Identifying Novel Anti-Osteoporosis Leads with a Chemotype-Assembly Approach

Article abstract of DOI:10.1021/acs.jmedchem.9b00517

In this paper, we applied a chemotype-assembly approach for ligand-based drug discovery (LBDD) to discover novel anti-osteoporosis leads. With this new approach, we identified 12 chemotypes and derived 18 major chemotype assembly rules from 245 known anti-osteoporosis compounds. Then, we selected 19 compounds from an in-house compound library using chemotype-assembly approach for anti-osteoporosis assays, which resulted in 13 hits. Based on structural features in these 13 compounds, we synthesized 50 possible anti-osteoporosis compounds from the anti-osteoporosis chemotypes by means of click chemistry techniques and discovered a compound (10a, IC₅₀ = 2 nM) with nanomolar activity. Compound 10a was then proved to be an anti-osteoporosis lead since it can prevent bone loss in vivo.

Full text of DOI:10.1021/acs.jmedchem.9b00517

Subscriber access provided by Stockholm University Library
Article
Identifying novel anti-osteoporosis leads
with a chemotype-assembly approach
Chao Zhao, Dane Huang, Ruyue Li, Yida Xu, shimin su, Qiong Gu, and Jun Xu
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00517 • Publication Date (Web): 24 May 2019
Downloaded from http://pubs.acs.org on May 28, 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 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Identifying novel anti-osteoporosis leads
with a chemotype-assembly approach
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
†
†,‡
‡
†
†
,†
,†,§
Chao Zhao , Dane Huang , Ruyue Li , Yida Xu , Shimin Su , Qiong Gu* , Jun Xu*
†Research Center for Drug Discovery, School of Pharmaceutical Sciences, Sun Yat-Sen
University, Guangzhou 510006, China
‡Guangdong Province Engineering Technology Research Institute of T.C.M.,
Guangdong Provincial Key Laboratory of Research and Development in Traditional
Chinese Medicine, Guangzhou 510095, China
§School of Biotechnology and Health Sciences, Wuyi University, 99 Yingbin Road,
Jiangmen 529020, China
ABSTRACT
In this paper, we applied a chemotype-assembly approach for ligand-based drug
discovery (LBDD) to discover novel anti-osteoporosis leads. With this new approach,
we identified 12 chemotypes and derived 18 major chemotype assembly rules from 245
known anti-osteoporosis compounds. Then, we selected 19 compounds from an in-
house compound library using chemotype-assembly approach for anti-osteoporosis
assays, which resulted in 13 hits. Based on structural features in these 13 compounds,
we synthesized 50 possible anti-osteoporosis compounds from the anti-osteoporosis
chemotypes by means of click chemistry techniques and discovered a compound (10a,
IC = 2 nM) with nanomolar activity. Compound 10a was then proved to be a anti-
5
0
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 65
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
osteoporosis lead since it can prevent bone loss in vivo.
INTRODUCTION
0
1
2
3
4
5
6
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
Ligand-based drug discovery (LBDD) is an interesting rational drug design method
1
to accelerate the earlier stages of drug discovery. LBDD usually begins with a
collection of known molecules that have the same target or treat same diseases. This
collection is then used to perform quantitative structure-activity relationship (QSAR)
studies, substructure retrieval or similarity searches against commercial or in-house
databases.2-3 A key limitation of LBDD is that newly identified molecules are often
close analogues of known ligands and novelty therefore is limited.
Recently, we reported identification of the privileged fragments (substructures of
known compounds) and the assembly rules for liver X receptor-β (LXRβ) from known
4
LXRβ agonists. These fragments and rules are then successfully applied to discover
new agonists. This approach encourages us using substructures for LBDD and
overcomes the limitations of LBDD. However, a remaining question is whether this
approach can be applied to wider range of diseases. This is an important problem since
5
many diseases lack of validated molecular targets. We hypothesize that the active
agents related to a disease are constructed from different chemotypes according to some
assembly rules. The chemotype is defined as “a chemical structure motif or primary
substructure that is common to a group of compounds”.6
Osteoporosis affects more than 25% of women and more than 10% of men in their
lifetimes and is a disease that is characterized by loss of bone mass, leading to increased
2
ACS Paragon Plus Environment

Page 3 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
risk of fractures. It was selected to test our hypothesis.7-9 Anti-osteoporosis drugs
including bisphosphonates,10 cathepsin K inhibitors11 and parathyroid hormone
12
mimetics are available but challenges in osteoporosis therapy still remain, because no
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
drugs have been found that reduce the risk of fractures significantly.13
Here we applied a previously described de novo substructure generation algorithm
4
(DSGA) to extract privileged fragments from a library of known anti-osteoporosis
compounds. DSGA allows us to discover frequent substructures or fragments from a
14
chemical structure library. The fragments that remained after ‘rule of 3’ filtering were
clustered according to our previously described scaffold-based classification approach
1
5
(
SCA). SCA can provide the scaffold and complexity of a structure. The complexity
presents the heavy atom number of substituents on this scaffold. These two parameters
6
are applied to cluster the fragments. One cluster was identified as a chemotype and
then chemotype assembly rules were generated by analysis of the linkages between two
neighboring chemotypes. From the chemotypes and their assembly rules, a series of
substructures, which present by Markush queries, were constructed and used to screen
our in-house library. A number of candidate lead compounds were identified according
to the bioassay results. Further optimization with click reactions and chemotype
assembly rules gave a putative anti-osteoporosis lead compound. The flowchart for this
process is shown in Figure 1.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 65
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. The flowchart of candidate anti-osteoporosis lead discovery through assembly of chemotypes.
The process begins with a de novo substructure generation algorithm (DSGA) which, after inspection of
2
45 known anti-osteoporosis compounds identifies 502 chemical fragments. Chemotype analysis permits
clustering of the fragments into 12 chemotypes by SCA. By analyzing the linker between two
neighboring chemotypes in a known anti-osteoporosis compound, chemotype assembly rules can be
defined. Both the chemotypes and rules are used to construct a series of substructure. By searching our
in-house library with substructure retrieval, 19 compounds were selected for bioassay. Lead candidates
were further optimized by click chemistry and the anti-osteoporosis leads were validated with in vitro
and in vivo experiments.
RESULTS
Chemotypes Derived from Known Anti-osteoporosis Agents. A total of 245
known anti-osteoporosis agents were collected from published sources. These
compounds were divided into 28 classes based on their associated targets: 162
compounds have specific protein targets, 19 compounds are involved in signal
transduction pathways, 60 compounds are cell-based regulators, and 4 compounds have
been tested on animal models but lack target information (Tables S1, S2 in
Supplementary Information - SI).
Employing DSGA on this library resulted in 502 privileged fragments (Table S3).
According to the “rule of three”, fragments were eliminated if: (1) they are acyclic, (2)
their molecular weight is >300 and the number of hydrogen bond donors >3 and the
4
ACS Paragon Plus Environment

Page 5 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
number of hydrogen bond acceptors >3 and cLogP >3,14, 16 and (3) the smallest set of
1
7
smallest rings >1. This resulted in 54 fragments, which were classified by SCA into
1
2 chemotypes (Table 1).
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
Table 1. Anti-osteoporosis chemotypes, their frequency and appearance in known anti-
osteoporosis agents
Chemotype
IDa
Example
molecules
b
b
Topology Frequency EF
Targets
Ref.
N
O
A1
SERMd
A: C6AR1
B: C6AR
C: C6AR2
228
191
105
1
1
1
N
18
OH
HO
Bazedoxifene
N
T
CF3
SARMe
NC
F3C
19
20
CF3
Ligandrol
O
H
N
N
H
A
O
CN
cathepsin K
inhibitor
A1
N
N
Balicatib
HO
OH
O
P
HO
A
O
HO
A
A
P
BPsf
21
22
A
67
54
1
1
D: C5AR
E: C6HR
N
OH
Q
N
Minodronic acid
O
OH
O
S
N
NH
A1
D2R
Agonistg
A
N
Quinagolide
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 65
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
H
N
N
H
O
O
N
cathepsin K
inhibitor
O
N
F: C6HAR
47
1
23
S
O
A
O
N
0
1
2
3
4
5
6
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
Relacatib
OH
G: C6R
H: C5R
I: C6AR3
54
45
40
1
1
1
SARM
SARM
24
25
26
O
Methyltestosterone
OH
A1
A
A
O
O
Oxandrolone
HN
F
N
O
A
NH2
O
DPP-IV
Inhibitorh
A
O
F
A1
F
Evogliptin
CF3
N
O
J: C6RE
K: C6R1
34
28
1
1
CF3
SARM
27
20
A
A
O
N
HO
A
H
LGD2941
O
H
N
N
H
A1
O
CN
cathepsin K
inhibitor
N
A
N
Balicatib
N
O
A
A
A
L: C7R
10
1
N
SERM
18
OH
HO
Bazedoxifene
aThe general formula of chemotype ID is Cm(H)(A)R(n), where C = chemotype, m = number of ring
members, H = hetero (not essential), A = aromatic (not essential), R = ring and n = length of side chain (not
b
c
2
essential). EF = enrichment factor. The meaning of letters in topology structures: A = C, O, N, S, CO, SO ;
A
1
= non-terminal groups, such as O, N, S, CO, SO
2
3
; T = terminal substituents, such as H, Cl, F, Br, CF ,
d
CN, NO ; A-A
2
1
or A-A can be C=C, C≡C; Q = non-carbon heavy atom, such as O, N, S. SERM = selective
e
f
estrogen-receptor modulator. SARM = selective androgen-receptor modulator. BPs = bisphosphonates.
6
ACS Paragon Plus Environment

Page 7 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
gD2R agonist = dopamine receptor D
agonist. DPP-IV inhibitor = dipeptidyl peptidase-4 inhibitor.
h
2
As shown in Table 1, a chemotype is a core to a fragment family, in which ring can
be a part, but a chemotype is not equivalent to a ring. One compound usually
constructed by two or more chemotypes. For example, bazedoxifene, a selective
estrogen receptor modulator, consists of 5 chemotypes (3 x C6AR1, 1 x C7R, and 1 x
C5AR (Table 1 row 1); The cathepsin K inhibitor, balicatib consists of 3 chemotypes
(1 x C6HR, 1 x C6AR2, and 1 x C6R1 (Table 1 row 3) and methyltestosterone consists
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
of 4 chemotypes (1 x C6RE, 2 x C6R, and 1 x C5R (Table 1 row 7). The enrichment
factor (EF) of the chemotypes were calculated by following formulas28:
EF = (Ha × D) ÷ (Ht × A)
Where, Ht = total number of compounds with selected chemotype, Ha = the total
number of active molecules in the Ht, D = total number of molecules in library, and A
=
the total number of actives in the database.
Chemotype Assembly Rules Derived from Known Anti-osteoporosis Agents. The
assembly rules define two chemotypes and the linker between them. To explore
chemotype assembly rules systematically, we use the substructures which present
chemotypes as queries against the known anti-osteoporosis molecules. For the two
2
chemotype combinations, there are 퐶 +12= 78 queries, including homo- and hetero-
1
2
combinations, and these queries were used to retrieve compounds from the anti-
osteoporosis library. The structural moiety connecting two neighboring chemotypes is
identified as the linker and an example is described in Figure 2. In this way, a total of
five different linkers, shown in Figure 2 were identified. These are coupling (L1), two
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 65
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
chemotypes directly connected; alkane chain (L2), two chemotypes connected by an
alkyl chain with 1-3 carbons; carbonyl containing chain (L3), two chemotypes
connected by an alkyl chain with 1-3 carbons and containing a carbonyl group, such as
ketone, amide or ester; ring fusion (L4), two chemotypes form one fused ring; and ring
merging (L5), two 6 member ring overlapped to give a multisubstituted ring.
Theoretically, there are 78 × 5 or 390 assembly rules. Analysis of the anti-
osteoporosis library resulted in 131 assembly rules, and their frequencies were also
calculated (Figure S1). The assembly rules with L1 ~ L3 usually construct the
scaffold,29-30 L4 provides fused ring and rules involving L5 provide multisubstituted
rings. Therefore, L1~L3 are preferred for discovering new scaffold leads.
0
1
2
3
4
5
6
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. Linker identification process and assembly rules. Analysis of assembly rules with bazedoxifene.
(1) Results of chemotype substructure retrieval, red represents chemotype A; blue represents chemotype
D. (2) Identification of the linkers, L1, L2 and L4.
Performance Validation of the Chemotype Assembly Rules. A library of
chemicals associated with chemotype A consisting of 32 anti-osteoporosis compounds
was curated from published papers. The decoy library (8,558 compounds) was
constructed with the following procedure: 8,558 compounds (known drugs or agents
under clinical trials) were derived from DrugBank database. Anti-osteoporosis
8
ACS Paragon Plus Environment

Page 9 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compounds were excluded. The two libraries were combined to form an external testing
data set of 8,590 compounds for validation of the performance of the chemotype
assembly rules in the anti-osteoporosis lead discovery process. To test whether
assembly rules are more efficient than other approaches, single chemotype A (method
a), chemotype A with one other chemotype (method b) and substructures constructed
according to assembly rules were all used for screening against this test set. As shown
in Figure 3A, there are 3,630 molecules containing chemotype A in the test set (blue in
Figure 3A). Among these, 32 compounds are anti-osteoporosis agents (green). When
method b was applied, the number of retrieved molecules was reduced to 3,618 (yellow)
including all 32 osteoporosis agents. When the assembly rules of chemotype A were
applied (red), the number of retrieved molecules was significantly reduced to 610, still
including the 32 known osteoporosis agents. The hit rates of three methods are 0.88%,
0.88% and 5.25% respectively (Figure 3A). Thus, when assembly rules were applied,
the hit rate was increased more than 5 times. This validation was also applied to
chemotype D and chemotype H. A total of 28 anti-osteoporosis compounds associated
with chemotype D and decoys were used in the validation test (Figure 3B). The hit rates
of method a, method b and chemotype D with assembly rules are 1.43%, 1.57% and
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
5
.02% respectively. These results show that assembly rules of chemotype D increase
the hit rate by a factor of >3. A total of 8 chemotype H associated anti-osteoporosis
compounds and decoys were used in the validation test (Figure 3C). The hit rate of
chemotype H with assembly rules is 4.82%, about 10 times that of other 2 methods.
These data indicate that assembly rules can increase the efficiency in hit discovery. The
EF of our approach is also increased significantly compare to other two methods.
Accordingly, the chemotype-assembled substructures based on associated rules were
applied to the discovery of anti-osteoporosis leads.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Results of ligand-based virtual screening against an external test set. (A) Screening with
chemotype A and its assembly rules. (B) Screening with chemotype D and its assembly rules. (C)
Screening with chemotype H and its assembly rules. The screening with assembly rules increased the hit
rate by ~3 compared to other methods. The meaning of abbreviations in the Tables: D = number of
molecules in the external testing set; A = number of active compounds in the external testing set. Ha =
number of active molecules in the hit list; Ht = number of hit molecules from the database. Hit rate =
[(Ha/A) × 100]. EF = enrichment factor.
Identification of Anti-osteoporosis Lead Candidates through Screening against
an In-house Compound Library using Chemotype D Assembly Rules. Chemotype
1
0
ACS Paragon Plus Environment

Page 11 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
D and its assembly rules with L1~L3 was selected for further study for two reasons
because chemotype D has a high frequency of occurrence, more than one fourth of anti-
osteoporosis compounds contain chemotype D. Moreover, many clinical and
commercial anti-osteoporosis drugs including strontium ranelate, raloxifene,
bazedoxifene and relacatib are constructed by chemotype D. A second reason is that
chemotype D represents 5-membered aromatic rings, such as triazole, which can be
prepared by one step click chemistry. As shown in Figure 4A, chemotype D has 18
different assembly rules with chemotypes A, B, C, D, E, F, G, H, I, J and L and L1 ~
L3. For further study, we selected the most important assembly rules of chemotype D
with each of the chemotypes, including ‘D+L1+A’, ‘D+L1+B’, ‘D+L1+C’, ‘D+L3+D’,
‘D+L3+E’, ‘D+L1+F’, ‘D+L3+F’, ‘D+L1+G’, ‘D+L2+G’, ‘D+L2+H’, ‘D+L1+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
‘D+L2+I’ (Figure 4B). These assembly rules were then applied to construct a set of
substructures for drug discovery (Table 3).
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 65
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. (A) The assembly rules of chemotype D. The thickness of line represents the degree of use of
assembly rules, the thicker of line, the more frequent use of assembly rules. (B) Assembly rules between
chemotype D and other chemotypes with L1~L3. The numbers in the box represent the instance of
assembly rules. ‘X’ represents that there is no connection between two chemotypes.
Table 3. Substructures constructed by chemotype D assembly rules
ID
Substructures
Assembly rules
D+L1+A
D+L1+B
D+L1+C
D+L1+F
A
A
A
A
A
Q1
A
A
A
A
A
A
D+L1+A
D+L1+C
D+L1+I
A
A
A
A
A
A
A
A
A
n
Q2
A
A
A
n = 1~4
D+L2+I
A
A
A
A
Q3
Q4
A
A
D+L1+D
D+L3+D
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
n
n = 1~4
A
D+L3+E
D+L3+F
D+L2+G
A
A
A
A
A
A
A
n
A
Q5
A
A
A
n = 1~3
1
2
ACS Paragon Plus Environment

Page 13 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
A
Q6
Q7
A
A
D+L1+G
A
A
A
A
A
A
A
A
A
A
A
A
A n
D+L2+H
n = 1~4
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
Here, our in-house library containing 1,902 compounds were applied to discover anti-
osteoporosis hit compounds. The in-house library consists of 321 natural products and
1
581 synthetic compounds with 901 different scaffolds, and 41 compounds that have a
thiazole amide substructure. The substructures were then used to search for compounds
against osteoporosis in this library through substructure retrieve. After removal of
duplicates, this gave 128 compounds with 30 different scaffolds. After comparison with
known anti-osteoporosis agents, 19 hit compounds with novel scaffolds (Figure S2)
remained and were tested against receptor activator of nuclear factor-kappa B ligand
(
RANKL)-induced bone marrow macrophage cells (BMMs). Of these, 13 compounds
were confirmed as anti-osteoporosis active hits with IC values ≤10 μM (Table 4).
5
0
Primary structure-activity relationship (SAR) studies on the hits show that one of the
privileged substructures leading to the active compounds is either the thiazole amide
(
red in Table 4) or the thiophene amide (red). In our in-house library, only 41
compounds have a thiazole amide substructure. Here, 12 of them were tested and 10
showed activity. The XOS00048 and XOS00881 which also contain thiazole amide
exhibited no inhibition effects. This may be caused by the dioxole and nitrile groups,
since dioxole group is more rigid (XOS00048 vs. XOS00057) and nitrile group is a
potent electron withdrawing group compare to ethoxyl group on XOS00057. These data
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 65
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
suggest that further optimization of the thiazole amide compounds may afford anti-
osteoporosis leads.
Table 4. Structure and anti-osteoclastogenesis activity of selected compounds obeying
chemotype D assembly rules
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
HO
O
O
OMe O
N
O
O
S
N
N
S
N
O
N
O
H
O
OH
O
N
S
OMe
N
H
N
XOS00702
1.19 ± 0.26 μMa
XOS00055
2.14 ± 0.33 μMa
XOS00011
2.70 ± 0.18 μMa
XOS00700
3.15 ± 1.01 μMa
O
N
O
O
N
O
O
O
O
S
N
S
N
N
N
H
O
N
H
H
S
O
OH
S
N
H
COOEt
XOS00046
.98 ± 0.85 μMa
XOS00049
3.98 ± 1.44 μMa
XOS00056
4.03 ± 0.87 μMa
XOS00033
4.24 ± 1.71 μMa
3
O
N
O
N
O
N
O
N
N
S
EtO
EtO
O
O
H
S
N
S
N
N
S
H
O
H
H
O
O
N
OEt
O
XOS00036
.13 ± 1.06 μMa
XOS00057
6.75 ± 1.15 μMa
XOS00063
8.11 ± 0.99 μMa
XOS00067
10.01 ± 2.13 μMa
5
OMe
OMe
O
O
N
O
N
NH
N
H
S
O
O
S
N
N
N
N
O
N
H
N
O
N
CF3
CF3
XOS00064
2.59 ± 2.15 μMa
XOS00281
Inactiveb
XOS00048
Inactiveb
XOS00189
Inactiveb
1
N
OMe
OMe
N
N
O
N
N
O
NH
N
S
N
H
N
N
O
N
NC
CF3
CF3
XOS00280
Inactiveb
XOS00285
Inactiveb
XOS00881
Inactiveb
1
4
ACS Paragon Plus Environment

Page 15 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
aActivity data are IC50 values against RANKL-induced BMM cells. Inactive means that compounds have IC50
b
values >30 μM.
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
Optimization of Lead Candidates with Assembly Rules and Click Chemistry.
According to the assembly rule ‘D+L3+D’, a thiazole amide is constructed with a
triazole ring. A molecule with these three chemotypes can be easily prepared by click
chemistry. As shown in Figure 5, there are four kinds of scaffold that can be made in
this way, and scaffold IV was chosen based on the availability of starting materials and
ease of synthesis. Chemotypes A, B, C, D, and F were applied as the third chemotype
assembled with a triazole ring through L1.
Figure 5. Design of new compounds with thiazole amide (red) according to assembly rules.
In the synthetic strategy shown in Scheme 1, we acquired 5 reagents for R , and 13
1
reagents for R and 50 new compounds were synthesized.
2
Scheme 1. Synthesis of compounds 6-10.
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 65
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
R2
N
O
O
Et N, CH Cl
1. NaN , CuI, DMSO, r.t.
3
2
2
3
O
R1 NH2 + Br
R1
Br
+
R2
Br
r.t.
N
2. DBU, 70 °C
R1
N
H
N
N
H
1-5
6-10
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In Table 5, series 6 was assembled with different substituted thiazole or isoxazole
amides as R and R as chemotype A. Two different chemotypes A (p-methoxyphenyl
1
2
and m-aminophenyl) were introduced, and total of 8 compounds were obtained.
Osteoclastogenesis assay of these compounds on BMMs indicated that the isoxazole
ring-containing compound 6e is inactive, but the remaining compounds have IC50
values in the range of 0.25~3.89 μM .
Table 5. The structures of compounds 6a-6h and their activity of anti-
osteoclastogenesis
R2
N
R2
N
O
O
R1
N
R1
N
N
H
N
N
H
N
IC50
μM)a, b
IC50
Compd.
R1
R2
Compd.
6e
R1
R2
(
(μM)a, b
N
O
N
6
a
0.51 ± 0.15
3.89 ±1.44
Inactivec


OMe
OMe

OMe
S



NH2
N
N
N
6
b
6f
6g
6h
1.27 ± 0.31
0.25 ± 0.06
1.11 ± 0.39
S





S
NH2
NH2
N


OMe
OMe
6
c
2.33 ± 1.31
1.33 ± 0.58
EtO2C
S

S
N
N
6d
S

S

aConcentration (μM) for 50% inhibition of osteoclast differentiation in RANKL-induced BMMs. The IC50 values are the
b
mean ± SEM from at least three experiments. All active compounds were tested at a non-cytotoxic concentration and their
c
CC50 values have been provided in the SI (Table S4). Inactive means that compounds have IC50 values >30 μM.
1
6
ACS Paragon Plus Environment

Page 17 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Compounds in series 7 were assembled with substituted thiazole or isoxazole amides
and 7 different chemotypes B. A total of 26 compounds were synthesized (Table 6),
and 19 of them were active, with IC values in the range 0.06~9.32 μM. All the
5
0
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
chemotypes B except m-chlorophenyl exhibited good activity. Thiazole and o-
chlorophenyl combined to form a potent compound (7t) with an IC value of 0.06 μM.
5
0
A hydroxyapatite resorption assay indicated that 7t can discourage osteoclast function
significantly (Figure S3). Compounds containing benzothiazole and isoxazole moieties
(7m, 7n, 7r, 7s, 7w and 7x) show little or no activity.
Table 6. The structures of compounds 7a-7z and their anti-osteoclastogenesis activity
R2
R2
O
O
N
N
R1
N
R1
N
N
H
N
N
H
N
IC50
IC50
Compd
R1
R2
Compd
7n
R1
R2
(
μM)a, b
(μM)a, b
N
O
N


NO2
NO2

CN
7
a
1.04 ± 0.22
0.19 ± 0.44
9.32 ± 1.38
0.19 ± 0.06
S



N
N
N

F
7
b
7o
7p
7q
S


S
N



NO2
NO2


F
F
7
c
1.84 ± 0.45
3.28 ± 0.74
0.25 ± 0.06
0.66 ± 0.22
EtO2C
S

S
N
N
7
d
EtO2C
S

S

O
N
N
NO2


F
F
Inactivec
Inactivec
7e
1.22 ± 0.60
1.10 ± 0.29
0.52 ± 0.11
7r
7s
7t

S

N
O
N


Cl
7
f
S
S



Cl
N
N
Cl
7g
0.06 ± 0.01
S


1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cl
Cl
Cl
Cl
N
N



Cl
Cl
7
h
0.18 ± 0.05
0.13 ± 0.03
1.91 ± 0.71
0.27 ± 0.09
7u
7v
7w
7x
7y
7z
1.24 ± 0.33
EtO C
2




S

S

N
N
7
i
0.13 ± 0.04
Inactivec
Inactivec
EtO C
2
S

S

0
1
2
3
4
5
6
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
O
N
N
Cl
7
j

S

N
O
N

CN
CN
CN
7
k
S


Cl
Cl
N
N
7
l
0.31 ± 0.13
Inactivec


EtO C
2
S


S

O
N
N
Inactivec
Inactivec
7m


S

aConcentration (μM) for 50% inhibition of osteoclast differentiation in RANKL-induced BMMs. The IC50 values are the
b
mean ± SEM from at least three experiments. All active compounds were tested at a non-cytotoxic concentration and their
c
CC50 values have been given in SI (Table S4). Inactive means that compounds have IC50 values >30μM.
Compounds 8a-8f are constructed with 3 different chemotypes C, including p-
carbomethoxyphenyl, a p-phenylcarboxylic acid and a p-ethylnylphenyl group (Table
7
). The bioassay results indicated that chemotype C shows no anti-osteoclastogenesis
activity.
Table 7. The structures of compounds 8a-8f and their anti-osteoclastogenesis activity
R2
R2
O
O
N
N
R1
N
R1
N
N
N
N
N
H
H
IC50
(μM)a, b
IC50
(μM)a, b
Compd
R1
R2
Compd
8d
R1
R2
N
N
8
a
Inactivec
CO H
2
Inactivec

CO2Me

S

S

1
8
ACS Paragon Plus Environment

Page 19 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
N
N


CO2Me
CO2Me
Inactivec


Inactivec
8
b
8e
8f
S

S

O
O
N
N
Inactivec
Inactivec
8
c


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
aConcentration (μM) for 50% inhibition of osteoclast differentiation in RANKL-induced BMMs. The IC50 values
b
are the mean ± SEM from at least three experiments. All active compounds were tested at a non-cytotoxic
c
concentration and their CC50 values are given in the SI (Table S4). Inactive means that compounds have IC50 values
>
30μM.
Table 8 shows compounds assembled with chemotypes D or F (9-10). A total of 5
compounds (9a-9e) have chemotype D, and 2 of these exhibit potent activity with IC50
values of 0.04 μM (9b) and 0.02 μM (9c) respectively. Compounds 10a-10e, which
were constructed using chemotype F showed the most potent anti-osteoclastogenesis
effects compared to other chemotypes. All the compounds showed potent activity with
IC values in the range 0.002-0.43μM. Among all the newly synthetic compounds,
5
0
compound 10a exhibited most potent activity with IC = 0.002 μM. Compound 10a
5
0
showed 50-fold increase in potency compare to 10b. However, the difference of 10a
and 10b is only a methyl group. This is an example of the activity cliff, which is mainly
caused by LogP changes (0.97 for 10a and 1.7 for 10b). Also, the methyl group at 10b
may cause a steric clash against the receptor. Thus, the binding affinity of 10b and its
receptor is significantly reduced.
Table 8. The structures of compounds 9-10 and their anti-osteoclastogenesis activity
R2
R2
O
O
N
N
R1
N
R1
N
N
N
N
N
H
H
IC50
a, b
Compd
R1
R2
Compd
R1
R2
IC (μM)
50
(
μM)a, b
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 65
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
N
N
N
N


N
N
9
a
Inactivec
10a
10b
0.002 ± 0.0007
S
S




S
S
S
S


9
b
0.04 ± 0.008
0.10 ± 0.035
0.12 ± 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
N
N

N
9
c
0.02 ± 0.006
10c
EtO C
S
EtO C
2
2

S

S

N
N


N
N
9
d
21.35 ± 5.57
Inactivec
10d
10e
0.05 ± 0.02
0.43 ± 0.11
S
S


S

S

O
O
N
N
9
e


aConcentration (μM) for 50% inhibition of osteoclast differentiation in RANKL-induced BMMs. The IC50 values are the mean
b
±
SEM from at least three experiments. All active compounds were tested at a non-cytotoxic concentration and their CC50
c
values have been given in SI (Table S4). Inactive means that compounds have IC50 values >30μM.
In an effort to understand that how chemotypes affect anti-osteoclastogenetic activity,
we conducted a chemotype-activity relationship (CAR) analysis. Our current scaffold
consists of a thiazole amide and a triazole linker with a group R, which is in a position
where new chemotypes could be introduced (Figure 6). Based on the aforementioned
assembly rules and the experimental results, chemotypes A, B, D and F have a positive
impact on the activities and these effects, by magnitude are in the order F >D >B >A,
blue in Figure 6. On the other hand, chemotype C would nullify the desired activities.
Conventional SAR studies have also been conducted on these active compounds
(
yellow). For chemotypes A, B, and F, adding groups at positions 2’ and 4’ can increase
the activity, but adding groups to chemotype B at the 3’ position can diminish or
eliminate the desired activity. Substituent groups in the thiazole amide ring do not
significantly affect the activity, but replacing thiazole by isoxazole significantly
reduces or eliminates the activity. By analyzing the results of CAR and SAR, new lead
2
0
ACS Paragon Plus Environment

Page 21 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
optimizations are developed and can be used in future studies.
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. CAR and SAR of the discovered active anti-osteoporosis compounds. Blue background
represents CAR and yellow represents SAR.
Compound 10a Prevents Bone Absorption in vitro and Has No Effect on
Osteoblast Function. Compound 10a, the most potent osteoclastogenesis inhibitor
found so far, with IC = 2 nM, was selected for in vivo experiments and further studies
5
0
of gene expression. Compound 10a dose-dependently inhibits RANKL-induced
osteoclast differentiation in BMMs (Figure 7A). The osteoclastogenesis-related genes
(
TRAcP, c-Fos, MMP-9, and NFATc131-33) are highly expressed after RANKL
treatment, but are significantly suppressed when compound 10a is incubated with the
BMMs (Figure 7B). It was shown in this way that compound 10a inhibits osteoclast
differentiation. The effect of compound 10a on osteoclast function was further tested
with a hydroxyapatite resorption assay (Figure 7C). In cells treated with 10a, the area
of osteoclast hydroxyapatite resorption is significantly reduced, compared to the
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 65
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
RANKL-treated group. Alizarin red staining reveals that compound 10a exhibits no
cytotoxicity and does not affect osteoblast differentiation and function in C3H10T1/2
cells (Figure 7D and Figure S5). All of the results from experiments in vitro indicate
that compound 10a should be subjected to further studies in vivo.
0
1
2
3
4
5
6
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 7. (A) Compound 10a attenuates RANKL-induced osteoclastogenesis in a dose-dependent
manner. (B) Compound 10a (0.1 μM) down-regulates osteoclastogenic-related mRNA expression,
including TRAcP, NFATc1, MMP9 and c-Fos. **, P≤0.01 compared with RANKL. (C) Representative
image of bone resorption on hydroxyapatite-coated plates shows that 10a can prevent bone resorption in
vitro. (D) Determination of calcium deposition in differentiated osteoblasts by Alizarin red staining.
Compound 10a Prevents Bone Mass Loss in vivo. Studies of 10a in vivo were made
2
2
ACS Paragon Plus Environment

Page 23 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in an ovariectomized (OVX) rat model, and Fosamax (sodium alendronate), a widely
used and generally available drug was used as a positive control. Micro-CT results
showed that compound 10a significantly reduces bone mass loss (Figure 8A). Rats
treated with 10a show a loss of bone mineral density (BMD) of 8% on average. In the
OVX and Fosamax-treated groups, the BMD loss in rats is on average, 19% and 3%
respectively (Figure 8B). The level of collagen I C-terminal telopeptide (CTX-I), a bone
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
2
turnover marker which is released during bone resorption was determined and was
found to have decreased in rats treated with 10a (Figure 8D), compared to the OVX
group. The concentrations of osteocalcin and procollagen-I N-terminal propeptide
(
PINP)34-35 in the rats treated with 10a are close to those in untreated rats (Figure 8D)
and it was concluded that 10a activates new bone formation. The RANKL level in
serum was not affected by either 10a or Fosamax. Routine blood tests indicate that rats
treated with 10a exhibit results similar to those of normal rats (sham group) (Figure
S6). Compound 10a has no organ toxicity (Figure S7) and, judging by the in vivo data,
compound 10a is a bona fide anti-osteoporosis agent.
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 65
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 8. In vivo results for compound 10a. (A) Micro-CT images of trabecular bone of femurs from
sham-operated rats, ovariectomized rats (OVX) and rats treated with 10a. (B) Femur BMD determined
by dual-energy X-ray absorption. (C) Quantitative measurement of bone loss in femurs by micro-CT,
BV/TV, bone volume as a fraction of total bone volume; Tb.N, Trabecular Number; Tb.Th, Trabecular
Thickness; Tb.Sp, Trabecular Spacing. (D) Serum concentration of bone resorption and formation
#
#
markers, CTX-I, osteocalcin, PINP and RANKL. Error bars are mean ± s.d. , P≤0.01 compared with
control; ###, P≤0.005 compared with sham; *, P≤0.05 compared with OVX; **, P≤0.01 compared
with OVX; ***, P≤0.001 compared with OVX.
CONCLUSIONS AND DISCUSSION
The chemotype-assembly approach combines fragments into a ligand-based drug
discovery with superior performance. It is divided into three steps: (1) derivation of
chemotypes and their assembly rules from a library of known compounds with a
common target, a protein or a disease, (2) generation of compounds with affinity for a
2
4
ACS Paragon Plus Environment

Page 25 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
target or disease based on given chemotype-assembled substructures, and (3)
optimization of the identified leads using click chemistry, and in vitro and in vivo testing.
In this way, we successfully assembled 69 anti-osteoporosis candidate compounds,
confirmed 52 of these as anti-osteoporosis agents, and discovered 4 lead compounds
with new scaffolds and nanomolar activity. The efficacy of the best compound (10a)
was confirmed by in vivo experiments.
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 chemotype-assembly approach allows getting the smallest active substructures
for a target or diseases rationally and easily. By assembling the chemotypes through
associated rules, the substructures can be easily obtained. These substructures can lead
us to discover the hits of variety novel scaffolds with high hit rate. In this study, we
discovered 7 novel anti-osteoporosis scaffold which are never reported before.
Click chemistry is a convenient technology with which to quickly generate a
compound library. The triazole ring formed by click chemistry belongs to chemotype
D, therefore, it can be used to optimize the candidates through the associated assembly
rules. In this study, we successfully use click chemistry to optimize thiazole amide to a
potent anti-osteoporosis lead compound (10a). Click chemistry also can be applied to
discover de novo lead compounds according to the chemotype D assembly rules.
Compound 10a significantly inhibits osteoclastogenesis in vitro and prevents bone
loss in vivo. The serum PINP concentration of 10a treated osteoporosis rats is closer to
normal rats compare to Fosamax treated one. This result indicates compound 10a may
active new bone formation in vivo, since PINP is highly related to this function.
Therefore, compound 10a is a valuable lead for the development of potent anti-
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 65
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
osteoporosis drugs which can reduce the risk of fractures.
In conclusion, chemotype-assembly based drug discovery may offer a new approach
to ligand-based drug discovery.
0
1
2
3
4
5
6
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
EXPERIMENTAL SECTION
Materials. Chemicals were purchased from Energy Chemical Co., Sigma-Aldrich
Co. or Tokyo Chemical Industry Co. and utilized without further purification. Solvents
were distilled prior to use. For in vitro and in vivo assay, α-Modified Minimal Essential
Medium (α-MEM), Fetal bovine serum (FBS) and Phosphate buffered saline (PBS)
were purchased from Gibco (Gibco, NY, USA). Tartrate resistant acid phosphatase
(TRAcP) enzymatic activity was detected using the Leukocyte acid phosphatase
staining kit (Sigma, St. Louis, MO, USA). Recombinant macrophage colony
stimulating factor (M-CSF) was obtained from R&D Systems (Minneapolis, MN, USA).
Glutathione S-transferase (GST)-rRANKL160–318 (GST-rRANKL) recombinant
3
6
protein was expressed and purified as previously described . Cell Counting Kit-8
(CCK-8) was obtained from Beyotime (Beyotime technology, China). Cell dissociation
solution, dexamethasone, ascorbic acid, β-glycerol-phosphate and Alizarin red S was
obtained from Sigma Aldrich (Sigma Aldrich, St Louis, MO, USA). Hydroxyapatite-
coated 24 well plates were obtained from Corning (Corning, USA). TRIzol reagent was
obtained from Invitrogen (Invitrogen, NY, USA). PrimeScript™ MixRT Master Mix
reverse transcriptase kit and was TB Green™ premix EX Taq™Ⅱ kit obtained from
Takara (Takara Biotechnology, Kusatsu, Japan).
2
6
ACS Paragon Plus Environment

Page 27 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
13
NMR spectroscopy and Mass spectrometry. H-NMR and C-NMR spectra were
collected on a BRUKER 400 MHz spectrometer and were calibrated with
tetramethylsilane. The NMR data are displayed as follows: chemical shifts (δ) are
recorded as ppm, coupling constants (J) in hertz (Hz), integrity as the number of protons,
and multiplicity as s (singlet), d (doublet), t (triplet), q (quartet), quintet, sextet, and m
(multiplet). Mass spectra were obtained on a Shimadzu LCMS-2010EV utilizing the
electron-spray ionization method (ESI-MS). Thin-layer chromatography (TLC) was
carried out using plates coated with silica gel 60 F254, purchased from Merck. Column
chromatography was performed using Merck silica gel 60 (70 - 230 mesh).
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
Animals. We performed all procedures on mice and rats with approval by the Sun
Yat-sen University Institutional Animal Care and Use Committee. The C57BL/6 mice
and female Sprague-Dawley (SD) rats were obtained from Guangdong Medical Lab
Animal Center. Mice and rats were maintained in pressurized ventilated cages under
conditions of repeated controlled illumination (12 h dark; 12 h light) with ad libitum
access to sterilized water and food.
General procedure A for synthesis of compounds 1-5. Different substituted amino
thiazoles and Et N was dissolved in CH Cl . Bromoacetyl bromide was added into the
3
2
2
mixture by drop-wise at 0 ℃, and stirred at r.t. for 8h, followed by extraction with
CH Cl . The extract was then washed with water, and brine, dried over Na SO , and
2
2
2
4
evaporated in vacuo. The residue was purified by column chromatography to give
products 1-5 as a solid. All compounds were found to have ≥95% purity by HPLC
(
Takara Biotechnology, Kusatsu, Japan)
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 65
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2-Bromo-N-(thiazol-2-yl)acetamide (1). The general procedure A was applied with
2
-amino-thiazole (100 mg, 1 mmol), Et N (218 mg, 2 mmol) and bromoacetyl bromide
3
1
(
301 mg, 1.5 mmol) to afford product 1 as a white solid (200 mg, 91.3%). H-NMR
0
1
2
3
4
5
6
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
(400 MHz, CDCl ) δ: 7.53 (d, J = 4.0 Hz, 1H), 6.74 (d, J = 4.0 Hz, 1H), 5.21 (s, 2H).
3
2
-Bromo-N-(4-methylthiazol-2-yl)acetamide (2). The general procedure A was
applied with 2-amino-4-methylthiazole (114 mg, 1 mmol), Et N (218 mg, 2 mmol) and
3
bromoacetyl bromide (301 mg, 1.5 mmol) to afford product 2 as an off-white solid (185
1
mg, 78.7%). H-NMR (400 MHz, CDCl ) δ: 6.82 (s, 1H), 5.22 (s, 2H), 2.28 (s, 3H).
3
Ethyl 2-(2-bromoacetamido)-4-methylthiazole-5-carboxylate (3). The general
procedure A was applied with ethyl 2-amino-4-methylthiazole-5-carboxylate (186 mg,
1 mmol), Et N (218 mg, 2 mmol) and bromoacetyl bromide (301 mg, 1.5 mmol) to
3
1
afford product 3 as a brown solid (177 mg, 57.7%). H-NMR (400 MHz, CDCl ) δ:
3
5
.22 (s, 2H), 4.22 (q, J = 8.0 Hz, 2H), 2.53 (s, 3H), 1.26 (t, J = 8.0 Hz, 3H).
N-(Benzo[d]thiazol-2-yl)-2-bromoacetamide (4). The general procedure A was
applied with 2-benzothiazolamine (150 mg, 1 mmol), Et N (218 mg, 2 mmol) and
3
bromoacetyl bromide (301 mg, 1.5 mmol) to afford product 4 as a yellow solid (120
1
mg, 44.3%). H-NMR (400 MHz, CDCl ) δ: 7.77 (d, J = 8.0 Hz, 1H), 7.74 (d, J = 8.0
3
Hz, 1H), 7.40 (t, J = 8.0 Hz, 1H), 7.28 (t, J = 8.0 Hz, 1H), 5.23 (s, 2H).
2-Bromo-N-(5-methylisoxazol-3-yl)acetamide (5). The general procedure A was
applied with 5-methylisoxazol-3-amine (99 mg, 1 mmol), Et N (218 mg, 2 mmol) and
3
bromoacetyl bromide (301 mg, 1.5 mmol) to afford product 5 as a white solid (195 mg,
1
8
9.9%). H-NMR (400 MHz, CDCl ) δ: 6.58 (s, 1H), 4.99 (s, 2H), 2.41 (s, 3H).
3
2
8
ACS Paragon Plus Environment

Page 29 of 65
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
General procedure B for synthesis of compounds 6–10. The bromo-substituted
compounds 1-5, CuI and sodium azide was dissolved in DMSO, stirred at r.t. for 0.5 h.
Then different alkyne compounds and DBU was added into the mixture, stirred at 70 ℃
for 8 h. The mixture was followed by extraction with EtOAc. The extract was then
washed with water, and brine, dried over Na SO , and evaporated in vacuo. The residue
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
2
4
was purified by column chromatography to give products 6-10 as a solid.
-(4-(4-Methoxyphenyl)-1H-1,2,3-triazol-1-yl)-N-(thiazol-2-yl)acetamide (6a).
2
The general procedure B was applied with compounds 1 (221 mg, 1 mmol), NaN (130
3
mg, 2 mmol), CuI (19 mg, 0.1 mmol), 4-ethynylanisole (132 mg, 1 mmol) and DBU
1
(
304 mg, 2 mmol) to provide product 6a as white solid (93 mg, 29.5%). H-NMR (400
MHz, DMSO-d6) δ 12.58 (s, 1H), 7.54 (d, J = 4.0 Hz, 1H), 7.27-7.25 (m, 2H), 7.04-
1
3
7
.02 (m, 2H), 6.81 (s, 1H), 6.71 (d, J = 4.0 Hz, 1H), 5.53 (s, 2H), 3.88 (s, 3H). C-
NMR (100 MHz, DMSO-d6) δ 131.7, 119.4, 107.8, 100.0, 60.1, 45.4, 41.8, 33.6, 19.3.
ESI-MS: 316.0 [M+H]+.
2
-(4-(4-Methoxyphenyl)-1H-1,2,3-triazol-1-yl)-N-(4-methylthiazol-2-
yl)acetamide (6b). The general procedure B was applied with compounds 2 (233 mg,
1
1
mmol), NaN (130 mg, 2 mmol), CuI (19 mg, 0.1 mmol), 4-ethynylanisole (132 mg,
3
mmol) and DBU (304 mg, 2 mmol) to provide product 6b as white solid (142 mg,
1
43.2%). H-NMR (400 MHz, DMSO-d6) δ 12.62 (s, 1H), 11.29 (s, 1H), 8.07-8.02 (m,
1
3
4
H), 6.82 (s, 1H), 5.53 (s, 2H), 3.88 (s, 3H), 2.28 (s, 3H). C-NMR (100 MHz, DMSO-
d6) δ 166.4, 145.7, 135.6, 130.4, 129.2, 125.7, 124.8, 108.8, 52.6, 52.2, 17.3. ESI-MS:
30.0 [M+H]+.
3
2
9
ACS Paragon Plus Environment