Green fluorescent protein chromophore derivatives as a new class of aldose reductase inhibitors

Article abstract of DOI:10.1016/j.ejmech.2016.10.016

A number of (Z)-4-arylmethylene-1H-imidazol-5(4H)-ones, which are related to the fluorescent chromophore of the Aequorea green fluorescent protein (GFP), have been synthesized and evaluated their in vitro inhibitory activity against recombinant human aldo

Full text of DOI:10.1016/j.ejmech.2016.10.016

Accepted Manuscript
Green fluorescent protein chromophore derivatives as a new class of aldose
reductase inhibitors
Ryota Saito, Maiko Hoshi, Akihiro Kato, Chikako Ishikawa, Toshiya Komatsu
PII:
S0223-5234(16)30878-9
10.1016/j.ejmech.2016.10.016
EJMECH 8981
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 20 May 2016
Revised Date: 14 September 2016
Accepted Date: 7 October 2016
Please cite this article as: R. Saito, M. Hoshi, A. Kato, C. Ishikawa, T. Komatsu, Green fluorescent
protein chromophore derivatives as a new class of aldose reductase inhibitors, European Journal of
Medicinal Chemistry (2016), doi: 10.1016/j.ejmech.2016.10.016.
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.

ACCEPTED MANUSCRIPT
Green fluorescent protein chromophore derivatives as a new class of aldose reductase inhibitors
Ryota Saito,a,b* Maiko Hoshi,a Akihiro Kato,a Chikako Ishikawa,c and Toshiya Komatsud
aDepartment of Chemistry, Toho University, 2-2-1 Miyama, Funabashi, Chiba, 274-8510 Japan;
bResearch Center for Materials with Integrated Properties, Toho University, 2-2-1 Miyama, Funabashi,
Chiba 274-8510, Japan; cFaculty of Pharmaceutical Science, Toho University, 2-2-1 Miyama, Funabashi,
d
Chiba, 274-8510 Japan; Kowa Company, Ltd., 3-4-14 Nihonbashi-honcho, Chuo-ku, Tokyo 103-8433,
Japan
Graphical Abstract
O
COOH
N
N
hALR2 IC50 = 0.10 µM

Green fluorescent protein chromophore derivatives as a new class of aldose reductase inhibitors
ACCEPTED MANUSCRIPT
Ryota Saito,a,b* Maiko Hoshi,a Akihiro Kato,a Chikako Ishikawa,c and Toshiya Komatsud
aDepartment of Chemistry, Toho University, 2-2-1 Miyama, Funabashi, Chiba, 274-8510 Japan;
bResearch Center for Materials with Integrated Properties, Toho University, 2-2-1 Miyama, Funabashi,
Chiba 274-8510, Japan; cFaculty of Pharmaceutical Science, Toho University, 2-2-1 Miyama, Funabashi,
d
Chiba, 274-8510 Japan; Kowa Company, Ltd., 3-4-14 Nihonbashi-honcho, Chuo-ku, Tokyo 103-8433,
Japan
Abstract. A number of (Z)-4-arylmethylidene-1H-imidazol-5(4H)-ones, which are related to the
fluorescent chromophore of the Aequorea green fluorescent protein (GFP), have been synthesized and
evaluated their in vitro inhibitory activity against recombinant human aldose reductase for the first time.
The GFP chromophore model 1a, with a p-hydroxy group on the 4-benzylidene and a carboxymethyl
group on the N1 position, exhibited strong bioactivity with an IC50 value of 0.36 µM. This efficacy is
higher than that of sorbinil, a known highly potent aldose reductase inhibitor. Compound 1h, the
2-naphtylmethylidene analogue of 1a, exhibited the best inhibitory effect among the tested compounds
with an IC50 value of 0.10 µM. Structure-activity relationship studies combined with docking
simulations revealed the interaction mode of the newly synthesized inhibitors toward the target protein
as well as the structural features required to gain a high inhibitory activity. In conclusion, the GFP
chromophore model compounds synthesized in this study have proved to be potential drugs for diabetic
complications.
Keywords
Aldose reductase inhibitor; Drug candidates for diabetic complications; GFP chromophore model;
(Z)-4-Arylmethylidene-1H-imidazol-5(4H)-ones; Docking study
1. Introduction
Aldose reductase (ALR2; EC 1.1.1.21) is an NADPH-dependent enzyme that catalyzes the reduction
of
D-glucose to D-sorbitol using NADPH as a reductant in the polyol pathway (Figure 1) [1, 2].
Ordinarily, most glucose is metabolized via the tricarboxylic acid (TCA) cycle in the glycolytic system.
In patients with hyperglycemia, however, glucose actively fluxes not only into the glycolytic system but
1

also into the polyol pathway, where ALR2 is activated to consume the flooded glucose [1, 2]. Therefore,
ACCEPTED MANUSCRIPT
high levels of
D
-sorbitol are formed intracellularly. Since the oxidation of
D-sorbitol to D-fructose is
very slow, the intracellular concentration of
D
-sorbitol increases and a large amount of NADPH is
consumed, resulting in an osmotic pressure imbalance that causes some diabetic complications such as
peripheral neuropathy, cataract, nephropathy, and angiopathy [1, 3-8]. Hence, the inhibition of ALR2
activity may offer a promising option for the alleviation or prevention of complications and symptoms
associated with chronic hyperglycemia [7, 9-11].
Figure 1. Schematic representation of the polyol pathway of glucose metabolism
Various types of ALR2 inhibitors (ARIs) have been reported and tested in clinical trials, but many of
them presented low activity or undesirable side effects [4, 12-14]. Currently, only epalrestat (Figure 2) is
launched on the market as the therapeutic medicine for diabetic peripheral neuropathy in Japan [14-17].
Owing to the limited number of drugs available for the treatment of diabetic complications, the
development of new ARIs has still been strongly desired.
ARIs exhibiting strong ALR2 inhibitory effects generally fall into two structural categories; those
containing an N-carboxymethyl (NCH2CO2H) moiety and those with a cyclic imide represented by a
spirohydantoin or related ring system [2, 11, 13, 14, 16, 18, 19]. These acidic functional groups are
believed to dissociate and form the corresponding anionic conjugate bases under physiological pH
conditions to interact, through hydrogen bonds, with the catalytic site of the protein composed of Tyr48,
His110, Trp20, and Trp111 side chains, adjacent to which the coenzyme NADPH is placed. The catalytic
site is also called an “anion binding pocket” since His110 is positively charged under physiological pH
conditions by protonation and, therefore, attracts the carbonyl oxygens of substrates, or negatively
charged substituents, such as carboxylates. Adjacent to the catalytic site, ALR2 has a lipophilic pocket
composed of the residues Trp20, Phe122, and Trp219, and, therefore, highly potent ARIs require having
a hydrophobic moiety [2, 20-23]. Epalrestat possesses exactly these key units.
The well-known green fluorescent protein (GFP) from the jellyfish Aequorea victoria has a
(Z)-4-(p-hydroxyphenylmethylene)-1H-imidazol-5(4H)-one ring system as a chromophore, formed by
2

autocatalytic posttranslational cyclization and oxidation of the tripeptide Ser65-Tyr66-Gly67 in the
ACCEPTED MANUSCRIPT
protein sequence (Figure 2) [24-27].
Figure 2. Structures of epalrestat, the chromophore moiety in the Aequorea green fluorescent protein
(GFP), and the GFP chromophore model compound 1a.
Interestingly, the GFP chromophore model structure contains the NCH2CO2H fragment, and has a
similar molecular skeleton and electronic structure to epalrestat. In spite of these features, no studies
have investigated the inhibitory activity of the GFP chromophore models against ALR2 thus far.
In this study, we synthesized several GFP model compounds, evaluated their in vitro inhibitory
activity against the recombinant human ALR2, and provided the first example of structure-activity
relationship among this family of compounds.
2. Results and discussion
2.1. Synthesis of (Z)-4-arylmethylene-1H-imidazol-5(4H)-ones
(Z)-4-Arylmethylidene-1H-imidazol-5(4H)-ones were synthesized using the procedure reported by
Niwa et al.[28-30] Azlactones 4a–k were first prepared by reacting N-acetylglycine with appropriate
arylaldehydes in acetic anhydride in the presence of sodium acetate [28, 31-36], and afterward reacted
with aqueous ammonia to give the imidazolones 5a–k (Scheme 1). Imidazolones 5a–c, containing
hydroxy groups, were protected with tert-butyldimethyl silyl (TBS) groups to form 5n–p, respectively.
The obtained imidazolones 5c–i and 5n–p were then reacted with ethyl bromoacetate in acetone in the
presence of K2CO3 to yield 6c–i and 6n–p, respectively [27]. The TBS protecting groups in 6n–p were
then removed with tetra-n-butylammonium fluoride (TBAF) to give the corresponding 6a–c. Afterward,
6a–j were hydrolyzed to give the corresponding carboxylic acids 1a–j. The reaction of unprotected 5a
with 2 equivalents of ethyl bromoacetate yielded 6l, which was then hydrolyzed to form compound 1m.
Compound 2, containing the β-alanine fragment (NCH2CH2CO2H) as a substitute of the glycine unit
(NCH2CO2H) present in 1a, was synthesized by reacting 5a with ethyl acrylate in the presence of
cesium fluoride in acetonitrile, followed by a hydrolysis of the resulting ester (Scheme 2).
3

ACCEPTED MANUSCRIPT
Scheme 1. Reagents and conditions: (i) 28% NH4OH, K2CO3, EtOH, reflux; (ii) TBSCl, imidazole,
DMF, r.t.; (iii) BrCH2CO2Et, K2CO3, acetone, reflux; (iv) TBAF, MeOH, r.t.; (v) NaOH, EtOH, then
H3O+.
Scheme 2. Reagents and conditions: (i) CH2=CHCO2Et, CsF, MeCN, reflux; (ii) NaOH, EtOH, then
H3O+.
The structures of 1a–j, 1m, and 2 were assigned on the basis of their spectroscopic data. In the
1H-NMR spectra, each compound showed a singlet peak around 6.98–7.28 ppm, assignable to the
olefinic proton, which indicates that these are cis-isomers, while the corresponding trans-isomers exhibit
the resonance of the olefinic protons with about 0.25-ppm higher chemical shift values [37]. X-ray
crystallographic analysis of 6k, which was prepared as depicted in Scheme 1, further confirmed the cis
configuration: Figure 3 shows an ORTEP diagram of 6k, clearly showing that the C=C at the C4
position of the imidazole ring takes a cis configuration. Unfortunately, several attempts to deprotect the
methyl protecting groups in 6k have only resulted in recovery or decomposition of the compound.
Figure 3. An ORTEP diagram of 6k
4

2.2. In vitro aldose reductase inhibitory activity of (Z)-4-arylmethylene-1H-imidazol-5(4H)-ones
ACCEPTED MANUSCRIPT
The ALR2 inhibitory activity of the synthesized imidazolones (1a–j, 1m,and 2) was evaluated in
vitro by measuring their inhibitory effects on the reduction of D,L-glyceraldehyde with the recombinant
human ALR2 in the presence of the cofactor NADPH [38-40]. The reaction was monitored with the use
of a spectrophotomete by measuring NADPH consumption, as previously reported [41]. The respective
IC50 values, which express the 50% inhibition concentration of the compounds on the bioreduction, are
shown in Table 1. The reference compound epalrestat was also assayed in the same manner, and the
result is compiled in Table 1.
Table 1. In vitro inhibitory activity of (Z)-4-arylmethylene-1H-imidazol-5(4H)-ones 1a–j, 1m, 2, 6a, 6c,
6f, 6g, 6l, and epalrestat against recombinant human ALR2
IC50 (µM) b
0.36
Compound
Ar
4-OH-C6H4
3-OH-C6H4
C6H5
πa
R
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
CH2CO2H
1a
1b
1c
1d
1e
1f
1g
1h
1i
3.20
0.58
0.51
0.65
0.27
0.24
0.10
0.54
0.13
0.31
1.96
1.94
2.10
2.67
2.82
3.07
4-OMe-C6H4
4-F-C6H4
4-Cl-C6H4
4-Br-C6H4
2-naphthyl
1-naphthyl
3,5-diCH3-4-OH-C6H4
4-OCH2CO2H-C6H4
4-OH-C6H4
4-OH-C6H4
C6H5
1j
1m
2
(CH2)2CO2H 10.7
CH2CO2Et
CH2CO2Et
CH2CO2Et
CH2CO2Et
CH2CO2Et
23
112
16% (13µM)c
n.d.d
6a
6c
6f
4-Cl-C6H4
4-Br-C6H4
4-OCH2CO2Et-C6H4
6g
6l
109
epalrestat
0.085
a Hydrophobicity parameters of Ar groups [42, 43]. b 50% inhibition concentrations. c % inhibition at the
d
given concentration. The value could not be determined owing to the poor solubility of the test
compound toward the assay medium.
The IC50 value of 1a was estimated to be 0.36 µM, indicating about 6-fold higher level of activity
than sorbinil, a highly potent ARI (IC50 = 0.9–20 µM against various ALR2) [22, 44, 45]. This result
strongly indicates that the GFP chromophore model represents a good ARI candidate.
A closer examination of the obtained ALR2 inhibition data gives insights into several
structure-activity relationship. First, the ethyl esters 6a, 6c, 6f, 6g and 6l inhibited ALR2 less than the
5

corresponding carboxylic acids 1a, 1c, 1f, 1g, and 1m, indicating that the carboxylic acid moiety is
ACCEPTED MANUSCRIPT
essential for the ALR2 inhibition activity as expected. Replacing the glycine unit (NCH2CO2H) in 1a
with the β-alanine fragment (NCH2CH2CO2H), i.e. extending the methylene chain, resulted in a drastic
decrease in the bioactivity (2, IC50 = 10.7 µM). These results indicate that the carboxymethyl group at
the N1 position is important and that the methylene-chain length is essential for the interaction with the
catalytic site of the enzyme.
Next, we examined the relationship between the structure of the arylmethylidene moiety at the C4
position and the ALR2 inhibitory activity of some of the compounds. The removal of the hydroxyl
group on the phenyl ring in 1a, resulting in compound 1c, reduced the bioactivity (IC50 = 0.58 µM). The
methyl protection of the hydroxyl group in 1a led to the less effective derivative 1d (IC50 = 0.51 µM).
Moving the hydroxyl group of 1a to position 3 of the 4-phenylidene ring (1b) markedly diminished the
inhibitory activity (IC50 = 3.20 µM). In contrast, compound 1m with a para carboxymethoxy group on
4-phenylidene produced almost equal inhibitory activity (IC50 = 0.31 µM) to that of 1a. From these
results, it is possible to deduce that a substituent capable of hydrogen bonding at the para position of the
4-phenylidene moiety may be important to achieve a significant inhibitory activity. However, a
paradoxical behavior was observed with the halogenated analogues 1e–g. The fluoro derivative 1e was
about two-fold less effective (IC50 = 0.65 µM) than 1a, whereas the isosteric replacement of the fluorine
atom in 1e with chlorine (1f, IC50 = 0.27 µM) and bromine atoms (1g, IC50 = 0.24 µM) increased the
bioactivity. Based on these observations, the effects of halogen atoms at the para position of the
4-phenylidene moiety on the ALR2 inhibitory activity can be ranked in the following order: of Br>Cl>F.
These results give us another insight into structure-activity relationship, as the above chemical
modifications decrease the molecular polarity of the compounds and increase their hydrophobicity. In
fact, the hydrophobic fragmental constants for F, Cl, and Br atoms in n-octanol are 0.399, 0.922 and
1.131, respectively, indicating hydrophobicity in the order of Br>Cl>F [46].
Taking these factors and the above observations into account, our understanding is that chemically
modifying 1a with substituents that increase the hydrophobicity around the 4-arylmethylidene moiety
results in a more profound ALR2 inhibitory activity of the GFP chromophore model family.
Substituents at the para position of the 4-phenylidene capable of forming hydrogen bonds improve the
inhibitory effect to a certain degree, but the hydrophobicity-raising modifications predominantly
enhance the activity probably by increasing the affinity of this moiety toward the lipophilic pocket of the
enzyme. Supplementary data supporting this consideration were obtained with 1h and 1j. In 1h, the
introduction of 2-naphthylmethylidene at the C4 position enhanced the inhibitory activity (IC50 = 0.10
6

µM) to a value similar to that of the positive control epalrestat, and resulted in the best ARI for the
ACCEPTED MANUSCRIPT
efficacy among the tested compounds in this study. Compound 1j has two methyl groups adjacent to the
hydroxyl group exhibited 3-fold higher inhibitory activity (IC50 = 0.13 µM) than that of the
corresponding analogue 1a. Altering the substitution pattern in 1h from
β to α (derivative 1i) diminished
the inhibitory activity (IC50 = 0.54 µM). These results strongly suggest that not only the acetic acid unit
at the N1 position but also an arylmethylidene fragment with increased hydrophobicity and appropriate
shape to fit the lipophilic pocket (e.g., the bicyclic aromatic systems), on the C4 position of the central
imidazolone skeleton is favorable for improving the ALR2 inhibitory activity of the GFP model
compounds.
2.3. Docking studies
In order to obtain better insight in the interactions between the imidazolones and ALR2, docking
experiments were performed using known high-resolution X-ray crystal structures of the human ALR2.
Nearly one hundred X-ray crystal structures of human ALR2 with various kinds of inhibitors have
already been reported. Based on these experimental data, the ALR2 binding site can be divided into
three sub-regions. Two of them are the catalytic and lipophilic pockets as mentioned above, while the
third region is defined as a specificity pocket, and is adjacent to the other two pockets [2]. While the
structure of the catalytic pocket is almost identical in the majority of the reported crystal structures, the
specificity pocket has a flexible structure to accommodate various ligands. This ligand-induced-fit
function is achieved by a movement of the Cys298–Cys303 loop region, and this brings two distinct
sates of the specificity pocket, “open” and “closed.” According to Klebe and co-workers, the
ligand-binding conformations of ALR2 can be classified into five types, and these are exemplified by the
PDB structures 1PWM, 1Z3N, 2FZD, 2NVC, and 2NVD [47, 48]. These structures have been frequently
used in ALR2 docking studies [23, 49, 50], and were used here as well. It should be noted that none of
these structures contains epalrestat as a ligand. Recently, the first report of the X-ray crystal structure of
ALR2 with epalrestat (PDB ID: 4JIR) has finally been published by Zhang et al., and showed that the
specificity pocket is in the “closed” conformation and that the hydrophobic moiety of epalrestat is
located in the lipophilic pocket [51]. Considering the structural similarities of our compounds with
epalrestat, this PDB structure was also used in the present study. Here, we also include an extra ALR2
crystal structure that was very recently reported by Klebe and co-workers (PDB ID: 4IGS) [52], as the
binding moiety of the ligand is a phenolic hydroxyl group, which is present in some of the imidazolones
synthesized in the this study. Interestingly, the protein conformation of 4IGS is almost similar to that of
7

1PWM except for the positions of Trp20, Phe122, and Trp219, composing the lipophilic pocket. These
ACCEPTED MANUSCRIPT
small changes seem to affect the docking results of our ligands since the imidazolones in this study
structurally resemble epalrestat, which certainly interacts with this pocket.
The optimized structures and atom charges of the selected ligands for docking simulations were
calculated by the semi-empirical PM7 calculation method with the COSMO solvation model [53-55].
The above X-ray crystal structure obtained with 6k was adopted as the initial geometry for each ligand
calculation, and the dielectric constant of water was used to calculate the properties of the ligands in
water. Density Functional Theory (DFT) calculations [56, 57] were also performed at the
B3LYP/6-31G(d) level [58-61] to obtain the optimized structures of the ligands, but in all cases the
optimized geometries were such that the aryl group was almost planar to the exomethylene moiety. This
conformation certainly gives rise to a collision between the vinyl proton and the ortho proton on the aryl
group, showing that the geometries optimized by the DFT calculations are impractical. Therefore, the
PM7 results were adopted for the following docking study.
All the docking experiments were performed using the GOLD Suite software package [62, 63].
ChemPLP scoring function [64] was used to predict the docking poses, and the obtained poses were
rescored with Astex Statistical Potential (ASP) [65], known as a good scoring function for proteins
required to make specific hydrogen bonds with ligands.[66] The best pose for each ligand was
determined by considering the results obtained with both of these scoring functions (both ChemPLP and
ASP give positive values, with the higher score indicating the more likely pose).
Table 2 shows the docking results of selected imidazolones (1a–j, 1m, 2, and 6a), expressed in terms
of ChemPLP fitness scores. In order to assess whether these results match the experimental data, the
coefficients of determination (r2) between these fitness scores and the pIC50 values were calculated and
the results are compiled in Table 2.
The docking into proteins in the “open” conformation, i.e. 1Z3N and 2FZD, gave relatively low
fitness scores and the correlation between these scores and the pIC50 values were very poor (r2 of 0.018
and 0.143, respectively). Hence, it is possible to postulate that above compounds are not capable of
opening the specificity pocket, and might better fit into a protein with the specificity pocket “closed”.
Almost all compounds showed the best fitness score with the 1PWM structure. However, some of the
docking poses in 1PWM are inconsistent with the experimental results. For instance, the best score was
obtained for 1h, whose inhibitory activity was about two-fold less than that of 1i. Moreover, compound
1b exhibited ten-fold lower activity, but better docking score than 1a. Furthermore, when we examined
the r2 values, the correlation between the pIC50 values and the docking scores for the protein-ligand
8

complexes was poor for 1PWM (r2 = 0.367), but, on the other hand, that was strong for 4IGS (r2 =
ACCEPTED MANUSCRIPT
0.908).
Table 2.a ChemPLP fitness scores for the docking experiments with thirteen selected compounds into
seven different protein structures obtained from the Protein Data Bank (PDB) compiled with the
coefficients of determination (r2) between pIC50 values and ChemPLP
Protein structure (state of the specificity pocket)
Compound
1PWM
(closed)
1Z3N
(open)
2FZD
(open)
2NVC
(closed)
2NVD
(closed)
4IGS
(closed)
4JIR
(closed)
1a
1b
1c
1d
1e
1f
147.9677 121.0379 118.4920 122.6749 106.3769 132.4328 122.4033
148.8442 116.7325 122.4086 132.3547 110.9174 127.3476 118.8936
146.4784 120.4499 119.1567 123.3144 106.6123 131.2452 122.7660
149.7175 123.0365 117.1841 126.5613 107.9176 134.6136 123.0212
148.6791 122.3372 116.6743 124.8679 106.1159 136.5031 122.2602
148.3229 122.1758 116.4419 126.7387 106.9029 136.0539 123.6788
148.3499 121.7783 116.4413 126.0770 106.8894 135.7583 123.7556
163.8831 125.6343 125.1224 135.8675 112.0374 145.9120 128.9951
167.0596 126.2255 123.2132 143.3077 114.4054 135.9552 129.1376
159.7231 123.7278 118.3947 132.4810 112.8971 142.0384 128.7544
153.6158 120.0431 119.0217 136.7273 109.7848 135.6587 124.9088
138.4267 134.5372 118.1440 129.4450 96.3939 121.5386 119.7465
145.5941 113.2006 113.0078 119.7632 102.9124 118.6954 112.2170
1g
1h
1i
1j
1m
2
6a
r2
0.367
0.018
0.385
0.143
0.143
0.112
0.270
0.436
0.319
0.908
0.211
0.749
r2 (flexible)b 0.038
<0.001
a Values for ligands that were docked into the binding site of the protein with a “flipped” conformation
are written in italics. b Coefficients of determination between the ChemPLP docking scores obtained by
flexible docking simulations and the experimental pIC50 values.
Closer examination of the docking poses for compounds 1a and 1c–j, which exhibited
sub-micromolar IC50 values, revealed that the orientation of the carboxylates at the N1 position obtained
with the 4IGS structure were different from those obtained with the 1PWM. In both structures, one
carboxylate oxygen atom forms a hydrogen bond with His110; the other one, however, is orientated
toward Trp111 (in 4IGS) or toward Trp20 (in 1PWM), as exemplified by 1c in Figure 4. We then
consulted the Protein Data Bank (PDB) and found 31 ALR2 X-ray crystal structures, in which
carboxylate-type ligands are incorporated. When we compared the predicted carboxylate orientations of
our compounds to those observed in the PDB database, we found that the poses of the carboxylate
groups obtained for 4IGS structure are similar to those predominantly found in the experimental crystal
structures. It is therefore conceivable that the docking poses calculated using the 1PWM structure are
inadequate to explain the experimental data and the docking poses with the 4IGS structure are indeed
9

possible.
ACCEPTED MANUSCRIPT
Figure 4. Depiction of the predicted binding modes for 1a and 1c–j docked into (a) 1PWM and (b) 4IGS
exemplified by 1c. Graphics were created by the Discovery Studio Visualizer.[67]
Further inspection of the docking poses for 1a and 1c–j in 4IGS revealed that all of these poses
superimpose well with each other, except for 1i, and the 4-arylidene scaffolds are placed parallel to
Trp219, forming π-π interactions (Figure 5). Among these compounds, the most potent 1h shows the
best overlap between the naphthyl residue and the indole ring of Trp219. These results suggest that the
hydrophobic interaction of the aromatic rings with Trp219, as well as the hydrogen-bonding interactions
in the anion-binding pocket, plays an important role in ligands binding to ALR2 described in the present
study. The aformentioned interactions for 1h have been visualized using the Discovery Studio Visualizer
program [67] as shown in Figure 6. The program detects a hydrogen-bond donor region in the protein
close to the carboxylate group of 1h, an aromatic face-to-face interaction region around the Trp219, and
a lipophilic contact around Leu300 near the naphthyl residue of 1h.
Figure 5. Superposition of the predicted docking poses for 1a and 1c–j docked into the binding site of
4IGS. All ligands are colored in cyan except 1i in magenta.
10

ACCEPTED MANUSCRIPT
(a)
(b)
(c)
Figure 6. The contour maps of (a) hydrogen-bonding, (b) aromatic, and (c) hydrophobic interaction
surfaces of the binding-site residues in 4IGS superposed with the predicted docking pose for compound
1h. All surfaces were detected and visualized by the Discovery Studio Visualizer [67].
The program SuperStar, which empirically generates interaction maps within protein binding sites
based on experimental crystal structure databases such as the Cambridge Structural Database (CSD) and
PDB [68-70], was used to visualize the knowledge-based protein-ligand interaction area within the
active site of 4IGS. Figure 7 depicts the PDB-based SuperStar propensity maps of the 4IGS binding site,
in which the hydrophobic interaction area (magenta) was detected using the aliphatic CH probe and the
hydrogen-bonding donor region (yellow) was simulated with the C=O probe. Overlapping the predicted
docking pose of 1h into this figure clearly shows that the carboxylate group and the naphthyl ring in the
ligand correlates well with the areas of predicted interaction propensity. Based on these observations, the
inhibitory activity of 1c–h, whose docking modes are quite similar and superimposable to each other, is
likely to vary because of difference in the degree of lipophilic contact between the aryl moiety and the
11

enzyme hydrophobic pocket (Trp219 and Leu300). In order to assess this consideration quantitatively,
ACCEPTED MANUSCRIPT
the correlation between the pIC50 values and the hydrophobicity parameter for the aryl groups of 1c–h
(π) was examined, and the following equation was obtained with a high correlation coefficient.
pIC50 = 5.062 + 0.589 π (r2 = 0.900)
(1)
This result clearly demonstrates that the experimental inhibitory activity of 1c–h can be described by the
hydrophobicity of the 4-arylidene moiety. There might be a criticism that the biggest term in this
equation is the constant value. As for this point, there have been some quantitative structure-activity
relationship (QSAR) models for aldose reductase inhibitors reported with larger constant values than
that in the equation 1. For example, Soni and Kaskhedikar reported a QSAR model for
2,4-dioxo-5-(naphtha-2-ylmethylene)-3-thiazolidinyl acetic acids and 2-thioxo analogues as follows:
pIC50 = 6.634 + 0.504 Σπ + 0.659 Ix (r2 = 0.726)
where Σπ represents the sum of hydrophobicity of the substituents on the naphthyl group and Ix is the
indicator variable for the presence of sulfur atom on the C2 position [71]. More recently, Nantasenamat
et al. reported a QSAR model with a much larger constant value (6.766) and much smaller coefficients
of explanatory valuables (0.097–0.332) to predict aldose reductase inhibitory activity of
sulfonylpyridazinones [72]. Thus one can safely say that the equation 1 is reasonable.
Compound 1a was excluded from this assessment because an additional and hydrogen-bonding
interaction with the backbone NH of Leu301 was predicted in addition to the interactions with the anion
binding site and Trp219. This additional interaction is likely to raise the bioactivity, as discussed above.
The 1-naphthyl group of 1i also faces to Trp219, but the area forming the π-π interaction is smaller
than that of 1h and similar to that of 1b. Moreover, because of its shape, the position of 1i within the
protein slightly deviates from those of the other derivatives. These might be the reasons behind the low
activity of 1i.
12

ACCEPTED MANUSCRIPT
Figure 7. PDB-based SuperStar propensity map of the 4IGS binding site using the aliphatic CH
(magenta) and C=O (yellow) probes accompanied with the predicted docking pose of 1h.
The docking pose of the ester 6a, which exhibited about 60-times less activity (IC50 = 23 µM) than the
corresponding carboxylic acid 1a (IC50 = 0.36 µM), with 4IGS was quite different from those of 1c–j.
The orientation of the imidazolone and phenyl rings is also flipped in comparison to the ones of 1c–j,
and the phenolic hydroxyl group is placed in the anion-binding pocket to form a hydrogen bond with
His110 (Figure 8a). A similar “flipped” pose was obtained for compound 2 (IC50 = 10.7 µM), which
exhibits low inhibitory activity despite the presence of a carboxyl group. As shown in Figure 8b, the
anionic carboxylate group in 2 does not bind to the catalytic pocket of 4IGS, but instead makes a
hydrogen bond with the backbone NH of Ser302. In contrast, the phenol moiety of 2 is inserted into the
anion-binding pocket and forms a hydrogen bond with His110, the carbonyl group at the C5 position of
the imidazolone ring forms a hydrogen bond with the backbone NH of Leu300, and the imidazolone ring
faces to Trp219. It is conceivable that these additional interactions in part improve the bioactivity of 2
compared to that of 6a, but its inhibitory effect is still lower than those of the carboxylic acid series 1,
probably because of the loss of the tight binding of the carboxylate with the catalytic pocket. The
docking pose for 1m in 4IGS was also solved as the “flipped” one (Figure 8c). In this pose, the
carboxylate group of the glycine unit at the N1 position of the imidazolone ring points toward the
outside of the protein, but instead the other carboxylate group at the para position of the phenyl ring
forms a tight net of hydrogen bonds with Tyr48, His110 and Trp111 (the anion-binding site), providing a
possible explanation for its good efficacy (IC50 = 0.31 µM).
For what concerns compound 1b, a meta analogue of 1a, we found a pose in which the carboxyl
group was incorporated into the catalytic pocket through a hydrogen bond with His110, and the
13

meta-hydroxyl group formed hydrogen bonds with the backbone NH of Ala299 and Leu300 (Figure 8d).
ACCEPTED MANUSCRIPT
However, the entire pose was not superimposable with those obtained for 1c–j. The amide carbonyl
group at the C5 positions of the imidazolone ring in 1b was flipped compared to 1c–j, and thus, the
hydrogen bond with Cys298, commonly observed in the poses of 1c–j, was lost (Figure S1). Moreover,
the predicted conformation of 1b in 4IGS was not identical to the optimized geometry (Figure S2),
namely the phenyl ring was rotated almost 180° in the docking experiments. According to a simple
molecular mechanics calculation, the rotation barrier was estimated to be 2.0 kcal/mol, showing that
extra energy is required for 1b to be incorporated into the binding site with this particular pose. These
results suggest that the binding of 1b with the protein is energetically unfavorable compared to the
others of series 1, and this may explain the low activity of 1b.
(a)
(b)
(c)
(d)
Figure 8. Predicted binding modes for (a) 6a, (b) 2, (c) 1m, and (d) 1b docked into 4IGS. Graphics were
created by the Discovery Studio Visualizer [67].
Since the ALR2 binding site shows a high degree of flexibility to incorporate diverse structures of
carbonyl substrates, we also performed induced-fit docking simulations [23, 73, 74] using the Side
14

Chain Flexibility function in GOLD. The same five protein structures described in Table 2 were used.
ACCEPTED MANUSCRIPT
The results indicate that the docking scores were not drastically improved, and that the correlation
between the docking scores and pIC50 values was poor (r2 < 0.4) in all cases (Table 2). This result
suggests that the predicted docking poses using the 4IGS structure with no side chains flexibility
account for the present experimental results best.
By considering all these data, we conclude that correlating the docking results with the experimental
IC50 values is a valid approach.
3. Conclusion
We have demonstrated that some (Z)-4-Arylmethylene-1H-imidazol-5(4H)-ones related to the GFP
chromophore act as good ARIs with the IC50 values less than 0.36 µM. Among these new GFP
chromophore models, 1h exhibited the best inhibitory activity with an IC50 of 0.10 µM, supporting
further work on evaluating its in vivo efficacy. The structure-activity relationship study indicated the
important structural features needed to the GFP chromophore model series to exhibit high ALR2
inhibitory activity. These include the carboxylate group at the N1 position and the hydrophobicity of the
4-arylmethylidene moiety.
We have explored a way to obtain reliable docking results by correlating the docking scores with the
IC50 values. In our docking study, the obtained docking poses for the imidazolones in the 1PWM
structure gave the best fitness scores, but exhibited poor correlation with the efficacy of the ligands. The
docking scores obtained using the 4IGS structure, on the other hand, correlated well with the
experimental inhibitory activity. The predicted docking poses with 4IGS suggest that the
4-arylmethylidene moiety forms a π-π interaction with the Trp219 residue as well as lipophilic contact
with Leu300, rationally explaining the importance of the hydrophobicity. In fact, a linear correlation was
observed between the hydrophobicity of the 4-arylmethylidene moiety and the experimental inhibitory
activity.
Some reports claimed the usefulness of induced-fit docking (IFD) methods to simulate the
induced-fit-type movement of the binding site of ALR2 in accommodating substrates with diverse
structures [23, 74]. Although this approach seems to be rational for modeling the complex of inhibitors
with the induced-fit-type protein, none of the examples demonstrated a correlation between the docking
scores and the experimental inhibitory activity in a quantitative fashion. In our system, the IFD
simulation did not yield any satisfactory results that would rationally explain the experimental data,
while the fully knowledge-based docking method in which the experimentally obtained protein
15

structures with different active site conformation were used, yielded scores and poses that accounted for
ACCEPTED MANUSCRIPT
the experimental results.
By considering these results comprehensively, we conclude that it is advisable to correlate between
the docking results with the experimental IC50 values quantitatively, in order to gain reliable information
of ligand-protein interactions, independently of the docking method used.
Acknowledgement
This research was supported by a Grant-in-Aid for Scientific Research C (No. 25410179) from the Japan
Society for the Promotion of Science (JSPS). Partial financial support to R.S. by the Nukada Memorial
Scholarship Fund of Toho University is also gratefully appreciated. The authors thank Dr. Yoichi
Habata and Dr. Shunsuke Kuwahara for carrying out the X-ray diffraction experiment and refinement of
the data.
Conflicts of interest
The authors declare no other conflicts of interest.
Appendix. Supplementary data
Supplementary
data
related
to
this
article
can
be
found
at
http://
dx.doi.org/10.1016/j.ejmech.xxxx.xx.xxx
References
[1] M. Brownlee, Biochemistry and molecular cell biology of diabetic complications, Nature 414 (2001)
813–820.
[2] Z. Changjin, Aldose reductase inhibitors as potential therapeutic drugs of diabetic complications, in:
O. Oguntibeju (Ed.), Diabetes mellitus - Insights and perspectives, InTech, 2013, pp. 17–46.
[3] K.H. Gabbay, N. Spack, S. Loo, H.J. Hirsch, A.A. Ackil, Aldose reductase inhibition: Studies with
alrestatin, Metabolism 28 (1979) 471–476.
[4] P.F. Kador, J.H. Kinoshita, N.E. Sharpless, Aldose reductase inhibitors - A potential new class of
agents for the pharmacological control of certain diabetic complications, J. Med. Chem. 28 (1985) 841–
849.
[5] D. Dvornik, D. Porte, Aldose reductase inhibition: An approach to the prevention of diabetic
complications, Biomedical Information Corp., 1987.
[6] T. Tanimoto, K. Maekawa, S. Okada, C. Yabe-Nishimura, Clinical analysis of aldose reductase for
differential diagnosis of the pathogenesis of diabetic complication, Anal. Chim. Acta 365 (1998) 285–
292.
[7] N. Hotta, T. Toyota, K. Matsuoka, Y. Shigeta, R. Kikkawa, T. Kaneko, A. Takahashi, K. Sugimura, Y.
Koike, J. Ishii, N. Sakamoto, S.D.N.S. Group, Clinical efficacy of fidarestat, a novel aldose reductase
16

inhibitor, for diabetic peripheral neuropathy - A 52-week multicenter placebo-controlled double-blind
ACCEPTED MANUSCRIPT
parallel group study, Diabetes Care 24 (2001) 1776–1782.
[8] M. Lorenzi, The polyol pathway as a mechanism for diabetic retinopathy: Attractive, elusive, and
resilient, Exp. Diabetes Res. (2007) Article ID 61038, 61010 pages. DOI:10.1155/2007/61038.
[9] N. Hotta, New approaches for treatment in diabetes: Aldose reductase inhibitors, Biomed.
Pharmacother. 49 (1995) 232–243.
[10] C. Yabe-Nishimura, Aldose reductase in glucose toxicity: A potential target for the prevention of
diabetic complications, Pharmacol. Rev. 50 (1998) 21–33.
[11] L. Costantino, G. Rastelli, P. Vianello, G. Cignarella, D. Barlocco, Diabetes complications and their
potential prevention: Aldose reductase inhibition and other approaches, Med. Res. Rev. 19 (1999) 3–23.
[12] M.A. Pfeifer, M.P. Schumer, Clinical-trials of diabetic neuropathy - Past, present, and future,
Diabetes 44 (1995) 1355–1361.
[13] L. Costantino, G. Rastelli, G. Cignarella, P. Vianello, D. Barlocco, New aldose reductase inhibitors
as potential agents for the prevention of long–term diabetic complications, Expert Opin. Ther. Pat. 7
(1997) 843–858.
[14] L. Costantino, G. Rastelli, M.C. Gamberini, D. Barlocco, Pharmacological approaches to the
treatment of diabetic complications, Expert Opin. Ther. Pat. 10 (2000) 1245–1262.
[15] N. Hotta, N. Sakamoto, Y. Shigeta, R. Kikkawa, Y. Goto, Clinical investigation of epalrestat, an
aldose reductase inhibitor, on diabetic neuropathy in japan: Multicenter study (reprinted from excerpta
medica international congress, series 1084, 1995), J. Diabetes Complicat. 10 (1996) 168–172.
[16] S. Miyamoto, Recent advances in aldose reductase inhibitors: Potential agents for the treatment of
diabetic complications, Expert Opin. Ther. Pat. 12 (2002) 621–631.
[17] M.A. Ramirez, N.L. Borja, Epalrestat: An aldose reductase inhibitor for the treatment of diabetic
neuropathy, Pharmacotherapy 28 (2008) 646-655.
[18] A. Ishii, T. Kotani, Y. Nagaki, Y. Shibayama, Y. Toyomaki, N. Okukado, K. Ienaga, K. Okamoto,
Highly selective aldose reductase inhibitors. 1. 3-(arylalkyl)-2,4,5-trioxoimidazolidine-1-acetic acids, J.
Med. Chem. 39 (1996) 1924–1927.
[19] T. Negoro, P. Murata, S. Ueda, B. Fujitani, Y. Ono, A. Kuromiya, M. Komiya, K. Suzuki, J.
Matsumoto,
Novel,
highly
potent
aldose
reductase
inhibitors:
(R)-(–)-2-(4-bromo-2-fluorobenzyl)-1,2,3,4-tetrahydropyrrolo[1,2-α]pyrazine-4-spiro-3'-pyrrolidine-1,2',
3,5'-tetrone (as-3201) and its congeners, J. Med. Chem. 41 (1998) 4118–4129.
[20] G. Rastelli, P. Vianello, D. Barlocco, L. Costantino, A. DelCorso, U. Mura, Structure-based design
of an inhibitor modeled at the substrate active site of aldose reductase, Bioorg. Med. Chem. Lett. 7
(1997) 1897–1902.
[21] M. Oka, Y. Matsumoto, S. Sugiyama, N. Tsuruta, M. Matsushima, A potent aldose reductase
inhibitor, (2S,4S)-6-fluoro-2',5'-dioxospiro[chroman-4,4'-imidazolidine]-2-carboxamide (fidarestat): Its
absolute configuration and interactions with the aldose reductase by x-ray crystallography, J. Med.
Chem. 43 (2000) 2479–2483.
[22] S. Miyamoto, Molecular modeling and structure-based drug discovery studies of aldose reductase
inhibitors, Chem-Bio Inform. J. 2 (2002) 74–85.
[23] R. Maccari, R. Ottana, R. Ciurleo, D. Rakowitz, B. Matuszczak, C. Laggner, T. Langer, Synthesis,
induced-fit docking investigations, and in vitro aldose reductase inhibitory activity of non-carboxylic
acid containing 2,4-thiazolidinedione derivatives, Bioorg. Med. Chem. 16 (2008) 5840–5852.
[24] H. Morise, O. Shimomura, F.H. Johnson, J. Winant, Intermolecular energy-transfer in
bioluminescent system of aequorea, Biochemistry 13 (1974) 2656–2662.
[25] O. Shimomura, Structure of the chromophore of aequorea green fluorescent protein, Febs Lett. 104
(1979) 220–222.
17

[26] C.W. Cody, D.C. Prasher, W.M. Westler, F.G. Prendergast, W.W. Ward, Chemical-structure of the
ACCEPTED MANUSCRIPT
hexapeptide chromophore of the aequorea green-fluorescent protein, Biochemistry 32 (1993) 1212–
1218.
[27] H. Niwa, S. Inouye, T. Hirano, T. Matsuno, S. Kojima, M. Kubota, M. Ohashi, F.I. Tsuji, Chemical
nature of the light emitter of the aequorea green fluorescent protein, Proc. Natl. Acad. Sci. U.S.A. 93
(1996) 13617–13622.
[28] S. Kojima, H. Ohkawa, T. Hirano, S. Maki, H. Niwa, M. Ohashi, S. Inouye, F.I. Tsuji, Fluorescent
properties of model chromophores of tyrosine-66 substituted mutants of aequorea green fluorescent
protein (GFP), Tetrahedron Lett. 39 (1998) 5239–5242.
[29] P. Busca, F. Paradisi, E. Moynihan, A.R. Maguire, P.C. Engel, Enantioselective synthesis of
non-natural amino acids using phenylalanine dehydrogenases modified by site-directed mutagenesis,
Org. Biomol. Chem. 2 (2004) 2684–2691.
[30] K.M. Khan, U.R. Mughal, M.T.H. Khan, U. Zia, S. Perveen, M. Iqbal Choudhary, Oxazolones:
New tyrosinase inhibitors; synthesis and their structure–activity relationships, Bioorg. Med. Chem. 14
(2006) 6027–6033.
[31] H.N.C. Wong, Z.L. Xu, H.M. Chang, C.M. Lee, A modified synthesis of (±)-β-aryllactic acids,
Synthesis 1992 (1992) 793–797.
[32] H. Hoshina, H. Tsuru, K. Kubo, T. Igarashi, T. Sakurai, Formation of isoquinoline derivatives by the
irradiation of N-acetyl-α-dehydrophenylalanine ethyl ester and its derivatives, Heterocycles 53 (2000)
2261–2274.
[33] K. Takrouri, T. Chen, E. Papadopoulos, R. Sahoo, E. Kabha, H. Chen, S. Cantel, G. Wagner, J.A.
Halperin, B.H. Aktas, M. Chorev, Structure–activity relationship study of 4EGI-1, small molecule
eIF4E/eIF4G protein–protein interaction inhibitors, Eur. J. Med. Chem. 77 (2014) 361–377.
[34] M. Blanco-Lomas, I. Funes-Ardoiz, P.J. Campos, D. Sampedro, Oxazolone-based photoswitches:
Synthesis and properties, Eur. J. Org. Chem. 2013 (2013) 6611–6618.
[35] K. Maekawa, T. Igarashi, K. Kubo, T. Sakurai, Electron transfer-initiated photocyclization of
substituted N-acetyl-α-dehydro(1-naphthyl)alanines to 1,2-dihydrobenzo[f]quinolinone derivatives:
Scope and limitations, Tetrahedron 57 (2001) 5515–5526.
[36] M. Bhalla, P.K. Naithani, T.N. Bhalla, A.K. Saxena, K. Shanker, Novel imidazole congeners as
antiinflammatory agents, J. Indian Chem. Soc. 69 (1992) 594–595.
[37] V. Voliani, R. Bizzarri, R. Nifosi, S. Abbruzzetti, E. Grandi, C. Viappiani, F. Beltram, Cis-trans
photoisomerization of fluorescent-protein chromophores, J. Phys. Chem. B 112 (2008) 10714–10722.
[38] C. Nishimura, T. Yamaoka, M. Mizutani, K. Yamashita, T. Akera, T. Tanimoto, Purification and
characterization of the recombinant human aldose reductase expressed in baculovirus system, Biochim.
Biophys. Acta. 1078 (1991) 171–178.
[39] C. La Motta, S. Sartini, S. Salemo, F. Simorini, S. Taliani, A.M. Marini, F. Da Settimo, L. Marinelli,
V. Limongelli, E. Novellino, Acetic acid aldose reductase inhibitors bearing a five-membered
heterocyclic core with potent topical activity in a visual impairment rat model, J. Med. Chem. 51 (2008)
3182–3193.
[40] R. Saito, M. Tokita, K. Uda, C. Ishikawa, M. Satoh, Synthesis and in vitro evaluation of
botryllazine b analogues as a new class of inhibitor against human aldose reductase, Tetrahedron 65
(2009) 3019–3026.
[41] A. Del Corso, L. Costantino, G. Rastelli, F. Buono, U. Mura, Aldose reductase does catalyse the
reduction of glyceraldehyde through a stoichiometric oxidation of nadph, Exp. Eye Res. 71 (2000) 515–
521.
[42] R.P. Verma, Cytotoxicity of heterocyclic compounds against various cancer cells: A quantitative
structure–activity relationship study, Top. Heterocycl. Chem. 9 (2007) 53–86.
18

[43] C. Hansch, A. Leo, D.H. Hoekman, Exploring qsar: Fundamentals and applications in chemistry
ACCEPTED MANUSCRIPT
and biology, American Chemical Society, 1995.
[44] K. Kato, K. Nakayama, M. Mizota, I. Miwa, J. Okuda, Properties of novel aldose reductase
inhibitors, M16209 and M16287, in comparison with known inhibitors, ONO-2235 and sorbinil, Chem.
Pharm. Bull. 39 (1991) 1540–1545.
[45] U. Dhagat, S. Endo, A. Hara, O. El-Kabbani, Inhibition of 3(17)α-hydroxysteroid dehydrogenase
(AKR1C21) by aldose reductase inhibitors, Bioorg. Med. Chem. 16 (2008) 3245–3254.
[46] R.F. Rekker, The hydrophobic fragmental constant. Its derivation and application: A means of
characterizing membrane systems, Elsevier Scientific Pub. Co.: distributors for the U.S. and Canada,
Elsevier/North Holland, Amsterdam; New York, 1977.
[47] C.A. Sotriffer, O. Kramer, G. Klebe, Probing flexibility and "induced-fit" phenomena in aldose
reductase by comparative crystal structure analysis and molecular dynamics simulations, Proteins 56
(2004) 52–66.
[48] H. Steuber, M. Zentgraf, C. La Motta, S. Sartini, A. Heine, G. Klebe, Evidence for a novel binding
site conformer of aldose reductase in ligand-bound state, J. Mol. Biol. 369 (2007) 186–197.
[49] R. Maccari, R. Ottanà, R. Ciurleo, M.G. Vigorita, D. Rakowitz, T. Steindl, T. Langer, Evaluation of
in vitro aldose redutase inhibitory activity of 5-arylidene-2,4-thiazolidinediones, Bioorg. Med. Chem.
Lett. 17 (2007) 3886–3893.
[50] X. Chen, C. Zhu, F. Guo, X. Qiu, Y. Yang, S. Zhang, M. He, S. Parveen, C. Jing, Y. Li, B. Ma,
Acetic acid derivatives of 3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxide as a novel class of potent
aldose reductase inhibitors, J. Med. Chem. 53 (2010) 8330–8344.
[51] L. Zhang, H. Zhang, Y. Zhao, Z. Li, S. Chen, J. Zhai, Y. Chen, W. Xie, Z. Wang, Q. Li, X. Zheng, X.
Hu, Inhibitor selectivity between aldo–keto reductase superfamily members AKR1B10 and AKR1B1:
Role of Trp112 (Trp111), Febs Lett. 587 (2013) 3681–3686.
[52] A. Cousido-Siah, F.X. Ruiz, A. Mitschler, S. Porte, A.R. de Lera, M.J. Martin, S. Manzanaro, J.A.
de la Fuente, F. Terwesten, M. Betz, G. Klebe, J. Farres, X. Pares, A. Podjarny, Identification of a novel
polyfluorinated compound as a lead to inhibit the human enzymes aldose reductase and akr1b10:
Structure determination of both ternary complexes and implications for drug design, Acta Crystallogr D
70 (2014) 889–903.
[53] J.J.P. Stewart, Optimization of parameters for semiempirical methods vi: More modifications to the
nddo approximations and re-optimization of parameters, J. Mol. Model. 19 (2013) 1–32.
[54] A. Klamt, G. Schuurmann, Cosmo: A new approach to dielectric screening in solvents with explicit
expressions for the screening energy and its gradient, J. Chem. Soc., Perkin Trans. 2 (1993) 799–805.
[55] R. Saito, T. Hirano, S. Maki, H. Niwa, Synthesis and chemiluminescent properties of
6,8-diaryl-2-methylimidazo[1,2-α]pyrazin-3(7H)-ones: Systematic investigation of substituent effect at
para-position of phenyl group at 8-position, J. Photochem. Photobiol. A-Chem. 293 (2014) 12–25.
[56] M.W. Schmidt, K.K. Baldridge, J.A. Boatz, S.T. Elbert, M.S. Gordon, J.H. Jensen, S. Koseki, N.
Matsunaga, K.A. Nguyen, S.J. Su, T.L. Windus, M. Dupuis, J.A. Montgomery, General atomic and
molecular electronic-structure system, J. Comput. Chem. 14 (1993) 1347–1363.
[57] F. Jensen, Introduction to computational chemistry, 2nd ed., John Wiley and Sons, Ltd., Wast
Sussex, UK, 2007.
[58] C.T. Lee, W.T. Yang, R.G. Parr, Development of the colle-salvetti correlation-energy formula into a
functional of the electron-density, Phys. Rev. B 37 (1988) 785–789.
[59] A.D. Becke, Density-functional thermochemistry. 3. The role of exact exchange, J. Chem. Phys. 98
(1993) 5648–5652.
[60] P.J. Stephens, F.J. Devlin, C.F. Chabalowski, M.J. Frisch, Ab-initio calculation of vibrational
absorption and circular-dichroism spectra using density functional force fields, J. Phys. Chem. 98 (1994)
19

11623–11627.
ACCEPTED MANUSCRIPT
[61] R. Hertwig, W. Koch, On the parameterization of the local correlation functional. What is
Becke-3-LYP?, Chem. Phys. Lett. 268 (1997) 345–351.
[62] Gold Version 5.2; The Cambridge Crystallographic Data Centre (CCDC): Cambridge, UK.
http://www.ccdc.cam.ac.uk.
[63] M.L. Verdonk, J.C. Cole, M.J. Hartshorn, C.W. Murray, R.D. Taylor, Improved protein–ligand
docking using gold, Proteins 52 (2003) 609–623.
[64] O. Korb, T. Stützle, T.E. Exner, Empirical scoring functions for advanced protein-ligand docking
with plants, J. Chem. Inf. Model. 49 (2009) 84–96.
[65] W.T.M. Mooij, M.L. Verdonk, General and targeted statistical potentials for protein-ligand
interactions, Proteins 61 (2005) 272–287.
[66] Cambridge Crystallographic Data Centre, Gold scoring function performance against the dud
decoy/active set. https://www.ccdc.cam.ac.uk/support-and-resources/ccdcresources/VS_workcase.pdf,
2012 (14.09.16).
[67] Discovery studio visualizer Release 4.1; Dassault Systèmes BIOVIA: San Diego, USA.
http://accelrys.com/products/collaborative-science/biovia-discovery-studio/visualization.html.
[68] M.L. Verdonk, J.C. Cole, R. Taylor, Superstar: A knowledge-based approach for identifying
interaction sites in proteins, J. Mol. Biol. 289 (1999) 1093–1108.
[69] M.L. Verdonk, J.C. Cole, P. Watson, V. Gillet, P. Willett, Superstar: Improved knowledge-based
interaction fields for protein binding sites1, J. Mol. Biol. 307 (2001) 841–859.
[70] D.R. Boer, J. Kroon, J.C. Cole, B. Smith, M.L. Verdonk, Superstar: Comparison of csd and
pdb-based interaction fields as a basis for the prediction of protein-ligand interactions1, J. Mol. Biol.
312 (2001) 275–287.
[71] L.K. Soni, S.G. Kaskhedikar, Exploring structural requirements for aldose-reductase inhibition by
2,4-dioxo-5-(naphth-2-ylmethylene)-3-thiazolidinyl acetic acids and 2-thioxo analogues: Fujita-ban and
hansch approach, Arch. Pharm. 339 (2006) 327–331.
[72] C. Nantasenamat, T. Monnor, A. Worachartcheewan, P. Mandi, C. Isarankura-Na-Ayudhya, V.
Prachayasittikul, Predictive qsar modeling of aldose reductase inhibitors using monte carlo feature
selection, Eur. J. Med. Chem. 76 (2014) 352–359.
[73] M. Zentgraf, H. Steuber, C. Koch, C. La Motta, S. Sartini, C.A. Sotriffer, G. Klebe, How reliable
are current docking approaches for structure-based drug design? Lessons from aldose reductase, Angew.
Chem. Int. Ed. 46 (2007) 3575–3578.
[74] S.V. Jain, K.S. Bhadoriya, S.B. Bari, Qsar and flexible docking studies of some aldose reductase
inhibitors obtained from natural origin, Med. Chem. Res. 21 (2012) 1665–1676.
20

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT