Design and evaluation of azaindole-substituted N-hydroxypyridones as glyoxalase i inhibitors

Article abstract of DOI:10.1016/j.bmcl.2012.10.045

We conducted a high throughput screening for glyoxalase I (GLO1) inhibitors and identified 4,6-diphenyl-N-hydroxypyridone as a lead compound. Using a binding model of the lead and public X-ray coordinates of GLO1 enzymes complexed with glutathione analogues, we designed 4-(7-azaindole)-substituted 6-phenyl-N-hydroxypyridones. 7-Azaindole's 7-nitrogen was expected to interact with a water network, resulting in an interaction with the protein. We validated this inhibitor design by comparing its structure-activity relationship (SAR) with that of corresponding indole derivatives, by analyzing the binding mode with X-ray crystallography and by evaluating its thermodynamic binding parameters.

Full text of DOI:10.1016/j.bmcl.2012.10.045

Bioorganic & Medicinal Chemistry Letters 22 (2012) 7486–7489
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Design and evaluation of azaindole-substituted N-hydroxypyridones
as glyoxalase I inhibitors
Takashi Chiba, Jun Ohwada, Hiroshi Sakamoto, Takamitsu Kobayashi, Takaaki A. Fukami, Machiko Irie,
⇑
Takaaki Miura, Kazuhiro Ohara, Hiroshi Koyano
Research Division, Chugai Pharmaceutical Co., Ltd, 200 Kajiwara Kamakura, Kanagawa 247-8530, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
We conducted a high throughput screening for glyoxalase I (GLO1) inhibitors and identified 4,6-diphenyl-
N-hydroxypyridone as a lead compound. Using a binding model of the lead and public X-ray coordinates
of GLO1 enzymes complexed with glutathione analogues, we designed 4-(7-azaindole)-substituted
6-phenyl-N-hydroxypyridones. 7-Azaindole’s 7-nitrogen was expected to interact with a water network,
resulting in an interaction with the protein. We validated this inhibitor design by comparing its struc-
ture-activity relationship (SAR) with that of corresponding indole derivatives, by analyzing the binding
mode with X-ray crystallography and by evaluating its thermodynamic binding parameters.
Ó 2012 Elsevier Ltd. All rights reserved.
Received 17 August 2012
Revised 3 October 2012
Accepted 9 October 2012
Available online 17 October 2012
Keywords:
Glyoxalase I (GLO1)
Structure-based drug design (SBDD)
N-Hydroxypyridone
7-Azaindole
Glyoxalase I (GLO1) is a zinc enzyme that catalyzes the isomer-
ization of a hemithioacetal, formed from glutathione (GSH) and
methylglyoxal (MG), to lactic acid thioester. The potential of
GLO1 as a drug target for cancer and other diseases has been dis-
cussed over the last several decades.1 GSH analogues,2 flavonoids,3
curcumins,4 and a benzothiazole derivative5 are reported to show
moderate to potent GLO1 inhibitory activity and are expected to
be evaluated further for practical uses.
We conducted a discovery program of non-peptidic GLO1 inhib-
itors through lead generation from high throughput screening
(HTS) followed by rational inhibitor design using human GLO1/
GSH analogue cocrystal structural data deposited in Protein Data
Bank (PDB).6–8 We analyzed the ligand–protein interactions ob-
served in those structures and found that the following interac-
tions are commonly observed: (1) hydrophobic interactions of
ligands’ substituents on the sulfur atom in a hydrophobic pocket;
(2) a hydrogen bond between cysteine NH of ligands and a water
molecule (hereafter referred to as W1) bound to Thr101A and
them, 4,6-diphenyl-N-hydroxypyridone (1) was found to have
more potent GLO1 inhibitory activity (IC50 = 1.2 M) than a well-
known GLO1 inhibitor, S-(p-bromobenzyl)glutathione (pBBG)2a
(IC50 = 21
M, in-house data),11 as well as good stability in serum
l
l
and against liver microsomal oxidation (Fig. 2).
His126B
Gln33A
HN
O
Glu172B
Leu160B
N
O
2+
O
Zn
Glu99A
O
hydrophobic
pocket
O
Met183B
R
S
Phe162B
Thr101A
H2N
W1
O
OH
O
S
H
N
H
N
H
N
O
NH2
Glu99A; (3) CH/
p
interaction between alkyl chain of
c
-glutamic
O
Phe62A
Lys156B
HN
O Met157B
O
O
NH
acid of ligands and Phe67A of the protein; (4) ionic interactions be-
H2N
Arg37A
O
O
tween the
c-glutamyl carboxylate/ammonium groups of ligands
HN
and Arg37A/Asn103A/Arg122B of the protein (Fig. 1).9
NH2
NH2
H2N
Phe67A
We conducted an HTS and identified a few hundred compounds
whose GLO1 IC50 values were less than 10 lM. The majority of
them had apparent zinc binding functional groups, which was
H2N
Asn103A
NH
O
Arg122B
anticipated from the nature of the GLO1 catalytic site. Among
Figure 1. Schematic drawing of the interactions found in X-ray cocrystal structures
of GLO1 and GSH analogs (R group, PDB code): R = benzyl, 1FRO;6 R = n-hexyl,
1BH5;7 R = 4-I-Ph-N(OH)C(O)–, 1QIN;8 R = 4-NO2–BnOC(O)–, 1QIP.8 GLO1 catalytic
site is composed of two peptide chains (A and B).10
⇑
Corresponding author.
E-mail address: koyanohrs@chugai-pharm.co.jp (H. Koyano).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.10.045

T. Chiba et al. / Bioorg. Med. Chem. Lett. 22 (2012) 7486–7489
7487
O
Table 1
1
HO
Serum T1/2 (min)
GLO1 IC50 = 1.2 µM
GLO1 inhibitory activity of 6-phenyl-N-hydroxypyridones with various substituents
N
Mouse: 440
at 4-position
3'
4'
4
6
Human: 2900
5'
Liver microsomal clearance (mg/min/mL):
Mouse: 28 (NADPH+), 6.8 (NADPH-)
Human: 16 (NADPH+), 6.6 (NADPH-)
O
HO
N
R4
Figure 2. Structure and property of 4,6-diphenyl-N-hydroxypyridone (1).
A binding model of compound 1 was generated by its docking
into the GLO1 catalytic site of a complex with zinc-coordinating
S-(N-hydroxy-N-(p-iodophenyl)carbamoyl)glutathione (PDB co-
de:1QIN)8 in which the position of the two zinc-coordinating oxygen
atoms was maintained. The probable binding mode is depicted in
Figure 3. Another mode of zinc-coordination with alternative posi-
tions for N–O and C@O was denied because of severe steric repulsion
between two phenyl groups and protein. In the model, 4-phenyl
group of 1 was situated in the hydrophobic pocket (composed of
Phe62A, Cys60A, Ile88A, Leu92A, Met157B, Met179B, Leu182B
and Met183B) making an edge-to-face aromatic interaction with
Phe62A. 6-Phenyl group of 1 was situated at the space which
Compound
R4
GLO1 IC50
1.19
(lM)
*
1
*
*
2a
2b
1.52
8.28
S
N
*
*
2c
8.13
8.40
CF3
Ph
c-glutamyl moiety of GSH analogues occupied (GSH binding
2d
pocket). Phe162B made an edge-to-face interaction with the
6-phenyl ring. No steric repulsionof 1 with W1 was observed (Fig. 3).
Since the zinc-coordination of N-hydroxypyridone fixed the vec-
tor of 4- and 6-substituents, we thought structure-activity relation-
ship (SAR) of each substituent is independent and the combination
of each optimized substituent would results in the best optimized
compound. The first step of our lead optimization was to find alter-
native groups for the 4-phenyl group of 1 to enhance hydrophobic
interaction in the hydrophobic pocket while still keeping the 6-phe-
nyl group. Creighton et al. reported a positive correlation between
competitive inhibition constant (Ki) and the Hansch hydrophobicity
constant of aryl and alkyl group of S-(N-hydroxy-N-aryl/alkyl-
carbamoyl)glutathiones.13 Therefore we prepared several 6-phenyl
N-hydroxypyridones bearing a hydrophobic substituent at 4-
position. Among them (1, 2a–2f), 1 was the most potent. 3-Thienyl
(2a), a bioisostere of the phenyl group, kept its activity but 3-
pyridinyl (2b) decreased in activity, presumably because of its
hydrophilicity. 4-Substituted phenyl groups (2c and 2d) also de-
creased activity indicating limited space around the 4-position of
the phenyl group. Non-aromatic substituents, 1-butynyl and butyl
group (2e and 2f, respectively) lost activity. This result indicates
that the edge-to-face aromatic interaction between the phenyl
*
*
2e
2f
>10
>10
group and Phe62A is important (Table 1). Accordingly, we
kept the phenyl group at 4-position during the following
modifications.
Next, we designed substituents of the 6-position. Edge-to-face
aromatic interaction between aromatic ring of ligand and
Phe162B needed to be retained because the interaction seems to
be important for tight binding. 30-Position of the 6-phenyl group
of 1 directs to a hydrophilic space where W1 exists. On the other
hand, 40-and 50-positions of the 6-phenyl ring directs to a hydro-
phobic pocket surrounded by side-chains of residues Lys156B,
Met157B and Phe162B. From these observations, we designed 7-
azaindole whose 7-nitrogen faces the hydrophilic space and 5-
membered ring occupies the hydrophobic pocket. The substituent
on 1 N-position, which directs to the solvent-accessible region and
therefore may contribute to the solubility of inhibitors, should be
explored. To understand the function of 7-azaindole, we prepared
the corresponding indole derivatives and evaluated them in
parallel.
We did not employ the ionic interactions with Arg37A/
Asn103A/Arg122B in our inhibitor design because introducing a
carboxylate and/or an ammonium functional group to the
N-hydroxypyridone scaffold would result in inhibitors with poor
cell permeability.
The SAR of azaindole derivatives is shown in Table 2 along with
the corresponding indole derivatives. 1 N-unsubstituted 7-
azaindole (3a) and indole (4a) equally showed better activity than
1. 3-Carbamoyl-benzyl (3b) and methoxyalkyl groups (3c and 3d)
on 1 N-position of 7-azaindole showed significant improvement in
activity but the same substituents on indole (4b–4d) did not show
such an effect.
We solved a cocrystal structure of 3d/GLO1 complex at a res-
olution of 1.5 Å (Fig. 4, PDB code:3VW9). An asymmetric unit
contains two active sites located at the interface of two peptide
chains (A and B). The structure revealed that N-hydroxypyridone
of compound 3d coordinated a zinc cation (d(NOꢀ ꢀ ꢀZn) = 2.4 Å,
d(COꢀ ꢀ ꢀZn) = 2.2 Å). 4-Phenyl group, which made the edge-to-face
Figure 3. Binding model of 1. The model was created using MOLOC.12 Color surface
is shown in green (hydrophobic), blue (donor), red (acceptor) and purple
(hydrophilic).
interaction with Phe62A (d(Phe62Aꢀ ꢀ ꢀ4-Ph
p face) = 4.3 Å), was

7488
T. Chiba et al. / Bioorg. Med. Chem. Lett. 22 (2012) 7486–7489
Table 2
GLO1 inhibitory activity of 4-phenyl-N-hydroxypyridones bearing 7-azaindole or
indole moiety at 6-position
O
HO
N
X
R
N
Compound
X
R
GLO1 IC50 (lM)
3a
4a
N
CH
H
0.28
0.25
*
3b
4b
N
CH
O
0.040
0.48
*
H2N
O
Figure 5. Thermodynamic parameters of 3d and its indole counterpart 4d.
3c
4c
N
CH
0.014
0.26
*
3d
4d
N
CH
0.011
0.30
*
O
The effect of oxygen in the 1 N-substituent was investigated by
the comparison with carbon analogues (Table 3). Simple alkyl
groups (3e and 3f) were not effective in enhancing activity. Since
oxygen in the alkyl chain of 3d stayed in the solvent-accessible re-
gion, simple alkyl groups were unfavorable in such a hydrophilic
environment.
Preparation of 3d was performed starting from 2,6-
dichloropyridineoxide (5) via iridium-catalyzed bolyration,15
Suzuki coupling, de-chlorinating hydrolysis, and methoxymethyl
(MOM) protection to give an intermediate 8. Suzuki coupling of
8 with boronate 9 followed by alkylation at 1 N-position under
basic conditions and subsequent deprotection afforded 3d
(Scheme 1).16
In summary, we discovered 7-azaindole substituted N-hydroxy-
pyridone 3d as a potent zinc-binding GLO1 inhibitor (IC50 = 11 nM)
through lead generation from HTS and structure-based inhibitor
design. The X-ray cocrystal structure and the comparison of bind-
ing energies with the indole counterpart 4d revealed that 3d bound
to the water network and its binding was enthalpy-driven.
Methoxypropyl chain of 3d is essential for potent activity and this
chain seems to stabilize the water network by strengthening
neighboring hydrogen bonds. Other approaches of lead optimiza-
tion for new GLO1 inhibitors will be discussed in future
publications.
situated in the hydrophobic pocket. Azaindole was situated at the
GSH binding pocket, forming an edge-to-face interaction with
Phe162B (d(Phe162Bꢀ ꢀ ꢀ7-azaindole
p
face) = 4.0 Å). Phe67A was
face) = 4.5 Å).
near 7-azaindole
p
face (d(Phe67Aꢀ ꢀ ꢀ7-azaindole
p
Azaindole pyridine nitrogen made a hydrogen bond with a water
(d(Nꢀ ꢀ ꢀO) = 2.8 Å) in the hydrogen bond network including W1. A
methoxypropyl group directed to Phe162B (d(Phe162Bꢀ ꢀ ꢀ-
CH2OMe) = 3.9 Å) and Trp170B (d(Trp170Bꢀ ꢀ ꢀCH2OMe) = 3.5 Å).
Oxygen in the alkyl chain reached the solvent-accessible region.
Thermodynamic parameters of inhibitor binding (3d and 4d) to
GLO1 were determined by isothermal titration calorimetry (Fig. 5).
As expected from the hydrogen bond network including azaindole
observed in the X-ray structure, the binding of 3d was enthalpic
driven
Such enthalpy-driven binding was not observed for indole 4d
H = ꢁ6.3 1.7 kcal/mol, ꢁT S = ꢁ2.0 1.4 kcal/mol). Since the
(D
H = ꢁ11.1 1.2 kcal/mol, ꢁTDS = 2.1 1.0 kcal/mol).
(D
D
1 N-unsubstituted derivatives 3a and 4a did not show a difference
in activity, the stabilization of the water network is affected by
substituents at this position. The dependence of hydrogen bond
strengths on microenvironment polarity is reported in stabilization
of protein structure.14 We assume that the stabilization of the
water network was achieved by 3d, and resulted in improved
activity.
Table 3
The effect of oxygen on alkyl chain
O
HO
N
N
R
N
Compound
R
GLO1 IC50
0.014
(lM)
*
*
O
3c
*
*
3d
0.011
O
3e
3f
0.30
0.51
Figure 4. X-ray cocrystal structure of 3d/GLO1 (PDB ID:3VW9).

T. Chiba et al. / Bioorg. Med. Chem. Lett. 22 (2012) 7486–7489
7489
Cl
O
O
Cl
Cl
N+
O
O
O
N+
O
N+
a
b
c
N
85%
90%
94%
OH
Cl
Cl
Cl
B
Cl
OH
5
6
7
8
O
O
O
O
O
N
O
N
O
B
N
d
e
N
8
+
O
N
47%
89%
N
HN
HN
O
9
10
11d
O
HO
N
f
78%
N
N
O
3d
Scheme 1. Preparation of 3d.a aReagents and conditions: (a) [Ir(OMe)COD]2, dtbpy, bis(pinacolato)diboron; (b) PdCl2(PPh3)2, iodobenzene, Na2CO3 aq, NMP; (c) (i) 1 M NaOH
aq, DMSO, 70 °C; (ii) MOMCl, DIPEA, CH2Cl2, rt; (d) PdCl2(PPh3)2, Na2CO3 aq, NMP, 70 °C; (e) MeO(CH2)3Cl, 5 M NaOH aq, DMF; (f) 4 M HCl/AcOEt, ethyleneglycol, THF.
5. Takasawa, R.; Tao, A.; Saeki, K.; Shionozaki, N.; Tanaka, R.; Uchiro, H.;
Takahashi, S.; Yoshimori, A.; Tanuma, S. Bioorg. Med. Chem. Lett. 2011, 21, 4337.
6. Cameron, A. D.; Olin, B.; Ridderström, M.; Mannervik, B.; Jones, T. A. EMBO J.
Acknowledgments
We thank Yasushi Sakaguchi, DKSH Japan K.K., for his help in
conducting ITC measurements.
1997, 16, 3386.
7. Ridderström, M.; Cameron, A. D.; Jones, T. A.; Mannervik, B. J. Biol. Chem. 1998,
273, 21623.
8. Cameron, A. D.; Ridderström, M.; Olin, B.; Kavarana, M. J.; Creighton, D. J.;
Mannervik, B. Biochem. 1999, 38, 13480.
9. For a detained discussion of non-covalent interactions, see: Bissantz, C.; Kuhn,
B.; Starhl, M. J. Med. Chem. 2010, 53, 5061.
10. He, M. M.; Clugston, S. L.; Honek, J. F.; Matthews, B. W. Biochemistry 2000, 39,
8719.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl.2012.
10.045.
11. GLO1 inhibitory activity was evaluated by examining the decrease in reduced
glutathione (GSH). Recombinant human GLO1 protein is expressed in
Escherichia coli BL21-CodonPlus. The test compound was incubated with
sodium phosphate (50 mM, pH 7.0) containing MgCl2 (12 mM), reduced GSH
(0.1 mM), methylglyoxale (0.5 mM), and purified human GLO1 protein
(8.75 ng) for 1 h at 37 °C. 5,50-Dithiobis(2-nitrobenzoic acid) (Ellman0s
reagent) (0.67 mM) in sodium phosphate (10 mM, pH 7.0) was added at
room temperature. After 10 min, the absorbance was measured at 450 nm.
12. Gerber, P. R.; Müller, K. J. Comput. Aided Mol. Des. 1995, 9, 251.
13. Kalsi, A.; Kavarana, M. J.; Lu, T.; Whalen, D. L.; Hamilton, D. S.; Creighton, D. J. J.
Med. Chem. 2000, 43, 3981.
14. Gao, J.; Bosco, D. A.; Powers, E. T.; Kelly, J. W. Nat. Struct. Mol. Biol. 2009, 16, 684.
15. Ishiyama, T.; Miyaura, N. Pure Appl. Chem. 2006, 78, 1369.
16. Spectral data of 3d: 1H NMR (400 MHz, DMSO-d6) d: 11.51 (br s, 1H), 8.55 (d,
J = 1.99 Hz, 1H), 8.30 (d, J = 1.99 Hz, 1H), 7.8 (m, 2H), 7.65 (d, J = 3.52 Hz, 1H),
7.40-7.54 (m, 3H), 6.85 (d, J = 2.63 Hz, 1H), 6.58 (d, J = 2.63 Hz, 1H), 6.56 (d,
J = 3.52 Hz, 1H), 4.28 (t, J = 7.05 Hz, 2H), 3.32 (t, J = 6.17 Hz, 2H), 3.24 (s, 3 H),
2.06 (tt, J = 6.17, 7.05 Hz, 2H). MS (ESI+) 376 [M+H]+.
References and notes
1. Cell biology of glyoxalase in microbial systems and human disease: Thornalley,
P. J.; Rabbani, N. Semin. Cell Dev. Biol. 2011, 22, 261.
2. (a) Vince, R.; Daluge, S. J. Med. Chem. 1971, 14, 35; (b) More, S. S.; Vince, R. J.
Med. Chem. 2009, 52, 4650; (c) Al-Timari, A.; Douglas, K. T. Biochim. Biophys.
Acta 1986, 870, 160; (d) Lo, T. W.; Thornalley, P. J. Biochem. Pharmacol. 1992, 44,
2357; (e) Hamilton, D. S.; Creighton, D. J. J. Biol. Chem. 1992, 267, 24933; (f)
Murthy, N. S.; Bakeris, T.; Kavarana, M. J.; Hamilton, D. S.; Lan, Y.; Creighton, D.
J. J. Med. Chem. 1994, 37, 2161.
3. Takasawa, R.; Takahashi, S.; Saeki, K.; Sunaga, S.; Yoshimori, A.; Tanuma, S.
Bioorg. Med. Chem. 2008, 16, 3969.
4. Santel, T.; Pflug, G.; Hemda, N. Y. A.; Schäfer, A.; Hollenbach, M.; Buchold, M.;
Hintersdorf, A.; Lindner, I.; Otto, A.; Bigl, M.; Oerlecke, I.; Hutschenreuter, A.;
Sack, U.; Huse, K.; Groth, M.; Birkemeyer, C.; Schellenberger, W.; Gebhardt, R.;
Platzer, M.; Weiss, T.; Vijayalakshmi, M. A.; Krüger, M.; Birkenmeier, G. PLoS
One 2008, 3, e3508.