Application of Fluorine- And Nitrogen-Walk Approaches: Defining the Structural and Functional Diversity of 2-Phenylindole Class of Cannabinoid 1 Receptor Positive Allosteric Modulators

Article abstract of DOI:10.1021/acs.jmedchem.9b01142

Cannabinoid 1 receptor (CB1R) allosteric ligands hold a far-reaching therapeutic promise. We report the application of fluoro- and nitrogen-walk approaches to enhance the drug-like properties of GAT211, a prototype CB1R allosteric agonist-positive allosteric modulator (ago-PAM). Several analogs exhibited improved functional potency (cAMP, β-arrestin 2), metabolic stability, and aqueous solubility. Two key analogs, GAT591 (6r) and GAT593 (6s), exhibited augmented allosteric-agonist and PAM activities in neuronal cultures, improved metabolic stability, and enhanced orthosteric agonist binding (CP55,940). Both analogs also exhibited good analgesic potency in the CFA inflammatory-pain model with longer duration of action over GAT211 while being devoid of adverse cannabimimetic effects. Another analog, GAT592 (9j), exhibited moderate ago-PAM potency and improved aqueous solubility with therapeutic reduction of intraocular pressure in murine glaucoma models. The SAR findings and the enhanced allosteric activity in this class of allosteric modulators were accounted for in our recently developed computational model for CB1R allosteric activation and positive allosteric modulation.

Full text of DOI:10.1021/acs.jmedchem.9b01142

Subscriber access provided by The Univ of Iowa Libraries
Article
Application of Fluorine- And Nitrogen-Walk Approaches: Defining
the Structural and Functional Diversity of 2-Phenylindole
Class of CB1 Receptor Positive Allosteric Modulators
Sumanta Garai, Pushkar M Kulkarni, Peter Carlton Schaffer, Luciana Leo, Asher L. Brandt, Ayat Zagzoog,
Tallan Black, Xiaoyan Lin, Dow P. Hurst, DAVID R JANERO, Mary E Abood, Anaelle Zimmowitch,
Alex Straiker, Roger G. Pertwee, Melanie Kelly, Anna-Maria Szczesniak, EILEEN DENOVAN-
WRIGHT, Ken Mackie, Andrea Hohmann, Patricia H. Reggio, Robert B. Laprairie, and Ganesh A Thakur
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01142 • Publication Date (Web): 22 Nov 2019
Downloaded from pubs.acs.org on November 22, 2019
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 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 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.
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 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Application of Fluorine- and Nitrogen-Walk Approaches: Defining the
Structural and Functional Diversity of 2-Phenylindole Class of CB1 Receptor
Positive Allosteric Modulators
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
†‡
†‡
†
€
Φ
Sumanta Garai, Pushkar M. Kulkarni, Peter C. Schaffer, Luciana M. Leo, Asher L. Brandt,
Ayat Zagzoog, Φ Tallan Black, Xiaoyan Lin , Dow P. Hurst, David R. Janero, Mary E. Abood,
Φ
£
¥
§
€
Anaelle Zimmowitch, Alex Straiker, £ Roger G. Pertwee, Melanie Kelly, Anna-Maria
£
#
^
^
^
£
£
¥
Szczesniak, Eileen M. Denovan-Wright , Ken Mackie , Andrea Hohmann , Patricia H. Reggio,
Robert B. Laprairie, * Φ,^and Ganesh A. Thakur*,†
†
Department of Pharmaceutical Sciences, School of Pharmacy, Bouvꢀ College of Health Sciences,
Northeastern University, Boston, Massachusetts 02115, United States.
Φ
College of Pharmacy and Nutrition, University of Saskatchewan, 104 Clinic Pl, Saskatoon, SK,
Canada – S7N2Z4.
€
Center for Substance Abuse Research, Lewis Katz School of Medicine, Temple University,
Philadelphia, PA 19140, United States
¥
Center for Drug Discovery, University of North Carolina Greensboro, Greensboro, North
Carolina 27402, United States
§
Department of Pharmaceutical Sciences, Bouvꢀ College of Health Sciences, Department of
Chemistry and Chemical Biology, College of Science, and Health Sciences Entrepreneurs;
Northeastern University, Boston, Massachusetts 02115, United States.
£Program in Neuroscience, Indiana University, Bloomington, Indiana; Psychological and Brain
Sciences, Indiana University, Bloomington, Indiana; Gill Center for Biomolecular Science,
Bloomington, IN 47405, United States.
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 81
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
^
Department of Pharmacology, Faculty of Medicine, Dalhousie University, 5850 College St,
Halifax, NS, Canada – B3H4R2
#
School of Medicine, Medical Sciences and Nutrition, Institute of Medical Sciences, University
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
of Aberdeen, Aberdeen AB25 2ZD, Scotland, U.K.
‡Authors contributed equally to this work.
*
Corresponding Authors.
KEYWORDS Allosteric modulator, CB1 Cannabinoid receptors, G-protein coupled receptor,
Positive allosteric modulator (PAM), cAMP, GTPγS, Fluorine walk, Nitrogen walk, Structure-
activity relationship, CFA model of pain, NEE model of glaucoma, MMC studies.
ACS Paragon Plus Environment
2

Page 3 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Graphical Abstract:
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
ABSTRACT:
Cannabinoid 1 receptor (CB1R) allosteric ligands hold far-reaching therapeutic promise. We
report application of fluoro- and nitrogen-walk approaches to enhance the drug-like properties of
GAT211, a prototype CB1R allosteric agonist-positive allosteric modulator (ago-PAM). Several
analogs exhibited improved functional potency (cAMP, arrestin2), metabolic stability, and
aqueous solubility. Two key analogs, GAT591 (6r) and GAT593 (6s), exhibited augmented
allosteric-agonist and PAM activities in neuronal cultures, improved metabolic stability, and
enhanced orthosteric agonist binding (CP55,940). Both also exhibited good analgesic potency in
the CFA inflammatory-pain model with longer duration of action over GAT211 while devoid of
adverse cannabimimetic effects. Another analog, GAT592 (9j), exhibited moderate ago-PAM
potency and improved aqueous solubility with therapeutic reduction of intraocular pressure in
murine glaucoma models. The SAR findings and the enhanced allosteric activity in this class of
allosteric modulators were accounted for in our recently developed computational model for CB1R
allosteric activation and positive allosteric modulation.
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 81
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
INTRODUCTION:
The endocannabinoid system (ECS) is an ubiquitous information transducing network comprised
of endocannabinoid lipid-signaling molecules [mainly, 2-arachidonoylglycerol (2-AG) and
anandamide (AEA)], type 1 and 2 cannabinoid G protein-coupled receptors (GPCRs) (CB1R and
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
CB2R), and endocannabinoid-synthesizing and -inactivating enzymes. Widely distributed
throughout the human body, the ECS helps control and modulate multiple physiological processes
1
such that aberrant ECS activity has been implicated in a broad range of disease states. The
prominence of CB1R for regulating synaptic activity in the central nervous system and its varied
roles in the periphery ranging from metabolic control to modulation of smooth muscle tone have
focused significant attention on the role of abnormal CB1R-mediated signal transduction in disease
etiology and the potential of pharmacological CB1R modulation as treatment modality. Indeed,
this GPCR, is recognized as a promising therapeutic target for several conditions including pain,
neuropathy, cachexia, glaucoma, and post-traumatic stress disorder. 2-11
9
The CB1R partial agonist and psychoactive plant (phyto)cannabinoid, ∆ -tetrahydrocannabinol
(THC), is approved in Canada and the European Union for treating cachexia and as a third-line
pain treatment.12, 13 As an orthosteric agonist, THC binds to the same site in CB1R as the
endogenous cannabinoids, thereby directly activating the receptor. Its demonstrated medical
utility notwithstanding, pharmacological CB1R activation by targeted orthosteric agonists carries
undesirable side-effect risks including intoxication, delirium, abuse liability, and development of
tolerance and dependence, which have significantly limited their therapeutic utility.
As a promising alternative to orthosteric agents for therapeutic GPCR targeting, allosteric
modulators have been pursued. These ligands bind to an allosteric site on the target GPCR that is
ACS Paragon Plus Environment
4

Page 5 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
subtype-specific and topographically distinct from the orthosteric ligand-binding pocket. The
cooperative interaction between bound orthosteric and allosteric ligands offers the potential to tune
the efficacy/affinity of the orthosteric ligand in a site- and event-specific manner. Binding of the
allosteric modulator is believed to induce structural changes in the GPCR that affects the receptor’s
affinity for and/or the efficacy of an orthosteric ligand to generate receptor conformational states
with unique structural and functional properties that may be therapeutically exploitable.14-21 In
particular, allosteric CB1R modulation by small-molecule ligands has several appealing features
that could translate into improved safety, more flexible therapeutic dose-ranging, and avoidance
of the inherent side-effect risks of orthosteric CB1R ligands. CB1R allosteric modulators that
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
potentiate [positive allosteric modulators (PAMs)] or inhibit [negative allosteric modulators
(NAMs)] orthosteric signaling have recently been described (Fig. 1). CB1R (ago)-PAMs have
garnered attention as potential treatments for pain and neurodegenerative diseases without the risk
of psycho-behavioral and other side effects carried by direct orthosteric CB1R activators.22-24
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 81
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
We recently described the pharmacological characterization of GAT211, a CB1R allosteric agonist
and PAM (i.e. ago-PAM) and the enantiospecific interaction of its enantiomers (GAT228 and
GAT229) with CB1R. It is noteworthy to mention that this is the first small molecule CB1R
modulator from the 2-phenylindole structural class that was correctly characterized as a CB1R
ago-PAM. With human-recombinant CB1R (hCB1R) and native receptor, GAT211 (racemic)
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
3
enhanced both [ H]CP55,940 binding and agonist-mediated CB1R signaling, whereas GAT228
(R-enantiomer) was an allosteric CB1R partial agonist and GAT229 (S-enantiomer) was a potent
CB1R PAM.25 Although GAT211 was first reported as a CB1R PAM by Astra Zeneca through
an HTS campaign, it was only partially characterized in vitro and was not studied further. Another
2
2
CB1R PAM ZCZ-011 was also first characterized as PAM , and later it was shown to behave as
ago-PAM.23, 26 GAT211 was instrumental in establishing the role of CB1R PAMs in treating
chronic and neuropathic pain, glaucoma, Huntington’s disease, and other CNS disorders. In vivo,
2
7
GAT211 showed preclinical therapeutic efficacy against neuropathic pain and did not produce
cardinal signs of CB1R activation associated with orthosteric cannabinoid receptor agonists. In
murine disease models, GAT211 and GAT229 attenuated IOP, a major risk factor for glaucoma, a
progressive neurodegenerative disease in which damage to the optic nerve results in loss of retinal
ganglion cells, and irreversible vision loss and blindness26, 28. GAT211 and GAT229 also reduced
the signs and symptoms of Huntington’s disease, while delaying its progression.29 GAT211,
however, has moderate PAM potency, low aqueous solubility and rapid metabolic clearance.
These findings have fostered an interest in understanding design parameters for productive
structure-activity relationships (SAR) around GAT211 and exploring the potential for additional,
novel candidates with comparable, or improved, drug-like properties while retaining or enhancing
CB1R PAM activity. In general, SAR studies of GPCR allosteric modulators have been reported
ACS Paragon Plus Environment
6

Page 7 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
to be challenging, yielding a flat-SAR.30 The fluorine- and nitrogen-walk approaches, which
respectively involve the replacement of a hydrogen atom with fluorine or a CH-function with
nitrogen have been reported as successful approaches to optimize biological activity, metabolic
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
stability, and overall physicochemical properties.31-34
Here, we report the application of fluorine- and nitrogen-walk approaches to the GAT211 scaffold
to develop potent and efficacious CB1R allosteric modulators with improved physicochemical
properties. We detail the development of the resulting novel analogs and their in vitro
characterization through functional assays (cAMP, -arrestin2, GTPS), including
electrophysiological and radioligand binding experiments. The three key compounds resulting
from these studies exhibited therapeutic efficacy in preclinical animal models of glaucoma (ocular
hypertensive model and NEE mice model) and pain (CFA model).
Ligand Design:
We exploited observations that systematic introduction or “walk” of fluorine around the core
structure of an allosteric ligand may retain, or enhance, chemical properties, pharmacological
activity, metabolic stability, and/or bioavailability. 35, 36 As a scaffold-hopping approach in a lead-
optimization drive, the “fluorine walk” (F-walk) has met success for identifying regions of the
allosteric ligand (in)tolerant to functionalization or contributory to a “flat” or “steep” SAR as
routinely observed for allosteric modulators and illuminating the electronic properties of the
ligand-binding site.30, 35 Based on this, we designed analogs in which each aromatic ring hydrogen
atom of GAT211 was replaced with a fluorine atom one at a time at a various position on sites I,
II and III of GAT211. The functional activity of 10 such mono-fluorinated analogs (Table 1)
further guided the design of four di-fluorinated and four tri-fluorinated analogs (Table 1).
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 81
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
Likewise, substitution of CH groups with nitrogen in (hetero-)aromatic rings, termed “nitrogen
walk” or aza walk (N-walk), can enhance the physicochemical and pharmacological properties of
a drug, mainly by altering intramolecular and/or ligand-GPCR noncovalent interactions to effect
positive conformational change upon the ligand and/or liganded GPCR.34, 36 Introduction of a basic
nitrogen atom into a ring generally decreases lipophilicity and increases the compound’s volume
of distribution, thereby prolonging its serum half-life and inviting the prospect of once-daily oral
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
3
4
dosing. Thus, we designed 10 aza-analogs of GAT211, one of which could not be synthesized
and isolated due to its instability. We also designed a representative hybrid analog in which both
nitrogen and fluorine atoms were incorporated together at the most-tolerable positions in the same
ring. All analogs retained the GAT211 nitro functionality, as its substitution with different groups
e.g., amino, acid, cyano, hydroxyl, trifluoromethyl) compromised pharmacological activity.23, 61
(
The aliphatic nitro group was well tolerated and did not cause apparent toxicity in the several
animal models in which GAT211 was profiled.27-29
RESULTS AND DISCUSSION:
CHEMISTRY:
The positional variation of fluorine at the Site III was accomplished according to Scheme 1.
Different fluorinated β-nitrostyrenes (2a-d) were synthesized as reported37 from each
corresponding mono-substituted fluoro-benzaldehydes (1a-d). These nitrostyrenes were further
treated for Michael addition with 2-phenyl indole (5) based on our in-house developed synthetic
3
8
methodology to furnish target compounds 6a-d in good yields.
ACS Paragon Plus Environment
8

Page 9 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 1. Synthesis of mono fluoro derivatives 6a-6d.
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
Similarly, a positional variation of fluorine at the Site I was accomplished according to Scheme 2.
Commercially available fluoro-substituted indoles (3e-h) were coupled with phenyl boronic acid
39
under a Suzuki type reaction condition in the presence of Pd(OAc) in acetic acid under an
2
oxygen atmosphere to yield fluorinated indoles (5e-h) in good yields (51%-67%). Similarly,
indoles 5i-k were synthesized by coupling commercially available fluoro phenylboronic acid 4i-k
with 2-phenylindole in good yields. All synthesized mono-fluoro 2-phenylindoles (5e-k) were
treated with β-nitrostyrene and trifluoro ammonium acetate under microwave irradiation to furnish
the corresponding Michael adducts 6e-k in 51-85% yields.
Difluoro-substituted analogs (6l-6o) were synthesized from 5g and 5h with 2-F and 4-F-
nitrostyrene (2a and 2c), respectively, using the same strategy (Scheme 2). Trifluorinated analogs
(6p-6s) were synthesized by following the same protocol that involved difluoro indoles (5l-5m)
and the corresponding fluoro nitrostyrene (2a and 2c).
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 81
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
Scheme 2. Synthesis of mono, di- and tri- fluoro derivatives 6e-6q.
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 nitrogen-walk analogs at Site III, we synthesized the corresponding pyridine nitrostyrenes 8a-
c by following the reported protocol40,41,48 which involved nitro-aldol condensation of
nitromethane and pyridine aldehyde 7a-c under basic condition followed by dehydration. Michael
adducts 9a-c were obtained in good yields by treating pyridine nitrostyrenes 8a-c with 2-
phenylindole in the presence of nBu
NI.41 On the other hand, pyridinyl indoles 11a-c were
4
synthesized by using a Fisher indole synthesis strategy. This involved synthesis of
phenylhydrazones from pyridine aldehydes 10b-c and phenylhydrazine followed by their
42
intramolecular cyclization in the presence of PPA to yield corresponding indoles. Indoles 11b-c
were treated with β-nitrostyrene in the presence of Et NBr in dioxane: water (9:1) under reflux
4
condition to yield Michael adducts 9d-e in good yields (Scheme 3). We were not able to isolate
the Michael adduct product derived from indole 11a and β-nitrostyrene under various attempted
ACS Paragon Plus Environment
1
0

Page 11 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
conditions, presumably due to its instability. The aza-indoles required for the synthesis of
nitrogen-walk analogs at Site I were obtained using different synthetic strategies (Scheme 4). Thus
4- and 6-aza-2-phenylindole 11d and 11f respectively were synthesized by treating 2-
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
methylpyridin-3-amine 12 and 4-methylpyridin-3-amine 14 with ethyl benzoate and sec-BuLi.43
For the synthesis of 5-aza-2-phenyl-1H-indole 11e, we used Sonogashira
Scheme 3. Synthesis of indoles and GAT 211 aza-analogs for site III and site IV variation
44
coupling followed by a 5-endo-dig cyclization strategy. . Accordingly, 3-iodopyridin-4-amine 13
was treated with phenyl acetylene, Pd(PPh
3
4
) , CuI in DMF to obtain the aryl-alkyne coupled
4
5
t
46
product , which was further cyclized with KO Bu to obtain the 4-aza-2-phenyl indole 11e in
8
4% overall yield. When 4-methylpyridin-3-amine was treated with ethyl benzoate for one-pot
ACS Paragon Plus Environment
1
1

Journal of Medicinal Chemistry
Page 12 of 81
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
indole formation, we obtained the corresponding amide compound 15, which was purified and
then cyclized with sec-BuLi to yield 7-aza indole 11g in 51% yield.
Once all corresponding aza-indoles were prepared for the synthesis of Site I variation analogs, 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
38
attempted our earlier developed method or other literature reported conditions for the Michael
addition reaction between 2-phenyl aza-indoles and β-nitrostyrene. However, for aza-indoles,
these approaches failed to produce the corresponding aza Michael adducts, a likely reflection of
nitrogen’s inductive effect, which makes the indole ring electron deficient and less reactive.
Scheme 4. Synthesis of aza analogs of GAT211 for Site I variations
To increase the indole ring’s electron density and foster the likelihood of its participating in
Michael addition with nitrostyrene, we generated the corresponding indolium anions by treating
t
n
47
the aza-indoles with a strong base (NaH, KO Bu, BuLi). We observed that this objective was
n
best achieved with BuLi. The indolium anion was then reacted in situ with nitrostyrene, and the
ACS Paragon Plus Environment
1
2

Page 13 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
corresponding Michael adduct was isolated in moderate to good yields (Scheme 5). This approach
represents a novel and useful strategy to synthesize Michael adducts between electron deficient
indoles and β-nitrostyrene. Using this optimized reaction condition, we synthesized the desired
aza-derivatives 9f-i (Scheme 4) in moderate to good yields.
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
Scheme 5: Alternative strategy employed for Michael addition reaction with electron-deficient
aza-indoles.
For the synthesis of the fluoro-aza hybrid compound 9j, the nitrostyrene 8d was synthesized from
t
6
-fluoropicolinaldehyde 7d. Aldehyde 7d was treated with nitromethane and KO Bu to obtain
corresponding nitro alcohol, which was further dehydrated with Ac
2
O/DMAP to get nitrostyrene
COONH under
4
8
8d in good yield. Nitrostyrene 8d was treated with 2-phenylindole with CF
reflux to obtain the corresponding Michael adduct 9j in 90% yield (Scheme 6).
Scheme 6. Synthesis of fluoro and aza hybrid analog 9j.
3
4
ACS Paragon Plus Environment
1
3

Journal of Medicinal Chemistry
Page 14 of 81
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
STRUCTURE-ACTIVITY RELATIONSHIP STUDIES: The influence of fluorine- and
nitrogen-walks on biological activity of GAT211 was indexed in vitro as CB1R-dependent PAM
inhibition of cAMP formation and arrestin2 recruitment in Chinese hamster ovary (CHO) cells
stably-expressing hCB1R (DiscoveRx HitHunter and PathHunter, respectively) + CP55,940
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
5
(
EC20) according to previously described methods (Table 1 and 2). Cellular cAMP formation
reflects G protein-mediated signaling, whereas arrestin2 recruitment reflects G protein-
25
independent information pathways. In the cAMP assay, inhibition of forskolin-stimulated cAMP
is quantified through the production of a chemiluminescent signal in the HitHunter cell line
proportional to cAMP abundance.25 Similarly, arrestin2 recruitment is quantified in the
PathHunter assay through arrestin2 enzyme complementation between CB1R and arrestin2,
each of which is expressed as a fusion protein with a fragment of −galactosidase, thus interaction
between CB1R and arrestin2 yields a complete enzyme whose activity is quantified via detection
2
5
of a chemiluminescent signal. Previous data have shown that G protein-mediated signaling via
CB1R is associated with sustained CB1R levels and cellular viability, whereas arrestin2
2
9
recruitment is associated with downregulation of CB1R and reduced cell viability. Thus, ideal
candidate compounds are likely to display high potency and efficacy for G protein-mediated
effects relative to arrestin2 recruitment. Our data thus provide insight into the bioactivity of our
newly synthesized compounds on the two principal mechanisms of GPCR/CB1R signaling. The
comprehensive in vitro SAR in these cell-based assays for GAT211 congeners obtained from the
fluorine-walk are given in Table 1, and those from nitrogen-walk are shown in Table 2.
In comparison to parent compound 6d, introduction of fluorine at C4, C5, C6 and C7 positions of
the indole ring at Site I (6e-h) showed incrementally improved PAM activity in the cAMP assay
with modest, if any, influence on arrestin2 recruitment. Among these, 6g (cAMP: EC50 = 110
ACS Paragon Plus Environment
1
4

Page 15 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
nM, Emax = 120%; arrestin2: EC50 = 1,100 nM, Emax = 38%) and 6h, (cAMP: EC50 = 22 nM, Emax
= 120%; arrestin2: EC50 = 750 nM, Emax = 45%), exhibited improved potency and signaling
selectivity.
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
Compounds 6i-k were derived from the fluorine-atom introduction at the ortho-, meta-, and para-
positions of the phenyl ring at the C2 position of indole (Site II). All analogs retained PAM activity
in the cAMP and arrestin2 assays, and the meta-fluoro analog 6j (cAMP: EC50 = 80 nM, Emax
40%; arrestin2: EC50 = 1,300 nM, Emax = 55%) displayed improved potency relative to the parent
compound 6d and selectivity for cAMP relative to arrestin2 recruitment.
=
1
The remaining fluorine-analogs of this “fluoro-walk” series reflect the introduction of fluorine-
atom into the phenyl ring (Site III). Intriguingly, all analogs (6a-c) showed significant variation
in PAM activity. Ortho-fluoro-substitution at Site III, 6a (cAMP: EC50 = 43 nM, Emax = 110%;

arrestin2: EC50 = 1,060 nM, Emax = 63%), was more potent than parent compound 6d and
somewhat selective for cAMP relative to arrestin2. Meta- and para- fluorine substitution at this
site led to analogs with decreased PAM activity (6b and 6c).
ACS Paragon Plus Environment
1
5

Journal of Medicinal Chemistry
Page 16 of 81
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. PAM activity of fluoro-analogs of GAT211
cAMP
arrestin2
Compound
Structure
cLogP
EC50 (nM)
E
Max (%)
EC50 (nM)
EMax (%)
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
9
40
6
d (GAT211)
230 (140-370)
110 ± 6.8
46 ± 9.5§
5.03
5.16
(540-1,800)^
900
37 ± 1.3§
6
e
130 (58-280)
120 ± 9.0
(
730-1,100)^
1
91 (96-380)
130 ± 8.5
120 ± 6.0
740 (540-990)^
23 ± 1.2§
5.16
5.16
6
f
1
,100
6
g
110 (59-190)
22 (8.4-58)*
38 ± 1.3§
(
880-1,300)^
45 ± 1.3§
6
h
120 ± 6.1
750 (630-880)^
5.16
5.16
1400
7
.0 ±
6
i
140 ± 6.2
710 (417-1,210)
.54†
,
†† §
0
(
1,100-1,700)*
ACS Paragon Plus Environment
1
6

Page 17 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
,300
6
j
80 (29-220)
140 ± 13†
120 ± 6.2
110 ± 3.6
55 ± 2.0§
5.16
(
1,100-1,600)^
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
1,600
-2.2 ±
6
k
a
>10,000
5.16
.7†††,§
6
(
1,200-2,300)*
1
,060
6
43 (28-66)*
63 ± 2.3§
5.16
5.16
(
840-1,300)^
2
,500
6
b
730 (280-1,900)
130 ± 18
19 ± 3.3†,§
(1,200-5,400)
6
c
440 (280-700)
110 ± 6.6
>10,000
8.0 ± 2.7†††,§
5.16
1
,140
6l
58 (34-97)*
110 ± 5.0
54 ± 2.8§
5.3
(
870-1,500)^
ACS Paragon Plus Environment
1
7

Journal of Medicinal Chemistry
Page 18 of 81
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
5
.3
6m
21 (14-30)*
28 (16-49)*
100 ± 3.2
93 ± 4.2
93 ± 4.9
680 (580-790)^
750 (570-980)^
510 (340-760)^
76 ± 1.8†††
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
6
r
s
71 ± 3.3
5.43
5.43
6
9
.1 (3.4-24)*
82 ± 4.8††
1,300
1,100
6
n
47 ± 5.6†††
121 ± 5.4
91 ± 12
32 ± 0.99
27 ± 1.9§
56 ± 2.0§
5.3
(
330-6,000)
(890-1,300)
7
00
6o
270 (190-380)
5.3
(
440-1,100)^
1
,150
2,100
6
p
5.43
(430-3,100)*
(1,800-2,600)
ACS Paragon Plus Environment
1
8

Page 19 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1,100
43 ± 1.6§
6
q
230 (160-340)
110 ± 5.3
5.43
(
910-1,400)^
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
CB1 PAM activity was quantified for cAMP inhibition using the DiscoveRx HitHunter assay (CHO hCB1R) in cells
treated with CP55,940 at EC20 + GAT211 analog for 30 min, and for arrestin2 recruitment using the DiscoveRx
PathHunter assay (CHO hCB1R) in cells treated with CP55,940 at EC20 + GAT211 analog for 90 min. Data were fit
to a variable slope (4 parameter) non-linear regression in GraphPad (v. 7). Data are mean with 95% confidence interval
(CI) (EC50) or mean ± SEM, n = 1-3 independent experiments performed in duplicate. Statistical analyses were by
non-overlapping CI or two-way ANOVA followed by Bonferroni’s post-hoc test. *P < 0.05 relative to GAT211 within
assay; ^P < 0.05 relative to cAMP assay within compound; P < 0.05, ††P < 0.01, †††P<0.001 relative to GAT211
†
§
within assay; P < 0.001 relative to cAMP assay within compound.
These GAT211 fluoro-walk data demonstrate the importance of the fluorine moiety and identified
positions of active sites in each of the parent compound’s rings (Sites I, II, and III) that can tolerate
substitution without compromising biological activity as CB1R PAMs and, in some cases, that
enhance activity and alter signaling bias between cAMP vs. arrestin2 recruitment.
To evaluate the potential for the cumulative effect of multiple fluoro-substitutions that were well
tolerated at each site studied, we synthesized a series of four di-fluoro GAT211 analogs (6l-o). The
di-fluoro analogs 6l and 6m were more potent and efficacious hCB1R PAMs than the parent
compound 6d and retained cAMP-selectivity, with 6m (C7 fluorinated analog) the more-potent of
the two. In contrast difluoro analogs 6n (cAMP: EC50 = 1,300 nM, Emax = 47%, arrestin2: EC50
=
1
,100 nM, Emax = 32%) and 6o (cAMP: EC50 = 270 nM, Emax = 121%, arrestin2: EC50 = 700 nM,
E
max = 27%) exhibited reduced activity. The tri-fluoro analogs 6r (cAMP: EC50 = 28 nM, Emax =
9
3%; arrestin2: EC50 = 750 nM, Emax = 71%) and 6s (cAMP: EC50 = 9.1 nM, Emax = 93%;
ACS Paragon Plus Environment
1
9

Journal of Medicinal Chemistry
Page 20 of 81
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

arrestin2: EC50 = 510 nM, Emax = 82%), were highly potent and fully efficacious CB1R PAMs
relative to 6d with notable bias towards cAMP inhibition relative to arrestin2 recruitment.
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
Unlike 6r and 6s, the tri-fluoro analogs 6p and 6q (para-substitution at Site III) exhibited unaltered
or reduced PAM activity relative to parent compound 6d. These combined data are congruent with
the SAR of the mono-fluoro and di-fluoro analogs and lead to the general conclusion that fluoro-
substitution at the para-position of Site III on the GAT211 scaffold is poorly tolerated.
Similar to GAT211 (6d), some 40-70% of active pharmaceuticals exhibit poor aqueous solubility,
a property which would compromise oral bioavailability without, for example, a suitable excipient
or delivery system such as those that have been recognized and approved by regulatory authorities.
4
9
With this in mind, we synthesized and profiled several nitrogen-walk analogs of 6d for their
PAM activity in the cAMP-inhibition and arrestin2-recruitment assays and calculated their
solubility with the aim of reducing lipophilicity (i.e., improving aqueous solubility) while
maintaining, if not improving, potency (Table 3).
The lack of activity exhibited by all aza compounds 9f-i at Site I indicates that the polarity of the
nitrogen substitution in Site I is not tolerated, suggesting that this region of the molecule interacts
with hydrophobic residues in hCB1R. Notably, compound 9i demonstrated a huge increase in
aqueous solubility compared to any other aza-analog from this series.
ACS Paragon Plus Environment
2
0

Page 21 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 2. PAM activity of aza-analogs of GAT211
Thermodyna
mic Aqueous
solubility
cAMP
arrestin2
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
Compound
Structure
cLogP
EC50
E
Max
EC50
E
Max
μg/
μM
(nM)
(%)
(nM)
(%)
mL
9
40
2
30 (140-
70)
6
d
114 ± 6.8
46 ± 9.5§
30 ± 1.1§
2.8 ± 2.4
5.03
7.8
2.7
3.8
(540-
3
1,800)^
1
,600
9f
>10,000
> 10,000
81 ± 4.4
4.16
11.2
(
1,300-
,000)
2
4
.0 ±
9
9
g
>10,000
>10,000
>10,000
3.91
3.91
4.16
23.6
8.1
2.3
2
.5†††
4
50 (47-
,200)
5.2 ±
3.5†††
5.1 ±
0.35
h
6.8
4
1
,100
7.0 ±
0.89
9i
39 ± 12†
159.0
54.6
(
1,100-
,700)*
1
ACS Paragon Plus Environment
2
1

Journal of Medicinal Chemistry
Page 22 of 81
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
1
,100
9
.2 ±
.7§
9
d
150 ± 12
>10,000
3.91
4.9
1.7
(630-
8
1,700)*
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
3
,100
1
,600
9
e
88 ± 18
19 ± 1.7§
3.91
2.1
0.7
(
1,400-
,700)*
(
980-2,700)
6
2
60 (140-
480)
5.8 ±
3.8§
9
a
130 ± 7.8
>10,000
4.15
2.8
1.0
3
,100
(634-
-8.1 ±
7.9§
9b
170 ± 39
>10,000
3.91
1.1
0.4
15,000)*
4
,000
510 (330-
780)
9c
110 ± 6.4
25 ± 4.4§
3.91
4.28
36.7
12.6
(
2,200-
,500)^
7
1
,900
-1.4 ±
9
j
95 ± 16
>10,000
--
--
(630-
2
.7§
3,700)*
ACS Paragon Plus Environment
2
2

Page 23 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
PAM activity was quantified for cAMP inhibition using the DiscoveRx HitHunter assay (CHO hCB1R) in cells treated
with CP55,940 at EC20 + GAT211 analog for 30 min, and for arrestin2 recruitment using the DiscoveRx PathHunter
assay (CHO hCB1R) in cells treated with CP55,940 at EC20 + GAT211 analog for 90 min. Data were fit to a variable
slope (4 parameter) non-linear regression in GraphPad (v. 7). Data are mean with 95% confidence interval (CI) (EC50
)
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
or mean ± SEM, n =1- 3 independent experiments performed in duplicate. Statistical analyses were by non-overlapping
CI or two-way ANOVA followed by Bonferroni’s post-hoc test. *P < 0.05 relative to GAT211 within assay; ^P <
†
0
.05 relative to cAMP assay within compound; P < 0.05, ††P < 0.01, †††P<0.001 relative to GAT211 within assay;
§P < 0.001 relative to cAMP assay within compound.
Similarly, aza-substitution of the aromatic ring at Site II (9d, 9e) reduced potency and efficacy
relative to the GAT211 parent 6d. Among the Site III aza analogs tested, 9a (ortho-substitution)
was the only analog that retained hCB1R PAM potency or efficacy in the cAMP inhibition assay
(cAMP: EC50 = 260 nM; Emax = 130%; arrestin2: EC50 >10 µM) compared to 6d. Whereas other
Site-III aza substitutions were not well tolerated (9b, 9c).
A fluoro-aza-hybrid compound (9j) was synthesized bearing the well-tolerated Site-III aza ortho-
substitution observed in 9a and the fluoro ortho-substitution observed in 6a. However, 9j displayed
reduced potency and efficacy in both cAMP and arrestin2 assays compared to parent 6d and both
6
6
a and 9a (Table 2). Interestingly, it had improved aqueous solubility over the parent compound
d. In comparison with parent compound GAT211, there was also a significant decrease in the
liver microsomal clearance of compound 6s in both human and rat (Table 3). Thus, these functional
activity-based SAR study identified, the two tri-fluoro compounds 6r and 6s as the potent CB1R
ago-PAMs with improved metabolic stability and compound 9j as a moderate PAM with improved
aqueous solubility over the parent compound 6d.
ACS Paragon Plus Environment
2
3

Journal of Medicinal Chemistry
Page 24 of 81
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 3. Metabolic stability and kinetic aqueous solubility of GAT211(6d), 6r, 9j and 6s.
Human
microsomal
Clint
(µL/min/mg
protein)
Rat
microsomal
Clint
(µL/min/mg
protein)
Kinetic
aqueous
solubility
(M)
Compound
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
t1/2 (min)
t1/2 (min)
GAT211 (6d)
81.2
28.4
368
6.27
1.56
6
r
44.9
84.5
44.4
51.5
27.4
52.1
248
556
94.0
9.31
4.08
24.6
1.56
3.13
1.56
9
j
6
s
Further in vitro and in vivo Characterization:
Agonism and positive allosteric modulation in HEK293 cells:
Following our initial SAR assessment, the key analogs 6r, 6s, and 9j were characterized for cAMP
inhibition in HEK293 cells stably expressing GFP-tagged rat CB1R (GFP-rCB1R). As with our
initial assessment, cells were treated with compound for 30 min; unlike the initial assessment,
however, cAMP levels were quantified via time-resolved fluorescence resonance energy transfer
(TR-FRET) in a distinct cell line for both agonist activity (Fig. 2) and concentration-dependent
PAM activity (Table 4; Fig. 3).
ACS Paragon Plus Environment
2
4

Page 25 of 81
Journal of Medicinal Chemistry
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
Figure 2. Agonist effect of cAMP inhibition for 6r, 9j, and 6s. HEK293 cells stably-expressing GFP-rCB1R were
treated with 1 pM – 10 μM key compounds alone for 30 min in the presence of 5μM Forskolin and cAMP inhibition
was measured. Data are expressed as % Forskolin response and were fit to a nonlinear regression (4 parameter model,
GraphPad v. 7) for statistics in Table 4. Data are mean ± S.E.M., n ≥ 3 independent experiments performed in triplicate.
Compounds 6r and 6s both displayed greater potency and efficacy as allosteric agonist than the
parent compound 6d for cAMP inhibition (Table 4; Fig. 2). 9j displayed greatly reduced agonist
activity relative to GAT211 (Table 4; Fig. 2) as observed in the initial assessment. Increasing
concentrations of all ago-PAMs produced a demonstrable downward and leftward shift in the
CP55,940-dependent concentration-response curve, as expected for a CB1R ago-PAM. The
magnitude of both potency and efficacy shift was greatest for 6r and 6s and least for 9j (Table 4;
Fig. 3). For both agonist and PAM activity evaluations, 6s was slightly, but not significantly, more
potent than 6r, which was consistent with initial SAR evaluations (Tables 1, 2).
ACS Paragon Plus Environment
2
5

Journal of Medicinal Chemistry
Page 26 of 81
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
Figure 3. Positive allosteric modulation of cAMP inhibition for 6r, 9j, and 6s. HEK293 cells stably-expressing GFP-
rCB1R were treated with 1 pM – 10 μM CP55,940 and 10, 30, 100, or 300 nM GAT compounds alone for 30 min in
the presence of 5μM Forskolin and cAMP inhibition was measured. Data are expressed as % Forskolin response and
were fit to a nonlinear regression (4 parameter model, GraphPad v. 7) for statistics in Table 4. Data are mean ± S.E.M.,
n ≥ 3 independent experiments performed in triplicate.
Radioligand binding and GTPS assay: The parent compound GAT211 (6d) was previously
3
3
shown to enhance the binding of [ H]CP55,940 and reduce the binding of [ H]SR141716A at
25
hCB1R in membranes isolated from CHO cells stably-expressing hCB1R. This paradoxical
enhancement of agonist binding and
ACS Paragon Plus Environment
2
6

Page 27 of 81
Journal of Medicinal Chemistry
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
3
3
Figure 4. Effect of 6r, 9j, and 6s on the binding of a) 1 nM [ H]SR141716A or b) [ H]CP55,940 to membranes
35
obtained from CHO cells stably-expressing human CB1R; or the binding of [ S]GTPS to G protein in membranes
obtained from CHO cells stably-expressing human CB1R and treated with c) CP55,940 and 1 μM GAT compound,
i
or d) GAT compound alone. K (nM) and EC50 (nM) are mean with 95% CI. EMax data are mean ± SEM. Data were fit
to a variable slope (4 parameter) nonlinear regression model, mean ± SEM, n = 4-6 independent experiments
performed in duplicate. ***P < 0.001 Student’s one sample t-test.
displacement of antagonist binding has been observed for all other CB1R allosteric modulators
described to date, including the NAMs Org27569 and PSNCBAM-1. The tri-fluorinated
3
compounds 6r and 6s reduced [ H]SR141716A binding at hCB1R with low nanomolar affinities
ACS Paragon Plus Environment
2
7

Journal of Medicinal Chemistry
Page 28 of 81
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
of 4.8 nM and 5.2 nM, respectively (Fig. 4a) in accordance with their high potency and efficacy in
3
vitro (Table 1). The aza-fluoro-hybrid 9j reduced [ H]SR141716A binding in the micromolar
3
range (2.3 µM; Fig. 4a). Compounds 6r, 6s, and 9j enhanced [ H]CP55,940 binding to hCB1R
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
similar to 6d (K
s, and 9j at 1M concentration significantly increased CP55,940-dependent G-protein coupling
(0.1 and 1 nM) (Fig. 4c). Both 6r and 6s enhanced G-protein coupling with nanomolar potency
i
= 88 nM for 6r, K
i
= 94 nM for 6s, K
i
= 110nM for 9j) (Fig. 4b). Compounds 6r,
6
(EC50 = 47 nM for 6r, K
i
= 33 nM for 6s), whereas 9j was less potent than the parent compound
(EC50 = 170 nM for 9j) when tested alone (Fig. 4d). These data support a SAR-driven correlation
between potent CB1R PAM activity, enhancement of agonist binding, and reduction of antagonist
binding.
Agonism, positive allosteric modulation, and bias in CHO cells: Our initial assessments of
CB1R-dependent PAM activity utilized a high throughput screening approach to gather
preliminary data in the presence of CP55,940 at EC20 for cAMP inhibition at 30 min and arrestin2
recruitment at 90 min. Given the potential for our compounds to exhibit mixed agonist/PAM
activity (i.e. ago-PAMs), it was important to assess their bias and PAM activity at a higher
concentration of orthosteric agonist, because the bias activity of PAMs – particularly ago/PAMs –
50
may change as a function of orthosteric agonist concentration. Further, we needed to have greater
statistical power for assessing ligand bias between cAMP inhibition and arrestin2 recruitment.
Therefore, the lead ago-PAMs (6r, 6s, and 9j) were assessed with greater power (n ≥ 6) (i) alone
to quantify intrinsic agonist activity, (ii) in the presence of 100 nM CB1R agonist (CP55,940) to
ACS Paragon Plus Environment
2
8

Page 29 of 81
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 4. Agonist activity and PAM-dependent shift in CP55,940 potency and efficacy for lead
GAT211 derivative compounds.
Compound Agonist
EC50 (nM)
EMin (% Forskolin)
Activity
CP55,940
18 ± 0.11
29 ± 3.5
39 ± 7.5
32 ± 3.5
55 ± 8.4
30 ± 3.9
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
GAT211 (6d)
240 ± 0.17
6
r
27 ± 0.11
9j
690 ± 2.6
6
s
26 ± 0.13
Positive Allosteric Modulator Activity
CP55,940 EC50 (nM) at Log[Compound] M
Compound
Vehicle (0)
-8
-7.5
-7
-6.5
∆EC50 (nM)
-5.1 ± 4.6
-12 ± 5.7
GAT211 (6d)
17 ± 0.92 17 ± 2.6
17 ± 5
19 ± 11
17 ± 1.7
10 ± 7.5
17 ± 3
6
r
20 ± 2.6
22 ± 10
14 ± 3.8
27 ± 11
2
2 ± 6.0
9j
20 ± 5.9 18 ± 1.9
-4.9 ± 3.8
-12 ± 8.0
6
s
12 ± 3.5 10 ± 3.9 9.0 ± 2.0
CP55,940 EMax (% Forskolin) at Log[Compound] M
Compound
6d
Vehicle (0)
-8
101 ± 7.4 100 ± 11 82 ± 6.7
76 ± 6.4
101 ± 9.0 108 ± 13 103 ± 12 83 ± 9.3
74 ± 5.7 63 ± 3.7 52 ± 1.5 33 ± 13
-7.5
-7
-6.5
∆EMax (% Forskolin)
75 ± 11
25 ± 3.0
68 ± 4.3
17 ± 10
67 ± 15
6
r
58 ± 9.2 46 ± 6.6 32 ± 4.7
1
00 ± 6.4
9
6
j
s
Compound activity was quantified for cAMP inhibition in cells treated with 10 - 300 nM GAT compound + CP55,940.
Data were fit to a variable slope (4 parameter) non-linear regression in GraphPad (v. 7). n ≥ 3 independent experiments
performed in triplicate from Figures 2 and 3. EMin from figure 2 refers to the lower plateau (i.e. bottom) of the
concentration-response curves observed with high concentrations of compound. EMax from Figure 3 refers to the
upper plateau (i.e. top) of the concentration-response curves observed at low concentrations of orthosteric agonist
CP55,940. Data are expressed as the difference between the agonist (CP55,940) alone and the maximal concentration
of GAT compound used, mean ± SEM.
determine the extent of PAM enhancement at greater orthosteric agonist concentrations, (iii) at
matched time points (90 min for both cAMP inhibition and arrestin2 recruitment), and (iv) for
signaling bias. cAMP inhibition data were normalized to % compound maxima for each respective
compound in order to fit data to the operation model and estimate signaling bias (Table 4; Fig. 5).
Compounds 6r, 6s, and 9j each directly activated CB1R-dependent cAMP inhibition (90 min),
ACS Paragon Plus Environment
2
9