Novel antitumor adamantane-azole gold(I) complexes as potential inhibitors of thioredoxin reductase

Article abstract of DOI:10.1007/s00775-016-1338-y

Gold complexes that could act as antitumor agents have attracted great attention. Heterocyclic compounds and their metal complexes display a broad spectrum of pharmacological properties. The present study reports the preparation and characterization of fo

Full text of DOI:10.1007/s00775-016-1338-y

J Biol Inorg Chem
DOI 10.1007/s00775-016-1338-y
ORIGINAL PAPER
Novel antitumor adamantane–azole gold(I) complexes as potential
inhibitors of thioredoxin reductase
1
1
1
Adriana Garcia · Rafael Carvalhaes Machado · Richard Michael Grazul ·
2
1
1
Miriam Teresa Paz Lopes · Charlane Cimini Corrêa · Hélio F. Dos Santos ·
Mauro Vieira de Almeida · Heveline Silva
1
1
Received: 8 July 2015 / Accepted: 9 January 2016
©
SBIC 2016
Abstract Gold complexes that could act as antitumor
agents have attracted great attention. Heterocyclic com-
pounds and their metal complexes display a broad spectrum
of pharmacological properties. The present study reports
the preparation and characterization of four novel gold(I)
complexes containing tertiary phosphine and new ligands
essential to stabilize the ligand–enzyme complex prior the
formation of covalent bond with gold center.
Graphical abstract The structure of the new gold com-
pounds was established on the basis of spectroscopic data,
DFT calculations and X-ray diffraction. TrxR inhibition
was evaluated and the results correlated with the assays in
tumor cells, suggesting the TrxR as possible target for these
compounds.
5
-adamantyl-1,3-thiazolidine-2-thione, 3-methyladaman-
tane–1,3,4-oxadiazole-2-thione. Spectroscopic data suggest
that gold is coordinated to the exocyclic sulfur atom in all
cases, as confirmed by X-ray crystallographic data obtained
for complex (1) and supported by quantum–mechanical
calculations. The cytotoxicity of the compounds has been
evaluated in comparison to cisplatin and auranofin in three
different tumor cell lines, colon cancer (CT26WT), meta-
static skin melanoma (B16F10), mammary adenocarci-
noma (4T1) and kidney normal cell (BHK-21). The gold
complexes were more active than their respective free
ligands and able to inhibit the thioredoxin reductase (TrxR)
enzyme, even in the presence of albumin. Molecular mod-
eling studies were carried out to understand the interaction
between the compounds and the TrxR enzyme, considered
as a potential target for new compounds in cancer treat-
ment. The docking results show that the adamantane ring is
Keywords Thiazolidine · Oxadiazol · Phosphine ·
Thioredoxin reductase · Crystallography · Molecular
docking · DFT calculations
Abbreviations
DFT
Density functional theory
Dimethysulfoxide
Dithiobisnitrobenzoic acid
(3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetra-
zolium bromide
Electronic supplementary material The online version of this
article (doi:10.1007/s00775-016-1338-y) contains supplementary
material, which is available to authorized users.
DMSO
DTNB
MTT
*
Heveline Silva
hevelinequi@gmail.com
NADPH Nicotinamide adenine dinucleotide phosphate
NMR
PEt3
PPh3
Nuclear magnetic resonance
Triethylphosphine
Triphenylphosphine
1
2
Departamento de Química, ICE, Universidade Federal de
Juiz de Fora, Juiz De Fora, MG 36036-900, Brazil
Departamento de Farmacologia, ICB, Universidade Federal
de Minas Gerais, Belo Horizonte, MG 31270-901, Brazil
RPMI
Roswell Park Memorial Institute Medium
1
3

J Biol Inorg Chem
TMS
TNB
TrxR
Tetramethylsilane
5-thionitrobenzol
Thioredoxin reductase
[19–22]. On the other hand, heterocyclic ligands such as
1,3-thiazolidines and 1,3,4-oxadiazolines belong to classes
of compounds that exhibit a broad spectrum of pharmaco-
logic activity. There are reports of antimicrobial, antifungal
and antihelminthic activity. Furthermore, they are present
in many natural products and pharmaceutical drugs such as
nesapidil, penicillin and pioglitazone [23–27]. In a previous
work we reported synthesis, characterization and biological
activity for four gold complexes containing thiazolidinic
and oxadiazolic moieties demonstrating cytotoxic activity
and high selectivity when investigated against tumoral and
non-tumoral cells [28]. In that study the structure–activ-
ity relationship was not clear and the complexes were not
tested as TrxR inhibitors; therefore the action mechanism
could not be completely investigated. Other phosphane
gold complexes with thiazole groups were investigated
with respect to their interactions with the TrxR enzyme and
cytotoxicity revealing important biological effects [29].
The hydrophilic and lipophilic balance of the aryl ligand
is another important property which should be consid-
ered since increasing the lipophilicity enhances the rate
of cellular uptake and the cytotoxic activity [30]. Some
previous structure–activity relationship studies of gold(I)
compounds and auranofin have shown that complexes coor-
dinated to tertiary phosphines are more active than similar
compounds with no phosphine substituents. The introduc-
tion of a tertiary phosphine also increases the lipophilicity
which, in turn, increases the permeability through the cell
membrane [31].
Introduction
Gold has been used in medicine since ancient times [1],
but only after 1935, when Forestier described the anti-
arthritic properties of gold(I) compounds [2], the scien-
tific community started the investigation of their beneficial
and toxic effects. Compounds like myochrysin (sodium
gold(I)thiomalate), solganal (gold(I)thioglucose) and
auranofin (triethylphosphine-(2,3,4,6-tetra-O-acetyl-β-1-D-
thiopyranosato-S)gold(I)) are still used in the treatment of
rheumatoid arthritis [3].
Auranofin has been tested as antitumor drug in several
cell lines, showing potent cytotoxic activity, especially for
leukemia [4, 5]. The malfunction of biochemical pathways
is related to the cellular pathophysiology of rheumatoid
arthritis, which is also important in cancer development
[6]. The possibility of gold complexes to act as antitumor
agents has attracted great attention. Many of the tested
compounds were effective to inhibit the growth of tumor
cells and have potential for treating cisplatin-resistant
tumors [7–9]. Nonetheless, the mechanism of action for
gold complexes is still unclear; many studies suggested
that the biological effect could be mediated by an anti-
mitochondrial mechanism, anti-inflammatory pathways
decreasing TNF-α (tumor necrosis factor) or even increas-
ing production of reactive oxygen species (ROS) inducing
apoptosis [3, 5, 10]. The enzyme thioredoxin reductase
In the present study, we describe the synthesis, charac-
terization, including, single-crystal X-ray diffraction data,
cytotoxic activity against tumor cell lines and the inhibi-
tion of TrxR for four new gold(I) complexes with ligands
derived from adamantane containing the heterocyclic
1,3-thiazolidines or 1,3,4-oxadiazolines and tertiary phos-
phine (PEt or PPh ). Molecular modeling was conducted
(
TrxR) is over expressed in many cancer cells indicating its
involvement in proliferation of tumor tissues [11–13]. TrxR
plays an important role in the regulation of intracellular
redox balance [5, 11]. Phosphine gold(I) complexes, such
as auranofin, also interact with glutathione, serum proteins
as albumin, cellular proteins and other small molecular
weight thiols-containing biomolecules [3, 14]. Auranofin
has also been shown to inhibit ubiquitin–proteasome sys-
tem that mediates protein degradation through a cascade
process, an important target for cancer and other diseases
3
3
to assist experimental data analysis, including structural
prediction, vibrational and NMR spectroscopies and the
reactivity of the complexes against a selenocystein (Sec)
residue. Docking studies were also performed for the main
derivatives considering the TrxR as the biological target.
[
15].
Several gold-based compounds have been reviewed over
the last decades as potential candidates as anticancer agents
16, 17]. Using molecules that exhibit biological activ-
Experimental
[
Materials and methods
ity as ligands could enhance the activity of the complex
by a synergistic mechanism or via multiple mechanism of
action [18]. Compounds derived from adamantane are pre-
sent in commercial drugs such as amantadine and meman-
tine which are used to treat Parkinson’s and Alzheimer’s
diseases, respectively. These molecules also present other
biological actions, including potential anticancer activity
All reagents and solvents were used without further purifi-
cation. Infrared (IR) spectra were recorded on a BRUKER
ALPHA FT-IR Spectrometer, in the region of 4000–
−1
400 cm as a KBr pellet, with a spectral resolution of
−1
4 cm and an average of 64 scans. Raman spectra were
obtained using a Bruker RFS 100 FT-Raman instrument
1
3

J Biol Inorg Chem
Scheme 1 Synthesis of ligands
a and b
S
3
NH2
NH
2
NH
N
4
O
OH
OH
O
OCH3
O
1
O
1) CS2
2) HCl
5
6
MeOH
NH NH2
2
1
5
7
1
4
+
-
12
H / ∆
EtOH / ∆
EtOH / OH / ∆
13
8
1
1
10
9
A
5
4
1
H
S
3
N
N 2
I
OH
6
S
7
Ethanolamine
CS
/OH-
16
8
9
I , PPh , imidazole
2
2
3
1
5
13
1
4
toluene / ∆
DMSO / ∆
∆
1
2
1
1
1
0
B
equipped with a germanium detector refrigerated by liq-
uid nitrogen, with excitation at 1064 nm from a Nd:YAG
laser, in the range between 4000 and 50 cm and a spec-
Uiso(H) = 1.2Ueq(C). The structures were drawn by
ORTEP-3 for Windows [36] and MERCURY softwares [37].
Ultraviolet (UV) spectra were recorded on a SHIMADZU
UV-1800 spectrophotometer in quartz cuvettes.
−1
−
1
1
1
tral resolution of 4 cm , with an average of 500 scans. H
3
NMR (300 MHz) and C NMR (75 MHz) spectra were
recorded in CDCl3 solutions on a BRUKER AVANCE
DRX/300 spectrometer. The chemical shifts were expressed
Synthesis of ligands (Scheme 1)
31
in the δ scale (ppm) relative to TMS internal standard.
P
Ligand (A) was previously reported and prepared accord-
ing to the method described in the literature [23, 27, 38].
Adamantane-1-carboxylic acid was esterified with metha-
nol under acidic conditions for 24 h at 90 °C. The ester was
reacted with hydrazine (hydrate 64 %) in ethanol at 80 °C
for 5 days and the hydrazide thus formed was treated with
carbon disulfide under basic conditions in ethanol, fol-
lowed by subsequent acidification with HCl (6 M) to pH 6
at 80 °C for 5 days. The final product (A) was purified by
recrystallization and the obtained crystals were analyzed by
single crystal X-ray diffraction.
NMR (202 MHz) spectra were recorded in acetone solu-
tions on a Avancer III HD BRUKER 500 MHz. The chemi-
cal shifts were expressed in the δ scale (ppm) relative to
H PO external standard. The TG/DTA curves were obtained
3
4
with a SHIMADZU DTG-60 Simultaneous DTA-TG appa-
ratus under a dynamic nitrogen atmosphere (heating rate
−1
1
0 °C min and temperature range from room tempera-
ture to 500 °C). Mass spectra were obtained on an AXIMA
MALDI-TOF-TOF Shimadzu Biotech instrument. A nitro-
gen laser (λmax = 337 nm) was employed for excitation in
an α cyano matrix. 200 scans were accumulated with 20 rep-
etitions each. Elemental analyses were performed at Central
Analítica, USP-Brazil. Diffraction data for single crystals
were collected using an Oxford GEMINI A Ultra diffrac-
tometer with Mo Kα (λ = 0.71073 Å) and temperature of
(A): yellow solid. TG/DTA: 190–191 °C (mp). IR ν
max
−1
KBr (cm ): 3122, 3092, 2914, 2850, 1605, 1504, 1452,
1
1181, 752. H NMR [300 MHz, CDCl , δ (ppm)]: 1.7–2.0
3
1
3
(m, 15H, H-adamantyl); 11.7 (s, 1H, N–H). C NMR
[75 MHz, CDCl , δ (ppm)]: 26.7; 28.5; 36.2; 40.9 (C-ada-
3
298 K. Data collection, reduction and cell refinement were
mantyl); 170.6 (C5); 178.5 (C2).
carried out by CRYSALISRED, Oxford diffraction Ltd.,
Version 1.171.32.38 software [32]. The structures were
solved and refined using SHELXS-97 and SHELXL-97,
respectively [33]. An empirical isotropic extinction param-
eter x was refined, according to the method described by
Larson [34]. A Multiscan absorption correction was applied
Ligand (B) was prepared from 1-adamantane methanol as
described in the literature [39]. The alcohol was treated ini-
tially with triphenylphosphine, imidazole and iodine in tolu-
ene for 24 h at 90 °C. The replacement of the iodine in the
1-adamantane methyl iodide was effected by reaction with
ethanolamine in DMSO for 24 h at 100 °C. The product was
purified by column chromatography. The final step consisted
of the cyclization of the amino alcohol with carbon disulfide
under basic conditions for 12 h at 80 °C. The final product
(B) was purified by recrystallization and the obtained crys-
tals were analyzed by single crystal X-ray diffraction.
(B): white solid. TG/DTA: 168–169 °C (mp). IR
[35]. Anisotropic displacement parameters were assigned to
all non-hydrogen atoms. The O- and N-bound H atoms were
initially located in a difference Fourier map, then added in
idealized positions and further refined according to the riding
model, with N–H = 0.86 Å, and with Uiso(H) = 1.5Ueq(O)
or 1.2Ueq(N). C-bound H atoms were included in the
riding-model approximation, with C–H = 0.95 Å and
−1
1
ν
KBr (cm ): 2905, 2845, 1481, 1413, 1296, 1229. H
max
1
3

J Biol Inorg Chem
Scheme 2 Synthesis of com-
plexes (1), (2), (3) and (4)
A
B
NMR [300 MHz, CDCl , δ (ppm), J (Hz)]: 1.6–2.0 (m,
evidenced by thin layer chromatography (TLC) (eluent:
9:1 dichloromethane/methanol). The product was purified
by a liquid/liquid extraction with dichloromethane/water
to eliminate carbonate and the solvent of the organic
phase was removed under reduced pressure to give the
desired compounds (1) and (2) in 88 and 92 % yields,
respectively.
3
1
2
5H, H-adamantyl); 3.2 (t, 2H, H5, J = 7.4); 3.4 (s,
5
,4
13
H, H6); 4.1 (t, 2H, H4, J5,4 = 7.4). C NMR [75 MHz,
CDCl , δ (ppm)]: 27.4; 28.2; 36.7; 41.3 (C-adamantyl);
3
3
6.6 (C5) 59.9 (C4) 61.0 (C6); 198.5 (C2).
Synthesis of the gold(I) complexes (Scheme 2)
Complexes (3) and (4) were obtained from ligand (B)
(0.4 mmol) dissolved in dichloromethane (5 mL) and added
to a solution of Au(PPh )Cl (0.4 mmol) or Au(PEt )Cl
Au(PPh )Cl was synthesized from K[AuCl ] according
3
4
to the literature [40] and Au(PEt )Cl is commercially
3
3
3
available. Complexes (1) and (2) were obtained from
ligand (A) (0.4 mmol) dissolved in dichloromethane
(0.4 mmol) in dichloromethane (5 mL). After stirring for
12 h at room temperature the complete consumption of the
starting materials was evidenced by thin layer chromatog-
raphy (TLC) (eluent: 9:1 dichloromethane/methanol). The
solvent was removed under reduced pressure to furnished
the desired compounds (3) and (4) in 82 and 90 % yields,
respectively.
(
(
5 mL) which was added to a solution of Au(PPh )Cl
3
0.4 mmol) or Au(PEt )Cl (0.4 mmol) in dichlorometh-
3
ane (5 mL). After stirring for 5 h at room temperature,
K CO (0.2 mmol) was added to adjust the pH. After 7 h
2
3
the complete consumption of the starting materials was
1
3

J Biol Inorg Chem
1
(
1): beige solid. TG/DTA: 218–219 °C (mp). IR νmaxKBr
C4); 61.2 (s, C6); 198.5 (s, C2). 31P{ H} NMR [202 MHz,
acetone, δ (ppm)]: 33.43 (s).
−1
(
5
(
cm ): 3051, 2916, 2848, 1452, 1145, 1101, 748, 692,
40, 501. H NMR [300 MHz, CDCl , δ (ppm)]: 1.7–2.1
m, 15H, H-adamantyl); 7.2–7.7 (m, 15H, Ar-PPh3).
NMR [75 MHz, CDCl , δ (ppm), J (Hz)] : 27.7; 35.1; 36.4;
1
Anal. Calc. for [C H NPS Au]Cl: C, 38.87; H, 5.87;
3
20 36
2
1
3
C
N, 2.27. Found: C, 38.92; H, 5.47; N, 2.09 %.
MS (MALDI): m/z Calc. for [C H NPS Au] [M]
+
+
3
20 36
2
1
3
9.7 (s, C-adamantyl); 128.9 (d, J (C,P) = 62.4, ipso-C
Calc. (582.6) found 582.4.
3
from PPh ); 129.5 (d, J (C,P) = 12.0, meta-C from PPh );
3
3
4
2
1
(
32.2 (d, J (C,P) = 2.7, para-C from PPh ); 134.4 (d, J
Cytotoxicity assay
3
C,P) = 13.8, ortho-C from PPh ); 160.3 (s, C5); 176.2 (s,
3
3
1
1
C2). P{ H} NMR [202 MHz, acetone, δ (ppm)]: 25.22
Cytotoxicity activity was investigated against the non-
tumor cell BHK-21—Baby Hamster Kidney and the fol-
lowing tumor cell lines: CT26-WT–murine colon can-
cer cells, B16-F10–mouse metastatic melanoma and
4T1–mouse metastatic mammary adenocarcinoma.
Cells were harvested and seeded in RPMI 1640 culture
medium, pH 7.4, supplemented with 10 % Fetal Bovine
Serum (FBS) at cell densities that varied depending on
(
s).
Anal. Calc. for [C H N OPSAu]: C, 51.88; H, 4.35; N,
.03. Found: C, 51.49; H, 4.35; N, 3.98 %.
MS (MALDI): m/z Calc. for [C H N OPSAu]
3
0
30
2
4
3
0
30
2
+
[
M + H] Calc. (695.6) found 695.5.
(
2): beige solid. TG/DTA: 144–145 °C (mp). IR
−1
1
ν
KBr (cm ): 2964, 2904, 2848 1454, 1145, 769. H
max
3
3
NMR [300 MHz, CDCl , δ (ppm)]: 1.2–1.3 (m, 9H, CH );
.6–2.0 (m, 21H, H-adamantyl, P(CH –CH ) ). C NMR
the cell line, 0.5 × 10 and 2 × 10 cells/100 µL/well in a
3
3
1
3
1
plate with 96 wells and were then incubated at 37 °C in a
2
3 3
[
75 MHz, CDCl , δ (ppm), J (Hz)]: 9.2 (s, P(CH –CH ) );
humidified atmosphere of 5 % CO for 24 h for optimum
3
2
3 3
2
1
8.1 (d, J (C–P) = 35, from PEt ); 27.5; 36.0; 38.8; 39.6 (s,
adhesion. Stock solutions of all compounds in DMSO were
serially diluted (100–0.1 µM) in cell culture medium (<1 %
DMSO). After drug exposure for 72 h the cells were incu-
bated with MTT (5 µg/10 µL/well) for 4 h. MTT is metabo-
lized by viable cells resulting in a violet complex product
that, after being solubilized in 100 µL DMSO/well, can be
quantified through colorimetric assay using a plate reader
(absorbance at 570 nm). Cisplatin and auranofin were used
as positive control against these cell lines. The raw data
were normalized to cell viability of the negative control
(culture medium with no compounds) as 100 %. The IC50
values were calculated by four parametric nonlinear regres-
sions using GraphPad Prism 5.0 software.
An equivalent experiment was performed using 4T1
cells and BHK-21 with some modifications [41]: 24 h after
the cells were seeded, bovine serum albumin (BSA, non-
toxic final concentration, 2 mg/mL) was added followed by
addition of gold compounds. IC50 values were obtained as
described above.
3
3
1
1
C-adamantyl);168.5 (s, C5); 174.6 (s, C2). P{ H} NMR
[
202 MHz, acetone, δ (ppm)]: 38.90 (s).
Anal. Calc. for [C H N OPSAu]: C, 39.28; H, 5.49; N,
1
8
30
2
5
.09. Found: C, 39.29; H, 5.42; N, 4.91 %.
MS (MALDI): m/z Calc. for [C H N OPSAu]
1
8
30
2
+
[
M + H] Calc. (551.4) found 551.3.
(
3): yellow solid. TG/DTA: 162–163 °C (mp). IR
−1
ν
KBr (cm ): 3058, 2906, 2844, 1481, 1413, 1296,
max
−
230, 1149, 1101, 748, 692, 545, 501. Raman ν (cm ):
max
29 (Au–S). H NMR [300 MHz, CDCl , δ (ppm), J (Hz)]:
1
1
3
1
3
1
1
3
.6–2.0 (m, 15H, H-adamantyl); 3.2 (t, 2H, H5, J5,4 = 7.4);
.4 (s, 2H, H6); 4.1 (t, 2H, H4, J = 7.4); 7.5–7.6 (m,
4
,5
1
3
5H, Ar-PPh3). C NMR [75 MHz, CDCl , δ (ppm), J
3
(
Hz)]: 27.8; 28.4; 36.9; 41.5 (s, C-adamantyl); 37.0 (s, C5);
1
6
0.4 (s, C4); 61.4 (s, C6); 128.9 (d, J (C,P) = 62.5, ipso-C
3
from PPh ); 129.4 (d, J (C,P) = 11.9, meta-C from PPh );
3
3
4
2
1
32.2 (d, J (C,P) = 2.3, para-C from PPh ); 134.4 (d, J
3
3
1
1
(
C,P) = 13.7, ortho-C from PPh ); 198.9 (s, C2). P{ H}
3
NMR [202 MHz, acetone, δ (ppm)]: 33.19 (s).
Anal. Calc. for [C H NPS Au]Cl: C, 50.43; H,
Inhibition of thioredoxin reductase (TrxR) activity
3
2
36
2
4
.76; N, 1.84. Found: C, 50.71; H, 4.85; N, 1.91 %. MS
+
+
(
(
MALDI): m/z Calc. for [C H NPS Au] [M] Calc.
726.7) found 726.5.
4): yellow solid. TG/DTA: 112–113 °C (mp). IR ν
Rat liver TrxR (Sigma) was used to determine TrxR inhibi-
tion by the compounds. The assay was performed accord-
ing to the manufacture’s instructions (Sigma T9698)
with appropriate modifications. Initially, 20 µL of com-
pounds (final concentration 5 μM in 5 % DMSO/potas-
sium phosphate buffer, pH 7.0) were added to 20 µL of
TrxR water solution (approximately 0.10 units) and incu-
bated for 1 h at 37 °C. After incubation, 600 µL of reac-
tion mixture containing substrate dithiobisnitrobenzoic acid
(DTNB) was added and the formation of 5-thionitroben-
zol (TNB) was monitored in SHIMADZU UV–Visible
3
2
36
2
(
max
−1
KBr (cm ): 2964, 2906, 2844, 1483, 1415, 1296, 1230,
−
1
1
1
149, 771. Raman νmax (cm ):314 (Au–S).). H NMR
[
1
300 MHz, CDCl , δ (ppm), J (Hz)]: 1.1–1.2 (m, 9H, CH );
3
3
.6–1.9 (m, 21H, H-adamantyl, P(CH –CH ) ); 3.2 (t, 2H,
2 3 3
H5, J = 7.4); 3.4 (s, 2H, H6); 4.1 (t, 2H, H4, J = 7.4).
5
,4
4,5
1
3
C NMR [75 MHz, CDCl , δ (ppm), J (Hz)]: 8.9 (s,
3
P(CH –CH ) ); 18.2 (d, J (C–P) = 37 Hz, from PEt ); 27.6;
2
3 3
3
2
8.3; 36.7; 41.4 (s, C-adamantyl); 36.8 (s, C5); 60.1 (s,
1
3

J Biol Inorg Chem
Spectrophotometer (UV-1601PC) at 412 nm for 4 min. The
reaction mixture consisted of 200 µL of 500 mM EDTA
solution pH 7.5, 800 µL of DTNB solution (63 mM) in
ethanol, 50 µL of nicotinamide adenine dinucleotide phos-
phate (NADPH) (48 mM), 100 µL of 20 mg/mL bovine
serum albumina(BSA), 1 mL of 1.0 M potassium phos-
phate buffer and 7.85 mL of water. The absorbance was
obtained from two independent experiments in triplicate.
The enzymatic activity of the samples was calculated in
percentage using the absorbance of the positive control (no
inhibitor) as maximum activity (100 %). The blank solu-
tions were prepared as samples described above for each
compound without TrxR.
its neutral zwitterionic form, regardless of the pK of the
a
side chain. The transition state (TS) structures were first
optimized and characterized as first-order saddle point on
the PES through harmonic vibrational analysis (only one
imaginary frequency). Then, the activation free energy
(ΔG ) and reaction free energy (ΔG ) were obtained using
a
r
infinitely separate reactants. The reaction paths were cal-
culated in aqueous solution (IEF-PCM, ε = 78.355) at the
B3LYP/6-31+G(2d)/SDD(f) level with T = 298.15 K and
p = 1 atm. The quantum mechanical calculations were car-
ried out using Gaussian 09, Revision A.02 [48].
®
The docking studies were conducted using the GOLD
package [49–51]. The human TrxR1 crystal structure coded
2
J3 N in the PDB was taken as a target [52]. This struc-
Glutathione interaction
ture has the Sec-498 residue (at the end of the C-terminal
arm) mutated to Cys-498, which is at the solvent exposure
region in the reduced form of the enzyme and represents
the main local binding site. The unit cell is composed of
three dimers and only the pair C–D was considered with
Phosphate buffer pH 7.0 solutions containing gold com-
plexes were prepared (DMF: 0.5 % v/v) at 500 μM. Aque-
ous solutions of reduced glutathione (GSH) from Sigma
were used at 250 μM. The experiment was performed in a
+
two FAD and two NADP molecules in addition to the
9
2
6-well plate. Each well received 25 μL of GSH solution,
5 μL of buffer containing gold complexes or DMF as con-
ligand 2-methyl-2,4-pentanediol (MPD). The enzyme was
first prepared by adding hydrogens using standard protona-
tion states at physiological pH and removal of the NADP
+
trol, 25 μL of 100 mM aqueous EDTA solution pH 7.5 and
incubated at 37 °C for 1 h. To each well, 200 μL of reaction
mixture (1000 μL reaction mixture consisted of 620 μL
potassium phosphate buffer pH 7.0, 80 μL, 100 mM EDTA
solution pH 7.5, and 300 μL distilled water) was added,
and the reaction was started with the addition of 25 μL of
a 20 mM ethanolic solution of DTNB. The formation of
cofactor and MPD ligand. The docking site was chosen at
the C-terminal motif defined as a spherical region of 15 Å
centered on the sulfur atom of the Cys-498 residue. The
®
GOLD program uses genetic algorithm (GA) to explore
the ligand–protein binding space [49–51]. Here, 100 GA
runs were carried out for the free ligands (A) and (B) and
the top 100 docking poses scored using the piecewise linear
potential (PLP) scoring function. The docking for ligands
on the TrxR binding site was performed without any con-
straint applied. For the gold complexes (1–4), 50 GA runs
were analyzed and the corresponding poses scored with
PLP scoring function. As the gold atom is not formally
accounted for in the searching algorithms and scoring
function, the distance between the sulphur atom of Cys-
498 (considered the center of the enzyme binding site) and
gold atom was constrained within the 1.5–3.5 Å range in
the docking procedure with the spring constant (k) set to 5
5
-TNB was monitored in a microplate reader at 405 nm
[13].
DFT and docking studies
Calculations were carried out at distinct levels depend-
ing on the property desired. Firstly the geometries of the
ligands (A and B) and gold(I) complexes (1–4) were opti-
mized in the gas phase at a density functional theory (DFT)
level with hybrid three-parameter B3LYP functional [42]
and 6-31+G(2d) basis sets for all atoms, except gold,
which was treated with the ECP-SDD [43] augmented with
®
(in arbitrary units). In the GOLD program [49–51], if the
one set of f polarization function (α = 1.13863280) [44–
constrained distance is found outside the specified bounds,
f
2
4
6] (hereafter abbreviated as B3LYP/6-31+G(2d)/SDD(f)).
spring energy is used to reduce the fitness score (E = kx :
Vibrational spectra were calculated and assigned in the gas
phase at the very same level of theory used for obtaining the
x is the difference between the bounds and actual distance
values). The intermolecular interactions of the best poses
were analyzed using the Discovery Studio 4.0 software
[53], which was also used to produce the images.
1
13
geometries. H and C NMR spectra were predicted at the
B3LYP/6-31+G(2d,p)/SDD(f) level in CHCl (ε = 4.711)
3
where the solvent was included through the IEF-PCM for-
malism [47] using the UFF radii to define the solute cavity.
The chemical shift was referenced to TMS.
Results and discussion
The ligand exchange reaction paths for all four com-
plexes were calculated considering the Sec amino acid as
the entering group. The Sec structure was considered in
All compounds were synthesized in good yields (over
80 %) and after recrystallization, ligands and complex (1)
1
3

J Biol Inorg Chem
were obtained as single crystals. All compounds were char-
acterized by NMR, infrared and Raman spectroscopies, ele-
mental analyses, mass spectroscopy and X-ray diffraction.
According to the results of TG/DTA, all complexes were
found to be thermally stable up to 200 °C and endothermal
decomposition was preceded by melting. The mass spectra
influence from gold(I) coordination; however, it is not clear
experimentally because of the coupling with the δ (NNH)
ip
deformation in the free ligand. Similar arguments can be
used for the ν(N3N4) stretching, which is predicted to shift
−1
−51 cm from ligand (A) to complex (1), but could not be
assigned experimentally due to the overlapping of intense
+
−1
showed pseudomolecular ions [M+H] of compounds (1)
phosphine vibrations at 1034 cm . The assignment of the
+
and (2), and [M] for compounds (3) and (4) in accordance
low frequency region is quite complex due to the strong
coupling of vibrational modes and low intensity. Experi-
mentally, the ν(AuS) was attributed for complex (1) at
with the structures proposed in scheme 2.
−1
IR and Raman vibrational analysis
377 cm . Theoretically two modes have contribution from
−1
ν(AuS) stretching at 294 and 369 cm . For complex (2),
−
1
Ligand (A) can exist in two tautomeric forms with
the thione form (NH) predominating in solution. At
B3LYP/6-31+G(2d,p)/SDD(f) level, the thione form was
only theoretical data are provided at 295 and 370 cm and
do not differ significantly from those calculated for com-
plex (1).
−1
1
1.8 kcal mol more stable than the thiol tautomer (SH) in
The Raman spectra for ligand (B) exhibited several
CHCl solution. The IR spectra for complexes (1) and (2)
bands, most of them assigned as CH vibrations (Table
3
2
were analyzed in comparison to the IR of the ligand (A) and
complexes (3) and (4) were compared to the IR of ligand
S2), which overlap with the most relevant vibrations
from the heterocyclic ring. These main bands are at 1481
(νC2N3), 1298 (νC4N3), 1232 (νC6N3), 1061 (νC2S),
(
B). The infrared spectrum of ligand (A) shows absorp-
−1
−1
tions corresponding to υ
The 2914 and 2850 cm regions show absorptions due to
at 3122, 3092 and 1505 cm .
987 (νC4C5) and 700 cm (νC5S1) for the free ligand
N–H
−
1
(B). In general, the position of these bands is not very sen-
sitive to gold coordination, except for the one assigned to
ν(C2N3), which is predicted to shift to the blue region by
CH and CH groups. Other absorptions that were observed
2
−
1
−1
are: υCN/δNH (1605 and 1181 cm ) and υ (750 cm ). In
CS
−1
the IR spectra of complexes (1) and (2), in addition to the
absorptions observed for the ligands, we have noticed the
absence of υN–H and thioamide bands [54].
~20 cm , though, this was not observed experimentally.
Indeed, an increase of the C2N3 bond order is expected
after the gold coordination due to the electron delocaliza-
tion over the N3–C2–S moiety, passing from N–C=S to
N=C–S canonical form. The C2–N3 bond lengths for (B)
and (4) were calculated as 1.36 and 1.33 Å, respectively.
The ν(AuS) stretching was experimentally assigned at 330
Ligand (B) shows absorptions in the regions of 2905 and
−
845 cm assigned to CH and CH. The thioamide bands
2
1
2
−
1
were observed at 1481 and 1296 cm resulting from vibra-
−
1
tional coupling of N–C=S. The absorptions at 1229 cm
−1
were attributed to δ [54]. In the IR spectra of complexes
and 314 cm for (3) and (4), respectively. Theoretically,
CN
(
3) and (4) no significant changes were observed. The IR
two vibrations are assigned as AuS stretching around 230
−1
spectra of complexes (1) and (3) show absorptions attribut-
able to PPh in the regions of 3050, 1145, 1101, 748, 692,
5
and 335 cm . These values are lower than the correspond-
ing frequencies assigned for complexes (1) and (2), which
support the experimental trend and agree with the predicted
structures where the Au–S bond length for complexes (2)
and (4) were 2.38 and 2.40 Å, respectively.
3
−
1
40, 501 cm .
The main Raman assignments are shown in Tables S1
and S2 for the series of compounds derived from ligands
(
(
(
A) and (B), respectively. Experimental data for ligand
A) and complex (1) are given in Table S1. For complex
2) the experimental Raman spectrum was of low qual-
NMR studies
1
ity and could not be assigned, therefore only theoretical
data are provided. The oxadiazoline vibrational modes
were assigned experimentally for ligand (A) at 1606
In the H NMR spectra, the hydrogen signals of the CH
2
and CH of the adamantine ligand appear in the δ 1.5–2.0
region. For ligand (A), a signal at δ 11.7 was also observed
which was attributed to NH and disappears in the spectra
of complexes (1) and (2). The calculated chemical shift for
NH of ligand (A) was 8.0 ppm, which is in closer accord-
ance with the experimental data than that predicted for
the SH tautomer (4.9 ppm). These data support the pre-
dominance of the NH tautomer in solution, as previously
(
νC5N4), 1441 (νC2N3), 1169 (νC2O1), 1062 (νN3N4)
−1
and 1044 cm (νC5O1) with the former being quite sensi-
tive to the gold(I) coordination. In complex (1), for which
a Raman spectrum was obtained, the ν(C5N4) stretching
frequency was at 1566 cm , a shift of 40 cm compar-
ing to the free ligand (A). A red shift was also predicted
theoretically from 1646 to 1606 cm , with calculated
frequencies overestimated as expected for the harmonic
oscillator approach. The ν(C2N3) mode also suffers some
−
1
−1
−
1
1
assessed from thermodynamic data. The H NMR spec-
tra of ligand (B) show the signal of adamantane and three
more signals: a singlet at δ 3.4 and two triplets at δ 3.2 and
1
3

J Biol Inorg Chem
Table 1 13C NMR chemical shift for the main atoms involved with
the coordination to gold metal
For the gold(I) complexes the 13C NMR signals do
not differ much from those predicted for the free ligands,
except for the set of peaks due the phosphine ligand that
C2
C5
C6
C4
were observed between 129 and 134 ppm to PPh and
Ligand (A)a 178.5 (178) 170.6 (169)
3
–
–
–
–
13
31
8
–18 ppm to PEt . Due to C– P coupling, the phenyl
3
(
1)
2)
176.2 (172) 160.3 (171)
174.6 (172) 168.5 (170)
–
carbons of complexes (1) and (3) appear as doublets reso-
nances as described in the literature [28]. The calculated
(
–
Ligand (B)a 198.5 (199)
36.6 (35.6) 61 (62)
37.0 (37.5) 61.4 (66)
59.9 (61)
60.4 (64.9)
spectra show the PPh signals in the 125–135 ppm region
3
(3)
4)
198.9 (201)
198.5 (201)
and PEt in the 8–25 ppm region for all four complexes,
3
(
36.8 (37.1) 61.2 (65.5) 60.1 (64.5)
which is in accordance with the experimental data. The
adamantane carbons are assigned to the 27–42 ppm region
as predicted for the free ligand and supported by the theo-
retical spectra. For complexes (1) and (2), an upfield shift
was observed for the signals corresponding to C2 (C=S)
and C5 (C=N) 176/160 (1) and 174/168 ppm (2), respec-
tively for C2 and C5 atoms. The calculated values follow
the experimental trend: 172/171 ppm (1) and 172/170 ppm
(2). Concerning complexes (3) and (4) the carbons C4, C6
and C2 (C=S) do not differ significantly from ligand (B).
The values are in δ scale (ppm) relative to TMS. Calculated values
sa or el u gt ii ov ne n) in brackets (B3LYP/6-31 + G(2d,p)/SDD(f) level in CHCl3
a
The atoms numbering is given in Schemes 1, 2
4
.1 attributable to the CH outside and inside the heterocy-
2
clic ring, respectively. For the complexes (1) and (3) signals
were observed in the δ 7.2–7.7 region, corresponding to
1
3
aromatic hydrogens of the PPh group. For complexes (2)
The C NMR chemical shift and assignments are summa-
rized in Table 1.
3
and (4) the signal attributed to CH of PEt group appears
3
3
For the gold(I) complexes, the 31P NMR spectrum pre-
sents a single resonance at δ 25.22 (1), 38.90 (2), 33.19 (3)
in the δ 1.1–1.3 region.
The 13C NMR spectrum of the ligand (A) shows the sig-
nals of adamantane in the δ 26.7–40.9 region. The signal at
δ 170.6 is attributed to the carbon directly bounded to the
N and O in the heterocyclic ring (C5) and at δ178.5 to the
C(=S) (C2) (see Scheme 2 for numbering). These assign-
ments are supported by theoretical calculations, which
predicted the corresponding signals at δ 169 (C5) and 178
and 33.43 (4) differing slightly from PPh AuCl (δ 33.17)
3
and Et PAuCl (δ 33.45). In accordance to the other analy-
3
sis, the complexes with oxadiazole derivative ligands (1
and 2) promote bigger interference in NMR signals than
complexes with thiazoline derivative ligands (3 and 4).
In short, the Raman and NMR spectroscopic analyses,
supported by DFT calculations, and mass spectroscopy,
suggest the structures represented in Figs. 1, 2, 3, and 4.
Some of them were confirmed by X-ray diffraction which
was taken as a reference to assess the confidence limit of
the molecular modeling studies. These are discussed in the
next section.
(
C2). The chemical shifts for the carbon atoms in the ada-
mantane moiety are calculated on the range of 35–44 ppm.
For the ligand (B) the observed signals of adamantane were
in the δ 27.4–41.3 region, and signals at δ 59.9 and δ 61.0
assigned to the CH groups directly bound to the N3 (C4
2
and C6, respectively). The C(=S) (C2) signal is observed
1
3
at δ 198.5. The theoretical C NMR predictions for ligand
(
B) are in perfect accordance to the experimental assign-
X‑ray structural determination
ments, with the main signals at δ 61 (CH of the hetero-
2
cyclic ring—C4), δ 62 (CH connecting the adamantane to
Crystals were obtained by recrystallization, from metha-
2
the heterocyclic ring—C6) and δ 199 (C=S—C2).
nol for ligand (A), from ethanol for ligand (B) (Fig. 1) and
Fig. 1 ORTEP drawing of the
crystal structures for ligand
(
a) and (b). The ellipsoids are
drawn at the 50 % probability
level and hydrogen atoms are
shown as spheres of arbitrary
radii
1
3

J Biol Inorg Chem
of the complex (1) (Fig. 2) consists of one Au(I) ion, one
PPh molecule and one ligand (A). The Au(I) ion is coor-
3
dinated by one P atom of the phosphine and one S atom
of the ligand (A) in a slightly distorted linear geometry
with coordination angle of 176.05(4)º. The asymmetric unit
interacts with each other by non-classical hydrogen bonds
of type C–H···N (2.75 Å) and C–H···O (2.64 Å), generating
a three-dimensional arrangement as shown in Fig. 3 along
the ac plane.
For ligands (A) and (B), the C2–S bond distances
are 1.653(3) and 1.657(3) Å, respectively, which are
shorter than the corresponding bonds in complex (1),
C2–S = 1.715(4) Å. The predicted values for the C2–S
bond length were 1.64 (A) and 1.65 Å (B), with bond order
~1.6. For complex (1) the C2–S bond was 1.73 Å long with
bond order ~1.2. For complexes (1) and (2) two forms (s
and a) were considered which differ by torsion around the
C2–S bond. The conformer with the dihedral O1-C2–S–Au
close to 180° (1-a and 2-a forms, Fig. 4) was the global
minimum, favored by ~3.6 kcal mol regardless the phos-
phine ligand. This has been confirmed by X-ray diffraction
for the complex (1), wherein the twist angle O1-C2-S–Au
is 178.1(3)º. Some bond lengths and angles for complex
(1) can be compared to the calculated values (in brackets):
Fig. 2 ORTEP drawing of the crystal structures of complex (1). The
ellipsoids are drawn at the 50 % probability level and hydrogen atoms
are shown as spheres of arbitrary radii
from a mixture of dimethylsulfoxide and dichloromethane
for complex (1), and isolated by filtration (Fig. 2). The
structures were determined using single-crystal X-ray dif-
fraction. Crystal data, data collection and structure refine-
ment details are summarized in Table 2. Ligand (A) crys-
−1
tallizes in the monoclinic crystal system and P2 /c space
1
group. Ligand (B) crystallizes in the monoclinic crystal
system and C2/c space group. The complex (1) crystallizes
in the orthorhombic crystal system and Pbca space group.
Crystallographic analysis indicates that the asymmetric unit
Table 2 Crystal data of ligands
Compound
Ligand (A)
Ligand (B)
Complex (1)
(
A) and (B) and complex (1)
Formula
C H N OS
C H NS
C H AuN OPS
1
2
16
2
14 21
2
30 30
2
−
1
Formula weight/g mol
236.33
298 (2)
267.44
694.56
Temperature/K
298(2)
298(2)
Crystal system
Monoclinic
P21/c
Monoclinic
C2/c
Orthorhombic
Pbca
Space group
a/Å
10.5545 (16)
11.0918 (14)
10.2864 (17)
90.00
25.225 (2)
9.9815 (6)
10.8885 (8)
90
19.3397 (5)
10.5189 (3)
26.2493 (8)
90
b/Å
c/Å
α/º
β/º
100.330 (14)
90.00
98.063 (7)
90
90
γ/º
90
V/Å3
1184.69 (31)
4
2714.4 (3)
8
5240.0 (3)
8
Z
Crystal size/mm
0.85 × 0.38 × 0.28
1.325
0.41 × 0.14 × 0.08
1.309
0.31 × 0.13 × 0.11
1.728
−
3
Dcalc/g cm
µ(Mo Kα)/cm
−
1
0.254
0.371
5.675
Transmission factors (min/max)
Reflections measured/unique
Observed reflections
0.886/0.944
5322/2411
1573
0.936/0.981
13870/2776
2083
0.989/0.995
38442/5463
4024
Nº. of parameters refined
R[F > 2σ(F )]
149
214
326
0.053
0.054
0.029
o
o
2
wR[F 2 > 2σ(F ) ]
0.152
0.136
0.048
o
o
S
1.002
1.033
1.060
RMS peak/
0.095
0.064
0.093
1
3

J Biol Inorg Chem
Fig. 3 View of the molecular
packing diagram along ac plane
Fig. 4 Calculated geometries for the gold(I) complexes (gas phase at
B3LYP/6-31+G(2d)/SDD(f) level). For complexes (1) and (2), two
distinct conformer are shown (“s” and “a”), differing by the relative
position around the C2–S bond (the relative energy including ZPE
correction is also provided). Some bond lengths and angle are shown
with experimental values in brackets
Au–S–C2 = 98.88(13)º [97.1º]; S–Au = 2.3122(11) Å
bonds (distance ~2.7 Å), which lead to the formation of a
two-dimensional array in solid state and, therefore, stabiliz-
ing the overall structure.
[
2.374 Å] and Au-P = 2.2584(11) Å [2.324 Å]. The S–
Au–P angle was 176° in the solid state and in gas phase,
suggesting that the intermolecular forces do not change sig-
nificantly the geometry of complex (1) around the coordi-
nation sphere. Similar geometry is predicted for complex
For complexes (3) and (4) only one conformer was con-
sidered as represented in Fig. 4. The coordination through
S1 (see Schemes 1, 2 for numbering sequence) was also
−
1
(
2) as seen in Fig. 4. Interestingly, for the analogue com-
considered, but the resulting structure was 15 kcal mol
plex containing the 5-phenyl-1,3,4-oxadiazole-2-thione
ligand (complex 4 in Ref. [28]), the X-ray structure was
assigned as the form “s” (similar to 2-s in Fig. 4) with
O1-C2–S–Au torsion angle of 13.60°. This finding was
mainly due to weak unconventional C–H···N hydrogen
higher in energy (not shown). The C2–S bond length for
complexes (3) and (4) was 1.71 Å, slightly shorter than
the corresponding bonds in (1) and (2), with bond order
~1.3. The N3-C2–S–Au dihedral was around 171° and
the C2–S–Au angle close to 107°. As expected, the ligand
1
3

J Biol Inorg Chem
Table 3 Cytotoxic activities against cell lines, inhibition activity of TrxR and GSH interaction
a
Compounds Tumor cells IC (µM ± SD)
BN Ho nK- t2u 1m or cells TrxR Inhibition (% ± SD) GSH Interaction (% ± SD)
5
0
B16-F10 SI CT26-WT SIb 4T1
b
SIb
c
c
c
c
c
A
1
>100
ND 55.0 ± 17 ND >100
ND >100
51.7 ± 2.1
40.7 ± 1.5
51.6 ± 1.6
35.7 ± 2.2
57.3 ± 0.7
60.2 ± 0.7
<0.1
5.7 ± 0.5 3.2 5.7 ± 0.9 3.2 6.6 ± 0.5 2.8 18.5 ± 2.9
1.2 ± 0.2 4.5 1.8 ± 0.8 3.0 1.6 ± 0.5 3.4 5.5 ± 0.1
9.0 ± 0.6 1.9 34.2 ± 0.9 0.5 30.9 ± 3.2 0.5 17.2 ± 6.7
1.8 ± 0.5 3.8 1.8 ± 0.3 3.8 3.0 ± 1.8 2.3 6.9 ± 0.8
1.0 ± 0.1 5.8 0.9 ± 0.1 6.4 1.1 ± 0.2 5.3 5.8 ± 0.1
21.1 ± 2.0
33.3 ± 1.0
<0.1
2
B
3
11.0 ± 4.1
51.2 ± 1.5
c
d
e
4
3
2
2.0 ± 1.7
7.0 ± 0.5
AuPPh Cl
6.6 ± 0.1 3.4 12.1 ± 2.8 1.9 10.3 ± 2.3 2.2 23.0 ± 0.3
2.3 ± 0.5 9.7 9.0 ± 0.7 2.5 9.5 ± 1.1 2.4 22.5 ± 0.2
0.5 ± 0.4 3.2 0.5 ± 0.4 3.2 0.6 ± 0.2 2.7 1.6 ± 0.5
–
–
14.9 ± 3.2
52.0 ± 2.0
16.5 ± 1.0
3
AuPEt Cl
3
c
d
e
Auranofin
59.7 ± 1.5
6
5
1.9 ± 2.0
9.7 ± 1.5
Cisplatin
6.0 ± 1.0 3.0 5.0 ± 1.7 3.6 6.2 ± 2.5 2.9 18.1 ± 10.9
–
–
a
SD standard deviation of quadruplicate of two independent experiments
b
c
d
e
SI selectivity index– the ratio of the IC50 obtained from the experiment on normal cells vs. tumor cells
Concentration of compounds in TrxR assay (5 µM)
Concentration of compounds in TrxR assay (1 µM)
Concentration of compounds in TrxR assay (0.5 µM)
ND not determined
arrange in a linear geometry around the gold center, with
noticed that the presence of adamantyl-heterocyclic ligand
improved cytotoxicity and justifies the designing new
ligands for Au(I) complexes. All complexes were more effi-
cient and selective than cisplatin to murine melanoma cell
line (B16-F10). With respect to CT26-WT (murine colon
carcinoma), the complexes (3) and (4) were more active
∠
S–Au–P = 176° for all complexes.
Biological tests
The cytotoxic activity of the compounds was evaluated
in comparison with auranofin and cisplatin in three tumor
cell lines to assess the activity in different embryonic ori-
gin (epithelial and fibroblast) and also in a normal kidney
cell that was used to evaluate the selectivity index. IC50 val-
ues, calculated from the cell viability dose response curves
obtained after 72 h drug treatment in the MTT test and per-
centage of inhibition of TrxR are shown in Table 3.
(IC50 = 1.8 ± 0.3 µM and IC = 0.9 ± 0.1 µM, respec-
50
tively) than cisplatin (IC = 5.0 ± 1.7 µM). In metastatic
5
0
mouse mammary adenocarcinoma (4T1) the complexes
with triethylphosphine (2) (IC50 = 1.6 ± 0.5 µM) and (4)
(IC50 = 1.1 ± 0.2 µM) present better results when com-
pared to cisplatin (IC = 6.2 ± 2.5 µM), even for selectiv-
5
0
ity. Complex (4) showed IC50 values close to the auranofin;
however, it has a greater SI for the tested cell lines that
presents a combination of alkyl phosphine and thiazolidine
ligands.
All gold complexes were more active in the tested cell
lines than their respective ligands and phosphane gold pre-
cursor complexes [Et PAuCl] and [Ph PAuCl] presenting
3
3
higher selectivity in most of the cases. In general, oxadia-
zole complexes were up to 80 times more active than the
free ligand whereas thiazolidine complexes were up to 38
times more active than the free ligand. The thiazolidine
complexes were more active than oxadiazole complexes
in all cases. These results can be explained in part by the
activity presented by the free ligand. Comparing different
phosphine ligands, complexes with triethylphosphine (2)
and (4) were more active than their triphenylphosphine
analogues (1) and (3), which is reflected in the same pro-
file presented by [Et PAuCl] and [Ph PAuCl] (Table 3). We
We also observed the effect of albumin upon cytotox-
icity of the compounds. Albumin has one free cysteine
residue that can promote sulfhydryl exchange reactions.
In cells exposed concomitantly to albumin and gold com-
pounds, the IC50 obtained showed a protective effect of
albumin increasing values up to ten times in tumor cells
(4T1) and up to five times in normal cells (BHK-21). These
results are showed in Fig. 5 and suggest the metabolic role
of albumin in gold complexes that was previously reported
for gold anti-arthritic drugs. More than 80 % of the circu-
lation gold in treated patients is bound to plasma protein
3
3
1
3

J Biol Inorg Chem
Fig. 5 Effect of albumin in
cytotoxic activities against
tumor cell line 4T1 and normal
cell line BHK-21
albumin [41]. From structure–activity analyses, we noticed
that the triethylphosphine derivatives remain more active
than the triphenylphosphine analogues even in presence of
albumin.
The assays of inhibition of TrxR can be related with
cytotoxicity, except for ligand (A) that even inhibited TrxR
but did not show any activity in analyzed cells. The differ-
ence may be explained by transport into the cell. The results
show values between 36 and 60 % of enzymatic inhibition.
The triethylphosphine complexes are slightly more active
than the triphenylphosphine analogues, and thiazolidine
respectively, reaching the same Auranofin inhibition activ-
ity only at 5 μM, Fig. 6.
Glutathione is another –SH target for gold drugs. The
results in Table 3 show, as expected, no significant interac-
tion for ligands and different potential interaction with gold
compounds. An interesting structure–reactivity relationship
is observed where the triethylphosphine analogues (2,4) are
more susceptible for reaction with glutathione, which are in
line with the TrxR inhibition.
Different sulfur targets can interact with gold drugs as it
is already knew for auranofin and, therefore, some compe-
tition can occur protecting cells from their action, but these
interactions are not able to deactivate 100 % of the drug
that still is active in tumor cells.
are more potent than the oxadiazole analogues. [Et PAuCl]
3
and [Ph PAuCl] are well known as TrxR inhibitors and this
3
has been previously discussed in the literature [55]. Com-
plex (4), the most cytotoxic complex, also showed the best
inhibition when compared to auranofin. Auranofin showed
higher inhibition effect even at 0.5 and 1.0 μM (58 %) con-
centration levels, which complex (4) inhibit 27 and 32 %,
DFT and docking analyses of the ligand‑TrxR binding
Molecular modeling analyses including DFT calculations
and drug-receptor docking were conducted in the present
study aiming to gain some insights into the binding modes
of the novels TrxR inhibitors described herein. It is well
accepted that the C-terminal moiety of TrxR (–G496–
C497–U498–G499–) is the main target for gold-containing
inhibitors due to the very attractive soft base Sec-498 resi-
due [55]. Firstly, the ligand exchange reactions for com-
plexes (1–4) were investigated where the ligands (A) or (B)
were taken as the leaving groups and the Sec amino acid
as the entering group. The sulfur ligand is more likely to
be replaced due to the high trans effect of the phosphine
moiety. An associative mechanism, with the proton trans-
fer observed upon nucleophilic attack, was assumed as
described previously for auranofin [46] where a distorted
trigonal planar geometry is proposed for the transitions
state (TS). The reactive species for the direct reactions,
including the TS structures, are represented in Fig. 7.
The structural parameters around the first coordination
sphere are also shown in addition to the Gibbs free energy
1
00
Auranofin
Complex 4
80
60
40
20
0
0
.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-1
Concentration (µmol.L )
Fig. 6 Inhibition of TrxR activity by Auranofin and Complex 4 in
different concentrations (0.5, 1.0, 5.0 μM). Experiments were per-
formed in triplicate and the results represent the mean ± SD
1
3

J Biol Inorg Chem
Fig. 7 Reagents (R), transition states (TS) and products (P) cal-
culated for the ligand exchange reactions of the gold complexes.
The structures were optimized in aqueous solution at B3LYP/6-
31+G(2d)/SDD(f). The free energy barrier (ΔGa,aq) and reaction free
−1
energy (ΔG ) are in kcal mol and bond lengths in Å
r,aq
barrier (ΔGa,aq) and reaction Gibbs free energy (ΔGr,aq).
The structures and energies were calculated at B3LYP/6-
It is well accepted that ligand exchange rate might be rel-
evant for the final biological response of metal-based drugs.
In our previous study, a structure–activity relationship
3
1+G(2d)/SDD(f) level in aqueous solution.
1
3

J Biol Inorg Chem
Fig. 7 continued
(
SAR) was established for a series of Pt(II) complexes using
as that found for Sec nucleophilic attack could be estab-
lished, the hydrolysis process being slower than the attack
of the Sec amino acid (data not shown). To illustrate, for
complex (2) the reaction with Sec residue was four times
faster than the reaction with water, whereas for complex (4)
the rate ratio was almost three orders of magnitude greater.
Moreover, it might be concluded based on solid theoretical
grounds that although complexes (3) and (4) react quickly
with the Sec residue, they are relatively stable in aqueous
solution. The same is true for oxadiazoline complexes (1)
mutagenicity and the hydrolysis rate constants as depend-
ent and independent variables, respectively [56]. From
Fig. 7, it is evident that the ligand-exchange reactions for
thiazolidine complexes (3) and (4) are much faster (lower
−
1
energy barrier—ΔGa,aq ~ 17 kcal mol ) than for oxadiazo-
−1
line complexes (1) and (2) (ΔGa,aq ~ 24 kcal mol ), even
though the processes for the latter are slightly more favora-
ble. The reactivity of these complexes using water as nucle-
ophile was also theoretically predicted and the same trend
1
3

J Biol Inorg Chem
Fig. 8 Scheme of the TrxR binding site composed by Subunits C (C-terminal motif) and D (N-terminal motif). The ligand-receptor binding
modes for the compounds studied, as predicted from docking analysis, are shown
1
3

J Biol Inorg Chem
and (2). The reaction of auranofin with Sec was described
previously [46] using the very same level of theory applied
Nonetheless, the Au–S(C498) distances were around 3.6
and 4.0 Å for oxadiazoline and thiazolidine complexes,
respectively, which are in line with the corresponding dis-
tances calculated for the TS structures shown in Fig. 7. For
the complexes (1) and (3) an additional π-stacking inter-
action is predicted involving the phenyl ring from P(Ph)3
ligand and H472 residue.
For all complexes analyzed, the docked structure
assumes a “folded” form, in between the “s” and “a” con-
formers, with the torsion angle around the C-S bond ~85°–
130° in the enzyme complex compared to ~171–180° in the
isolated complex. This finding suggests that this “folded”
form fits better to the TrxR binding site and thus puts the
complex in a more favorable position for nucleophilic
attack, which is an important event to improve TrxR inhibi-
tion. For complexes (2) and (4) (Fig. 8d, e, f), the strongest
TrxR inhibitors according Table 3, the PLP score was 52
and 57, which is in line with observed biological responses
quoted in Table 3.
From previous analyses, we may suggest some molecu-
lar features for the mechanism of action for the compounds
reported herein considering TrxR as the primary target: (1)
the compounds prepared should interact in a competitive
mode with the thioredoxin (trx) binding site, namely the
pocket around the C-terminal motif of TrxR enzyme; (2)
no specific anchoring points are observed over the docking
mode, with ligand (A) being the only compound showing a
moderate hydrogen bond with the W407 residue, remem-
bering that ligand (A) is more potent as an enzyme inhibitor
than ligand (B); (3) for all four gold complexes the docking
modes are driven by hydrophobic interaction; (4) As gener-
ally accepted for metallodrugs, the chelating ability is the
main feature responsible for the biological response. For
the described gold complexes, the best biological response
(TrxR inhibition) was observed for complexes (3) and
(4), which were fairly reactive with the Sec residue; (5) A
“folded” form in between the “s” and “a” conformers of the
complexes seems to play a role in the docking mode.
−
1
here. The results were (in kcal mol ) ΔG
= 29.8 and
a,aq
ΔGr,aq = −5.5, which are closer to those found for oxa-
diazoline derivatives (see Fig. 7). For analogue compounds,
the kinetic parameters might be related with biological
responses as shown for platinum complexes [56]. From
Fig. 7 and Table 3 we note that the most reactive complexes
(
3) and (4) show higher cytotoxicity and higher enzyme
inhibition compared to the corresponding (1) and (2) oxadi-
azoline derivatives, which is evidence that the rate of ligand
exchange might play a role on drug–enzyme binding and
thereby contribute to the overall biological response. The
structure–reactivity relationship proposed previously [45,
4
6] is also observed here, with the energy barrier decreas-
ing with an increase of the “reactive angle” between the
entering and leaving groups (see angles values in Fig. 7).
Lastly, the ligand–receptor docking analysis was per-
formed for the free ligands and gold(I) complexes. The best
pose for each compound (highest score) is represented in
Fig. 8 with the main amino acid residues defining the bind-
ing pocket also included. Only one of the two active sites
in dimeric TrxR enzyme is represented, which is located
at the interface between the Subunits C (C-terminal motif
–
G496–C497–C498–G499–) and D (N-terminal motif –
C59–V60–N61–V62–G63–C64–). For the free ligands
Fig. 8a, b) the binding modes involve one hydrogen bond
(
in addition to several hydrophobic contacts. For ligand (A)
a relatively strong hydrogen bond is predicted with resi-
due W407, which acts as a hydrogen donor. The N–H···O1
distance was 1.84 Å. For ligand (B), a weak hydrogen
bond is also observed involving the residue L409, with
the N–H···S=C hydrogen bond distance equal to 2.03 Å.
It is noteworthy that for such small molecules where spe-
cific anchoring points are not found, the prediction of the
best pose is a challenge. Nonetheless, the strong hydrogen
bond predicted for ligand (A) might provide additional
evidence for the higher biological response of ligand (A)
compared to ligand (B) when TrxR inhibition is consid-
ered, even though, it was not observed in cytotoxicity stud-
ies (see Table 3). When considering gold(I) complexes, the
hydrophobic interactions dominate and the best binding
poses predicted are similar for all complexes (Fig. 8c, d,
e, f). The adamantane ring lies over the Subunit C which
acts as a hydrophobic cushion, and the phosphine ligand
is pointed towards the N-terminal motif. Both ligands are
involved in van der Waals contacts, with electrostatic inter-
action observed only between the metal center and C498
and Y116 residues. It is important to bear in mind that
these docking runs were conducted under Au–S(C498) dis-
tance constraint, therefore the C498-Au short contact is not
meaningful and was used as a bouldering condition to set
up the gold complex close to the enzyme reactive center.
Concluding remarks
The present work describes the synthesis and characteriza-
tion of two ligands and four novel gold(I) complexes con-
sidered as potential anticancer drugs. The structural char-
acterization of the compounds was established on the basis
of elemental analysis and spectroscopic data with the aid of
DFT calculations. For complex (1), the X-ray structure was
obtained and used as reference to propose the structures
of the other complexes and assess the theoretical protocol.
Complex (4) was shown to be more selective than auranofin
and quite similar in cytotoxicity. Moreover, albumin inter-
ference, glutathione interaction and TrxR inhibition was
1
3

J Biol Inorg Chem
tested and correlated with assays in tumor cells. Even for
gold complexes interacting to other thiol containing mol-
ecules, the results suggest TrxR as possible target for these
compounds. Complex (4) with triethylphosphine and thia-
zolidine ring was the most cytotoxic and active against
TrxR among the tested analogues.
9. Marzano C, Gandin V, Folda A, Scutari G, Bindoli A, Rigobello
MP (2007) Free Radical Biol Med 42:872–881
1
0. Messori L, Scaletti F, Massai L, Cinellu MA, Gabbiani C, Ver-
gara A, Merlino A (2013) Chem Commun 49:10100–10102
11. Liu C, Liu Z, Li M, Li X, Wong Y-S, Ngai S-M, Zheng W, Zhang
Y, Chen T (2013) Plos One 8:e53945
1
2. Ortego L, Cardoso F, Martins S, Fillat MF, Laguna A, Meire-
les M, Villacampa MD, Gimeno MC (2014) J Inorg Biochem
Molecular modeling was investigated in order to
obtain insights on the mechanism of action. The first
approach was to evaluate the reactivity of the gold com-
plexes against the Sec residue. The results showed that
the most active complexes (3,4) react with Sec much
faster than complexes (1,2). The rate constant for the
1
30:32–37
13. Rubbiani R, Kitanovic I, Alborzinia H, Can S, Kitanovic A,
Onambele LA, Stefanopoulou M, Geldmacher Y, Sheldrick
WS, Wolber G, Prokop A, Woelfl S, Ottt I (2010) J Med Chem
5
3:8608–8618
1
4. Meyer A, Gutierrez A, Ott I, Rodriguez L (2013) Inorg Chimica
Acta 398:72–76
15. Liu N, Huang H, Dou QP, Liu J (2015) Oncoscience 5:457–466
16. Garcia-Moreno E, Gascon S, Atrian-Blasco E, Rodriguez-Yoldi
MJ, Cerrada E, Laguna M (2014) Eur. J Med Chem 79:164–172
7. Lessa JA, Guerra JC, de Miranda LF, Romeiro CFD, Da Silva
JG, Mendes IC, Speziali NL, Souza-Fagundes EM, Beraldo H
(2011) J Inorg Biochem 105:1729–1739
8. Meyer A, Oehninger L, Geldmacher Y, Alborzinia H, Wolfl S,
Sheldrick WS, Ott I (2014) ChemMedChem 9:1794–1800
9. Sun SY, Yue P, Hong WK, Lotan R (2000) Cancer Res
60:7149–7155
−
1
−1
ligand exchange process was in the order of 10 M
s
−
5
−1 −1
for complexes (3,4) and 10
M
s for complexes
(
1,2). Therefore, the higher biological response for com-
1
plexes (3,4) might be due to the faster ligand exchange
upon interaction with the TrxR binding site. In the second
approach, molecular docking was accomplished for both
free ligands and all gold complexes. Concerning the free
ligands, the best TrxR inhibitor was observed to be ligand
1
1
(
A), which showed a short hydrogen bond (1.84 Å) with
20. Horvat M, Uzelac L, Marjanovic M, Cindro N, Frankovic O,
Mlinaric-Majerski K, Kralj M, Basaric N (2012) Chem Biol
Drug Des 79:497–506
the W407 residue. For ligand (B) and all gold complexes,
the hydrophobic contacts dominate, with the adamantane
ligand playing a role for ligand-receptor stabilization.
Interestingly, for all complexes a conformation change is
predicted from “a” to a “folded” form, characterized by a
torsion around the C–S bond. Even though we do not have
a causative relationship, the “folded” form might serve as
template to design molecules with potential for inhibiting
the TrxR enzyme.
2
1. Parkes JD, Baxter RC, Marsden CD, Rees JE (1974) J Neurol
Neurosurg Psychiatr 37:422–426
22. Bleich S, Wiltfang J, Kornhuber J (2003) N Engl J Med
49:609–610
3. Manjunatha K, Poojary B, Lobo PL, Fernandes J, Kumari NS
2010) Eur J Med Chem 45:5225–5233
3
2
(
24. Kadi AA, El-Brollosy NR, Al-Deeb OA, Habib EE, Ibrahim TM,
El-Emam AA (2007) Eur J Med Chem 42:235–242
2
2
2
2
5. Kadi AA, Al-Abdullah ES, Shehata IA, Habib EE, Ibrahim TM,
El-Emam AA (2010) Eur J Med Chem 45:5006–5011
6. Bektas H, Ceylan S, Demirbas N, Alpay-Karaoglu S, Sokmen
BB (2013) Med Chem Res 22:3629–3639
7. Patel MB, Modi NR, Raval JP, Menon SK (2012) Org Biomol
Chem 10:1785–1794
8. Chaves JDS, Neumann F, Francisco TM, Corrêa CC, Lopes
MTP, Silva H, Fontes APS, De Almeida MV (2014) Inorg Chim-
ica Acta 414:85–90
Acknowledgments The authors wish to thank CNPq, FAPEMIG
and CAPES for financial supports and fellowships. HFDS would like
to thank CNPq (485779/2013-7) and HS (455548/2014-5) for provid-
ing support for this study. This work is a collaborative research project
with members of Rede Mineira de Química (RQ-MG) supported by
FAPEMIG (Project: CEX - RED-0010-14) and HS (APQ-01648-14).
2
3
9. Serebryanskaya TV, Lyakhov AS, Ivashkevich LS, Schur J, Frias
C, Prokop A, Ott I (2015) Dalton Trans 44:1161–1169
0. Silva H, Barra CV, Costa CF, Almeida MV, César ET, Silveira
JN, Paula FCS, Maia ECP, Fontes APS (2008) J Inorg Biochem
References
1
02:767–772
1
2
3
4
.
.
.
.
Sutton BM (1986) Gold Bull 19:15–16
Forestier J (1934) The Lancet 646-648
3
3
1. Scheffler H, You Y, Ott I (2010) Polyhedron 29:66–69
2. V. 1.171.35.21 (release 17–11-2008 Oxford Diffraction Ltd.,
Berners-Price SJ, Filipovska A (2011) Metallomics 3:863–873
Madeira JM, Gibson DL, Kean WF, Klegeris A (2012) Inflam-
mopharmacology 20:297–306
Gandin V, Fernandes AP, Rigobello MP, Dani B, Sorrentino F,
Tisato F, Bjornstedt M, Bindoli A, Sturaro A, Rella R, Marzano
C (2010) Biochem Pharmacol 79:90–101
1
3:58:09) CrysAlis171. NET
3
3
3. Sheldrick GM (2008) Acta Cryst A64:112–122
4. Byrnes MJ, Chisholm MH, Clark RJH, Gallucci JC, Hadad CM,
Patmore NJ (2004) Inorg Chem 43:6334–6344
5. Sun YG, Jiang B, Cui TF, Xiong G, Smet PF, Ding F, Gao EJ, Lv
TY, Van den Eeckhout K, Poelman D, Verpoort F (2011) Dalton
Trans 40:11581–11590
5
6
7
8
.
.
.
.
3
Rubbiani R, Schuh E, Meyer A, Lemke J, Wimberg J, Metzler-
Nolte N, Meyer F, Mohr F, Ott I (2013) Med Chem Commun
3
3
3
6. Blessing RH (1995) Acta Crystallogr Sect A51:33–38
7. Farrugia LJ (1997) J Appl Crystallogr 30:565
4
:942–948
Messori L, Marchetti L, Massai L, Scaletti L, Guerri A, Landini
I, Nobili S, Perrone G, Mini E, Leoni P, Pasquali M, Gabbiani C
8. Al-Deeb A, Al-Omar MA, El-Brollosy NR, Habib EE, Ibrahim
TM, El-Emam AA (2006) Arzneim Forsch Drug Res 56:40–47
9. Sullivan FAV, Lindaw AC (1965) U.S. Patent 3.215.703
0. Baenziger NC, Bennet WE, Soboroff DM (1976) Acta Crystal-
logr Sect B 32(3):962–963
(
2014) Inorg Chem 53:2396–2403
Rigobello MP, Folda A, Dani B, Menabo R, Scutari G, Bindoli A
2008) Eur J Pharmacol 582:26–34
3
4
(
1
3

J Biol Inorg Chem
49. Jones G, Willett P, Glen RC (1995) J Mol Biol 245:43–53
50. Jones G, Willett P, Glen RC, Leach AR, Taylor R (1997) J Mol
Biol 267:727–748
51. Verdonk ML, Cole JC, Hartshorn MJ, Murray CW, Taylor RD
(2003) Proteins Struct Funct Genet 52:609–623
52. Fritz-Wolf K, Urig S, Becker K (2007) J Mol Biol 370:116–127
53. Discovery Studio v3.5.0.12158. (2005) Accelrys Inc
4
1. Garcia-Orad A, Arizti P, Sommer F, Silvestro L, Massiot P,
Chevallier P, Gutierrez-Zorrilla JM, Martinez de Pacorbo M,
Colacio E, Tapiero H (1993) Biomed Pharmacother 47:363–370
2. Becke AD (1993) J Chem Phys 98:5648–5652
3. Andrae D, Haussermann U, Dolg M, Stoll H, Preuss H (1990)
Theor Chim Acta 77:123–141
4
4
4
4
4. Dos Santos HF, Paschoal D, Burda JV (2012) Chem Phys Lett
5
48:64–70
5. Dos Santos HF, Paschoal D, Burda JV (2012) J Phys Chem A
16:11015–11024
54. Rao CNR, Venkataraghavan
18:541–547
R (1962) Spectrochim Acta
1
55. Bindoli A, Rigobello MP, Scutari G, Gabbiani C, Casini A, Mes-
sori L (2009) Coord Chem Rev 253:1692–1707
56. Costa LAS, Rocha WR, De Almeida WB, Dos Santos HF (2005)
J Inorg Biochem 99:575–583
4
4
4
6. Dos Santos HF (2014) Comput Theor Chem 1048:95–101
7. Scalmani G, Frisch MJ (2010) J Chem Phys 132:114110
8. Frisch MJ et al. (2009) Gaussian 09, Revision A.02, Gaussian
Inc, Wallingford CT
1
3