Discovery of a Natural-Product-Derived Preclinical Candidate for Once-Weekly Treatment of Type 2 Diabetes

Article abstract of DOI:10.1021/acs.jmedchem.8b01491

Poor medication adherence is one of the leading causes of suboptimal glycaemic control in approximately half of the patients with type 2 diabetes mellitus (T2DM). Long-acting antidiabetic drugs are clinically needed for improving patients' compliance. Dipeptidyl peptidase-4 (DPP-4) inhibitors play an increasingly important role in the treatment of T2DM because of their favorable properties of weight neutrality and hypoglycemia avoidance. Herein, we report the successful discovery and scale-up synthesis of compound 5, a structurally novel, potent, and long-acting DPP-4 inhibitor for the once-weekly treatment of T2DM. Inhibitor 5 has fast-associating and slow-dissociating binding kinetics profiles as well as slow clearance rate and long terminal half-life pharmacokinetic properties. A single-dose oral administration of 5 (3 mg/kg) inhibited >80% of DPP-4 activity for more than 7 days in diabetic mice. The long-term antidiabetic efficacies of 5 (10 mg/kg, qw) were better than those of the once-weekly trelagliptin and omarigliptin, especially in decreasing the hemoglobin A1c level.

Full text of DOI:10.1021/acs.jmedchem.8b01491

Subscriber access provided by EKU Libraries
Article
Discovery of a Natural-Product-Derived Preclinical
Candidate for Once-Weekly Treatment of Type 2 Diabetes
Shiliang Li, Chun Qin, Shichao Cui, Hongling Xu, Fangshu Wu, Jiawei Wang,
Mingbo Su, Xiaoyu Fang, Dan Li, Qian Jiao, Ming Zhang, Chunmei Xia, Lili Zhu,
Rui Wang, Jia Li, Hualiang Jiang, Zhenjiang Zhao, Jing-Ya Li, and Honglin Li
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01491 • Publication Date (Web): 29 Jan 2019
Downloaded from http://pubs.acs.org on January 30, 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 52
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 52
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
Discovery of a Natural-Product-Derived Preclinical Candidate for
Once-Weekly Treatment of Type 2 Diabetes
1
,†
1,†
2,3,†
1
1
1
Shiliang Li , Chun Qin , Shichao Cui , Hongling Xu , Fangshu Wu , Jiawei Wang ,
0
1
2
3
4
5
6
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
1
2
1
1
2
1
Mingbo Su , Xiaoyu Fang , Dan Li , Qian Jiao , Ming Zhang , Chunmei Xia , Lili Zhu , Rui
1
2
2
1,
2,3,
1,
Wang , Jia Li , Hualiang Jiang , Zhenjiang Zhao *, Jingya Li *, Honglin Li *
1
Shanghai Key Laboratory of New Drug Design, State Key Laboratory of Bioreactor
Engineering, School of Pharmacy, East China University of Science and Technology,
Shanghai 200237, China.
2
State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Chinese
Academy of Sciences. Shanghai 201203, China.
3
University of Chinese Academy of Sciences, No.19A Yuquan Road, Beijing, 100049, P. R.
China
†
Authors contributed equally to this work
*To whom correspondence should be addressed. Email: zhjzhao@ecust.edu.cn,
jyli@simm.ac.cn, hlli@ecust.edu.cn
Abstract
Poor medication adherence is one of the leading causes of suboptimal glycaemic control
in approximately half of the patients with type 2 diabetes mellitus (T2DM). Long-acting
antidiabetic drugs are clinically needed for improving patients’ compliance. DPP-4 inhibitors
play an increasingly important role in the treatment of T2DM for their favorable properties of
weight neutrality and hypoglycemia avoidance. Herein, we report the successful discovery
and scale-up synthesis of compound 5, a structurally novel, potent and long-acting DPP-4
inhibitor for the once-weekly treatment of T2DM. Inhibitor 5 has fast-associating and
slow-dissociating binding kinetics profiles as well as slow clearance rate and long terminal
half-life pharmacokinetic properties. A single dose oral administration of 5 (3 mg/kg)
inhibited >80% DPP-4 activity for more than 7 days in diabetic mice. The long-term
antidiabetic efficacies of 5 (10 mg/kg, q.w) were better than the once-weekly trelagliptin and
omarigliptin, especially in decreasing the hemoglobin A1c (HbA1c) level.
Keywords: Drug discovery, DPP-4 inhibitor, Natural product, Diabetes, Once-weekly
treatment
ACS Paragon Plus Environment

Page 3 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
Diabetes mellitus is one of the largest health emergencies globally. The global
healthcare costs to treat diabetes and related complications have reached 727 billion dollars 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
1
2017. The worldwide prevalence of diabetes mellitus for adults is growing at a continuous
rate and has reached a total of 425 million in 2017, and the number will rise to 629 million by
2045. Among the diagnosed patients, approximately 90% of cases are type 2 diabetes
mellitus (T2DM). Diabetes causes significant morbidity and mortality. Prevention of
diabetes-related micro- and macrovascular complications such as nephropathy, retinopathy,
neuropathy, and stroke in diabetic patients requires tight glycemic control. An increasing
number of antihyperglycemic agents have been approved for the treatment of T2DM.
Compared with the traditional antidiabetic drugs (mainly biguanides, sulfonylureas,
thiazolidinediones, meglitinides, α-glucosidase inhibitors and insulin) that usually bring
undesirable side effects including weight gain, hypoglycemia and gastrointestinal distress,
newer therapies such as glucagon-like peptide-1 (GLP-1) receptor agonists, DPP-4 inhibitors
and sodium-glucose co-transporter 2 (SGLT2) inhibitors have more advantages because of
2
their efficacy and safety. Nevertheless, in spite of their potency in lowering HbA1c level
without causing hypoglycemia, a particular inconvenience with GLP-1 receptor agonists
therapy is its administration by injection, while SGLT2 inhibitors may potentially lower
blood pressure and increase genital infection. Thus, DPP-4 inhibitors tend to draw more
attention.3, 4
The DPP-4 enzyme is a widely expressed serine protease that belongs to the
α/β-hydrolases (family S9B). It can selectively inactivate peptides with an alanine, proline, or
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 52
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
5
serine residue at the penultimate position from the N-terminus. Inhibition of the catalytic
activity of DPP-4 helps to elevate GLP-1 level, which is an important incretin secreted by the
small intestine in response to meal ingestion, and amplifies the insulin secretory response to
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
nutrients through their cognate G protein-coupled receptors in the pancreatic β cells. GLP-1
regulates glucose lowering activities in a glucose-dependent manner, however, it has short
half-lives in vivo (< 2 min), and is rapidly degraded to their inactive form by DPP-4.6
Therefore, by inhibiting the catalytic activity of DPP-4, DPP-4 inhibitor helps to maintain the
endogenous GLP-1 concentration, and is a clinically proven, therapeutically novel approach
to treat T2DM.7
Nowadays, eleven DPP-4 inhibitors have been approved worldwide (Fig. S1). Classical
sitagliptin (Fig. 1a), saxagliptin, alogliptin, and linagliptin et al. are all DPP-4 inhibitors given
once per day. Since T2DM is a chronic, lifelong condition that requires patients to take
anti-diabetic drugs regularly to control glucose levels, long-acting DPP-4 inhibitors have
8
been clinically needed for improving patients’ compliance. Trelagliptin and omarigliptin are
the only two gliptins that have recently received marketing authorization in Japan in 2015 for
once-weekly dosing regimen.9, 10 Patient satisfaction and compliance improved after
switching from taking once-daily DPP-4 inhibitor to once-weekly DPP-4 inhibitor in Japan.11,
12
However, manufacturers of both drugs have announced that they discontinued their plans
for seeking regulatory approval in countries other than Japan. Omarigliptin was reported to
have a higher rate of serious adverse events than sitagliptin, showing elevated rates of
cardiovascular issues, hepatobiliary events and prostate cancer.13-16 Although currently
available data of trelagliptin suggest that trelagliptin is a safe and efficacious medication with
ACS Paragon Plus Environment

Page 5 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
7
a similar safety profile to the once-daily alogliptin, further studies are needed to investigate
1
8
the effect of trelagliptin on cardiovascular events, since its close analog alogliptin had been
19
warned by the U.S. FDA for an increased risk of heart failure. It is still unclear whether the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
excessive heart failure effect is target related or compound related, because sitagliptin did not
increase cardiovascular risk in T2DM patients.20, 21 Structurally distinct DPP-4 inhibitors may
serve as new tools to address this question. Therefore, it is still a necessity and challenge to
find novel and safe DPP-4 inhibitors with long-acting antidiabetic efficacy, especially in
China where more than 110 million adults have T2DM, accounting for about 25% of all
T2DM patients worldwide.
Previously, based on a natural product lead iso-daphnetin (Fig. 1a), we have discovered
a novel series of 3-phenyl-2,3-dihydro-1H-benzo[f]chromen-2-amine derivatives as potent
2
2
DPP-4 inhibitors by scaffold hopping. Among those inhibitors, compound 1 (Fig. 1a) was
the most promising one, which could inhibit > 80% DPP-4 activity for more than 24 h with a
single oral dose of 3 mg/kg in diabetic mice. However, the long-acting antidiabetic efficacy
of 1 still cannot afford a once-weekly dosing regimen. Both the ligand-protein binding
kinetics and pharmacokinetic profiles of 1 should be optimized because those two profiles are
reported to directly influence the duration of a drug’s in vivo efficacy.23-26
Herein, based on the lead compound 1, we report the successful discovery, scale-up
synthesis and biochemical, structural, pharmacokinetic as well as pharmacological evaluation
of a natural-product-derived DPP-4 inhibitor 5 for once-weekly treatment of T2DM with
better and longer antidiabetic efficacy than trelagliptin and omarigliptin.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 52
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
Fig. 1 Discovery and design of long-acting DPP-4 inhibitor 5. (a) Chemical structures of
representative DPP-4 inhibitors. (b) Binding mode of DPP-4 inhibitor 1 (PDB code: 5J3J),
key residues are shown in grey sticks, while Arg358 from 1X70 binding sitagliptin as a
ligand is shown in purple for comparison. Hydrogen bonds are displayed as black dashes. (c)
Pair interaction energy (PIE) between the residues at the active site of DPP-4 and inhibitor 1
ACS Paragon Plus Environment

Page 7 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and (d) Pair interaction energy decomposition analysis (PIEDA) for these residues. The
electrostatics, charge-transfer, dispersion, and exchange-repulsion PIE terms are represented
in green, orange, yellow, and blue, respectively. (e) Structures of three close analogs of 1:
compounds 2-4. (f) PIEs between residues at the S2 extended subsite of DPP-4 and 1-4. (g)
Correlations between experimentally measured affinities (DPP-4, pIC ) and sums of PIEs
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
0
between residues at the S2 extended subsite (PIE-sumS2 ext.) of DPP-4 and inhibitors 1-4. (h)
Structure of long-acting candidate 5 and the electrostatic surface potential (ESP) of its
fragment of 8-cyano-9-methoxy-naphthalene. (i) Simulated binding mode of 5 with Arg358
in a closed-up conformation for FMO calculation. (j) PIEDA for the residues at the active site
of DPP-4 and inhibitor 5.
Results
Design of the potent and long-acting DPP-4 inhibitor
In our previous study, we found that both shape and electrostatic complementarities at
the S2 extended subsite of the DPP-4 catalytic site influence the bioactivity of the inhibitors
2
2
in our series. The flexible Arg358 can be attracted by the negative electrostatic potential
(ESPmin = -36.41 kcal/mol) produced by the two methoxyl oxygen atoms in 1, and rotated its
guanidine group into a closed-up conformation to generate hydrogen bonds with the
methoxyl oxygen atoms (Fig. 1b). Therefore, compound 1 has a nanomolar binding affinity
-9
−3 -1
(K = 2.38×10 M) against DPP-4 and a slower dissociation rate (k = 3.43×10 s ) than
D
off
sitagliptin (koff = 1.49×10-2 s-1) whose Arg358 flips away in an open conformation.
Conformational changes were reported to be determinative features of many protein-ligand
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 52
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
7
systems with slow dissociation kinetics. Thus, we speculated that interacting with Arg358
may be a key factor influencing the dissociation rate of our compounds in this series since
Arg358 is flexible likes a door which can hinder the ligand from leaving.
0
1
2
3
4
5
6
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 order to investigate how Arg358 contributes to the ligand-protein binding, the
fragment molecular orbital (FMO) quantum-mechanical (QM) method28-31 was applied to
quantify the interactions formed between 1 and the pocket-lining residues of DPP-4. After
FMO calculation, we found that dispersion interactions are prominent at the S1 and S’1
subsites (Fig. 1c and 1d), indicating that hydrophobic interactions are major inter-molecular
interactions at those regions including residues His740, Tyr547, Try631, Tyr666, Val656,
Tyr662, Val711, Asn710. Strong electrostatic interactions were presented at the S2 subsite,
especially between 1 and residues Glu205 and Glu206, which are the two most important
residues in the binding pocket that most of the reported DPP-4 inhibitors are interacting with.
At the S2 extended subsite including residues Val207, Ser209, Phe357 and Arg358, we were
surprised to observe that apart from the strong π-π stacking interactions between 1 and
Phe357 informed by the pair interaction energy decomposition analysis (PIEDA), pair
interaction energy (PIE) between 1 and Arg358 is comparable to that of Glu206 (-34.49 vs
-
32.94 kcal/mol), indicating that Arg358 contributes significantly to the binding stability of
the complex. To explore whether such a great contribution made by Arg358 to the binding
affinity is unique to the scaffold of 1, we subsequently performed FMO calculation on the
X-ray complex structure of sitagliptin and DPP-4 (PDB: 1X70) in which Arg358 adopts an
open conformation to accommodate the -CF3 group of sitagliptin. Results showed that PIE
between sitagliptin and Arg358 is -9.59 kcal/mol (Fig. S2), only a third of that of the
ACS Paragon Plus Environment

Page 9 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1-Arg358 pair. The above results suggested that the ligand-induced conformation change of
Arg358 plays an important role in the binding of 1 to DPP-4, and contributes significantly to
the binding affinity.
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
Expecting to keep Arg358 in a closed-up conformation in the subsequent structural
optimization, the specific contribution made by the two methoxyl groups should be studied to
illuminate us which methoxyl group is appropriate for substituting at the S2 extended subsite.
Then, FMO calculations were performed on three close analogs of 1 (Fig. 1e): compound 2
1
2
with only one methoxyl group at R , compound 3 with only one methoxyl group at R , and
compound 4 with neither of the methoxyl groups. Preparations of the complex structures of
2
-4 with DPP-4 were described in the experimental section. For the four residues at the S2
extended subsite, major PIE differences were observed against Arg358 (Fig. 1f and Fig. S2).
1
2
We found that methoxyl group at R and methoxyl group at R play an almost equally
important role in interacting with Arg358, because PIEs of the 2-Arg358 and 3-Arg358 pairs
were close to each other (-18.52 vs -19.40 kcal/mol). Without two methoxyl groups, 4
interacted Arg358 with PIE value of -4.16 kcal/mol, a more than 4 fold decrease than that of
2 or 3. Whereas, the occurrence of the two methoxyl group brought 1 the most strong
interaction with Arg358 (PIE = -34.49 kcal/mol). The sums of the PIEs (PIE-sumS2 ext.
)
between each analog and residues at the S2 extended subsite correlated well with the four
2
inhibitors’ biological potencies (R = 0.84, Fig. 1g). Those results suggested that functional
1
2
groups at R and R should cooperate with each other to attract Arg358 to form a closed-up
conformation, and two methoxyl groups in 1 made nearly equal contributions. Compared
1
2
with R position that has limited space, R positon is open wide enough, thus we decided to
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 52
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
decorate at R with the attempt to keep Arg358’s closed-up conformation.
2
The methoxyl oxygen atom of R in 1 that makes H-bond with Arg358 is 1.4 Å away
from the attached aromatic carbon atom. In order to mimic the role of that oxygen atom, an
H-bond acceptor that attached close to the aromatic ring but with greatly increased
electro-withdrawing profile is preferred. A linear and rigid cyano group, which is 2.4 Å long
with an electronegative nitrogen atom that has a set of lone pair electrons, emerged in our
0
1
2
3
4
5
6
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
2
mind immediately. The ESPmin of the fragment containing R of -OCH and R of -CN was
3
more negative than that of the fragment in 1 (-44.17 vs -36.41 kcal/mol) (Fig. 1h), indicating
that 5 may be capable to fix Arg358 in a closed-up conformation by strong dipole-dipole
interaction. We then constructed the complex structure of 5 and DPP-4 (Fig. 1i) and
performed the FMO calculation. PIEDA results informed us that the overall energy of the
electronic, charge transfer and dispersion interactions of the 5-Arg358 pair (-49.45 kcal/mol)
is comparable to that of the 1-Arg358 pair (-48.90 kcal/mol). Meanwhile, the introduction of
the cyano group made the molecule more polar in nature, leading to strengthened electrostatic
interactions between the amino group and residues Glu205 and Glu206 at the S2 subsite (Fig.
1j). Those calculating data suggested that 5 is of great possibility to lock Arg358 in a
closed-up conformation. Additionally, substituting -OCH with -CN may also improve the
3
pharmacokinetic features of the compound. Subsequently, compound 5 was synthesized for
further experimental verification.
Scale-up synthesis of compound 5
ACS Paragon Plus Environment

Page 11 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CHO
F
F
F
O
O
CHCl OCH /TiCl
4
O
O
2
3
o
CH Cl , 0-10 C
2
2
OHC
NC
NC
1
0
Yield 90%
12
11
MeNO ,NaOH
KI/AlCl3
2
o
o
0
-10 C
MeCN, 80 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
Yield 78%
Yield 89%
F
2
F
F
NO
CHO
OH
L-Pipecolinic acid
F
F
F
O
+
o
O
O
Toluene, 120 C
NO2
NC
Yield 75%
NC
9
8
7
F
F
F
F
F
NO
NH
2
2
o
NaBH , 0-10 C
4
DIPEA
Zn/HCl
EtOH
F
O
O
O
O
CH OH/THF
Ethanol
3
Yield 80%
NC
NC
6
(two pair of enantiomers)
5 (one pair of trans-enantiomers)
F
F
F
NH
2
2
3
(
+)-Dibenzoyl-D-tartaric acid
Acetone/H O
Yield 37%
O
O
ee > 99%
total yield: 17%
2
NC
5
Scheme 1. Scale-up synthesis of compound 5.
Due to the excellent research and development prospects of 5, pilot-scale synthesis of 5
22
must be solved. If we use the previously reported synthetic route of 1 to synthesize 5, a total
yield of less than 1% would be obtained, and finally only mg-level product could be acquired.
Besides the low selectivity of demethylation for intermediate 11 and low condensation
efficiency of intermediate 7, the separation of the enantiomers by preparative HPLC in the
synthetic route of 1 also hampered to provide Kg-level product for further studies.
Herein, we reported an optimized route for the preparation of 5 with the overall yield
improved to 17% (Scheme 1). In this optimized approach, the AlCl /KI-mediated cleavage of
3
the methyl-oxygen bond in compound 11 was proceeded regio-selectively and isolated as a
white crystalline solid in more than 90% purity through a simple filtration at a yield of 89%
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 52
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
which is much higher than the previous 40%. The key intermediate 7 was successfully
obtained by Michael addition and Henry reaction of substituted salicylaldehyde 8 on
nitroalkene 9. After optimization by trying out a variety of bases and ligands, we found that
the use of L-(-)-piperidine-2-carboxylic acid or morpholine instead of DABCO as catalyst
gave the corresponding coupling product 7 more effectively in 75% yield which is much
better than the ~20% of the original process, and the crude product was used in the next step
0
1
2
3
4
5
6
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
without further purification. Reduction of 7 with NaBH gave a mixture of cis/trans
4
dihydrochroman derivatives 6, then configuration transformation of the cis isomers to the
trans isomers could occur in one-pot upon exposure to DIPEA in ethanol at room
temperature. Subsequently, reduction of the nitro group of trans isomers of 6 to an amine
proceeded with Zn and HCl in ethanol formed enantiomers of 5 smoothly. The two-step
reaction of NaBH reduction and configuration transformation was completed in one step,
4
and the yield was 80%, which is significantly higher than the yield of less than 50% of the
original process.
The absolute configuration of 5 was determined to be (2S, 3R) by X-ray crystallographic
analysis. Having established a scale-up method to access trans-enantiomers 5, we explored
several commercially available resolving agents to afford optically pure 5. Through a rational
screening approach, (+)-Dibenzoyl-D-tartaric acid was selected as the resolving agent to
achieve the desired salts of 5 with ee value 99% and yield 37% after re-crystallization twice
from acetone/H O, which is much more convenient than the separation of enantiomers 5 by
2
chiral preparative HPLC in the original process. The total yield of the entire synthetic route
was increased from less than 1% to nearly 20% in the present study, and the operation
ACS Paragon Plus Environment

Page 13 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
process has been greatly simplified. Through the Kg-level production procedure above
described, we have finished nearly 10 Kg product of 5 for ongoing preclinical 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
Inhibitor 5 binds DPP-4 with fast-associating and slow-dissociating profiles
Compound 5 inhibited DPP-4 activity (Table S1) reversibly and showed a clear
characteristic of fast-associating and slow-dissociating inhibition (Table 1, Fig. 2a) with the
7
-1 -1
association rate constant (k ) of 1.29×10 M s and dissociation rate constant (k ) of
on
off
-3 -1
2
.29×10 s . Compared with 1, the association rate constant of 5 has increased by an order of
magnitude, while the dissociation rate constant decreased slightly. Hence, 5 has a smaller KD
-10
-9
value than 1 (1.77×10 vs 2.38×10 M). Moreover, compound 5 binds tighter than sitagliptin
-8
-9
(
K = 1.27×10 M) and omarigliptin (K = 2.75×10 M) to the DPP-4 enzyme (Fig. S3). The
D D
kinetics studies also demonstrated that compound 5 binds faster than both sitagliptin and
omarigliptin to the DPP-4 enzyme. The fast on-rate of 5 could increase its rebinding to
DPP-4, which is the situation where a ligand, after dissociated from the target, binds to the
target again rather than diffuses away from it.24, 32, 33 Rebinding of 5 could lead to a high local
drug concentration, which is beneficial to decrease the rate of decline of target occupancy,
thus resulting in a prolonged in vivo duration of action.24, 34 More importantly, the
dissociation of bound 5 from the DPP-4 enzyme was 6.5 times slower than sitagliptin,
demonstrating that the residence time (1/k ) of 5 was 6.5 times longer than sitagliptin.
off
Drug-target residence time has been reported to be a favorable parameter for long-lasting in
vivo efficacy.25, 35 Collectively, the fast on-rate and slow off-rate kinetics profiles of 5 would
ensure an extended overall duration of target occupancy, thus leading to a long-acting DPP-4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 52
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 efficacy in vivo (Fig. S4).
Table 1. Binding kinetics profiles of 5
0
1
2
3
4
5
6
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 -1
-1
Compound
k (M s )
on
k (s )
K (M)
D
2.38×10
off
6
7
-3
-9
1
1.44±0.022×10
3.43±0.045×10
-3
-10
-8
5
1.29±0.055×10 2.29 ± 0.078×10 1.77×10
6
-2
sitagliptin
1.18±0.010×10 1.49 ± 0.012×10
1.27×10
2.75×10
5
-4
-9
omarigliptin 2.58±0.022 ×10
7.09±0.12×10
Fig. 2 (a) Kinetics analysis of the binding of 5 to DPP-4. (b) Crystal structure of DPP-4 in
complex with candidate 5. 2Fo-Fc electron density is contoured at 1σ. The electron densities
of both 5 and Arg358 are clear, and their position and pose are unambiguous. (c) Single
crystal X-ray structure of 5, and comparison of the bound, unbound and global minimum
conformations of 5. The three conformations are superimposed well with each other.
Inhibitor 5 binds DPP-4 with Arg358 in a closed-up conformation
In order to verify whether 5 makes close dipole-dipole interactions with residue Arg358
in a closed-up conformation as expected, we solved the X-ray crystal structure of 5 in
complex with DPP-4 (Table S2). The complex crystal structure (Fig. 2b) shows us that 5
binds at the catalytic pocket of the DPP-4 enzyme with its amino group on the dihydropyran
ACS Paragon Plus Environment

Page 15 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ring positioned appropriately to generate charge-reinforced hydrogen-bonding interactions
with the side chains of Glu205, Glu206 and Tyr662 of the recognition site for the peptide
substrate N-terminus. The 2,4,5-trifluorobenzene warhead penetrates into the hydrophobic S1
subsite adjacent to the catalytic triad (Ser630, Asp708, and His740). The naphthalene
fragment of 5 contacts Phe357 with favorable face-to-face π-π stacking interactions. Our
previous work had found that shape matching and electrostatic complementarity are two
crucial factors for DPP-4 inhibitors to accommodate the widely opened S2 extended subsite
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
22
that covered with relatively strong positive electrostatic potential (Fig. S5). For inhibitor 5,
1
at the S2 extended subsite, the -OCH at R could insert into the concave surface formed by
3
residues Glu206, Val207 Ser209 and Arg358 to make favorable van der Waals (vdW)
1
2
interactions. Accompanied by -OCH at R , the electron-withdrawing -CN at R attracts the
3
flexible Arg358 and forms charge-reinforced hydrogen bond with it just as expected.
Moreover, the electron densities of Arg358 are clear, and its position and pose in a closed-up
pattern are unambiguous. Collectively, those binding characteristics made 5 a very
tight-binding DPP-4 inhibitor.
By superimposing the modeled complex structure of 5 in DPP-4 with the X-ray crystal
structure, an RMSD value of 0.246 Å was obtained for the pocket-lining residues with both
Arg358 and 5 aligned well with each other (Fig. S6), suggesting that the FMO-QM based
drug design strategy is reasonable.
Negligible conformational reorganization between unbound and bound state of 5
2
The introduction of a linear and rigid -CN at R of our scaffold not only provides a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 52
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
functional group that generates relatively strong electrostatic interactions with Arg358 but
also makes 5 more rigid by reducing one rotatable bond compared with our -OCH3
substituted lead compound 1. In solution, 5 preferentially assumes the same liner-shaped
conformation (Fig. 2c, Fig. S7, Table S3) as that required for DPP-4 complexation. The
RMSD value between the unbound and bound state of 5 is 0.01 Å, indicating that 5 incurred
almost no conformational reorganization when going from the solution unbound state to the
DPP-4 bound state. Thus, little reorganization free energy penalty is needed to be
counterbalanced by the strong intermolecular interactions of the ligand and receptor, resulting
in a subnanomolar binding affinity of 5. Furthermore, both the unbound and bound
conformations of 5 are very similar to its global minimum conformation (RMSD < 0.16 Å,
Fig. 2c, Table S4), as are their conformational energies (ΔSPE < 0.32 kcal/mol, Table S5).
The negligible conformational strain energy of 5 in its bioactive conformation could greatly
contribute to its tight and stable binding to DPP-4.
0
1
2
3
4
5
6
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
Inhibitor 5 has a long half-life across preclinical species
Compound 5 has a long half-life (> 24h) in both mouse and rat (Table 3). In ICR mice, 5
achieved the maximal plasma concentration slowly (Tmax = 10.20 h), displayed a large
exposure (AUCs), and had a very long terminal half-life of 25.31 h for PO dosing. In SD rats,
5
had a larger exposure (15377.72 vs 3415.00 μg/L*h), wider distribution (2.07 vs 1.66 L/kg)
and slower plasma clearance rate (0.066 vs 0.30 L/h/kg) than 1, leading to a longer terminal
half-life (T ) for iv dosing (11.99 vs 3.87 h). While for the PO administration, T was not
1
/2
1/2
observed within the experimental time period of 24 h, which indicated that 5 can maintain a
ACS Paragon Plus Environment

Page 17 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
high plasma concentration for a long time (> 24 h) in vivo. In addition, the terminal half-life
of 5 is much longer than sitagliptin (1.7 h) and omarigliptin (11 h) in rats with the same doses
for iv and po administration.36, 37 The sustained high plasma concentration of 5 would
undoubtedly contribute to a long duration of action as an inhibitor of plasma DPP-4 in vivo.
The oral bioavailability of 5 is very high (98.54%) in rats, indicating that 5 has good
gastrointestinal permeability.
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 3. Preclinical pharmacokinetic profiles of 5.
Group
Mouse a
Rat b
Administration
Dose (mg/kg)
po
po
iv
2
2
NC
1
c
T (h)
25.31
10.20
1773.37
11.99 ± 1.25
NC
2007.65 ± 217.16
1/2
Tmax (h)
6.67 ± 1.15
1387.47 ±145.99
C /C (μg/L)
0
max
AUC (μg/L*h) 82006.97 23162.26 ± 3041.78 11752.88 ± 962.01
0
-t
AUC0-∞
(μg/L*h)
8
5621.45
NC
15377.72 ± 1776.19
Cl (L/h/kg)
MRT0-∞ (h)
NC
40.68
NC
NC
NC
NC
0.066 ± 0.007
15.87 ± 1.96
2.07 ± 0.18
Vd (L/kg)
ss
a
b
c
Values are mean (n = 4). Values are mean ± SD (n = 3). Not observed within the
experimental time period of 24 h. NC: Not calculated.
Pharmacodynamic profiling in ICR mice, type 2 diabetic ob/ob mice
On the basis of the superior properties of 5 including potent inhibitory potency in vitro,
long residence time and favorable pharmacokinetic characteristics, it was advanced into
further in vivo efficacy study. First, the plasma DPP-4 inhibition study of 5 was primarily
evaluated in ICR mice. As shown in Figure 3a, single dose oral treatment of 5 resulted in a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 52
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
time and dose-dependent inhibition of plasma DPP-4 activity in ICR mice, and 5 (3 mg/kg)
displayed sustained up to 80% inhibitory effect of DPP-4 activity in 168 h (7 days) after
single dose administration, which is superior to the performance of the long-acting control
omarigliptin (3 mg/kg). Those results inspired us to conduct another experiment to further
validate whether 5 can have a sustained bioactivity against DPP-4 in type 2 diabetic ob/ob
mice (Fig. S8). As a result, 5 inhibited the DPP-4 activity in a dose-dependent manner (Fig.
0
1
2
3
4
5
6
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
b), and >80% inhibitory effect of DPP-4 activity was observed through 168 h after single
dose administration, which is better than trelagliptin and omarigliptin.
Fig. 3 Inhibition of DPP-4 activity in serum or plasma obtained from the normal and diabetic
animal. (a) Inhibition of DPP-4 activity in serum obtained from ICR mice after a single oral
ACS Paragon Plus Environment

Page 19 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
administration of 5 and omarigliptin, respectively (n = 10, male). Data are presented as the
mean ± SE. (b) Inhibition of DPP-4 activity in serum obtained from type 2 diabetic ob/ob
mice after a single oral administration of 5, trelagliptin and omarigliptin, respectively (n = 10,
5 male and 5 female). Data are presented as the mean ± SE.
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
Effect of acute administration of 5 on serum glucose, active GLP-1 and insulin levels
Inhibition of DPP-4 augments the level of active GLP-1 and insulin by inhibiting the
degradation of active GLP-1, and it returns glucose homeostasis toward physiological control
levels.38 Considering the favorable in vivo DPP-4 inhibition activity of 5, it was further
advanced to evaluate its ability to increase active GLP-1 levels, insulin levels and improve
the glucose clearance capacity. In ICR mice, 5 administered 60 min before a glucose
challenge markedly reduced blood glucose AUC0-120 min by 16.4% and 27.2% at the dose of
0
.3 and 3 mg/kg (P < 0.05), respectively, which were similar with omarigliptin (14.6% and
22.1%) (Fig. 4a-4b). Oral glucose tolerance test (OGTT) in db/db mice showed that
single-dose oral treatment with 5 (10 mg/kg) significantly improved the glucose tolerance
capacity by decreasing the AUC0-120 min of 18.3% compared with vehicle group (P <0.05), and
the pharmacodynamic effect was equipotent to trelagliptin (ΔAUC0-120 min = 19.3%) but
superior to omarigliptin (ΔAUC0-120 min = 8.3%) (Fig. 4c-4d, Fig. S9). We also evaluated the
GLP-1 and insulin secretion effect of 5. Upon treatment of ICR mice with 5 at 3 mg/kg,
active GLP-1 level significantly increased by 2.0 and 2.6-fold compared with the vehicle
control (P < 0.01) at 5 min and 10 min after the glucose challenge, which were better than
omarigliptin(1.7 and 1.6-fol) (Fig. 4e). Likewise, the insulin levels were also increased by
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 52
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
administration of 5, which is comparable to the efficacy of omarigliptin (Fig. 4f). This
demonstrated that a single oral dose of 5 inhibited serum DPP-4 activity, which increases
active GLP-1 and insulin levels and improves glucose homeostasis in normal mice and
genetic-type 2 diabetic mice.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 21 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Fig. 4 Effects of a single oral administration of 5 on plasma glucose, active GLP-1, and
insulin levels. (a-b) Blood glucose changes in OGTT and AUC of blood glucose between 0
and 120 min in ICR mice (n = 10, male). Compound 5, trelagliptin, omarigliptin or vehicle
was orally administered to 8h-fasted ICR mice 60 min prior to oral glucose challenge (3
g/kg). Data are presented as the mean ± SE. (c-d) Blood glucose changes in OGTT and AUC
of blood glucose between 0 and 120 min in db/db mice (n = 10, 5 male and 5 female).
Compound 5, trelagliptin, omarigliptin or vehicle was orally administered to 6h-fasted db/db
mice 60 min prior to an oral glucose challenge (1.5 g/kg). Data are presented as the mean ±
SE. (e-f) Plasma active GLP-1 and insulin levels following OGTT in ICR mice (n = 10,
male). Compound 5, trelagliptin, omarigliptin or vehicle was orally administered to 8h-fasted
ICR mice 60 min prior to an oral glucose load (3 g/kg). Data are presented as the mean ± SE.
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
(*) P < 0.05 vs vehicle; (**) P < 0.01 vs vehicle. (b.d) One-Way ANOVA with Fisher’s LSD
test for multi-comparisons. (e.f) Two-Way ANOVA with Fisher’s LSD test for
multi-comparisons.
Chronic antidiabetic effects of 5 on type 2 diabetic db/db mice
Encouraged by the results from the acute glucose-lowering activity, long-term
antidiabetic effects of 5 once weekly were evaluated in genetic-type 2 diabetes animal models
(3 and 10 mg/kg, q.w) for 8-weeks. Administration of 5 with the dose of 3 and 10
mg/kg/week caused significant reductions in FBG levels (Fig. 5a-b), and the average
reduction during the entire treatment period was 24.7% and 33.1%, respectively, which were
better than trelagliptin (16.8%) and omarigliptin (26.6%). Chronic administration of 5 (10
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 52
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
mg/kg/week) in db/db mice also significantly improved the glucose tolerance of the diabetic
mice (P < 0.05), which is comparable to omarigliptin and better than trelagliptin (Fig. 5c-d).
Remarkably, an glucose stimulated insulin secretion test revealed that treatment with 5
significantly increased insulin level and raised by 2.6 and 2.2-fold compared to the vehicle
control (P < 0.01) in AUC0-120 min, which are much larger than those of trelagliptin and
omarigliptin at the same dose of 10 mg/kg/week (1.7 and 1.8-fold vs the vehicle control, P <
0
1
2
3
4
5
6
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
0
0
.05) (Fig. 5e-f). Notably, the 10 mg/kg/week 5 performed the best (HbA1c% = 3.56%, P <
.01) when compared with the long-acting controls trelagliptin and omarigliptin (Fig. 5g).
Collectively, the long-term antidiabetic effect of 5 for once-weekly treatment is superior to
both trelagliptin and omarigliptin in db/db mice. Since trelagliptin and omarigliptin were
approved for antidiabetic treatment with a once-weekly dosing regimen, our candidate 5, with
longer and better antidiabetic efficacies in preclinical species than both trelagliptin and
omarigliptin, is likely to be amenable for once-weekly dosing in human.
GLP-1 and its analogs have been demonstrated to reduce the body weight in both
diabetic animal models and clinical studies. Inhibition of DPP-4 activity prevented
3
9
degradation and enhanced the biological activity of active GLP-1. However, the fasting
body weight and postprandial body weight after chronic administration of 5 increased
gradually as those of vehicle, omarigliptin and trelagliptin, demonstrating that 5 (both 3 and
10 mg/kg/week) is well tolerated in db/db mice, and do not cause abnormally weight gain.
These results are similar to the reports of several other studies of DPP-4 inhibitors. The
possible reasons might be due to the less effect of the enhanced endogenous active GLP-1
caused by DPP-4 inhibition rather than that of the injection of exogenous GLP-1 or its
ACS Paragon Plus Environment

Page 23 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
analogs.39 In addition, all animals tolerated exposure to therapeutic concentrations of
compound 5 without any signs of distress or discomfort.
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 52
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
ACS Paragon Plus Environment

Page 25 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Fig. 5 Long-term antidiabetic efficacies of 5 in diabetic db/db mice with 3 and 10
mg/kg/week for 8 weeks (n = 10, 5 male and 5 female). (a) Fasting blood glucose (FBG)
changes, (b) Mean reduction rate of FBG, (c) Blood glucose changes in OGTT, (d) AUC of
blood glucose between 0 and 120 min (the areas under the curve are indicators of glucose
clearance), (e) Plasma insulin levels following OGTT, (f) AUC of insulin between 0 and 60
min (the areas under the curve are indicators of insulin secretion), (g) Plasma HbA1c levels,
and (h) Fasting body weight (FBW). OGTT was conducted after 7 weeks treatment of 5 after
an oral glucose load (1.5 g/kg). Plasma insulin levels after an oral glucose load (1.5 g/kg)
were performed after 8 weeks treatment of 5. Data are presented as the mean ± SE. (*) P <
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
0
.05 vs vehicle; (**) P < 0.01 vs vehicle. (a,e.h) Two-Way ANOVA with Fisher’s LSD test
for multi-comparisons. (d,f,g) One-Way ANOVA with Fisher’s LSD test for
multi-comparisons.
Preclinical safety studies
Blockage of the human potassium channel hERG (human Ether-à-go-go-Related Gene)
can prolong the cardiac action potential and the QT-interval, which is a major concern in drug
discovery and development. The selectivity of 5 against hERG was tested using Patch-clamp
experiments, and the results showed that the IC value of 5 on hERG is larger than 40.00
5
0
μM.
We also conducted preclinical safety studies of 5 in ICR mice. For a single dose oral
toxicity study, 5 was well tolerated at 300 mg/kg with no obvious toxicity. For 14
consecutive days of oral toxicity study, 5 was well tolerated at a dose of 30 mg/kg once daily
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 52
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
based on clinical signs, but no histopathology was performed. Body weight and organ index,
blood parameter examination and serum biochemical metabolites of the ICR mice were
comparable to those of the control group (Table S6-S9).
0
1
2
3
4
5
6
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
Discussion and Conclusions
Our previously discovered inhibitor 1 unexpectedly induced Arg358 to form a closed-up
conformation, which may be one crucial structural determinant for 1’s slow binding kinetics
profile. With the attempt to optimize both the binding kinetics and pharmacokinetic
properties of 1 to get a super long-acting antidiabetic candidate for once-weekly dosing, we
intended to lock Arg358 in a closed-up conformation in the following structural optimization.
The FMO-QM method helped us quantitatively evaluate the specific contribution made by
the two -OCH groups in 1 that attract Arg358, and further guided us to perform structural
3
2
optimization at R . By elaborately introducing a linear 2.4 Å -CN group, which is an
2
electronegative H-bond acceptor, to mimic the role of the -OCH group at R cooperating
3
1
with the other -OCH at R in 1 to attract Arg358, 5 displayed a slightly stronger electronic,
3
charge transfer and dispersion interactions with Arg358 than 1. X-ray crystal structure of 5 in
complex with DPP-4 disclosed that 5 is expectedly inducing Arg358 into the same closed-up
conformation as that in the complex crystal structure of 1. Subsequent SPR assay
demonstrated that 5 binds faster and tighter to DPP-4 than 1 and sitagliptin, while the
disassociating rate is slightly slower than 1 but an order of magnitude slower than sitagliptin.
Pharmacokinetic studies provided that 5 has a larger exposure, wider distribution and slower
clearance rate than 1, thus leading to a longer half-life than 1. Those distinctive binding
ACS Paragon Plus Environment

Page 27 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
kinetics and pharmacokinetic characteristics made 5 a potent and orally available DPP-4
inhibitor with longer and better antidiabetic efficacy than 1 and the once-weekly controls
trelagliptin and omarigliptin in preclinical species, suggesting that compound 5 holds promise
for an efficacious and safe clinical candidate for the once-weekly treatment of T2DM to
improve ease of use and compliance.
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
Both the fast-associating and slow-dissociating binding kinetics profiles and the
pharmacokinetic profiles of 5 contribute to its long-acting antidiabetic efficacy in vivo. The
fast on-rate of 5 can increase its rebinding to DPP-4. Rebinding is helpful to maintain a high
degree of target occupancy. The slow off-rate of 5 can ensure a long drug-target residence
time, which is beneficial for a long duration of action in vivo. The long terminal half-life is
favorable for sustaining a high plasma concentration of 5 to inhibit the soluble DPP-4 enzyme
in the blood. With a large volume of distribution, 5 may also inhibit the membrane-bound
DPP-4 in the tissues, such as liver and small intestine. This may also be conductive to the
long duration of action. Therefore, both the special binding kinetic profiles and
pharmacokinetic profiles are supposed to be main drivers of the long-lasting antidiabetic
efficacy of 5 as an inhibitor of DPP-4 in vivo.
Arg358 is one of the most important residues of the S2 extended subsite, which is
40
flexible in apo and holo structures of the DPP-4 enzyme, and its orientation influences the
4
1
shape and physicochemical properties of the S2 extensive subsite of DPP-4. Although many
DPP-4 inhibitors have been discovered, as far as we know, there is no specific
structure-based drug design work conducted against Arg358 with the aim of fixing its flexible
sidechain in a closed-up conformation. In our case, we successfully locked Arg358 in a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 52
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
2
closed-up conformation to accommodate the substitutions at R and R of our scaffold by
using the FMO method. And this induced conformational change of Arg358 is supposed to be
one structural determinant that contributes to the slow dissociation kinetics of the inhibitors.
The FMO-QM method can be served as a useful tool in guiding structural optimization where
some key residues are preferentially targeted in a ligand-protein complex.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 29 of 52
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Experimental Section
1
. Computational Methods
The FMO-QM method was developed by Kitaura and co-workers, which can accurately
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
quantify the binding interactions of ligand-residue pairs.28-31 It has been successfully applied
in many biological systems to study ligand-protein interactions, in which the ab initio FMO
method clearly outperformed the molecular mechanics (MM) based scoring functions in
predicting binding energies.42-45 Compared with MM based methods, the major advantages of
FMO method is its consideration of dynamic polarization and charge transfer of the
ligand-protein system when computing electrostatic interaction.
Structure Preparation. Molecular Operating Environment (MOE, version 2014.09)
program was used to prepare the complex structure of the ligand and DPP-4 for the FMO
calculation. All water molecules were removed. The missing hydrogen atoms of the crystal
structure of l in complex with DPP-4 (PDB: 5J3J, chain B) were added and the protonation
state of the acidic and basic amino acid residues was predicted using the Protonate 3D
module. Default settings including pH of 7.0, a cutoff of 15 Å for the generalized
Born/volume integral methodology and solvent dielectric constant of 80 were applied. Then,
a structural minimization was conducted to the complex with each atom restrained to deviate
up to 0.5 Å from its original position in the crystal structure by using the semi-empirical
Amber10:EHT forced field that implemented in the MOE program. The refined complex
structure was then used for the subsequent FMO calculation.
Compounds 2-5 were close analogs to 1, so they were likely to exhibit a binding pattern
similar to 1. Therefore, we generated the binding conformations for compounds 2-5 in
ACS Paragon Plus Environment