Gold(I) complexes with aryl-thiosemicarbazones: Molecular modeling, synthesis, cytotoxicity and TrxR inhibition

Article abstract of DOI:10.1016/j.poly.2017.05.004

Thiosemicarbazones and their metal complexes are of considerable interest because of their chemical and biological properties. Gold(I) compounds form a family of potential anticancer agents that are currently undergoing intense preclinical investigations. Novel gold(I) complexes were obtained through the reaction of aryl-thiosemicarbazone ligands with KAuCl₄ and Au(PPh₃)Cl. The complexes were characterized by elemental analyses, Raman, IR and NMR spectroscopies, electrospray ionization mass spectrometry, TG/DTA, cyclic voltammetry and with the aid of molecular modeling. The cytotoxicity of the complexes was assessed against tumor cell lines (B16-F10 and CT26.WT) and non-tumor cell line (BHK-21). Most of the tested complexes showed satisfactory cytotoxic activity, some of them being even more cytotoxic in tumor cells than cisplatin with a high selectivity index in the tested cells. The gold complexes with phosphine derivatives showed a better biological response and higher selectivity indices than their respective non-phosphine gold complexes. A preliminary drug-receptor docking study suggests the interaction of thiosemicarbazone ligands with Y116 and E30, which are part of the catalytic N-terminal motif of thioredoxin reductase (TrxR). All compounds were evaluated for their ability to inhibit the activity of the TrxR enzyme. The non-substituted phosphine gold complexes turned out to be effective inhibitors of TrxR when compared to the phosphine analogues. Both complexes, in most of the cases, were more efficient TrxR inhibitors than the free ligands.

Full text of DOI:10.1016/j.poly.2017.05.004

Accepted Manuscript
Gold(I) complexes with aryl-thiosemicarbazones: molecular modeling, synthe-
sis, cytotoxicity andTrxR inhibition
Tatiane Teixeira Tavares, Gustavo Chevitarese Azevedo, Adriana Garcia,
Arthur G. Carpanez, Pâmela Matos Lewer, Diego Paschoal, Bruno L. Müller,
Hélio F. Dos Santos, Renato Camargo Matos, Heveline Silva, Richard Michael
Grazul, Ana Paula Soares Fontes
PII:
S0277-5387(17)30326-1
DOI:
http://dx.doi.org/10.1016/j.poly.2017.05.004
POLY 12617
Reference:
Polyhedron
To appear in:
Received Date:
Accepted Date:
9 March 2017
1 May 2017
Please cite this article as: T.T. Tavares, G.C. Azevedo, A. Garcia, A.G. Carpanez, P.M. Lewer, D. Paschoal, B.L.
Müller, H.F. Dos Santos, R.C. Matos, H. Silva, R.M. Grazul, A.P.S. Fontes, Gold(I) complexes with aryl-
thiosemicarbazones:molecular modeling, synthesis, cytotoxicity andTrxR inhibition, Polyhedron (2017), doi:http://
dx.doi.org/10.1016/j.poly.2017.05.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Gold(I) complexes with aryl-thiosemicarbazones: molecular modeling, synthesis,
cytotoxicity andTrxR inhibition
Tatiane Teixeira Tavaresa, Gustavo Chevitarese Azevedob, Adriana Garciab,
Arthur G. Carpanezb, Pâmela Matos Lewerb, Diego Paschoalc, Bruno L. Müllerb,
Hélio F. Dos Santosb, Renato Camargo Matosb, Heveline Silvad, Richard Michael
Grazulb, Ana Paula Soares Fontesb
ꢀ
1Departamento de Ciências Exatas e da Terra, Universidade do Estado de Minas
Gerais, 36500-000 Ubá, MG, Brazil
2Departamento de Química, Universidade Federal de Juiz de Fora, 36036-900 Juiz de
Fora, MG, Brazil
3NQTCM: Núcleo de Química Teórica e Computacional de Macaé, Polo Ajuda,
Universidade Federal do Rio de Janeiro, Campus UFRJ-Macaé, 27.971-525, Brazil
4Departamento de Química, Universidade Federal de Minas Gerais, 31270-901, Belo
Horizonte, MG, Brazil
______________________
ꢀCorresponding author:
ana.fontes@ufjf.edu.br
Fax 55.32.2102.3310
1

Abstract
Thiosemicarbazones and their metal complexes are of considerable interest because of
their chemical and biological properties. Gold(I) compounds form a family of potential
anticancer agents that are currently undergoing intense preclinical investigations. Novel
gold(I) complexes were obtained through the reaction of aryl-thiosemicarbazone ligands
with KAuCl4and Au(PPh3)Cl. The complexes were characterized by elemental analyses,
Raman, IR and NMR spectroscopies, electrospray ionization mass spectrometry,
TG/DTA, cyclicvoltammetry and with the aid of molecular modeling. The cytotoxicity
of the complexes was assessed against tumor cell lines (B16-F10 andCT26.WT) and
non-tumor cell line (BHK-21). Most of the tested complexes showed satisfactory
cytotoxic activity, some of them being even more cytotoxic in tumor cells than cisplatin
with a high selectivity index in the tested cells. The gold complexes with phosphine
derivatives showed a better biological response and higher selectivity indices than their
respective non-phosphine gold complexes. A preliminary drug-receptor docking study
suggests the interaction of thiosemicarbazone ligands with Y116 and E30, which are
part of the catalytic N-terminal motif of thioredoxin reductase (TrxR). All compounds
were evaluated for their ability to inhibit the activity of the TrxR enzyme. The non-
substituted phosphine gold complexes turned out to be effective inhibitors of TrxR
when compared to the phosphine analogues. Both complexes, in most of the cases, were
more efficient TrxR inhibitors than the free ligands.
Keywords Thiosemicarbazones; Au(I) Complexes; Cytotoxicity; Molecular Modeling;
Thioredoxin Reductase
2

Abbreviations
DTA Differential thermal analysis
DFT Density functional theory
DMSO Dimethysulfoxide
DMF Dimethylformamide
DTNB Dithiobisnitrobenzoic acid
MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide
NADPH Nicotinamide adenine dinucleotide phosphate
NMR Nuclear magnetic resonance
RPMI Roswell Park Memorial Institute Medium
TG Thermogravimetry
TMS Tetramethylsilane
TNB 5-thionitrobenzol
TrxR Thioredoxinreductase
3

Introduction
Metal complexes continue to find applications for the treatment of a wide variety
of human ailments [1]. The most prominent metal complex in medicine is the anticancer
platinum drug, cisplatin. This compound is one of the most widely employed agents in
the treatment of head, neck, lung, ovarian and testicular cancer [2, 3]. In the past few
decades, considerable efforts have been made in the search of novel metal-based
anticancer agents that might display innovative mechanisms of action and, thus offer
significant pharmacological advantages over the clinically established antitumor
platinum drugs (cisplatin, carboplatin, and oxaliplatin). Accordingly, several new
platinum and non-platinum compounds were prepared, characterized, and evaluated as
potential cytotoxic and antitumor drugs [4, 5].
The phenomenon of gold in medicine dates back to antiquity with both Arabic
and Chinese physicians using gold preparations to treat a variety of ailments. In more
−
recent times, after Koch demonstrated the bacteriostatic effects of Au(CN)2 , thereby
providing a scientific basis for the pharmacological activity of gold compounds, gold
thiolates have been used in the treatment of rheumatoid arthritis [6]. Subsequently, gold
compounds have been investigated for their antitumour [7-10] and anti-HIV [11]
activities. Auranofin, triethylphosphine-(2,3,4,6-tetra-O-acetyl-β-1-D-thiopyranosato-
S)-gold(I), is the first metal phosphine complex introduced into clinical practice for
chrysotherapy, the treatment of rheumatoid arthritis with gold-based drugs [12, 13],
following pioneering studies conducted with gold(I) thiolate compounds [14].
Gold complexes have also attracted significant attention as potential anticancer
agents since many gold(I) and gold(III) compounds have been found to inhibit tumor
cell growth [15].The mechanism of action is still not well understood. Mechanistic
4

studies suggested that, in contrast to cisplatin, DNA could not be the primary target and
that their cytotoxicity was mediated by their ability to alter mitochondrial function and
inhibit protein synthesis by interfering with DNA-protein interactions [16, 17]. This
could be partially explained by the higher affinity of gold(I) towards the so-called soft
ligands with sulfur (thiolates) and phosphorus (phosphines), and the lower affinity for
nitrogen and oxygen containing ligands [18, 19].
Disulfide reductase enzymes, such as thioredoxin reductase (TrxR) and
glutathione reductase (GR), are considered potential targets for anticancer gold
complexes, since these enzymes are over expressed in cancer cells and play a role on the
proliferation of tumor tissues [20]. These enzymes are involved in the cell redox
regulation and, therefore, their inhibition might trigger anti-mitochondrial effects
leading to cell apoptosis [21-23]. Besides, some others thiol-containing enzymes have
also been labeled as potential target for metal drugs, including cysteine proteases,
tyrosine phosphatases [20] and, more recently, the membrane channel protein aquaporin
[24].
Metal coordination to biologically active molecules can be used as a strategy to
enhance their activity and overcome resistance. For instance, metal complexes of
thiosemicarbazones can be more active than the free ligand, or they can be employed as
a vehicle for activation of the ligand as the cytotoxic agent [25, 26]. The coordination
chemistry of thiosemicarbazones has invited considerable interest in view of their
variable donor ability, structural diversity and biological applications [27, 28]. They can
bind metal ions as bidentate NS- or tridentate NNS-donor ligands forming five-
membered chelate rings. Transition metal complexes of thiosemicarbazones are of
considerable interest in chemistry because of their bioinorganic relevance.
Thiosemicarbazones are among the most potent inhibitors of ribonucleotidereductase
5

activity which catalyses the synthesis of DNA and is responsible for maintaining a
balanced supply of the deoxyribonucleotides required for synthesis and repair. Strong
positive correlation has been established between ribonucleotidereductase activity and
antitumor activity [29].
Thiosemicarbazones may incorporate aromatic moieties and have attracted a
great deal of research interest. They have been described to present nematocidal,
insecticidal, antibacterial, antifungal, antiviral, and anti-inflammatory, especially in the
case of furan and thiophene groups [30]. On the other hand, gold compounds containing
tertiary phosphines with a linear S-Au-P arrangement have been found to be more active
than similar compounds with no phosphinic substituents [31]. As such the study of
compounds containing gold as a metal ion and ligands derived from thiosemicarbazones
and phosphines remains a relevant area of research.
Both gold(I) and gold(III) thiosemicarbazone complexes have been reported to
present interesting biological activity [7, 32-36]. As part of our continuing
investigations on metal complexes as anticancer [37, 38], we report herein the synthesis,
characterization and cytotoxic activity of twelve novel gold(I) complexes containing
aryl-thiosemicarbazones as ligands carrying different aromatic moieties. The inhibition
of TrxR, a disulfide reductase acting as potential target for gold complexes, was also
tested with the drug-receptor binding modes discussed through molecular docking
analysis.
6

Experimental
Materials and methods
All reagents and solvents were used as obtained from commercial sources
(Sigma-Aldrich) without further purification. The thiosemicarbazone ligands were
prepared as previously reported in the literature [39]. Au(PPh3)Cl was synthesized from
K[AuCl4] according to the literature [40, 41].
The melting points were determined with a MQAPF-Microquímica hot stage
apparatus. Elemental analyses were performed at the University of São Paulo, Brazil. IR
spectra were obtained on a Bomem FT IR MB-102 spectrometer as KBr pellets.
Fourier-transform Raman spectroscopy was carried out using a Bruker RFS 100
instrument equipped with an Nd3+/YAG laser operating at 1064 nm in the near infrared
and a CCD detector cooled with liquid nitrogen with an average of 1000 scans and a
1
spectral resolution of 4 cm−1. H NMR (300 MHz) and 13C NMR (75 MHz) spectra
were recorded on a Bruker spectrometer and were obtained by dissolving the complexes
in DMSO-d6. The chemical shifts are expressed as δ (in ppm) from the internal
reference standard TMS. Thermogravimetry was carried out on a Shimadzu TG-60
with thermo balance in a flow of nitrogen gas and at a heating rate of 10 K min-1 in an
alumina crucible. Cyclic voltammograms were obtained at a µAutolab Type III
potentiostat (Eco Chemie, Netherlands) with GPES version 4.9 software. The
voltammetric measurements were made using a glassy carbon electrode as working
electrode, a platinum wire and a Ag/AgCl(s),KCl(sat) as auxiliary and reference
electrodes, respectively. Cyclic voltamograms were obtained between -0.60 and 1.00 V
at 50 mV s-1. Solutions of compounds (1 x 10-3mol L-1) were recorded with NaClO4 (0.1
7

mol L-1) (supporting electrolyte) dissolved in DMF. Mass spectra were obtained on an
AXIMA MALDI-TOF-TOF Shimadzu Biotech instrument. A nitrogen laser (λmax=337
nm) was employed for excitation in an α-cyano matrix. 200 scans were accumulated
with 20 repetitions each. UV/Vis spectra were recorded on a Shimadzu 1800 UV-Vis
spectrophotometer.
Synthesis of ligands a-f (Scheme 1)
The aryl-thiosemicarbazone ligands were prepared and characterized according
to the literature procedure [39, 42].
Synthesis of complexes 1-6 (Scheme 1)
The gold(I) complexes were prepared by mixing the desired ligand (1 mmol) in
methanol (5 mL) with methanolic solution (5 mL) of KAuCl4 (0.5 mmol). The resulting
solution was stirred for 2h at room temperature after which time a beige solid
precipitated out. The product was isolated by filtration, washed with ethanol and ether,
and dried. Yields: (1) 52%, (2) 56%, (3) 41%, (4) 34%, (5) 59%, (6) 62%.
1) IR νmaxKBr (cm-1): 3363, 3225, 3155, 2963, 1599, 1548, 1091, 811; Raman
1
(cm1): 1599, 1573, 1091, 581, 256. H NMR (300 MHz, DMSO-d6) δ: 12.30 (s,
1H, NH); 9.10 and 8.89 (s, 2H, NH2); 8.31 (s, 1H, HC=N), 7.81, 7.80 and 7.79
(t, 2H, CH, Ph), 7.42, 7.41 and 7.39 (t, 3H, CH, Ph); 13C NMR (75 MHz,
DMSO-d6) δ: 171.83 (C-S), 147.65 (HC=N), 146.10, 132.98, 131.44, 129.38,
127.32 and 126.55 (C, Ph); Anal. Calc. For [C16H18N6S2Au]Cl.HCl: C, 30.62;
8

H, 3.03; N, 13.39; Cl, 11.32. Found: C, 30.01; H, 3.02; N, 12.87, Cl, 10.82. MS
(MALDI): m/z calc. for [C16H18N6S2Au][M+H]+ calc. 555.4, found 555.2.
2) IR νmaxKBr (cm-1): 3408, 3271, 3143, 2955, 1597, 1511, 1026, 828; Raman
1
(cm1): 1599, 1569, 1104, 804, 436, 258. H NMR (300 MHz, DMSO-d6) δ:
12.33 (s, 1H, NH); 9.19 and 8.78 (s, 2H, NH2); 8.34 (s, 1H, HC=N), 7.84 and
13
7.80 (d, 2H, CH, Ph), 7.01 and 6.98 (d, 2H, CH, Ph), 3.80 (s, 3H, CH3); C
NMR (75 MHz, DMSO-d6) δ: 170.13 (C-S), 161.53 (C-OCH3), 148.13 (HC=N),
129.78, 125.79 and 114.41 (C, Ph), 67,01 (C, CH3); Anal. Calc. For
[C18H22N6O2S2Au]Cl.2HCl: C, 29.85; H, 3.32; N, 11.61; Cl, 14.70. Found: C,
30.72; H, 3.12; N, 10.94; Cl, 13.00. MS (MALDI): m/z calc. for
[C18H22N6O2S2Au][M+H]+ calc. 615.5, found 615.2.
3) IR νmaxKBr (cm-1): 3415, 3270, 3167, 2979, 1587, 1553, 1049, 819; Raman
1
(cm1): 1603, 1516, 1082, 857, 526, 262. H NMR (300 MHz, DMSO-d6) δ:
12.25 (s, 1H, NH); 9.07 and 8.38 (s, 2H, NH2); 8.49 (s, 1H, HC=N), 7.76 and
7.75, 7.57 and 7.56, 7.17, 7.16 and 7.14 (d, d and t, 3H, CH, thiophene); 13C
NMR (75 MHz, DMSO-d6) δ: 170.85 (C-S), 141.85 (HC=N), 136.98, 131.87,
129.92 and 127.48 (C, thiophene); Anal. Calc. For [C12H14N6S4Au]Cl.2HCl: C,
21.32; H, 2.36; N, 12.44; Cl, 15,77. Found: C, 21.13; H, 2.02; N, 12.27; Cl,
14.99. MS (MALDI): m/z calc. for [C12H14N6S4Au][M+H]+ calc. 567.5, found
567.1.
4) IR νmaxKBr (cm-1): 3398, 3151, 2989, 1615, 1546, 1016, 828; Raman (cm1):
1
1617, 1551, 1087, 827, 557, 266. H NMR (300 MHz, DMSO-d6) δ: 12.19 (s,
9

1H, NH); 9.03 and 8.40 (s, 2H, NH2); 8.20 (s, 1H, HC=N), 7.88, 7.09 and 6.67
(m, 3H, CH, furan); 13C NMR (75 MHz, DMSO-d6) δ: 171.91 (C-S), 136.45
(HC=N), 148.59, 145.74, 114.60 and 112.42 (C, furan); Anal. Calc. For
[C12H14N6O2S2Au]Cl.2HCl: C, 22.38; H, 2.49; N, 13.05; Cl, 16.55. Found: C,
23,12; H, 2.24; N, 13,51; Cl, 15,83. MS (MALDI): m/z calc. for
[C12H14N6O2S2Au][M+H]+ calc.535.4 found 535.1.
5) IR νmaxKBr (cm-1): 3416, 3074, 2977, 2766, 1608, 1535, 1081, 818; Raman
1
(cm1): 1638, 1590, 1087, 878, 549, 266. H NMR (300 MHz, DMSO-d6) δ:
12.39 (s, 1H, NH); 9.01and 8.80 (s, 2H, NH2); 8.07 (s, 1H, HC=N), 8.59 and
13
8.57, 8.15 and 8.13 (d, 4H, CH, pyridine); C NMR (75 MHz, DMSO-d6) δ:
174.63 (C-S), 148.56 (HC=N), 142.35, 140.49 and 123.73 (C, pyridine); Anal.
Calc. For [C14H16N8S2Au]Cl.2HCl: C, 25.24; H, 2.70; N, 16.83; Cl, 16,00.
Found: C, 25.55; H, 2.93; N, 16.02; Cl, 15.72. MS (MALDI): m/z calc. for
[C14H16N8S2Au][M+H]+ calc. 557.4, found 557.2.
6) IR νmaxKBr (cm-1): 3407, 3287, 3158, 3013, 1596, 1553, 1100, 819; Raman
1
(cm1): 1599, 1586, 1091, 822, 418, 284. H NMR (300 MHz, DMSO-d6) δ:
12.32 (s, 1H, NH); 10.21 (s, 1H, OH); 9.15 and 8.71 (s, 2H, NH2); 8.33 (s, 1H,
HC=N), 7.80 and 7.78, 6.87 and 6.86 (d, 4H, CH, Ph), 7.01 and 6.98 (d, 2H, CH,
13
Ph), 3.13 (s, 1H, OH); C NMR (75 MHz, DMSO-d6) δ: 170.52 (C-S), 160.12
(C-OH), 147.88 (HC=N), 129.78, 124.22 and 115.63 (C, Ph); Anal. Calc. For
[C16H18N6O2S2Au]Cl.HCl: C, 29.14; H, 2.88; N, 12.75, Cl: 10.77. Found: C,
29.99; H, 3.02; N, 12.91, Cl, 12.01. MS (MALDI): m/z calc. for
[C16H18N6O2S2Au][M+H]+ calc. 587.5, found 587.1.
10

Synthesis of complexes 7-12 (Scheme 1)
To a solution of Au(PPh3)Cl (1mmol) in dichloromethane (5 mL), the respective
ligand (1mmol) dissolved in acetone (5 mL) was slowly added during 1h. After stirring
for 6h at room temperature in the dark, the solvent was removed under reduced pressure
to furnish a white residue which was purified by preparative chromatography (eluent:
dichloromethane) to give the desired compounds 7-12. Yields: (7) 35%, (8) 38%, (9)
44%, (10) 35%, (11) 30%, (12) 46%.
7) IR νmaxKBr (cm-1): 3423, 3261, 3142, 2962, 1594, 1519, 1092, 870, 536, 425;
1
Raman (cm1): 1598, 1567, 1175, 868, 617, 542, 250; H NMR (300 MHz,
DMSO-d6) δ: 11.50 (s, 1H, NH); 8.26 (s, 1H, HC=N), 8.06 (s, 2H, NH2); 7.80,
7.79 (d, 2H, CH, Ph, J = 3.6 Hz) and 7.39 (m, 3H, CH, Ph); 7.57 (m, 15H,
PPh3); 13C NMR (75 MHz, DMSO-d6) δ: 177.46 (C-S), 142.39 (HC=N), 132.23,
129.80, 129.51, 128.59 and 127.27 (C, Ph); 134.10 (3C, CH, PPh3), 133.76 (d,
1C, CH, PPh3, J = 18.8 Hz), 129.64 (d, 2C, CH, PPh3, J = 11.7 Hz). Anal. Calc.
For [C26H24N3SPAu]Cl: C, 46.33%; H, 3.56%; N, 6.24%. Found: C, 46.76%; H,
3.75%; N, 6.49%. MS (MALDI): m/z calc. for [C26H24N3SPAu][M+H]+ calc.
638.5, found 638.3.
8) IR νmaxKBr (cm-1): 3407, 3288, 3150, 2971, 1604, 1545, 1168, 1074, 835, 519,
1
442; Raman (cm1): 1601, 1570, 1170, 877, 527, 426, 257; H NMR (300 MHz,
DMSO-d6) δ: 11.47 (s, 1H, NH); 8.27 (s, 1H, HC=N), 8.05 (s, 2H, NH2); 7.76,
7.74 (d, 2H, CH, Ph, J = 8.4 Hz); 7.58 (m, 15H, CH, PPh); 6.98 and 6.96 (d, 2H,
Ph3, J = 8.7 Hz); 3.79 (s, 3H, H3CO); 13C NMR (75 MHz, DMSO-d6) δ: 176.71
11

(C-S), 160.43 (H3COC); 142.44 (HC=N), 131.98, 128.70, 127.73, 126,48 and
113.89 (5C, CH, Ph); 133.18 (3C, CH, PPh3), 133.54 (d, 1C, CH, PPh3), 129.30
(d, 2C, CH, PPh3); 55.03 (H3CO). Anal. Calc. For [C27H26N3OSPAu]Cl: C,
46.06%; H, 3.70%; N, 5.97%. Found: C, 46.53%; H, 3.55%; N, 5.89%. MS
(MALDI): m/z calc. for [C27H26N3OSPAu][M+H]+ calc. 668.5, found 668.6.
9) IR νmaxKBr (cm-1): 3415, 3270, 3150, 2903, 1587, 1545, 1100, 827, 528, 494;
1
Raman (cm1): 1593, 1509, 1096, 829, 656, 528, 428, 262; H NMR (300 MHz,
DMSO-d6) δ: 11.60 (s, 1H, NH); 8.34 (s, 2H, NH2), 7.67 (m, 15H, CH, PPh),
13
7.53 (s, 1H, HC=N); 7.20, 7.14 and 7.00 (s, 3H, CH, thiophene); C NMR (75
MHz, DMSO-d6) δ: 176.38 (C-S), 138.09 (HC=N), 137.60, 131.48, 128.30 and
127.35 (3C, CH, thiophene); 133.20 (d, 1C, CH, PPh3, J = 10.87), 129.98 (3C,
CH, PPh3), 129.01 (d, 2C, CH, PPh3, J = 10.95); Anal. Calc. For
[C24H22N3S2PAu]Cl: C, 42.38%; H, 3.24%; N, 6.18%. Found: C, 43.18%; H,
3.47%; N, 6.50%. MS (MALDI): m/z calc. for [C24H22N3S2PAu][M+H]+ calc.
644.5, found 644.5.
10) IR νmaxKBr (cm-1): 3449, 2920, 2849, 1608, 1528, 1061, 835, 541, 489; Raman
(cm1): 1628, 1555, 1099, 887, 671, 440, 265; 1H NMR (300 MHz, DMSO-d6) δ:
11.49 (s, 1H, NH); 8.26 and 7.08 (s, 2H, NH2), 7.98 (s, 1H, HC=N); 7.57 (m,
15H, CH, PPh), 7.81; 6.97 and 6.62 (s, 3H, CH, furan); 13C NMR (75 MHz,
DMSO-d6) δ: 176.94 (C-S), 149.11 (HC=N),144.64, 132.58, 112.52 and 111.99
(3C, CH, furan); 133.45 (d, 1C, CH, PPh3, J = 13.87), 131.73 (3C, CH, PPh3),
129.27 (d, 2C, CH, PPh3, J = 11.47); Anal. Calc. For [C24H22N3OSPAu]Cl: C,
12

43.41%; H, 3.32%; N, 6.33%. Found: C, 43.02%; H, 3.44%; N, 6.15%. MS
(MALDI): m/z calc. for [C24H22N3OSPAu][M+H]+ calc. 628.5, found 628.4.
11) IR νmaxKBr (cm-1): 3427, 3270, 3157, 2917, 1603, 1550, 1107, 882, 544, 499;
1
Raman (cm1): 1597, 1582, 1102, 887, 689, 328, 254; H NMR (300 MHz,
DMSO-d6) δ: 11.77 (s, 1H, NH); 8.37 and 8.21 (s, 2H, NH2), 8.02 (s, 1H,
HC=N), 8.58 and 7.77 (s, 4H, CH, pyridine), 7.52 (m, 15H, CH, PPh); 13C NMR
(75 MHz, DMSO-d6) δ: 177.76 (C-S), 149.49 (HC=N), 141.07, 139.35, 128.29,
127.47 and 120.46 (5C, CH, pyridine); 133.14 (d, 1C, CH, PPh3, J = 13.50),
131.63 (3C, CH, PPh3), 129.98 (d, 2C, CH, PPh3, J = 10.57); Anal. Calc. For
[C25H23N4SPAu]Cl: C, 44.48%; H, 3.41%; N, 8.30%. Found: C, 44.62%; H,
3.26%; N, 7.98%. MS (MALDI): m/z calc. for [C25H23N4SPAu][M+H]+ calc.
639.5, found 639.4.
12) IR νmaxKBr (cm-1): 3474, 3370, 3129, 1598, 1511, 1100, 835, 537, 499; Raman
(cm1): 1605, 1586, 1164, 877, 646, 413, 257; 1H NMR (300 MHz, DMSO-d6) δ:
11.53 (s, 1H, NH); 10.13 (s, 2H, NH2), 8.33 (s, 1H, HC=N), 8.11, 8.07 (d, 2H,
CH, Phenol, J = 12.9 Hz), 6.93, 6.90 (d, 2H, CH, Phenol, J = 7.8 Hz); 7.71 (m,
15H, CH, PPh); 13C NMR (75 MHz, DMSO-d6) δ: 176.38 (C-S), 159.18 (HOC);
143.29 (HC=N), 133.50 (d, 1C, CH, PPh3, J = 13.05), 129.36 (d, 2C, CH, PPh3,
J = 13.05), 128.79 (3C, CH, PPh3), 132.00, 124.67 and 115.28 (5C, CH,
Phenol). Anal. Calc. For [C26H24N3OSPAu]Cl: C, 45.25%; H, 3.48%; N, 6.09%.
Found: C, 44.89%; H, 3.33%; N, 6.19%. MS (MALDI): m/z calc. for
[C26H24N3OSPAu][M+H]+ calc. 654.5, found 654.5.
13

Insert Scheme 1
Cytotoxicity assay
Cell lines,B16-F10 - mouse metastatic skin melanoma, and CT26.WT – colon
cancer cells, and the non-tumor cell lines BHK-21 - Baby Hamster Kidney, were
maintained at 37ºC in a 5% CO2 atmosphere in RPMI supplemented with 10% fetal
bovine serum (FBS). Cells were seeded in 96-well tissue culture plates (100 µL/well) at
0.5×103, 1×103 and 2×103 viable cells/well (respectively to BHK-21, B16-F10 and
CT26.WT), after 24h, compounds were added in serially diluted culture medium (< 1%
DMSO) and incubated for 72h at 37°C and 5% CO2. Finally, MTT was aseptically
added (0.01 M in water solution - 10 µL/well) for 4h at 37°C and 5% CO2 was
quantified through colorimetric assay (570 nm) to measure MTT metabolized by viable
cells resulting in a violet product previous solubilized in 100 µL of DMSO. The raw
data were normalized to the untreated control cells. IC50 values were calculated by four
parametric nonlinear regression using GraphPad Prism 5.0 software and compared to
cisplatin.
Inhibition of thioredoxinreductase(TrxR) activity
Rat liver TrxR (Sigma) was used to determine TrxR inhibition by the
compounds. The assay was performed according to the manufacturer’s instructions
(Sigma product information sheet T9698) with appropriate modifications. Initially, 20
µL of the TrxR rat liver solution (containing approximately 0.10 units of the enzyme)
were added to a solution of the complex in 5% DMSO/potassium phosphate buffer pH
14

7.0 (20 µL) and the resulting solution was incubated during 1h at 37°C. For all
compounds the final concentration was 5 µM. After incubation, 600 µL of the reaction
mixture were added. The reaction mixture consisted of 200 µL of 500 mM EDTA
solution pH 7.5, 800 µL of 63 mM dithiobisnitrobenzoic acid (DTNB) in ethanol, 50 µL
of 48 mM nicotinamide adenine dinucleotide phosphate (NADPH), 100 µL of 20
mg/mL bovine serum albumin, 1 mL of 1.0 M potassium phosphate buffer pH 7.0 and
7.85 mL of water. The blank solution consisted of a solution of the complex in 5%
DMSO/potassium phosphate buffer pH 7.0 (20 µL) added to water (20 µL) and the
reaction mixture (600 µL). The formation of 5-thionitrobenzol (TNB) was monitored in
UV-Visible Spectrophotometer (UV-1601PC SHIMADZU) at 412 nm for 4 minutes.
The enzymatic activity of the samples was calculated in percentage using the
absorbance of the positive control (20 µL of the TrxR rat liver solution and 20 µL of
potassium phosphate buffer pH 7.0 added 600 µL of reaction mixture) as maximum
activity (100% of activity). Measurements were performed in triplicate.
DFT calculations
The geometries of the free ligands (a-f) and gold complexes (1-12) were
optimized in gas phase at PBEh1PBE/6-31+G(2d)/SDD(f) level of theory. The 6-
31+G(2d) basis set was applied for all atoms except gold, which was treated with the
Stuttgart/Dresden ECP for 60 inner electrons and the D95 basis set for 19 valence
electrons [43]. The valence shell was augmented with a set of “f” polarization functions
(αf=1.1386328) proposed previously [44] for predicting structure and reactivity of gold
complexes. Similar level of theory was used previously for gold complexes providing
satisfactory results for their structures and molecular properties [7, 19, 45-46]. In a
15

recent paper concerning gold(I) complexes [45] we showed that the mean error found at
DFT level for bond lengths and angles was lower than 4%, which is considered
excellent if we bear in mind that the experimental data came from solid state samples.
Therefore, even though crystal structures were not obtained for the complexes studied
herein, the DFT structures can be taken as a good approximation. The NMR spectra
were further predicted in DMSO using the PBEh1PEB/6-31+G(2d,p)/SDD(f) level of
theory, which differ from that used for geometry by only the set of “p” functions added
to the hydrogen atoms. The solvent effect was accounted for by the PCM approach
using the IEFPCM formalism [47]. Chemical shifts were obtained as the difference
between shielding constants calculated through GIAO method [48] and using the TMS
as internal standard. The calculations were carried out using the Gaussian 09 Rev. A.02
[49].
Drug-receptor docking
The gold complexes synthetized here were experimentally tested as inhibitors of
the TrxR enzyme. This enzyme has been considered a potential target for gold
complexes due to the presence of cysteine and selenocysteine residues in the active site,
which are good ligands for soft metals such as gold(I) [50]. Aiming to get some insights
on the drug-receptor binding mode, molecular docking studies were accomplished. The
TrxR structure was obtained from Protein Data Bank under code 2J3N and prepared
according to the Protein Preparation Wizard Tool (standard protocol) implemented into
GOLD-5.0 software [51-53] using receptor-rigid method. The searching for the best
docked poses, was carried out using 10 Å of radius centered at C498 residue,
sufficiently to cover all binding cavity. One-hundred genetic algorithm (GA) runs were
16

carried out with high efficiency (set to 200%), generating 100 poses for each compound
for further analysis. The ChemScore scoring function (SF) was used for scoring and
then, ChemPLP, GoldScore and ASP scoring functions for re-scoring. All these are
empirical scoring functions, which the final score is estimated from an empirical
regression function with coefficients fitted to reproduce the binding energy (or some
other related property such as the ligand binding position) for a set of testing drug-
receptor complexes [53]. The scoring terms are often based on energy, but their meaning
is not “pure” as in the force-field-based scoring functions. Therefore, the fitness score,
taken as the negative of the sum of the energy terms, is the property of interest, i.e., the
larger the score the better is the drug-receptor pose [54].The disadvantages of empirical
SF is that they are constructed using different numbers and types of energy terms, which
make them quite sensitive to the system investigated. Moreover, binding energy and free
energy is not always accessible for mostly of empirical SF. One way to improve the
scoring and the prediction of bound poses is to combine the scores from several
empirical SF in a single consensus scoring. In the present study, the Z-score function
(Eq. 1) was used for generating a consensus ranking (Z) [55]. The quantities ꢀ and σ in
Eq. (1) are the average score and standard deviation over 100 poses, respectively (Table
S1). The summation in Eq. (1) is over the four scoring functions used.
ꢂ = ꢅꢇꢈꢉ ꢄꢅꢆ ∴ꢄꢅꢆ = ꢊ ꢍꢊ ; ꢎ = 0,1, … 100
(1)
ꢋꢌ
ꢃ
σ
For the gold complexes (6–12), the same docking protocol applied for the free
ligands was used. However, as the gold atom is not formally accounted for in the
searching algorithms and scoring functions, the distance between the sulphur atom of
17

Cys-498 and gold atom was constrained within the 1.5–3.5 Å bond range. This was
done by setting the spring constant k=5(in arbitrary units). In the GOLD® program [51-
53], if the constrained distance is found outside the specified bounds, spring energy is
used to reduce the fitness score. Finally, intermolecular interactions of the best poses
were analyzed using Discovery Studio 4.0 software [56]. Similar protocol was
successfully used in Ref. [7] for gold(I) complexes. It is worth mentioning that most of
gold complexes act as TrxR inhibitors by coordinating to Cys and Sec residues in the
active site [57, 58]. The binding mode is often irreversible, with the final structure
containing the gold atom coordinated only to the enzyme residues [58]. Thus, the gold
complex is degraded and, in principle, the type of ligand would not play a role on the
overall biological response. Nonetheless, the ligand exchange rate and thermodynamics
might contribute to the process. Moreover, prior the gold-enzyme binding, the entire
complex has to anchor on the active site in a suitable conformation, which is certainly,
driven by the ligand structure [7].
Results and Discussion
Reactions of thiosemicarbazone and KAuCl4 in MeOH at room temperature
produced precipitates of the general composition [Au(L)2]+ (complexes 1-6). Reduction
of gold(III) to gold(I) took place in all cases, as previously observed for other gold
complexes with thiosemicarbazones reported in the literature [8, 59, 60].
Complexes 7-12 were synthesized by adding the corresponding ligands,
previously dissolved in acetone, to a solution of Au(PPh3)Cl in dichloromethane at
room temperature, and were isolated by preparative plate chromatography.
18

Based on the experimental characterization data and considering that gold(I)
complexes are generally linear and that the metal ion is a soft acid, it was proposed that
the coordination sphere of the metal center is filled only by the sulfur atoms from the
two thiosemicarbazone ligands for complexes 1-6 or by a sulfur atom from the
thiosemicarbazone and the phosphorus atom from triphenylphosphine for complexes 7-
12. Complexes of analogous structures were reported in the literature [8, 11, 61, 62].
The DFT optimized structures are represented in Figure S1 (Supporting
Information) for complex 1(S1a-b) and 7 (S1c) and the main structural parameters for
all complexes are shown in Table S2. For complexes 1-6, two conformations were
proposed, differing by rotation around C-S bonds. The folded form (S1a) is less stable
than the extended one (S1b) by ~3 kcal mol-1 in gas phase and by ~5 kcal mol-1 in
DMSO, regardless of the complex. Therefore, our results suggest the extended form of
complexes 1-6 (see Figure S1b for complex 1) as predominant in the condensed phase.
The extended form has already been reported previously for an analogous gold(I)
thiosemicarbazone complex [63].
For complexes 7-12, the general structure is as represented in Figure S1c for
complex 7. The values in Table S2 show that the main difference in the structure of the
thiosemicarbazone moiety is in the Au-S bond distance, which is longer for complexes
7-12 due to the strong trans effect of the phosphine ligand. For complexes 1-6, the Au-S
bond was ~2.33 Å and for complexes 7-12 it was around 2.37 Å. The experimental Au-
S bond length for the analogue 1.4MeOH described by Castiñeiras [63] which has the
same coordination sphere as complexes 1-6, was 2.297 Å, in satisfactory accordance
with our calculated values.
The ligands are arranged linearly in the gold coordination sphere as expected for
gold(I) complexes, with the S(P)-Au-S angle in the range of 176-177°. Interestingly, for
19

complexes 1-6 the planes of the two ligands are perpendicular to each other, with the
H2N-S-S-NH2 dihedral angle varying from 99 to 109°. A distinct form was reported in
the solid state for the analogue 1.4MeOH where both thiourea groups are found in the
same plane, with the H2N-S-S-NH2 dihedral angle equal 180°[63]. The parallel form for
complex 1 was calculated as a transition state and found only 0.4 kcal mol-1 higher in
energy than the perpendicular structure in DMSO solution. This energy barrier is lower
than the thermal energy at room temperature (RT=0.592 kcal mol-1) and, therefore, both
structures are likely to coexist in solution. It is also worth noting that these structures
show the unconventional Au…H-N hydrogen bonds with the average Au…H, Au…N
and Au…H-N parameters found 2.65 Å, 3.30 Å and 122° for complexes 1-6. These
values are in satisfactory accordance with experimental data for the 1.4MeOH [63]:
2.89 Å, 3.42 Å and 120.4°, respectively. In a recent review by Schmidbaur et al. [64],
66 hits containing Au…H-N contacts from CSD search were analyzed with most of
molecules showing structural parameters within the ranges 2.5-3.0 Å (Au…H-N), 3.0-
3.6 Å (Au…N) and 100-130° (Au…H-N).
Spectroscopic characterization
The IR spectra of the complexes are characterized by strong broad absorptions in
the 3474−3129 cm−1 range assigned to ν(N-H). The absorption attributed to ν(C=N)
around 1600 cm-1 in the ligand spectra is not modified in the spectra of the complexes,
indicating that the imine nitrogen atoms are not attached to the gold(I) center. This is
supported by the calculated IR spectra. To illustrate, for the ligand (a) and complex (1),
the ν(C=N) mode is predicted at 1715 and 1718 cm-1, respectively. This same finding is
20

predicted for the others ligands and complexes. Based on the structural data (Table S2),
the C=N bond length is ~1.28 Å for the free ligands and all the analyzed complexes.
The spectra of all the ligands showed medium intensity bands in the 1090-835 cm−1
range attributed to ν(C=S) vibrations, which were shifted to lower frequencies on the
coordination of the thione sulfur to gold(I) ion [65]. This is also supported on the basis
of geometrical analysis. It can be observed that the C-S bond increases from 1.65 to
1.70 Å upon gold coordination (Table S2). These values are consistent with
experimental data for the gold(I) complex with the thiosemicarbazone derivative
reported in reference [63] where the C-S bond length increases from 1.69 to 1.72 Å. The
ν(C=S) mode is calculated at 788 cm-1 for ligand (a) and 745 cm-1 for complex (1).
Despite the predicted shift of 43 cm-1, this vibration is of low intensity and, therefore,
difficult to assign experimentally. The ν(Au-S) vibration around 260 cm-1 was observed
in the Raman spectra of all the complexes, which is expected based on data available in
the literature [32, 61, 66]. Theoretically, for complexes (1-6) the symmetric and
asymmetric Au-S vibrations are calculated over the ranges 305-320 cm-1 and 314-346
cm-1, respectively. For complexes (7-12), ν(Au-S) is calculated at 302-334 cm-1 and the
ν(Au-P) around 543 cm-1. As noted, calculated vibrational frequencies are
systematically overestimated due mainly to the use of the harmonic oscillator approach.
1
The broad signal around 11.5 ppm in the H NMR spectra of the ligands
persisted in the spectra of the complexes, indicating the protonated nature of the =N–
NH group [61]. The signals of the HC=N, which appeared as singlets around δ 8.1 in
the spectra of the ligands, did not suffer relevant changes after coordination to gold,
consistent with the non-coordination of the imine nitrogen to the metal center. The
signals of the aromatic protons of the complexes and free ligands appeared in the δ 6.6-
21

8.8 region.The signal observed around δ 7.6 in the spectra of complexes 7-12 can be
attributed to the aromatic hydrogens of the PPh3 group [67]. The NH2 signal in the
spectra of complexes 1-6 was observed as a doublet around δ 8.3-9.1 [32] and for
complexes 7-12, it appeared around δ 8.5. The calculated 1H NMR chemical shifts show
two signals for NH2 hydrogens around 5.6 and 6.6 ppm for all complexes. The predicted
chemical shift is underestimated relative to the experimental due to NH2…DMSO
hydrogen bonds, which are not considered in our implicit solvent model. The downfield
value is assigned to the hydrogen pointed towards the gold center. For the ligands only
one signal is expected ~5.0 ppm, therefore, the non-equivalence of NH2 hydrogens in
the complexes is not only due to the hindered rotation around the C-N bond, but also
due to the Au…H-N hydrogen bond discussed previously. Indeed, the C-NH2 bond
length decreases upon complex formation, from 1.38 to 1.34 Å, and consequently, the
bond-order slightly increases from 1.2 to 1.3.
In the 13CNMR spectra of the complexes, the signals of the HC=N group appear
in the range of δ149-133. The C=S signal, observed around δ 178 in the spectra of the
free ligands, showed a downfield shift for the complexes supporting the coordination of
the thione sulfur to the metal center [68].The aromatic carbon signals were observed in
theδ161-115 region. The signals corresponding to the aromatic carbons of the PPh3
group (complexes 7-12) were observed in the δ 133-128 region. Theoretical analysis
predicts the HC=N signal in the range of 142-131 ppm for free ligands, with the lowest
values calculated for the ligands c and d at 137 and 131 ppm, respectively. For the
complexes, a downfield shift is calculated around 149-135 ppm, with lower values
found for complexes with ligands c and d, at 142 and 135 ppm, respectively. The C=S
chemical shift was calculated as ~178 and 171 ppm for the free ligands and complexes,
respectively, with the ligand playing a minor role.
22

Thermal analysis (TG and DTA)
TG and DTA curves for the complexes were consistent to the proposed
structures. For instance, complex 1was stable until about 145 °C followed by the first
weight loss in the 145 °C to 174 °C range which was attributed to the loss of one mol of
HCl (calculated: 5.8%; experimental: 6.1%). Subsequently, two successive weight
losses were observed related to decomposition of the organic ligand, followed by an
exothermic event at 210 °C. The final residue was consistent with one mol of metallic
gold (calculated: 31.4%; experimental: 32.1%).
Voltammetric characterization
The voltammetric characterization was performed to assure that gold was
reduced during the preparation of the complexes 1-6, and to confirm that gold(I)
maintained its oxidation state in complexes 7-12.
The cyclic voltammograms of complexes 1-6, ligands and KAuCl4 were
obtained between -0.60 and 1.00 V starting from -0.6 V, at 50 mV s-1. Figure 1A shows
the electrochemical behaviour of complexes 1, 2 and 3 compared to that of KAuCl4. In
the cyclic voltammogram of complex 2, a quasi-reversible process was observed, with
→
an anodic peak at 0.60 V vs Ag/AgCl(s),KCl(sat) which may be attributed to gold(I)
gold(II) oxidation and a cathodic peak centered at -0.22 V vs Ag/AgCl(s),KCl(sat) related
→ gold(I) reduction
to gold(II)
[69], which is absent in cyclic voltammograms of the
→ go
ligands. For KAuCl4, only the gold(II)
ld(I) process was observed. Similar
response was observed for all other complexes (Table 1). Thus the results suggest that
23

the synthesized complexes have Au(I) as central ion, in agreement with other results
presented in this work.
Insert Figure 1
Insert Table 1
The cyclic voltammograms of complexes 7-12, ligands and triphenylphosphine
were obtained between 1.70 and -1.50 V starting from 1.70 V at 200 mV s-1. The results
were compared with that of KAuCl4, to confirm the presence of gold(I). There were no
similarities between the voltammograms of complexes and KAuCl4. Cyclic
voltammograms of the complexes exhibit one irreversible wave in the cathodic range
→ gold(0) and three one
-electron diffusion
corresponding to the reduction of gold(I)
controlled irreversible waves at the anodic range.
Voltammetry curves of complexes 7, 8 and 9 are presented in Figure 1B. The
values of the anodic and cathodic peak potentials are listed in Table 1, suggesting the
presence of gold(I) in complexes 7-12.
Cytotoxic activity
The cytotoxicity of the ligands and complexes was evaluated in comparison to
cisplatin in two tumor cell lines: mouse metastatic skin melanoma (B16-F10) and colon
cancer cells (CT26.WT). The complexes were also examined for their cytotoxic
properties in one non-tumor cell line, Baby Hamster Kidney (BHK-21) [70]. The IC50
values, calculated from the dose survival curves obtained after 72h of drug exposure
(MTT test), are shown in Table 2. These cells were chosen to compose a panel of
different tumor types such as carcinoma, adenocarcinoma and a non-tumor cell to assess
24

the activity in different embryonic origins such as epithelial or fibroblasts to overcome
and to try to control any other differences.
Insert Table 2
In the B16-F10 cell line, no activity was observed for the ligands while
complexes 1, 2, and 4 showed some activity. These data suggest that coordination to
gold enhanced the cytotoxicity of the thiosemicarbazones. However, this statement
could not be further extended since the ligands were interestingly more cytotoxic and
selective than the respective gold complexes in the CT26.WT cell line.
The phosphine gold complexes (complexes 7-12) were much more cytotoxic and
more selective than the respective non-phosphine gold complexes 1-6 in the B16-F10
cell line being in the same range of cisplatin activity. In the cell CT26.WT, gold
complexes also showed significant enhancement in activity for the phosphine
complexes when compared to their non-phosphine analogues. All of them showed
activity at concentrations below 2 µM. The lipophilicity of the compounds, introduced
by the presence of the phenylphosphine group, could be responsible for the increased
cytotoxicity since this property could facilitate cell membrane penetration. These results
did not provide any noticeable differences when comparing the different substituted
thiosemicarbazones investigated.
TrxR Inhibition
All compounds have shown to be able to inhibit the TrxR enzyme at least 25%.
We can highlight ligands (e) and (f). The latter can inhibit the enzymatic activity by
25

99% under the experimental conditions. The results obtained from the TrxR inhibition
assay are in contrast to the cytotoxicity results. Gold complexes 1-6 that inhibited TrxR
almost completely (up to 98%) did not show considerable cytotoxicity in the tested cell
lines (Table 2). Phosphine gold complexes showed the opposite response. The results
showed values between 56% and 78% for enzymatic inhibition for such complexes (7-
12). The difference may be explained by transport into the cell. The TrxR assay is
performed with isolated enzyme and the assay conditions exclude the main pathway for
drugs, namely cellular uptake. The high lipophilicity and high cellular penetration
promoted by phosphine groups is well known [71]. The phosphine gold complexes (7-
12) exhibited similar or better enzymatic inhibition when compared to auranofin (60%
under the same experimental conditions) which resulted in improved cytotoxicity while
combining good cellular uptake and TrxR inhibition.
Drug-receptor docking analysis
The drug-receptor binding modes for the free ligands are illustrated in Figure 2,
where the poses were those selected from consensus (highest Z-score). It is noticeable
that all derivatives bind in a similar mode with TrxR enzyme at the principal catalytic
site. Two main intermolecular interactions were common for all ligands, through π-
stacking with Y116 and H-bonding with E30, both residues found at the N-terminal
motif of the pattern chain. For the ligand (f), which showed the best TrxR inhibitory
activity among the tested ligands (99%, Table 2), three anchor points are highlighted in
Figure 2b, including an electrostatic interaction (weak H-bond) between the hydroxyl
group at p-position of (f) with S22 (rHO=2.82Å), besides the hydrophobic interaction
with Y116 (rcentroid=5.87 Å) and H-bond with E30 (rHO=1.93Å) residues. The drug-
26

receptor binding free energy (∆Gb) cannot be properly estimated within the consensus
approach used in this work, once only the ChemScore SF calculates this property
explicitly. However, a qualitative picture of the terms contributing to the binding energy
can be drawn by using the ChemScore analysis for the consensus pose represented in
Figure 2b, even though, this is not necessarily the highest score pose found by
ChemScore SF. For ligand (f), ∆Gb=17.0 kJ mol-1 which has 19% due H-bond
contribution (-3.27 kJ mol-1) and 64% due lipophilic contribution (-10.8 kJ mol-1).
Interestingly, the Y116 residue has been considered a key residue in the catalytic
mechanism of TrxR1 [72] by weakening the Se-S bond in the oxidized form of C-
terminal motif and, thereby, mediating the reduction by the N-terminal C59 and C64
residues. Therefore, the ligands prepared here may act as a TrxR inhibitor by blocking
the participation of Y116 in the enzyme catalytic mechanism.
Similar conclusions were drawn for the gold complexes as represented in Figure
3 for the compounds 6 and 12, which contain the strongest TrxR inhibitor ligand,
namely ligand (f). Complex 6 was the most potent inhibitor (98%) among the series 1-6
and complex 12 the second most potent inhibitor (68%) containing phosphine as a
ligand (see Table 2). Moreover, it should bear in mind that complex 12 is much less
active as a TrxR inhibitor than 6.
Insert Figure 2
Insert Figure 3
Figure 3a shows the three anchor points (H-bonds) for complex 6, involving the
residues C497 (C-terminal, subunit C, rNO=2.52 Å), E30 (rOO=2.82 Å) and Y116 (N-
terminal, subunit D, rNO=2.84 Å). The latter two were also found for the free ligand (f),
27

but in a distinct binding mode, as shown in Figure 2. For complex 6, the
thiosemicarbazone ligand interacts with Y116 and E30 through H-bond, which is
stronger than the π-stacking predicted for the free ligand (f) (Figure 2b). In addition, a
strong drug-receptor H-bond is also observed for complex 6 with C497 residue. As a
result, the H-bond contribution to the binding free energy was larger than that found for
free ligand (f), account for -9.31 kJ mol-1 (34%). In spite of the increasing in the
electrostatic nature of drug-receptor interaction for complex 6, due H-bonds, the
lipophilic term still dominates the overall binding energy, contribution with 62% or
-16.83 kJ mol-1. It is worth emphasizing the common anchor points Y116 and E30 for
complex 6 and free ligand (f), and the role of the former for enzymatic catalysis [72].
Concerning the binding mode of complex 12, only two anchor points (H-bond) are
observed with residues C497 (rNO=3.17 Å) and E477 (rOO=2.88 Å). Both H-bonds are
longer than those found for complex 6, leading to H-bond contribution for binding free
energy of only -6.71 kJ mol-1 (18%). Complex 12 is the most lipophilic molecule among
those studied here, which is also one of the most cytotoxic according Table 2, and,
therefore, showed the highest lipophilic contribution to drug-receptor binding energy,
equal to -28.8 kJ mol-1 (74%). Moreover, the interactions with E30 and mainly Y116 are
missing, which lead to a lower Z-score for complex 12 (6.7) compared to 6 (8.5) (Table
S1). This finding is in agreement with experimental TrxR inhibition potency: 68% (12)
and 98% (6). Finally, based on experimental data for TrxR inhibition and calculated
contributions to the binding free energy, we may suggest that the H-bonds between the
ligand and TrxR should drive the binding mode and drug-receptor stability.
28

Concluding Remarks
This work described the synthesis and characterization of novel
thiosemicarbazone gold(I) complexes. The structural characterization of the compounds
was established on the basis of elemental analyses, Raman, IR, NMR spectroscopies,
mass spectrometry, TG/DTA, cyclic voltammetry, and DFT calculations. We have
investigated the cytotoxicity of the free ligands and complexes. It was also observed
that the free ligands already have important cytotoxicity and selectivity in colon cancer
cells and impressive TrxR inhibition capacity, especially ligand (f). Phosphine
complexes were more cytotoxic than the non-phosphine complexes even though their
ability to inhibit TrxR enzyme is lesser.
Drug-receptor docking analysis provided some molecular insights in the binding
mode of the synthesized compounds and the TrxR enzyme, which is considered a
potential non-DNA target for gold-complexes in cancer chemotherapy. Both ligands and
gold complexes have similar binding modes with Y116 and E30 residues as the main
anchor points. The Y116 is implicated in participating directly in the enzyme catalysis,
therefore, blocking this amino acid might be a way to inhibit the enzyme activity.
Interestingly, the desirable interaction of the drug with Y116 in the catalytic site is
somehow driven by the interaction E30 residue; thus, those ligands whose size can fit in
between the Y116 and E30 distance (~14 Å) containing donor or acceptor H-bonds at
the edges are considered promising TrxR inhibitors.
Among all tested compounds in the present study, ligand (f) and complex 6 are
the most potent TrxR inhibitors, with complex 12 showing good activity with the series
containing -PPh3 as ligand.
29