Discovery of Small-Molecule Inhibitors of Receptor Activator of Nuclear Factor-κB Ligand with a Superior Therapeutic Index

Article abstract of DOI:10.1021/acs.jmedchem.0c01316

Receptor activator of nuclear factor-κB ligand (RANKL) constitutes the master mediator of osteoclastogenesis, while its pharmaceutical inhibition by a monoclonal antibody has been approved for the treatment of postmenopausal osteoporosis. To date, the pur

Full text of DOI:10.1021/acs.jmedchem.0c01316

Subscriber access provided by Karolinska Institutet, University Library
Article
Discovery of Small-Molecule Inhibitors of Receptor Activator of
Nuclear Factor-#B Ligand (RANKL) with Superior Therapeutic Index
Vagelis Rinotas, Athanasios Papakyriakou, Foteini Violitzi, Christos Papaneophytou,
Maria-Dimitra Ouzouni, Polyxeni Alexiou, Alexandros T. Strongilos, Elias
Andreas Couladouros, George A Kontopidis, Elias Eliopoulos, and Eleni Douni
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c01316 • Publication Date (Web): 21 Sep 2020
Downloaded from pubs.acs.org on September 21, 2020
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 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of Small-Molecule Inhibitors of Receptor Activator of Nuclear Factor-
κB Ligand (RANKL) with Superior Therapeutic Index
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
Vagelis Rinotas,1,2 Athanasios Papakyriakou, Foteini Violitzi, Christos
3
2
Papaneophytou,4,5 Maria-Dimitra Ouzouni, Polyxeni Alexiou, Alexandros Strongilos,
6
6
7
6
4
1
1,2,*
Elias Couladouros, George Kontopidis, Elias Eliopoulos, Eleni Douni
1Laboratory of Genetics, Department of Biotechnology, Agricultural University of Athens,
5 Iera Odos, 11855, Athens, Greece
2Institute for Bioinnovation, Biomedical Sciences Research Center "Alexander Fleming",
4 Fleming Street, 16672, Vari, Greece
3Institute of Biosciences and Applications, National Centre for Scientific Research
Demokritos", 15341, Agia Paraskevi, Athens, Greece
4Department of Biochemistry, Veterinary School, University of Thessaly, 224 Trikalon,
3131, Karditsa, Greece
7
3
"
4
5Department of Life and Health Sciences, School of Sciences and Engineering, University
of Nicosia, 46 Makedonitissas Avenue, 2417, Nicosia, Cyprus
6Laboratory of General Chemistry, Department of Food Science and Human Nutrition,
Agricultural University of Athens, 75 Iera Odos, 11855, Athens, Greece
7proACTINA SA, 20 Delfon Street, 15125, Marousi-Athens, Greece
*Corresponding author: Eleni Douni, Laboratory of Genetics, Department of
Biotechnology, Agricultural University of Athens, 75 Iera Odos, 11855, Athens, Greece;
douni@aua.gr/ Institute for Bioinnovation, Biomedical Sciences Research Center
"
Alexander Fleming", 34 Fleming Street, 16672, Vari, Greece; douni@fleming.gr
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
Receptor activator of nuclear factor-κB ligand (RANKL) constitutes the master mediator
of osteoclastogenesis, while its pharmaceutical inhibition by a monoclonal antibody has
been approved for the treatment of postmenopausal osteoporosis. To date, the pursuit of
pharmacologically more favorable approaches using low-molecular-weight inhibitors has
been hampered by low specificity and high toxicity issues. This study aimed to discover
small-molecule inhibitors targeting RANKL trimer formation. Through a systematic
screening of 39 analogs of SPD-304, a dual inhibitor of TNF and RANKL trimerization,
we identified 4 compounds (1b, 3b, 4a, and 4c) that selectively inhibited RANKL-induced
osteoclastogenesis in a dose-dependent manner, without affecting TNF activity or
osteoblast differentiation. Based on structure-activity observations extracted from the
most potent and less toxic inhibitors of RANKL-induced osteoclastogenesis, we
synthesized a focused set of compounds that revealed 3 potent inhibitors (19a, 19b and
20a) with remarkably low cell-toxicity and improved therapeutic indexes as shown by the
LC to IC ratio. These RANKL-selective inhibitors are an excellent starting point for the
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
50
50
development of small-molecule therapeutics against osteolytic diseases.
2
ACS Paragon Plus Environment

Page 3 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Bone integrity is maintained through a dynamic process, known as bone remodeling,
resulting from a continuous balanced interplay between bone resorption caused by
osteoclasts and bone formation employed by osteoblasts. An increase in bone
remodeling activity occurs in osteoporosis, the most common bone disease, that is
characterized by low bone density, reduced bone quality and increased risk of fractures,
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
causing serious health consequences. Receptor activator of nuclear factor-κB ligand
(RANKL), a tumor necrosis factor (TNF) superfamily member, constitutes the master
mediator of bone resorption as it promotes osteoclast differentiation, activity and
survival.2–4 It is a type II transmembrane protein that contains a conserved extracellular
TNF-like core domain forming an antiparallel β-sheet that is predicted to assemble into
homotrimers required for binding to its cognate receptor RANK.5,6 Trimeric soluble
RANKL is also produced either by proteolytic cleavage of the transmembrane form or by
alternative splicing.7,8 Binding of RANKL to RANK initiates downstream signaling
cascades, including nuclear factor κB (ΝF-κB), protein kinases, nuclear factor of activated
9
T-cell c1 (NFATc1) and c-Fos, that activate osteoclastogenic genes. RANKL is also
implicated in diverse in vivo biological processes such as immune regulation, mammary
gland development, hormone-induced breast cancer induction, and body
thermoregulation.10–13 Genetic ablations of RANKL and RANK result in severe autosomal
recessive osteopetrosis, a disease caused by osteoclast deficit, demonstrating that the
RANKL/RANK system is indispensable for osteoclastogenesis and bone resorption.14–18
RANKL is physiologically inhibited by the natural decoy receptor osteoprotegerin (OPG)
that prevents its binding to the RANK receptor and thus impairs the process of
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 54
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
osteoclastogenesis.19 An imbalance at the RANKL/OPG ratio caused by abundant
RANKL levels is believed to be a major determinant in the development of bone loss
diseases, including postmenopausal osteoporosis, bone metastasis, and multiple
myeloma.20,21 Our research group has previously demonstrated that overexpression of
human RANKL in transgenic mice leads to progressive bone resorption and osteoporosis
in both sexes, establishing unique genetic models of osteoporosis for understanding the
underlying pathogenic mechanisms and for preclinical evaluation of novel therapeutics.22
During the past decades, many anti-osteoporotic therapeutic interventions have been
established targeting mainly inhibition of bone resorption and to a lesser extent
stimulation of bone formation.23,24 However, many drawbacks including poor clinical
responses, undesired side-effects and high market cost have stimulated the research for
new compounds with improved specificity and minimal adverse effects.25 The discovery
of the RANKL/RANK/OPG system offered the possibility of developing novel therapeutics
that target specifically the main bone-resorbing factor, RANKL. Denosumab, a human
monoclonal antibody to RANKL that specifically blocks its binding to RANK constitutes
the first RANKL inhibitor approved by FDA for the treatment of postmenopausal
osteoporosis.26–28 However, functional limitations of therapeutic antibodies have come to
light, such as inadequate pharmacokinetics, tissue accessibility and immunogenicity,
which point to areas where additional research is needed.29
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
Alternative therapeutic approaches using peptide-mimics,30,31 natural product inhibitors32
or small molecules that inhibit RANKL function have been pursued in parallel.33–37
Inhibition of RANKL using small-molecules has gained considerable attention due to the
advantages of their low-cost production, desirable pharmacokinetic properties, and the
4
ACS Paragon Plus Environment

Page 5 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
potential for oral administration. On these grounds, SPD-304 (Scheme 1) has been
previously identified as a potent TNF inhibitor promoting dissociation of trimer
assembly , which also inhibits RANKL-induced osteoclastogenesis18,36, whereas its
preclinical use has been hampered by its high toxicity.36,38 SPD-304 contains a potentially
toxic 3-alkyindole moiety that can be activated by cytochrome P450 enzymes to produce
38
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
39
α,β-unsaturated iminium species. Indeed, toxicological studies have shown that SPD-
304 can be dehydrogenated by CYP3A4 to the electrophilic 3-methyleneindolenine
intermediate, in addition to other hydroxylation, N-dealkylation and epoxidation
40
metabolites. With the aim to develop analogs of SPD-304 with improved toxicity profiles,
we have previously employed structure-based design and synthesis of the compounds
shown in Scheme 1.41,42 Their design strategy comprised of: (i) substitution of the
trifluoromethylphenyl moiety of SPD-304 with electron-withdrawing groups, such as the
phenylsulfonyl or the 3-nitrophenyl groups; (ii) conversion of the diamine linker into the
corresponding diamide, or cyclization to 1-4-piperazinylone; and (iii) elimination of the
methyl groups of the chromone moiety, or its substitution by other aromatic rings. Their
screening against TNF revealed several equally potent inhibitors of TNF with respect to
42
SPD-304, while incorporation of electron-withdrawing substituents at the indole moiety
in conjunction with elimination of the 6΄-methyl group of the 4-chromone moiety, led to
the identification of compound 1c as a significantly less toxic inhibitor of TNF/TNF-R1.41
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 54
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
Scheme 1. Structures of the compounds designed as TNF inhibitors and used here in a
systematic screening against RANKL-induced osteoclastogenesis.
6
ACS Paragon Plus Environment

Page 7 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Herein, we report the identification of 7 novel RANKL inhibitors through RANKL-induced
osteoclastogenesis assays, further supported by binding affinity and cytotoxicity assays.
The efficacy of the compounds and the IC50 were estimated through a quantitative
RANKL-induced osteoclastogenesis assay, while the MTT assay was used to measure
cell viability and the LC on primary osteoclast precursors stimulated with compounds in
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
50
a range of concentrations exceeding those tested for their efficacy in the absence of
43
RANKL. By measuring the LC to IC ratio (therapeutic index) , we evaluated whether
50
50
the anti-osteoclastogenic effects of the compounds were attributed either to cellular
toxicity unrelated to RANKL (ratio close to 1) or specific inhibition of the RANKL function
(ratio >>1). Based on the systematic screening of 39 SPD-304 analogs previously
screened for TNF inhibition41,42 (Scheme 1, Table 1), we identified 4 compounds (1b, 3b,
4a, and 4c) that displayed total inhibition of RANKL-induced osteoclastogenesis with low
toxicity. We also report the development of 3 potent inhibitors (19a, 19b, 20a) of human
RANKL-induced osteoclastogenesis with minimal toxicity in primary cell cultures of
osteoclast precursors derived from bone marrow cells. We also suggest their potential
binding mode at the interface of human RANKL dimer through ensemble docking
calculations using representative structures from a series of molecular dynamics
simulations.
Table 1. List of 39 compounds evaluated as RANKL inhibitors. With bold are highlighted the
a
4
most active compounds. Quantification of binding affinity to TNF by fluorescence assay,
b
c
Mean± SD (n=3 independents experiments); p<0.01. Quantification of inhibition of TNF-induced
d
death in L929 cells. Quantification of binding affinity to RANKL by fluorescence assay.
eQuantification of compounds toxicity in BMΜ cells. n.d.= not determined due to no inhibition of
osteoclastogenesis, inactive= no binding affinity, toxic= cell toxicity is observed at 5μM, * Values
taken from 41,42
.
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 54
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
Inhibition of
RANKL
binding
TNF binding
Inhibition of
osteoclastogenesis
at 5 μM
BMΜ
viability
(LC50, μM)
Compou
nd ID
TNF
activity*
a
e
Affinity*
c
d
affinity
(K , μΜ)
d
b
(K
d
, μΜ)
b
b
(IC50, μΜ)
SPD-304
1a
5.36 ± 0.21
8.50±0.56
7.60±0.62
5.10±0.23
21.00±0.78
10.12 ± 0.67
4.78 ± 0.45
5.12 ± 0.61
12.22 ± 0.85
25.36 ± 1.25
35.23 ± 1.92
14.12 ± 0.69
7.00 ± 0.44
12.81 ± 0.78
3.16 ± 0.21
16.82 ± 0.51
0.95 ± 0.06
5.45 ± 0.62
>35
5.00 ± 0.15
7.50±0.15
50.00 ± 0.11
10.00±0.15
10.00±0.15
10.00 ± 1.90
30.00 ± 3.10
inactive
13.80±0.70
8.71±0.44
5.24±0.12
4.90±0.31
6.62±0.27
>10
toxic
toxic
total
total
<20
n.d
1
b
>20
<20
n.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
1
c
1
d
toxic
total
2a
<20
n.d
2
b
>11
none
total
3
a
4.60±0.23
2.18±0.11
>25
<20
>20
n.d
3
b
inactive
total
none
none
none
total
partial
total
total
3c
inactive
3
d
inactive
17.61±0.78
19.06±1.47
4.60±0.34
11.04±0.77
4.60±0.38
inactive
n.d
3
e
10.00 ± 1.30
>60
n.d
4
a
>20
> 20
>20
<20
n.d
4b
15.00 ± 1.70
20.00 ± 1.60
20.00 ± 2.10
25.00 ± 1.80
20.00 ± 1.90
inactive
4
c
4
d
4
e
22.30±1.24
inactive
none
none
none
none
none
partial
none
total
4f
n.d
5
a
3.55±0.22
inactive
n.d
5
b
inactive
inactive
n.d
5
c
17.27 ± 1.26
inactive
inactive
>15
n.d
5d
inactive
12.50±0.56
inactive
>20
n.d
5
e
inactive
inactive
6
a
13.28 ± 1.15
inactive
inactive
2.82±0.31
inactive
<20
n.d
6
b
inactive
none
partial
none
none
partial
none
partial
total
6c
15.61 ± 1.46
inactive
40.00 ± 3.80
inactive
9.31±0.25
inactive
>20
n.d
7
a
7
b
15.17 ± 1.34
>35
inactive
inactive
n.d
7
c
inactive
5.34±0.41
inactive
<20
n.d
7d
inactive
inactive
8
a
1.61 ± 0.15
inactive
10.00 ± 0.90
inactive
8.49±0.34
inactive
>20
<20
n.d
8
b
8
c
7.82 ± 1.06
inactive
20.00 ± 1.80
inactive
inactive
none
partial
partial
none
none
none
partial
none
8d
inactive
>20
>20
n.d
8
e
2.11 ± 0.16
5.20 ± 0.65
18.70 ± 1.90
inactive
15.00 ± 1.10
20.00 ± 1.90
inactive
18.64±0.63
inactive
9
a
9
b
18.35±0.95
inactive
n.d
9c
inactive
n.d
1
0a
17.85 ± 1.28
10.74 ± 1.32
inactive
12.80±0.86
15.80±0.64
>20
n.d
1
0b
inactive
8
ACS Paragon Plus Environment

Page 9 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
RESULTS AND DISCUSSION
SPD-304 is a dual inhibitor for TNF and RANKL
SPD-304 (Scheme 1) has been discovered as a potent inhibitor of TNF, which promotes
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
8
the dissociation of active TNF trimers. A crystallographic structure of TNF in complex
with SPD-304 revealed the inhibitor bound at the interface of dimeric TNF intermediate,
thus blocking the formation of trimers. To examine whether SPD-304 could also serve as
an inhibitor of RANKL we performed a series of biochemical and biological assays. Using
an in vitro fluorescence binding assay we measured the affinity of SPD-304 for human
RANKL, which revealed a dissociation constant (K = 13.8 μΜ) that is higher than the
d
41
corresponding values reported for TNF (K = 5.4 μΜ). By employing a quantitative
d
osteoclastogenesis assay, we found that SPD-304 is a potent inhibitor of human RANKL-
induced osteoclastogenesis in a dose-dependent manner (IC =1.0 μΜ, Figure 1A,B).
50
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 54
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. Functional characterization of SPD-304 as inhibitor of RANKL and
molecular modeling of SPD-304 interaction with RANKL. (A) Osteoclastogenesis
cultures based on BM cells treated with various concentrations of SPD-304 in the presence
of RANKL (40 ng/mL) and M-CSF (25 ng/mL) for 5 days upon staining with TRAP. (B) IC50
calculation of SPD-304 effect on osteoclastogenesis was based on TRAP activity measured
at day 4. (C) LC50 calculation of SPD304 effect (0-10 μΜ) on BM cell viability by MTT assay.
10
ACS Paragon Plus Environment

Page 11 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
D) Recombinant soluble human RANKL and SPD-304 were preincubated and crosslinked
with DSS. The crosslinked complex was separated in 12% SDS-PAGE and immunoblotted
using an anti-RANKL polyclonal antibody. (E) Representation of the X-ray crystal structure
of TNF dimer in complex with the small-molecule inhibitor SPD-304 (PDB ID: 2AZ5)38.
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
The two TNF subunits are shown in green and cyan for chains A and B, respectively,
while the inhibitor is color-coded using yellow C, red O, blue N and cyan F atoms. Inset
is a surface representation of the surrounding residues that form the inhibitor-binding
pocket, with an asterisk indicating residues that differ in human RANKL. (F) Molecular
model of human RANKL dimer based on the TNF dimer and the X-ray structure of RANKL
complex with the N-terminal fragment of its decoy receptor osteoprotegerin (PDB ID:
3
URF).19 Inset is a close-up view of the RANKL interface, in the same orientation as
shown for TNF, illustrating a favorable bound pose of the dual inhibitor SPD-304.
However, SPD-304 proved to be of equally high toxicity to osteoclast precursors derived
from bone marrow cells (LC = 3.2 ± 0.1 μΜ) (Figure 1C, Table 2). To understand the
50
molecular basis of human RANKL inhibition by SPD-304, we investigated its effect at the
level of RANKL trimerization using chemical cross-linking experiments. In this
experimental setup, pre-incubated soluble human RANKL with SPD-304 at equimolar, or
excess molar ratio, were cross-linked and analyzed in SDS polyacrylamide gel
electrophoresis (PAGE) for the detection of RANKL trimers, dimers, and monomers. A
dramatic increase of RANKL monomers was detected in the presence of SPD-304
(
Figure 1D), indicating that the inhibitor promotes human RANKL subunit disassembly.
Taken together, we demonstrate that SPD-304 is a dual inhibitor of human TNF and
RANKL, which mediates inactivation of the cytokines through dissociation of the
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 54
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
43
biologically active trimers, but with a low therapeutic index (LC / IC = 3.2), indicating
50
50
high toxicity (Table 2).
Table 2. Evaluation of potency and toxicity of most effective RANKL inhibitors.
a
b
Quantification of inhibition of RANKL-induced TRAP Activity in BM cells, Quantification of
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
c
compounds toxicity in BMΜ cells, Values are mean ± standard deviation from 3 independents
d
experiments, Calculated as the ratio of LC to IC .
50
50
a
Toxicityb
LC50 (µM)
Compound TRAP Activity
Therapeutic
c
c
d
ID
IC50 (µM)
Index
SPD-304
1.00 ± 0.20
3.20 ± 0.10
58.50 ± 5.00
44.25 ± 2.64
62.33 ± 2.59
32.20 ± 3.55
>300
3.2
23.4
27.8
21.8
6.3
1
3
4
4
b
b
a
c
2.50 ± 0.67
1.88 ± 0.25
2.86 ± 1.16
5.07 ± 0.87
2.70 ± 0.67
2.00 ± 0.19
1.90 ± 0.26
1.40 ± 0.67
5.10 ± 1.10
6.40 ± 0.88
19a
19b
>112
>250
9.5
>500
1
2
2
2
9c
0a
0b
0c
18.30 ± 1.10
123.00 ± 4.00
52.60 ± 7.20
118.00 ± 26.00
84.8
10.4
18.5
Structural model of RANKL bound to SPD-304
SPD-304 has been shown to bind TNF dimer mostly through a shape-driven interaction
within a hydrophobic pocket comprising 6 aromatic tyrosine residues (Tyr59, Tyr119, and
Tyr152 from each monomer, Figure 1E). To investigate whether SPD-304 could operate
via a similar mechanism as inhibitor of RANKL trimerization, we prepared a model of
human RANKL dimer based on the TNF dimer as template and the crystallographic
12
ACS Paragon Plus Environment

Page 13 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
structure of human RANKL complex with the N-terminal fragment of its decoy receptor
OPG.19 Examination of the model revealed a significant change in the shape of the
corresponding pocket at the interface of human RANKL dimer, mainly due to the change
of Tyr119 and Leu57 in TNF to Asn276 and Tyr215 in RANKL, respectively (Figure 1F).
It is thus reasonable to assume that SPD-304 does not bind RANKL in such a compact
conformation as displayed in the co-crystal structure in complex with TNF dimer. Indeed,
molecular docking calculations of SPD-304 using the energy minimized model of RANKL
dimer revealed a large number of diverse binding modes (clusters of conformations),
displaying comparable binding affinities with respect to the standard error of the method
(see Computational Methods for more detail). This challenging task was also the case
when re-docking of SPD-304 to the X-ray structure of the TNF dimer was performed with
the widely-used AutoDock software.44 In particular, the top-ranked and the most
populated clusters of conformations failed to identify the crystallographic bound pose of
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
SPD-304 within a root-mean-square deviation (RMSD ) of 2.5 Å (Supporting Information
ref
Table S1 and Figure S1), indicating the large conformational space available on the dimer
interface.
To tackle the even more challenging case of human RANKL dimer, for which there is no
structural information of small-molecule inhibitors to date, we performed a combination
of molecular dynamics simulations (MDs) with docking calculations, a method termed
ensemble docking.45,46 Briefly, a representative set of 5 conformations of human RANKL
dimer were extracted from atomistic MDs in explicit solvent using two clustering criteria
(Supporting Information Figure S2); one by considering only the interface residues and
the second using all residues of the dimer. A characteristic analysis of the MDs revealed
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 54
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
a notable stability of the RANKL dimer model at the 100-ns timescale of the simulations
(Supporting Information Figures S3–S4). Docking of SPD-304 at the initial, energy-
minimized structure of human RANKL dimer (designated as RANKL-min) and the 5
representative conformations from the MDs (Supporting Information Figure S2) displayed
a significant number of low populated clusters of conformations (Supporting Information
Table S2). However, including the dynamic structural information using multiple
conformations of RANKL allowed for a more systematic selection of the most probable
ligand binding mode. Specifically, the highest-populated clusters among the top-10
ranked solutions revealed very similar bound poses of SPD-304 in 5 out of the 6
conformations of RANKL employed (Supporting Information Figure S5). We selected the
most populated cluster rather than the highest-ranked conformation for each target
considering that this choice has been shown to increase the predictive power of docking
in several cases.47,48 The representative model of RANKL-min in complex with SPD-304
displays the ligand in a significantly different conformation than in the X-ray structure of
the TNF dimer (Figure 1E,F), which is accommodated at two shallow pockets separated
by Asn276 and flagged by Tyr215 and Tyr273 residues of both RANKL monomers. In this
pose, the ligand could form 2 hydrogen bonds between the two chromenone oxygens of
SPD-304 as acceptors and the phenolic group of Tyr273 and the main chain amide of
Asn276 as donors. Although lack of an experimentally determined structure renders this
analysis highly speculative, taken together with our results from the binding assay and
the RANKL trimerization cross-linking experiment, suggest a putative mechanism of
action similar to that proposed for TNF.38 Thus, SPD-304 could potentially bind to the
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
14
ACS Paragon Plus Environment

Page 15 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
active RANKL trimer in a more extended conformation, forming an intermediate complex
that mediates the subunit dissociation process.
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
Discovery of 4 selective inhibitors of human RANKL
The finding that SPD-304 inhibits human RANKL and TNF effectively, prompted us to
investigate a series of SPD-304 analogs that have been developed previously as TNF
inhibitors (Scheme 1).41,42 Their systematic screening comprised measurement of their
42
binding affinity for human RANKL in comparison with the reported affinity for TNF, and
evaluation of their efficacy to inhibit RANKL-mediated osteoclastogenesis at a single
concentration of 5 μM (Table 1). Compounds that displayed partial or total inhibition were
further evaluated through MTT viability assay. Analysis of our screening results revealed
10 compounds that completely prevented osteoclast formation and 8 compounds that
partially inhibited RANKL-induced osteoclastogenesis. 19 compounds had no effect on
RANKL-induced osteoclastogenesis, while 2 compounds displayed similar toxicity to
SPD-304 (Table 1). From the 10 most potent RANKL inhibitors (total inhibition of
osteoclastogenesis at 5 μM), 4 compounds displayed a relatively low toxicity of LC >20
50
μM (1b, 3b, 4a and 4c, Scheme 1). Interestingly, these 4 compounds had not been
identified as potent inhibitors of TNF activity (IC > 20 μM, Table 1), and thus selected
50
for further investigation. Their assessment comprised of osteoclastogenesis assays
quantitative TRAP activity, phalloidin staining), toxicity assays, osteoblast differentiation
(
(alkaline phosphatase activity), RANKL signaling (NFATc1, c-Fos) and RANKL
trimerization.
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 54
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
Compound 1b was developed by substitution of the m-trifluoromethylphenyl ring of SPD-
304 by a phenylsulfonyl group, in addition to removal of both methyl groups at the 4-
chromone moiety (Scheme 1). The binding affinity of 1b (K = 5.24 ± 0.12 μΜ) displayed
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
a 2.7-fold increase compared to SPD-304 (Table 1). Τhe effect of 1b on RANKL-induced
osteoclastogenesis demonstrated a dramatic inhibition of osteoclast formation at 5 and
10 μΜ, as shown by the absence of multinuclear TRAP+ cells (Figure 2A), and the
decrease of phalloidin-labeled actin ring formation (Figure 2B-C). Further evaluation
revealed a RANKL inhibitory potency of IC = 2.5 μM, with a significantly lower toxicity
50
than SPD-304 (LC = 58.5 ± 5.0 μΜ for 1b vs 3.2 ± 0.1 μΜ for SPD-304) (Figure 2D-E).
50
Cross-linking assays showed that preincubation of 1b with human RANKL protein
promoted a substantial release of monomers in a dose-dependent manner (Figure 2F),
suggesting an interference with the RANKL trimerization process. We further investigated
whether 1b mediated undesirable inhibitory effects on osteoblasts. To this scope, we
examined the effects of 1b on the differentiation of pre-osteoblastic MC3T3-E1 cells
through quantification of alkaline phosphatase (ALP) activity, a widely recognized marker
for osteoblastic differentiation and activity. MC3T3-E1 cells were treated with 1b at
indicated concentrations (Figure 2G) for 14 days and ALP activity was measured at the
endpoint. Our analysis showed that 1b at a concentration range of 2–10 μM did not
interfere with osteoblast differentiation and activity (Figure 2G). Therefore, 1b emerged
as a potent RANKL inhibitor with an increased therapeutic index compared to SPD-304
(23.4 vs 3.2, Table 2), indicating low toxicity.
16
ACS Paragon Plus Environment

Page 17 of 54
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. Characterization of the small-molecule 1b as RANKL inhibitor. (A)
Osteoclastogenesis cultures based on BM cells treated with compound 1b at 5 and 10
μM in the presence of RANKL (40 ng/mL) and M-CSF (25 ng/mL) for 5 days upon staining
with TRAP. (B) Actin ring organization in osteoclastogenesis cultures upon exposure to
compound 1b at 5 and 10 μM as shown by phalloidin staining (green) and DAPI to
visualize nuclei (blue). (C) Quantification of intact F-actin rings per mm2 on
osteoclastogenesis cultures treated with compound 1b at 5 and 10 μM. (D) IC50
calculation for compound 1b effect on RANKL-induced osteoclastogenesis based on
TRAP activity measured at day 4. (E) LC calculation of compound 1b effect (0–100 μΜ)
50
on BMM cell viability by MTT assay. (F) Recombinant soluble RANKL and compound 1b
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 54
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
were preincubated either at equal molar ratio or excess and then crosslinked with DSS.
The crosslinked complex was separated in 12% SDS-PAGE and immunoblotted using
an anti-RANKL polyclonal antibody. (G) Effect of compound 1b (2, 5 and 10 μM) on the
differentiation of pre-osteoblastic MC3T3-E1 cells as measured by alkaline phosphatase
activity (ALP). Undifferentiated MC3T3-E1 cells are marked as c. All experimental assays
were repeated at least three times. Results are presented as means ± SD (n>3). *p>0.05,
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
**p>0.01.
Compound 3b comprises a diamide linker instead of the diamine moiety in SPD-304, and
a 3-nitrophenyl instead of the trifluoromethylphenyl group (Scheme 1). 3b displayed the
highest binding affinity (K = 2.18 ± 0.11 μΜ) among the compounds tested (Table 1), and
d
inhibited RANKL-induced osteoclastogenesis in a dose-dependent manner (Figure 3A-
D), as detected by staining of cultures with TRAP and phalloidin, as well as by quantitative
TRAP activity (IC = 1.88 ± 0.25 μM). Regarding cell toxicity, 3b displayed a superior
50
profile compared to SPD-304 (LC = 44.25 ± 2.64 μM) (Figure 3E). Moreover, 3b at a
50
concentration range of 2–10 μM had no effect on MC3T3-E1 differentiation, as measured
by ALP activity (Figure 3F). Cross-linking biochemical assays showed that 3b promoted
a dramatic release of human RANKL monomers, even at 1:3 molar ratio (Figure 3G), an
effect that is in agreement with the high binding affinity of 3b for RANKL. We then
investigated whether 3b affects the signaling pathways activated downstream of RANKL
by examining the levels of the transcription factors NFATc1 and c-Fos in RAW264.7 cells
through western blot.49–51 Our results indicate that 3b diminished the NFATc1 and c-Fos
protein levels in RAW264.7 cells treated with RANKL (Figure 3H). Taken together, these
data indicate that 3b promotes inhibition of RANKL-induced osteoclastogenesis and as
18
ACS Paragon Plus Environment

Page 19 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a consequence of RANKL-induced signaling pathways involving NFATc1 and c-Fos,
while its toxicity is much lower than SPD-304 based on the therapeutic index (27.8 vs
3.2, Table 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
Figure 3. Characterization of the small-molecule 3b as RANKL inhibitor. (A)
Osteoclastogenesis cultures treated with compound 3b at 5 and 10 μM in the presence
of RANKL (40 ng/mL) and M-CSF (25 ng/mL) for 5 days upon staining with TRAP. (B)
Actin ring organization in osteoclastogenesis cultures upon exposure to compound 3b at
5
and 10 μM as shown by phalloidin staining and DAPI. (C) Quantification of intact F-
2
actin rings per mm on osteoclastogenesis cultures treated with compound 3b at 5 and
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 54
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
0 μM. (D) IC50 calculation for compound 3b effect on RANKL-induced
osteoclastogenesis based on TRAP activity measured at day 4. (E) LC50 calculation of
compound 3b effect (0–100 μΜ) on BMM cell viability by MTT assay. (F) Effect of
compound 3b (2, 5 and 10 μM) on the differentiation of pre-osteoblastic MC3T3-E1 cells
as measured by ALP activity. (G) Recombinant soluble RANKL and compound 3b were
preincubated either at equal molar ratio or excess and then crosslinked with DSS. The
crosslinked complex was separated in 12% SDS-PAGE and immunoblotted using an anti-
RANKL polyclonal antibody. (H) Western blot showing the effect of preincubated
compound 3b with RANKL at the indicated concentrations on the induction of NFATc1
and c-Fos in RAW264.7 cells. Antibody specific for β-actin was used for normalization.
Numbers below NFATc1 and c-Fos blots indicate relative expression compared to
positive control. All experimental assays were repeated at least three times. Results are
presented as means ± SD (n>3). *p>0.05, **p>0.01.
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 4a is the closest analog of 1b, but with diamide moiety as in 3b (Scheme 1).
Biochemical evaluation revealed that 4a displayed a 3-fold higher binding affinity than
SPD-304 (Table 1), and effectively inhibited osteoclastogenesis in a dose-depended
manner (IC = 2.86 ± 1.16 μM, Supporting Information Figure S6A-D). More importantly,
50
4a was found to be 20-fold less toxic than SPD-304 (LC = 62.33 ± 2.59 μM), without
50
affecting differentiation of the pre-osteoblastic cell line MC3T3-E1 (Supporting
Information Figure S6E,F).
The fourth potent inhibitor identified, 4c, is a 3-nitrophenylsulfonyl derivative of 4a
(Scheme 1). Similarly, 4c showed an improved binding affinity with respect to SPD-304
(K = 4.60 ± 0.38 μΜ), displayed inhibition of osteoclast formation in a dose-dependent
d
manner (IC = 5.07 ± 0.87 μM, Supporting Information Figure S7A-D), with a 10-fold
50
lower cell-toxicity compared to SPD-304 (LC = 32.20 ± 3.55 μM, Supporting Information
50
20
ACS Paragon Plus Environment

Page 21 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure S7E). 4c also effectively reduced the induction of the transcription factors NFATc1
and c-Fos in RAW264.7 cells, while not affecting differentiation of pre-osteoblastic cells
(Supporting Information Figures S7F,G), indicating a specific effect of 4c in RANKL-
induced osteoclastogenesis. Taken together, compounds 1b, 3b, 4a and 4c inhibit
RANKL-induced osteoclastogenesis and display an increase of the therapeutic index, 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
8
times compared to SPD-304 (Table 2).
Structure-activity relationships of the identified RANKL inhibitors
Considering the structural characteristic of the 4 most potent and less toxic inhibitors
identified, we can make the following observations. With respect to the parent compound
(SPD-304) all 4 potent inhibitors of RANKL comprise the 4-chromone moiety but without
its two methyl substituents. In 3 of these (1b, 4a and 4c) the trifluoromethylphenyl group
is replaced by a phenylsulfonyl moiety, whereas 3b contains the 3-nitrophenyl moiety.
Interestingly, although the nitro group is considered as a toxicophore, albeit its wide use
52
in therapeutics, 2 out of the 4 RANKL inhibitors (3b and 4c) contain the 3-nitrophenyl
moiety and are >10-fold less toxic than SPD-304. Lastly, except for 1b that comprise the
diamine linker as in SPD-304, this moiety is converted to the more rigid diamide in the
other 3 inhibitors. To gain further insight into their structure-activity relationships, we
predicted their binding mode to human RANKL dimer following the same methodology as
described above (Figure 4). As in the case of SPD-304, the docking results were
evaluated by means of the most populated clusters of conformations in the 6
representative structures of the RANKL dimer (Supporting Information Tables S3–S6 and
Figures S8–S10). Specifically, the most favorable bound pose of 1a was selected in
conjunction with those displayed by 4a, similarly to the selection of the most favorable
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 54
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
conformations for 3b and 4c. For the first couple of inhibitors (Figure 4A,C), the presence
of the phenylsulfonyl-1H-indole moiety favors its interaction at an aromatic-rich pocket
comprising of Tyr215 (both chains A and B), Tyr217(B) and Tyr307(B). In this bound
pose, their sulfonyl group can accept 2 hydrogen bonds from the side-chain amide of
Asn276(B) and the phenolic group of Tyr215(B). Although conversion of the diamine of
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
1b to diamide in 4a decrease its conformational freedom, the linker with the 4-chromone
moiety is accommodated accordingly to accept a hydrogen bond from the main chain
amide of Asn276(A) and probably interact with Gly278(A) as well (Figure 4A,C). For
inhibitors 3b and 4c, it is possible that the presence of the nitro group can mediate their
binding at an inverted orientation with respect to 1b and 4a (Figure 4B,D). This effect is
probably due to the favorable hydrogen-bonding interactions with the two side chain
amines of Lys262 (chains A and B), in addition to a potential hydrogen bond with the
phenolic group of Tyr273(A). The sulfonyl group of 4c display the potential to form
hydrogen bonds with the hydroxyl groups of Tyr273(B) and probably Ser274(A), which
are not possible for 3b (Figure 4B), which in contrast, exhibits the potential to form a
hydrogen bond with the main-chain amide of Asn276(B) (Figure 4D). For both 3b and 4c,
the 4-chromone moiety is predicted to interact with Tyr215 (both A and B) without
exhibiting any hydrogen bonds, whereas the adjacent carbonyl group of their linker can
form a hydrogen bond with the amide of Gly278(A).
22
ACS Paragon Plus Environment

Page 23 of 54
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 4. Close-up view of human RANKL dimer models in complex with the 4
potent inhibitors 1b (A), 3b (B), 4a (C) and 4c (D). Putative hydrogen bonding
interactions are indicated with yellow dashed lines and the heavy atom distance in Å,
whereas atom colors are as described in Figure 1. The models of RANKL dimer were
based on the X-ray structure of human RANKL with PDB ID: 3URF.19
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 54
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
Development of 3 potent inhibitors of RANKL-induced osteoclastogenesis with low
cell-toxicity
Based on the above observations, we decided to investigate the effect of substituting the
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
4-chromone and its adjacent amide moiety with a phenylsulfonylamide group. Although
substitution of the 4-chomone ring of SPD-304 by other aryl or heteroaryl rings (e.g. 5a-
e, 7a-d, 8c-e, 9b-c and 10a-b in Scheme 1) did not reveal any potent inhibitor of RANKL,
rather than partial inhibitors of osteoclastogenesis (7c, 8d-e and 10a in Table 1), our
choice was mainly driven by the positive effect of the phenylsuflonyl group displayed by
the most potent inhibitors 1b, 4a and 4c. By considering the potential hydrogen bonding
interactions of the 3-nitrophenyl group of 4c (Figure 4D), we prepared compounds 19a-c
according to Scheme 2.
24
ACS Paragon Plus Environment

Page 25 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Scheme 2. Synthesis of compounds 19a-c and 20a-c.Reagents and conditions: (i) KOH,
EtOH; then acetone at r.t.; (ii) POCl , DMF at r.t; then aq. KOH reflux; (iii) NaClO ,
3
2
H NSO , t-BuOH/H O at r.t.; (iv) C₆F₅OH, EDC, DCM at r.t.; (v) Et N, DCM at r.t.
3
3
2
3
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
In particular, the 4-chromone ring and the adjacent amide of 4c have been substituted
by phenylsulfonylamide (19a), 4-cyano-phenylsulfonylamide (19b) and 3-nitro-
phenylsulfonylamide (19c). We also prepared the corresponding cyclized piperazine
derivatives 20a, 20b, and 20c (Scheme 2), with the aim to investigate the effect of the
additional conformational restriction imposed by the cyclization of the linker moiety, and
based on the results obtained for the cyclized analogs of two potent RANKL inhibitors.
Specifically, 4b and 4d (the piperazine analogs of 4a and 4c, Scheme 1) displayed partial
and total inhibition of osteoclastogenesis, but with low and high toxicity, respectively
(Table 1).
Experimentally, all 6 compounds effectively inhibited RANKL-induced osteoclast
formation within an IC range of 1.40–6.40 μΜ and were less toxic than SPD-304 (Figure
50
5A-C, Supporting Information Figure S11, Table 2). Notably, three compounds (19a, 19b,
and 20a) displayed an increase of the therapeutic index (LC / IC ratio), more than 25
50
50
times compared to SPD-304 (Table 2). In particular, 19a and 19b displayed a dramatic
decrease of cellular toxicity, exhibiting LC50 values of more than 300 and 500 μΜ,
respectively, while retaining
a
high inhibitory effect in RANKL-induced
osteoclastogenesis. Compound 20a displayed the lowest IC50 values in TRAP activity
and attenuated NFATc1 and c-Fos induction upon RANKL treatment with a significant
(40-fold) decrease in toxicity with respect to SPD-304 (Figure 5D, Table 2). On the other
hand, 19c, 20b and 20c displayed a mild increase of the therapeutic index (LC / IC
50
50
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 54
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
ratio), 3-6 times compared to SPD-304 (Table 2). Taken together our results indicate that
despite the small number of compounds tested, we succeeded in developing 3 new
inhibitors of RANKL-induced osteoclastogenesis with a high therapeutic index (LC / IC
50
50
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
>
85).
Figure 5. Compounds 19a, 19b and 20a inhibit human RANKL-induced
osteoclastogenesis and exhibit dramatic decrease of toxicity. (A)
a
Osteoclastogenesis cultures treated with compounds 19a, 19b and 20a at 5μM in the
presence of RANKL (40 ng/mL) and M-CSF (25 ng/mL) for 5 days upon staining with
26
ACS Paragon Plus Environment

Page 27 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
TRAP. (B) IC50 calculation for each compound effect on RANKL-induced
osteoclastogenesis based on TRAP activity measured at day 4. (C) LC50 calculation for
each compound on BMM cell viability by MTT assay. (D) Western blot showing the effect
of preincubated compound 20a with RANKL at the indicated concentrations on the
induction of NFATc1 and c-Fos in RAW264.7 cells. All experimental assays were
repeated at least three times.
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
Examination of the molecular models of the most potent compounds (Supporting
Information Tables S7–S9 and Figures S12–S13) reveals that introduction of the second
phenylsulfonyl moiety in 19a and 19b, the 2 less toxic compounds identified (Figure 6),
retains the overall binding geometry with respect to that predicted for 3b and 4c (Figure
4). However, the increased versatility of the phenylsulfonyl group with respect to the rigid
chromenone ring exhibits a potential to be accommodated inside the shallow pocket of
RANKL, in addition to the formation of a putative hydrogen bond with the side chain of
Tyr215(A). The 4-cyano substituent of 19b shows an additional hydrogen bonding
potential as acceptor from the phenolic group of Tyr307(A) at the base of the pocket
(Figure 6B). Notably, the selected model of a cyclized analog, 20a, revealed a very similar
mode of binding with 19a-b (Figure 6C), albeit the conformational restraint imposed by
the 1,4-substituted piperazine moiety. Still, however, experimental verification of these
predictions has to wait for the determination of a co-crystal structure with the human
RANKL dimer.
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 54
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 6. Close-up views of the predicted inhibitor binding interface RANKL in
complex with 19a (A), 19b (B), and 20a (C). Putative hydrogen bonding interactions are
indicated with yellow dashed lines and the heavy atom distance in Å, whereas atom colors
are as described in Figure 1. The models of RANKL dimer were based on the X-ray
structure of human RANKL with PDB ID: 3URF.19
28
ACS Paragon Plus Environment

Page 29 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CONCLUSION
The development of RANKL inhibitors constitutes the most efficient approach to treat
osteoporosis and other bone resorption diseases. To date, the monoclonal antibody
denosumab has been approved for postmenopausal osteoporosis and bone metastasis
from solid tumors26–28, while only a few selective small-molecule inhibitors of RANKL have
been identified as alternative therapeutics36,37. Unfortunately, the majority of the potential
RANKL inhibitors were screened in activity and viability cell assays at similar
concentration ranges without providing IC50 and LC50 values, arising concerns about
interference of toxicity with the observed effectiveness.37,53 Furthermore, the use of
immortalized cell lines such as the preosteoclastic RAW264.7 cells for the LC50
evaluation35,53,54 could lead to underestimation of the compound toxicity compared to
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
primary cell cultures. Thus, the therapeutic index (LC50 to IC ratio) is an important
50
indicator of the probability of the successful development of a drug. SPD-304, despite its
promising effect as a dual inhibitor of RANKL and TNF41,42, displays a low therapeutic
index (LC / IC = 3.2), indicating high toxicity.
50
50
In the present study, we evaluated systematically 39 compounds against human RANKL,
using quantitative biochemical and cellular assays based on primary pre-osteoclasts to
determine their binding affinity, inhibition potency and toxicity, including specific cellular
assays to elucidate their mechanism of action. Four compounds, 1b, 3b, 4a, and 4c,
exhibited a dose-dependent inhibition of RANKL-induced osteoclastogenesis and
displayed substantially lower toxicity compared to SPD-304 (LC /IC range between 6.3
50
50
and 27.8). At the mechanistic basis of inhibition, we demonstrated that compounds 1b
and 3b interfered with the trimerization of human RANKL protein, similarly to SPD-304.
29
ACS Paragon Plus Environment