer containing 25 μM NKH-477 (a water-
soluble analogue of forskolin) to stimulate adenylate cyclase
and enhance basal cAMP levels. For compound 11, the cells
were pre-incubated with increasing concentration of the sam-
ple, and incubated with the agonist challenge (JWH-133) at
its EC80 concentration (3 μM, previously determined in sepa-
rate experiments) in the presence of NKH-477. Luminescence
was measured using a GloMax Multi Detection System
(Promega, Italy). Data are normalized to the maximal and
minimal responses observed, and are reported as mean
SEM of three independent experiments conducted in tripli-
cate. Data analysis was done using PRISM software
(GraphPad Software Inc, San Diego, CA).
Author contributions
MT and EC conceived the study; MT and BT designed and
synthesized all the molecules and provided their structural
characterization; EC performed the computational studies;
AL and AR performed the binding experiments; AM
performed the functional assays; MT, EC, PF and AL analyzed
and discussed the results; MT and EC wrote the manuscript.
In silico evaluation of pharmacokinetic properties
The prediction of ADMET properties was performed using
the Advanced Chemistry Development (ACD) Percepta plat-
form (www.acdlabs.com). Any ADMET descriptor was evalu-
ated by Percepta using training libraries implemented in the
Conflicts of interest
There are no conflicts to declare.
2052 | Med. Chem. Commun., 2018, 9, 2045–2054
This journal is © The Royal Society of Chemistry 2018

View Article Online
MedChemComm
Research Article
Acknowledgements
L. Hodzic, S. St-Onge, C. Godbout, D. Salois and K. Payza,
Bioorg. Med. Chem. Lett., 2008, 18, 3695–3700.
17 D. Pagé, M.-C. Brochu, H. Yang, W. Brown, S. St-Onge, E.
Martin and D. Salois, Lett. Drug Des. Discovery, 2006, 2, 298–303.
18 (a) F. Sparatore, V. Boido and F. Fanelli, Farmaco, Ed. Sci.,
1968, 23, 344–359; (b) G. Paglietti, M. A. Pirisi, M. Loriga,
G. E. Grella, F. Sparatore, M. Satta and P. Manca, Farmaco,
Ed. Sci., 1988, 43, 215–226.
19 M. Tonelli, P. Paglietti, V. Boido, F. Sparatore, F. Marongiu,
E. Marongiu, P. La Colla and R. Loddo, Chem. Biodiversity,
2008, 5, 2386–2401.
20 M. Tonelli, M. Simone, B. Tasso, F. Novelli, V. Boido, F.
Sparatore, G. Paglietti, S. Pricl, G. Giliberti, S. Blois, C. Ibba,
G. Sanna, R. Loddo and P. La Colla, Bioorg. Med. Chem.,
2010, 18, 2937–2953.
This work was financially supported by the University of
Genoa, Italy. The authors would like to thank Mr. V. Ruocco
for the informatics support to calculations and Mr. O.
Gagliardo for performing the elemental analysis. The experi-
ments performed by Alì M. Mahmoud are part of his INCIPIT
PhD programme, which is co-funded by the COFUND scheme
Marie Skłodowska-Curie Actions.
References
1 R. Kaur, S. R. Ambwani and S. Singh, Curr. Clin. Pharmacol.,
2016, 11, 110–117.
2 A. Ligresti, L. De Petrocellis and V. Di Marzo, Physiol. Rev.,
2016, 96, 1593–1659.
3 A. C. Howlett, F. Barth, T. I. Bonner, G. Cabral, P. Casellas,
W. A. Devane, C. C. Felder, M. Herkenham, K. Mackie, B. R.
Martin, R. Mechoulam and R. G. Pertwee, Pharmacol. Rev.,
2002, 54, 161–202.
4 S. S.-J. Hu and K. Mackie, Handb. Exp. Pharmacol.,
2015, 231, 59–93.
5 I. B. Adams and B. R. Martin, Addiction, 1996, 91,
1585–1614.
6 D. Piomelli, A. Giuffrida, A. Calignano and F. Rodríguez de
Fonseca, Trends Pharmacol. Sci., 2000, 21, 218–224.
7 J. A. López-Moreno, G. González-Cuevas, G. Moreno and M.
Navarro, Addict. Biol., 2008, 13, 160–187.
21 M. Tonelli, F. Novelli, B. Tasso, I. Vazzana, A. Sparatore, V.
Boido, F. Sparatore, P. La Colla, G. Sanna, G. Giliberti, B.
Busonera, P. Farci, C. Ibba and R. Loddo, Bioorg. Med.
Chem., 2014, 22, 4893–4909.
22 M. Tonelli, B. Tasso, L. Mina, G. Paglietti, V. Boido and F.
Sparatore, Mol. Diversity, 2013, 17, 409–419.
23 M. S. Moolten, J. B. Fishman, J. C. Chen and K. R. Carlson,
Life Sci., 1993, 52, PL199–PL203.
24 E. Cichero, A. Ligresti, M. Allarà, V. Di Marzo, Z. Lazzati, P.
D'Ursi, A. Marabotti, L. Milanesi, A. Spallarossa, A. Ranise
and P. Fossa, Eur. J. Med. Chem., 2011, 46, 4489–4505.
25 F. Sparatore, V. Boido and F. Fanelli, Farmaco, Ed. Sci.,
1968, 23, 344–352.
8 G. Navarro, P. Morales, C. Rodríguez-Cueto, J. Fernández-
Ruiz, N. Jagerovic and R. Franco, Front. Neurosci., 2016, 10,
406.
9 M. D. Van Sickle, M. Duncan, P. J. Kingsley, A. Mouihate, P.
Urbani, K. Mackie, N. Stella, A. Makriyannis, D. Piomelli,
J. S. Davison, L. J. Marnett, V. Di Marzo, Q. J. Pittman, K. D.
Patel and K. A. Sharkey, Science, 2005, 310, 329–332.
10 M. T. Viscomi, S. Oddi, L. Latini, N. Pasquariello, F.
Florenzano, G. Bernardi, M. Molinari and M. Maccarrone,
J. Neurosci., 2009, 29, 4564–4570.
11 J. M. Frost, M. J. Dart, K. R. Tietje, T. R. Garrison, G. K.
Grayson, A. V. Daza, O. F. El-Kouhen, L. N. Miller, L. Li,
B. B. Yao, G. C. Hsieh, M. Pai, C. Z. Zhu, P. Chandran and
M. D. Meyer, J. Med. Chem., 2008, 51, 1904–1912.
12 J. M. Frost, M. J. Dart, K. R. Tietje, T. R. Garrison, G. K.
Grayson, A. V. Daza, O. F. El-Kouhen, B. B. Yao, G. C. Hsieh,
M. Pai, C. Z. Zhu, P. Chandran and M. D. Meyer, J. Med.
Chem., 2010, 53, 295–315.
26 F. I. Carroll, R. W. Handy, J. A. Kepler and J. A. Gratz,
J. Heterocycl. Chem., 1967, 4, 262–267.
27 G. Paglietti and F. Sparatore, Farmaco, Ed. Sci., 1972, 27,
333–342.
28 R. L. Shriner and A. H. Land, J. Org. Chem., 1941, 6,
888–894.
29 E. Cichero, M. Tonelli, F. Novelli, B. Tasso, I. Delogu, R.
Loddo, O. Bruno and P. Fossa, J. Enzyme Inhib. Med. Chem.,
2017, 32, 375–402.
30 G. Paglietti, V. Boido and F. Sparatore, Farmaco, Ed. Sci.,
1975, 30, 504–511.
31 S. Han, J. Thatte, D. J. Buzard and R. M. Jones, J. Med.
Chem., 2013, 56, 8224–8256.
32 T. Tuccinardi, P. L. Ferrarini, C. Manera, G. Ortore, G.
Saccomanni and A. Martinelli, J. Med. Chem., 2006, 49,
984–994.
33 H. Waterbeemd and E. Gifford, Nat. Rev. Drug Discovery,
2003, 2, 192–204.
13 P. Diaz, J. Xu, F. Astruc-Diaz, H. M. Pan, D. L. Brown and M.
Naguib, J. Med. Chem., 2008, 51, 4932–4947.
34 M. Tonelli, S. Espinoza, R. R. Gainetdinov and E. Cichero,
Eur. J. Med. Chem., 2017, 127, 781–792.
14 P. Diaz, S. S. Phatak, J. Xu, F. Astruc-Diaz, C. N. Cavasotto
and M. Naguib, J. Med. Chem., 2009, 52, 433–444.
15 D. Pagé, H. Yang, W. Brown, C. Walpole, M. Fleurent, M.
Fyfe, F. Gaudreault and S. St-Onge, Bioorg. Med. Chem. Lett.,
2007, 17, 6183–6187.
16 D. Pagé, E. Balaux, L. Boisvert, Z. Liu, C. Milburn, M.
Tremblay, Z. Wei, S. Woo, X. Luo, Y. X. Cheng, H. Yang, S.
Srivastava, F. Zhou, W. Brown, M. Tomaszewski, C. Walpole,
35 S. Franchini, L. I. Manasieva, C. Sorbi, U. M. Battisti, P.
Fossa, E. Cichero, N. Denora, R. M. Iacobazzi, A. Cilia, L.
Pirona, S. Ronsisvalle, G. Aricò and L. Brasili, Eur. J. Med.
Chem., 2017, 125, 435–452.
36 MOE, Chemical Computing Group Inc., Montreal H3A 2R7
Canada.
37 H. J. Böhm, J. Comput.-Aided Mol. Des., 1992, 6, 61–78.
38 H. J. Böhm, J. Comput.-Aided Mol. Des., 1994, 8, 243–256.
This journal is © The Royal Society of Chemistry 2018
Med. Chem. Commun., 2018, 9, 2045–2054 | 2053

View Article Online
Research Article
MedChemComm
39 M. Rarey, B. Kramer, T. Lengauer and G. Klebe, J. Mol. Biol.,
1996, 261, 470–489.
40 L. Bichmann, Y. T. Wang and W. B. Fischer, Comput. Biol.
Chem., 2014, 53, 308–317.
41 I. Reulecke, G. Lange, J. Albrecht, R. Klein and M. Rarey,
ChemMedChem, 2008, 3, 885–897.
43 M. Tonelli, L. Naesens, S. Gazzarrini, M. Santucci, E.
Cichero, B. Tasso, A. Moroni, M. P. Costi and R. Loddo, Eur.
J. Med. Chem., 2017, 135, 467–478.
44 V. Francesconi, L. Giovannini, M. Santucci, E. Cichero, M. P.
Costi, L. Naesens, F. Giordanetto and M. Tonelli, Eur. J. Med.
Chem., 2018, 155, 229–243.
42 N. Schneider, S. Hindle, G. Lange, R. Klein, J. Albrecht, H.
Briem, K. Beyer, H. Claußen, M. Gastreich, C. Lemmen and
M. Rarey, J. Comput.-Aided Mol. Des., 2012, 26, 701–723.
45 N. A. Osman, A. Ligresti, C. D. Klein, M. Allarà, A. Rabbito,
V. Di Marzo, K. A. Abouzid and A. H. Abadi, Eur. J. Med.
Chem., 2016, 122, 619–634.
2054 | Med. Chem. Commun., 2018, 9, 2045–2054
This journal is © The Royal Society of Chemistry 2018