Probing the Existence of a Metastable Binding Site at the β2-Adrenergic Receptor with Homobivalent Bitopic Ligands

Article abstract of DOI:10.1021/acs.jmedchem.9b00595

Herein, we report the development of bitopic ligands aimed at targeting the orthosteric binding site (OBS) and a metastable binding site (MBS) within the same receptor unit. Previous molecular dynamics studies on ligand binding to the β₂-adrenergic receptor (β₂AR) suggested that ligands pause at transient, less-conserved MBSs. We envisioned that MBSs can be regarded as allosteric binding sites and targeted by homobivalent bitopic ligands linking two identical pharmacophores. Such ligands were designed based on docking of the antagonist (S)-alprenolol into the OBS and an MBS and synthesized. Pharmacological characterization revealed ligands with similar potency and affinity, slightly increased β₂/β₁AR-selectivity, and/or substantially slower β₂AR off-rates compared to (S)-alprenolol. Truncated bitopic ligands suggested the major contribution of the metastable pharmacophore to be a hydrophobic interaction with the β₂AR, while the linkers alone decreased the potency of the orthosteric fragment. Altogether, the study underlines the potential of targeting MBSs for improving the pharmacological profiles of ligands.

Full text of DOI:10.1021/acs.jmedchem.9b00595

Subscriber access provided by BUFFALO STATE
Article
Probing the existence of a metastable binding site at the
2-adrenergic receptor with homobivalent bitopic ligands
#
Birgit I. Gaiser, Mia Danielsen, Emil Marcher-Rørsted, Kira Røpke Jørgensen,
Tomasz Marcin Wróbel, Mikael Frykman, Henrik Johansson, Hans Bräuner-
Osborne, David E. Gloriam, Jesper Mosolff Mathiesen, and Daniel Sejer Pedersen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00595 • Publication Date (Web): 12 Jul 2019
Downloaded from pubs.acs.org on July 13, 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 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Probing the existence of a metastable binding site at the β -adrenergic receptor with
2
homobivalent bitopic ligands
a
a
Birgit I. Gaiser, Mia Danielsen, Emil Marcher-Rørsted, Kira Røpke Jørgensen, Tomasz M.
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
Wróbel, Mikael Frykman, Henrik Johansson, Hans Bräuner-Osborne, David E. Gloriam, Jesper
*
,b
*,b
Mosolff Mathiesen, Daniel Sejer Pedersen
Department of Drug Design and Pharmacology, University of Copenhagen, Jagtvej 162, 2100
Copenhagen (Denmark)
ABSTRACT
Herein we report the development of bitopic ligands aimed at targeting the orthosteric (OBS)
and a metastable binding site (MBS) within the same receptor. Previous molecular dynamics
studies on ligand binding to the β -adrenergic receptor (β AR) suggested that ligands pause at
2
2
transient, less-conserved MBSs. We envisioned that MBSs can be regarded as allosteric
binding sites and targeted by homobivalent bitopic ligands linking two identical
pharmacophores. Such ligands were designed based on docking of the antagonist (S)-alprenolol
into the OBS and a MBS, and synthesized. Pharmacological characterization revealed ligands
with similar potency and affinity, slightly increased β /β AR-selectivity and/or substantially
2
1
slower β AR off-rates compared to (S)-alprenolol. Truncated bitopic ligands suggested the
2
major contribution of the metastable pharmacophore to be a hydrophobic interaction with the
β2AR, while the linkers alone decreased the potency of the orthosteric fragment. Altogether,
the study underlines the potential of targeting MBSs for improving the pharmacological
profiles of ligands.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 141
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
G protein-coupled receptors (GPCRs) constitute the largest and most diverse family of
eukaryotic membrane-bound receptors and are activated by a broad variety of both endogenous
and exogenous ligands, including small molecules, lipids, peptides, proteins, ions, light and
odorants.1–3 Consequently, GPCRs mediate numerous cellular responses and are involved in
many pathophysiological processes, making them attractive targets in human drug therapy.
Almost 500 drugs representing around 35% of all FDA approved drugs act at GPCRs.3–7 Since
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
8
the publication of the first mammalian GPCR crystal structure in 2000, an increasing number
of GPCR structures have been solved providing important insights into the conformation of
inactivated as well as activated GPCRs, thus facilitating GPCR structure-based drug design.2,9–
1
1
To date, drug discovery has focused mostly on targeting orthosteric binding sites (OBSs) of
GPCRs. Such binding sites are generally highly conserved across receptor subtypes, and hence
orthosteric ligands commonly evoke undesired side effects due to targeting of closely-related
receptors.12,13 Today it is generally acknowledged that all GPCR families contain additional,
topographically distinct allosteric binding sites (ABSs) that are located in less conserved areas
among receptor subtypes. Drug discovery efforts with the aim of targeting ABSs thus bear the
potential to provide subtype selective ligands. However, this approach is hampered by the lack
of knowledge with regards to the exact location and shape of ABSs.11,12,14–17 To improve
subtype selectivity one recent approach has been to design bitopic ligands. Such ligands contain
both an orthosteric and an allosteric pharmacophore that are connected via a linker, thus
allowing the molecule to simultaneously bind the OBS and an ABS. In theory, bitopic ligands
combine the advantages of orthosteric and allosteric ligands by connecting a high affinity
pharmacophore with a subtype selective pharmacophore, respectively. Moreover, such ligands
may provide additional advantageous properties such as increased residence time or biased
ACS Paragon Plus Environment

Page 3 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
signalling profiles. However, as for monomeric allosteric modulators, it remains a challenge to
design bitopic ligands due to the lack of structural information on ABSs.14,18–20 Nevertheless,
2
1
several examples of bitopic ligands have been reported for a number a GPCRs, including the
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
22
23–31
adenosine A receptor, the dopamine D and D receptors
and the muscarinic M and M
1 2
1
2
3
receptors.17,32–36
During the past decade, GPCR structure elucidation and computational advances have enabled
extensive molecular dynamics (MD) simulations of entire proteins and their surroundings.
Several independent MD studies investigating GPCR ligand recognition, have reported that
ligands in silico pause at different transient binding sites before entering the OBS to adopt their
final pose.20,37–45 Such binding sites have been termed metastable binding sites (MBSs) and
may be located in less conserved areas among receptor subtypes. Dror et al. were among the
2
0
first to thoroughly investigate ligand association with a GPCR and to describe MBSs. It was
proposed that the in silico discovery and mapping of MBSs might be used in the design of
allosteric ligands, thus overcoming the lack of structural information on ABSs. Recently, we
proposed that MBSs may serve as a source of ABSs and could be used in the design of
homobivalent bitopic ligands.21 Unlike heterobivalent bitopic ligands that simultaneously
target the OBS and an ABS, we hypothesized that homobivalent bitopic ligands comprised of
two identical pharmacophores could be designed to simultaneously target the OBS and a MBS
(Figure 1), and that such ligands could exhibit receptor subtype selectivity as well as improved
potency and binding kinetics. Moreover, since homobivalent bitopic ligands (referred to as
bitopic ligands in the following) are comprised of identical pharmacophores, their syntheses
should in theory be relatively simple. Since MBSs to date have only been reported
computationally, our aim was also to probe the existence of MBSs in vitro.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 141
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 1. Heterobivalent bitopic ligands consisting of an orthosteric and an allosteric
pharmacophore target the orthosteric binding site (OBS) and an allosteric binding site (ABS)
at the receptor. We hypothesize that metastable binding sites (MBSs) can serve as a source of
ABSs and that homobivalent bitopic ligands consisting of two identical pharmacophores could
be designed to simultaneously target the OBS and a MBS to provide high-potency ligands with
improved receptor subtype selectivity and binding kinetics. Reprinted with permission from
Fronik, P. et al., J. Med. Chem. 2017, 60 (10), 4126-4134.21 Copyright 2017 American
Chemical Society.
ACS Paragon Plus Environment

Page 5 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESULTS & DISCUSSION
Ligand design
2
0
Ligand design was based on the MD work by Dror et al. that utilized an inactive X-ray crystal
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
structure of the β -adrenergic receptor (β AR) and the antagonist (S)-alprenolol (ALP) amongst
2
2
other β-blockers. In several unbiased and unguided MD simulations, ligands were found to
2
0
pause at two distinct MBSs en route to the OBS. We found one of these MBSs to be
particularly suitable for the design of a bitopic ligand due to the distance and orientation of the
ligand relative to the OBS and a predicted affinity of 5 µM. Two major contributions to ligand
recognition were proposed by Dror et al. for this MBS: i) a salt bridge between the ligand amine
with the carboxylate side-chain of Asp300, and ii) a hydrophobic interaction of the ligand’s
aromatic ring with the receptor surface.20
Analogous to Dror et al.20 we utilized the β AR-ALP X-ray co-crystal structure (PDB:
2
4
6
3
NYA). Despite a relatively low resolution of 3.1 Å, this structure was used for docking
because it retains the overall receptor-fold compared to other inactive β AR-ligand co-crystal
2
structures of higher resolution. Furthermore, ALP and other β AR ligands bind to the receptor
2
in the same general pose.
During the protein preparation process for ligand docking, special attention was drawn to
aspartic acid Asp300 in extracellular loop (ECL) 3 (Figure 2). From the 3NYA Asp300 rotamer
library, the side-chain rotamer with a carboxylate orientated towards the MBS was selected.
This modification was considered consistent with the relatively flexible nature of ECL3. After
optimization and restrained minimization, a Ramachandran plot was generated that showed no
residues in disallowed regions. (S)-ALP was drawn in the 2D sketcher of the Schrödinger 2015-
1 small molecule drug dsi suite from which its 3D structure was generated.47–51 It was prepared
with the LigPrep tool using the OPLS_2005 force field. After minimization, (S)-ALP was
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 141
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
docked into the prepared protein in two consecutive steps with the Glide docking tool. First, a
receptor grid was generated based on the position and size of the crystallized ligand. (S)-ALP
was docked into the OBS applying a rigid protein docking protocol. The (S)-ALP output poses
were found to match well with the co-crystallized (S)-ALP. Subsequently, a second grid was
generated from the receptor containing (S)-ALP in the OBS. This time, position and size of the
grid were manually curated to cover the entire extracellular vestibule. Applying the same
docking protocol except for the selection of the grid, (S)-ALP was found in the MBS with its
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
0
ammonium nitrogen interacting with Asp300 as proposed by Dror et al. (Figure 2). Based on
visual comparison of the docking poses with the orientation of (S)-ALP in the MD simulation
by Dror et al., a conformational analysis and docking scores, eight out of the ten output
structures were deemed of equal quality. From the model with an (S)-ALP ligand in both the
OBS and the MBS, two possibilities to link the molecules became apparent leading to two
different types of bitopic ligands (Figure 2). Linkage via the allyl chains would give rise to
symmetric bitopic ALP ligands, whereas linking the orthosteric ligand’s isopropyl group and
the metastable ligand’s aromatic ring would lead to non-symmetric bitopic ALP ligands. The
crevice depicted left in Figure 2A was the preferred ALP entry pathway in the simulations by
2
0
Dror et al. and furthermore an ALP derivative extending from the allyl group through that
5
2
cleft has been reported previously. Thus, despite its narrowness, this crevice was expected to
be able to accommodate the linkers of symmetric bitopic ligands. Moreover, we anticipated
that such symmetric ligands would be more amenable to synthesis, and thus, we selected to
move forward with this class of bitopic ligands.
ACS Paragon Plus Environment

Page 7 of 141
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. (A) The β AR represented as its surface (PDB: 3NYA) co-crystallized with (S)-ALP
2
(in cyan) and two docked (S)-ALP ligands in the orthosteric and the metastable binding site (in
purple). The side chain of aspartic acid 300 is shown as lines and its interaction with the ligand
in the metastable binding site is represented as a yellow dashed line. Yellow dashed lines
between the ligand in the orthosteric and metastable binding sites indicate potential linkages
between the two pharmacophores and the distance. (B) Generalized structures of potential
bitopic ligands based on the docked poses. Adapted with permission from Fronik, P. et al., J.
2
1
Med. Chem. 2017, 60 (10), 4126-4134. Copyright 2017 American Chemical Society.
In the design of bivalent ligands, the linker attachment point, length and composition need to
be considered.18,19,53 From structure-activity relationship studies on β-adrenergic receptor
antagonists, it is known that a broad variety of substituents on the aromatic ring and the amino
nitrogen are well-tolerated, provided that the aryloxypropanolamine motif remains intact.54,55
Based on our docking study (Figure 2A) we decided to replace the allyl side chains with a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 141
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
linker to connect the two aromatic rings ortho to the propanolamine (PA) substituent (Figure
2
B, top). In the model, the distance between the termini of the two allyl substituents was found
to be 8.0 Å (Figure 2A). To bridge a gap of this length a chain of ca. 5 atoms would be required
assuming a bond length of ~1.5 Å and a bond angle of 109.5°). However, based on the inherent
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
(
flexibility of the allyl substituents that allows them to adopt an orientation towards each other,
the linker spanning the two aromatic motifs was estimated to be 9-11 atoms. Concerning linker
composition, it is important that the linker is sufficiently flexible to allow the two
pharmacophores to adopt the required conformation for binding, but at the same time
1
8
sufficiently rigid to keep the conformational energy penalty of binding at a minimum. With
the aim of covering different degrees of flexibility and other physicochemical properties, all-
carbon, polyethylene glycol (PEG), urea, diamide and triazole linkers were selected as
synthetically feasible linkers that utilize several common intermediates, which would simplify
synthesis (Table 1). To facilitate the evaluation of the linker design we decided to span a range
of 6-15 atoms, which based on our calculations would render some ligands too short and others
too long. Additionally, it is known that the therapeutic effect of aryloxypropanolamine-based
β-blockers relies almost entirely on the (S)-enantiomer even though they generally are
administered as racemates.56,57 As we were interested in the importance of the configuration of
the stereocenter in the PA motif of our bitopic ligands, we docked (R)-ALP into the β AR in a
2
manner analogous to the one describe above for (S)-ALP. (R)-ALP was found to adopt an
almost identical pose in the two binding sites, including the salt bridge and the hydrophobic
interaction with the receptor surface in the metastable binding site (data not shown). To test the
in vitro pharmacological effect of inverting the stereocenters, the (R,R)-configured version of
the diamide linked ligand 4d was included in the study.
ACS Paragon Plus Environment

Page 9 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 1. Proposed symmetric bitopic (S)-ALP ligands. The suggested ligands cover different
linker lengths and a spectrum of physicochemical properties. Linker lengths are given as the
number of atoms spanning the two aromatic rings. *For the diamide linked ligand 4d it was
also proposed to prepare the enantiomer (R,R)-4d.
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
Linker length
6
7
8
9
10
11
12
13
14
15
OPA
OPA
1a
1b
1c
1d
n
n=6
n=8
n=10
n=12
PA=
OPA
OPA
2a
2b
HN
O
O
O
O
HO
n
n=1
n=2
OPA
OPA
3
a
3b
3c
3d
n
n
N
N
H
H
n=2
n=3
n=4
n=5
4
a
4b
4c
4d*
4e
4f
n=3
OPA
OPA
H
H
N
m
n
n
N
n=1
n=2
n=2
n=3
n=3
O
O
m=2
m=2 m=3 m=2 m=3 m=4
OPA
OPA
5a
5b
5c
5d
n
N
n
N
H
N
N
N
N
N
n=0
n=1
n=2
n=3
Synthesis
Synthesis of optically active alprenolol
Initially, we set out to synthesize (S)- and (R)-ALP to characterize these pharmacologically and
confirm that (S)-ALP is in fact the more potent enantiomer.56,57 Moreover, the ALP monomers
would be utilized as reference compounds for the potential bitopic ligands. The synthesis was
achieved according to a published procedure using the activating agent m-nitrobenzene
sulfonyl (mNs) chloride that suppresses racemization during alkylation reactions (Scheme
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 141
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
).58,59 Glycidols (R)- and (S)-6 were activated to give the corresponding mNs-glycidols (S)-
and (R)-7. The propanolamine (PA) motif was then installed by alkylation of 2-allyl phenol
with (S)- and (R)-7 to produce (S)- and (R)-8, respectively, and subsequent treatment of the
epoxides with isopropylamine yielding the two ALP enantiomers (S)- and (R)-9.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Synthesis of (R)- and (S)-ALP, construction of the propanolamine (PA) motif.a
O O
S
a
O
HO
O
O
H
H
NO2
(
R)-6
(S)-6
(S)-7
(R)-7
HN
O
HO
O
OH
O
b
c
(
S)-8
(R)-8
(S)-9 = (S)-ALP
(R)-9 = (R)-ALP
a
Reagents and conditions: (a) mNsCl, Et N, toluene, −20 °C, 18-25 h; 80-95%; (b) K CO ,
3
2
3
i
acetone, reflux, 1-2 h, then: (S)- or (R)-7, reflux, 13-24 h, 65-70%; (c) PrNH , EtOH, reflux,
2
1
-23 h, 80-85%, e.r. ((S)-ALP) = 95:5, e.r. ((R)-ALP) = 99:1.
The synthesis of (S)- and (R)-ALP worked as expected and the enantiomeric purities were
determined by chiral HPLC to 95% and 99% for (S)- and (R)-ALP, respectively (see Supporting
Information, 4. NMR and HPLC data). The enantiomeric purities reflect that of the
commercially available glycidols, thus no detectable stereochemical leakage occurs in
5
9
accordance with that reported by Sharpless and co-workers. The PA motif was installed using
essentially the same chemistry in all following syntheses. Thus, it is reasonable to assume that
ACS Paragon Plus Environment

Page 11 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
no significant stereochemical leakage occurred in the synthesis of all ligands reported in the
following. Furthermore, none of the bitopic ligands gave rise to mixtures of diastereoisomers
that could be detected by NMR or HPLC, supporting that the stereochemical integrity was not
compromised.
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
Synthesis of bitopic ligands
Initially we envisaged that the all-carbon linked ligands could be synthesized by a strategy
involving an Fe-catalyzed aryl-alkyl cross-coupling followed by Grubbs metathesis and final
construction of the PA motif (Scheme 2, top). Two different Fe-catalyzed aryl-alkyl cross-
coupling methods reported by Cahiez et al. and Czaplik et al. were explored.60,61 However,
despite repeated experiments and careful attention to detail, all attempts using both methods
with 2-bromoanisole (10) and 5-bromo-1-pentene (11) as the coupling partners failed. Attempts
at lowering the reaction temperature did not improve the outcome. At best, trace amount of the
desired product was detected, the main components being starting material, the aryl-aryl or
alkyl-alkyl homo-coupled products. Based on these poor results we decided to explore a
double-Sonogashira strategy instead using 1,7-octadiyne and 2-iodoanisole (13) (Scheme 2).
To remove molecular oxygen from the reaction mixture it was necessary to heat the catalysts
extensively under vacuum before flushing with argon and adding the substrates and base.62
Under these conditions, the desired double Sonogashira coupling product 14 could be obtained.
However, yields were repeatedly found to be >>100% even after flash column chromatography
that removed unreacted starting material. Since NMR analysis only showed the desired
product, we speculate that the unidentified impurity is of inorganic nature. It was attempted to
carry the synthesis on with the aim of removing the impurity after catalytic hydrogenation on
the alkyne moieties, but to our surprise catalytic hydrogenation of diyne 14 to give the fully
saturated analogue 15 did not proceed at all under standard conditions. Screening of reaction
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 141
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
conditions including the use of Pd/C and Pearlman’s catalyst (Pd(OH) /C), high catalyst
2
loadings, various solvents, elevated temperatures, increased pressure and the use of NH HCO
4
2
in a microwave assisted catalytic transfer hydrogenation, did not provide any detectable amount
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
63
of the desired bis-anisole or any of the partly reduced intermediates. It remains unclear what
the cause for this lack of reactivity was, but we speculate that the unidentified impurity
hampered the reaction or that aggregation and/or folding of the molecules due to π-π-stacking
may prevent association of the diyne 14 with the palladium catalyst in a productive way
(Scheme 2, Box). Eventually, the synthesis of the all-carbon linked ligands was realized using
Wittig chemistry (Scheme 2, bottom). The required alkyl bis-phosphonium iodides 18a-d were
generated from the corresponding bis-bromides 16a-d.64–66 In one instance, this was achieved
67
directly from bromide 16a via an in situ Finkelstein reaction, but as formation of potassium
bromide required separation of the latter from the Wittig salt, the corresponding two step
sequence with isolation of the alkyl iodide intermediates 17b-d was preferred for the remaining
three cases (16b-d). The Wittig salts were treated with potassium tert-butoxide and 2-
benzyloxybenzaldehyde to give the desired bis-alkenes 19a-d. The byproduct
triphenylphosphine oxide was effectively separated from the reaction mixture by treatment of
the crude products with anhydrous magnesium chloride in a mixture of toluene and n-heptane
6
8
according to the method by Lukin et al. Catalytic hydrogenation of the Wittig products
saturated the double bonds and removed the benzyl groups simultaneously to give bis-phenols
20a-d. Finally, the bis-phenols were subjected to the construction of the PA motif as described
above (Scheme 1).
ACS Paragon Plus Environment

Page 13 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 2. Synthesis of the all-carbon linked ligands 1a-d.a
OMe
OMe
OMe
OMe
Br
a
R
R
+
Br
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
10
11
12
OMe
OMe
OMe
OMe
OMe
8
I
4
c
b
13
14
15
n-2
OBn
OBn
n
n
I
d
e
I
n
g
Br
Br
I
I
Ph P
PPh3
3
16b-d
17b-d
18a-d
19a-d
4
f
Br
Br
h
16a
HO
O
HO
O
i
i
NH Pr
NH Pr
OH
OH
a n = 4
b n = 6
c n = 8
i
n
d n = 10
n
2
0a-d
1a-d
a
Reagents and conditions: (a) Method A: Fe(acac) , Mg, TMEDA, HMTA, THF; Method B:
3
FeCl , Mg, TMEDA, THF, 0 °C; (b) 1,7-octadiyne, Pd(PPh ) Cl , CuI, Et N, rt, 23 h, >95%
3
3 2
2
3
(
crude); (c) Attempted hydrogenation conditions: (1) Pd(OH) /C or Pd/C, H , AcOH, 80 °C,
2 2
20 h; (2) Pd/C, H , DMF/AcOH, 80 °C, 26 h; (3) Pd/C, H , dioxane/AcOH, 80 °C, 22 h; (4)
2
2
Pd/C, H , dioxane, rt, 5.5 bar, 21 h; (5) Pd(OH) /C, AcOH, 60 °C, 4 bar, 22 h; (6) Pd/C,
2
2
NH HCO , THF, 100 °C (sealed microwave vial), 1 h; (d) NaI, acetone, rt, 20-22 h; (e) PPh ,
4
2
3
MeCN, reflux, 20-24 h, 84-95% (over 2 steps); (f) KI, PPh , MeCN, 120 °C, 26 h, >95%
3
t
(
crude); (g) KO Bu, 2-benzyloxybenzaldehyde, THF, 3Å molecular sieves, 0 °C to reflux, 18-
2
6 h, 33-72%; (h) Pd/C, H , THF, rt, 20 h to 4 days, 70-95%; (i) By the general method for
2
synthesis of the PA motif (Scheme 1) ligands 1a-d were isolated in 15%, 34%, 9% and 10%
yield, respectively. 19a-d were a mixture of geometrical isomers. For clarity only the (E,E)-
isomer is depicted.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 141
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
The two PEG linked ligands 2a and 2b were synthesized from di- and triethylene glycol (21a,b)
that were activated with methanesulfonyl (Ms) chloride (Scheme 3). The resulting mesylates
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
2a and 22b were connected to two 2-benzyloxyphenol moieties followed by removal of the
benzyl groups by catalytic hydrogenation to give 23a,b and 24a,b, respectively. In the case of
2
a, the PA motif was attached as described above (Scheme 1), whereas for 2b it was installed
i
using NaH and (S)-7, followed by PrNH in DMF.
2
Scheme 3. Synthesis of the PEG linked ligands 2a,b.a
a n=1
b n=2
n
OR
OR
O
OR
b
O
O
O
n
OR
R=Bn 23a,b
R=H 24a,b
R=PA 2a,b
R=H 21a,b
R=Ms 22a,b
c
a
d
a
Reagents and conditions: (a) MsCl, Et N, DCM, 0 °C to rt, 16-64 h; (b) K CO , 2-
3
2
3
benzyloxybenzaldehyde, acetone, reflux, 23 h; (c) Pd/C, H , THF, rt, 26 h, 16-26% (over 3
2
steps); (d) By the general method for synthesis of the PA motif (Scheme 1) ligand 2b was
i
isolated in 44% yield; for 2a: NaH, (S)-7, DMF, rt, 3 h, then: PrNH , 80 °C, 2 h, 10%.
2
The synthesis of diamide linked ligands 4a-f was achieved from lactones 26-28 by ring-opening
with a series of diamines (Scheme 4). Lactones 26 and 27 were commercially available and 28
was prepared from α-tetralone (25) by Baeyer-Villiger oxidation.69 Ring-opening of the
lactones with the diamines led to the corresponding bis-phenols 29a-f. In some cases, closely
eluting byproducts hampered the isolation of the products and thus, only poor yields could be
ACS Paragon Plus Environment

Page 15 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
obtained. The bis-phenols were carried on to the final products 4a-f and (R,R)-4d following the
above described reaction sequence for the construction of the PA motif.
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 4. Synthesis of the diamide linked ligands 4a-f and (R,R)-4d.a
O
a n=1, m=2 d n=3, m=2
b n=2, m=2 e n=3, m=3
c n=2, m=3 f n=3, m=4
25
a
OR
OR
H
H
n
n
O
N
N
b
O
m
O
O
n
26 n=1
27 n=2
28 n=3
R=H
R=PA
29a-f
4a-f
R=(R)-PA (R,R)-4d
c
aReagents and conditions: (a) for 28: mCPBA, DCM, rt, 42 h, >95% (crude); (b) ethane-1,2-
diamine, propane-1,3-diamine or butane-1,4-diamine, EtOH, rt to reflux, 16-80 h, 10-95%;
(c) By the general method for synthesis of the PA motif (Scheme 1) ligands 4a-f and (R,R)-
4
d were isolated in 35%, 20%, 25%, 3%, 45%, 45% and 34% yield, respectively.
Next, we turned our attention to the triazole linked ligands 5a-d. Unlike the bitopic ligands
synthesized thus far, we intended to prepare these from monomers with the PA motif
preinstalled. Hence, the preparation of these ligands required the syntheses of the relevant azide
monomers 32, 38, 43 and 47 that would be subjected to a double copper-catalyzed azide alkyne
cycloaddition (CuAAC) (Scheme 5 bottom). Azide monomer 32 was prepared from 2-
aminophenol (30) that was converted to the corresponding azide 31 in a diazotransfer
7
0
reaction, followed by construction of the PA motif. For the synthesis of azide 38, the same
reaction sequence was envisioned using 2-aminomethyl phenol (33) as the starting material. A
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 141
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
diazotransfer reaction gave the desired azide 34, however, we were unable to alkylate the
phenol in the first step of the PA motif construction. Different reaction conditions were
screened including the use of a variety of bases, solvents, temperatures and modes of addition,
but to no avail. We speculate that the in situ generated phenolate polymerizes via nucleophilic
attack on the benzylic carbon with expulsion of the azide rather than with the intended activated
glycidol derivative. To circumvent this problem amine 33 was instead protected with tert-
butyloxycarbonyl (Boc) prior to attachment of the PA motif. Removal of the Boc group of 37
gave the corresponding amine, which was successfully converted to azide 38 in a diazotransfer
reaction.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 17 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 5. Synthesis of the azide monomers 32, 38, 43 and 47 and synthesis of bis-triazole
linked bitopic ligands 5a-d using double CuAAC reactions.a
OH
OH
OPA
NH2
N3
31
N3
a
b
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
30
32
OH
OH
O
O
c
N3
a
NH2
N3
33
34
35
d
OH
OPA
OPA
e,a
b
N3
NHBoc
NHBoc
36
38
37
OR
OH
R
OH
f
a
OH
N3
39
R=CN
R=CH NH 41
40
R=H
42
g
b
2
2
R=PA 43
h
i, j
O
O
OH OH
OR N3
44
R=H
46
45
b
R=PA 47
OPA
OPA
OPA
n
n
n
k
N3
N
N
H
N
N N
N N
32 n=0
38 n=1
43 n=2
47 n=3
5a n=0
5b n=1
5c n=2
5d n=3
a
Reagents and conditions: (a) NaHCO or K CO , CuSO , 1H-imidazole-1-sulfonyl azide
3
2
3
4
hydrochloride, MeOH/H O (4:1, v/v), rt, 3-18 h, 25-95%; (b) By the general method for
2
synthesis of the PA motif (Scheme 1) compounds 32, 37, 43 and 47 were isolated in 46%,
6
2
8%, 68% and 64% yield, respectively; (c) Attempted conditions: (1) K CO , acetone, 70 °C,
2 3
t
3 h; (2) K CO , DMSO, 60 °C, 20 h; (3) KO Bu, DMF, 80 °C, 16 h; (4) NaH, DMF, rt, 2 h;
2
3
(5) Et N, DCM, 50 °C, 24 h; (6) Et N, THF, 70 °C, 23 h; (d) (Boc) O, Et N, MeOH, rt, 17 h,
3
3
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 141
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
7
5%; (e) TFA, TIS, DCM, 0 °C to rt, 21-48 h, 65%; (f) NaCN, DMF, 160 °C, 14 h, 50%; (g)
BH ×SMe , THF, 70 °C, 16 h, >95% (crude); (h) LiAlH , THF, 0 °C to reflux, 2.5 h, 44%; (i)
3
2
4
TsCl, pyridine, DCM, 0 °C to rt, 23 h, >95% (crude); (j) NaN , DMSO, rt, 19 h, >95%
3
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
t
(crude); (k) dipropargylamine, CuSO , sodium ascorbate, BuOH/H O (1:2, v/v), rt, 20-26 h,
4
2
1
0-60%.
The synthesis of azide 43 started from 2-(hydroxymethyl)phenol (39) which was transformed
7
1
to the corresponding nitrile 40. Attempts at the following reduction to furnish amine 41 in a
Pd/C catalyzed hydrogenation gave very poor yields. In another effort the reduction was carried
out with lithium aluminium hydride, however, the following aqueous workup was complicated
because the aminophenol could not be extracted into an organic phase. Ultimately, the
reduction was achieved using borane dimethyl sulfide complex to give the desired amine 41 in
a high crude yield. A diazotransfer reaction produced azide 42, followed by the usual
construction of the PA motif. For the synthesis of the final azide 47, dihydrocoumarin was
reduced with lithium aluminium hydride to give alcohol 45. Subsequent tosylation followed by
substitution with sodium azide gave azide 46 and installation of the PA motif provided azide
4
7. The synthesized azide monomers were finally linked in a double CuAAC with
dipropargylamine to give the four desired bis-triazole ligands 5a-d.72
For the generation of the urea-linked ligands 3a-d (Scheme 6) we decided to employ aza-Wittig
conditions which would allow us to use some of the already available azides (Scheme 5). Azide
47, required for urea 3b was used to generate amine 50 by catalytic hydrogenation. Only a poor
yield could be obtained, likely due to the high hydrophilicity of the product that complicated
the work-up and following chromatographic purification. The subsequent aza-Wittig reaction
7
3
of 47 and 50 afforded the desired urea, however the reaction led to the formation of a variety
ACS Paragon Plus Environment

Page 19 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of by-products. Presumably, not only the primary amine of 50, but also the other nucleophiles
present in the substrates, had reacted with the isocyanate intermediate.
To decrease the product’s hydrophilicity and prevent side reactions, we decided to change the
synthetic strategy and construct the urea linker prior to attachment of the PA motif for the
remaining urea linked ligands. Aminophenol 41 required for the synthesis of urea 3a was
already available (Scheme 5) and the corresponding azide 48 was prepared from 41 by a
diazotransfer reaction. The required amine 52 and azide 53 for the synthesis of urea 3c were
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
7
4
prepared from Baeyer-Villiger product 28. By the method of Ley and co-workers lactone 28
was treated with magnesium nitride in a sealed microwave vial to give to the corresponding
primary amide 51. The reaction provided an efficient route for the synthesis of unsubstituted
amide 51 from lactone 28, but was limited to approximately half gram scale because the
reaction is highly exothermic and pressure build up was observed on scale-up resulting in an
explosion. The initial yield of the reaction was unsatisfying, but was considerably improved by
work-up optimization using sulfate buffer for neutralization prior to extraction with ethyl
acetate. Amide 51 was reduced to amine 52 using borane dimethyl sulfide complex before
transformation to the azide 53 by a diazotransfer reaction. The azides 48 and 53 and amines 41
and 52, respectively, were thus reacted under aza-Wittig conditions to give the desired urea
bisphenols 49 and 54. Without the presence of the PA motif the product mixtures were
substantially easier to work-up and purify. For the longest urea analogue 3d, we decided to
deprotect the phenol functionality after generation of the urea to avoid handling of the
zwitterionic aminophenol. Thus, phenol derivative 56 that was obtained from a Wittig reaction
of 2-benzyloxybenzaldehyde with phthalimide protected bromoamine 55 was deprotected to
7
5
give amine 57 and converted to the corresponding azide 58. The aza-Wittig reaction gave
urea 59 that was subjected to a catalytic hydrogenation removing the benzyl groups and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 141
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
saturating the double bonds simultaneously leading to urea 60. The bis-phenols 49, 54 and 60
were carried on to the construction of the PA motif to provide final compounds 3a, c and d.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 21 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 6. Synthesis of the urea linked ligands 3a-d.a
OH
OH
OR
O
OR
2
2
2
2
a
b
NH2
N3
N
N
H
H
41
48
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
R=H 49
R=PA 3a
c
OPA
OPA
OH
OPA
O
OPA
3
3
b
3
3
d
NH2
N3
N
N
H
H
50
47
3
b
O
OH
a
O
3
4
NH2
f
e
NH2
O
51
52
28
OH
OR
O
OR
4
4
4
b
N3
N
H
N
H
5
3
R=H 54
R=PA 3c
c
OBn
O
O
3
i
Br
g,h
OBn
R
N
4
N
3
55
56
R = NH 57
O
O
2
a
R = N3 58
b
OR
O
OR
OBn
O
OBn
5
5
3
3
d
N
N
H
N
H
N
H
H
59
R=H 60
R=PA 3d
c
a
Reagents and conditions: (a) NaHCO or K CO , CuSO , 1H-imidazole-1-sulfonyl azide
3
2
3
4
hydrochloride, MeOH/H O (4:1, v/v), rt, 3-19 h, 36-85%; (b) nBu P, THF, 3Å molecular
2
3
sieves, rt, 1-2 h, then: amine (41, 50, 52 or 57) CO , rt, 20-23 h, 15-54%; (c) By the general
2
method for synthesis of the PA motif (Scheme 1) ligands 3a, 3c and 3d were isolated in 15%,
9
% and 45% yield, respectively; (d) Pd/C, H , THF, rt, 2-20 h, for 50: 29%, for 60: 90%; (e)
2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 141
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
Mg N , MeOH, 80 °C, 24 h, 61%; (f) BH ×SMe , THF, reflux, 18 h, 73%; (g) PPh , MeCN,
3
2
3
2
3
reflux, 18 h; (h) K CO , 18-crown-6, 2-benzyloxybenzaldehyde, toluene, 3Å molecular
2
3
sieves, reflux, 14 h, 33% (over 2 steps); (i) hydrazine hydrate, EtOH, reflux, 2 h, 80%.
Alkenes were a mixture of geometrical isomers. For clarity only the (E)- or (E,E)-isomer is
depicted.
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
Finally, a series of truncated analogues of urea 3b and diamide 4d were synthesized for probing
the binding mode in the extracellular vestibule. Truncated analogues of urea 3b were
synthesized using phenol 46 and different amines in the aza-Wittig reaction (Scheme 7). 3-
Phenylpropan-1-amine and propan-1-amine were commercially available, whereas amine 63
was prepared from carboxylic acid 61 by reduction with lithium aluminium hydride followed
by tosylation of the primary alcohol to give tosylate 62, substitution with sodium azide, and
finally catalytic hydrogenation of the resulting azide yielding the hydrochloride of 63. Phenols
6
4-66, obtained from the aza-Wittig reactions, were carried on to the final products 3b-I, 3b-
II and 3b-III as described above.
Scheme 7. Synthesis of truncated analogues of urea linked ligand 3b.a
ACS Paragon Plus Environment

Page 23 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
O
O
O
O
a,b
c,d
OH
OTs
62
NH2
61
63xHCl
OH
O
OPA
O
OH
3
3
R
3
3
R
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
e
f
N
N
N
N
N3
H
H
H
H
3
3
3
b-I R = 2-(OMe)Ph
b-II R = Ph
b-III R = H
4
6
64 R = 2-(OMe)Ph
5 R = Ph
6 R = H
6
6
a
Reagents and conditions: (a) LiAlH , THF, 0 °C to rt to reflux, 44 h, 84%; (b) TsCl,
4
pyridine, DCM, 0 °C to rt, 18 h, 63%; (c) NaN , DMSO, rt, 22 h, 81%; (d) Pd/C, H , THF, rt,
3
2
18 h, 39%; (e) nBu P, THF, 3Å molecular sieves, 0 °C to rt, 1-2 h, then: 3-phenylpropan-1-
3
amine or propan-1-amine or amine HCl salt (63) and DIPEA, CO , rt, 18-40 h, 33-36%; (f)
2
By the general method for synthesis of the PA motif (Scheme 1) ligands 3b-I, 3b-II and 3b-
III were isolated in 34%, 40% and 15% yield, respectively.
Truncated analogues of diamide 4d started either from lactone 28 or from carboxylic acids 67
and 68 (Scheme 8). For the synthesis of analogue 4d-I, 28 was ring-opened with N-Boc-
ethylenediamine and the resulting phenol was methylated using potassium carbonate and
methyl iodide yielding anisole 69. In the two other cases, 67 and 68 were coupled to N-Boc-
ethylenediamine using EDC hydrochloride and DIPEA. After deprotection of the amines 69-
7
1, reaction of these with lactone 28 provided phenols 72-74. Finally, the PA motif was
introduced on the phenols as described above to give truncated ligands 4d-I, 4d-II and 4d-III.
Scheme 8. Synthesis of truncated analogues of diamide linked ligand 4d.a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 141
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
O
O
O
6
9 R = 2-(OMe)Ph
a,b
BocHN
R 70 R = Ph
71 R = H
N
H
3
28
d,e
O
OH
O
c
H
3
7
7
2 R = 2-(OMe)Ph
3 R = Ph
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
R
3
N
R
3
HO
N
H
O
74 R = H
67 R = Ph
68 R = H
f
OPA
O
H
3
4
d-I R = 2-(OMe)Ph
4d-II R = Ph
N
R
3
N
H
O
4d-III R = H
a
Reagents and conditions: (a) N-Boc-ethylenediamine, EtOH, reflux, 20 h, 62%; (b) K CO ,
2
3
acetone, 30 min, then: 0 °C, MeI, rt, 20 h, 37%; (c) EDC hydrochloride, DIPEA, N-Boc-
ethylenediamine, DCM, rt, 20-22 h, 50%; (d) TFA, TIS, DCM, 0 °C to rt, 12 h, 48-95%; (e)
8, EtOH, 40 °C to reflux, 3-26 h, 40-71%; (f) By the general method for synthesis of the PA
2
motif (Scheme 1) ligands 4d-I, 4d-II and 4d-III were isolated in 44%, 45% and 20% yield,
respectively.
Pharmacology
Monomer potency and affinity
To determine the reference compound potencies, rac-, (S)- and (R)-ALP were tested for their
ability to block β AR and β AR agonist-induced cAMP production using membranes of
1
2
HEK293 cell lines stably overexpressing either of the two receptor subtypes. In addition a
7
6,77
whole cell assay of the β AR-overexpressing HEK293 cell line was used.
Isoproterenol
2
(ISO)-induced cAMP production was antagonized by rac-ALP at both receptors, and in
accordance with previous studies, (S)-ALP was substantially more potent than (R)-ALP at both
receptors (Table 2, Figure 3A-C).57
ACS Paragon Plus Environment

Page 25 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To assess ligand affinities, competition binding experiments were carried out using the same
membrane preparations. The ligands competed concentration-dependently with the radioligand
3
3
H-dihydroalprenolol ( H-DHA), and their affinities (K values) were estimated using the
i
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
Cheng-Prusoff equation with the experimentally determined K ’s of H-DHA at the two
D
receptor subtypes (0.85 ± 0.04 nM at the β AR (n = 3), and 0.39 ± 0.16 nM at the β AR (n =
1
2
4
)). In agreement with the literature, rac-ALP had an affinity of approximately 1 nM, similar
3
55
to that of the radioactive analogue H-DHA. As anticipated, (S)-ALP had a higher affinity
than (R)-ALP at both receptors (Table 3, Figure 3D-E).
Figure 3. Inhibition and displacement curves of rac-, (R)- and (S)-alprenolol (ALP) at the β AR
1
and β AR. Potencies were determined in functional cAMP assays using membranes (A-B) or
2
whole cells (C) in presence of isoproterenol (ISO), while affinities were estimated in a
3
competition binding assay using membranes and the radioligand H-DHA (D-E). The potency
of ALP is attributed to the (S)-isomer, which has a potency and affinity comparable to rac-
ALP. The (R)-isomer is less potent. Data represents mean ± SD from one representative out of
3-14 independent experiments carried out in duplicate.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 141
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
Bitopic ligand potency and affinity
Potencies and affinities of the bitopic ligands at the β AR and β AR were determined using the
1
2
same approach as described for the monomers (Table 2-3, Figure 4). The bitopic ligands
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
antagonized ISO-induced cAMP production of both receptors and displaced H-DHA similar
to the monomers.
Several ligands with equipotency to (S)-ALP were identified, e.g. 1c, 3c, 3d, 4d, 5c and 5d,
but none had substantially higher potencies compared to the monomer (Table 2, Figure 4). In
line with the functional data, the majority of the tested bitopic ligands were similar in affinity
to (S)-ALP, but none of them had significantly higher affinity than (S)-ALP. In general, there
was a good correlation between potency (Table 2) and affinity (Table 3) for the bitopic ligands
(Supporting Information Figure S1).
Typically, a bitopic binding mode affords a substantial increase in binding affinity when going
from monomeric to bivalent ligands. However, for the bitopic ligands described above, no
noteworthy increases in affinities or potencies were observed. It is conceivable that the high
potency and affinity of the (S)-ALP pharmacophore makes it difficult to significantly increase
the potency or affinity in bitopic ligands based on (S)-ALP. Since (R)-ALP has a much lower
potency than (S)-ALP we hypothesized that a bitopic ligand based on the (R)-ALP
stereochemistry could be a better model for demonstrating an increase in potency due to a
bitopic binding mode. To investigate this hypothesis, we selected one of the most potent ligands
4d as the model system. However, pharmacological evaluation of the bis-amide linked ligand
(R,R)-4d revealed a much lower potency (Table 2) than expected when compared to (R)-ALP.
This is a rather surprising finding since the potency of (S)-ALP was maintained for most (S,S)-
bitopic ligands including 4d which has the same linker type and length as (R,R)-4d. The result
indicates that contrary to the in silico results mentioned above, ligand chirality may in fact be
ACS Paragon Plus Environment

Page 27 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
an important determinant for in vitro binding to the MBS, and that (R)-ALP has less favorable
interactions with the extracellular vestibule compared to (S)-ALP.
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
Linker length and composition
Data from the cAMP TR-FRET assay (Figure 4) shows that the optimal linker length varies
considerably between the linker types and that it deviates from the outcome of our ligand
design. This could be a result of differing degrees of flexibility between linker types. To
achieve potencies similar to that of (S)-ALP, shorter linkers were required for the highly
flexible all-carbon and PEG linked ligands (1a-d, 2a-b) whereas longer linkers were needed
for the more rigid diamide and triazole linked ligands (4a-f, 5a-d). Based on these observations,
we conclude that an increase in linker rigidity requires concomitant increase in linker length
for potency to be retained. Furthermore, the potencies of the all-carbon linked ligands (1a-d)
did not vary substantially between the different linker lengths, whereas noticeable changes
were observed for the urea, diamide and triazole ligands (3a-d, 4a-f, 5a-d). Potentially, this is
the consequence of the unique physicochemical properties of the all-carbon linker compared
to the other linker types. The all-carbon linked ligands might easily adopt the preferred binding
conformation irrespective of linker length due to their high flexibility. Furthermore, it is
conceivable the hydrophobicity and smaller structure of the all-carbon linkers renders them
easier to accommodate in the channel connecting the OBS and the MBS and with less
detrimental receptor linker interactions compared to the other linker types studied herein. A
more detailed discussion of linker length and composition can be found in the Supporting
Information Section 3.1.2. Yet another explanation for the observed behavior of the ligands
could be related to the targeting of other ABSs beyond the MBSs reported by Dror et al.20
depending on the length of the linker.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 141
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 4. Correlation between potencies determined in the membrane-based cAMP assay and
linker lengths for the different bitopic ligand groups at the β AR (A) and β AR (B). Data
1
2
represents mean pIC ± SEM from 3-8 independent experiments carried out in duplicate. The
50
grey dotted line represents the mean pIC of (S)-ALP.
5
0
Receptor subtype selectivity
Because MBSs are located in less conserved receptor areas, we hypothesized that receptor
subtype selectivity could be achieved with bitopic ligands targeting a MBS compared to ligands
solely binding to the OBS. Based on selectivity plots for the most promising ligands
(
Supporting Information Section 3.1.3), the most interesting ligands in the context of β AR
2
selectivity were the ureas 3b-c, the diamides 4b-d, 4f and (R,R)-4d, and the triazole 5a. Of
these, 3b-c, 4c and 4f are particularly interesting since their potencies and affinities are similar
to that of (S)-ALP. There was no apparent increase in β AR selectivity for the all-carbon and
2
the PEG linked ligands. The improved selectivity could be due to targeting of less conserved
receptor areas of the β AR, e.g. the extracellular vestibule and the cleft accommodating the
2
linker. Interestingly, the tendency for increased potency with longer linkers that was observed
for the diamide, triazole and urea linked ligands was reversed with respect to β AR selectivity.
2
ACS Paragon Plus Environment

Page 29 of 141
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Hence, for these bitopic ligands the highest β AR selectivity was obtained with the shorter
2
linkers. A possible explanation for the opposite trends is differences in the extracellular
vestibules of the β AR and β AR. To investigate this hypothesis, we utilized the docking output
1
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
obtained during design of our ligands and superposed it with an inactive-state human β AR
1
7
8
homology model. The superposition showed that non-conserved residues make the passage
proposed to accommodate the linkers more narrow and longer in the β AR compared to the
1
β AR (Figure 5). Specifically, the residues Trp196 and Lys347 in the β AR are larger than the
2
1
corresponding Tyr174 and His296 in the β AR. Additionally, the Asp192-Lys305 salt bridge
2
observed in the β AR is absent in the β AR, and thus leads to the extracellular vestibule
2
1
extending further towards helix 2 in the β AR (not shown). In conclusion, the extracellular
1
vestibule appears to be shifted further away from the OBS in the β AR compared to the β AR,
1
2
which could be the reason for the observed β AR selectivity of ligands with shorter linkers.
2
ACS Paragon Plus Environment