Amide linkage isomerism as an activity switch for organometallic osmium and ruthenium anticancer complexes

Article abstract of DOI:10.1021/jm900731j

We show that the binding mode adopted by picolinamide derivatives in organometallic OsII and RuII half-sandwich complexes can lead to contrasting cancer cell cytotoxicity. N-Phenyl picolinamide derivatives (XY) in OsII (1, 3-5, 7, 9) and RuII (2, 6, 8, 10

Full text of DOI:10.1021/jm900731j

J. Med. Chem. 2009, 52, 7753–7764 7753
DOI: 10.1021/jm900731j
Amide Linkage Isomerism As an Activity Switch for Organometallic Osmium and Ruthenium
Anticancer Complexes
Sabine H. van Rijt,† Andrew J. Hebden,‡ Thakshila Amaresekera,‡ Robert J. Deeth,† Guy J. Clarkson,† Simon Parsons,§
Patrick C. McGowan,*,‡ and Peter J. Sadler*,†
†Department of Chemistry, University of Warwick, Gibbet Hill Road, Coventry CV4 7AL, U.K., ‡School of Chemistry, University of Leeds, Leeds
LS2 9JT, U.K., and §School of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3JJ, U.K.
Received May 28, 2009
We show that the binding mode adopted by picolinamide derivatives in organometallic OsII and RuII
half-sandwich complexes can lead to contrasting cancer cell cytotoxicity. N-Phenyl picolinamide
derivatives (XY) in OsII (1, 3-5, 7, 9) and RuII (2, 6, 8, 10) complexes [(η6-arene)(Os/Ru)(XY)Cl]nþ
,
where arene=p-cymene (1-8, 10) or biphenyl (9), can act as N,N- or N,O-donors. Electron-withdrawing
substituents on the phenyl ring resulted in N,N-coordination and electron-donating substituents in
N,O-coordination. Dynamic interconversion between N,O and N,N configurations can occur in
solution and is time- and temperature- (irreversible) as well as pH-dependent (reversible). The neutral
N,N-coordinated compounds (1-5 and 9) hydrolyzed rapidly (t1/2 e min), exhibited significant
(32-70%) and rapid binding to guanine, but no binding to adenine. The N,N-coordinated compounds
1, 3, 4, and 9 exhibited significant activity against colon, ovarian, and cisplatin-resistant ovarian human
cancer cell lines (3 . 4>1>9). In contrast, N,O-coordinated complexes 7 and 8 hydrolyzed slowly, did
not bind to guanine or adenine, and were nontoxic.
Introduction
metal ions in high (gþ3) oxidation states due to its strong
σ-donor capability,14 and also some N,N coordination com-
plexes of RuII are known.15 In the present work, we have
investigated the influence on cancer cell cytoxicity of nitrogen,
nitrogen (N,N) versus nitrogen, oxygen (N,O) coordination of
picolinamide derivatives containing an anchoring pyridyl ring
in OsII and RuII arene complexes.
A number of ruthenium compounds has previously been
shown to display promising anticancer activity,16 and two
RuIII complexes have entered clinical trials, trans-[RuCl4-
(DMSOa)(Im)]ImH (NAMI-A, where Im=imidazole) and
trans-[RuCl4(Ind)2]IndH (KP1019, where Ind=indazole). It
is believed that the activity of the RuIII compounds is depen-
dent on the in vivo reduction to the more reactive RuII
species.17 This has led to increased interest into the anticancer
potential of RuII compounds, and RuII “half-sandwich” arene
complexes, for example, exhibit both in vitro and in vivo
activity.18 Also, our recent work has shown that analogues
containing the heavier congener osmium can also be potently
cytotoxic toward cancer cells with activity that is comparable
to the clinical drugs carboplatin and cisplatin.19,20
There is an intriguing interplay between oxygen and nitro-
gen binding of amides to metal ions in coordination com-
plexes.1 Amide groups are usually planar, with the C-N bond
possessing some 40% double-bond character. Metal ions tend
to interact only weakly with the carbonyl oxygen of a neutral
amide group, but this is strengthened if there is a nearby
anchoring group so that a chelate ring can form. However,
this interaction remains weak. The deprotonated amide nitro-
gen is a much stronger metal binding site (ca. 1016 more basic
than the neutral amide); a site which becomes more favorable
at higher pH, although it has to compete with metal ion
hydrolysis. Binding to a deprotonated amide N is also
strengthened when there is an anchoring group that allows a
5- or 6-membered chelate ring to form. There are only a
few reported examples of pairs of linkage isomers; these
include N- and O- bonded monodentate amide adducts of
(NH3)5CoII,2,3 dienPtII,4-6 and (NH3)4RuIII,7,8 however these
could not be interconverted.
Metal complexes with picolinamide ligands have previously
received attention because of their relevance to peptide
chemistry.9-11 These ligands are known to bind to metal ions
either as neutral N,O-donors, or as monoanionic N,N-donors
via loss of the amide proton. For example, complexes with the
first row transition metals FeII, CoII, NiII CuII, and ZnII
mostly involve N,O-donation.12,13 The anionic N,N-donor
form of the picolinamide ligand is able to stabilize first row
We show here that the binding mode exhibited by picolin-
amide derivatives in OsII and RuII arene complexes can have
a major effect on the cancer cell cytotoxicity of these com-
plexes and attempt to relate this not only to their solid-state
structures but also to their hydrolysis rates, to the acidity of
the resulting aqua adducts, and nucleobase binding. This
a Abbreviations: bip, biphenyl; p-cym, para- cymene; cisplatin, cis-
diamminedichloridoplatinum(II); RPMI, Roswell Park Memorial
Institute; DMSO, dimethylsulfoxide; IC50, concentration inhibiting cell
growth by 50%; SRB, sulforhodamine B; 9EtG; 9-ethyl guanine; 9EtA,
9-ethyl adenine.
*To whom correspondence should be addressed. For P.J.S.: phone,
(þ44) 024 7652 3818; fax, (þ44) 024 76523819; E-mail, p.j.sadler@
warwick.ac.uk. For P.C.M.: phone, (þ44) 0113 3436404; fax, (þ44) 0113
343 6565; E-mail: P.C.McGowan@leeds.ac.uk.
r
2009 American Chemical Society
Published on Web 09/30/2009
pubs.acs.org/jmc

7754 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
Chart 1. Osmium and Ruthenium Arene Complexes Studied in
This Work and the Structures of 9-Ethyl Guanine and 9-Ethyl
Adenine
6-trimethyl-Ph-picolinamide)Cl]PF6 (8 CHCl3), Figure 1A-D
3
and Figure S1A,B (Supporting Information, SI). All complexes
adopt the familiar pseudo-octahedral “three-legged piano stool”
geometry. Ruthenium and osmium are π-bonded to the
p-cymene ring (“the seat”) and σ-bonded to a chloride and to
a chelated N-Ph-picolinamide ligand. The latter is bound
through both nitrogens (pyridyl and amidinato nitrogen,
N,N-coordination) in 1, 2, 3, and 6 and through the pyridyl
nitrogen and carboxamide oxygen (N,O-coordination) in 7 and
8. These atoms, together with Cl, constitute the three legs of the
piano stool. Crystallographic data, selected bond lengths, and
angles are given in Tables S1 and S2 of the SI.
Intermolecular H-bonding between C26-H F17 (2.479
3 3 3
˚
A) and intramolecular H-bonding between C15-H F14
3 3 3
˚
˚
(2.547 A) and between C24-H O8 (2.212 A) is observed in
3 3 3
the crystal structure of 1 (Figure S2A, SI). In addition, there
is an offset π-stacking interaction between the p-cymene
arene and the substituted phenyl of adjacent molecules with
˚
a centroid-to-centroid distance of 3.71 A (plane tilt of 14°).
The molecular structure (Figure S1A, SI) and crystal cell
data (Table S1, SI) obtained for 2 are similar to those of Os
congener 1, although the M-Cl bond is significantly shorter
˚
˚
(i.e., M-Cl: 2.4434(14) A for 1, 2.4126(4) A for 2). The crystal
structure of 3 contains a molecule of methanol in the asym-
metric unit. There is an offset weak π-stacking interaction
between two pyridyls of adjacent molecules with a centroid-
˚
to-centroid distance of 4.06 A (Figure 2SB, SI). In addition,
there is H-bonding between the amidinato carboxylate and
˚
methanol solvent molecule (1.93 A, Figure S2B, SI).
study demonstrates that the substituents on the ligands in this
type of complex can have marked effects on the configuration
adopted by the complexes and their reactivity and might
provide a route to “fine-tune” and switch their biological
activity.
Remarkably, the bond distances of N,N-coordinated
osmium complex 3 are more comparable to N,N-coordinated
ruthenium complex 2 than to N,N-coordinated osmium com-
plex 1, with in particular the M-Cl bond length (i.e., M-Cl:
˚
˚
2.4434(14) A for 1, 2.4050(6) A for 2, and 2.4126(4) for
3 MeOD, Table S2, SI).
Results
3
Synthesis and Characterization. Novel half-sandwich
osmium(II) and ruthenium(II) arene complexes with
N-Ph-picolinamide derivatives (Chart 1) were synthesized
from the chlorido-bridged dimers, [(η6-p-cym)MCl2]2 (where
M is Os or Ru) and [(η6-bip)OsCl2]2. Osmium complexes
containing electron-withdrawing substituents on the phenyl
ring could be isolated as their neutral N,N-coordinated
forms (1, 3, 4, and 9 and RuII complex 2, Chart 1). The
isolation of N,O-coordinated picolinamide complexes was
successful only when the Ph-picolinamide ligand contained
electron-donating methyl groups for both the osmium (7)
and ruthenium compounds (8) (Chart 1). This N,O-binding
mode was confirmed by their X-ray crystal structures. Both
N,N- and N,O- coordination modes were observed for
The crystal structure of the dimethoxyphenylpicolinamide
-
ruthenium complex 6 contains half a PF6 anion in the
asymmetric unit. Inspection of the remainder of the structure
˚
indicated a short intramolecular distance of 2.425 A between
the oxygen atoms O7 in neighboring symmetry-related mole-
cules. This suggests that there is an O-H-O bond linking
two neighboring ruthenium fragments that would provide
appropriate charge balance for the anion (Figure 1C). How-
ever, this hypothetical H atom could not be located in the
Fourier map. For the related complex, [(η6-p-cym)Ru(N-
2-fluoro-Ph-picolinamide)Cl] (10), the X-ray structure of the
dimer [10-H-10]PF6 (Figure S3, Table S3, SI) also has the
-
O-H-O motif and a PF6 anion and the H atom in the
O-H-O motif could be located in the Fourier map, result-
ing in a linear O-H-O bond. Accordingly, for 6, the H atom
in the O-H-O bond was placed in a fixed calculated
position such that the bond O-H-O was linear (Figure 1C).
The X-ray structures of 7 and 8 revealed N,O-coordina-
tion for the picolinamide ligand. The cationic Os compound
-
compound 5 when no PF6 anion was added. However,
after addition of PF6-, N,N-coordination was observed by
1H NMR. This can be accounted for the binding of a proton
as a bridge between the carboxamide oxygens of two mole-
cules of the complex forming a cationic dimeric species, so
explaining the role of the counteranion PF6-. Such a dimeric
structure was also found in the crystal structure of the
ruthenium analogue 6 (Figure 1C) and is described below.
X-ray crystal structures were obtained for compounds
[(η6-p-cym)Os(N-2,4-difluoro-Ph-picolinamide)Cl] (1), [(η6-
p-cym)Ru(N-2,4-difluoro-Ph-picolinamide)Cl] (2), [(η6-p-
cym)Os(N-4-nitro-Ph-picolinamide)Cl] MeOD (3 MeOD),
-
7, where PF6 acts as a counterion, crystallizes with one
solvent molecule of dichloromethane. The phenyl ring of the
N-Ph-picolinamide ligand is almost perpendicular to the
pyridine ring with a plane angle of 82.19° (72.64° for 1,
Figure 1D). The Os-Cl bond distance in compound
7 CH2Cl2 is short compared to compound 1 (i.e., Os-Cl
3
˚
2.3878(10) A for 7 CH2Cl2, 2.4434(14) A for 1). A 1:1
˚
3
3
3
[(η6-p-cym)Ru(N-2,4-dimethoxide-Ph-picolinamide)Cl]2HPF6
([6-H-6]PF6),
[(η6-p-cym)Os(N-2,4,6-trimethyl-Ph-picolina-
mide)Cl]PF6 CH2Cl2 (7 CH2Cl2), and [(η6-p-cym)Ru(N-2,4,
enantiomeric ratio of SOs/Ru:ROs/Ru was observed in the
crystal lattices of all compounds. The same ratio was ob-
1
served for 7 in solution by H NMR (CDCl3, 298 K) on
3
3

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7755
Figure 1. X-ray structures and atom numbering schemes for complexes (A) [(η6-p-cym)Os(N-2,4-difluoro-Ph-picolinamide)Cl] (1), (B) [(η6-p-
cym)Os(N-4-nitro-Ph-picolinamide)Cl] MeOD (3 MeOD), (C) [(η6-p-cym)Ru(N-2,4-dimethoxide-Ph-picolinamide)Cl]2HPF6 ([6-H-6]PF6),
3
3
and (D) [(η6-p-cym)Os(N-2,4,6-trimethyl-Ph-picolinamide)Cl]PF6 CH2Cl2 (7 CH2Cl2). Hydrogen atoms, except the bridging H in 6 (C),
solvents, and counterions are omitted for clarity.
3
3
addition of the chiral anionic shift reagent TRISPHAT
(Figure S4, SI). No attempt was made to separate the
enantiomers. The cationic ruthenium compound 8, with
temperature range 173-323 K. From 173 K up to 283 K, the
ratio between N,O- and N,N-coordination remained at
about 50:50. However, above 283 K, the N,N-coordinated
linkage isomer of 3 was favored, and by 323 K, a full
conversion to the N,N-coordinated compound was observed
(Figure S5, SI). This process was independent of the deuter-
ated solvent (D2O, DMSO-d6, MeOD-d4, CDCl3 at 298 K)
and was irreversible after a 48 h period at ambient tempera-
ture after recording the 1H NMR spectrum at 298 K. Also, it
was found to be possible to obtain a higher ratio of N,O-
coordination versus N,N-coordination by synthesizing 3 at
lower temperatures. This typically resulted in the formation
of only the N,N-coordinated linkage isomer after 3-day
storage of the mixture in solution at ambient temperature.
To investigate the effect of pH on the 50:50 equilibrium
mixture of the configurational isomers of 3, 1H NMR spectra
of 3 in D2O were recorded over the pH* (pH meter reading
for a D2O solution) range of 2-9. When the pH was increased
from 2 to 9, a full conversion from N,O- to N,N-coordination
was observed (Figure S6, SI). At an acidic pH* of 2, only the
N,O-isomer of 3 was observed, while at pH* 8.5, only the N,N-
analogue of 3 was present. This process was found to be re-
versible. The onset of conversion of N,O- into N,N-coordination
appeared to occur when the pH* was raised above ca. pH* 7.
Hydrolysis Data. The rate of hydrolysis of compounds
1-5 and 7-9 in a 5% MeOD-d4/95% D2O mixture was
-
PF6 as the counterion, crystallizes with one solvent mole-
cule of chloroform (Figure S1B, SI). The methyl and iso-
propyl groups on the p-cymene ring are disordered over two
positions with occupancy of 0.5. The Ru-Cl and Ru-O
bond lengths of compound 8 are very similar to that of the
˚
osmium analogue 7 (i.e., M-Cl: 2.3853(11) A for 8 CHCl3,
3
˚
2.3878(10) A for 7 CH2Cl2: and M-O: 2.118(3) for 8 CHCl3,
3
3
2.117(2) for 7 CH2Cl2). The Ru-Cl bond of 8 is shorter than
3
for Ru compounds 2 and [6-H-6]PF6 (i.e., Ru-Cl 2.3853(11)
˚
˚
A for 8 CHCl3, 2.4050(6) A for 2, and 2.4114(11) A for
˚
3
[6-H-6]PF6)
Kinetic NMR studies. During the preparation of com-
pounds 1-4 and 9, an equilibrium between the two config-
urations was often observed by 1H NMR (in MeOD-d4 and
DMSO-d6). However, upon longer reaction times and re-
crystallization from dichloromethane, Os complexes 1, 3, 4,
and 9 and Ru complex 2 could be isolated as their neutral N,
N-coordinated forms (Chart 1). To investigate the nature of
the equilibrium between the N,O- and N,N-coordination
observed for products isolated from the syntheses of com-
plexes 1-4 and 9, a 50:50 mixture of 3 (obtained as a sticky
solid) was studied as a function of temperature, time, and
1
pH. H NMR spectra of the N,O- and N,N-coordinated
mixture of 3 (50:50) in MeOD-d4 were recorded over the
1
monitored by H NMR at 288 K by observation of new

7756 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
Table 1. Hydrolysis Data for Complexes 1-5 and 7-9 at 288 K
Table 3. In Vitro Growth Inhibition in the A2780, A2780cis and
HCT116 Cell Lines for Compounds 1-5 and 7-9 and Cisplatin
(CDDP) as Control, IC50 ( μM)a
Determined by 1H NMR
compd
k, h-1
g5.9
g5.9
t1/2, h
e0.12
e0.12
compd
A2780
A2780cis
HCT116
1
2
3
4
5
7
8
9
1
24.5 ( 1.0
>50b
30.1 ( 0.8
>50b
20.2 ( 0.3
NDc
2
g5.9
g5.9
e0.12
e0.12
e0.12
3
11.8 ( 0.1
21.1 ( 4.6
>50
7.8 ( 0.4
15.9 ( 0.4
>50
2.6 ( 0.4
10.3 ( 0.1
>50
4
g5.9
5
0.16 ( 0.03
1.63 ( 0.05
g5.9
4.33 ( 0.68
0.42 ( 0.02
e0.12
7
>50
>50
>50
>50
>50
>50
8
9
CDDP
33.2 ( 0.3
2.36 ( 0.1
31.9 ( 0.8
16.9 ( 0.5
22.0 ( 2.5
2.7 ( 0.5
a Drug treatment period was 24 h. Each value represents the mean (
SD for three independent experiments. b Testing hampered by low
aqueous solubility of 2. c ND = not determined
Table 2. pKa* Values for the Deprotonation of the Coordinated D2O in
AquaAdductsof Complexes 1, 7 and of the Amide Proton in Compound 7
compd
pKa* aqua adduct
pKa* NH
1
7
7.33
7.08
Table 4. Time Dependence of Nucleobase Adduct Formation for
Compounds 1-5 and 7-9 Determined by 1H NMR
7.62
compd
time/h
% 9EtG adduct
% 9EtA adduct
peaks over time due to aqua adduct formation. To ensure
that the hydrolysis rates were determined for the N,N-
coordinated isomers (e.g., for complex 3, the N,O-isomer is
formed reversibly at low pH), we determined the hydrolysis
rates at pH ∼ 7. All complexes, except N,O-coordinated
complexes 7 and 8, hydrolyzed rapidly with half-lives too
short to measure by 1H NMR (t1/2 e 10 min; time needed to
record the first 1H NMR spectrum). The percentage of aqua
peak formation for 7 and 8 was plotted against time and
fitted to pseudo-first-order kinetics to give half-lives of 4.33
and 0.42 h, respectively (Table 1).
1
0.17
24
56
56
∼34a
∼34a
31
43
57
57
70
75
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
3
4
5
7
8
9
0.17
24
0.17
24
0.17
24
0.17
24
0.17
24
0
1
pKa* Determination. Changes in the H NMR chemical
0.17
24
0
shifts for the protons of the coordinated p-cymene ring in
compounds 1 and 7, present in equilibrium aqueous solu-
tions of 1 or 7 containing their aqua adducts as the major
species, were followed over the pH* range 2-9 (Figure S7,
SI). When the pH* values of the solutions were increased, the
major set of NMR peaks assigned to the aqua adducts of 1
and 7 gradually shifted to higher field in the spectrum. For
compound 7, the two sets of peaks, assigned to the aqua and
chlorido adducts, shifted to high field as a result of amide
proton loss with increasing pH*. The chemical shift was
plotted against pH*, and the resulting pH* titration curves
were fitted to the modified Henderson-Hasselbalch equa-
tion.21,22 This gave rise to similar pKa* values for the
coordinated water of 7.3 for 1 and 7.1 for 7. The amide
proton in compound 7 gave a pKa* value of 7.6 (Table 2).
Cancer Cell Cytotoxicity. The cytotoxicity of the com-
plexes toward colon HCT116, ovarian A2780, and ovarian
cisplatin resistant A2780cis cancer cell lines was investigated.
Complexes 7, 8, and 5 were found to be nontoxic in all three
cell lines up to the highest test concentration of 50 μM. Their
IC50 values (concentration at which 50% of the cell growth is
inhibited) are therefore likely to be >100 μM, and the
compounds can therefore be described as inactive. However,
promising activity was observed for the neutral N,N-coordi-
nated compounds 1, 3, 4, and 9 (Table 3, Figure S8, SI). A
trend in cytotoxic activity is observed in all three cell lines,
following the order of 3 . 4>1>9 (Chart 1). Complex 3,
[(η6-p-cym)Os(N-4-nitro-Ph-picolinamide)Cl], shows simi-
lar activity as cisplatin in the HCT116 (colon) cell line and
promising activity in the A2780cis (ovarian) cell line
(Table 3). Surprisingly, the ruthenium analogue of 1
(complex 2) showed no activity up to the highest concentra-
tion tested (50 μM).
0
0.17
24
68
68
a Low solubility caused low signal-to-noise ratio in spectra
Binding to Model Nucleobases 9-Ethyl Guanine (9EtG) and
9-Ethyl Adenine (9EtA). Nucleobase binding reactions using
the model compounds 9-ethyl guanine (9EtG) and 9-ethyl
adenine (9EtA; Chart 1) were investigated. Solutions of all
arene compounds (1 mM; containing an equilibrium mixture
of the chlorido and respective aqua adducts as the major
species) with 1 mol equiv of 9EtG or 9EtA in D2O were
prepared and 1H NMR spectra were recorded at various time
intervals. The percentage of nucleobase binding based on 1H
NMR peak integrals is shown in Table 4. The addition of 1
mol equiv 9EtG to aqueous solutions of N,N-coordinated
compounds 1-5 and 9 gave rise to new peaks due to nucleo-
base adduct formation after ca. 10 min (33-70%, Table 4)
with no further changes observed after 24 or 72 h, except for
complex 3, which showed 10% more binding to 9EtG after 24
h. Compound 2, the ruthenium analogue of osmium complex
1, showed reduced nucleobase binding compared to 1, how-
ever due to its low solubility (<1 mM), low signal-to noise
ratios were obtained. The addition of 1 mol equiv 9EtA to
aqueous solutions of the N,N-coordinated compounds re-
sulted in no new peaks even after 72 h. Also the N,O-
coordinated complexes 7 and 8 did not give rise to any new
peaks upon reaction with 9EtG or 9EtA for 72 h (Table 4).
Computation. The minimum energy structures of nucleo-
base adducts of N,N-coordinated compound 1, [(η6-p-
cymene)Os(N-2,4-difluoro-Ph-picolinamide)9EtG-N7]þ
(1-9EtG), and [(η6-p-cymene)Os(N-2,4-difluoro-Ph-picolin-
amide)9EtA-N7]þ, (1-9EtA), and N,O-coordinated compound

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7757
on the phenyl ring of the picolinamide ligand. Electron-with-
drawing substituents such as nitro and fluoro groups on
the phenyl ring appear to favor coordination through both
nitrogens for OsII and RuII. However, an equilibrium between
the two configurations was observed under certain conditions.
Variable temperature NMR experiments on a 50:50 mix-
ture of N,N and N,O forms of [(η6-p-cym)Os(N-2-nitro-
Ph-picolinamide)Cl] (3) showed that the N,O-coordinated
complex is the kinetic product while N,N-coordination is
thermodynamically favored (Figure S5, SI). Indeed, for
compounds 1-4 and 9, neutral monomers containing
N,N-coordination could be obtained. Generally, osmium-
nitrogen bonds are more stable than osmium-oxygen
bonds. The increased acidity of the amide proton caused by
the electron-withdrawing (difluoro and nitro) groups on
the phenyl ring for 1-4, facilitates proton loss in solution,
favoring N,N-coordination. Complex 9 does not contain
electron-withdrawing substituents on the phenyl ring, but
biphenyl as the arene is less electron-rich compared to the
p-cymene arene present in 1-4 and was also isolated as the
N,N-complex. The X-ray crystal structure of active com-
pounds 1 and 3 show weak intermolecular π-π stacking
(Figure S2, SI). Similar π-π interactions could contribute
to the cytotoxicity of these systems involving intercalation
into DNA.
Figure 2. Optimized geometries for the cations (A) [(η6-p-cymene)Os-
(N-2,4-F-Ph-picolinamide)9EtG-N7]þ (1-9EtG), (B) [(η6-p-cymene)-
Os(N-2,4-F-Ph-picolinamide)9EtA-N7]þ (1-9EtA), (C) [(η6-p-cymene)-
Os(N-2,4,6-Me-Ph-picolinamide)9EtG-N7]2þ (7-9EtG), and (D)
[(η6-p-cymene)Os(N-2,4,6-Me-Ph-picolinamide)9EtA-N7]2þ (7-9EtA).
The observed N,O-coordination in compounds 7 and 8,
both of which contain the N-2,4,6-trimethyl-Ph-picolinamide
ligand, can be explained by the presence of sterically demand-
ing methyl groups, which, in combination with their electron-
donating effect (i.e., making the amide proton less acidic),
Table 5. Gas Phase and Solution (COSMO) 9EtG and 9EtA Binding
Energies for Complexes 1-9EtG, 1-9EtA, 7-9EtG, and 7-9EtAa
compound
9EtG kcal/mol
9EtA kcal/mol
1
7
gas
48.7
65.7
31.8
52.9
1
favor N,O-coordination. A H NMR peak for the amide
NH of both 7 and 8 was observed in their spectra but
accompanied by line broadening of all the peaks, suggesting
the presence of chemical exchange due to ring-opening and
closing; this demonstrates the lability of the Os/Ru-O bonds,
something that has previously been observed in acetylaceto-
nate complexes.23
1
7
COSMO
64.4
80.4
43.3
62.7
a Binding energies for structures depicted in Figure 3.
7, [(η6-p-cymene)Os(N-2,4,6-trimethyl-Ph-picolinamide)9EtG-
N7]2þ (7-9EtG), and [(η6-p-cymene)Os(N-2,4,6-trimethyl-Ph-
picolinamide)9EtA-N7]2þ (7-9EtA), obtained from DFT
calculations, are shown in Figure 2. The total binding
energy of 9EtG and 9EtA (see Chart 1 for their structures)
in both adducts is shown in Table 5. Both nucleobases show
significant binding energies toward both compounds 1 and
7; the bonding of 9-ethyl guanine is more favorable than for
9-ethyl adenine (by 16.9 kcal/mol and 12.7 kcal/ mol for
1-9EtG and 7-9EtG, respectively). The role of solvation
was investigated by calculating the binding energies with
COSMO, which simulates aqueous environments. The
effect of solvation was similar for all four nucleobase
adducts; it is notable that guanine is better solvated
(Table 5).
The presence of the PF6- anion had a directing effect on the
configuration of isolated complex 5. When the product was
isolated without NH4PF6 addition, a mixture of N,N- and N,
O-coordinated products was observed. When the PF6- anion
was present, dimerization of two osmium complexes through
H-bridging between the two carbonyl groups of adjacent
picolinamide ligands appeared to occur. The bridging proton
confers an overall þ1 charge, and PF6- acts as the counterion.
The electron-donating methoxy functional groups may pre-
vent facile amide proton loss in solution to form exclusively
the neutral N,N-coordinated product but do not causeenough
steric hindrance (the 6-position on phenyl is still un-
substituted) for the exclusive formation of the N,O-linkage
isomer. In the amide tautomeric form (HO-CdN), the O-H
group may associate with another monomer forming the
cationic dimeric species in the presence of PF6-, creating a
binding mode that is intermediate between N,O- and N,N-
coordination. This unusual motif also appears to be present in
the X-ray crystal structure of ruthenium analogue 6 ([6-H-
6]PF6, Figure 1C) and in complex 10, ([(η6-p-cym)Ru(N-2-
fluoro-Ph-picolinamide)Cl]2HPF6, [10-H-10]PF6, Figure S3,
SI), where the H atom in the O-H-O motif was located in the
Fourier map resulting in a linear O-H-O bond. A similar
motif has been reported for [Cu{6,60-bis(2-hydroxyphenyl)-
2,20-bipyridine}](HPF)0.5 H2O,24 where the O O distance
Discussion
OsII and RuII complexes 1-9 (Chart 1) containing N-Ph-
picolinamide derivatives that can act as N,N- (through pyridyl
and amidinato nitrogen) or N,O-donors (through the
pyridyl nitrogen and carbonyl oxygen) have been synthesized
and characterized. In an attempt to understand their contrast-
ing cytotoxic behavior toward cancer cells, we have investi-
gated their aquation rates, the acidity of aqua adducts, and
nucleobase binding in relationto the natureof the substituents
on the phenyl ring of the picolinamide ligand.
3
3 3 3
˚
The structures of the complexes depend intimately on
electronic effects and steric demands of the functional groups
is 2.397 A and the O-H-O angle almost linear at 173.85°. In
0
˚
the case of complex 6, the O7 O7 distance is 2.425 A. A
3 3 3

7758 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
complex 7, over 36 times slower at 288 K compared to the
N,N-coordinated compounds 1-5 and 9 (Table 1). Therefore,
the coordinated chelating ligand can play a major role in
controlling the reactivity of these types of complexes.
The substituents on the phenyl ring of the picolinamide
ligand may also play a role in their aqueous reactivity.
However, all the N,N-coordinated Os compounds hydrolyzed
rapidly and reacted with nucleobases to a similar extent. This
appears to be the case regardless of the electronic and steric
nature of the substituents on the phenyl ring. The difference
observed in the hydrolysis rates can be explained by the strong
donor character of the amidate nitrogen and by the overall
positive charge on 7 and 8, which hinders chloride release
compared to the neutral N,N-coordinated complexes 1-5 and
9. Indeed, in the X-ray crystal structures, the Os-Cl bond in 7
˚
is 0.06 A shorter than the Os-Cl bond in complex 1 (Table S2,
SI). Changing OsII in 7 to RuII (8) increased the aqueous
reactivity (∼10 times, Table 1), in line with the generally faster
kinetics observed for the second row Ru comparedto the third
row Os transition metal.
The N,O-coordinated compounds 7 and 8 and N,N-coordi-
nated complex 5 were noncytotoxic up to the highest concen-
tration tested (50 μM). The difference between 5 and the other
N,N-coordinated compounds is that it contains strong elec-
tron-donating substituents on the phenyl ring, whereas the
other N,N-coordinated compounds (1-4) contain electron-
withdrawing substituents. The strong donating capabilities of
the methoxy substituents is demonstrated by the protonation
of the carbonyl oxygen and formation of the hydrogen
bridged dimer in the crystal structure of 6, the Ru analogue
of 5 (Figure 1C). Such protonation may weaken the metal-N
bond and make the complex susceptible to reactions with
other biomolecules in cells. In the case of complexes 7 and 8,
their slow aqueous reactivity might prevent the complexes
reacting with targets such as, DNA. Promising activity was
observed for the neutral N,N-coordinated compounds 1, 3, 4,
and 9 in the HCT116 (colon), A2780 (ovarian), and A2780cis
(ovarian, cisplatin resistant) cell lines (Table 3, Figure S8, SI).
A trend in cytotoxic activity was observed in all three cell lines:
electron-withdrawing substituents (F, NO2) on the uncoordi-
nated phenyl ring influence the biological activity signifi-
cantly, with activity decreasing in the order of 3 . 4>1>9
(where 9 has an unsubstituted phenyl ring, Chart 1). The
activity of 9 is probably enhanced by the presence of biphenyl
as the coordinated arene instead of p-cymene, which is the
arene in the other compounds because biphenyl can act as a
DNA intercalator, contributing to its cytotoxicity.27-29 Com-
plex 3, [(η6-p-cym)Os(N-4-nitro-Ph-picolinamide)Cl], shows
similar activity as cisplatin in the HCT116 (colon) cell line and
shows promising activity in the A2780 (ovarian) cisplatin
resistant cell line (Table 3). The different cytotoxicity profiles
of the four active compounds indicate that the nature of the
substituent is important in the mechanism of action of these
complexes. In addition, the difference in cytotoxicity observed
between isomers 4 (o-NO2) and 3 (p-NO2) indicates that also
the position of the substituent on the phenyl ring is important
for the cytotoxic activity of these complexes. Complex 2 (the
ruthenium analogue of 1) appears to be nontoxic, however its
low solubility hampered the testing.
Figure 3. Summary of reaction pathways giving rise to N,N- or N,O
coordination for osmium and ruthenium (M = Os or Ru) picolin-
amide complexes.
summary of the different reaction pathways which could give
rise to N,N- or N,O-coordinated osmium and ruthenium
picolinamide complexes is presented in Figure 3.
Dynamic NMR studies on a 50:50 mixture of the two
configurational isomers of compound 3 showed that the
equilibrium between the N,O- and N,N-coordination is tem-
perature- as well as pH-dependent. At low temperatures, the
kinetic N,O-coordinated product remained as 50% of the
mixture, while the thermodynamic N,N-coordinated product
was exclusively and irreversibly formed at a temperature of
323 K or higher (Figure S5, SI). Changing the pH of the same
mixture of linkage isomers to an acidic pH* of 2 resulted in the
formation of only the N,O-coordinated product. Under acidic
conditions, the amide nitrogen remains protonated, favoring
donation through the lone pairs of oxygen. Surprisingly, even
at pH 2.4, the equilibrium mixture contained some of the
thermodynamic N,N-coordinated product. Full, reversible,
conversion to the N,N-form was observed over the pH range
of 7.1-8.6 (Figure S6, SI). These observations show the high
thermodynamic stability of the N,N-coordinated linkage iso-
mer and suggest the possibility that appropriate ligand design
could allow controllable switching between N,O and N,N
configurations under biological conditions.
Previous studies on the hydrolysis rates of OsII arene
compounds of the type [(η6-arene)Os(XY)Cl]nþ where XY is
a N,N-coordinated chelating ligand gave an aqueous half-life
of 6.4 h for XY=ethylenediamine, ca. 40 times slower than for
its ruthenium analogue.25 The rate of hydrolysis was slowed
down even further (by a factor of ca. 7) by incorporating a
π-acceptor chelating ligand such as 2,20-bipyridine or 1,10-
phenanthroline as XY.26 In the present study, however, in
which the N,N-coordinated ligand is anionic, the N,N-coordi-
nated compounds displayed very fast hydrolysis rates, too fast
to measure by 1H NMR even at 288 K (t1/2 e 0.12 h, Table 1).
In contrast, the N,O-coordinated compounds 7 and 8 hydro-
lyzed comparatively slowly, in the case of OsII arene
Binding studies of RuII and OsII arene complexes with
nucleobases are of special interest because DNA is considered
to be the main target for classical transition-metal anticancer
drugs.30 Distortions of DNA structure are critical for protein
recognition and downstream processing and often correlate

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7759
with anticancer activity.31,32 To gain insight into the reactivity
of these types of complexes with DNA, the nucleobase
derivatives 9-ethyl guanine (9EtG) and 9-ethyl adenine
(9EtA) were used as models (Chart 1). Previously we found
that the OsII biphenyl complex containing the N,N-chelator
ethylenediamine (en) as ligand reacted only slowly with 9EtG
and only to a limited extent with adenosine (Ado).25 All
neutral N,N-coordinated compounds studied in this work
(1-4 and 9, Chart 1) showed significant and rapid binding
to 9EtG (detectable after just 10 min; 32-70%, Table 4). The
cytotoxic activity of these complexes might therefore be aided
by their fast reaction with guanine bases in the target DNA.
Surprisingly, inactive compound 5, containing ortho and para
methoxy substituents on the phenyl ring of the chelating
ligand, also binds to 9EtG to a large extent, suggesting that
its inactivity is not due to its reactivity toward DNA but
perhaps due to inactivating side reactions with other bio-
molecules or possibly reduced cell uptake due to its different
charge distribution. However, we cannot rule out the possi-
bility that DNA is not the primary target for these types of
complexes. Ruthenium complex 2, the analogue of the active
osmium complex 1, appeared to show reduced nucleobase
binding compared to 1 (Table 4), however NMR studies were
hampered by its low solubility. None of the active cytotoxic N,
N-coordinated compounds nor inactive complex 5 showed
any binding toward 9-ethyl adenine, a similar preference to
that observed for the ethylenediamine OsII arene complexes
mentioned earlier. N,O-coordinated complexes 7 and 8
did not bind to the guanine or adenine model nucleobases.
This is attributable to their kinetic inertness and suggests
that they would not be effective in targeting DNA. Bar
charts that summarize the effect of N,N- versus N,O-binding
on the hydrolysis rates, guanine and adenine binding, and
cytotoxic activity of compounds 1, 3, 4 and 7-9 are shown in
Figure 4.
Computational methods were used to gain insight into the
guanine-specific binding observed for compounds 1-5 and 9
and the lack of nucleobase reactivity for compounds 7 and 8
(Chart 1, Table 5). The total bonding energies in the gas phase
and withwater solvation(COSMO) ofthe nucleobases 9-ethyl
guanine (9EtG) and 9-ethyl adenine (9EtA, Chart 1) in the
optimized structures of [(η6-p-cymene)Os(N-2,4-difluoro-
Ph-picolinamide)9EtG-N7]þ (1-9EtG), [(η6-p-cymene)Os-
(N-2,4-difluoro-Ph-picolinamide)9EtA-N7]þ (1-9EtA), [(η6-
p-cymene)Os(N-2,4,6-trimethyl-Ph-picolinamide)9EtG-N7]2þ
(7-9EtG), and [(η6-p-cymene)Os(N-2,4,6-trimethyl-Ph-pico-
linamide)9EtA-N7]2þ (7-9EtA) (see Figure 2 for their struc-
tures) are shown in Table 5. The 9EtG nucleobase adduct of
active compound 1 (1-9EtG) is thermodynamically preferred
by 16.9 kcal mol-1 compared to the 9EtA adduct of 1
(1-9EtA), with a bonding energy of 48.7 kcal mol-1 compared
to 31.8 kcal mol-1 for the adenine adduct (Table 5). This
significant difference may explain the guanine-specific binding
observed for the neutral N,N-coordinated compounds 1-5
and 9. Surprisingly, inactive compound 7 displays high bond-
ing energies toward both nucleobases, showing a similar
thermodynamic preference for guanine over adenine by
12.7 kcal mol-1, with a bonding energy of 65.7 kcal/mol for
7-9EtG compared to 52.9 kcal/mol for 7-9EtA (Table 5). The
high bonding energies calculated for both nucleobase adducts
of 7indicate that its observed inactivity toward the nucleobases
is not due to the thermodynamic stability of the nucleobase
adducts but most likely due to high kinetic barriers. Because
nucleobase binding is likely to require initial aquation, the slow
Figure 4. Bar chart illustrating the effect of N,N- versus N,O-
binding of picolinamide on the hydrolysis rates, guanine and
adenine binding, and cytotoxic activity in A2780 (ovarian),
A2780cis (ovarian cisplatin resistant), and HCT116 (colon) cancer
cell lines, of complexes 1, 3, 4, 7, 8, and 9.
hydrolysis rates of 7 and 8 will presumably hinder nucleobase
binding. The high nucleobase bonding energies found for 7
cannot be attributed to a difference in solvation effects for 1
and 7 (which are differently charged, þ1/þ2, respectively)
because the solvation effects calculated using COSMO to
simulate aqueous conditions are similar for both nucleobase
adducts of 1 and 7. Guanine is notably more strongly solvated
(Table 5). This higher solvation energy for guanine (4.2 kcal
mol-1 and 5.0 kcal mol-1 higher than 9EtA for 1 and 7,
respectively) can be attributed to the high charge concentration
at the oxo group in guanine.
Conclusions
This study demonstrates that the coordination mode (N,N
versus N,O) and substituents on the phenyl ring of N-phenyl-
picolinamide ligands in both RuII and OsII arene complexes
can have a dramatic effect on the aqueous and biological

7760 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
activity of the complexes. Both N,N- and N,O-configurations
were characterized for both metals and are determined by
electronic effects and steric demands of the functional
groups on the phenyl ring of the picolinamide ligand. In
addition, an unusual binding mode, involving proto-
nation of the amidinato carbonyl, was observed in the
N,N-complexes 5 and 6 (see Figure 3 for a summary).
Depending on the synthetic method, an equilibrium between
the N,O- and N,N-coordinated products was obtained that is
time and temperature- (irreversible) as well as pH-dependent
(reversible). This suggests that appropriate ligand design
could be used to optimize the interconversion between these
two coordination modes under biological conditions.
In contrast to complexes in the family containing neutral N,
N donors [(η6-arene)Os(XY)Cl]þ, where XY is ethylenedi-
amine, 2,20-bipyridine, or 1,10-phenanthroline, all the N,N-
coordinated picolinamide compounds in this study, displayed
rapid hydrolysis rates (t1/2 e 0.12 h). The N,O-coordinated
compounds 7 and 8 hydrolyzed much more slowly, over 36
times slower for 7 (at 288 K) than the N,N-coordinated
compounds (1-5 and 9). This large difference in aqueous
reactivity between complex 7 and complexes 1, 3, 4, and 9 is
remarkable and demonstrates not only the importance of the
chelatingligand inthese complexesbut also suggeststhat these
complexes can be “tuned” to obtain biologically favorable
ligand exchange rates.
Experimental Section
Materials. 1,4-Dihydrobiphenyl and the dimers, [(η6-p-cym)-
Os/RuCl2]2 and [(η6-bip)OsCl2]2, were prepared by previously
reported procedures.25,33 9-Ethyl guanine and 9-ethyl adenine
were purchased from Sigma-Aldrich. OsCl3 nH2O and
RuCl3 nH2O were purchased from Alfa Aesar. All deuterated
3
3
solvents were obtained from Aldrich. Ethanol and methanol
were distilled over magnesium/iodine prior to use. Complexes
1-10 were synthesized from the dimeric precursors, [(η6-p-
cym)MCl2]2 (where M is Os or Ru) and [(η6-bip)OsCl2]2, using
procedures similar to those reported previously for other half-
sandwich arene complexes.34,35 The purities of compounds
1-10 were all determined to be g95% by elemental analysis
and are reported in the SI. In addition, 1H, 13C NMR, and
ESI-MS data for compounds 1-10 and the preparations and
characterization data for the N-Ph-picolinamide ligands are in
the SI.
Preparation of the Complexes: [(η6-p-cym)Os(N-2,4-difluoro-
Ph-picolinamide)Cl] (1). A suspension of [(η6-p-cym)OsCl2]2 (51
mg, 0.086 mmol) and N-2,4-difluoro-Ph-picolinamide (2 equiv,
32 mg) in dry and degassed EtOH (25 mL) was refluxed under
argon for 2 h, after which time the solution had turned color
from brown to yellow. The solution was filtered while hot into a
filtered solution of NH4PF6 (∼5 mol equiv, 110 mg) in 5 mL of
EtOH. This mixture was stirred for another hour at ambient
temperature. The filtrate was brought to dryness on a rotary
evaporator, and the orange solid was redissolved in a minimum
-
of dichloromethane; excess of the PF6 salt formed a white
Promising cytotoxic activity for OsII complexes followed
the order 3 . 4 > 1 > 9 in all three cell lines tested, and
compound 3 shows similar activity as cisplatin in HCT116
(colon) cell line and shows promising activity in the A2780cis
(ovarian cisplatin resistant) cell line.
precipitate. This was filtered off and the volume of the filtrate
reduced on a rotary evaporator until an orange precipitate
began to form. The mixture was left standing at 278 K to allow
precipitation to occur. The orange crystalline powder was
recovered by filtration and air-dried to give a final yield of
42.9 mg (56%) Crystals of 1 suitable for X-ray diffraction were
obtained by slow evaporation from MeOH at ambient tempera-
ture.
All active compounds showed significant binding to
9EtG with 32-70% binding after just 10 min. Their rapid
binding to guanine may aid binding to DNA and contribute
to their cytotoxicity. No binding to adenine (9EtA) was
observed for any of the active compounds. N,O-coordinated
complexes 7 and 8 did not show any binding to guanine or
adenine model nucleobases, indicating that their cytotoxic
inactivity is likely to be due to their overall slow kinetics,
hindering their binding to DNA. The anomalous behavior of
N,N-complex 5, which binds rapidly to 9-ethylguanine but is
noncytotoxic, seems likely to be due to the presence of the
electron-donating methoxy groups on the phenyl ring and
high basicity of the carbonyl oxygen, which was partially
protonated in the crystal structure of the isolated dimer. In
DFT calculations, the 9EtG nucleobase adduct of active
complex 1 is thermodynamically preferred compared to its
9EtA adduct by 16.9 kcal mol-1 (Table 5), explaining the
guanine-specific binding observed experimentally for all the
N,N-coordinated compounds (1-5 and 9). Surprisingly, DFT
calculations with inactive compound 7 show high bonding
energies toward both nucleobases. This indicates that its
unreactivity toward both nucleobases is most likely caused
by high kinetic barriers, as indicated by its slow hydrolysis
rate.
[(η6-p-cym)Ru(N-2,4-difluoro-Ph-picolinamide)Cl] (2). A sus-
pension of N-2,4-difluoro-Ph-picolinamide (75 mg, 0.32 mmol)
was added to a Schlenk tube containing a stirred solution of [(η6-
p-cym)RuCl2]2 (0.1 g, 0.16 mmol) in dry ethanol (50 mL). The
mixture was warmed at 323 K for 15 min until dissolution and
then filtered over NH4PF6 (0.1 g, excess). The mixture was
stirred under dinitrogen for 18 h to give an orange solution. The
solvent was removed in vacuo to leave an orange solid. The solid
was washed with petrol (3 ꢀ 10 mL) and dried in vacuo to give an
orange powder. The product was dissolved in MeOH for
recrystallization. Red-orange prism-shaped crystals were isolated,
washed in petrol (3 ꢀ 10 mL), and dried in vacuo, Yield:
56 mg (68.8%). Crystals of 2 suitable for X-ray diffraction were
obtained by slow evaporation from MeOH at ambient temperature.
[(η6-p-cym)Os(N-4-nitro-Ph-picolinamide)Cl] (3). A suspen-
sion of [(η6-p-cym)OsCl2]2 (52.7 mg, 0.087 mmol) and N-4-nitro-
Ph-picolinamide (2.1 mol equiv, 35.5 mg) in dry and degassed
EtOH (25 mL) was refluxed mildly for 3 h, after which no
significant color change was observed. The solution was filtered
while hot into a filtered solution of NH4PF6 (∼5 mol equiv,
110 mg) in 5 mL of EtOH. This mixture was stirred for another
2 h at ambient temperature. The orange solution was reduced in
volume on a rotary evaporator until an orange precipitate began
to form. The mixture was left standing at 278 K to allow further
precipitation. The orange crystals were recovered by filtration
and air-dried to give a final yield 15.3 mg (19%). Crystals of
On the basis of the examples studies here, it appears that N,
N- and N,O-coordinated N-phenyl picolinamide OsII and
RuII arene complexes can exhibit dramatic differences in
their biological reactivity. The N,N-coordinated compounds
hydrolyze quickly, show selective binding to guanine, and
exhibit cancer cell cytotoxicity, while the N,O-coordinated
compounds hydrolyze slowly, display no reactivity toward
guanine or adenine, and are noncytotoxic, as summarized in
Figure 4.
3 MeOD suitable for X-ray diffraction were obtained by slow
3
evaporation from MeOD at ambient temperature.
[(η6-p-cym)Os(N-2-nitro-Ph-picolinamide)Cl] (4). A suspen-
sion of [(η6-p-cym)OsCl2]2 (50 mg, 0.083 mmol) and N-2-nitro-
Ph-picolinamide (2 mol equiv, 35 mg) in dry and degassed EtOH
(25 mL) was refluxed for 2 h, after which no significant color
change was observed. The solution was filtered while hot into a
filtered solution of NH4PF6 (∼5 mol equiv, 110 mg) in 5 mL of

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7761
EtOH. This mixture was stirred for another 2 h at ambient
temperature. The solvent was removed on a rotary evaporator
and the residue redissolved in dichloromethane. This mixture
was filtered, and again the solvent was removed on a rotary
evaporator. The orange solid was redissolved in a minimum of
MeOH and was left standing at 278 K to allow precipitation to
occur. The orange crystals were recovered by filtration and air-
dried to give a final yield 27.5 mg (36%).
under argon for 1 h. The brown solution was filtered while hot
into a filtered solution of NH4PF6 (∼5 mol equiv, 110 mg) in
5 mL of EtOH. This mixture was stirred for another 3 h at
ambient temperature. The mixture was reduced in volume on a
rotary evaporator until a precipitate began to form and was left
standing at 278 K. The orange crystalline powder was recovered
as two associating monomers with PF6 as a counterion by
filtration and air-dried to give a final yield of 44.4 mg (64%).
[(η6-p-cym)Os(N-2-fluoro-Ph-picolinamide)Cl] (10). N-2-Fluoro-
Ph-picolinamide (0.069 g, 0.32 mmol) was added to a Schlenk tube
containing a stirred solution of dichloro-p-cymene ruthenium dimer
(0.1 g, 0.16 mmol) in dry ethanol (50 mL). The mixture was warmed
at 323 K for 15 min until dissolution and then filtered over to
NH4PF6 (0.1 g, excess). The mixture was stirred under dinitrogen
for 18 h to give an orange solution. The solvent was removed in
vacuo to leave an orange solid. The solid was washed with petrol
(3 ꢀ 10 mL) and dried in vacuo to give an orange powder. The
product was dissolved in MeOH for recrystallization. Red-orange
prism-shaped crystals were isolated, washed in petrol (3 ꢀ 10 mL),
and dried in vacuo. The crystals were suitable for X-ray crystal-
lographic analysis (0.060 g, 0.054 mmol, 67.1%).
[(η6-p-cym)Os(N-2,4-dimethoxide-Ph-picolinamide)Cl]2PF6
(5). A suspension of [(η6-p-cym)OsCl2]2 (50 mg, 0.081 mmol)
and N-2,4-dimethoxide-Ph-picolinamide (2 mol equiv, 34 mg) in
dry and degassed EtOH (25 mL) was refluxed for 2.5 h, after
which no significant color change was observed. The solution
was filtered while hot into a filtered solution of NH4PF6 (∼5 mol
equiv, 110 mg) in 5 mL of EtOH. This mixture was stirred for
another hour at ambient temperature. An orange precipitate
formed in this period. The volume was reduced on a rotary
evaporator upon which more precipitate began to form. The
orange powder was recovered as two associating monomers
with PF6 as a counterion by filtration and air-dried to give a final
yield of 57.6 mg (74%).
[(η6-p-cym)Ru(N-2,4-dimethoxide-Ph-picolinamide)Cl]2PF6
(6). N-2,4-Dimethoxide-Ph-picolinamide (0.083 g, 0.32 mmol)
was added to a Schlenk tube containing a stirred solution of
dichloro-p-cymene ruthenium dimer (0.1 g, 0.16 mmol) in dry
ethanol (50 mL). The mixture was warmed at 323 K for 15 min
until dissolution and then filtered over to NH4PF6 (0.1 g,
excess). The mixture was stirred under nitrogen for 18 h to give
an orange solution. The solvent was removed in vacuo to leave
an orange solid. The solid was washed with petrol (3 ꢀ 10 mL)
and dried in vacuo to give an orange powder. The product was
dissolved in MeOH for recrystallization. The coffin shaped
crystals were suitable for X-ray crystallographic analysis
(0.053 g, 0.044 mmol, 55%).
Methods and Instrumentation: NMR Spectroscopy. 1H NMR
spectra were acquired in 5 mm NMR tubes at 298 K (unless
stated otherwise) on either a Bruker DMX 500 (1H = 500.13
MHz) or AVA 600 (1H=599.81 MHz) spectrometer. 1H NMR
chemical shifts were internally referenced to (CHD2)(CD3)SO
(2.50 ppm) for DMSO-d6, CHCl3 (7.26 ppm) for chloroform-d1,
CD2HOD (3.34 ppm) for methanol-d4, and to 1,4-dioxane (3.75
ppm) for aqueous solutions. For NMR spectra recorded for
aqueous solutions, water suppression was carried out using
Shaka or presaturation methods.36 All data processing was
carried out using TOPSPIN version 2.0 (Bruker U.K. Ltd.).
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 or by Warwick Analytical Service using an Exeter
analytical elemental analyzer (CE440) and the School of Chem-
istry Microanalytical Service, University of Leeds. Mass spectra
were obtained on a VG Autospec mass spectrometer by the
Department of Chemistry Mass Spectrometry Service, Univer-
sity of Leeds.
X-ray Crystallography. Diffraction data for 1 and 3 were
collected by the EPSRC National Crystallography Service
(Southampton) using a Bruker-Nonius FR591 anode X-ray
generator and a Bruker-Nonius Kappa CCD area detector.
The crystal was held at 120(2) K with an Oxford Cryosystem
cryostream cooler.37 Absorption corrections for all data sets
were performed with the multiscan procedure SADABS;38
structures were solved using either Patterson or direct methods
(SHELXL39 or DIRDIF40); complexes were refined against F or
F2 using SHELXTL, and H-atoms were placed in geometrically
calculated positions. X-ray diffraction data for 7 were collected
with Mo KR radiation on a Bruker SMART Apex diffract-
ometer equipped with an Oxford Cryosystems low-temperature
device operating at 150 K. Following application of a multiscan
correction for systematic errors (SADABS38), the structure was
solved by direct methods (SIR9241) and refined by full-matrix
least-squares against F2 (CRYSTALS42). H atoms were placed
geometrically and then refined subject to distance restraints and
finally constrained to ride on their host atoms for the final cycles
of refinement. All non-H atoms were refined with anisotropic
displacement parameters. The modeling program Diamond
3.020 was used for production of graphics.
[(η6-p-cym)Os(N-2,4,6-trimethyl-Ph-picolinamide)Cl]PF6 (7).
A suspension of [(η6-p-cym)OsCl2]2 (50 mg, 0.083 mmol) and N-
2,4,6-trimethyl-Ph-picolinamide (2 mol equiv, 29.9 mg) in dry
and degassed EtOH (25 mL) was refluxed under argon for 2 h,
after which the solution turned color from brown to bright
yellow. The solution was filtered while hot into a filtered
solution of NH4PF6 (∼5 mol equiv, 110 mg) in 5 mL of EtOH.
This mixture was stirred for another half hour at ambient
temperature. The filtrate was brought to dryness on a rotary
evaporator, and the yellow solid was redissolved in a minimum
of dichloromethane; excess of the PF6 salt formed a white
precipitate. This was filtered off and the volume of the filtrate
was reduced on a rotary evaporator until an orange precipitate
began to form. The mixture was left standing at 278 K to allow
precipitation to occur. The orange crystals were recovered by
filtration and air-dried to give a final yield of 59 mg (78%).
Crystals of 7 CH2Cl2 suitable for X-ray crystallography were
obtained by slow evaporation from CH2Cl2 at ambient tem-
perature.
3
[(η6-p-cym)Ru(N-2,4,6-trimethyl-Ph-picolinamide)Cl]PF6 (8).
A suspension of [(η6-p-cym)RuCl2]2 (78 mg, 0.127 mmol) and N-
2,4,6-trimethyl-Ph-picolinamide (2 equiv, 67 mg) in dry and
degassed EtOH (25 mL) was refluxed under argon for 2.5 h. The
orange solution was filtered while hot into a filtered solution of
NH4PF6 (∼5 equiv, 110 mg) in 5 mL of EtOH. This mixture was
stirred for 16 h at ambient temperature. The orange solution was
reduced in volume on a rotary evaporator until an orange
precipitate began to form. The mixture was left standing at
278 K to allow further precipitation. The orange crystals were
recovered by filtration and air-dried to give a final yield 56 mg
X-ray diffraction data for 2, 6, 8, and 10 were collected with a
Nonius KappaCCD area-detector diffractometer using graphite
(69%). Crystals of 8 CHCl3 suitable for X-ray crystallography
were obtained by slow evaporation from ethanol at ambient
temperature.
3
˚
monochromated Mo KR radiation (λ=0.71073 A). The crystal
was cooled to 150 K by an Oxford Cryosystems low temperature
device37 before data collection using 1.0° j rotation frames. The
images were processed using the DENZO and SCALEPACK
programs,43 followed by structure solution by direct methods
[(η6-bip)Os(N-Ph-picolinamide)Cl] (9). A suspension of [(η6-
bip)OsCl2]2 (50 mg, 0.087 mmol) and N-Ph-picolinamide (2 mol
equiv, 25 mg) in dry and degassed EtOH (25 mL) was refluxed

7762 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
via one of the SHELXS8644 SIR92,41or SIR9745 programs. All
structures were refined by full-matrix least-squares on F2 using
SHELXL97.39 Molecular graphics were plotted using POVray46
via XSeed.47 All non-hydrogen atoms were refined anisotropi-
cally. Hydrogen atoms were constrained with a riding model;
U(H) was set at 1.2ꢀ (1.5ꢀ for methyl groups) Ueq for the parent
atom. There were disorder problems involving solvent of crys-
tallization in the structures of 2 and 8, and details are given
below.
In the case of 2, the methyl and isopropyl groups (involving
C1, C2, C3, C7, and C13) of the p-cymene ligand to Ru1 exhibit
large ADPs, indicating that the p-cymene arene as a whole is
displaying some disorder, but this behavior was not modeled.
This disorder results in some bond length and angle anomalies in
the p-cymene arene. For 8, the severe disorder in the PF6- anion
could not be resolved, resulting in high anisotropic thermal
parameters. In addition, the methyl and isopropyl groups
(involving C20, C27, C28, and C29) of the p-cymene arene to
Ru1 exhibit large ADPs. This indicates that the p-cymene ligand
as a whole is displaying some disorder, but this behavior was not
modeled.
X-ray crystallographic data for compounds 1, 2, 3, 6, 7, 8, and
10 have been deposited in the Cambridge Crystallographic Data
Centre under the accession numbers CCDC 708837, 723343,
733992, 723341, 723344, 723340, and 723342, respectively.
pH* Measurements. pH* values (pH meter reading from D2O
solution without correction for effects of D on glass electrode) of
NMR samples in D2O were measured at ca. 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.
the sample was adjusted if necessary so as to remain close to 7.4
(physiological). 1H NMR spectra of these solutions were re-
corded at 298 K after various time intervals. In between, the
samples were kept at 310 K.
Computation. DFT calculations were carried out using the
Amsterdam density functional (ADF)50 program (version
2007.01). The coordinates of complexes used for the calculations
were directly obtained from the X-ray crystal structures. Modi-
fications 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 sca-
lar ZORA relativistic correction.51-55 The general numerical
integration was 4.0. The frozen core approximation56 was
applied using triple-ζ plus polarization (TZP) basis sets. Default
convergence criteria were applied for self-consistent fields (SCF)
and geometry optimization. Geometry and energies were con-
firmed with frequency calculations. The conductor-like screen-
ing model (COSMO), as implemented in the ADF program, was
used to simulate aqueous environments with ε = 78.4 and a
ꢀ
˚
probe radius=1.9 A. 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. Single-point calculations were carried out for the
guanine and adenine fragments of the cations of complexes
1-9EtG [(η6-p-cymene)Os(N-2,4-difluoro-Ph-picolinamide)-
9EtG-N7]þ, 1-9EtA [(η6-p-cymene)Os(N-2,4-difluoro-Ph-pi-
colinamide)9EtA-N7]þ, 7-9EtG [(η6-p-cymene)Os(N-2,4,
6-trimethyl-Ph-picolinamide)9EtG-N7]2þ, and 7-9EtA [(η6-p-
cymene)Os(N-2,4,6-trimethyl-Ph-picolinamide)9EtA-N7]2þ
and the fragments without 9EtG and 9EtA. The energies of the
separate fragments were subtracted from the energy of the entire
cations of complexes 1-9EtG, 1-9EtA, 7-9EtG, and 7-9EtA
in the single point calculation of all complexes to obtain the total
binding energy of 9EtG and 9EtA for both complexes.
Hydrolysis. The kinetics of hydrolysis of complexes 1-5 and
7-9 were followed by 1H NMR at 288 K. 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
Acknowledgment. We thank EPSRC National Crystallo-
graphy Service (University of Southampton) for collecting the
diffraction data for complex 1 and 3 and Dr. Michael Khan
and Dr. Ana Pizarro (Warwick University) for help and
advice on cell culture. This research was supported by the
Engineering and Physical Sciences Research Council
(EPSRC) and Science City (Advantage West Midlands/
European Regional Development Fund). We also acknowledge
our participation in the EU COST Action D39, which enabled
us to exchange regularly the most recent ideas in the field of
anticancer metallodrugs with several European colleagues.
1
(pH* of ca. 7). H NMR spectra were recorded after various
intervals using the presaturation method. The rate of hydrolysis
was determined by fitting plots of concentrations (determined by
1H NMR peak integrals) versus time to a pseudo-first-order rate
equation using ORIGIN version 5.0 (Microcal Software Ltd.).
pKa* Calculations. For determinations of pKa* values (pKa
values for solutions in D2O), the pH* values of the aqua com-
plexes of 1 and 7 in D2O (formed in situ by dissolution of
the parent chloro complexes) were varied from ca. pH* 2-10
by the addition of dilute NaOD and HNO3 and 1H 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* þ 0.42, as suggested by Krezel and
Bal,48 for comparison with related values in literature.
Cancer Cell Cytotoxicity. After plating, human ovarian
A2780 and cisplatin-resistant A2780cis cancer cells were treated
with OsII and RuII complexes on day 3, and human colon
HCT116 cancer cells on day 2, at concentrations ranging from
0.1 to 100 μM. Solutions of the OsII complexes were made up in
0.125% 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.49
Supporting Information Available: Crystallographic data and
selected bond lengths for complexes 1, 2, [6-H-6], (7 CH2Cl2),
3
and (8 CHCl3), crystallographic data and selected bond lengths
3
and angles for [(η6-p-cym)Ru(N-2-fluoro-Ph-picolinami-
de)Cl]2HPF6 ([10-H-10]PF6), X-ray structures of compounds 2
and 8, X-ray structure of 1 and 3 showing intermolecular π-π
stacking and short contact interactions, X-ray structure and
atom numbering scheme for [(η6-p-cym)Ru(N-2-fluoro-Ph-
picolinamide)Cl]2HPF6 ([10-H-10]PF6), low field region of
the 1H NMR spectrum of 7 with 3 equiv of TRISPHAT showing
the presence of two stereoisomers. Effect of temperature and of
1
pH on a mixture of the two linkage isomers of 3, H NMR
pH titration of 1 and 7, IC50 curves for compounds 1, 3, 4, and
9 in human ovarian, ovarian cisplatin-resistant, and colon
cancer cells and the preparation of the ligands and spectroscopic
data, ESI-MS, and CHN analysis of complexes 1-10. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
Interactions with Nucleobases. The reactions of 1-5 and 7-9
with model nucleobases typically involved addition of a solution
containing 1 mol equiv of nucleobase in D2O to an equilibrium
solution of the complex in D2O (>90% aqua). The pH* value of
(1) Sigel, H.; Martin, R. B. Coordinating properties of the amide bond.
Stability and structure of metal ion complexes of peptides and
related ligands. Chem. Rev. 1982, 82, 385–426.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23 7763
(2) Angel, R. L.; Fairlie, D. P.; Jackson, W. G. Linkage isomerization
of (formamide-N)- and (acetamide-N)pentaamminecobalt(III)
ions in water, dimethyl sulfoxide, and sulfolane. Inorg. Chem.
1990, 29, 20–28.
(3) Fairlie, D. P.; Angus, P. M.; Fenn, M. D.; Jackson, W. G. Factors
influencing the nitrogen vs oxygen bonding mode of amides bound
to pentaamminecobalt(III) and the kinetics and mechanism of
rearrangement. Inorg. Chem. 1991, 30, 1564–1569.
20-bipyridine: intramolecular hydrogen-bonding in complexes of
Cu(II), and a dinuclear double helicate with Ag(I). Polyhedron
1999, 18, 2633–2640.
(25) 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. Tuning the reactivity of osmium(II) and ruthenium(II) arene
complexes under physiological conditions. J. Am. Chem. Soc. 2006,
128, 1739–1748.
(4) Fairlie, D. P.; Woon, T. C.; Wickramasinghe, W. A.; Willis, A. C.
Amide-iminol tautomerism: effect of metalation. Inorg. Chem.
1994, 33, 6425–6428.
(5) Woon, T. C.; Fairlie, D. P. Amide complexes of (diethylenetri-
amine)platinum(II). Inorg. Chem. 1992, 31, 4069–4074.
(6) Woon, T. C.; Wickramasinghe, W. A.; Fairlie, D. P. Oxygen versus
nitrogen coordination of a urea to (diethylenetriamine)platinum-
(II). Inorg. Chem. 1993, 32, 2190–2194.
(7) Ilan, Y.; Kapon, M. Crystal structures of (glycinamido)- and
(glycinato)tetraammineruthenium(III) chelates. Effects of the me-
tal center electronic configuration on coordinated amido and
carboxylato groups. Inorg. Chem. 1986, 25, 2350–2354.
(8) Ilan, Y.; Taube, H. Isomeric forms of the complexes of tetra-
ammineruthenium(III) and (II) with glycinamide and derivatives.
Inorg. Chem. 1983, 22, 1655–1664.
(9) Chou, M. H.; Szalda, D. J.; Creutz, C.; Sutin, N. Reactivity and
coordination chemistry of aromatic carboxamide RC(O)NH2 and
carboxylate ligands: properties of pentaammine ruthenium(II)
and -(III) complexes. Inorg. Chem. 1994, 33, 1674–1684.
(10) Ford, P. C.; Rudd, D. P.; Gaunder, R.; Taube, H. Synthesis and
properties of pentaamminepyridineruthenium(II) and related pen-
taammineruthenium complexes of aromatic nitrogen heterocycles.
J. Am. Chem. Soc. 1968, 90, 1187–1194.
(26) Peacock, A. F. A.; Habtemariam, A.; Moggach, S. A.; Prescimone,
A.; Parsons, S.; Sadler, P. J. Chloro half-sandwich osmium(II)
complexes: influence of chelated N,N-ligands on hydrolysis, gua-
nine binding, and cytotoxicity. Inorg. Chem. 2007, 46, 4049–4059.
(27) Chen, H.; Parkinson, J. A.; Parsons, S.; Coxall, R. A.; Gould, R.
O.; Sadler, P. J. Organometallic ruthenium(II) diamine anticancer
complexes: arene-nucleobase stacking and stereospecific hydro-
gen-bonding in guanine adducts. J. Am. Chem. Soc. 2002, 124,
3064–3082.
(28) Liu, H. K.; Wang, F. Y.; Parkinson, J. A.; Bella, J.; Sadler, P. J.
Ruthenation of duplex and single-stranded d(CGGCCG) by orga-
nometallic anticancer complexes. Chem.;Eur. J. 2006, 12, 6151–
6165.
(29) Novakova, O.; Kasparkova, J.; Bursova, V.; Hofr, C.; Vojtiskova,
M.; Chen, H.; Sadler, P. J.; Brabec, V. Conformation of DNA
modified by monofunctional Ru(II) arene complexes: recognition
by DNA binding proteins and repair. relationship to cytotoxicity.
Chem. Biol. 2005, 12, 121–129.
(30) Zhang, C. X.; Lippard, S. J. New metal complexes as potential
therapeutics. Curr. Opin. Chem. Biol. 2003, 7, 481–489.
(31) Brabec, V.; Novakova, O. DNA binding mode of ruthenium
complexes and relationship to tumor cell toxicity. Drug Resist.
Updates 2006, 9, 111–122.
(11) Singh, A. K.; Balamurugan, V.; Mukherjee, R. Synthesis and
characterization of low-spin and cation radical complexes of
ruthenium(III) of a tridentate pyridine bis-amide ligand. Inorg.
Chem. 2003, 42, 6497–6502.
(12) Jacob, W.; Mukherjee, R. Synthesis, structure, and properties of
monomeric Fe(II), Co(II), and Ni(II) complexes of neutral
N-(aryl)-2-pyridinecarboxamides. Inorg. Chim. Acta 2006, 359,
4565–4573.
(13) Patra, A. K.; Mukherjee, R. Bivalent, trivalent, and tetravalent
nickel complexes with a common tridentate deprotonated pyridine
bis-amide ligand. molecular structures of nickel(II) and nickel(IV)
and redox activity. Inorg. Chem. 1999, 38, 1388–1393.
(14) Canty, A. J.; Lee, C. V. Relative σ donor ability of pyridines,
imidazoles, and pyrazoles. Inorg. Chim. Acta 1981, 54, L205–L206.
(15) Das, A.; Peng, S.-M.; Lee, G.-H.; Bhattacharya, S. Synthesis,
structure and electrochemical properties of a group of ruthenium-
(III) complexes of N-(aryl)picolinamide. New J. Chem. 2004, 28,
712–717.
(32) Jung, Y.; Lippard, S. J. Direct Cellular Responses to Platinum-
Induced DNA Damage. Chem. Rev. 2007, 107, 1387–1407.
(33) Stahl, S.; Werner, H.
A new family of (arene)osmium(0)
and (arene)osmium(II) complexes. Organometallics 1990, 9,
1876–1881.
(34) Morris, R. E.; Aird, R. E.; Murdoch, P. D.; Chen, H. M.; Cum-
mings, J.; Hughes, N. D.; Parsons, S.; Parkin, A.; Boyd, G.; Jodrell,
D. I.; Sadler, P. J. Inhibition of cancer cell growth by ruthenium(II)
arene complexes. J. Med. Chem. 2001, 44, 3616–3621.
(35) Fernandez, R.; Melchart, M.; Habtemariam, A.; Parsons, S.;
Sadler, P. L. Use of chelating ligands to tune the reactive site of
half-sandwich ruthenium(II)-arene anticancer complexes.
Chem.;Eur. J. 2004, 10, 5173–5179.
(36) Hwang, T. L.; Shaka, A. J. Water Suppression That Works;
Excitation Sculpting Using Arbitrary Wave-Forms and Pulsed-
Field Gradients. J. Magn. Reson. Ser. A 1995, 112, 275–279.
(37) Cosier, J.; Glazer, A. M. A nitrogen-gas-stream cryostat for
general x-ray diffraction studies. J. Appl. Crystallogr. 1986, 19,
105–107.
(16) Bratsos, I.; Jedner, S.; Gianferrara, T.; Alessio, E. Ruthenium
anticancer compounds: challenges and expectations. Chimia 2007,
61, 692–697.
(38) Sheldrick, G. M. SADABS version 2006-1, University of Gottingen,
Gottingen, Germany, 2006.
(17) Clarke, M. J.; Zhu, F. C.; Frasca, D. R. Non-platinum chemother-
apeutic metallopharmaceuticals. Chem. Rev. 1999, 99, 2511–2533.
(18) Melchart, M.; Sadler, P. J. Ruthenium arene anticancer complexes.
In Bioorganometallics, G. Jaouen, Ed.,Wiley-VCH: Weinheim
2006, pp 39–64.
(19) Peacock, A. F. A.; Sadler, P. J. Medicinal organometallic chem-
istry: designing metal arene complexes as anticancer agents. Chem.
Asian J. 2008, 3, 1890–1899.
(20) van Rijt, S. H.; Peacock, A. F. A.; Johnstone, R. D. L.; Parsons, S.;
Sadler, P. J. Organometallic osmium(II) arene anticancer com-
plexes containing picolinate derivatives. Inorg. Chem. 2009, 48,
1753–1762.
(21) Lee, S. A.; Eyeson, R.; Cheever, M. L.; Geng, J. M.; Verkhusha,
V. V.; Burd, C.; Overduin, M.; Kutateladze, T. G. Targeting of the
FYVE domain to endosomal membranes is regulated by a histidine
switch. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 13052–13057.
(22) Sigel, R. K. O.; Sabat, M.; Freisinger, E.; Mower, A.; Lippert, B.
Metal-modified base pairs involving different donor sites of purine
nucleobases: trans-[a(2)Pt(7,9-DimeG-N1)(9-EtGH-N7)](2þ) and
trans-[a(2)Pt(7,9-DimeG-N1)(9-EtG-N7)](þ) (a = NH3 or CH3-
NH2; 9-EtGH = 9-ethylguanine; 7,9-DimeG = 7,9-dimethylgua-
nine). Possible relevance to metalated DNA triplex structures.
Inorg. Chem. 1999, 38, 1481–1490.
(39) Sheldrick, G. M. SHELXL-97: Program for the Refinement of
Crystal Structures; University of Gottingen: Germany. 1997.
(40) 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, Nijmegen,
The Netherlands, 1996.
(41) Altomare, A.; Cascarano, G.; Giacovazzo, C.; Guagliardi, A. Early
finding of preferred orientation: a new method. J. Appl. Crystal-
logr. 1994, 27, 1045–1050.
(42) Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.;
Watkin, D. J. CRYSTALS version 12: software for guided crystal
structure analysis. J. Appl. Crystallogr. 2003, 36, 1487.
(43) Otwinowski Z.; W., M. In DENZO and SCALEPACK Programs;
Yale University: New Haven, CT. 1995.
(44) Sheldrick, G. M. Program for the Solution of Crystal Structures;
University of Gottingen: Gottingen, Germany 1985.
(45) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G. L.;
Giacovazzo, C.; Guagliardi, A.; Moliterni, A. G. G.; Polidori,
G.; Spagna, R. SIR97: a new tool for crystal structure determina-
tion and refinement. J. Appl. Crystallogr. 1999, 32, 115–119.
(46) POVray: Persistence of Vision Raytracer.6. In POVray: Persis-
tence of Vision Raytracer, 2004.
(47) Barbour, L. J. X-Seed;A software tool for supramolecular crys-
(23) Peacock, A. F. A.; Melchart, M.; Deeth, R. J.; Habtemariam, A.;
Parsons, S.; Sadler, P. J. Osmium(II) and ruthenium(II) arene
maltolato complexes: Rapid hydrolysis and nucleobase binding.
Chem.;Eur. J. 2007, 13, 2601–2613.
(24) Couchman, S. M.; Jeffery, J. C.; Ward, M. D. Synthesis and
coordination chemistry of the tetradentate ligands 6,60-bis-
(3-pyrazolyl)-2,20-bipyridine and 6,60-bis(2-hydroxyphenyl)-2,
tallography. J. Supramol. Chem. 2001, 1, 189–191.
(48) Krezel, A.; Bal, W. A formula for correlating pK(a) values deter-
mined in D2O and H2O. J. Inorg. Biochem. 2004, 98, 161–166.
(49) Skehan, P.; Storeng, R.; Scudiero, D.; Monks, A.; McMahon, J.;
Vistica, D.; Warren, J. T.; Bokesch, H.; Kenney, S.; Boyd, M. R.
New colorimetric cytotoxicity assay for anticancer-drug-screening.
J. Natl. Cancer Inst. 1990, 82, 1107–1112.

7764 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 23
van Rijt et al.
(50) Te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra,
C.; Van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. Chemistry
with ADF. J. Comput. Chem. 2001, 22, 931–967.
(51) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic regular
two-component Hamiltonians. J. Chem. Phys. 1993, 99, 4597–
610.
(52) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic total
energy using regular approximations. J. Chem. Phys. 1994, 101,
9783–9792.
equation using large components only. J. Chem. Phys. 1996, 105,
2373–2377.
(54) van Lenthe, E.; Ehlers, A.; Baerends, E.-J. Geometry optimizations
in the zero order regular approximation for relativistic effects.
J. Chem. Phys. 1999, 110, 8943–8953.
(55) Van Lenthe, E.; Van Leeuwen, R.; Baerends, E. J.; Snijders, J. G.
Relativistic regular two-component Hamiltonians. Int. J. Quantum
Chem. 1996, 57, 281–293.
(56) Baerends, E. J.; Ellis, D. E.; Ros, P. Theoretical study of the
interaction of ethylene with transition metal complexes. Theor.
Chim. Acta 1972, 27, 339–354.
(53) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Construction of the
Foldy-Wouthuysen transformation and solution of the Dirac