Molecular dynamics study-guided identification of cyclic amine structures as novel hydrophobic tail components of hPPARγ agonists

Article abstract of DOI:10.1016/j.bmcl.2014.06.023

We previously reported that a α-benzylphenylpropanoic acid-type hPPARγ-selective agonist with a piperidine ring as the hydrophobic tail part (3) exhibited sub-micromolar-order hPPARγ agonistic activity. In order to enhance the activity, we planned to carr

Full text of DOI:10.1016/j.bmcl.2014.06.023

Bioorganic & Medicinal Chemistry Letters 24 (2014) 4001–4005
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Molecular dynamics study-guided identification of cyclic amine
structures as novel hydrophobic tail components of hPPARc agonists
Yuta Tanaka a, , Kanae Gamo a, , Takuji Oyama b, Masao Ohashi a, Minoru Waki a, Kenji Matsuno a,
Nobuyasu Matsuura c, Hiroaki Tokiwa d, Hiroyuki Miyachi a,
⇑
a Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, 1-1-1, Tsushima-Naka, Kita-ku, Okayama 700-8530, Japan
b Department of Biotechnology, Faculty of Life and Environmental Sciences, University of Yamanashi, 4-3-37 Takeda, Kofu City, Yamanashi 400-8510, Japan
c Department of Life Science, Faculty of Science, Okayama University of Science, 1-1 Ridai-cho, Kita-ku, Okayama 700-0005, Japan
d Department of Chemistry, Rikkyo University, Nishi-Ikebukuro, Toshimaku, Tokyo 171-8501, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
We previously reported that a
piperidine ring as the hydrophobic tail part (3) exhibited sub-micromolar-order hPPAR
a
-benzylphenylpropanoic acid-type hPPAR
c
-selective agonist with a
agonistic activ-
ity. In order to enhance the activity, we planned to carry out structural development based on informa-
tion obtained from the X-ray crystal structure of hPPAR ligand binding domain (LBD) complexed with 3.
Received 17 May 2014
Revised 5 June 2014
Accepted 9 June 2014
Available online 24 June 2014
c
c
However, the shape and/or nature of the binding pocket surrounding the piperidine ring of 3 could not be
precisely delineated because the structure of the omega loop of the LBD was poorly defined. Therefore,
we constructed and inserted a plausible omega loop by means of molecular dynamics simulation. We
then used the reconstructed LBD structure to design new mono-, bi- and tricyclic amine-bearing com-
pounds that might be expected to show greater binding affinity for the LBD. Here, we describe synthesis
Keywords:
Peroxisome proliferator-activated receptor
gamma
Cyclic amine
and evaluation of
sized compounds exhibited more potent hPPAR
than 3. Some of these compounds also showed comparable aqueous solubility to 3.
a
-benzylphenylpropanoic acid derivatives 8. As expected, most of the newly synthe-
Molecular dynamics simulation
c
agonistic activity and greater hPPAR binding affinity
c
Ó 2014 Elsevier Ltd. All rights reserved.
Structural biology data obtained by X-ray crystallographic anal-
ysis of target protein(s) complexed with specific ligand(s) are a
powerful tool for medicinal chemists, offering insight into the
shape of the ligand-binding pocket, critical amino acids for ligand
binding, specific interaction modes of ligands, determinants of
potency and/or selectivity, and so on. Such information provides
a basis for medicinal chemists to design structural frameworks
for the creation of new lead compounds, and to identify appropri-
ate functional groups for obtaining high potency and/or selectivity.
In some cases however, probably due to molecular flexibility or
thermodynamic instability, sufficient information about parts of
the structure may be unavailable from crystallographic study
(Fig. 2A–C).1–3 This proved to be the case for the omega loop part
in the crystal structure of the ligand binding domain (LBD) of
human peroxisome proliferator-activated receptor gamma
of molecular dynamics (MD) calculations. We also describe the
discovery of novel active hydrophobic tail structures for our
phenylpropanoic acid-type hPPARc agonists based on the MD-
developed hPPAR model structure.
c
We have been engaged in structural development studies of
various synthetic hPPARs ligands,4–7 and one of our current focuses
is creation of hPPAR
and selective hPPAR
in vivo use. We recently showed that a
acid-type hPPAR -selective agonist with a piperidine ring as a
c
c
agonists and hPPAR
affinity and sufficient aqueous solubility for
-benzylphenylpropanoic
c antagonists with potent
a
c
hydrophobic tail part (3) exhibited sub-micromolar-order hPPAR
c
agonistic activity with moderate aqueous solubility (Fig. 1).
In order to improve the activity, we planned to modify 3 based
on the X-ray crystal structure of hPPAR LBD complexed with 3
c
(PDB: 3VSP). However, we found that the structure of some of
the amino acids involved in part of the binding pocket surrounding
the piperidine ring of 3 was insufficiently well defined (Fig. 2D),
because the structure of the omega loop part8 (265Lys-His-Ile-
Thr-Pro-Leu-Gln-Glu-Gln-Ser-275Lys) of the binding pocket was
unclear. This is not a special case, because all the hPPAR subtype
LBD-ligand complexes examined by our group lack a well-defined
omega loop structure, presumably due to molecular flexibility or
(hPPARc) complexed with the selective agonist 3. Here, we
describe construction of a plausible structure of the disordered
peptide fragment, that is, the omega loop part of hPPARc, by means
⇑
Corresponding author. Tel.: +81 086 251 7930.
E-mail address: miyachi@pharm.okayama-u.ac.jp (H. Miyachi).
These authors contributed equally.
http://dx.doi.org/10.1016/j.bmcl.2014.06.023
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

4002
Y. Tanaka et al. / Bioorg. Med. Chem. Lett. 24 (2014) 4001–4005
O
O
CO2H
CO2H
N
H
N
H
H
H
N
O
O
N
1:MEKT-1
2:MEKT-21
O
CO2H
N
H
H
N
O
3:MEKT-28
Figure 1. Some representative structures of our previously developed PPAR
c
agonists. Compound 1: MEKT-1. Compound 2: MEKT-21. Compound 3: MEKT-28.
A
B
C
D
E
F
G
H
I
Phe264
His266
Glu272
Gly284
Gln283
Arg280
Figure 2. (A)–(D) Crystal structures of hPPAR
and hPPAR LBD–3 complex (D) (PDB: 3VSP). (E) Whole structure of the omega-loop-inserted hPPAR
structure is depicted as a green ribbon-model. (F) Zoomed view of the reconstructed omega loop structure of hPPAR
are depicted as blue cylinder models. (G) The surface structure of the binding pocket surrounding the piperidine ring of 3 in the hPPAR
structure of the binding pocket surrounding the piperidine ring of 3 in the reconstructed, omega-loop-inserted hPPAR LBD–3 complex model. (I) Zoomed view of the
piperidine-interacting amino acids in the omega loop-inserted hPPAR LBD–3 complex model.
c
LBD–TIPP-703 complex (A) (PDB: 2ZNO), hPPAR
c
LBD–1 complex (B) (PDB: 3AN4), hPPAR
LBD–3 complex model. The main chain of the inserted omega loop
LBD. The amino acids of the reconstructed omega loop
LBD–3 complex. (H) The surface
c LBD–2 complex (C) (PDB: 3VSO)
c
c
c
c
c
c
thermodynamic instability, as illustrated in Figure 2A–C. Therefore,
we decided to construct and insert a plausible omega loop part by
using molecular dynamics (MD) simulation. MD-assisted construc-
tion of the omega loop part involved three steps, as follows. (1) We
used the molecular operating system (MOE) program suite to
insert the eleven amino acids of the omega loop, Lys-His-Ile-Thr-
Pro-Leu-Gln-Glu-Gln-Ser-Lys, between the end of the H20 helix
and the beginning of the H3 helix in the PDB: 3VSP protein struc-
ture based on the homology alignment method with the use of the
template structure (PDB ID: 3VSP) to construct the initial geometry
of the omega loop of hPPAR
c
. (2) Next, compound 3 was docked
into this initial geometry of hPPAR
c
by the ASEDock9 procedure

Y. Tanaka et al. / Bioorg. Med. Chem. Lett. 24 (2014) 4001–4005
4003
O
O
R1
R2
a
b
c
CF3
OH
NH
R1
R1
N
R2
N
R2
4a-j
5a-j
6a-j
O
O
O
O
O
N
OH
d
O
N
H
N
H
R1
R1
N
R2
O
N
R2
O
Bn
7a-j
8a-j
N
N
N
N
N
N
N
N
N
N
a,
b,
c,
d,
e,
f,
g,
h,
I,
j
Scheme 1. Synthetic route to the present series of compounds. Reagents and conditions: (a) 2,2,2,40-tetrafluoroacetophenone, TEA, DMSO, 100 °C, 16 h; (b) NaOH, DMF/H2O,
100 °C, 2 h, 43–79% (2 steps); (c) (S)-3-((R)-2-(3-(aminomethyl)-4-propoxybenzyl)-3-phenylpropanoyl)-4-benzyloxazolidin-2-one hydrochloride, HATU, DIEA, DMF, 16 h,
55–95%; (d) LiOH-H2O, 30% H2O2, THF/H2O, 3.5 h, 46–94%.
Table 1
Values of hPPARs transactivation activity (EC50), binding affinity (KD) and aqueous solubility of the present series of derivatives
K
D (nM)
3
R² = 0.9747
O
2.5
2
CO2H
N
H
H
1.5
1
R1
O
N
R2
0.5
0
0
10
20
30
40
50
60
EC㧡㧜 (nM)
a
b
a
b
Compd
R1R2N-
EC50 (nM)
KD (nM)
Sol.pH7.4c
10
(lg/mL)
Compd
R1R2N-
EC50 (nM)
2.2
KD (nM)
Sol.pH7.4c
<10
(lg/mL)
N
N
3
52.7
2.85
8f
0.40
0.64
N
N
N
8a
8b
6.7
1.9
n.t
n.t
8g
8h
6.7
5.3
10
N
0.1
<10
0.41
0.80
<10
N
N
N
8c
38.9
1.82
n.t
8i
11.7
10
N
8d
8e
7.0
n.t
n.t
<10
<10
8j
1
2.2
3.6
0.11
N.T.
<10
<1
N
10.4
a
b
c
EC50 values represent half-maximal effective concentration (nM). These data were taken from the two independent duplicated experiments.
KD values were measured using the surface plasmon resonance (SPR) method (Biacore X100). These data were taken from the two independent duplicated experiments.
Solubility was measured using the 48 h shaking flask method. 1 mg of compound was ground in an agate mortar and taken up in 1.0 mL of an equal volume of a mixture of
phosphate buffer (pH 7.2–7.4) and EtOH. The suspension was shaken for 48 h at 25 °C. An aliquot was filtered through a Minisart RC 15 (0.45 lm). The filtrate was diluted in
CH3CN and subjected to HPLC with UV detection at 254 nm. The concentration of the sample solution was calculated using a previously determined calibration curve,
corrected for the dilution factor of the sample.
incorporated in the MOE program suite to obtain the structure of
the hPPAR complex with compound 3, and the complex structure
having the top docking score was adopted. (3) Thermal relaxation
of the complex was performed by means of isothermal (NVT) MD
simulations under standard conditions. The temperature and pres-
sure of the system were controlled by Nos’e-Poincar’e-Andersen
c

4004
Y. Tanaka et al. / Bioorg. Med. Chem. Lett. 24 (2014) 4001–4005
O
O
CO 2H
CO 2
H
N
H
N
H
H
H
(3)
(8b)
Phe
A
C
B
N
O
N
O
Gly
Arg
Gln
(8f)
(8g)
D
Phe
Glu
Gly
Leu
Arg
Arg
Gln
Gly
Gln
Figure 3. (A)–(D) Zoomed view of the molecular modeling structures of 3, 8b, 8f and 8g complexed with the omega-loop-inserted hPPARc LBD. The amino acids of the
binding cavity are depicted as magenta cylinder models, and the structures of the ligands are depicted as cyan cylinder models.
formalism (details will be reported elsewhere). The structure of the
omega loop-reconstructed hPPAR LBD complexed with 3 is
As expected, the introduction of a dimethyl group at the 3- and
4-positions of the piperidine ring of 3 was also effective in increas-
c
depicted in Figure 2E, and a magnified view of the reconstructed
omega loop structure is shown in Figure 2F. The surface structures
of the binding pocket surrounding the piperidine ring of 3 in the
ing hPPARc agonistic activity. The 3,3-dimethylpiperidine- and
4,4-dimethylpiperidine derivatives (8d, 8e) exhibited superior
activity, and the potency of 3,3,5,5-tetramethylpiperidine deriva-
tive (8f) was more than 10 times greater than that of 3.
Based on these results, we next synthesized bicyclic and tricy-
clic amine structures (8g–8j). All these compounds exhibited
hPPAR
hPPAR
c
c
LBD–3 complex and in the omega loop-reconstructed
LBD–3 complex model are depicted in Figure 2G and H,
respectively.
The binding pocket surrounding the piperidine ring is some-
what narrower in the MD-assisted binding model, as compared
with the model obtained from the X-ray crystal structure. How-
ever, there is still remaining hydrophobic space, especially around
the 2,3,5,6 positions of the piperidine ring of 3. Therefore, we spec-
ulated that compounds bearing ring-expanded, bicyclic, and tricy-
greater hPPAR
pounds, 8b, 8f and 8j, exhibited potent hPPAR
comparable to that of a previously reported -benzylphenylpropa-
c
agonistic activity than 3. Representative com-
c
agonistic activity,
a
noic acid-type hPPARc-selective agonist with a bulky adamantyl
group as the hydrophobic tail part (1).1
To understand the structure–activity relationship of the present
series of compounds, we constructed binding models of represen-
tative compounds (8b, 8f and 8g) complexed with the omega loop-
clic amine derivatives, 8a–j, might exhibit superior hPPAR
c
agonistic activity.
To test this idea, we synthesized the series of compounds illus-
trated in Scheme 1. The appropriate amine10 was treated with
2,2,2,40-tetrafluoroacetophenone in the presence of triethylamine
in DMSO to afford the amino group-substituted 2,2,2-trifluoroace-
tophenone, and then alkaline hydrolysis11 afforded the key inter-
mediate benzoic acids 6a–j. Compounds 6a–j were condensed
with (S)-3-((R)-2-(3-(aminomethyl)-4-propoxybenzyl)-3-phenyl-
propanoyl)-4-benzyloxazolidin-2-one hydrochloride,3 followed by
LiOOH hydrolysis12 to afford the desired compounds 8a–j.
First of all, we focused on the ring size-expanded series, 8a–8c.
As shown in Table 1, the ring size of the cyclic amino group corre-
lated with hPPARc agonistic activity, which increased in the ring-
size order of 6 (3) < 7 (8a) < 8 (8b). The 8-membered cyclic amino
group (8b) might be optimal, because the 9-membered cyclic
reconstructed hPPARc LBD, using the computational ligand-dock-
ing program ASEdock incorporated in the MOE program suite.
We calculated the interaction energy of each conformer with the
omega loop-reconstructed hPPAR
c LBD, and the top energy-
minimized conformer was selected in each case. The results are
summarized in Figure 3A–D, in addition to that for the omega
loop-reconstructed hPPARc LBD–3 complex model.
As described above, the piperidine ring of 3 is hosted in the
hydrophobic binding cavity composed of the side chains of amino
acid residues Glu259, Phe264, His266, Glu272, Arg280, Ile281,
Gln283 and Gly284 (Fig. 3A). Expansion of the ring size of the
hydrophobic tail part amino moiety from 6 (3) to 8 (3b) enhanced
the hydrophobic interaction, especially with the side chains of
Phe264, Arg280, Gln283 and Gly284 (Fig. 3B). On the other hand,
in the case of 3c, both methylenes adjacent to the nitrogen atom
of the cyclononyl amino moiety showed short contacts with the
side chains of Phe264 and Gly284. The introduction of tetramethyl
amino group (8c) exhibited decreased hPPAR
c
agonistic activity,
although it still possessed potent hPPARc agonistic activity, com-
parable to that of 3.

Y. Tanaka et al. / Bioorg. Med. Chem. Lett. 24 (2014) 4001–4005
4005
groups at the 3,3,5,5 positions of the piperidine ring of 3 enhanced
the hydrophobic interactions with Phe264, Glu272, Arg280,
Gln283 and Gly284 (Fig. 3C). The formation of a bicyclic ring sys-
tem also enhanced the hydrophobic interactions with the side
chains of Arg280, Gln283 and Gly284 (Fig. 3D). All these data indi-
cated that the introduction of appropriate hydrophobic substitu-
ents at suitable positions on the piperidine ring of 3 effectively
enhances the hydrophobic interactions with the binding cavity of
Informatics, and Structural Life Science from the Ministry of Educa-
tion, Culture, Sports, Science and Technology, Japan.
References and notes
1. Kasuga, J.; Yamasaki, D.; Ogura, K.; Shimizu, M.; Sato, M.; Makishima, M.; Doi,
T.; Hashimoto, Y.; Miyachi, H. Bioorg. Med. Chem. Lett. 2008, 18, 1110.
2. Ohashi, M.; Oyama, T.; Nakagome, I.; Sato, M.; Nishio, Y.; Nobusada, H.; Hirono,
S.; Morikawa, K.; Hashimoto, Y.; Miyachi, H. J. Med. Chem. 2011, 54, 331.
3. Ohashi, M.; Oyama, T.; Putranto, E. W.; Waku, T.; Nobusada, H.; Kataoka, K.;
Matsuno, K.; Yashiro, M.; Morikawa, K.; Huh, N. H.; Miyachi, H. Bioorg. Med.
Chem. 2013, 21, 2319.
4. Ban, S.; Kasuga, J.; Nakagome, I.; Nobusada, H.; Takayama, F.; Hirono, S.;
Kawasaki, H.; Hashimoto, Y.; Miyachi, H. Bioorg. Med. Chem. 2011, 19, 3183.
5. Kasuga, J.; Yamasaki, D.; Araya, Y.; Nakagawa, A.; Makishima, M.; Doi, T.;
Hashimoto, Y.; Miyachi, H. Bioorg. Med. Chem. 2006, 14, 8405.
6. Kasuga, J.; Nakagome, I.; Aoyama, A.; Sako, K.; Ishizawa, M.; Ogura, M.;
Makishima, M.; Hirono, S.; Hashimoto, Y.; Miyachi, H. Bioorg. Med. Chem. 2007,
15, 5177.
7. Kasuga, J.; Oyama, T.; Nakagome, I.; Makishima, M.; Hirono, S.; Morikawa, K.;
Hashimoto, Y.; Miyachi, H. ChemMedChem 2008, 3, 1662.
the omega loop-reconstructed hPPAR
In order to confirm that the increase in hPPAR
ity is due to increased binding affinity for hPPAR
c LBD.
c
c
agonistic activ-
owing to intro-
duction of the additional alkyl groups, we then evaluated the direct
binding affinity (KD) of representative compounds to hPPAR
13 (we
used the nitrilotriacetic acid (NTA) sensor chip method to avoid
non-specific binding of the ligand to hPPAR LBD). As shown in
c
c
Table 1, compound 3 exhibited nanomolar-order KD value, and rep-
resentative compounds of the present series, 8b, 8f, 8g, 8h, 8i and
8j, all exhibited sub-nanomolar-order KD values, except 8c. The
8. Waku, T.; Shiraki, T.; Oyama, T.; Maebara, K.; Nakamori, R.; Morikawa, K. EMBO
J. 2010, 29, 3395.
in vitro hPPAR
the binding affinity (KD) (r2 = 0.97), suggesting that the increase
in hPPAR agonistic activity of the present series of compounds
c agonistic activity (EC50) was well correlated with
9. Goto, J.; Kataoka, R.; Muta, H.; Hirayama, N. J. Chem. Inf. Model. 2008, 48, 583.
10. (a) (a) 3,3,5,5-Tetramethylpiperidine: Mizuno, M.; Manabe, A., WO
2004099160 (A1).; (b) (b) 8-Azabicyclo[3,2,1]octane hydrochloride: Mueller,
R.; Street, L., WO 2009/023126 (A2).; 8-Azabicyclo[3,3,1]nonane
hydrochloride: (c) Markus, B. L.; Shannon, S. S. ACS Catal. 2013, 3, 2612; 3-
Azabicyclo[3,2,1]octane hydrochloride: (d) Dong, J.; Han, H.; Geng, B.; Li, X.;
Gong, Z.; Liu, K. J. Pept. Res. 2005, 65, 440; 2-Azaadamantane: (e) Shibuya, M.;
Sasano, Y.; Tomizawa, M.; Hamada, T.; Kozawa, M.; Nagahama, N.; Iwabuchi, Y.
Synthesis 2011, 21, 3418. The other cyclic amines were commercial products.
11. Delgado, A.; Clardy, J. Tetrahedron Lett. 1992, 20, 2789.
c
is indeed mainly due to increased binding affinity for hPPAR
c
Table 1.
We then evaluated the thermodynamic solubility of representa-
tive compounds.14 Although most of the potent compounds exhib-
ited decreased aqueous solubility, 8-azabicyclo[3,2,1]octane
derivative (8g) showed aqueous solubility comparable to that of
the parent piperidine derivative (3).
12. Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc. 1981, 103, 2127.
13. We performed direct binding assay of representative compounds based on the
principle of the surface plasmon resonance, using a Biacore X 100 system with
In conclusion, we have succeeded in constructing a plausible
a hPPAR
follows.
c LBD-functionalized sensor-chip. The experimental protocol was as
omega loop structure for the hPPAR
tion. We used the reconstructed hPPAR
c
LBD by means of MD simula-
LBD model structure to
Reagents: Reagents (NiCl2, EDTA, NaOH) were purchased from Biacore
(Uppsala, Sweden). The assay compounds were prepared as 10 mM stock
solutions in DMSO. Immediately before analysis, each compound was diluted
with assay buffer containing 10 mM HEPES, 150 mM sodium chloride, 0.05%
TWEEN 20, pH 7.4 to yield a final DMSO concentration of 1% and a compound
concentration of 1000, 100, 10, 1, 0.1 or 0.01 nM.
c
design new mono-, bi- and tri-cyclic amine compounds that were
expected to show increased binding affinity for the LBD. As
expected, most of the newly designed and synthesized compounds
exhibited more potent hPPARc agonistic activity and greater
Capture of His-hPPAR
1 min by injecting 0.5 mM NiCl2. A 250
10 mM HEPES, 150 mM sodium chloride, pH 7.4) was injected for 2 min at a
flow rate of 10 L/min. Typically, this method resulted in immobilization of ca.
2500 RU PPAR LBD.
Kinetic analysis of ligand/hPPAR
data, a concentration series of each compound was injected at a flow rate of
30 L/min at 25 °C. For analysis, the association and the dissociation phases
c
LBD by NTA: Biacore X100 flow cells were activated for
hPPAR binding affinity than 3. Some of these compounds exhib-
c
l
g/mL His-hPPAR LBD solution (in
c
ited comparable aqueous solubility to 3. In order to ascertain the
precise binding mode of the representative compound 8g, we are
currently attempting an X-ray crystallographic analysis of the
l
c
c
LBD interactions: To collect detailed kinetic
hPPARc LBD complexed with 8g.
l
were fixed at 60 s.
Acknowledgments
Analysis: Affinity analysis of each ligand/receptor interaction was conducted by
fitting the response data to steady-state affinity according to supplier’s
protocol.
This work was supported in part by a Grant-in-Aid for Scientific
Research from The Ministry of Education, Culture, Sports, Science
and Technology, Japan, and Platform for Drug Discovery,
14. Ishikawa, M.; Hashimoto, Y. J. Med. Chem. 2011, 54, 1539.