Organometallic osmium(ii) arene anticancer complexes containing picolinate derivatives

Article abstract of DOI:10.1021/ic8020222

Chlorido osmium(ll) arene [(Η⁶-biphenyl)Os ^ll(X-pico)CI] complexes containing X = Br (1), OH (2), and Me (3) as ortho, or X = CI (4), CO₂H (5), and Me (6) as para substituents on the picolinate (pico) ring have been synthe

Full text of DOI:10.1021/ic8020222

Inorg. Chem. 2009, 48, 1753-1762
Organometallic Osmium(II) Arene Anticancer Complexes Containing
Picolinate Derivatives
Sabine H. van Rijt,† Anna F. A. Peacock,‡ Russell D. L. Johnstone,‡ Simon Parsons,‡
and Peter J. Sadler*,†
Department of Chemistry, UniVersity of Warwick, Gibbet Hill Road, CoVentry CV4 7AL, U.K., and
School of Chemistry, UniVersity of Edinburgh, West Mains Road, Edinburgh EH9 3JJ, U.K.
Received October 22, 2008
Chlorido osmium(II) arene [(η6-biphenyl)OsII(X-pico)Cl] complexes containing X ) Br (1), OH (2), and Me (3) as
ortho, or X ) Cl (4), CO2H (5), and Me (6) as para substituents on the picolinate (pico) ring have been synthesized
and characterized. The X-ray crystal structures of 1 and 6 show typical “piano-stool” geometry with intermolecular
π-π stacking of the biphenyl outer rings of 6. At 288 K the hydrolysis rates follow the order 2 . 6 > 4 > 3 > 5
. 1 with half-lives ranging from minutes to 4.4 h illustrating the influence of both electronic and steric effects of
the substituents. The pKa values of the aqua adducts 3A, 4A, 5A, and 6A were all in the range of 6.3-6.6. The
para-substituted pico complexes 4-6 readily formed adducts with both 9-ethyl guanine (9EtG) and 9-ethyl adenine
(9EtA), but these were less favored for the ortho-substituted complexes 1 and 3 showing little reaction with 9EtG
and 9EtA, respectively. Density-functional theory calculations confirmed the observed preferences for nucleobase
binding for complex 1. In cytotoxicity assays with A2780, cisplatin-resistant A2780cis human ovarian, A549 human
lung, and HCT116 colon cancer cells, only complexes 4 (p-Cl) and 6 (p-Me) exhibited significant activity (IC50
values < 25 µM). Both of these complexes were as active as cisplatin in A2780 (ovarian) and HCT116 (colon) cell
lines, and even overcome cisplatin resistance in the A2780cis (ovarian) cell line. The inactivity of 5 is attributed to
the negative charge on its para carboxylate substituent. These data illustrate how the chemical reactivity and
cancer cell cytotoxicity of osmium arene complexes can be controlled and “fine-tuned” by the use of steric and
electronic effects of substituents on a chelating ligand to give osmium(II) arene complexes which are as active as
cisplatin but have a different mechanism of action.
Introduction
A lack of understanding of the aqueous chemistry of
organometallic complexes, in particular of RuII and OsII arene
complexes, under biologically relevant conditions provides
an obstacle in current attempts to design anticancer drugs.
Knowledge of the aqueous chemistry of these types of
complexes may eventually lead to the control of their
pharmacological properties, including cell uptake, distribu-
tion, DNA binding, metabolism, and toxic side-effects.
For the OsII arene ethylenediamine (en) complex, [(η6-
biphenyl)Os(en)Cl]+, hydrolysis occurs with a half-life of
6.4 h at 298 K, which is about 40 times slower than that
Currently, there is much interest in ruthenium(II) arene
complexes of the type [(η6-arene)Ru(X)(YZ)] (where YZ is
a bidentate chelating ligand, and X a good leaving group,
e.g. Cl-) which exhibit both in vitro and in vivo anticancer
activity, in some cases even comparable to that of cisplatin.1
Yet the pharmacological potential of the heavier congener
osmium has been little explored. Previous work in our
laboratory on OsII arene complexes, [(η6-arene)Os(YZ)X]n+,
has shown that their aqueous reactivity is highly dependent
on the nature of the chelating ligand YZ.2,3
(2) Peacock, A. F. A.; Habtemariam, A.; Fernandez, R.; Walland, V.;
Fabbiani, F. P. A.; Parsons, S.; Aird, R. E.; Jodrell, D. I.; Sadler, P. J.
J. Am. Chem. Soc. 2006, 128, 1739.
(3) (a) Peacock, A. F. A.; Parsons, S.; Sadler, P. J. J. Am. Chem. Soc.
2007, 129, 3348. (b) Krostrhunova, H.; Florian, J.; Novakova, O.;
Peacock, A. F. A.; Sadler, P. J.; Brabec, V. J. Med. Chem. 2008, 51,
3635.
* To whom correspondence should be addressed. E-mail: p.j.sadler@
warwick.ac.uk.
†
University of Warwick.
University of Edinburgh.
‡
(1) Aird, R. E.; Cummings, J.; Ritchie, A. A.; Muir, M.; Morris, R. E.;
Chen, H.; Sadler, P. J.; Jodrell, D. I. Br. J. Cancer 2002, 86, 1652.
10.1021/ic8020222 CCC: $40.75  2009 American Chemical Society
Inorganic Chemistry, Vol. 48, No. 4, 2009 1753
Published on Web 01/15/2009

van Rijt et al.
Chart 1. Osmium Arene Complexes Studied in This Work
(IC50 ) 6 µM). These studies demonstrate that the kinetics
and thermodynamics of these types of complexes are
significant for their biological activity and importantly, that
these factors can be controlled by appropriate ligand design.
Here we investigate the optimization of the biological activity
of osmium(II) arene complexes where YZ is a picolinate
derivative (Chart 1). This has been approached by making
systematic changes to their design by placing different
substituents in the ortho- and para-positions of the pyridine
ring of the picolinate chelating ligand. Solid-state structures,
cytotoxicity data, and investigations of their aqueous chem-
1
istry using H NMR spectroscopy, including their rates of
hydrolysis, acidity of the resulting aqua adducts, and nucleo-
base binding studies, are discussed. This work shows that
substituents on the picolinate backbone can have significant
effects on the aqueous chemistry in osmium(II) compounds
of the type [(η6-bip)Os(YZ)(Cl)] providing great scope for
design for this class of compounds.
Experimental Section
Materials. 1,4-Dihydrobiphenyl and the dimer, [(η6-bip)OsCl2]2
were prepared by previously reported procedures.2,4 9-Ethyl guanine
and 9-ethyl adenine were purchased from Sigma-Aldrich.
OsCl3 ·nH2O and 6-hydroxo picolinic acid (>97%) were purchased
from Alfa Aesar. 2-Cyano-4-methyl pyridine (98%) and thionyl
chloride were obtained from Riedel de Hae¨n and Fluka, respectively.
6-Bromo picolinic acid (98%), 6-methyl picolinic acid (95%), 2,4
pyridinedicarboxylic acid monohydrate (98%), dicyclohexylcarbo-
diimide, and all deuterated solvents were obtained from Aldrich.
Ethanol and methanol were distilled over magnesium/iodine prior
to use.
Preparation of Complexes. Complexes 1-6 were synthesized
from the dimeric precursor, [(η6-bip)OsCl2]2, using procedures
similar to those reported previously for other half-sandwich OsII
arene complexes.5,6 Complex 7 was synthesized in EtOH using
dicyclohexylcarbodiimide (DCC) as coupling reagent.
4-Chloro Picolinic Acid. A suspension of picolinic acid (0.79
g, 6.35 mmol) and sodium bromide (1.30 g, 12.7 mmol) in 10 mL
of thionyl chloride was refluxed mildly for 20 h. The initially dark
green mixture changed to a dark red color. Excess SOCl2 was
removed by rotary evaporation, and the orange residue was taken
up in about 15 mL of CH2Cl2 and was filtered through celite to
remove any insoluble material. The orange solution was cooled to
271 K, and 20 mL of H2O (doubly distilled) was added dropwise
while stirring vigorously, keeping the temperature between 271 and
275 K. The color of the solution changed to a lighter orange, and
a white precipitate formed. The mixture was further stirred at
ambient temperature for 20 h. The CH2Cl2 and H2O were removed
by rotary evaporation. The solid was recrystallized from a minimum
amount of EtOH to give a yield of 0.28 g (28%). 1H NMR (DMSO-
d6): δ ) 8.72 (1 H, d, J ) 4.9 Hz), 8.09 (1 H, d, J ) 1.6 Hz), 7.84
(1 H, dd, J ) 4.9, 1.6 Hz).
of the RuII analogue. This highlights the lower reactivity of
OsII.2 Interestingly, despite its slow hydrolysis rate, [(η6-
biphenyl)Os(en)Cl]+ still exhibits promising activity against
the human ovarian cancer A2780 cell line (IC50 ) 9 µM).3b
Changing the ligand from the neutral N,N-chelator en to the
anionic O,O-chelator acetylacetonate (acac) has a marked
effect on the extent and rate of hydrolysis. The hydrolysis
rate of [(η6-arene)Os(acac)Cl], is too fast to monitor by NMR
at 298 K. However, hydrolysis of the acac compounds is
complicated by the formation of the hydroxo bridged dimer,
[(η6-arene)Os(µ2-OH)3Os(η6-arene)]+, with loss of the acac
ligand. This hydroxo-bridged dimer is the only observed
species at the micromolar concentrations in solutions similar
to those used in biological cell culture tests.2 On account of
formation of these hydroxo-bridged adducts, compounds
containing O,O-chelators are generally inactive toward the
human ovarian (A2780) and human lung (A549) cancer cell
lines.
Intermediate behavior to that of the complexes containing
N,N- and O,O-chelators is observed in aqueous solution for
some complexes containing anionic N,O-chelators. Com-
plexes containing a pyridine derivative as the N-donor atom
hydrolyze at an intermediate rate, are stable in aqueous
solution at micromolar concentrations, and are active toward
both A549 and A2780 cell lines.3 Notably, complexes
containing picolinate (pico) as the N,O-chelate, display
promising activity toward the human ovarian cancer cell line
with IC50 ) 4.5 µM, a value similar to that of carboplatin
4-Methyl Picolinic Acid. A solution of 2-cyano-4-methyl
pyridine (0.15 g, 1.27 mmol) in about 10 mL of 6 M HCl was
refluxed for 24 h. During this time, the initially light yellow solution
(4) Stahl, S.; Werner, H. Organometallics 1990, 9, 1876.
(5) Morris, R. E.; Aird, R. E.; Murdoch, P. D.; Chen, H. M.; Cummings,
J.; Hughes, N. D.; Parsons, S.; Parkin, A.; Boyd, G.; Jodrell, D. I.;
Sadler, P. J. J. Med. Chem. 2001, 44, 3616.
(6) Fernandez, R.; Melchart, M.; Habtemariam, A.; Parsons, S.; Sadler,
P. L. Chem.sEur. J. 2004, 10, 5173.
1754 Inorganic Chemistry, Vol. 48, No. 4, 2009

Organometallic Osmium(II) Arene Anticancer Complexes
changed to a clear solution. The solution was taken to dryness,
and a white solid remained. The solid was recrystallized from a
minimal amount of distilled water to give a yield of 148 mg (85%).
1H NMR (DMSO-d6): δ ) 8.65 (1H, s), 8.06 (1H, s), 7.66 (1H, s),
2.05 (3H, s).
(534.03): C, 42.73; H, 3.40; N, 2.62%. Found: C, 42.57; H, 2.99;
N, 2.58%. H NMR (DMSO-d6): δ ) 8.94 (1H, d, J ) 5.66 Hz),
7.74 (1H, br), 7.63 (2H, m), 7.52 (1H, d, J ) 4.53 Hz), 7.45 (3H,
m), 6.73 (1H, d, J ) 5.29 Hz), 6.69 (1H, d, J ) 4.91 Hz), 6.40
(1H, m), 6.37 (2H, m). Crystals suitable for X-ray diffraction were
obtained by slow evaporation from CH2Cl2 as 6·CH2Cl2 at ambient
temperature.
1
[(η6-bip)Os(6-Br-pico)Cl] (1). A solution of [(η6-bip)OsCl2]2
(51.9 mg, 0.06 mmol) in dry and degassed MeOH (10 mL) was
refluxed under argon for 1 h before adding a solution of sodium
methoxide (2.2 mol equiv, 7.2 mg) and 6-bromo picolinic acid (2.2
mol equiv, 27.2 mg) in 5 mL of dry and degassed MeOH. The
resulting mixture was refluxed mildly for 3 h, filtered, and solvent
reduced on a rotary evaporator until precipitate began to form and
was left standing at 278 K. The yellow powder was recovered by
filtration and was air-dried to give a final yield of 45.6 mg (63%).
Anal. Calcd for C18H13BrClNO2Os (580.94): C, 37.22; H, 2.26; N,
2.41%. Found: C, 36.91; H, 2.12; N, 2.25%. 1H NMR (CDCl3): δ
) 8.09 (1H, d, J ) 7.0 Hz), 7.83 (1H, d, J ) 7.0 Hz), 7.70 (1H,
t, 8 Hz), 7.52 (2H, d, J ) 7.9 Hz), 7.38 (3H, m), 6.71 (1H, d, J )
5 Hz), 6.63 (1H, d, J ) 6 Hz), 6.46 (1H, t, J ) 5.29 Hz), 6.35 (1H,
t, J ) 6.0 Hz), 6.32 (1H, t, J ) 5.29 Hz). Crystals suitable for
X-ray diffraction were obtained by slow evaporation from CHCl3
as 1·CHCl3 at ambient temperature.
[(η6-bip)Os(4-CO2Et-pico)Cl] (7). A solution of [(η6-bip)OsCl2]2
(47.1 mg, 0.056 mmol) and dimethylaminopyridine (3 mg) in dry
and degassed EtOH was placed under nitrogen and was stirred for
10 min after which it was cooled down to 273 K before adding
dicyclohexylcarbodiimide (1 mol equiv, 19 mg). The reaction
mixture was stirred at 273 K for a further 15 min, after which it
was left to stir at ambient temperature for another 2.5 h. This was
then filtered, and the solvent was removed on the rotary evaporator.
About 3 mL of dry DCM was added to the brown solid, and this
was filtered to remove the white precipitated urea side product.
The DCM filtrate was dried on a rotary evaporator, and the brown
solid was washed with ether to remove the remaining dimethy-
laminopyridine which resulted in a brown crystalline solid. Yield
1
2.2 mg (3.4%). H NMR (MeOD-d4): δ ) 9.10 (1H, d, J ) 5.52
Hz), 8.40 (1H, d, J ) 1.76 Hz), 8.05 (1H, dd, J ) 3.76, 2.01 Hz),
7.64 (2H, m), 7.44 (3H, m), 6.70 (1H, d, J ) 5.02 Hz), 6.66 (1H,
d, J ) 5.77 Hz), 6.39 (2H, q, J ) 4.76 Hz), 6.34 (1H, m), 4.48
(2H, q, J ) 7.03 Hz), 1.2 (3H, t, J ) 7.03 Hz).
[(η6-bip)Os(6-OH-pico)Cl] (2). Synthesis as for 1 using [(η6-
bip)OsCl2]2 (51.7 mg, 0.06 mmol), sodium methoxide (2.2 equiv,
6.71 mg), and 6-hydroxo picolinic acid (2.2 equiv, 19 mg). Yield:
37.3 mg (60%). Calcd for C18H14ClNO3Os (517.99): C, 41.74; H,
Instrumentation. X-ray Crystallography. Diffraction data for
compound 1 were collected at 150 K using a Bruker Smart Apex
CCD diffractometer. Diffraction data for 6 was collected at 120 K
by the EPSRC National Crystallography Service (Southampton)
using a Bruker-Nonius APEX II CCD camera on kappa-goniostat.
Absorption corrections for all data sets were performed with the
multiscan procedure SADABS.7 The structure of 1 was solved by
direct methods (SIR92),8 and that of 6 by Patterson methods
(DIRDIF).9 Refinement was against F2 using all data (SHELX-
TL).10 H-atoms were placed in geometrically calculated positions.
In 6 a molecule of dichloromethane is disordered about a crystal-
lographic inversion center. The C-Cl bond distances were re-
strained to 1.73(1) Å during refinement. The programs Diamond
3.020,11 Mercury 1.4.1,12 and ORTEP 3213 were used for analysis
of data and production of graphics. X-ray crystallographic data for
compounds 1 and 6 have been deposited in the Cambridge
Crystallographic Data Centre under the accession numbers CCDC
706164 and 706165, respectively.
1
2.72; N, 2.70%. Found: C, 41.22; H, 2.62; N, 2.65%. H NMR
(DMSO-d6): δ )13.12 (1H, br), 7.86 (1H, t, J ) 8 Hz), 7.65 (2H,
d, J ) 7.2 Hz), 7.44 (3H, m), 7.33 (1H, d, J ) 7.0 Hz), 7.11 (1H,
d, J ) 8.31 Hz), 6.74 (1H, d, J ) 5.29 Hz), 6.66 (1H, d, J ) 5.28
Hz), 6.42 (2H, m), 6.38 (1H, t, J ) 4.53 Hz).
[(η6-bip)Os(6-Me-pico)Cl] (3). Synthesis as for 1 using [(η6-
bip)OsCl2]2 (52.2 mg, 0.06 mmol), sodium methoxide (2.2 mol
equiv, 7.0 mg), and 6-methyl picolinic acid (2.2 mol equiv, 18 mg).
Yield: 50 mg (77%) Anal. Calcd for C19H16ClNO2Os (517.05): C,
44.22; H, 3.13; N, 2.71%. Found: C, 43.89; H, 2.65; N, 2.73%. 1H
NMR (CDCl3): δ ) 7.95 (1H, d, J ) 7.93 Hz), 7.72 (1H, t, J )
7.56), 7.43 (3H, m), 7.36 (3H, m), 6.51 (1H, d, J ) 5.29 Hz), 6.47
(1H, d, J ) 5.67 Hz), 6.30 (1H, t, J ) 5.29 Hz), 6.27 (1H, t, J )
4.91 Hz), 6.22 (1H, t, J ) 5.29 Hz).
[(η6-bip)Os(4-Cl-pico)Cl] (4). Synthesis as for 1 using [(η6-
bip)OsCl2]2 (51.5 mg, 0.06 mmol), sodium methoxide (2.2 mol
equiv, 7.4 mg), and 4-chloro picolinic acid (2.2 mol equiv, 21.5
mg). Yield: 37.1 mg (54%) Anal. Calcd for C18H13Cl2NO2Os·H2O
(554.45): C, 38.99; H, 2.73; N, 2.53%. Found: C, 38.66; H, 1.95;
1
NMR Spectroscopy. H NMR spectra were acquired in 5 mm
NMR tubes at 298 K (unless stated otherwise) on either a Bruker
1
DMX 500 (1H ) 500.13 MHz) or an AVA 600 (1H ) 599.81 MHz)
N, 2.60%. H NMR (DMSO-d6): δ ) 9.11 (1H, d, J ) 5.67 Hz),
1
7.93 (1H, d, J ) 1.89 Hz), 7.92 (1H, dd, J ) 6.04, 2.26 Hz), 7.65
(2H, m), 7.47 (3H, m), 6.76 (1H, d, J ) 5.29 Hz), 6.72 (1H, d, J
) 5.66), 6.46 (1H, t, J ) 5.29 Hz), 6.43 (1H, t, J ) 5.29 Hz), 6.40
(1H, t, J ) 4.91 Hz).
spectrometer. H NMR chemical shifts were internally referenced
to (CHD2)(CD3)SO (2.50 ppm) for DMSO-d6, CHCl3 (7.26 ppm)
for chloroform-d1, and to 1,4-dioxane (3.75 ppm) for aqueous
solutions. For NMR spectra of aqueous solutions, water suppression
[(η6-bip)Os(4-CO2H-pico)Cl] (5). Synthesis as for 1 using [(η6-
bip)OsCl2]2 (53.6 mg, 0.06 mmol), sodium methoxide (2.2 mol
equiv, 7.3 mg), and 2,4 pyridinedicarboxylic acid (2.2 mol equiv,
23.3 mg). Yield: 29.7 mg (41%). Anal. Calcd for C19H14ClNO4Os
(546.00): C, 41.80; H, 2.58; N, 2.57%. Found: C, 41.19; H, 2.12;
N, 2.55%. 1H NMR (DMSO-d6): δ ) 9.31 (1H, m), 8.14 (1H, m),
8.01 (1H, m), 7.65 (2H, m), 7.47 (3H, m), 6.76 (1H, d, J ) 5.67
Hz), 6.72 (1H, d, J ) 6.04 Hz), 6.46 (1H, t, J ) 5.29 Hz), 6.44
(1H, t, J ) 5.29 Hz), 6.41 (1H, t, J ) 4.91 Hz).
(7) Sheldrick, G. M. SADABS, Version 2006-1; University of Go¨ttingen:
Go¨ttingen, Germany, 2006.
(8) Altomare, A.; Cascarano, G.; Giacovazzo, C.; Guagliardi, A. J. Appl.
Crystallogr. 1994, 27, 1045.
(9) Beurskens, P. T., Beurskens, G.; Bosman, W. P.; de Gelder, R.; Garcia-
Granda; S.; Gould, R. O.; Israel, R.; Smits, J. M. M. Crystallography
Laboratory; University of Nijmegen: The Netherlands, 1996.
(10) Sheldrick, G. M. SHELXL-97. Program for the refinement of crystal
structures; University of Go¨ttingen: Germany, 1997.
(11) CrystalImpact DIAMOND, Version 3.0, Visual crystal structure
information system.; Crystal Impact GbR: Bonn, Germany, 2004.
(12) Macrae, C. F.; Edgington, P. R.; McCabe, P.; Pidcock, E.; Shields,
G. P.; Taylor, R.; Towler, M.; van de Streek, J. J. Appl. Crystallogr.
2006, 39, 453.
[(η6-bip)Os(4-Me-pico)Cl] (6). Synthesis as for 1 using [(η6-
bip)OsCl2]2 (53.1 mg, 0.06 mmol), sodium methoxide (2.2 mol
equiv, 7.4 mg), and 4-methyl picolinic acid (2.2 mol equiv, 19.2
mg). Yield: 35.6 mg (57%) Anal. Calcd for C19H16ClNO2Os·H2O
(13) Farrugia, L. J. J. Appl. Crystallogr. 1997, 30, 565.
Inorganic Chemistry, Vol. 48, No. 4, 2009 1755

van Rijt et al.
14
was carried out using Shaka or presaturation methods. All data
processing was carried out using XWIN-NMR version 3.6 (Bruker
U.K. Ltd.).
pH* Measurements. pH* values (pH meter reading without
correction for effects of D on glass electrode) of NMR samples in
D2O were measured at about 298 K directly in the NMR tube,
before and after recording NMR spectra, using a Corning 240 pH
meter equipped with a micro combination electrode calibrated with
Aldrich buffer solutions of pH 4, 7, and 10.
Computation. Density-functional theory (DFT) calculations were
carried out by using the Amsterdam density functional (ADF)17
program (version 2007.01). The coordinates of complexes used for
the calculations were obtained directly from the X-ray crystal
structures. Modifications to the structures were performed in
Chemcraft (version 1.5). Geometries and energies were obtained
by using the Perdew-Wang gradient-corrected functional (GGA)
with scalar ZORA relativistic correction.18-22 The general numer-
ical intergration was 4.0. The frozen core approximation23 was
applied using triple-ꢀ polarization (TZP). Default convergence
criteria were applied for self-consistent fields (SCF) and geometry
optimization. The COSMO, as implemented in ADF program, was
used to simulate the aqueous environment with ε ) 78.4 and a
probe radius ) 1.9 Å. The atomic radii used were Os ) 1.958, F
) 1.425, Cl ) 1.725, Br ) 1.850, O ) 1.517, N ) 1.608, C )
1.700, and H ) 1.350 Å.
Elemental Analysis. Carbon, hydrogen, and nitrogen (CHN)
elemental analysis were carried out at The University of St. Andrews
in the School of Chemistry on a Carlo Erba CHNS analyzer.
Methods. Kinetics of Hydrolysis. The kinetics of hydrolysis
for complexes 1-6 were followed by 1H NMR at different
temperatures. For this, solutions of the complexes with a final
concentration of 0.8 mM in 5% MeOD-d4/95% D2O (v/v) were
prepared by dissolution of the complexes in MeOD-d4 followed
by rapid dilution using D2O with a pH* of about 2 (acidified with
Results
1
HNO3), so that the aqua ligand was not deprotonated. H NMR
Synthesis and Characterization. The osmium compounds
were synthesized via the Cl-bridged dimer, [(η6-bip)OsCl2]2,
where bip ) biphenyl. Compound 7, however, was obtained
in low yield from compound 5 as starting material in dry
ethanol with DCC as coupling reagent. The X-ray crystal
structures of compounds [(η6-bip)Os(6-Br-pico)Cl]·CHCl3
(1·CHCl3) and [(η6-bip)Os(4-Me-pico)Cl] ·CH2Cl2 (6·CH2Cl2)
were determined. Both structures adopt the familiar pseudo-
octahedral “three-leg piano stool” geometry (Figure 1A,B).
The osmium is π-bonded to the arene ligand (“the seat”)
with a centroid Os-ring distance of 1.654 Å for 1 and 1.660
Å for 6, σ-bonded to the chloride (2.4065(15) Å (1),
2.4095(19) Å (6)) and σ-bonded to the chelating picolinate
ligand (pyridine nitrogen-Os; 2.154(5) Å (1), 2.088(5) Å (6),
oxygen-Os 2.077(4) Å (1), 2.082(4) Å (6)), which constitute
the three legs of the piano stool. Crystallographic data,
selected bond lengths, and angles are given in Tables 1 and
2. Complex 1 has a propeller twist of the biphenyl ligand of
about 38°, while for complex 6 a propeller twist of 42° is
observed. There is an intermolecular interaction between the
outer rings of the biphenyl arene of adjacent molecules in
compound 6, with a centroid-centroid distance of 3.7 Å
(Figure 2). This value indicates intermolecular ring stack-
ing.24
spectra were taken after various intervals using the presaturation
method for water surpression. The rate of hydrolysis was determined
by fitting plots of concentrations (determined from H NMR peak
integrals) versus time to a pseudo first-order equation using ORIGIN
version 5.0 (Microcal Software Ltd.). The Arrhenius activation
energy (Ea), activation enthalpies (∆Hq), and activation entropy
(∆Sq) for compound 6 were determined from the slopes of the
Arrhenius and intercepts of Eyring plots.
1
pKa* Calculations. For determinations of pKa* values (pKa
values determined from solutions in D2O), the pH* values of the
aqua complexes of 3-6 in D2O (formed in situ by dissolution of
the parent chloro complexes) were varied from about pH* 2 to 10
1
by the addition of dilute NaOD and HNO3, and H NMR spectra
were recorded. The chemical shifts of the arene ring protons
were plotted against pH*. The pH* titration curves were fitted
to the Henderson-Hasselbalch equation, with the assumption
that the observed chemical shifts are weighted averages according
to the populations of the protonated and deprotonated species.
These pKa* values can be converted to pKa values by use of the
equation pKa ) 0.929pKa* + 0.42 as suggested by Krezel and
Bal15 for comparison with related values in the literature.
Interactions with Nucleobases. The reaction of 1-6 with
nucleobases typically involved addition of a solution containing 1
mol equiv of nucleobase in D2O, to an equilibrium solution of 1-6
in D2O (>90% aqua complex). The pH* value of the sample was
adjusted if necessary so as to remain close to 7.4 (physiological).
1H NMR spectra of these solutions were recorded at 298 K after
various time intervals.
Cancer Cell Cytotoxicity. After plating, human ovarian A2780
and cisplatin resistant A2780cis cancer cells were treated with OsII
complexes on day 3, and human lung A549 and human colon
HCT116 cancer cells on day 2, at concentrations ranging from 0.5
µM to 100 µM. Solutions of the OsII complexes were made up in
0.125% (v/v) dimethylsulfoxide (DMSO) to assist dissolution
(0.03% final concentration of DMSO per well in the 96-well plate).
Cells were exposed to the complexes for 24 h, washed, supplied
with fresh medium, allowed to grow for three doubling times (72
h), and then the protein content measured (proportional to cell
survival) using the sulforhodamine B (SRB) assay.16
Kinetics of Hydrolysis. The rates of hydrolysis of
compounds 1-6 in a 5% MeOD-d4/95% D2O were moni-
1
tored by H NMR at 288 and 298 K by the observation of
new peaks over time due to aqua adduct formation. Five
percent MeOD was used to improve solubility, and acidic
(17) Te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra,
C.; Van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput.
Chem. 2001, 22, 931.
(18) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1993,
99, 4597.
(19) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1994,
101, 9783.
(20) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1996,
105, 2373.
(21) van Lenthe, E.; Ehlers, A.; Baerends, E.-J. J. Chem. Phys. 1999, 110,
8943.
(14) Hwang, T. L.; Shaka, A. J. J. Magn. Reson. Ser. A 1995, 112, 275.
(15) Krezel, A.; Bal, W. J. Inorg. Biochem. 2004, 98, 161.
(16) Skehan, P.; Storeng, R.; Scudiero, D.; Monks, A.; McMahon, J.;
Vistica, D.; Warren, J. T.; Bokesch, H.; Kenney, S.; Boyd, M. R.
J. Natl. Cancer Inst. 1990, 82, 1107.
(22) van Lenthe, E.; van Leeuwen, R.; Baerends, E. J.; Snijders, J. G. Int.
J. Quantum Chem. 1996, 57, 281.
(23) Baerends, E. J.; Ellis, D. E.; Ros, P. Theor. Chim. Acta 1972, 27,
339.
(24) Janiak, C. J. Chem. Soc., Dalton Trans. 2000, 3885.
1756 Inorganic Chemistry, Vol. 48, No. 4, 2009

Organometallic Osmium(II) Arene Anticancer Complexes
Figure 1. X-ray crystal structure and atom numbering scheme for (A) [(η6-bip)Os(6-Br-pico)Cl] ·CHCl3 (1·CHCl3), and (B) [(η6-bip)Os(4-Me-pico)Cl] ·CH2Cl2
(6·CH2Cl2). H atoms and solvent molecules have been omitted for clarity.
Table 1. Crystallographic Data for [(η6-bip)Os(2-Br-pico)Cl]·CHCl3
(1·CHCl3) and [(η6-bip)Os(4-Me-pico)Cl]·CH2Cl2 (6·CH2Cl2)
1·CHCl3
6·CH2Cl2
formula
C19 H14 Br Cl4 N O2 Os C19 H16 Cl N O2 Os
molecular weight 700.22
558.44
crystal description colorless block
colorless lath
size
λ (Å)
T/K
0.34 × 0.17 × 0.14 mm 0.16 × 0.06 × 0.02 mm
0.71073
150(2)
0.71073
120(2)
crystal system
space group
a (Å)
b (Å)
c (Å)
R (deg)
ꢁ (deg)
γ (deg)
volume (Å3)
Z
triclinic
P1
triclinic
P1
j
j
Figure 2. X-ray structure of [(η6-bip)Os(4-Me-pico)Cl]·CH2Cl2 (6·CH2Cl2)
6.8555(8)
11.0638(11)
13.9209(19) A
99.783(7)
98.387(8)
91.757(7)
1027.7(2)
2
8.065(5)
10.495(5)
11.672(5)
81.806(5)
78.860(5)
77.776(5)
942.0(8)
2
showing intermolecular ring stacking of the extended biphenyl ring.
energy (Ea), activation enthalpy (∆Hq), and activation entropy
(∆Sq) of compound 6 (Figure 3B) are listed in Table 4. The
large negative ∆Sq value for compound 6 is notable. The
extent of hydrolysis at equilibrium is high for all compounds,
with even 100% hydrolysis observed for compounds 4 and
5 and between 88% and 96% for 1, 3, and 6 (Figure 3A,B).
The hydrolysis rate of compound 1 (6-Br pico) was too slow
to monitor at 288 K, but its half-life was successfully
determined by increasing the temperature to 298 K (Table
3, Figure 3A). Compound 6 hydrolyzes the fastest with a
half-life of just under 1 h (288 K). Complexes 1, 3-5
hydrolyzed with half-lives ranging from 2.4 to 4.4 h at 288
K (Table 3).
R
Rw
GOF
0.0376
0.0868
0.997
0.0399
0.0837
1.120
Table 2. Selected Bond Lengths (Å) and Angles (deg) for
[(η6-bip)Os(2-Br-pico)Cl]·CHCl3 (1·CHCl3) and
[(η6-bip)Os(4-Me-pico)Cl]·CH2Cl2 (6·CH2Cl2)
bond length/ angle
compound 1·CHCl3
compound 6·CH2Cl2
Os-C(arene)
2.178(7)
2.207(7)
2.175(6)
2.177(6)
2.159(6)
2.208(6)
2.077(4)
2.154(5)
2.4065(15)
77.35(18)
84.82(12)
81.74(13)
2.164(6)
2.208(7)
2.193(6)
2.199(6)
2.173(6)
2.176(6)
2.082(4)
2.088(5)
2.4095(19)
77.10(18)
84.50(14)
83.30(15)
The effects of chloride concentrations typical of blood
plasma (100 mM), cell cytoplasm (22.7 mM), and cell
nucleus (4 mM) on the aqueous chemistry of 3 and 6 were
Os-O
Os-N
1
investigated. H NMR spectra of 3 or 6 (1 mM) were
Os-Cl
recorded within 10 min of sample preparation and after
incubation at 310 K for 24 h (Supporting Information, Figure
S2). On the basis of 1H NMR peak integrals, 35% of
hydrolyzed complex 3 (3A) and 10% of hydrolyzed complex
6 (6A) were found to be present in 100 mM [Cl] (pH* 6.22),
60% of 3A, and 53% of 6A at 22.7 mM [Cl] (pH* 6.50)
and 60% of 3A and 82% of 6A at 4 mM [Cl] (pH* 6.75)
after 10 min. For complex 6, no significant change was
observed after 24 h, while compound 3 had not reached
equilibrium after 24 h, see Table 5 and Supporting Informa-
tion, Figure S2.
O-Os-N
O-Os-Cl
N-Os-Cl
conditions (D2O, pH* 2) were used to prevent the deproto-
nation of the aqua complex as a secondary reaction. The
hydrolysis of compound 2 containing 6-hydroxy picolinate
1
was too rapid to be observed by H NMR at 288 K. The
percentage of aqua peak formation for 1, 3-6 was plotted
against time and was fitted to pseudofirst order kinetics
(Figure 3, Supporting Information, Figure S1), and their half-
life times were calculated (Table 3). The Arrhenius activation
Inorganic Chemistry, Vol. 48, No. 4, 2009 1757

van Rijt et al.
pH. The chemical shifts were plotted against pH*, Supporting
Information, Figure S4, and the resulting pH titration curves
were fitted to the modified Henderson-Hasselbalch equa-
tion.25,26 This gave rise to pKa* values between 6.30 and
6.60 (Table 6). For compound 5, the pKa* value of the para-
substituent carboxylic acid proton was determined to be 2.50.
Reactions with Nucleobase Models, 9-Ethyl Guanine
(9EtG) and 9-Ethyl Adenine (9EtA). Since DNA is often
believed to be the main target for transition metal antican-
cer drugs,27 nucleobase binding reactions of compounds
1, [(η6-bip)Os(6-Br-pico)Cl], 2, [(η6-bip)Os(6-OH-pico)Cl],
3, [(η6-bip)Os(6-Me-pico)Cl], 4, [(η6-bip)Os(4-Cl-pico)Cl],
-
5, [(η6-bip)Os(4-CO2 -pico)Cl], and 6, [(η6-bip)Os(4-Me-
pico)Cl], with nucleobase models 9-ethyl guanine (9EtG) and
9-ethyl adenine (9EtA), were investigated. Solutions of 1-6
(1 mM) (containing an equilibrium mixture of 1-6 and their
respective aqua adducts 1A-6A as the major species) with
1 mol equivalent of 9EtG or 9EtA in D2O were prepared,
and 1H NMR spectra were recorded at different time
intervals. The percentages of nucleobase adducts formed by
all compounds based on 1H NMR peak integrals are
displayed in Table 7A,B and Figure 4. The addition of 1
mol equiv 9EtG or 1 mol equiv 9EtA to aqueous solutions
of all compounds resulted in no new peaks after about 5
min. However, after incubation at 310 K for 24 h, new peaks
for all compounds were observed.
Figure 3. Time dependence for formation of the aqua complexes 1A, 3A,
4A, 5A, and 6A (based on 1H NMR peak integrals) during hydrolysis of 1,
3, 4, and 5 in acidic D2O (pH* 2) at (A) 298 K for 1A (9), 3A (b), 4A
(2), and 5A (1) and (B) at 298 K (1), 288 K (2), 285 K (9), and 278 K
(b) for 6A. The inset shows the Arrhenius plot, the slope of which gives
the Arrhenius activation energy Ea of 90.6 kJ mol-1
.
Compounds 4-6 formed 9EtG and 9EtA adducts to the
extent of 60%-100% completion after 24 h. Only compound
6 showed increased binding to 9EtG, with 20% more binding
compared to 9EtA. Compound 5 showed an exceptionally
high nucleobase affinity with 100% nucleobase adduct
formation with both 9EtG and 9EtA after 24 h. The ortho-
substituted pico compounds 1-3 were less reactive toward
the nucleobases compared to the para-substituted com-
pounds 4-6 (Table 7, Figure 4). Compound 3 (o-Me)
showed a preference for 9EtG, while compounds 1 (o-Br)
and 2 (o-OH) formed no adducts with 9EtG, but did form
9EtA adducts. In contrast to compounds 3-6, compounds 1
and 2 formed two adenine nucleobase adducts, most likely
through osmium binding to the N1 or N7 of adenine forming
9EtA-1/2a and b, Chart 1. It was not possible to determine
the precise adduct ratio for 9EtA-1 adducts because of peak
overlap. Compound 2 formed 9EtA adducts in a 14% and
22% ratio.
Table 3. Hydrolysis Data for Compounds 1-6 at 288 K/ 298 K,
1
Determined by H NMR
compound
T/ K
k/ h-1
t1/2/ h
1
2
3
298
288
288
298
288
298
288
298
288
298
0.345 ( 0.0002
g5.9
2.01 ( 0.001
e0.12
0.219 ( 0.005
0.676 ( 0.039
0.288 ( 0.001
0.656 ( 0.032
0.158 ( 0.011
0.384 ( 0.004
0.710 ( 0.012
2.32 ( 0.08
3.17 ( 0.07
1.03 ( 0.04
2.40 ( 0.08
1.06 ( 0.05
4.42 ( 0.32
1.81 ( 0.02
0.98 ( 0.02
0.30 ( 0.01
4
5
6
pKa* Determination. The changes in the 1H NMR
chemical shifts for the protons of the coordinated phenyl ring
in compounds 3-6, present in equilibrium solutions of 3-6
with their aqua adducts as the major species (3A-6A), were
followed with change in pH* over a range of 2-10
(Supporting Information, Figure S3A). When the pH* values
of the solutions were increased from about 2 to 10, the major
set of NMR peaks assigned to 3A, 4A, and 6A gradually
shifted to high field in the spectrum. In contrast, the signals
assignable to parent chloro species 3, 4, and 6 remained
unchanged with changing pH*, but their intensities increased
at lower pH*. The protonation of the para-carboxylate on
the picolinate ring of complex 5 caused the NMR peaks
assigned to both species (5 and 5A) to shift to high field at
low pH* (Supporting Information, Figure S3B), while at
higher pH* only the peaks assignable to 5A shifted to higher
field.
Cytotoxicity Data. The cytotoxicity of complexes 1-7
toward colon HCT116, human lung A549, ovarian A2780,
and ovarian cisplatin-resistant A2780cis cancer cell lines was
investigated (Table 8). Complexes 1, 2, 3, and 5 were non-
toxic up to the highest test concentration of 50 µM in all
four cell lines. The IC50 values (concentration at which 50%
of the cell growth is inhibited) of these complexes are
(25) Lee, S. A.; Eyeson, R.; Cheever, M. L.; Geng, J. M.; Verkhusha, V. V.;
Burd, C.; Overduin, M.; Kutateladze, T. G. Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 13052.
(26) Sigel, R. K. O.; Sabat, M.; Freisinger, E.; Mower, A.; Lippert, B.
Inorg. Chem. 1999, 38, 1481.
(27) Zhang, C. X.; Lippard, S. J. Curr. Opin. Chem. Biol. 2003, 7, 481.
(28) Peacock, A. F. A.; Habtemariam, A.; Moggach, S. A.; Prescimone,
A.; Parsons, S.; Sadler, P. J. Inorg. Chem. 2007, 46, 4049.
Because of the chirality at the osmium metal center, none
of the protons of the coordinated arene are equivalent, and
1
each gave rise to separate H NMR peaks at high and low
1758 Inorganic Chemistry, Vol. 48, No. 4, 2009

Organometallic Osmium(II) Arene Anticancer Complexes
Table 4. Rate Data and Activation Parameters for the Aquation of Active Complex 6 at Various Temperatures
compound
T/ K
k/ h-1
t1/2/ h
Ea /kJ mol-1
∆Hq /kJ mol-1
∆Sq/J K-1 mol-1
-55.6(1.6
6
278
285
288
298
0.17
0.44
0.71
2.32
4.06
1.57
0.98
0.30
90.6(1.0
87.7(2.1
Table 5. Percentage of Aqua Adduct Formation in a Solution of 1 mM
3 or 6 in D2O at Chloride Levels Typical of Blood Plasma (100 mM),
Cell Cytoplasm (22.7 mM), and Cell Nucleus (4 mM)
Table 8. In Vitro Growth Inhibition of A2780, A2780cis, A549, and
HCT116 Cell Lines for Compounds 1-7 and Cisplatin (CDDP) as
Control
% aqua adduct
A2780
A2780cis
A549
HCT116
compound
IC50 (µM)
IC50 (µM)
IC50 (µM)
IC50 (µM)
t ) 10 min
t ) 24 h
1
2
3
4
5
6
7
>50
>50
>50
4.8
>50
4.4
>50
>50
>50
5.45
>50
7.61
61
>50
>50
>50
24
>50
25.1
NDa
7.7
>50
>50
>50
2.8
[NaCl]
3A
6A
3A
6A
100 mM
22.7 mM
4 mM
35
60
60
10
53
82
27
49
70
20
53
81
>50
2.5
Table 6. pKa* and pKa Valuesa for the Deprotonation of the
Coordinated H2O in Complexes 3A-6A
44.4
3
NDa
2.7
CDDP
16.9
a ND ) not determined.
complex
pKa*
pKa
[(η6-bip)Os(6-Me-pico)OD2]+
[(η6-bip)Os(4-Cl-pico)OD2]+
[(η6-bip)Os(4-CO2H-pico)OD2]+
[(η6-bip)Os(4-Me-pico)OD2]+
3A
4A
5A
6A
6.34
6.30
6.61
6.42
6.31
6.27
6.56
6.38
6 overcome cisplatin resistance in the A2780cis cell line.
Compound 7, the ethyl ester of compound 5, shows moderate
activity in the human ovarian A2780 and A2780cis cell lines
(Table 8).
a pKa values calculated from pKa* according to Krezel and Bal.15
Table 7. Extent of 9EtG and 9EtA Adduct Formation for Compounds
1-6 at Different Time Intervals
Computation. Using computational methods we attempted
to gain insight into the steric effects caused by the ortho
substituents upon nucleobase binding and their effect on the
rate of hydrolysis. The minimum energy structures of
nucleobase adducts of compounds 1 [(η6-bip)Os(6-Br-pico)-
(9EtG/9EtA)]+, 3 [(η6-bip)Os(6-Me-pico)(9EtG/9EtA)]+, and
6 [(η6-bip)Os(4-Me-pico)(9EtG/9EtA)]+ obtained from DFT
calculations are shown in Supporting Information, Figure S5.
The 9EtG adduct of p-Me complex 6 is 6.6 kcal/mol lower
in energy, and the 9EtA adduct is 5.8 kcal/mol lower in
energy compared to similar adducts of its isomer the o-Me
complex 3.
Geometry optimizations for the nucleobase adducts of
complex 1 revealed H-bonding between the NH2 of adenine
and the caboxylate oxygen of the picolinate ligand (NH · · ·OC,
1.833 Å). In contrast, a close contact between guanine C6O
and the ortho-bromide substituent (3.260 Å) was observed
in the guanine adduct (Supporting Information, Figure S5).
compound
t ) 10 min
t ) 24 h
t ) 72 h
(A) % 9EtG Nucleobase Adduct
1
2
3
4
5
6
0
0
0
0
0
0
0
0
30
59
100
80
0
0
30
76
100
85
(B) % 9EtA Nucleobase Adduct
1
2
3
4
5
6
0
0
0
0
0
0
∼30a
14:20b
9
∼35a
14:20b
9
71
100
60
62
100
60
a Approximate value due to peak overlap. b Formation of two nucleobase
adducts (9EtA-2a and b, Chart 1).
Discussion
Mixed N,O-chelating ligands can be used to “fine-tune”
the aqueous reactivity of OsII arene compounds by the choice
of the types of N- and O-donor groups. The aim of this study
was to optimize the biological activity of osmium(II) arene
complexes by making systematic changes to their design
using different substituents in the ortho- and para-positions
of the pyridine ring of the picolinate chelating ligand.
Complexes 1-7 were synthesized and X-ray structures of
compound 1 [(η6-bip)Os(6-Br-pico)Cl], as a CHCl3 solvate, and
6 [(η6-bip)Os(4-Me-pico)Cl], as a CH2Cl2 solvate, were deter-
mined. Both compounds adopt the familiar pseudo-octahedral
“three-leg piano stool” geometry as found previously for other
osmium(II) biphenyl compounds (i.e., [(η6-bip)Os(YZ)(Cl)]),
for which a propeller twist of the biphenyl ligand of about 40°
and Os-Cl bond lengths of 2.40 Å are observed.2
Figure 4. Bar chart showing the extent of binding of compounds 1-6 to
the nucleobases 9EtG (dark gray) and 9EtA (light gray) after 48 h.
therefore likely to be >100 µM and then can be described
as inactive compounds. Compounds 4 (p-Cl) and 6 (p-Me),
however, exhibited promising toxicity with IC50 values
similar to those found for cisplatin in the human ovarian
A2780 and colon HCT116 cancer cell lines and moderate
cytotoxicity in the A549 cancer cell line (ca. 3 times less
active than cisplatin). Interestingly, both compounds 4 and
Inorganic Chemistry, Vol. 48, No. 4, 2009 1759

van Rijt et al.
The π-π stacking by the flexible extended biphenyl arene
ring system observed in the X-ray structure of compound 6
(Figure 2) could allow intercalation into DNA to contribute
to the cytotoxicity of this class of compounds.
and at the lowest chloride concentration of 4 mM close to
that of the cell nucleus, complex 6 is 81% present as the
reactive aqua species. These data indicate that in the cell
nucleus, complex 6 might be selectively activated through
hydrolysis as a mode of activation toward DNA binding,
while outside the cell and in particular in blood plasma
complex 6 is predominately present as its less reactive intact
chloro species.
For compound 3, the different chloride concentrations had
little effect on the concentration of hydrolyzed complex 3
(3A) after 10 min, although after equilibration for 24 h the
extent of hydrolysis was similar to that of compound 6. This
can be explained by the steric bulk of the ortho-methyl
substituent in complex 3 slowing down hydrolysis. It is
important to note that depending on the pH of the cell and
the pKa of the aqua adduct, the complex can be present in
either the less reactive hydroxo or in its reactive aqua form
(Table 5, Supporting Information, Figure S2).
Aqueous Solution Chemistry. Previous studies on the
hydrolysis rates of OsII arene compounds of the type [(η6-
arene)Os(YZ)Cl]n+ have shown that the aqueous reactivity
of these complexes is highly dependent on the nature of the
chelating ligand YZ.2,3,28 Compounds 1-6 (Chart 1) hydro-
lyze relatively fast with half-lives ranging from 0.98 to 4.4 h
at 288 K, apart from the o-OH compound 2, [(η6-bip)Os(6-
OH-pico)Cl], which hydrolyzes too fast to be monitored by
1H NMR at 288 K (Table 3). In addition, the equilibrium
for hydrolysis lies to a great extent toward the aqua adducts
for all compounds with even 100% hydrolysis observed for
compounds 2, 4, and 5, and between 88% and 96% for 1, 3,
and 6. The rapid hydrolysis of compound 2 might be aided
by the hydrogen bonding ability of the ortho-hydroxo
substituent which can stabilize the incoming water molecule
in the transition state, and in addition, the electron-donating
properties of the ortho-hydroxo substituent can contribute
to the weakening of the Os-Cl bond.
Compound 6 also hydrolyzes rapidly with a half-life of
just under 1 h (288 K), indicating an electronic substituent
effect of the para-methyl group. The electron-donating
properties of the methyl group give rise to increased electron
density around osmium which may facilitate chloride loss
and therefore result in a faster hydrolysis rate. This substitu-
ent effect is also observed for compound 4 which contains
the electron-withdrawing (through the inductive effect) para-
chloride group and for compound 5 with the electron-
withdrawing (through the resonance effect) carboxyl group
also in the para-position on the pyridine ring, with hydrolysis
half-lives of 2.4 and 4.4 h (288 K) for 4 and 5, respectively.
A similar trend is observed for compounds 1 and 3 which
contain bromide (1) and methyl (3) substituents in the ortho-
position of the pyridine ring. Compound 3, with an electron-
donating methyl group, hydrolyzes twice as fast as compound
1, which contains an electron-withdrawing bromide group.
However, important to note is the slower hydrolysis for
compound 3 compared to compound 6 (Table 3), both of
which contain a methyl substituent but in different positions
on the ring (i.e., ortho for 3 and para for 6). This can be
explained by the steric bulk of the ortho-methyl group around
the metal center, which may force a less favorable dissocia-
tive pathway. A more favorable associative pathway is
allowed in the transition state of compound 6 since here the
methyl group does not interfere with the reactive site. The
large negative activation entropy of hydrolysis (-55.6 J K-1
mol-1, Table 4, Figure 3B) for compound 6, suggests that
the mechanism of hydrolysis for this compound does indeed
involve an associative pathway.
Compounds 3A, [(η6-bip)Os(6-Me-pico)D2O]+, 4A, [(η6-
bip)Os(4-Cl-pico)D2O]+, 5A, [(η6-bip)Os(4-CO2H-pico)-
D2O]+, and 6A, [(η6-bip)Os(4-Me-pico)D2O]+, have similar
pKa* values, ranging from 6.30 to 6.61 (Table 6). This shows
that substituents on the picolinate chelating ligand have little
effect on the acidity of the bound water. These values are
significantly lower (ca. 1.5 units) than those for structurally
similar RuII arene aqua complexes.2,6 The pKa* values of
3A-6A (<7, Table 6) indicate that at physiological pH (7.4)
almost all of 3A-6A would be present in their less reactive
hydroxo forms. Furthermore, the carboxylic acid substituent
of compound 5 (pKa* 2.5), would be present in its depro-
tonated form at physiological pH giving an overall negative
charge on 5 (i.e., [(η6-bip)Os(4-CO2-pico)Cl]-).
Interaction with Nucleobase Derivatives; 9-Ethyl
Guanine (9EtG) and 9-Ethyl Adenine (9EtA). Nuclear
DNA is believed to be a major target for transition metal-
based anticancer complexes.27 Osmium arene complexes,
[(η6-arene)Os(XY)Cl]n+, containing a neutral N,N-chelate
(e.g., ethylenediamine) bind selectively to G nucleobases,
and those containing an anionic O,O-chelate (e.g., acetylac-
etonate) have similar affinities for both G and A nucleo-
bases.2
In this study, the interactions with nucleobase derivatives,
9-ethyl guanine (9EtG) and 9-ethyl adenine (9EtA) with
compounds 1-6 in D2O were investigated (Table 7, Figure
4). Compounds 4-6 bind significantly (60-100%) to both
nucleobases. Compound 6 appears to bind more weakly to
9EtA which indicates a slightly weaker binding of 6 to
adenine compared to guanine. Compound 5 shows excep-
tionally high nucleobase affinity with 100% nucleobase
adduct formation for both 9EtG and 9EtA after 24 h.
Compound 4 reacted more slowly with both nucleobases with
equilibration reached after 72 h, while the other compounds
reacted with the nucleobases within 24 h. (Table 7).
Compounds 1-3 containing ortho-substituents, reacted
significantly less with the model nucleobases (Figure 4),
indicating weaker binding to G and A bases for these
complexes. This weaker binding is most likely due to the
steric hindrance caused by the ortho-substituents on the
At high chloride concentrations typical of blood plasma
(100 mM), complex 6 is 80% present as the intact chloro
species, also referred to as the “prodrug” form since in this
form it is relatively unreactive. At a lower chloride concen-
tration resembling that in the cell cytoplasm (22.7 mM), 53%
of complex 6 is present in its active hydrolyzed form (6A)
1760 Inorganic Chemistry, Vol. 48, No. 4, 2009

Organometallic Osmium(II) Arene Anticancer Complexes
picolinate ring. This is more evident for 3 compared to
compound 6, which both contain a methyl substituent
displaying a similar electronic effect but with only 3 (ortho-
Me) causing steric hindrance at Os. Compound 3 is
significantly less reactive toward these nucleobases com-
pared to compound 6 with about 50% less binding
observed to both 9-ethyl guanine and 9-ethyl adenine after
24 h. ADF calculations confirm that the minimum energies
of geometry-optimized structures of the nucleobase ad-
ducts of 6 are significantly lower than those found for its
isomer complex 3; 9EtG-6 was found to be 6.6 kcal/mol
more stable than 9EtG-3, and 9EtA-6 was 5.8 kcal/mol
more stable than 9EtA-3, indicating that the nucleobase
adducts of complex 6 are thermodynamically more stable
than those of complex 3.
In contrast to compounds 3-6, compound 1 formed no
adduct with 9EtG but does, however, form adducts with
9EtA. Minimum energy structures of the 9EtA adduct with
complex 1, ([(η6-bip)Os(6-Br-pico)(9EtA)], Supporting In-
formation, Figure S5B), obtained from DFT calculations do
indeed show hydrogen bonding between the NH2 group of
adenine and the carboxylate of the picolinate ligand, stabiliz-
ing this adduct. In contrast, the minimum energy structure
of the 9EtG adduct, ([(η6-bip)Os(6-Br-pico)(9EtG)]), shows
a close contact between the bromide substituent and the
guanine C6O which might explain the lack of reactivity of
complex 1 with 9EtG (Supporting Information, Figure S5A).
This argument can also be used to explain the same observed
nucleobase specificity of complex 2. Also, both compounds
1 and 2 form two adducts with 9EtA, a behavior which is
not observed for any of the other osmium complexes 3-6
(Table 7).
Cytotoxicity. Compounds 1-3 and 5 were non-toxic
toward colon HCT116, human lung A549, human ovarian
A2780, and cisplatin resistant A2780cis cancer cell lines.
Compounds 4 and 6, however, exhibited promising toxicity,
with IC50 values similar to those for cisplatin in the human
ovarian A2780 and colon HCT116 cancer cell lines, and
moderate cytotoxicity in the A549 cancer cell line with IC50
values about three times that of cisplatin. Interestingly, both
compounds 4 and 6 overcome cisplatin resistance in the
A2780cis cell line indicating that the mechanism of action
of these osmium arene complexes is different from that of
cisplatin (Table 8).
The steric bulk present around the metal center caused by
the ortho-substituents in the picolinate compounds 1-3
accompanied by their reluctance to form nucleobase adducts
may account for their cytotoxic inactivity. This is further
supported by their reduced 9EtG and 9EtA binding, espe-
cially comparing complex 3 with complex 6 and the slower
hydrolysis rates observed for compounds 1 and 3. The
cytotoxic inactivity of compound 5 might arise from poor
cellular uptake as a result of the deprotonation of its para-
substituent carboxyl group (pKa* 2.5, Table 6), and overall
negative charge. This hypothesis is supported by the observed
activity of compound 7 which has an ester group in the para
position in the human ovarian A2780 and A2780cis cancer
cell lines (Table 8).
Conclusions
In this study, we have shown how steric and electronic
effects of substituents on the ortho- and para-position of
picolinate chelated to [(η6-arene)OsIICl]+ fragments can be
used to control hydrolysis rates, extent of nucleobase (G and
A) binding, and cancer cell cytotoxicity. Faster hydrolysis
is observed with an electron-donating methyl group (com-
pound 6) and slower hydrolysis for the electron-withdrawing
chloro and carboxyl groups (compounds 4 and 5, respec-
tively, see Table 3) in the para-position.
Of the 6 compounds tested, compounds 4 and 6 show
promising activity in the human ovarian A2780 and human
colon HCT116 cancer cell lines with similar IC50 values to
those of cisplatin (see Table 8) and even overcome cisplatin
resistance in the A2780cis cell line, indicating that the
mechanism of action of these osmium arene complexes is
different from that of cisplatin.
The cytotoxic inactivity of compounds 1-3 may be caused
by the presence of sterically demanding ortho-substituents
on the pyridine ring which hinder attack on the metal center.
The effect of the steric hindrance is further demonstrated
by their relatively slow hydrolysis rates and their reduced
and weaker binding to guanine and adenine especially for
complex 3 in comparison to complex 6 (Table 7, Figure 4),
which differ only by having the methyl substituent at the
ortho- and para-positions of the picolinate ring, respectively.
The cytotoxic inactivity of compound 5 might be caused
by the deprotonation of its para-substituent carboxyl group
(pKa* 2.5), resulting in an overall negative charge (in its
chloro form) in aqueous conditions hindering cellular uptake.
Also, even if complex 5 is taken up into the cell and activated
by hydrolysis to its neutral aqua form, the lack of a positive
charge might hinder its attack on DNA (which is negatively
charged). To investigate the influence of charge, the ethyl
ester, compound 7, was synthesized, and its cytotoxicity was
determined in the A2780 and A2780cis cell lines. Its
moderate cytotoxic activity (Table 8) suggests that the overall
negative charge in complex 5 does indeed contribute to its
inactivity.
This work shows that substituents on the picolinate
backbone can have significant effects on the aqueous
chemistry of osmium(II) compounds of the type [(η6-
bip)Os(YZ)(Cl)] allowing a great scope for design for this
class of compounds. The synthetic route for compound 7
illustrates as well a new way of functionalizing osmium arene
compounds which could be applied to a wide range of other
osmium compounds. This study shows that we are able to
introduce desirable features into these types of complexes
to optimize their design as anticancer drugs.
Acknowledgment. We thank the EPSRC National Crys-
tallography Service (Dr Peter Horton, University of Southamp-
ton) for collecting the diffraction data for complex 6. We
thank Dr. R. Deeth for advice on the use of ADF. We thank
Dr M. Khan (Warwick University), Dr A.M. Pizarro (War-
wick University), and Rhona Aird (General Hospital, Edin-
burgh) for help and advice on cell culture. This research was
supported by Engineering and Physical Sciences Research
Inorganic Chemistry, Vol. 48, No. 4, 2009 1761

van Rijt et al.
of Cl- ion on the hydrolysis of 3 and 6 (Figure S2), 1H NMR spectra
of the pH titration of 6A and 5A (Figure S3), plots of the 1H NMR
chemical shifts of the coordinated arene ring protons versus pH*
(Figure S4), and optimized geometries for the nucleobase adducts
of 1, 3, and 6 (Figure S5). This material is available free of charge
via the Internet at http://pubs.acs.org.
Council (EPSRC) and Warwick University. The authors also
acknowledge that their participation in the EU COST Action
D39 enabled them to exchange regularly the most recent
ideas in the field of anticancer metallodrugs with several
European colleagues.
Supporting Information Available: Time dependence for
hydrolysis of 3, 4, and 5 (Figure S1), effect of various concentrations
IC8020222
1762 Inorganic Chemistry, Vol. 48, No. 4, 2009