Highly cytotoxic trithiophenolatodiruthenium complexes of the type [(η6-p-MeC6H4Pr i ) 2Ru2(SC6H4-p-X)3] +: Synthesis, molecular structure, electrochemistry, cytotoxicity, and glutathione oxidation potential

Article abstract of DOI:10.1007/s00775-012-0911-2

A series of cationic dinuclear p-cymene ruthenium trithiophenolato complexes of the type [(η⁶-p-MeC₆H₄Pr ⁱ )₂Ru₂(SC₆H₄-p-X) ₃]⁺ (1 X is H, 2 X is Me, 3 X is Ph, 4 X is Br, 5 X is OH, 6 X is NO₂, 7 X is OMe, 8 X is CF₃, 9 X is F, 10 X is Pr ⁱ, 11 X is Bu ^t ) have been synthesized from the reaction of [(η⁶-p-MeC₆H₄Pr ⁱ )RuCl ₂]₂ with the corresponding thiol, isolated as the chloride salts, and further studied for their electrochemical properties, cytotoxicity towards human ovarian cancer cells, and catalytic activity for glutathione (GSH) oxidation. Complex 1 was also compared with the benzene and hexamethylbenzene analogues [(η⁶-C₆H₆)₂Ru ₂(SC₆H₅)₃]⁺ (12) and [(η⁶-C₆Me₆)₂Ru ₂(SC₆H₅)₃]⁺ (13). The most active compound [11]Cl was structurally studied by single-crystal X-ray diffraction analysis. The concentrations corresponding to 50 % inhibition of cancer cell growth (IC₅₀ values) in the A2780 and A2780cisR cell lines of these complexes except for 6 were in the submicromolar range, complex 11 showing an IC₅₀ value of 0.03 μM in both cell lines. The high in vitro anticancer activity of these complexes may be at least partially due to their catalytic potential for the oxidation of GSH, although there is no clear correlation between the IC₅₀ values and the turnover frequencies at about 50 % conversion. However, the cytotoxicity is tentatively correlated to the physicochemical properties of the compounds determined by the electronic influence of the substituents X (Hammett constants σ_p) and the lipophilicity of the thiols p-XC₆H₄SH (calculated log P parameters).

Full text of DOI:10.1007/s00775-012-0911-2

J Biol Inorg Chem (2012) 17:951–960
DOI 10.1007/s00775-012-0911-2
ORIGINAL PAPER
Highly cytotoxic trithiophenolatodiruthenium complexes
of the type [(g6-p-MeC6H4Pri)2Ru2(SC6H4-p-X)3]+: synthesis,
molecular structure, electrochemistry, cytotoxicity,
and glutathione oxidation potential
•
Federico Giannini Julien Furrer Anne-Flore Ibao
•
•
•
•
•
Georg Su¨ss-Fink Bruno Therrien Olivier Zava
ˇ
Mathurin Baquie Paul J. Dyson Petr Stepnicka
•
•
ˇ
ˇ
Received: 20 March 2012 / Accepted: 25 May 2012 / Published online: 16 June 2012
Ó SBIC 2012
Abstract A series of cationic dinuclear p-cymene ruthe-
nium trithiophenolato complexes of the type [(g6-p-
MeC6H4Pri)2Ru2(SC6H4-p-X)3]? (1 X is H, 2 X is Me, 3
X is Ph, 4 X is Br, 5 X is OH, 6 X is NO2, 7 X is OMe, 8
X is CF3, 9 X is F, 10 X is Pri, 11 X is But) have been
synthesized from the reaction of [(g6-p-MeC6H4Pri)-
RuCl2]2 with the corresponding thiol, isolated as the
chloride salts, and further studied for their electrochemical
properties, cytotoxicity towards human ovarian cancer
cells, and catalytic activity for glutathione (GSH) oxida-
tion. Complex 1 was also compared with the benzene and
hexamethylbenzene analogues [(g6-C6H6)2Ru2(SC6H5)3]?
(12) and [(g6-C6Me6)2Ru2(SC6H5)3]? (13). The most active
compound [11]Cl was structurally studied by single-crystal
X-ray diffraction analysis. The concentrations corresponding
to 50 % inhibition of cancer cell growth (IC50 values) in the
A2780andA2780cisRcelllinesofthesecomplexesexceptfor
6 were in the submicromolar range, complex 11 showing an
IC50 value of 0.03 lM in both cell lines. The high in vitro
anticanceractivity ofthesecomplexesmay beatleast partially
due to their catalytic potential for the oxidation of GSH,
although there is no clear correlation between the IC50 values
and the turnover frequencies at about 50 % conversion.
However, the cytotoxicity is tentatively correlated to the
physicochemical properties of the compounds determined by
the electronic influence of the substituents X (Hammett con-
stants rp) and the lipophilicity of the thiols p-XC6H4SH
(calculated log P parameters).
Dedicated to Prof. Wolfgang Beck on the occasion of his 80th
birthday.
Keywords Ruthenium ꢀ Cancer ꢀ Cytotoxicity ꢀ
Lipophilicity ꢀ Glutathione oxidation
Electronic supplementary material The online version of this
article (doi:10.1007/s00775-012-0911-2) contains supplementary
material, which is available to authorized users.
Introduction
F. Giannini ꢀ J. Furrer (&)
¨
Department fu¨r Chemie und Biochemie, Universitat Bern,
3012 Berne, Switzerland
e-mail: julien.furrer@dcb.unibe.ch
The field of antitumoral and antimetastatic arene ruthenium
complexes was pioneered independently by Sadler [1, 2]
and Dyson [3, 4], who reported the organometallic com-
plexes [(g6-PhC6H5)Ru(en)Cl]? (en is 1,2-diaminoethane)
[5] and [(g6-p-MeC6H4Pri)Ru(P-pta)Cl2] (pta is 1,3,5-
triaza-7-phosphatricyclo[3.3.1.1]decane) [6] in 2001 [7].
The mode of action by which arene ruthenium complexes
exert their antitumoral or antimetastatic effects is not yet
fully understood. By analogy with platinum complexes, it
was originally expected that DNA binding was also the
main reason for the anticancer activity of these ruthenium
complexes, but serum proteins have also been discussed as
possible targets [8]. Although the ability of ruthenium to
A.-F. Ibao ꢀ G. Su¨ss-Fink (&) ꢀ B. Therrien
´ ˆ
Institut de Chimie, Universite de Neuchatel,
ˆ
2000 Neuchatel, Switzerland
e-mail: georg.suess-fink@unine.ch
O. Zava ꢀ M. Baquie ꢀ P. J. Dyson
´
Institut des Sciences et Ingenierie Chimiques, (EPFL),
Ecole Polytechnique Federale de Lausanne,
´ ´
1015 Lausanne, Switzerland
ˇ
P. Stepnicka
ˇ
ˇ
Department of Inorganic Chemistry, Faculty of Science,
Charles University in Prague, 12840 Prague 2, Czech Republic
123

952
J Biol Inorg Chem (2012) 17:951–960
bind to DNA has been demonstrated [9, 10], in particular
for arene ruthenium 1,2-diaminoethane complexes [11–13],
it was observed that DNA binding of ruthenium was
weaker and different from that observed for platinum [14–
16]. These findings suggest different modes of action
depending on the type of complexes. Thus, arene ruthe-
nium 1,3,5-triaza-7-phosphatricyclo[3.3.1.1]decane com-
plexes have been found to inhibit thioredoxin reductase
and cathepsin B [3, 4]. Indeed, bioanalytical data com-
bined with an approximately 200 kDa crystal structure of
[(g6-p-MeC6H4Pri)Ru(P-pta)Cl2] binding to the nucleo-
some core particle show an overwhelming preference for
protein binding of this compound [17]. In addition to direct
DNA and protein interactions, another mode of action has
been found for cytotoxic arene ruthenium iodoazopyridine
complexes involving the catalytic oxidation of glutathione
(GSH), which is supposed to be at the origin of their
anticancer activity [18].
p-XC6H4SH [0.98 mmol; X is NO2 152 mg, X is OMe 120
lL, X is CF3 129 lL, X is F 104 lL, X is CH(CH3)2
152 lL, and X is C(CH3)3 169 lL] in technical grade eth-
anol (5 mL) was added dropwise to the hot solution. The
resulting mixture was refluxed for 18 h. After the mixture
had cooled to 20 °C, the solvent was removed under
reduced pressure. The oil obtained was purified by column
chromatography on silica gel using a mixture of dichloro-
methane and ethanol (5:1) as the eluent. The chloride salts
of cations 6–11 (Scheme 1) were isolated as air-stable
orange to red solids and dried in vacuo.
Spectroscopic and analytical data for [6]Cl
Red crystalline solid, yield 75 mg (48 %) C38H40N3O6-
Ru2S3ClꢀH2OꢀEtOH (1,032.61): calcd. C 46.53, H 4.69;
found C 46.74, H 4.61. ESI mass spectrometry (MS)
1
(MeOH): m/z 933.4 [M]?. H NMR (400 MHz, CDCl3):
We recently observed that water-soluble and air-stable
arene ruthenium complexes of the type [(g6-arene)2-
Ru2(SR)3]? are surprisingly cytotoxic despite their inert-
ness to ligand substitution [19], and that the most active
derivative [(g6-p-MeC6H4Pri)2Ru2(SC6H4-p-Me)3]? [20] is
a highly efficient catalyst for the oxidation of GSH
in aqueous solution, which may explain its high cytotox-
icity [21]. In this article, we report our systematic study of
[(g6-p-MeC6H4Pri)2Ru2(SC6H4-p-X)3]? complexes, most
of which are newly designed compounds, aiming at finding
correlations between anticancer activity, catalytic GSH
oxidation activity, and redox properties.
d = 8.31 (d, 3J = 8.0 Hz, 6H, SC6H4NO2), 8.20 (d,
3
3J = 8.0 Hz, 6H, SC6H4NO2), 5.77 (d, J = 6.0 Hz, 2H,
H–Ar), 5.63 (d, 3J = 6.0 Hz, 2H, H–Ar), 5.43 (m, 4H,
3
H–Ar), 1.99 [sept, J = 6.8 Hz, 2H, CH(CH3)2], 1.68 (s,
3
6H, CH3), 0.92 [d, J = 6.8 Hz, 6H, CH(CH3)2], 0.80 [d,
3J = 6.8 Hz, 6H, CH(CH3)2] ppm. 13C{1H} NMR
(100 MHz, CDCl3): d = 149.1, 148.1, 135.0, 125.0, 109.7,
102.5, 86.4, 85.7, 85.5, 32.3, 25.2, 22.1, 17.9 ppm.
Spectroscopic and analytical data for [7]Cl
Orange crystalline solid, yield 140 mg (93 %) C41H49-
O3Ru2S3ClꢀCH2Cl2 (966.04): calcd. C 53.32, H 5.35; found
1
C 53.17, H 5.96. ESI MS (MeOH): m/z 889.4 [M]?. H
3
Materials and methods
NMR (400 MHz, CDCl3): d = 7.79 (d, J = 8.8 Hz, 6H,
SC6H4OCH3), 6.91 (d, 3J = 8.8 Hz, 6H, SC6H4OCH3),
5.32 (d, 3J = 6.0 Hz, 2H, H–Ar), 5.19 (d, 3J = 6.0 Hz, 2H,
H–Ar), 5.10 (d, 3J = 6.0 Hz, 2H, H–Ar), 5.07 (d,
3J = 6.0 Hz, 2H, H–Ar), 3.88 (s, 9H, OCH3), 1.96 (sept,
3J = 6.8 Hz, 2H, CH(CH3)2), 1.62 (s, 6H, CH3), 0.91 (d,
Materials and analyses
The starting material [(g6-p-MeC6H4Pri)RuCl2]2 [22] and
the known complexes 1–5 (Scheme 1) [20, 23, 24] were
prepared according to published methods. All other
reagents were commercially available and were used
without further purification. Electrospray ionization (ESI)
mass spectra were obtained in positive-ion or negative-ion
mode with an LCQ Finnigan mass spectrometer. Microa-
nalyses were performed by the Mikroelementaranalytisches
Laboratorium, ETH Zu¨rich (Switzerland).
3
3J = 6.8 Hz, 6H, CH(CH3)2), 0.83 (d, J = 6.8 Hz, 6H,
CH(CH3)2) ppm. 13C{1H} NMR (100 MHz, CDCl3):
d = 160.1, 133.9, 128.6, 114.8, 107.3, 99.6, 85.4, 84.9,
84.7, 83.6, 55.7, 30.7, 22.7, 22.1, 17.8 ppm.
Spectroscopic and analytical data for [8]Cl
Orange crystalline solid, yield 120 mg (76 %) C41H40-
F9Ru2S3Cl (1,037.99): calcd. C 47.46, H 3.89; found C
Synthesis and characterization
1
of compounds [6]Cl–[11]Cl
47.65, H 4.11. ESI MS (MeOH): m/z 1,002.8 [M]?. H
3
NMR (400 MHz, CDCl3): d = 8.20 (d, J = 8.0 Hz, 6H,
The dinuclear complex [(g6-p-MeC6H4Pri)RuCl2]2 (0.16
mmol, 100 mg) was heated in technical grade ethanol
(50 mL). As soon as the starting material had completely
dissolved, a solution of the corresponding thiophenol
SC6H4CF3), 7.71 (d, 3J = 8.0 Hz, 6H, SC6H4CF3), 5.77 (d,
3J = 6.0 Hz, 2H, H–Ar), 5.37 (m, 4H, H–Ar), 5.26 (d,
3J = 6.0 Hz, 2H, H–Ar), 1.89 [sept, 3J = 6.8 Hz, 2H,
3
CH(CH3)2], 1.68 (s, 6H, CH3), 0.89 [d, J = 6.8 Hz, 6H,
123

J Biol Inorg Chem (2012) 17:951–960
953
CH(CH3)2], 0.75 [d, 3J = 6.8 Hz, 6H, CH(CH3)2] ppm. 19
F
85.1, 84.6, 84.1, 84.0, 58.6, 34.9, 31.3, 30.4, 22.5, 22.0,
17.8 ppm.
NMR (376 MHz, CDCl3): d = -62.5 ppm. 13C{1H} NMR
(100 MHz, CDCl3): d = 142.5, 133.2, 130.5, 126.1, 125.3,
107.7, 100.3, 85.9, 85.3, 85.0, 84.1, 30.7, 22.5, 21.7,
17.8 ppm.
X-ray crystal structure analysis of [11]Cl
A crystal of [11]Cl was mounted on a Stoe image plate
diffraction system equipped with a / circle goniometer,
using Mo Ka graphite monochromated radiation
Spectroscopic and analytical data for [9]Cl
˚
Orange crystalline solid, yield 65 mg (45 %) C38H40-
F3Ru2S3Clꢀ0.25CH2Cl2 (926.88): calcd. C 51.43, H 4.54;
found C 51.37, H 5.04. ESI MS (MeOH): m/z 853.2 [M]?.
1H NMR (400 MHz, CDCl3): d = 7.89 (m, 6H, SC6H4F),
(k = 0.71073 A) with / range 0°–200°, an increment of
˚
1.2°, and Dmax-Dmin = 12.45-0.81 A. The structure was
solved by direct methods using the program SHELXS-97
[25]. The refinement and all further calculations were done
using SHELXL-97 [25]. The hydrogen atoms were inclu-
ded in calculated positions and treated as riding atoms
using the SHELXL default parameters. All nonhydrogen
atoms were refined anisotropically, using weighted full-
matrix least squares on F2. Crystallographic details are
summarized in Table 1. An ORTEP drawing [26] of [11]Cl
is shown in Fig. 1.
3
7.07 (m, 6H, SC6H4F), 5.49 (d, J = 6.0 Hz, 2H, H–Ar),
5.23 (d, 3J = 6.0 Hz, 2H, H–Ar), 5.16 (m, 4H, H–Ar), 1.89
[sept, 3J = 6.8 Hz, 2H, CH(CH3)2], 1.58 (s, 6H, CH3), 0.86
[d, 3J = 6.8 Hz, 6H, CH(CH3)2], 0.75 [d, 3J = 6.8 Hz, 6H,
CH(CH3)2] ppm. 19F NMR (376 MHz, CDCl3): d =
-112.4 ppm. 13C{1H} NMR (100 MHz, CDCl3): d =
161.8, 134.6, 134.5, 132.9, 116.6, 116.4, 107.6, 100.1,
85.6, 85.2, 85.1, 83.9, 30.8, 22.6, 22.1, 17.9 ppm.
CCDC 866470 contains the supplementary crystallo-
graphic data for this article. These data can be obtained free
of charge from the Cambridge Crystallographic Data
Centre via http://www.ccdc.cam.ac.uk/data_request/cif.
Spectroscopic and analytical data for [10]Cl
Red crystalline solid, yield 170 mg (86 %) C47H61Ru2S3-
Clꢀ0.5CH2Cl2 (1,002.25): calcd. C 56.92, H 6.24; found C
56.96, H 6.38. ESI MS (MeOH): m/z 925.4 [M]?. 1H NMR
3
Table 1 Crystallographic and selected experimental data of [11]Cl
(400 MHz, CDCl3): d = 7.79 [d, J = 8.0 Hz, 6H, SC6H4
CH(CH3)2], 7.23 [d, 3J = 8.0 Hz, 6H, SC6H4C(CH3)3],
5.40 (d, 3J = 6.0 Hz, 2H, H–Ar), 5.19 (d, 3J = 6.0 Hz, 2H,
Chemical formula
Formula weight
Crystal system
C50H67ClRu2S3
1,001.81
3
H–Ar), 5.12 (d, J = 6.0 Hz, 2H, H–Ar), 5.06 (d, J =
3
Orthorhombic
P212121
3
6.0 Hz, 2H, H–Ar), 2.95 [sept, J = 6.8 Hz, 3H, SC6H4
Space group
3
CH(CH3)2], 1.87 [sept, J = 6.8 Hz, 2H, CH(CH3)2], 1.64
Crystal color and shape
Crystal size (mm3)
Red block
0.21 9 0.18 9 0.17
11.5549 (4)
20.2135 (7)
22.2663 (8)
5200.6 (3)
4
(s, 6H, CH3), 1.28 [d, 3J = 6.8 Hz, 18H, SC6H4
CH(CH3)2], 0.87 [d, 3J = 6.8 Hz, 6H, CH(CH3)2], 0.72 [d,
3J = 6.8 Hz, 6H, CH(CH3)2] ppm. 13C{1H} NMR
(100 MHz, CDCl3): d = 149.6, 134.8, 132.6, 127.3, 106.7,
100.4, 85.8, 84.8, 84.3, 84.0, 33.9, 30.5, 24.0, 22.6, 22.0,
17.8 ppm.
˚
a (A)
˚
b (A)
˚
c (A)
3
V (A )
˚
Z
T (K)
Dc (g cm-3
l (mm-1
293 (2)
Spectroscopic and analytical data for [11]Cl
)
1.279
)
0.782
Orange crystalline solid, yield 150 mg (92 %) C50H67-
Ru2S3ClꢀEtOH (1,047.92): calcd. C 59.60, H 7.02; found C
59.46, H 7.07. ESI MS (MeOH): m/z 967.4 [M]?. 1H NMR
Scan range (°)
3.66 \ 2h \ 58.50
14,045
Unique reflections
Reflections used [I [ 2r(I)]
Rint
3
11,010
(400 MHz, CDCl3): d = 7.78 [d, J = 8.0 Hz, 6H, SC6H4
C(CH3)3], 7.38 [d, J = 8.0 Hz, 6H, SC6H4C(CH3)3], 5.41
3
0.0953
(d, 3J = 6.0 Hz, 2H, H–Ar), 5.20 (d, 3J = 6.0 Hz, 2H,
Flack parameter
0.02 (4)
Final R indices [I [ 2r(I)]a
R indices (all data)
Goodness of fit
R1 0.0686, wR2 0.1079
R1 0.0954, wR2 0.1163
1.071
3
3
H–Ar), 5.12 (d, J = 6.0 Hz, 2H, H–Ar), 5.06 (d, J =
6.0 Hz, 2H, H–Ar), 1.84 [sept, 3J = 6.8 Hz, 2H, CH
(CH3)2], 1.65 (s, 6H, CH3), 1.35 [s, 27H, SC6H4C(CH3)3],
-3
3
3
˚
Max, min Dq/e (A
)
0.752, -1.014
0.86 [d, J = 6.8 Hz, 6H, CH(CH3)2], 0.69 (d, J = 6.8
Hz, 6H, CH(CH3)2) ppm. 13C{1H} NMR (100 MHz,
CDCl3): d = 151.9, 134.5, 132.3, 126.2, 106.5, 100.7,
a
2
2
2 2
2 2 1/2
Structures were refined on F0 : wR2 = [R[w(F0 -Fc ) ]/Rw(F0 ) ]
,
where w-1 = R F0 ? (aP)2 ? bP and P = [max(F0 ,0) ? 2Fc ]/3.
2
2
2
123

954
J Biol Inorg Chem (2012) 17:951–960
Table 2 Cytotoxicity of [1]Cl–[13]Cl and Cisplatin towards A2780
and A2780cisR cancer cell lines
Compound
IC50 for A2780 (lM)
IC50 for A2780cisR (lM)
[1]Cl
0.24 0.01
0.13 0.01
0.28 0.03
0.25 0.01
0.53 0.02
26.01 0.62
0.18 0.03
0.17 0.05
0.66 0.08
0.08 0.01
0.03 0.01
0.37 0.03
0.43 0.04
1.35 0.15
0.25 0.02
0.08 0.03
0.10 0.01
0.26 0.05
1.47 0.02
54.43 1.38
0.15 0.01
0.27 0.05
1.05 0.06
0.05 0.01
0.03 0.01
0.47 0.05
1.20 0.2
[2]Cl
[3]Cl
[4]Cl
[5]Cl
[6]Cl
[7]Cl
[8]Cl
[9]Cl
[10]Cl
[11]Cl
[12]Cl
[13]Cl
Cisplatin
Fig. 1 ORTEP drawing of [11]Cl with 50 % probability level
ellipsoids with hydrogen atoms omitted for clarity. Selected bond
˚
lengths (A) and angles (°): Ru1–Ru2 3.3536(5), Ru1–S1 2.3926(15),
Ru1–S2 2.4293(13), Ru1–S3 2.4015(13), Ru2–S1 2.3797(15), Ru2–
S2 2.4008(12), Ru2–S3 2.4248(14); Ru1–S1–Ru2 89.29(5), Ru1–S2–
Ru2 87.94(4), Ru1–S3–Ru2 88.03(5)
12.18 0.59
IC50 concentration corresponding to 50 % inhibition of cancer cell
growth
Electrochemistry
Electrochemical measurements were done with a lAUTO-
LAB III multipurpose polarograph (Eco Chemie, The Neth-
erlands) at room temperature using a Metrohm three-
electrode cell equipped with a platinum disc working elec-
trode (2-mm diameter), a platinum sheet auxiliary electrode,
and a double-junction Ag/AgCl (3 M KCl) reference elec-
trode. The compounds were dissolved in a dry solvent (all
from Sigma-Aldrich) to give a solution containing approxi-
mately 0.5 mM analyte and 0.1 M Bu4NPF6 (Fluka, p.a. for
electrochemistry). The solutions were deaerated with argon
prior to the measurement and then kept under an argon
blanket. Ferrocene was used as an internal reference.
cells at a final concentration of 0.25 mg mL-1 and the mix-
ture was incubated for 2 h; then the culture medium was
aspirated and the violet formazan (artificial chromogenic
precipitate of the reduction of tetrazolium salts by dehydro-
genases and reductases) was dissolved in DMSO. The optical
density of each well (96-well plates) was quantified three
times in triplicate at 540 nm using a multiwell plate reader
(iEMS Reader MF, Labsystems, USA), and the percentage of
surviving cells was calculated from the ratio of the absor-
bance of treated cells to theabsorbanceof untreatedcells. The
concentrations corresponding to 50 % inhibition of cancer
cell growth (IC50 values) were determined by fitting the plot
of the logarithmic percentage of surviving cells against the
logarithm of the drug concentration using a linear regression
function. The median value and the median absolute devia-
tion were obtained using ExcelTM (Microsoft), and the values
are reported in Table 2.
Cell culture and inhibition of cell growth
Human A2780 and A2780cisR ovarian carcinoma cells were
obtained from the European Collection of Cell Cultures
(ECACC, Salisbury, UK) and maintained in culture as
described by the provider. The cells were routinely grown in
RPMI 1640 medium with GlutaMAXTM containing 5 % fetal
calf serum and antibiotic (penicillin and streptomycin) at
37 °C and 5 % CO2. For the evaluation of growth inhibition
tests, the cells were seeded in 96-well plates (25 9 103 cells
per well) and grown for 24 h in complete medium. The
compounds were dissolved in dimethyl sulfoxide (DMSO)
and added at the required concentration to the cell culture for
72 h incubation. Solutions of compounds were applied by
diluting a freshly prepared stock solution of the correspond-
ing compound in aqueous RPMI medium with GlutaMAXTM
(20 mM). Following drug exposure, 3-(4,5-dimethylthiazol-
2-yl)-2,5-diphenyltetrazolium bromide (MTT) was added to
NMR spectroscopy
NMR data were acquired at 37 °C using a Bruker Avan-
ce II 500-MHz NMR spectrometer equipped with an
inverse dual channel (1H, X) z-gradient probe head
(broadband inverse) or using a Bruker Avance II 400-MHz
NMR spectrometer equipped with an inverse dual channel
(1H, X) z-gradient probe head (broadband inverse). One-
dimensional 1H NMR data were acquired with 16–64
transients as 32,768 data points over a width of 12 ppm
using a classical presaturation to eliminate the water res-
onance. A relaxation delay of 6 s was applied between the
transients. All NMR data were processed using Topspin
123

J Biol Inorg Chem (2012) 17:951–960
955
1
(version 2.1 or 3.0, Bruker, Switzerland). The H d scale
was referenced to the residual water signal at 4.637 ppm
(37 °C); the 13C and 19F d scales were referenced to
external tetramethylsilane and CFCl3, respectively.
To evaluate the catalytic performance of the complexes
for the oxidation of the reduced form of GSH to the
disulfide form (GSSG) under physiological conditions, the
complexes (approximately 0.2 lM) were dissolved in
0.6 mL of D2O, and 100 equiv of GSH was added to the
solution.
Br)3]? [23]. In accordance with the electron count, the
˚
Ru–Ru distance of 3.3536(5) A is clearly outside the range
˚
for a metal–metal single bond (2.28–2.95 A) [23].
The redox behavior of complex 2 as a representative was
studied first by cyclic voltammetry in the anodic region at a
platinum disc electrode using approximately 0.5 mM
solutions in various dry solvents containing 0.1 M
Bu4NPF6 as the supporting electrolyte (see the electronic
supplementary material). In acetonitrile, the compound
showed an irreversible anodic wave, which was followed
by several relatively weaker irreversible peaks (Fig. S1).
The redox response changed upon changing the solvent for
the less polar, noncoordinating 1,2-dichloroethane
(Fig. S2). The voltammograms recorded in this solvent
were complicated owing to extensive adsorption, particu-
larly in the case of 12 (Fig. S3), which has no substituents
at the arene ring and is thus probably more prone to
interactions with the electrode surface via its arene ligands.
Finally, the cyclic voltammograms of all compounds
were recorded in the highly polar and donating DMSO,
which is chemically more relevant to the conditions of
biological testing1. The compounds typically displayed
only one2 single irreversible, diffusion-controlled3 oxida-
tion at approximately 0.66–0.78 V versus ferrocene/ferro-
cenium (Fig. 2). However, even these waves were probably
associated with some chemical complications (e.g.,
adsorption phenomena) and typically shifted upon repeated
scanning, which precluded any exact determination of the
redox potentials. Indeed, the roughly estimated anodic peak
potentials increased with the electron-withdrawing char-
acter of the substituent at position 4 of the thiophenolato
ligand. However, because of uncertainty of the redox
potential determination, no reliable correlation could be
drawn between the redox potentials and the Hammett
constants (rp) [27].
1
The samples were subsequently analyzed by H NMR
spectroscopy. For all complexes, the 1H NMR spectra were
recorded immediately after sample preparation, and then
every 30 min until complete oxidation was evidenced by
the complete disappearance of the original resonances of
cysteine. To study a potential effect of the presence of
chloride ions, the experiments were conducted in 4 and
50 mM D2O solutions of NaCl.
Results and discussion
The p-cymene ruthenium dichloride dimer reacts in re-
fluxing ethanol with various thiophenols to give the cat-
ionic trithiophenolato complexes [(g6-p-MeC6H4Pri)2Ru2
(SC6H4-p-X)3]? (1 X is H, 2 X is Me, 3 X is Ph, 4
X is Br, 5 X is OH, 6 X is NO2, 7 X is OMe, 8 X is CF3,
9 X is F, 10 X is Pri, 11 X is But), which can be isolated in
high yields by column chromatography as air-stable chlo-
ride salts (Scheme 1). Compounds [1]Cl–[5]Cl have been
reported previously [19–21, 23]. The newly prepared
derivatives [6]Cl–[11]Cl were fully characterized by ele-
mental analysis and spectroscopic methods.
The known benzene and hexamethylbenzene analogues
of complex 1, [(g6-C6H6)2Ru2(SC6H5)3]? (12) and
[(g6-C6Me6)2Ru2(SC6H5)3]? (13), which were included in
this study for comparison, were synthesized in the same
way according to the published method [12].
The antiproliferative activity of complexes 1–11 was
evaluated towards the human ovarian A2780 cancer cell line
and its cisplatin-resistant derivative A2780cisR using the
MTT assay, which measures mitochondrial dehydrogenase
activityasanindication ofcellviability. TheIC50 valuesofthe
complexes are reported in Table 2 together with those of 12,
13, and cisplatin, which are included for comparison.
Suitable crystals for X-ray analysis were obtained for
the tert-butyl derivative [11]Cl by recrystallization from a
chloroform/diethyl ether mixture. The molecular structure,
shown in Fig. 1, contains a trigonal bipyramidal Ru2S3
framework, in which each ruthenium atom adopts a
pseudo-octahedral geometry owing to the presence of three
sulfur atoms and the p-cymene ligands that each formally
occupies three coordination sites. Selected bond lengths
and angles are listed in Fig. 1, and crystallographic details
are summarized in Table 1.
All complexes 1–11 with exception of the nitro deriv-
ative 6 are highly cytotoxic towards human ovarian cancer
cells, the IC50 values for the cell lines A2780 and
A2780cisR being in the nanomolar range. The most active
1
Addition of water (1 % v/v) to DMSO did not change the redox
response.
The Ru–S bond distances in cation 11 range from
˚
2.3797(15) to 2.4293(15) A and the Ru–S–Ru angles range
2
Some compounds (e.g., [5]Cl) showed complicated convoluted
from 87.91(4) to 89.29(5)°, similar to those found in the
known p-cymene derivatives [(g6-p-MeC6H4Pri)2Ru2
(SC6H5)3]? [23] and [(g6-p-MeC6H4Pri)2Ru2(SC6H4-p-
waves.
3
Anodic peak currents ipa increased linearly with the square root of
the scan rate.
123

956
J Biol Inorg Chem (2012) 17:951–960
+ Cl-
Scheme 1 Synthesis of
compounds [(g6-p-
X
MeC6H4Pri)2Ru2(SC6H4-p-
X)3]Cl (1 X is H, 2 X is Me,
3 X is Ph, 4 X is Br, 5 X is OH,
6 X is NO2, 7 X is OMe,
8 X is CF3, 9 X is F,
e
n
3 XC6H4SH
3 HCl
e
r
S
Cl
Cl
a
Cl
Ru
Ru
Ru
Ru
10 X is Pri, 11 X is But)
Cl
S
S
X
X
present in the reduced form (GSH) and less than 10 %
exists in the oxidized disulfide form (GSSG) [33, 34].
We therefore studied the possible interaction of GSH
with complex 2 and found it catalyzes the oxidation of
cysteine to cystine and also that of GSH to GSSG in water
[21]. Here we report a systematic NMR study of the cat-
alytic oxidative behavior of all 11 compounds, following
the reaction between complexes 1–11 with GSH in a 1:100
ratio. The 1H NMR spectra, recorded in a 4 mM NaCl
solution in D2O/DMSO-d6 (95:5) at pD 7 and 37 °C and in
an aerobic atmosphere (the GSH autoxidation in the pres-
ence of O2 being less than 5 % in 24 h), showed that the
incubation of 90 mM GSH with 0.9 mM 1–11 leads to the
complete oxidation of GSH to GSSG in 13–16 h, as evi-
denced by the disappearance of the b-CH2 resonances of
GSH at d * 3 ppm and the simultaneous appearance of
two new resonances at d * 3.1 ppm and d * 3.4 ppm,
corresponding to the b-CH2 of GSSG (see the electronic
supplementary material). To study a potential effect of the
presence of chloride ions, experiments were also conducted
in 50 mM D2O solutions of NaCl. The results for com-
plexes 8–11 show the turnover frequencies at about 50 %
conversion (TOF50) increase by only about 10 % upon
increasing the concentration of the chloride ions from 4 to
50 mM, thus confirming our findings for complex 2 [21].
Figure 3 shows the catalytic turnover as a function of
time for the most active complex, 11. The TOF50 values are
listed in Table 3; they were obtained from each catalytic
run by fitting the turnover frequencies as a function of time
with the exponential expression y = a - bcx for all com-
plexes. The turnover frequencies were calculated according
to the following equation: fIGSSG=ðIGSH þ IGSSGÞg ꢁ
Fig. 2 Cyclic voltammograms of [2]Cl recorded in dimethyl sulfox-
ide (scan rate 0.2 V s-1). Full line first scan, dashed line second scan
complex, 11, has an IC50 value of 0.03 lM for both cell
lines, which means that for cell line A2780 it is more than
100 times more cytotoxic than cisplatin. For the cisplatin-
resistant cell line A2780cisR, complex 11 is more than 500
times more cytotoxic than cisplatin. To the best of our
knowledge, these complexes are the most cytotoxic
ruthenium compounds reported so far [28–31]. Such strong
activity may appear surprising at first glance, since the
compounds are apparently chemically inert. However, a
recent study by Meier et al. [32] suggests an inverse cor-
relation between metallodrug–protein interaction and
cytotoxicity towards tumor cells. Indeed, despite their
inertness towards biomolecules such as amino acids and
nucleotides, complexes 1–11 have been found to be active
catalysts for the oxidation of GSH and potentially other
peptides/proteins, which could explain at least partially
their biological activity.
È
É
½GSHꢂ0=½complexꢂ , where IGSSG and IGSH are the inte-
gral intensities of the signals of GSSG and GSH respec-
tively. The turnover frequencies were obtained as a
derivative of the fitting function for x = 2 (after 2 h
incubation corresponding to 50 % conversion of GSH to
GSSG). All turnover frequencies obtained are in the range
The tripeptide GSH (Scheme 2) is the major endoge-
nous antioxidant produced by cells (up to 5 mM in living
cells). It participates directly in the neutralization of free
radicals and reactive oxygen compounds, and maintains the
levels of exogenous antioxidants such as vitamin C. In
healthy cells, more than 90 % of the total GSH pool is
from 5.5 to 8.5 h-1
.
Table 3 summarizes the catalytic turnover frequencies
(TOF50), the Hammett constants (rp) of the substituents R
123

J Biol Inorg Chem (2012) 17:951–960
957
Scheme 2 Oxidation of
reduced glutathione (GSH) to
give the disulfide GSSG
COO-
-OOC
COO-
NH2
N
H
N
H
O
H
O
H
2 e-
O
2
N
S-
N
H
O
N
S
S
O
O
2 e-
H
N
H2N
H2N
COO-
-OOC
COO-
(GSH)
(GSSG)
The data shown in Table 3 suggest there is no clear cor-
relation between the IC50 values for the cytotoxicity and the
TOF50 values for the catalytic oxidation of GSH to GSSG,
presumably owing to differences in cellular uptake of the
compounds, which is related to the lipophilicity (see below).
For instance, the TOF50 value of the most cytotoxic complex
of this series, complex 11 (IC50 = 0.03 lM towards both cell
lines) drops to 6.4 h-1, whereas it increases to 8.4 h-1 for
complex 3 (IC50 = 0.28 lM towards A2780 cells), but goes
down again to 7.5 h-1 for the least cytotoxic complex, 6
(IC50 = 26 and 54.4 lM towards A2780 and A2780cisR,
respectively). This is presumably because the IC50 data
express the anticancer activity of 1–11 for living cancer cells
in cell culture, whereas the TOF50 data relate to the catalytic
activity of 1–11 for the oxidation of GSH in an isolated sys-
tem. Accordingly, there is no correlation between the TOF50
data and the Hammett constants (rp) nor with the partition
coefficients (log P). However, such correlations can be found
between the IC50 data and rp and log P (see below). Indeed,
physicochemical properties such as electronic parameters and
lipophilicity appear to have a greater influence on the activity
of compounds in cells than the catalytic potential for the
oxidation of GSH.
100
80
60
40
20
TOF (per 1 Ru atom) = 6.4 h-1 (2h)
10 12 14 16 18
0
2
4
6
8
Time (h)
Fig. 3 Turnover frequency (TOF) of complex 11 for the catalytic
oxidation of reduced glutathione
reflecting their electronic influence [27], and the calculated
partition coefficients (log P) of the complexes reflecting
their lipophilicity. The partition coefficients (log P) were
calculated using ACD/ChemSketch [35]. As the p-cymene
ruthenium fragments are the same in complexes 1–11, the
lipophilicity of these complexes should vary with the RSH
log P parameters. Indeed, in a recent study, Achrem-
Achremovicz et al. [36] showed that these calculated
log P values match closely those obtain by experiment in
the case of semisynthetic botulin derivatives, and that the
values obtained can be used in correlations of biological
activities of the compounds and their structures.
From the data in Table 3, correlations between the
lipophilicity (log P) of the thiol ligands, the Hammett
constants (rp) of the substituents at sulfur, and the IC50
values for the A2780 and A2780cisR cell lines can be
extracted. In Fig. S6, the correlations between the IC50
values and the calculated partition coefficients (log P) are
shown for both cell lines. It can be clearly seen that the
lowest IC50 values are obtained for the complexes having
log P values between 3.0 and 4.2 for the A2780 cell line,
123

958
J Biol Inorg Chem (2012) 17:951–960
Table 3 Comparison of cytotoxicities and catalytic activities of [1]Cl––[11]Cl with physicochemical data for the corresponding thiols (the
log P values correspond to the values calculated for the thiol RSH groups)
Compound
IC50 for A2780 (lM)
IC50 for A2780cisR (lM)
TOF50 (h-1 a
)
rp (R)
LogP (RSH)
[1]Cl
[2]Cl
[3]Cl
[4]Cl
[5]Cl
[6]Cl
[7]Cl
[8]Cl
[9]Cl
[10]Cl
[11]Cl
0.24 0.01
0.13 0.01
0.28 0.03
0.25 0.01
0.53 0.02
26.01 0.62
0.18 0.03
0.17 0.05
0.66 0.08
0.08 0.01
0.03 0.01
0.25 0.02
0.08 0.03
0.10 0.01
0.26 0.05
1.47 0.02
54.43 1.38
0.15 0.01
0.27 0.05
1.05 0.06
0.05 0.01
0.03 0.01
7.2 0.36
7.1 0.35
8.4 0.42
8.3 0.42
7.3 0.37
7.5 0.38
7.8 0.39
5.7 0.29
7.8 0.36
7.6 0.38
6.4 0.32
0.00
-0.17
-0.01
0.23
2.52 0.28
2.98 0.28
4.28 0.33
3.53 0.39
1.68 0.29
2.61 0.30
2.46 0.30
3.49 0.36
2.81 0.39
3.86 0.28
4.21 0.29
-0.37
0.78
-0.27
0.54
0.06
-0.15
-0.20
TOF50 turnover frequency at about 50 % conversion
a
It can be assumed that the main source of error for the TOF50 values stems from the integral extracted in the NMR spectra. This error can be
rounded to 5 %, reflecting the signal-to-noise ratio of the 1H NMR spectra. The v2 values, which represent the accuracy of the fits, are provided in
the electronic supplementary material.
and between 2.8 and 4.3 for the A2780cisR cell line. A
general correlation trend, indicated by the black line in
Fig. S6, clearly emphasizes this correlation.
can be extracted from the data. First, the results suggest that
part of the high in vitro anticancer activity of these complexes
may be due to their catalytic potential for the oxidation of
GSH, as observed and proposed by Wang et al. [18]. It also
appears that some other properties/mechanisms are involved
since there is no correlation between the IC50 and TOF50
values. Second, the arene ruthenium complexes containing an
aliphatic substituent (1, 2, 10, 11) have higher cytotoxicities
than the other complexes. Moreover, the cytotoxicity of these
four complexes increases with the size of the substituent.
Third, the Hammett constants and the lipophilicity parame-
ters have a clear effect on the cytotoxicity, as shown by the
correlations observed for the compounds. It is worth noting
that the nanomolar IC50 values observed for some of these
compounds place them among the most cytotoxic arene
ruthenium compounds reported so far.
Taken together, the data shown in Figs. S6 and S7 suggest
that the complexes possessingHammett constants in the range
-0.2 \ rp \0 and log P values above 3.0 have the lowest
IC50 values, i.e., in the nanomolar region. The complexes with
Hammett constants outside the optimal range (-0.2 \
rp \ 0), namely, 4, 5, and 8, have significantly larger IC50
values than the other complexes, irrespective of their
log P values. The fluorine-containing complex 8 (R is CF3) is
an exception, with low IC50 values (0.17 and 0.27 lM for
A2780 and A2780cisR cells, respectively) despite having a
rather high Hammett constant (rp = 0.54).
A three-dimensional graphical representation of the
correlation between the IC50 values, the Hammett constants
(rp), and the lipophilicity parameters (log P) for the 11
complexes is displayed in Fig. 4. Interestingly, the opti-
mization problem admits a unique solution described by a
second-order polynomial function. The regression shows a
good nonlinear determination coefficient: R2 = 0.92. In
this representation, the optimal region encompassing the
most favorable values for both the Hammett constant
(-0.2 \ rp \ 0) and the log P values (log P [ 3.0) lead-
ing to the lowest IC50 values is apparent.
The lipophilicity undoubtedly plays an important role
for future design of metal drugs [37, 38], and it was
recently proposed that the level of activity can be corre-
lated to some extent to physicochemical properties of the
compounds, such as their RuIII/II redox potential and their
lipophilicity [39]. It is now well accepted that most of the
RuIII derivatives such NAMI-A and KP-1019 are actually
prodrugs that become cytotoxic only once they have been
activated by reduction [40, 41]. Correlation between the
Hammett constants, the lipophilicity parameters, and the
cytotoxicity data for our complexes could be established
but not really rationalized. As a reasonable hypothesis, one
could assume that the ruthenium compounds alter in some
way the behavior of certain enzymes in the cells after the
formation of reactive oxygen species, with uptake by the
cancer cells influenced by the lipophilicity properties. How
these physicochemical properties of the ruthenium
Conclusions
The 11 trithiophenolatodiruthenium complexes tested are
highly cytotoxic with comparable effects on both cisplatin-
sensitive and cisplatin-resistant human ovarian cancer cell
lines, with the exception of complex 6. Three general trends
123

J Biol Inorg Chem (2012) 17:951–960
959
Fig. 4 Three-dimensional
representation of the correlation
between the concentrations
corresponding to 50 %
inhibition of cancer cell growth
(IC50), the Hammett constants
(rp), and the lipophilicity
parameters (log P) for the 11
ruthenium complexes. a A2780
cell line, b A2780cisR cell line.
Regressions show a nonlinear
determination coefficient
(R2 = 0.92) in both cases
5. Morris RE, Aird RE, Murdoch PdS, Chen H, Cummings J,
Hughes ND, Pearsons S, Parkin A, Boyd G, Jodrell DI, Sadler PJ
(2001) J Med Chem 44:3616–3621
compounds are responsible for the cellular oxidative power
will be studied soon.
6. Allardyce CS, Dyson PJ, Ellis DJ, Heath SL (2001) Chem
Commun 1396–1397
7. Su¨ss-Fink G (2010) Dalton Trans 39:1673–1688
8. Melchart M, Sadler PJ (2006) In: Jaouen G (ed) Bioorganomet-
allics, Wiley-VCH, Weinheim, pp 39–64
Acknowledgments This work was financially supported by the
Swiss National Science Foundation (projects 200021-131867 and
200020-131844) and by the Ministry of Education, Youths and Sports
of the Czech Republic (project MSM 0021620857).
9. Bacac M, Hotze ACG, van der Schilden K, Haasnoot JG, Pacor S,
Alessio E, Sava G, Reedijk J (2004) J Inorg Biochem 98:402–412
10. Schluga P, Hartinger CG, Egger A, Reisner E, Galanski M, Ja-
kupec MA, Keppler BK (2006) Dalton Trans 1796–1802
11. Chen H, Parkinson JA, Morris RE, Sadler PJ (2003) J Am Chem
Soc 125:173–186
12. Wang F, Chen H, Parsons S, Oswald IDH, Davidson JE, Sadler
PJ (2003) Chem Eur J 9:5810–5820
13. Yan YK, Melchart M, Habtemariam A, Sadler PJ (2005) Chem
Commun 4764–4776
References
1. Dougan SJ, Sadler PJ (2007) Chimia 61:704–715
2. Pizarro AM, Habtemariam A, Sadler PJ (2010) Top Organomet
Chem 32:21–56
3. Dyson PJ (2007) Chimia 61:698–703
4. Casini A, Hartinger CG, Nazarov AA, Dyson PJ (2010) Top
Organomet Chem 32:57–80
123

960
J Biol Inorg Chem (2012) 17:951–960
14. Egger A, Arion VB, Reisner E, Cebrian-Losantos B, Shova S,
Trettenhahn G, Keppler BK (2005) Inorg Chem 44:122–132
15. Chen H, Parkinson JA, Novakova O, Bella J, Wang F, Dawson A,
Gould R, Parsons S, Brabec V, Sadler PJ (2003) Proc Natl Acad
Sci USA 100:14623–14628
16. Wang F, Xu J, Habtemariam A, Bella J, Sadler PJ (2005) J Am
Chem Soc 127:17734–17743
17. Wu B, Ong SM, Groessl M, Adhireksan Z, Hartinger CG, Dyson
PJ, Davey CA (2011) Chem Eur J 17:3562–3566
29. Mendoza-Ferri MG, Hartinger CG, Nazarov AA, Kandioller W,
Severin K, Keppler BK (2008) Appl Organomet Chem 22:326–332
30. Mendoza-Ferri MG, Hartinger CG, Mendoza MA, Groessl M,
Egger A, Eichinger RE, Mangrum JB, Farrell NP, Maruszak M,
Bednarski PJ, Klein F, Jakupec A, Nazarov AA, Severin K,
Keppler BK (2009) J Med Chem 52:916–925
31. Mendoza-Ferri MG, Hartinger CG, Nazarov AA, Eichinger RE,
Jakupec MA, Nazarov AA, Severin K, Keppler BK (2009)
Organometallics 28:6260–6265
18. Dougan SJ, Habtemariam A, McHale SE, Pearsons S, Sadler PJ
(2008) Proc Natl Acad Sci USA 105:11628–11633
19. Gras M, Therrien B, Su¨ss-Fink G, Zava O, Dyson PJ (2010)
Dalton Trans 39:10305–10313
32. Meier SM, Hanif M, Kandioller W, Keppler BK, Hartinger CG
(2012) J Inorg Biochem 108:91–95
33. Satoh N (2010) Biochem Genet 48:816–821
34. Pastore A, Lo Russo A, Greco M, Rizzoni G, Federici G (2001)
Clin Chem 47:1467–1469
´
20. Cherioux F, Thomas CM, Monnier T, Su¨ss-Fink G (2003)
Polyhedron 22:543–548
35. Advanced Chemistry Development (2012) ACD/ChemSketch,
version 12.0. Advanced Chemistry Development, Toronto
21. Giannini F, Su¨ss-Fink G, Furrer
50:10552–10554
J (2011) Inorg Chem
´
36. Achrem-Achremowicza J, Ke˛pczynskab E, Zylewskib M, Jan-
22. Bennett MA, Huang TN, Matheson TW, Smith AK (1982) Inorg
Synth 21:74–78
eczkoc Z (2010) Biomed Chromatogr 24:261–267
37. Scolaro C, Chaplin AB, Hartinger CG, Bergamo A, Cocchietto
M, Keppler BK, Sava G, Dyson PJ (2007) Dalton Trans
5065–5072
23. Mashima K, Mikami A, Nakamura
1795–1798
A (1992) Chem Lett
´
24. Cherioux F, Therrien B, Su¨ss-Fink G (2003) Eur J Inorg Chem
1043–1047
38. Renfrew AK, Juillerat-Jeanneret L, Dyson PJ (2011) J Organomet
Chem 696:772–779
25. Sheldrick GM (2008) Acta Crystallogr A 64:112–122
26. Farrugia LJ (1997) J Appl Crystallogr 30:565
27. Hansch C, Leo A, Taft RW (1991) Chem Rev 91:165–195
28. Mendoza-Ferri MG, Hartinger CG, Eichinger RE, Stolyarova N,
Severin K, Jakupec MA, Nazarov AA, Keppler BK (2008)
Organometallics 27:2405–2407
39. Fetzer L, Boff B, Ali M, Meng XJ, Collin JP, Sirlin C, Gaiddon
C, Pfeffer M (2011) Dalton Trans 40:8869–8878
40. Gianferrara T, Brastos I, Alessio
7588–7598
E (2009) Dalton Trans
41. Bergamo A, Gaiddon C, Schellen JHM, Beijnen JH, Sava G
(2012) J Inorg Biochem 106:90–99
123