Computer-assisted designed "selenoxy-chinolin": A new catalytic mechanism of the GPx-like cycle and inhibition of metal-free and metal-associated Aβ aggregation

Article abstract of DOI:10.1039/c5dt02130h

Using support from rational computer-assisted design, a novel series of hybrids (selenoxy-chinolin) designed by fusing the metal-chelating agent CQ and the antioxidant ebselen were synthesized and evaluated as multitarget-directed ligands. Most of the hybrids demonstrated significant ability to mimic GPx, which is highly consistent with the prediction results of DFT studies for the selenenyl sulfide intermediates in the computational design. Using ⁷⁷Se, ¹H and ¹³C NMR spectroscopy and high-resolution mass spectroscopy (HRMS), a novel catalytic mechanism, including a new selenium quinone active species, was first demonstrated. 2D NMR studies indicated that the typical hybrid has an effective interaction with Aβ. In addition, the optimal compound 12k was found to possess an excellent ability to scavenge peroxide and to inhibit self- and metal-induced Aβ aggregation, and an ability to disassemble preformed self- and metal-induced Aβ aggregates effectively. Furthermore, 12k was able to penetrate the central nervous system (CNS) and did not exhibit any acute toxicity in mice at doses up to 2000 mg kg^-1. Overall, we demonstrated that hybrid 12k, through rational structure-based computational design, shows a potential for development as a therapeutic agent in AD.

Full text of DOI:10.1039/c5dt02130h

Dalton
Transactions
View Article Online
View Journal | View Issue
PAPER
Computer-assisted designed “selenoxy–chinolin”:
a new catalytic mechanism of the GPx-like cycle
and inhibition of metal-free and metal-associated
Aβ aggregation†
Cite this: Dalton Trans., 2015, 44,
20913
Zhiren Wang, Yali Wang, Wenrui Li, Zhihong Liu, Zonghua Luo, Yang Sun, Ruibo Wu,
Ling Huang* and Xingshu Li*
Using support from rational computer-assisted design, a novel series of hybrids (selenoxy–chinolin)
designed by fusing the metal-chelating agent CQ and the antioxidant ebselen were synthesized and eval-
uated as multitarget-directed ligands. Most of the hybrids demonstrated significant ability to mimic GPx,
which is highly consistent with the prediction results of DFT studies for the selenenyl sulfide intermediates
1
in the computational design. Using 77Se, H and 13C NMR spectroscopy and high-resolution mass spec-
troscopy (HRMS), a novel catalytic mechanism, including a new selenium quinone active species, was first
demonstrated. 2D NMR studies indicated that the typical hybrid has an effective interaction with Aβ. In
addition, the optimal compound 12k was found to possess an excellent ability to scavenge peroxide and
to inhibit self- and metal-induced Aβ aggregation, and an ability to disassemble preformed self- and
metal-induced Aβ aggregates effectively. Furthermore, 12k was able to penetrate the central nervous
system (CNS) and did not exhibit any acute toxicity in mice at doses up to 2000 mg kg−1. Overall, we
demonstrated that hybrid 12k, through rational structure-based computational design, shows a potential
for development as a therapeutic agent in AD.
Received 5th June 2015,
Accepted 26th October 2015
DOI: 10.1039/c5dt02130h
www.rsc.org/dalton
which mimic the function of the antioxidant selenoenzyme
GPx, have been developed for the prevention of diseases such
Introduction
Glutathione (GSH), a major endogenous enzyme-catalyzed as AD and Parkinson’s disease (PD).7–14 Ebselen, a classic
antioxidant which is present at high concentrations of 1–2 mM example of a GPx mimic, exhibits numerous biological activi-
within the brain, plays a fundamental role in the detoxification ties and has undergone phase III clinical trials as an anti-
of ROS and regulates the intracellular redox environment.1–3 inflammatory agent (Fig. 1).15,16 In recent years, many ebselen
Postmortem human studies have demonstrated that the mRNA analogs or derivatives have been designed and studied for
levels of GPx, a mammalian selenoenzyme, which protects development as highly efficient GPx mimics.17–26
various organisms from oxidative damage by catalyzing the
According to the “amyloid hypothesis”, the production and
reduction of harmful peroxides in the presence of GSH or accumulation of oligomeric aggregates of Aβ in the brain are
other thiol cofactor systems, was elevated in the hippocampus central events in the pathogenesis of AD, and they are thought
and inferior parietal lobe of Alzheimer’s disease (AD) to initiate the pathogenic cascade, ultimately leading to neuro-
patients.4–6 Small-molecule organoselenium compounds, nal loss and dementia.27,28 Studies have indicated that remark-
ably high levels of biometals such as iron, zinc, and copper,
which can bind to Aβ peptides and promote their aggrega-
School of Pharmaceutical Sciences, Sun Yat-sen University, Guangzhou 510006,
China. E-mail: huangl72@mail.sysu.edu.cn, lixsh@mail.sysu.edu.cn;
Fax: +086-20-3994-3051, +086-20-3994-3050;
Tel: +086-20-3994-3051, +086-20-3994-3050
†Electronic supplementary information (ESI) available: 1H and 13C NMR data of
intermediate and target compounds, the biological assay method, the coupled
reductase assay method, determination of the catalytic parameters, scavenging
of H2O2 by GSH in the presence of the compound, mechanism of the GPx-like
catalytic cycle, computational chemistry, UV spectrum of metal chelation and
tables of results for the PAMPA. See DOI: 10.1039/c5dt02130h
Fig. 1 Structures of clioquinol (PBT1), PBT2 and ebselen.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20913

View Article Online
Paper
Dalton Transactions
Fig. 2 Drug design strategy for multitarget-directed ligands.
tion,29 were found in the brains of AD patients (Cu, 0.4 mM; action (lower positive charge on the selenium) in the selenenyl
Zn and Fe, 1 mM).27 Dysregulated redox-active metal ions sulfide, resulting in the low solvent-assisted proton exchange
bound or unbound to amyloid plaques resulted in concomi- (SAPE) activation barrier,41,42 was favorable for the thiol to
tant free radical generation and reactive oxygen species (ROS) attack the sulfur in the selenenyl sulfide to obtain the selenol
overproduction,30 giving rise to neuronal death.31–36 Thus, the intermediate. On the contrary, due to the strong Se⋯O inter-
modulation of these biometals in the brain has been proposed action (higher positive charge on the selenium), the attack
as a potential AD therapeutic strategy. Clioquinol (CQ, Fig. 1) of the thiol on the selenium in the selenenyl sulfide was
and its second-generation analog, PBT2 (Fig. 1), an 8-hydroxy- more preferred and led to undesired thiol exchange, which
quinoline derivative with moderate affinity for Cu and Zn that was the rate-determining step of the classical GPx cycle
inhibits metal-induced Aβ aggregation and reactive oxygen (Scheme 2).15,21,24,26,43
species generation, have been used for the treatment of AD
and have shown evidence of slowing cognitive deterioration attach to the A ring of the selenenyl sulfides to decrease the
and significantly decreasing plasma Aβ1-42 levels.37–39
Se⋯O interaction and increase the positive charge on the
Therefore, identification of the suitable substituents to
Considering the numerous biological activities of ebselen, sulfur atom was important for strengthening the antioxidant
especially its antioxidant and anti-inflammatory effects that potency of the organoselenium compounds.21 We focused on
are closely related to the etiology of AD,16,40 in this work, with introducing substituent groups at the meta or para positions of
computer-assisted design, we fused the pharmacophores of the A ring for reducing the Se⋯O interaction as many substitu-
CQ and ebselen into one molecule to obtain a “selenoxy– ent groups, including coordinating substituents for seleninate
chinolin” with a high GPx antioxidant ability, significantly esters,44–46 ortho to the selenium on the benzene ring have
inhibiting the activities of self-induced and metal-induced Aβ been reported as GPx mimics.18,19,21,23,24,47
aggregation (Fig. 2). The “selenoxy–chinolin” are, to the best of
DFT calculations were carried out to identify the suitable
our knowledge, the first single compounds obtained from substituents on the selenenyl sulfides, which were the crucial
computer-assisted design that target multipathogenic factors, intermediates in the GPx-like cycle for the hybrids (see
including GPx, Aβ, metal ions and metal–Aβ, and remarkably the ESI†).21,47
improve their effects.
The geometries were fully optimized using calculations at
the B3LYP/6-311+g(d) level, which have been widely used to
study the redox chemistry of small-molecule GPx mimics with
high success.15,17–25,47 The results in Table 1 shows that,
without a substituent group, the selenenyl sulfides have a dis-
tance of 2.605 Å between the oxygen and selenium atoms, and
11.41 kcal mol−1 of O⋯Se orbital interaction energy. When a
chloro group was introduced in the R1 position, a slightly
Results and discussion
Computational design and synthesis
According to the classical GPx-like cyclic mechanism
(Scheme 1), the weak intramolecular nonbonding Se⋯O inter-
strong O⋯Se interaction energy (10.61 kcal mol−1
) was
observed. However, replacement of the chloro group with both
Scheme 1 GPx-like cycle of ebselen (cycle A) and the classical GPx-like
Scheme 2 The rate-determining step for the classical GPx-like catalytic
cycle (cycle B).
cycle.
20914 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Table 1 Summary of DFT calculations and NBO analysis on the selenenyl sulfides at the B3LYP/6-311+G(d, p) level on the B3LYP/6-311+G(d) level-
optimized geometries
δ (77Se NMR)
ESe⋯O
—
R
X
rSe⋯O [Å]
θO⋯Se−S [°]
qSe
qS
(exptl) (ppm)a
[kcal mol−1
]
—
H
Cl
CF3
F
OMe
Cl
H
H
H
H
H
H
H
H
H
H
H
Cl
Ib
2.605
2.619
2.609
2.617
2.615
2.630
2.635
2.616
2.642
2.552
2.511
2.646
2.618
175.89
174.69
174.63
174.47
175.19
174.06
173.80
175.47
175.21
176.11
176.27
173.40
175.40
0.364
0.371
0.372
0.372
0.362
0.365
0.360
0.355
0.359
0.375
0.388
0.351
0.358
−0.032
−0.025
−0.021
−0.027
−0.029
−0.027
−0.026
−0.034
−0.030
−0.047
−0.040
−0.019
−0.032
553
563
564
568
562
556
564
529
529
563
517
524
533
11.41
10.61
11.11
10.75
11.32
9.93
9.90
11.12
8.48
R1
R2
F
OMe
OH
OMe
F
OH
OH
R1 = R2
R3
14.76
16.83
7.28
R2
10.60
a Selenenyl sulfides dissolved in d6-DMSO and Me2Se as external (77Se) standards. b Using Lanl2dz as the basis set for the I atom and 6-311+G(d,
p) for the other atoms.
electron-withdrawing (trifluoromethyl, fluoro) and -donating The natural charge (+0.351) on the selenium was less positive
(methoxy) groups led to a substantial increase in ESe⋯O values and the charge on sulfur (−0.019) was more positive than in
of 11.11, 10.75 and 11.32 kcal mol−1, respectively. A shift of other selenenyl sulfides. In addition, the 77Se NMR chemical
chloro, fluoro or methoxy from the R1 to the R2 position shift of the selenenyl sulfide (77Se NMR: δ 524) in solution was
showed a drop of ESe⋯O to 9.93, 9.90 and 11.12 kcal mol−1
,
significantly highfield compared with others. In contrast,
respectively. The results revealed that substituents at the para replacement of a chloro group with an iodo group generated a
position, regardless of being electron-donating or -withdraw- moderate interaction energy (10.60 kcal mol−1).
ing, resulted in lower interaction than those at the meta posi-
Synthesis
tion of the A ring in selenenyl sulfides. This could be further
demonstrated by a comparison of the 77Se NMR chemical shift
of selenenyl sulfides in solution; the R1 position substituted
selenenyl sulfides (77Se NMR: δ 563, 568, 562 ppm, respect-
ively) resulted in an apparent downfield shift of the 77Se NMR
chemical shift than in the R2 position (77Se NMR: δ 556, 564,
529 ppm, respectively). However, disubstituting the R1 and R2
positions with methoxy groups gave a greater interaction
energy (14.72 kcal mol−1) in comparison with only the R2 posi-
tion being substituted with a methoxy group (11.12 kcal mol−1),
which indicated that the effects of multiple methoxy
groups were not cumulative and the result was in accordance
with seleninate esters.46 It is interesting that the replacement
of the substituent at the para position (R2) with a hydroxyl
group led to a considerable decrease in the interaction energy
(8.48 kcal mol−1). Conversely, a shift of a fluoro group from the
R2 to the R3 position produced a sharp increase in the ESe⋯O
The target compounds 11a–11k, 12k and 13k were obtained by
the route described in Scheme 3. First, compounds 2a–2c and
3a were prepared as reported in the literature.14,48,49 Reduction
of 3a by SnCl2·2H2O gave amine 4a, and reduction of 3b and
3c with Na2S·9H2O in EtOH provided 4b and 4c, respectively.
Na2Se2 was reacted with the corresponding benzenediazonium
value to 16.83 kcal mol−1
.
Next, we maintained R2 as a hydroxyl group and modified
the B ring of the quinoline moiety. The introduction of a chloro
group resulted in a large decrease in ESe⋯O (7.28 kcal mol−1
)
and a significantly longer rSe⋯O (2.646 Å) than in other selene-
nyl sulfides. A natural bond orbital (NBO) analysis further con-
firmed the weaker O⋯Se interaction in the selenenyl sulfides. Scheme 3 Synthesis of 11a–11k, 12k and 13k.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20915

View Article Online
Paper
Dalton Transactions
salts, which were prepared from the substituted 2-amino- pounds 11c (chloro), 11e (fluoro) and 11i (methoxy), which
benzoic acids and NaNO2, to give the 2,2-diselenobisbenzoic acids contained a substituent para to the selenium atom, showed
6a–6k (intermediate 6l was prepared by treating 6k with acetic better catalytic activities than compounds 11b (chloro), 11d
anhydride in pyridine). A subsequent reaction with SOCl2 (fluoro) and 11h (methoxy) where the substituent group was in
under reflux provided the 2-chloroselenobenzoyl chlorides 7a– the meta position. Compound 12k, with a hydroxyl group para
7j, which were reacted with the corresponding amines in anhy- to the selenium atom, exhibited excellent activities (H2O2, v0 =
drous CH2Cl2 to provide the key intermediates 8a–8k, 9k and 236.0 μM min−1; t-BuOOH, v0 = 70.0 μM min−1; Cum-OOH,
10k. Treating 8k–10k with K2CO3 in CH3OH produced com- v0 = 198.2 μM min−1), which could be attributed to the weak-
pounds 8l–10l. Deprotection of the Boc group with piperidine ness of the Se⋯O noncovalent interaction (7.28 kcal mol−1) in
gave the target compounds 11a–11c, 11e, and 11i–11j. The the selenenyl sulfide intermediate (Table 1). This weakening of
hydrochloride salts 11d, 11f–11g, 11k, 12k and 13k were interaction overcomes the undesired thiol exchange at the
obtained by removal of the Boc group with HCl gas. The hydro- selenium and regenerates the catalytically active selenol. The
bromide salt of compound 11h was prepared from intermedi- outstanding activities of 12k suggested that the hydroxyl group
ate 10h by refluxing in HBr (47%).
para to the selenium was indispensable for high GPx activity.
Compounds 11k (182.9 μM min−1; 44.8 μM min−1; 152.6 μM min−1)
and 12k, with or without a chlorine atom on the quino-
line moiety, exhibited significantly higher catalytic activities
Glutathione peroxidase-like activity
The GPx-like activity of the clioquinol–ebselen hybrids was
evaluated by the coupled reductase method, which uses GSH
as the thiol cofactor and hydrogen peroxide (H2O2), tert-butyl
hydroperoxide (t-BuOOH), and cumene hydroperoxide (Cum-
OOH) as substrates.17,23,24,26
than ebselen. However, compound 13k (101.8 μM min−1
;
22.7 μM min−1; 66.3 μM min−1), with an iodine atom on the
quinoline moiety, showed much lower GPx-like activity. These
results were consistent with the DFT studies for selenenyl
sulfide intermediates in the computational design. It is worth
The catalytic activities in Table 2 are in high accordance
with the DFT calculations in all three peroxide systems. Com-
noting that compound 11f (249.3 μM min−1; 27.2 μM min−1
;
180.9 μM min−1), which contains a fluorine atom on the ben-
zoselenazol-3(2H)-one moiety (R3), also showed excellent GPx-
like activity.
Table 2 Initial rates, ν0 (μM min−1), for the reduction of peroxides by
GSH in the presence of CQ and ebselen hybrid derivatives
Determining the catalytic parameters
Kinetic experiments were conducted to further understand the
catalytic behavior of 11f and 12k, which exhibit prominent
GPx-like activities (Tables S1–S18† and Fig. S2–S19†).23,24 It
could be observed from Fig. S1† that, with the increase in the
concentration of GSH in the presence of catalysts and per-
oxide, there was a rapid increase in the initial rate at the begin-
ning, followed by saturation kinetics. These results confirmed
that compounds 11f and 12k followed typical saturation kine-
tics for native GPx.
The KM values (Table 3) obtained for 11f (0.76, 0.75 and
0.48 mM) and 12k (0.30, 0.30 and 0.33 mM) were much
smaller (especially 12k) than those for the parent compound
ebselen (1.05, 0.83 and 1.18 mM), suggesting that the weak
O⋯Se interaction significantly prevented the undesired thiol
exchange reactions in the selenenyl sulfide. The catalytic
efficiency of 12k obtained at various thiol concentrations was
more than five times of that of ebselen (2.23, 0.65 and
0.80 mM−1 min−1) and two times that of 11f (5.90, 1.05 and
4.78 mM−1 min−1). Similarly, the catalytic efficiency of 12k
(6.83, 1.46 and 5.28 mM−1 min−1) observed at various peroxide
concentrations was approximately five-fold as large as that of
ebselen (2.10, 0.22 and 1.59 mM−1 min−1) and two times that
of 11f (2.99, 0.40 and 1.98 mM−1 min−1). The relatively low
η (catalytic efficiencies) values of 11f relative to 12k were due to
the presence of a strong ESe⋯O in the selenenyl sulfide, which
led to thiol exchange reactions rather than to selenol
regeneration.
GPx-like v0 (μM min−1
)
SDa
Compd.
H2O2
t BuOOH
Cum-OOH
11a
11b
11c
11d·HCl
158.7 3.6
148.3 1.6
153.7 6.6
132.8 2.6
162.1 3.3
249.3 5.2
101.2 1.1
88.6 4.2
96.8 1.1
74.53 1.2
182.9 3.9
236.0 3.7
101.8 5.7
124.3 4.3
37.8 2.0
37.2 0.7
28.1 1.9
28.9 0.4
29.7 0.3
26.1 0.7
29.5 0.8
27.2 0.8
25.3 0.2
17.3 1.4
21.3 1.5
16.2 0.9
44.8 1.2
70.0 1.6
22.7 2.6
27.7 0.4
4.5 0.2
84.2 2.0
93.7 2.6
119.8 2.6
95.43 1.6
118.4 1.3
180.9 1.6
86.1 2.2
46.5 0.7
66.3 0.5
49.1 0.3
152.6 1.2
198.2 10.2
77.0 6.2
62.3 2.0
6.3 0.4
11e
11f·HCl
11g·HCl
11h·HBr
11i
11j
11k·HCl
12k·HCl
13k·HCl
Ebselen
Clioquinol
Controlb
5.0 0.8
5.8 1.2
a Assay conditions: reactions were carried out in phosphate buffer
(100 mM, pH 7.5) at 25 °C with 1 mM ethylene diamine tetraacetate
(EDTA), 2 mM GSH, 0.4 mM NADPH, 1.3 unit mL−1 glutathione
reductase (GR), 80 μM selenium compounds, and 1.6 mM peroxide. All
values were determined in triplicate for the initial 10 s, and the
average values with standard deviations were obtained. b Control
values were obtained from the reduction of peroxide by GSH in the
absence of the compound.
20916 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Table 3 Effect of peroxide and GSH concentrations on Vmax (maximum velocities), KM (Michaelis constants), kcat (catalytic constants), and η (cata-
lytic efficiencies) for ebselen, 11f and 12k
Vmax
KM
(mM)
kcat
η
Compd.
Peroxide
(μM·min−1
)
(min−1
)
(mM−1·min−1
)
Effect of peroxide concentrationa
a
Ebselen
11f·HCl
12k·HCl
H2O2
222.71
58.55
1.32
3.33
1.11
1.98
6.08
3.30
0.46
0.96
0.63
2.78
0.73
1.77
5.92
2.44
6.54
3.14
1.41
3.33
2.10
0.22
1.59
2.99
0.40
1.98
6.83
1.46
5.28
t-BuOOHa
Cum-OOHa
141.44
473.93
195.31
523.56
251.25
113.12
266.67
a
H2O2
t-BuOOHa
Cum-OOHa
a
H2O2
t-BuOOHa
Cum-OOHa
Effect of thiol concentrationb
b
Ebselen
11f·HCl
12k·HCl
H2O2
187.96
43.01
75.64
361.01
62.77
184.16
284.90
74.68
1.05
0.83
1.18
0.76
0.75
0.48
0.30
0.30
0.33
2.35
0.54
0.95
4.51
0.78
2.30
3.56
0.94
2.78
2.23
0.65
0.80
5.90
1.05
4.78
11.87
3.13
8.42
t-BuOOHb
Cum-OOHb
b
H2O2
t-BuOOHb
Cum-OOHb
b
H2O2
t-BuOOHb
Cum-OOHb
222.71
a Assay conditions: reactions were carried out in phosphate buffer (100 mM, pH 7.5) at 25 °C containing 2 mM GSH, 0.4 mM NADPH, 1 mM
EDTA, 1.3 U mL−1 GR, variable peroxide, and 80 μM test compound. b Assay conditions: reactions were carried out in phosphate buffer (100 mM,
pH 7.5) at 25 °C containing variable GSH, 0.4 mM NADPH, 1 mM EDTA, 1.3 U mL−1 GR, 1.6 mM peroxide, and 80 μM test compound.
Scavenging peroxide in the presence of GSH catalyzed by
clioquinol–ebselen hybrids
sulfide 14 (77Se NMR: δ 524 ppm), which strongly suggested a
weaker O⋯Se intramolecular interaction compared to that of
the selenenyl sulfide of ebselen (77Se NMR: δ 545 ppm).15 The
addition of another equivalent of GSH to the solution pro-
duced selenol 15, which was trapped by generating the sele-
nide 15a (77Se NMR: δ 251 ppm) with iodoacetic acid and
confirmed by HRMS (Fig. S35–67†). Selenol 15 and its sodium
selenolate (77Se NMR: δ 158 ppm), obtained by reduction with
NaBH4, were not stable and were immediately oxidized to di-
selenide 16 (77Se NMR: δ 432 ppm) in air. On the other hand,
upon addition of H2O2 (1.0 equiv.) and GSH (1.0 equiv.) to a
solution of selenenyl sulfide 14, only a peak for diselenide 16
appeared, which indicated that the diselenide was the first
product of the oxidation. Upon further addition of H2O2
(2.0 equiv.), the peak for diselenide 16 disappeared and a signal
for selenium quinone 17 (77Se NMR: δ 803 ppm) appeared,
which was further confirmed by 13C NMR, 1H NMR and HRMS
(Fig. S49–S53†). It was interesting that none of the other oxi-
dative species, such as the selenenic or seleninic acids
(expected δ: 1050–1350 ppm),15,18–21,23–26,50–52 were detected
when even more H2O2 was added to the reaction. This obser-
vation implies that the formation of selenium quinone 17 pre-
vents the further oxidation of the selenium, and allows the
nucleophilic attack of the thiol to be accelerated. Addition of
GSH to diselenide 16 or selenium quinone 17 also produced
selenenyl sulfide 14. Taken together (Scheme 5), at a high con-
centration of GSH, selenol 15 was transformed into diselenide
16 and then reformed through selenenyl sulfide 14. At a high
concentration of hydrogen peroxide, diselenide 16 could be
further oxidized to selenium quinone 17. This observation was
To further evaluate the catalyzed antioxidant activity of clio-
quinol–ebselen derivatives, kinetic reactions in the presence of
peroxide and GSH were conducted with ebselen and CQ as the
controls (Fig. S20–S34†). The results indicated that the full
scavenging of H2O2, Cum-OOH and t-BuOOH was obtained
within 180–210 minutes in the presence of catalyst 12k,
respectively (Fig. 3). In contrast, in the presence of 11f and
ebselen, less than 80% of the scavenging of H2O2 or Cum-
OOH, and 35% of the scavenging of t-BuOOH, were achieved
within 160 min (Fig. 3D), and the reaction did not proceed
when the time was extended. These effects indicated that 12k,
unlike the parent compound ebselen, was capable of proceed-
ing the GPx-like function at concentrations of diverse peroxide
as low as possible, which has significant importance for pro-
tecting the intracellular redox environment. All the results
further show that the unique structure of hybrid 12k results in
significant antioxidant activities compared to the leading com-
pounds ebselen and CQ.
Mechanism study of the GPx-like catalytic cycle
To gain further insight into the activity differences among the
hybrids with various substituents, a mechanistic study of the
GPx-like catalytic cycle for the hybrids was carried out using
1
77Se, H and 13C NMR spectroscopy and high-resolution mass
spectroscopy (HRMS) (see the ESI†).
As expected (Scheme 4), treatment of 12k (77Se NMR:
δ 903 ppm) with one equivalent of GSH afforded selenenyl
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20917

View Article Online
Paper
Dalton Transactions
Fig. 3 Scavenging of peroxide by GSH over time in the presence and absence of a catalyst. (A) Scavenging of H2O2. (B) Scavenging of t-BuOOH.
(C) Scavenging of Cum-OOH. (D) The maximal scavenging peroxide ability. Assay conditions: reactions were carried out in phosphate buffer
(100 mM, pH 7.5) at 25 °C with EDTA (1 mM), GSH (0.25 mM), NADPH (0.50 mM), GR (1.3 unit mL−1), peroxide (0.20 mM) and catalyst (10 μM). Values
reported are the means for three independent experiments. Statistical significance was analyzed by ANOVA: (**) P < 0.01, (***) P < 0.001.
than the disproportionation of selenenyl sulfide to generate
diselenide, which was a characteristic feature of the GPx cata-
lytic cycle of ebselen and its analogs.15,23,26
Nevertheless, for compound 11f (77Se NMR: δ 878 ppm),
upon the addition of GSH (1 equiv.), the newly formed selene-
nyl sulfide (77Se NMR: δ 517 ppm) underwent disproportiona-
tion to generate the corresponding diselenide (77Se NMR:
Scheme 4 Reaction of 12k with GSH and H2O2.
δ 455 ppm) immediately (Fig. S55 and S56†). These results
demonstrated that 11f, due to the strong Se⋯O interaction
present in the selenenyl sulfide intermediate, has the same
GPx-like mechanism of ebselen, which enhanced the thiol
exchange and facilitated a disproportionation reaction.15,26
The facile disproportionation reaction could be responsible for
the relatively higher initial reduction rate of 11f at high con-
centration of peroxide but was unable to proceed the GPx-like
cycle at concentrations of diverse peroxide as low as possible.
2D NMR study of hybrids with Aβ1-42 in solution
A 2D NMR study of 12k with the monomeric form of Aβ1-42 was
performed,53,54 which significantly indicated the direct inter-
actions of the hybrids with Aβ1-42 from the chemical shift or
broadening of the peaks of the Aβ1-42 residues R5, H6, H14
and F20 upon treatment with a stoichiometric amount of 12k
Scheme 5 Proposed catalytic cycle for the reduction of H2O2 by GSH
in the presence of 12k.
(Fig. 4). As these residues were in close contact with the puta-
tive metal coordination site of Aβ (H6, H13, and H14),55–57
in precise accordance with the computational results for the
selenenyl sulfide, which indicated that the structure of 12k
was different from the parent compound ebselen and was
favorable for producing the catalytically active selenol rather
these observations demonstrated that hybrid 12k not only has
a strong bonding ability with the Aβ1-42 monomer, but also
may easily chelate metal ions in Aβ. With an excess of 12k
added to Aβ1-42, remarkable chemical shift movements were
20918 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Fig. 4 Interactions of 12k with Aβ1-42 in solution. (a and b) 2D 1H–15N TROSY-HSQC NMR spectra of 12k-titrated monomeric Aβ1-42 (freshly dis-
solved Aβ1-42 (50 μM) in 5 mM phosphate buffer, 0.5 mM Na2EDTA, and 0.05 mM NaN3, pH 7.5, 4 °C, 950 MHz). (c) Chemical shift perturbations for
Aβ1-42 residues in the presence of 12k (Aβ1-42 : 12k = 1 : 2). * Denotes absent or overlapping signals. (d) Residues with the largest changes in chemical
shifts mapped onto the structure of Aβ1-42 (PDB 1Z0Q).
detected because of the possible change in the conformation between the compound and Fe3+. Furthermore, the maximum
of Aβ1-42. A smaller number of peak movements were observed absorption of 12k exhibited a significant shift from 259 nm to
with methionine 35 (M35), suggesting that 12k does not 263 nm when ZnCl2 was added. Additionally, the absorption
53
induce the oxidation of Aβ1-42
.
shoulder shifted from 339 nm to 334 nm when FeSO4 was
added, suggesting that 12k binds Zn2+ and Fe2+. A series of
UV-vis spectra of compound 12k titrated with Cu2+ are shown
in Fig. 5B, and the binding stoichiometry of compound 12k
with Cu2+ was determined by following the absorption
changes at 413 nm. The presence of an isosbestic point
revealed the formation of a unique Cu2+-12k complex (Fig. 5B),
and the titration analysis was consistent with a 1 : 2 Cu2+/
ligand molar ratio. To further confirm the components of the
complex, high-resolution mass spectroscopy (Fig. S68†) was
carried out in solution, and the result was positive (HRMS:
calcd for [(12k)2Cu–2H]2−, 420.8924; found, 420.8943). The
interaction of 12k with Cu2+ was further investigated by a UV-
vis spectrum in the presence of Aβ1–42 (Fig. 5C). When 12k was
added to the solutions containing Aβ1-42 and Cu2+, the optical
changes were similar to the samples containing 12k and Cu2+
alone, which demonstrated that hybrid 12k preferentially
Metal chelating properties of compounds
The ability of target compounds to chelate biometals such as
Cu(II), Zn(II), Fe(II) and Fe(III) was studied by UV-visible spec-
trometry. New optical bands were observed in the UV-vis
spectra for compounds 11k, 12k and 13k upon the addition of
CuSO4 or ZnCl2 and FeSO4, suggesting that these compounds
were able to interact with Cu2+/Zn2+/Fe2+ to form complexes
(Fig. S68–S70†). For example, upon incubation of CuSO4 with
ligand 12k in buffer (Fig. 5A), the new optical bands (in the
absence of Cu2+, the UV-vis spectrum of compound 12k
showed an absorption maximum at 259 nm and a shoulder at
339 nm; when Cu2+ was added, a new band associated with
the copper complex appeared at 413 nm) indicated complex
formation via metal chelation. Interestingly, after the addition
of FeCl3, there was no significant shift; however, the absorp-
tion decreased significantly, suggesting a possible interaction
bonds with Cu2+ in the presence of Aβ1-42
.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20919

View Article Online
Paper
Dalton Transactions
Fig. 5 (A) UV absorbance spectrum of compound 12k (50 μM) alone or in the presence of CuSO4 (50 μM), ZnCl2 (50 μM), FeSO4 (50 μM) or FeCl3
(50 μM) in a buffer (20 mM HEPES, 150 mM NaCl, pH 7.4). (B) UV-vis titration of compound 12k with Cu2+ in a buffer (20 mM HEPES, 150 mM NaCl,
pH 7.4) at room temperature. The concentration of Cu2+ varied from 0 to 144 μM. A breakpoint was observed at a 0.5 : 1 ratio. The concentration of
compound 12k was 80 μM. (C) UV absorbance spectrum of compound 12k (50 μM) alone or in the presence of Cu2+ (25 μM) and Aβ1-42 (25 μM) in a
buffer (20 mM HEPES, 150 mM NaCl, pH 7.4).
Modulation of metal-free and meta-associated Aβ aggregation
and disaggregation of Aβ
of metal ions (Fig. 6Ba, lane 3) than in the metal-free samples
(Fig. 6Ba, lane 2). On the other hand, upon incubation of Aβ1-42
with Cu(II) and 12k, the oligomerization was blocked almost
completely and the intensity of the monomer band increased
significantly (Fig. 6Ba, lane 5). More importantly, 12k (inhi-
bition: 91%; disaggregation: 70%) demonstrated a greater
effect on the control of Cu(II)-mediated Aβ aggregation and
transformed Cu(II)-associated Aβ aggregates than CQ (68%;
41%) in the ThT assay (Fig. 6Bb–Bc). Consistent with the ThT
binding assay results, the TEM images indicated that more
well-defined Aβ fibrils (Fig. 6Bd and Be, sequence 3) were
formed in the presence of Cu(II) compared with Aβ alone
(Fig. 6Bd and Be, sequence 2), 12k (Fig. 6Bd and Be, sequence
5) and CQ (Fig. 6Bd and Be, sequence 4). These results
revealed that 12k could significantly modulate Cu(II)-induced
Aβ1-42 aggregation and disassemble the preformed Cu(II)-
induced aggregates into relatively smaller, amorphous confor-
mations. Taken together, all the inhibition and disaggregation
results revealed that 12k possessed distinctive activity for the
control of self- and Cu(II)-induced Aβ aggregation.
The ability of 12k to inhibit the formation of Aβ1-42 aggregates
or to disassemble preformed Aβ1-42 aggregates was investigated
by gel electrophoresis (to visualize the changes in the size
distribution of low molecular weight (LMW) Aβ1-42 oligomers), a
thioflavin-T (ThT) assay (to measure the formation of Aβ aggre-
gates) and transmission electron microscopy (TEM) (to deter-
mine the morphological changes in the Aβ species) (Fig. 6). As
shown by the results in Fig. 6A, Aβ1-42 existed predominately as
a mixture of monomers and oligomers of the order 2–4
(Fig. 6Aa, lane 1; Fig. 6Ad, sequence 1). In contrast, monomers
and oligomers were no longer detected and fibrils were the
main form in the Aβ1-42 sample after incubating for 24 h at
37 °C (Fig. 6Aa, lane 2; Fig. 6Ad, sequence 2). Upon incubation
of 12k with the Aβ1-42 sample, the intensity of the monomer
band was strengthened remarkably and the oligomer band was
reduced substantially (Fig. 6Aa, lane 5). However, the intensity
of the monomer and oligomer bands in the Aβ1-42 sample
treated with CQ or ebselen was markedly reduced (Fig. 6Aa,
lanes 3 and 4). These results indicated that 12k had a better
ability to inhibit the self-induced Aβ1-42 species aggregation
than CQ or ebselen. The results from gel electrophoresis were
substantiated by the ThT assay (Fig. 6Ab and Ac) and TEM
(Fig. 6Ad and Ae). The ThT assay showed that 12k inhibited
84% (Fig. 6Ab) of the Aβ1-42 aggregation and disaggregated
81% (Fig. 6Ac) of the preformed Aβ1-42 aggregates. Compara-
tively, CQ (32%; 18%) and ebselen (less than 10%) showed
relatively lower activity than 12k. As observed by TEM, fewer Aβ
fibrils were detected in the 12k-treated Aβ sample (Fig. 6Ad
and Ae, sequence 5) than in the controls (Fig. 6Ad and Ae,
sequence 2), CQ (Fig. 6Ad and Ae, sequence 3) or ebselen
(Fig. 6Ad and Ae, sequence 4), which further demonstrated
that 12k can attenuate Aβ aggregation and disassemble pre-
formed Aβ aggregates effectively.
Blood–brain barrier permeation assay
To evaluate the brain penetration of the hybrids, the parallel
artificial membrane permeability assay (PAMPA) was used as
the blood–brain barrier (PAMPA-BBB) assay, with a modifi-
cation (Tables S19–S21 and Fig. S71†).14 Compound 12k, with
a Pe value of 5.3 × 10−6 cm s−1, would be able to penetrate the
BBB and reach its biological targets in the CNS.
Compound 12k·HCl demonstrated no acute toxicity
To evaluate the toxicity of 12k·HCl, twenty KM mice (KM mice,
which are common closed colony mice and most widely used
in biomedical research in China) were randomly divided into
4 groups and the test compound 12k·HCl was administered in
single oral doses of 0, 677, 1333, or 2000 mg kg−1. After admin-
istration of the compound, mice were monitored continuously
for the first 4 h for any abnormal behavior and mortality
changes, intermittently for the next 24 h, and occasionally
thereafter for 14 days for the onset of any delayed effects.
Furthermore, the capacity of 12k to modulate Cu(II)-
induced Aβ1-42 aggregation and transform the preformed
Cu(II)-induced Aβ1-42 aggregates was investigated (Fig. 6B). More
Cu(II)-triggered Aβ1-42 oligomers were observed in the presence
20920 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Fig. 6 Experiments of inhibition and disaggregation in the presence of 12k. (A) Self-induced Aβ1-42 aggregation experiments including (a) self-
induced Aβ1-42 species were visualized by PICUP, followed by Tricine-SDS and silver staining. Lanes: (1) fresh Aβ1-42, (2) Aβ1-42 alone, (3) Aβ1-42 + CQ,
(4) Aβ1-42 + ebselen, (5) Aβ1-42 + 12k. (b and c) Top: Scheme for the inhibition or disaggregation experiments. Bottom: the fluorescence intensity of
the ThT binding assay. Statistical significance was analyzed by ANOVA: (**) p < 0.01, (***) p < 0.001. (d and e) TEM image analysis of the inhibition of
Aβ1-42 aggregation and disaggregation of Aβ1-42 fibrils. Serial number: (1) fresh Aβ1-42, (2) Aβ1-42 alone, (3) Aβ1-42 + CQ, (4) Aβ1-42 + ebselen, (5) Aβ1-42
+
12k. Experimental conditions: Aβ1-42 (25 μM); CQ, ebselen or 12k (25 μM); PBS (50 mM); pH 7.4; 37 °C. (B) Cu2+-induced Aβ1-42 aggregation experi-
ments including (a) self-induced Aβ1-42 species were visualized by gel electrophoresis. Lanes: (1) fresh Aβ1-42, (2) Aβ1-42 alone, (3) Aβ1-42 + Cu2+
,
(4) Aβ1-42 + Cu2+ + CQ, (5) Aβ1-42 + Cu2+ + 12k. (b and c) Top: scheme for the inhibition or disaggregation experiments. Bottom: the fluorescence
intensity of the ThT binding assay. Statistical significance was analyzed by ANOVA: (**) p < 0.01, (***) p < 0.001. (d and e) TEM image analysis of the
inhibition of Cu2+-induced Aβ1-42 aggregation and disaggregation of Cu2+-induced Aβ1-42 fibrils. Serial number: (1) fresh Aβ1-42, (2) Aβ1-42 alone,
(3) Aβ1-42 + Cu2+, (4) Aβ1-42 + Cu2+ + CQ, (5) Aβ1-42 + Cu2+ + 12k. Experimental conditions: Aβ1-42 (25 μM); CuSO4 (25 μM); CQ, or 12k (50 μM);
HEPES (20 μM) and NaCl (150 μM); pH 6.6; 37 °C.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20921

View Article Online
Paper
Dalton Transactions
During the experimental period, no acute toxicity, such as (10 μL, 25 μM, final concentration) with or without copper
mortality, and significant abnormal changes in water or food (10 μL, 25 μM, final concentration) and the tested compound
consumption and body weight, were observed. Furthermore, (10 μL, 50 μM, final concentration) was incubated at 37 °C for
all mice were sacrificed at the 14th day of drug administration 24 h. Then 20 μL of the sample was diluted to a final volume
and no damage to the heart, liver and kidneys was macroscopi- of 200 μL with 50 mM glycine–NaOH buffer (pH 8.0) contain-
cally detected. Overall, compound 12k·HCl was non-toxic and ing thioflavin T (5 μM). The detection method was the same as
well tolerated at doses up to 2000 mg kg−1
.
that of the self-mediated Aβ1-42 aggregation experiment.
For the experiment on the disaggregation of self-induced Aβ
fibril, the Aβ stock solution was diluted with 10 mM phosphate
buffer (pH 7.4). The peptide (15 μL, 50 μM) was incubated at
37 °C for 24 h. The tested compound (15 μL, 50 μM) was then
added and incubated at 37 °C for another 24 h. Then 20 μL of
the sample was diluted to a final volume of 200 μL with
50 mM glycine–NaOH buffer (pH 8.0) containing thioflavin T
(5 μM). The detection method was the same as above.
For the experiment on the disaggregation of copper-
induced Aβ fibrils, the Aβ stock solution was diluted in 20 μM
HEPES (pH 6.6) with 150 μM NaCl. The mixture of the peptide
(10 μL, 25 μM, final concentration) with copper (10 μL, 25 μM,
final concentration) was incubated at 37 °C for 24 h. The
tested compound (10 μL, 50 μM, final concentration) was then
added and incubated at 37 °C for another 24 h. Then 20 μL of
the sample was diluted to a final volume of 200 μL with
50 mM glycine–NaOH buffer (pH 8.0) containing thioflavin T
(5 μM). The detection method was the same as above.
Conclusions
In summary, our study involved the computer-assisted design,
synthesis, GPx-like activity evaluation and catalytic mechanism
studies of a novel series of selenoxy–chinolin. The results indi-
cated that the excellent GPx mimic activities of the hybrids,
including the prominent catalytic parameters and complete
scavenging peroxide, which were highly consistent with those
of the DFT studies. In contrast to the GPx-like mechanism of
ebselen, a new GPx-like process has been demonstrated. In
particular, a new active species, selenium quinone, which pre-
vented the further oxidation of the selenium and accelerated
the nucleophilic attack of the thiol, was first reported and con-
firmed by 77Se, 13C, and H-NMR and HRMS. 2D NMR studies
1
demonstrated that 12k has an effective interaction with Aβ.
Finally, other properties of the optimal compound 12k, such
as the metal-chelating ligand and the potential inhibitor of
Aβ1-42 aggregation, were also studied. This compound was able
to penetrate the CNS, according to an in vitro blood–brain
barrier model, and toxicity tests in mice showed that 12k has
no acute toxicity at doses up to 2000 mg kg−1. All of these pro-
perties highlight the potential of compound 12k for use as a
lead compound for further study in the treatment of AD.
Chemical cross-linking and determination of oligomer
frequency distributions
The sample preparation was the same as that for the ThT
assay. Immediately after their preparation, photo-induced
cross-linking of unmodified proteins (PICUP) was performed
as described previously.59,60 Briefly, 1 μL of 40 mM ammonium
persulfate (APS, Aldrich) and 1 μL of 2 mM tris(2,2′-bipyridyl)-
dichlororuthenium(II) (Ru(Bpy)) (Alfa Aesar) were added to
18 μL of 25 μM Aβ1-42 sample in a clear PCR tube. The mixture
was irradiated for 1 s with visible light and then quenched
immediately by adding 4 μL of 1 M DTT which was dissolved
in Tricine sample buffer (Beyotime). Samples were then
analyzed using 1.5 mm thick, 10–20% Tris-Tricine gradient
gels and silver stained (Beyotime). Gels were analyzed using
ImageJ. Reported values are the averages of three separate gels.61
Biological assays
ThT assay.58 Aβ1-42 (Millipore, counter ion: NaOH) was dis-
solved in ammonium hydroxide (1% v/v) to give a stock solu-
tion (2000 μM), which was aliquoted into small samples and
stored at −80 °C.
For the experiment on the inhibition of self-mediated Aβ1-42
aggregation, the Aβ stock solution was diluted with 50 mM
phosphate buffer (pH 7.4) to 50 μM before use. A mixture of
the peptide (10 μL, 25 μM, final concentration) with or without
the tested compound (10 μL) was incubated at 37 °C for 48 h.
Blanks using 50 mM phosphate buffer (pH 7.4) instead of Aβ
2D NMR spectroscopy
with or without inhibitors were also carried out. The sample The interaction of monomeric Aβ1-42 with 12k was determined
was diluted to a final volume of 200 μL with 50 mM glycine– by two-dimensional (2-D) Translational Relaxation Optimized
NaOH buffer (pH 8.0) containing thioflavin T (5 μM). Then the Spectroscopy (TROSY) and 1H–15N Heteronuclear Single
fluorescence intensities were recorded five minutes later (exci- Quantum Correlation (HSQC) NMR measurements by titrating
tation, 450 nm; emission, 485 nm). The percent inhibition of a 50 μM solution of 15N-labeled Aβ1-42 with a 25 mM stock
aggregation was calculated by using the expression (1 − IFi/IFc) × solution of 12k·HCl in DMSO-d6.53,54 The concentration of
100%, in which IFi and IFc are the fluorescence intensities DMSO used for each sample was lower than 1% v/v (final con-
obtained for Aβ in the presence and absence of inhibitors after centration) so that the influence of DMSO on the spectrum of
subtracting the background, respectively.
Aβ was minor,62,63 and the hydrochloride of 12k was well solu-
For the experiment on the inhibition of copper-mediated bilized (with no precipitation at 4 °C) in the buffer which was
Aβ1-42 aggregation, the Aβ stock solution was diluted in 20 μM used for the experiment. The 15N-labeled Aβ1-42 was purchased
HEPES (pH 6.6) with 150 μM NaCl. The mixture of the peptide from rPeptide and stored at −80 °C. One sample (1.0 mg) was
20922 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
predissolved in aqueous NaOH solution (1 mM, pH = 10) in a (PVDF membrane, pore size 0.45 mm) and the acceptor micro-
1 : 1 ratio (mg : mL) with sonication for 1 min at 4 °C. The plate were both from Millipore. The 96-well UV plate
basic pH = 10 solutions of the 15N-labeled Aβ1-42 peptides were (COSTAR®) was from Corning Incorporated. The acceptor 96-
diluted with a 4 °C potassium phosphate buffered solution well microplate was filled with 300 µL PBS/EtOH (7 : 3), and
(5 mM, pH 7.5) that contained 0.50 mM ethylenediamine tetra- the filter membrane was impregnated with 4 µL PBL in
acetic acid (Na2EDTA) and 0.05 mM NaN3, verified to be pH dodecane (20 mg mL−1). Compounds were dissolved in DMSO
7.5 before the start of each titration. Each spectrum was at 5 mg mL−1 and diluted 50-fold in PBS/EtOH (7 : 3) to a final
obtained with 9.5 kHz and 2.7 kHz spectral width and 2048 concentration of 100 μg mL−1. Then, 200 µL of the solution
and 256 complex data points in the 1H and 15N dimensions, was added to the donor wells. The acceptor filter plate was
respectively, 96 scans per free induction decay, and 1.0 s relax- carefully placed on the donor plate to form a sandwich, which
ation delay on a Bruker Avance 950 MHz spectrometer at 4 °C. was left undisturbed for 10 h at 25 °C. After incubation, the
2D data were processed using TOPSPIN 3.2 (from Bruker) and donor plate was carefully removed, and the concentration of
1
analyzed with Sparky version 3.113. H–15N HSQC peaks were compounds in the acceptor wells was determined using a UV
assigned by comparison with literature values.53–55,64,65 The plate reader (Flexstation® 3). Every sample was analyzed at five
normalized chemical shift perturbations were calculated from wavelengths in four wells and in at least three independent
eqn (1).66
runs. Pe was calculated by using the following expression: Pe =
−Vd × Va/[(Vd + Va)A × t] × ln(1 − drugacceptor/drugequilibrium
where Vd is the volume of the donor well; Va, volume in the
)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
2
ðΔδHÞ þ ð0:2ðΔδNÞÞ
ΔδNꢀH
¼
ð1Þ
acceptor well; A, filter area; t, permeation time; drugacceptor, the
2
absorbance obtained in the acceptor well; and drugequilibrium
the theoretical equilibrium absorbance. The results are given
as the means standard deviation. In the experiment, 13
,
TEM assay67
For the metal-free experiment, Aβ stock solution was diluted
with a 10 mM phosphate buffer (pH = 7.4); for the copper-
induced experiment, Aβ stock solution was diluted with 20 μM
HEPES (pH = 6.6) and 150 μM NaCl. The sample preparation
was the same as that for the ThT assay.
Aliquots (10 μL) of the samples were placed on a carbon-
coated copper/rhodium grid for 2 min. Each grid was stained
with uranyl acetate (1%, 5 μL) for 2 min. After draining off the
excess staining solution, the specimen was transferred for
quality control standards (Table S19†) of known BBB per-
meability were included to validate the analysis set. A plot of
the experimental data versus literature values gave a strong
linear correlation, Pe (exp.) = 1.4574Pe (lit.) − 1.0773 (R2
=
0.9427) (Fig. S71†). From this equation and the limit estab-
lished by Di et al. (Pe (lit.) = 4.0 × 10−6 cm s−1) for blood–brain
barrier permeation, we concluded that compounds with a per-
meability greater than 4.7 × 10−6 cm s−1 could cross the blood–
brain barrier (Table S20†).
imaging into
a transmission electron microscope (JEOL
JEM-1400). All compounds were solubilized in the buffer
which was used for the experiment.
Acute toxicity
The procedures for acute toxicity study followed the similar
protocols of our previous studies.70,71 A total of 20 KM mice
(KM mice are common closed colony mice and are most
widely used in biomedical research in China) (male, 22 days,
18–20 g) purchased from the laboratory animal center of Sun
Yat-sen University (Guangzhou, China) were used to evaluate
the acute toxicity of compound 12k·HCl. Mice were maintained
with a 12 h light/dark cycle (light from 07:00 to 19:00) at
20–22 °C with a 60–70% relative humidity. Sterile food and
water were provided according to institutional guidelines.
Prior to each experiment, mice were fasted overnight and
allowed free access to water. Compound 12k·HCl was sus-
pended in 0.5% carboxymethyl cellulose sodium (CMC-Na)
salt solution (677, 1333, and 2000 mg kg−1) and administered
orally to the divided experimental groups. After the adminis-
tration of the compounds, the mice were observed continu-
ously for the first 4 h for any abnormal behavior and mortality
Coupled reductase assay
The GPx-like activity of the organoselenium compounds was
determined using a spectrophotometric method at 340 nm
described by Wilson et al.68 The test mixture contained GSH
(2 mM), EDTA (1 mM), glutathione reductase (1.3 unit mL−1
)
and NADPH (0.4 mM) in 100 mM potassium phosphate buffer,
pH 7.5. GPx samples (80 μM) were added to the test mixture at
25 °C, and the reaction was initiated by the addition of H2O2
(1.6 mM). The initial reduction rates were calculated from the
oxidation rate of NADPH at 340 nm. The initial reduction rate
was determined at least 3 times and calculated from the first
5–10% of the reaction using 6.22 mM−1 cm−1 as the extinction
coefficient for NADPH.
In vitro blood–brain barrier permeation assay
The blood–brain barrier penetration of compounds was evalu- changes, intermittently for the next 24 h, and occasionally
ated using the parallel artificial membrane permeation assay thereafter for 14 days for the onset of any delayed effects. All
(PAMPA) described by Di et al.69 Commercial drugs were pur- animals were sacrificed on the 14th day after drug adminis-
chased from Sigma and Alfa Aesar. Porcine brain lipid (PBL) tration and were macroscopically examined for possible
was obtained from Avanti Polar Lipids. The donor microplate damage to the heart, liver, and kidneys.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20923

View Article Online
Paper
Dalton Transactions
Metal-chelating study
9 G. Mugesh and H. B. Singh, Chem. Soc. Rev., 2000, 29, 347–
357.
10 M. P. Rayman, Lancet, 2000, 356, 233.
11 H. Tapiero, D. M. Townsend and K. D. Tew, Biomed.
Pharmacother., 2003, 57, 134–144.
12 J. P. d. Magalhães and G. M. Church, Exp. Gerontol., 2006,
41, 1–10.
13 J. Nordberg and E. S. J. Arnér, Free Radicals Biol. Med.,
2001, 31, 1287–1312.
14 Z. Wang, Y. Wang, W. Li, F. Mao, Y. Sun, L. Huang and
X. Li, ACS Chem. Neurosci., 2014, 5, 952–962.
15 B. K. Sarma and G. Mugesh, Chem. – Eur. J., 2008, 14,
10603–10614.
16 P. C. Trippier, K. Jansen Labby, D. D. Hawker, J. J. Mataka
and R. B. Silverman, J. Med. Chem., 2013, 56, 3121–
3147.
17 K. P. Bhabak and G. Mugesh, Chem. – Eur. J., 2007, 13,
4594–4601.
18 K. P. Bhabak and G. Mugesh, Chem. – Eur. J., 2008, 14,
8640–8651.
The chelating studies were performed with a UV-Vis spectro-
photometer. All compounds are solubilized in the buffer
which was used for the experiment. The absorption spectra of
each compound (50 μM, final concentration) alone or in the
presence of CuSO4, FeSO4, FeCl3 or ZnCl2 (50 μM, final concen-
tration) for 30 min in buffer (20 mM HEPES, 150 mM NaCl,
pH 7.4) were recorded at room temperature.
For the stoichiometry of the compound–Cu2+ complex, a
fixed amount of 12k (80 μM) was mixed with increasing
amounts of copper ions (0–144 μM), and the difference UV-vis
spectra were examined to investigate the ratio of ligand/metal
in the complex.
Statistical analysis
The results are expressed as the means SD for at least three
independent experiments. Data were subjected to a Student’s
t test or one-way analysis of variance (ANOVA) followed by a
Dunnett’s test. P Values less than 0.05 were considered to be
significant.
19 J. K. Pearson and R. J. Boyd, J. Phys. Chem. A, 2008, 112,
1013–1017.
20 B. K. Sarma, D. Manna, M. Minoura and G. Mugesh, J. Am.
Chem. Soc., 2010, 132, 5364–5374.
Acknowledgements
21 B. K. Sarma and G. Mugesh, J. Am. Chem. Soc., 2005, 127,
11477–11485.
22 K. Selvakumar, H. B. Singh and R. J. Butcher, Chem. – Eur.
J., 2010, 16, 10576–10591.
23 K. Selvakumar, P. Shah, H. B. Singh and R. J. Butcher,
Chem. – Eur. J., 2011, 17, 12741–12755.
24 V. P. Singh, H. B. Singh and R. J. Butcher, Eur. J. Org.
Chem., 2011, 5485–5497.
25 V. P. Singh, J.-f. Poon, R. J. Butcher and L. Engman, Chem.
– Eur. J., 2014, 20, 12563–12571.
We thank the National Natural Science Foundation of China
(no. 21302235 and 20972198), Guangdong Natural Science
Foundation (2014A030313124) and the Fundamental Research
Funds for the Central Universities (15ykpy04) for financial
support of this study. All NMR experiments for Aβ were per-
formed at the Beijing NMR Center and the NMR facility of the
National Center for Protein Sciences at Peking University. We
also thank the National Supercomputing Center in Guangzhou
for providing the computational resources. This research was
also supported in part by the Guangdong Province Key Labora-
tory of Computational Science and the Guangdong Province
Computational Science Innovative Research Team.
26 K. Satheeshkumar and G. Mugesh, Chem. – Eur. J., 2011,
17, 4849–4857.
27 G. Pepeu and M. G. Giovannini, Curr. Alzheimer Res.,
2009, 6, 86–96.
28 J. Hardy, J. Neurochem., 2009, 110, 1129–1134.
29 A. I. Bush, J. Alzheimer’s Dis., 2008, 15, 223–240.
Notes and references
1 S. P. Raps, J. C. K. Lai, L. Hertz and A. J. L. Cooper, Brain 30 X. Huang, C. S. Atwood, M. A. Hartshorn, G. Multhaup,
Res., 1989, 493, 398–401.
2 C. L. Hammond, T. K. Lee and N. Ballatori, J. Hepatol.,
2001, 34, 946–954.
L. E. Goldstein, R. C. Scarpa, M. P. Cuajungco, D. N. Gray,
J. Lim, R. D. Moir, R. E. Tanzi and A. I. Bush, Biochemistry,
1999, 38, 7609–7616.
3 R. Dringen, Prog. Neurobiol., 2000, 62, 649–671.
31 A. I. Bush, Trends Neurosci., 2003, 26, 207–214.
4 C. Jacob, G. I. Giles, N. M. Giles and H. Sies, Angew. Chem., 32 K. J. Barnham and A. I. Bush, Curr. Opin. Chem. Biol., 2008,
Int. Ed., 2003, 42, 4742–4758. 12, 222–228.
5 C. B. Pocernich and D. A. Butterfield, Biochim. Biophys. 33 S. C. Drew and K. J. Barnham, Acc. Chem. Res., 2011, 44,
Acta, 2012, 1822, 625–630. 1146–1155.
6 M. Y. Aksenov, H. Michael Tucker, P. Nair, M. V. Aksenova, 34 P. Faller, ChemBioChem, 2009, 10, 2837–2845.
D. Allan Butterfield, S. Estus and W. R. Markesbery, J. Mol. 35 G. Casadesus, M. A. Smith, Z. Xiongwei, G. Aliev,
Neurosci., 1998, 11, 151–164.
7 G. Mugesh, W.-W. du Mont and H. Sies, Chem. Rev., 2001,
101, 2125–2180.
8 K. P. Bhabak and G. Mugesh, Acc. Chem. Res., 2010, 43,
1408–1419.
A. D. Cash, K. Honda, R. B. Petersen and G. Perry,
J. Alzheimer’s Dis., 2004, 6, 165–169.
36 R. J. Castellani, K. Honda, X. Zhu, A. D. Cash,
A. Nunomura, G. Perry and M. A. Smith, Ageing Res. Rev.,
2004, 3, 319–326.
20924 | Dalton Trans., 2015, 44, 20913–20925
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
37 C. W. Ritchie, A. I. Bush, A. Mackinnon, et al., Arch. 54 K. Ono, L. Li, Y. Takamura, Y. Yoshiike, L. Zhu, F. Han,
Neurol., 2003, 60, 1685–1691.
38 L. Lannfelt, K. Blennow, H. Zetterberg, S. Batsman,
D. Ames, J. Harrison, C. L. Masters, S. Targum, A. I. Bush,
X. Mao, T. Ikeda, J. Takasaki, H. Nishijo, A. Takashima,
D. B. Teplow, M. G. Zagorski and M. Yamada, J. Biol.
Chem., 2012, 287, 14631–14643.
R. Murdoch, J. Wilson and C. W. Ritchie, Lancet Neurol., 55 L. Hou and M. G. Zagorski, J. Am. Chem. Soc., 2006, 128,
2008, 7, 779–786. 9260–9261.
39 P. A. Adlard, R. A. Cherny, D. I. Finkelstein, E. Gautier, 56 J. Danielsson, R. Pierattelli, L. Banci and A. Graslund, FEBS
E. Robb, M. Cortes, I. Volitakis, X. Liu, J. P. Smith, K. Perez, J., 2007, 274, 46–59.
K. Laughton, Q.-X. Li, S. A. Charman, J. A. Nicolazzo, 57 P. Faller and C. Hureau, Dalton Trans., 2009, 1080–1094,
S. Wilkins, K. Deleva, T. Lynch, G. Kok, C. W. Ritchie, DOI: 10.1039/B813398K.
R. E. Tanzi, R. Cappai, C. L. Masters, K. J. Barnham and 58 M. Rosini, E. Simoni, M. Bartolini, A. Cavalli, L. Ceccarini,
A. I. Bush, Neuron, 2008, 59, 43–55.
40 L. Xie, W. Zheng, N. Xin, J.-W. Xie, T. Wang and Z.-Y. Wang,
Neurochem. Int., 2012, 61, 334–340.
N. Pascu, D. W. McClymont, A. Tarozzi, M. L. Bolognesi,
A. Minarini, V. Tumiatti, V. Andrisano, I. R. Mellor and
C. Melchiorre, J. Med. Chem., 2008, 51, 4381–4384.
41 S. Antony and C. A. Bayse, Inorg. Chem., 2011, 50, 12075– 59 G. Bitan and D. B. Teplow, Acc. Chem. Res., 2004, 37, 357–364.
12084.
60 L. Pujadas, D. Rossi, R. Andres, C. M. Teixeira, B. Serra-
Vidal, A. Parcerisas, R. Maldonado, E. Giralt, N. Carulla
and E. Soriano, Nat. Commun., 2014, 5, 3443.
42 C. A. Bayse and A. Pavlou, Org. Biomol. Chem., 2011, 9,
8006–8015.
43 A. J. Mukherjee, S. S. Zade, H. B. Singh and R. B. Sunoj, 61 P. A. Novick, D. H. Lopes, K. M. Branson, A. Esteras-Chopo,
Chem. Rev., 2010, 110, 4357–4416.
44 Y. Nakashima, T. Shimizu, K. Hirabayashi, F. Iwasaki,
I. A. Graef, G. Bitan and V. S. Pande, J. Med. Chem., 2012,
55, 3002–3010.
M. Yamasaki and N. Kamigata, J. Org. Chem., 2005, 70, 62 J. J. Braymer, J. S. Choi, A. S. DeToma, C. Wang, K. Nam,
5020–5027.
J. W. Kampf, A. Ramamoorthy and M. H. Lim, Inorg. Chem.,
2011, 50, 10724–10734.
45 D. J. Press, E. A. Mercier, D. Kuzma and T. G. Back, J. Org.
Chem., 2008, 73, 4252–4255.
46 D. J. Press, N. M. R. McNeil, M. Hambrook and T. G. Back,
J. Org. Chem., 2014, 79, 9394–9401.
47 G. Mugesh, A. Panda, H. B. Singh, N. S. Punekar and
R. J. Butcher, J. Am. Chem. Soc., 2001, 123, 839–850.
48 N. B. Patel, J. C. Patel and S. H. Modi, J. Saudi Chem. Soc.,
2011, 15, 167–176.
63 S. Lee, X. Zheng, J. Krishnamoorthy, M. G. Savelieff,
H. M. Park, J. R. Brender, J. H. Kim, J. S. Derrick, A. Kochi,
H. J. Lee, C. Kim, A. Ramamoorthy, M. T. Bowers and
M. H. Lim, J. Am. Chem. Soc., 2014, 136, 299–310.
64 Y. Yan, S. A. McCallum and C. Wang, J. Am. Chem. Soc.,
2008, 130, 5394–5395.
65 Y. Yan and C. Wang, J. Mol. Biol., 2006, 364, 853–862.
49 U. K. Mazumder, M. Gupta, S. Bhattacharya, S. S. Karki, 66 S. S. Hindo, A. M. Mancino, J. J. Braymer, Y. Liu,
S. Rathinasamy and S. Thangavel, J. Enzyme Inhib. Med.
Chem., 2004, 19, 185–192.
S. Vivekanandan, A. Ramamoorthy and M. H. Lim, J. Am.
Chem. Soc., 2009, 131, 16663–16665.
50 K. P. Bhabak and G. Mugesh, Chem. – Eur. J., 2009, 15, 67 J.-S. Choi, J. J. Braymer, R. P. R. Nanga, A. Ramamoorthy
9846–9854.
and M. H. Lim, Proc. Natl. Acad. Sci. U. S. A., 2010, 107,
21990–21995.
51 V. Nascimento, E. E. Alberto, D. W. Tondo, D. Dambrowski,
M. R. Detty, F. Nome and A. L. Braga, J. Am. Chem. Soc., 68 S. R. Z. Wilson, P. A. Huang and R. R. C. Spector, J. Am.
2012, 134, 138–141. Chem. Soc., 1989, 111, 5936–5939.
52 S. S. Zade, H. B. Singh and R. J. Butcher, Angew. Chem., Int. 69 L. Di, E. H. Kerns, K. Fan, O. J. McConnell and G. T. Carter,
Ed., 2004, 43, 4513–4515. Eur. J. Med. Chem., 2003, 38, 223–232.
53 L. Hou, H. Shao, Y. Zhang, H. Li, N. K. Menon, 70 C. Lu, Y. Guo, J. Yan, Z. Luo, H.-B. Luo, M. Yan, L. Huang
E. B. Neuhaus, J. M. Brewer, I.-J. L. Byeon, D. G. Ray, and X. Li, J. Med. Chem., 2013, 56, 5843–5859.
M. P. Vitek, T. Iwashita, R. A. Makula, A. B. Przybyla and 71 Z. Luo, J. Sheng, Y. Sun, C. Lu, J. Yan, A. Liu, H.-b. Luo,
M. G. Zagorski, J. Am. Chem. Soc., 2004, 126, 1992–2005.
L. Huang and X. Li, J. Med. Chem., 2013, 56, 9089–9099.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans., 2015, 44, 20913–20925 | 20925