Design of potent dipeptidyl peptidase IV (DPP-4) inhibitors by employing a strategy to form a salt bridge with Lys554

Article abstract of DOI:10.1016/j.bmcl.2017.05.048

We report a design strategy to obtain potent DPP-4 inhibitors by incorporating salt bridge formation with Lys554 in the S1′ pocket. By applying the strategy to the previously identified templates, quinoline 4 and pyridines 16a, 16b, and 17 have been identified as subnanomolar or nanomolar inhibitors of human DPP-4. Docking studies suggested that a hydrophobic interaction with Tyr547 as well as the salt bridge interaction is important for the extremely high potency. The design strategy would be useful to explore a novel design for DPP-4 inhibitors having a distinct structure with a unique binding mode.

Full text of DOI:10.1016/j.bmcl.2017.05.048

Accepted Manuscript
Design of potent dipeptidyl peptidase IV (DPP-4) inhibitors by employing a
strategy to form a salt bridge with Lys554
Hironobu Maezaki, Michiko Tawada, Tohru Yamashita, Yoshihiro Banno,
Yasufumi Miyamoto, Yoshio Yamamoto, Koji Ikedo, Takuo Kosaka,
Shigetoshi Tsubotani, Akiyoshi Tani, Tomoko Asakawa, Nobuhiro Suzuki,
Satoru Oi
PII:
S0960-894X(17)30529-2
DOI:
http://dx.doi.org/10.1016/j.bmcl.2017.05.048
BMCL 24993
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
1 February 2017
9 May 2017
16 May 2017
Please cite this article as: Maezaki, H., Tawada, M., Yamashita, T., Banno, Y., Miyamoto, Y., Yamamoto, Y., Ikedo,
K., Kosaka, T., Tsubotani, S., Tani, A., Asakawa, T., Suzuki, N., Oi, S., Design of potent dipeptidyl peptidase IV
(DPP-4) inhibitors by employing a strategy to form a salt bridge with Lys554, Bioorganic & Medicinal Chemistry
Letters (2017), doi: http://dx.doi.org/10.1016/j.bmcl.2017.05.048
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Design of potent dipeptidyl peptidase IV (DPP-4) inhibitors
by employing a strategy to form a salt bridge with Lys554
Hironobu Maezakia,*, Michiko Tawadaa, Tohru Yamashitaa, Yoshihiro Bannoa, Yasufumi Miyamotoa, Yoshio
Yamamotoa, Koji Ikedoa, Takuo Kosakaa, Shigetoshi Tsubotania,b, Akiyoshi Tania, Tomoko Asakawaa,
Nobuhiro Suzukia, Satoru Oia,c.
a Pharmaceutical Research Division, Takeda Pharmaceutical Company Limited, 26-1, Muraoka-Higashi 2-
chome, Fujisawa, Kanagawa 251-8555, Japan
* To whom correspondence should be addressed. e-mail: hironobu.maezaki@takeda.com
b Present address: Sumika Chemical Analysis Service, Ltd. 17-85, Jusohomachi 2-chome, Yodogawa-ku,
Osaka 532-8686, Japan
c Present address: Takeda Pharmaceutical Company Limited, 17-85, Jusohomachi 2-chome, Yodogawa-ku,
Osaka 532-8686, Japan
Abstract
1

We report a design strategy to obtain potent DPP-4 inhibitors by incorporating salt bridge formation with
Lys554 in the S1' pocket. By applying the strategy to the previously identified templates, quinoline 4 and
pyridines 16a, 16b, and 17 have been identified as subnanomolar or nanomolar inhibitors of human DPP-4.
Docking studies suggested that a hydrophobic interaction with Tyr547 as well as the salt bridge interaction is
important for the extremely high potency. The design strategy would be useful to explore a novel design for
DPP-4 inhibitors having a distinct structure with a unique binding mode.
Keywords: dipeptidyl peptidase IV (DPP-4); S1' pocket; salt bridge formation; quinoline-based inhibitors;
pyridine-based inhibitors.
2

Diabetes mellitus is a serious chronic disease, and an estimated 422 million adults worldwide are living
with the disease as of 2014. The majority of the adults are affected by type 2 diabetes mellitus (T2DM). This
used to occur mainly among adults, but now occurs in children too.1,2 Recently, glucagon-like peptide-1
(GLP-1)-based therapy has been recognized as one of the most potential therapeutic options to treat this
disorder. GLP-1 is an incretin hormone whose active form (GLP-1[7–36]amide) has multiple antidiabetic
effects represented by glucose-dependent insulin secretion.3,4 Despite the beneficial effects, active GLP-1 is
rapidly inactivated by dipeptidyl peptidase IV (DPP-4, EC 3.4.14.5) known as a predominant enzyme
responsible for degradation of the hormone. Therefore, inhibition of DPP-4 can prevent the degradation and
maximize the antidiabetic effects.5–7 DPP-4 is a ubiquitously distributed serine protease that cleaves dipeptides
from the substrates with L-proline or L-alanine as a penultimate N-terminal residue. A number of DPP-4
inhibitors targeting the active site residues have been investigated clinically or launched as 'gliptins' so far.8–12
Many of them were identified through peptidomimetic approaches and involve interactions with the catalytic
center, i.e. Ser630. We have investigated non-peptidomimetic inhibitors having a distinct structure with a
unique binding mode. Our lead generation strategy for novel DPP-4 inhibitors is as follows: novel leads would
be obtained by introducing an additional interaction with S1' pocket residues into templates possessing a core
heterocycle and substituents that can bind to S1 and S2 pockets adjacent to the catalytic center. In parallel, an
isoquinolone derivative was selected from high-throughput screening hits as a start point that can be applied to
the strategy most appropriately.13 Scaffold hopping based on the cocrystal structure of optimized isoquinolone
1 (Protein Data Bank Code: 3OPM) led to the identification of quinoline-based and pyridine-based templates,
which were proven useful for exploring novel DPP-4 inhibitors and afforded potent inhibitors forming no
direct interactions toward Ser630 (e.g. 2 and 3 in Figure 1).14–16 In the course of these studies, we envisaged
that Lys554 in S1' pocket can be utilized as an unexplored target residue of DPP-4.13,14 Although Ikuma et al.
reported a docking study suggesting the interaction in 2012,17,18 there have been no other reports that target the
useful residue. Therefore, we have investigated the introduction of a salt bridge interaction with Lys554 into
our templates in order to expand the applications. In this letter, we describe the designs, syntheses, and
biological results of new series of DPP-4 inhibitors represented by 4.
3

Figure 1. Representative DPP-4 inhibitors derived from scaffolds
interacting with S1 and S2 pockets.
4

Scheme 1. Synthetic route for quinoline derivatives. Reagents and conditions: (a) halides, K2CO3, DMF; (b) NaOH, THF–
MeOH or THF–EtOH; (c) 3-bromo-1-propanol, K2CO3, DMF; (d) (1) (COCl)2, DMSO, CH2Cl2, −78 °C; (2) 2-methyl-2-
butene, NaClO2, NaH2PO4, THF–t-BuOH–H2O; (e) HCl or TFA; (f) N-phenyl bis(trifluoromethanesulfonimide), NaH, DMF,
0 °C; (g) (1) bis(pinacolato)diboron, Pd(dppf)Cl2∙CH2Cl2, NaOAc, DMSO, 100 °C; (2) 13a or 13b, Pd(PPh3)4, K2CO3,
toluene–MeOH–H2O, reflux.
The starting materials 5 and 6 and the intermediate 7 shown in Schemes 1 and 2 were prepared using the
previously reported procedures.14,15,19 As shown in Scheme 1, quinoline-based carboxylic acids 8a–e were
obtained by alkylation of 5 followed by alkaline hydrolysis of esters of 10a and 10c–e or oxidation of 10b and
subsequent N-Boc deprotection of 11a–e. Suzuki coupling of triflate 12 prepared from 5 with commercially
available halides 13a and 13b provided esters 14a and 14b, respectively. Saponification of 14a and 14b
followed by N-Boc deprotection gave the desired 4 and 9, respectively. Synthesis of pyridine derivatives is
outlined in Scheme 2. Oxidation of pyridin-3-ylmethanol 7 and subsequent Wittig reaction with a
phosphonium salt prepared from ethyl 2-chloro-1,3-thiazole-4-carboxylate (19) afforded ester 20a, which was
5

hydrogenated over palladium on carbon to yield ester 20b. Ether 22 was obtained by mesylation of 7,
6

Scheme 2. Synthetic route for pyridine derivatives. Reagent and conditions: (a) (1) DIBALH, toluene, −78 to 0 °C; (2) Boc2O,
NaOH; (b) (1) Dess–Martin reagent, CH2Cl2, (2) ((4-(ethoxycarbonyl)-1,3-thiazol-2-yl)methyl)(triphenyl)phosphonium
bromide, NaH, DMF; (c) NaOH, THF–EtOH or THF–MeOH; (d) HCl or TFA then HCl; (e) 10% Pd on charcoal, H2, EtOH;
(f) (1) MsCl, NEt3, THF; (2) (4-bromophenyl)methanol, NaH, THF: (g) CO2, Pd(dppf)Cl2∙CH2Cl2, NEt3, MeOH–DMF, 80 °C;
(h) (1) Boc2O, THF; (2) NaOH, MeOH, reflux; (j) halides or mesylate, K2CO3, DMF.
alkylation with 4-bromobenzyl alcohol, and palladium-catalyzed carbonylation using CO2. For the synthesis
of a series of nicotinates 18a–e, the starting nicotinate 6 was hydrolyzed and protected by a Boc group to yield
7

23, which was re-esterified to afford 24a, 24b, 24d, and 24e. For 24c, further esterification of the obtained 4-
bromobenzyl ester was performed by palladium-catalyzed carbonylation using CO2. The esters 20a, 20b, 22,
and 24a–e were converted to the desired acids 18a–e by alkaline hydrolysis and subsequent N-Boc
deprotection. Carboxamides 28a–c corresponding to carboxylic acids 11d, 15a, and 25b, respectively, were
prepared using a general coupling procedure followed by acidic N-Boc deprotection.
The synthesized compounds were evaluated for their in vitro inhibition against human DPP-4 using Gly-
Pro-p-NA as a substrate. IC50 results analyzed with sigmoidal dose response curves are outlined in Tables 1
and 2. To seek DPP-4 inhibitors that can form a salt bridge with Lys554, quinolines with a terminal carboxy
group in the 6-substituent were evaluated (8a–e in Table 1). The results indicated that the activity increased as
linker length elongated. The optimal 8d (n = 4) exhibited 30-fold more potent inhibition than 8a (n = 1).
Based on the reported X-ray cocrystal structure of isoquinolone-based inhibitor 1 bound to DPP-4,13 the
distance from the oxygen atom at the 6-position of isoquinolone 1 to the terminal amino group of Lys554 is
around 6.6 Å. Assuming that quinolines 8 take the similar binding mode to that of isoquinolone 1, the optimal
chain length in quinolines 8 would be that of 8a (n = 1) in an extended form. Therefore, the linker length of
the most potent analogue 8d (n = 4) seemed too long to allow the carboxy group to interact with Lys554. To
understand the potent DPP-4 inhibition of 8d, docking study was performed based on the cocrystal structure
of 1. The study provided multiple conformations and the most appropriate binding mode is displayed in Figure
2. In the model, 8d could adopt an essentially similar binding mode to the previously reported for other
quinoline derivatives.14 The result suggested that the relatively long linker
Table 1. SAR Summary of Linkers (L) in Quinoline-based Carboxylic Acids
Compound
L
DPP-4 IC50 (nM)a
140
8a
n = 1
8b
8c
2
3
29
17
8

4
5
4.6
6.8
8d
8e
28
9
4
0.38
1b
3b
220
22
a Inhibitory activity against human DPP-4 was measured using Gly-Pro-pNA as a substrate. b Free bases of 1
and 3 were used as positive controls.
Figure 2. Docking study between compound 8d and DPP-4 protein. (A) Lys554 (pink), Tyr547 (yellow), and other amino acids
(gray) in the binding site of 8d (green carbons). Major interactions are depicted as red dotted lines. Distances in angstroms. (B)
Superimposition of 8d (magenta) onto X-ray cocrystal structure of 1 (cyan).
9

Figure 3. Proposed pharmacophore using quinoline-based or pyridine-
based scaffold involves hydrophobic interaction with Tyr547 to
a
efficiently form a salt bridge with Lys554.
was packed within the space of S1' pocket by curving the chain. It was anticipated that flat aromatic ring
linkers instead of the long linear linkers would enable the carboxy group to locate at the suitable position for
Lys554 by the support of the interaction with Tyr547. Therefore, compounds incorporating the aromatic ring
linker moiety were hypothesized to achieve highly efficient binding to the enzyme. Based on the hypothesis,
we constructed a new pharmacophore using the quinoline and pyridine cores substituted with 3-aminomethyl,
2-isobutyl, and 4-p-tolyl groups as templates (Figure 3).
First, the ring linker was introduced to the 6-position of the quinoline template. On the basis of the above-
mentioned results of docking studies, five-membered heteroaromatic rings were selected as a linker at the 6-
position to orient the carboxy group toward the amino side chain of Lys554. 2-Furoic acid 9 and thiazole-4-
carboxylic acid 4 were evaluated for the DPP-4 inhibitory activity. As shown in Table 1, compounds 9 and 4
were found to be potent inhibitors. Remarkably, 4 exhibited highly potent subnanomolar inhibitory activity
(IC50 = 0.38 nM), which is the highest potency of this series.
For further simplification, we applied the pharmacophore model to the pyridine-based template. The high
inhibitory activity of quinoline 4 was retained by ring-opened analogues 16a and 16b, which were identified
to be nanomolar inhibitors (Table 2). In contrast, a lack of the heteroaryl linker resulted in a dramatic decrease
in potency (18a). By replacing the alkyl linker of 18a with a benzene ring, benzyl esters 18c, 18d, and 18e
showed 5- to 30-fold more potent inhibitory activity. Replacement of the linker with a cyclohexane ring also
resulted in ca. 6-fold potency increase (18b) when compared with 18a. The activity was slightly emphasized
by increasing flexibility of the linker in 18e, and ether 17 demonstrated a nanomolar inhibition. These results
indicate that general hydrophobic rings, such as an aryl and cycloalkyl, could be incorporated as a linker to
enhance the inhibitory activity of carboxylic acid derivatives as well as a thiazolyl linker.
10

Table 2. SAR Summary of Linkers (L1 and L2) in Pyridine-based Carboxylic Acids
Compound
L1
L2
DPP-4 IC50 (nM)a
5.5
16a
(E)-CHCH
16b
18a
18b
CH2CH2
CO2CH2
CO2CH2
2.6
470
80
CH2CH2
18c
CO2CH2
ortho
29
18d
18e
CO2CH2
CO2CH2
meta
para
87
17
CH2OCH2
7.2
17
a Inhibitory activity against human DPP-4 was measured using Gly-Pro-pNA as a substrate.
To understand the binding modes of these series of quinoline- and pyridine-based carboxylic acids, the
corresponding carboxamides were evaluated for DPP-4 inhibitory activity (Table 3). As a result, all the
representative carboxamides 28a–c exhibited potent inhibitory activity, which indicates that polar substituents
11

are favorable as the C-terminal moiety. Interestingly, these compounds tend to show slightly less potent
inhibitory activity than that of the corresponding carboxylic acids (ca. 3 to 5-fold). The results suggest that the
salt bridge formation would be involved in the stronger interaction between the enzyme and the inhibitors
targeting the S1' pocket.
Table 3. DPP-4 Inhibitory Activity of Carboxamides 28
Compound
L
Additive
none
DPP-4 IC50 (nM)a
13
28a
2HCl
2.1b
28b
28c
2HCl
8.5
a Inhibitory activity against human DPP-4 was measured using Gly-Pro-pNA as a substrate. b IC50 value of the
corresponding methyl ketone (28d) was 8.7 nM.
Next, compounds were docked into the DPP-4 model. Figure 4 displays the model of 4 as a representative.
Their core rings showed an excellent overlay with that of isoquinolone 1, predicting a very limited potency
loss in each interaction between the 2-, 3-, and 4-substituents and their target residues. The thiazole ring of 4
would be positioned appropriately for a π–π stacking interaction with the benzene ring of Tyr547 (ca. 4 Å
distance in 4).21 The stacking ring linkers would consequently make the carbonyl oxygen oriented to the
amino nitrogen of Lys554 in a suitable distance for a salt bridge interaction (2.8 Å). Interestingly, the docking
model suggested that 4 would make no direct interaction with the catalytic serine (Ser630). Taken together
with the SAR results, it is strongly suggested that 4 would exhibit the highly potent subnanomolar inhibition
12

by employing the hydrophobic interaction with Tyr547 as well as the salt bridge interaction with Lys554
without the interaction with the catalytic residue.
Figure 4. Docking study of compound 4 with DPP-4 protein. (A) Lys554 (pink), Tyr547 (yellow), and other amino acids (gray) in
the binding site of 4 (green carbons). Major interactions are depicted as red dotted lines. Distances in angstroms. (B)
Superimposition of 4 (magenta) onto X-ray cocrystal structure of 1 (cyan).
In addition to the promising DPP-4 inhibition profiles, reduced DPP-8 inhibition was consistently found in
analogues targeting Lys554. It is assumed that the isozyme selectivity is induced by the sequence differences
between the two enzymes in S2 pocket (Phe630 in DPP-4 vs His in DPP-8) and in S1' pocket (Lys554 in DPP-
4 vs Leu in DPP-8). Our templates utilize Phe357 to make an indispensable hydrophobic interaction with the
isobutyl group and generally lead to high selectivity against the isozyme. It is consistent that pyridine-based
inhibitor 315 shows very low DPP-8 inhibitory activity with an IC50 value of 11 μM (500-fold selective for
DPP-4) most likely by the unfavorable hydrophobicity of the 2-isobutyl group.22 By adding a polar substituent
that reaches Lys554 onto the 6-carboxy group of 3, compound 18e showed further potency reduction for DPP-
8 (IC50 > 100 μM). A similar result has been reported in quinoline 2, which can make a hydrogen bonding to
Lys554.14 The potency losses in the analogues targeting Lys554 could be explained by the other sequence
difference in S1' pocket, and the introduced hydrophilic substituent (e.g. carboxy and carbonyl groups) would
be unfavorable under the hydrophobic environment in the S1' pocket of DPP-8. Thus, the proposed
pharmacophore would also provide an attractive feature of higher selectivity against DPP-8.
Representative pharmacological profiles are shown in Figures 5 and 6. An oral dose of 1 mg/kg of 4
induced >50% inhibition of plasma DPP-4 activity during 6 h in an ex vivo test in Sprague Dawley rats
13

(Figure 5). In an oral glucose tolerance test in female Wistar fatty rats, 4 significantly reduced glucose
excursion and enhanced insulin secretion after oral glucose load at 1 mg/kg (Figure 6). These ex vivo and in
vivo results clearly demonstrate the potential for our design strategy to yield a new lead with in vivo efficacy.
Figure 5. Ex vivo study of compound 4 in Sprague-Dawley rats. Data points and error bars indicate means and SD, respectively, of the residual
DPP-4 activity in rat plasma after administration of 1 mg/kg of compound (n = 5). * p < 0.05, ** p < 0.001 versus control by Dunnett test.
Compound 1 was used as a positive control for the experiment.
Figure 6. Oral glucose tolerance test of compound 4 in female Wistar fatty rats. Plasma glucose (A) and plasma insulin levels (B) of rats administered 1 mg/kg
of compound or 0.5% MC 60 min before oral glucose load (1 g/kg). Data points indicate means, and the error bars indicate SD (n = 6). * p < 0.05, ** p < 0.01
versus control by Dunnett test. LAF-237 was used as a positive control for the experiment.
In summary, we have designed a series of carboxylic acids incorporated in quinoline-based and pyridine-
based templates of DPP-4 inhibitors. In the design, the terminal carboxy groups are appropriately directed
toward Lys554 in the S1' pocket to form a salt bridge, suggesting the importance of the hydrophobic
interaction between the aromatic ring linkers and Tyr547. The representative quinoline (4) and pyridines (16a,
16b, and 17) demonstrated subnanomolar to nanomolar inhibition. Interestingly, the docking studies suggested
that these potent DPP-4 inhibitors have no direct interaction with the active site residue Ser630 in contrast to
14

many of the reported inhibitors. Furthermore, the interaction with Lys554 can contribute to the isozyme
selectivity against DPP-8. Taken together, the design strategy using our templates would be useful to explore a
novel design for DPP-4 inhibitors having a distinct structure with a unique binding mode
Acknowledgements
The authors would like to thank Nobuo Cho for critical reading of the manuscript and for useful
suggestions to improve it. The authors would also like to thank Harumi Hattori and Yusuke Kamada for a
statistical analysis of biological data and Koji Takeuchi, Yu Momose, Yuji Ishihara, Shigenori Ohkawa,
Takashi Sohda, and Akio Miyake for helpful discussions throughout the study.
A. Supplementary data
Supplementary data associate with this article can be found, in the online version, at http://....
References and Notes
1.
2.
3.
4.
5.
6.
7.
WHO. Global Report on Diabetes. World Health Organization Library Cataloguing; 2016.
NCD Risk Factor Collaboration. Lancet. 2016;387:1513.
Edwards CM. Ann Med. 2005;37:314.
Salvatore T, Carbonara O, Cozzolino D, Torella R, Sasso FC. Curr Diabetes Rev. 2007;3:15.
Mentlein R, Gallwitz B, Schmidt WE. Eur J Biochem. 1993;214:829.
Drucker DJ. Expert Opin Investig Drugs. 2003;12:87.
Crepaldi G, Carruba M, Comaschi M, Del Prato S, Frajese G, Paolisso G. J Endocrinol Invest.
2007;30:610.
15

8.
Scheen AJ. Expert Opin Pharmacother. 2015;16:43.
9.
Kaku K. Expert Opin Pharmacother. 2015;16:2539.
10.
11.
12.
Nabeno M, Akahoshi F, Kishida H, et al. Biochem Biophys Res Commun. 2013;434:191.
Watanabe YS, Yasuda Y, Kojima Y, et al. J Enzyme Inhib Med Chem. 2015;30:981.
Arulmozhiraja S, Matsuo N, Ishitsubo E, Okazaki S, Shimano H, Tokiwa H. PLoS One.
2016;11:e0166275.
13.
14.
15.
16.
17.
18.
19.
20.
21.
Banno Y, Miyamoto Y, Sasaki M, et al. Bioorg Med Chem. 2011;19:4953.
Maezaki H, Banno Y, Miyamoto Y, et al. Bioorg Med Chem. 2011;19:4482.
Miyamoto Y, Banno Y, Yamashita T, et al. J Med Chem. 2011;54:831.
See supplementary SAR information Table S1.
Ikuma Y, Hochigai H, Kimura H, et al. Bioorg Med Chem. 2012;20:5864.
Ikuma Y, Hochigai H, Kimura H, et al. Bioorg Med Chem. 2015;23:779.
Miyamoto Y, Banno Y, Yamashita T, et al. Bioorg Med Chem. 2011;19:172.
Tsuzuki S, Honda K, Uchimaru T, Mikami M, Tanabe K. J Am Chem Soc. 2002;124:104.
Rummey C, Metz G. Proteins. 2007;66:160.
16

Graphical Abstract
Design of potent dipeptidyl peptidase IV
(DPP-4) inhibitors by employing a strategy
to form a salt bridge with Lys554
Leave this area blank for abstract info.
Hironobu Maezaki*, Michiko Tawada, Tohru Yamashita, Yoshihiro Banno, Yasufumi Miyamoto,
Yoshio Yamamoto, Koji Ikedo, Takuo Kosaka, Shigetoshi Tsubotani, Akiyoshi Tani, Tomoko Asakawa,
Nobuhiro Suzuki, and Satoru Oi
17