Highly selective binding of organometallic ruthenium ethylenediamine complexes to nucleic acids: Novel recognition mechanisms

Article abstract of DOI:10.1021/ja027719m

We have investigated the recognition of nucleic acid derivatives by organometallic ruthenium(II) arene anticancer complexes of the type [(η⁶-arene)Ru(II)(en)X] where en = ethylenediamine, arene = biphenyl (Bip), tetrahydroanthracene (THA), dihydroanthracene (DHA), p-cymene (Cym) or benzene (Ben), X = Cl^- or H₂O using ¹H, ³¹P and ¹⁵N (¹⁵N-en) NMR spectroscopy. For mononucleosides, {(η⁶-Bip)Ru-(en)}²⁺ binds only to N7 of guanosine, to N7 and N1 of inosine, and to N3 of thymidine. Binding to N3 of cytidine was weak, and almost no binding to adenosine was observed. The reactivity of the various binding sites of nucleobases toward Ru at neutral pH decreased in the order G(N7) > I(N7) > I(N1), T(N3) > C(N3) > A(N7), A(N1). Therefore, pseudo-octahedral diamino Ru(II) arene complexes are much more highly discriminatory between G and A bases than square-planar Pt(II) complexes. Such site-selectivity appears to be controlled by the en NH₂ groups, which H-bond with exocyclic oxygens but are nonbonding and repulsive toward exocyclic amino groups of the nucleobases. For reactions with mononucleotides, the same pattern of site selectivity was observed, but, in addition, significant amounts of the 5′-phosphate-bound species (40-60%) were present at equilibrium for 5′-TMP, 5′-CMP and 5′-AMP. In contrast, no binding to the phosphodiester groups of 3′, 5′-cyclic-GMP (cGMP) or cAMP was detected. Reactions with nucleotides proceeded via aquation of [(η⁶-arene)Ru(en)Cl]⁺, followed by rapid binding to the 5′-phosphate, and then rearrangement to give N7, N1, or N3-bound products. Small amounts of the dinuclear species, e.g., Ru-O(PO₃)GMPN7-Ru, Ru-O(PO₃)IMPN1-Ru, Ru-O(PO₃)TMPN3-Ru, Ru-N7IMPN1-Ru, and Ru-N7InoN1-Ru were also detected. In competitive binding experiments for [(η⁶-Bip)Ru(en)Cl]⁺ with 5′-GMP versus 5′-AMP or 5′-CMP or 5′-TMP, the only final adduct was [(η⁶-Bip)Ru(en)(N7-GMP)]. Ru-H₂O species were more reactive than Ru-OH species. The presence of Cl^- or phosphate in neutral solution significantly decreased the rates of Ru-N7 binding through competitive coordination to Ru. In kinetic studies (pH 7.0, 298 K, 100 mM NaClO₄), the rates of reaction of cGMP with {(η⁶-arene)Ru(II)(en)X}ⁿ⁺ (X = Cl^- or H₂O) decreased in the order: THA > Bip > DHA ? Cym > Ben, suggesting that N7-binding is promoted by favorable arene-purine hydrophobic interactions in the associative transition state. These findings have revealed that the diamine NH₂ groups, the hydrophobic arene, and the chloride leaving group have important roles in the novel mechanism of recognition of nucleic acids by Ru arene complexes, and will aid the design of more effective anticancer complexes, as well as new site-specific DNA reagents.

Full text of DOI:10.1021/ja027719m

Published on Web 11/26/2002
Highly Selective Binding of Organometallic Ruthenium
Ethylenediamine Complexes to Nucleic Acids: Novel
Recognition Mechanisms
Haimei Chen, John A. Parkinson, Robert E. Morris, and Peter J. Sadler*
Contribution from the Department of Chemistry, UniVersity of Edinburgh, West Mains Road,
Edinburgh EH9 3JJ, United Kingdom
Received July 16, 2002; E-mail: p.j.sadler@ed.ac.uk
Abstract: We have investigated the recognition of nucleic acid derivatives by organometallic ruthenium(II)
arene anticancer complexes of the type [(η6-arene)Ru(II)(en)X] where en ) ethylenediamine, arene )
biphenyl (Bip), tetrahydroanthracene (THA), dihydroanthracene (DHA), p-cymene (Cym) or benzene (Ben),
1
X ) Cl- or H2O using H, 31P and 15N (15N-en) NMR spectroscopy. For mononucleosides, {(η6-Bip)Ru-
(en)}2+ binds only to N7 of guanosine, to N7 and N1 of inosine, and to N3 of thymidine. Binding to N3 of
cytidine was weak, and almost no binding to adenosine was observed. The reactivity of the various binding
sites of nucleobases toward Ru at neutral pH decreased in the order G(N7) > I(N7) > I(N1), T(N3) >
C(N3) > A(N7), A(N1). Therefore, pseudo-octahedral diamino Ru(II) arene complexes are much more
highly discriminatory between G and A bases than square-planar Pt(II) complexes. Such site-selectivity
appears to be controlled by the en NH2 groups, which H-bond with exocyclic oxygens but are nonbonding
and repulsive toward exocyclic amino groups of the nucleobases. For reactions with mononucleotides, the
same pattern of site selectivity was observed, but, in addition, significant amounts of the 5′-phosphate-
bound species (40-60%) were present at equilibrium for 5′-TMP, 5′-CMP and 5′-AMP. In contrast, no
binding to the phosphodiester groups of 3′, 5′-cyclic-GMP (cGMP) or cAMP was detected. Reactions with
nucleotides proceeded via aquation of [(η6-arene)Ru(en)Cl]+, followed by rapid binding to the 5′-phosphate,
and then rearrangement to give N7, N1, or N3-bound products. Small amounts of the dinuclear species,
e.g., Ru-O(PO3)GMPN7-Ru, Ru-O(PO3)IMPN1-Ru, Ru-O(PO3)TMPN3-Ru, Ru-N7IMPN1-Ru, and Ru-
N7InoN1-Ru were also detected. In competitive binding experiments for [(η6-Bip)Ru(en)Cl]+ with 5′-GMP
versus 5′-AMP or 5′-CMP or 5′-TMP, the only final adduct was [(η6-Bip)Ru(en)(N7-GMP)]. Ru-H2O species
were more reactive than Ru-OH species. The presence of Cl- or phosphate in neutral solution significantly
decreased the rates of Ru-N7 binding through competitive coordination to Ru. In kinetic studies (pH 7.0,
298 K, 100 mM NaClO4), the rates of reaction of cGMP with {(η6-arene)Ru(II)(en)X}n+ (X ) Cl- or H2O)
decreased in the order: THA > Bip > DHA . Cym > Ben, suggesting that N7-binding is promoted by
favorable arene-purine hydrophobic interactions in the associative transition state. These findings have
revealed that the diamine NH2 groups, the hydrophobic arene, and the chloride leaving group have important
roles in the novel mechanism of recognition of nucleic acids by Ru arene complexes, and will aid the
design of more effective anticancer complexes, as well as new site-specific DNA reagents.
Introduction
trostatic interactions, can effectively enhance site- and base-
selective recognition of nucleic acids. Selective coordination
Studies of the molecular recognition of DNA and RNA are
of fundamental importance in post-genomic chemistry. The
rational design of new DNA-binding agents that recognize
specific sequences or structures, and can modify specific DNA
functions such as replication and transcription, provides an
effective approach for development of novel chemotherapeutic
anticancer drugs.1 Noncovalent interactions play key roles in
biological molecular recognition.2 Metal coordination, combined
with, for example, hydrogen bonding, hydrophobic and elec-
to nucleobases is influenced by structural and electronic factors
arising from the nucleobase and the complex itself.3 Both
hydrogen-bonding and nonbonding repulsive interactions be-
tween the exocyclic groups on the bases and the chelate ligands
in metal complexes can lead to selective reactions. Hydrophobic
interactions can increase DNA affinity and give rise to highly
specific recognition of DNA base sequences.4 Zn(II) macrocyclic
tetraamine (1,4,7,10-tetraazacyclododecane, cyclen) complexes
with quinolyl or naphthyl pendants selectively bind to AT-rich
regions of DNA mainly via Zn(II)-N(3)(thymine) coordination,5
(1) (a) Chaires, J. B. Curr. Opin. Struct. Biol. 1998, 314-320. (b) Jenkins, T.
C. Curr. Med. Chem. 2000, 7, 99-115.
(2) (a) Mu¨ller-Dethlefs, K.; Hobza, P. Chem. ReV. 2000, 100, 143-167. (b)
ComprehensiVe Supramolecular Chemistry; Lehn, J.-M., Atwood, J. L.,
Davies, J. E. D., MacNicol, D. D., Vo¨gtle, F., Eds.; Pergamon: Oxford,
1996.
(3) (a) Martin, R. B. Acc. Chem. Res. 1985, 18, 32-38. (b) Marzilli, L. G.;
Kistenmacher, T. J. Acc. Chem. Res. 1977, 10, 146-152.
(4) Erkkila, K. E.; Odom, D. T.; Barton, J. K. Chem. ReV. 1999, 99, 2777-
2795.
9
10.1021/ja027719m CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 173-186
173

A R T I C L E S
Chen et al.
Chart 1. Structures and Atom Numbering for Nucleobases and
whereas incorporation of anthraquinone pendants into the cyclen
ring allows recognition of G bases via a dominant π-π stacking
interaction.6 There is precedent for incorporating potential DNA
intercalators into metal anticancer complexes. Examples include
[Pt(AO)-(CH2)6-(en)Cl2]+, cis-[Pt(NH3)2(N9-AA)] (AO ) acri-
dine orange; N9-AA ) 9-aminoacridine),7,8 and trans-[PtCl2-
(NH3)L] (L ) planar aromatic N donor such as pyridine,
thiazole, or quinoline),9 where the additional intercalation or
hydrophobic interaction after covalent binding of the Pt to
nucleobases can lead to enhanced activity.
Their Derivatives.
We have found that organometallic ruthenium(II) complexes
of the type [(η6-arene)RuCl(X)(Y)] (X, Y are monodentate or
chelating ligands) are cytotoxic to cancer cells, including
cisplatin-resistant cell lines. Preliminary structure-activity data
have shown that compounds are active when X, Y belong to a
chelating diamine such as ethylenediamine (en) or its derivatives
containing NH groups in the amine. The complex [(η6-Bip)-
Ru(en)Cl][PF6] is active in vivo against the A2780 xenograft
model of human ovarian cancer, and is also active against
A2780cis, the cisplatin-resistant xenograft.10 The cytotoxicity
increases with increase in size of the η6-arene. For example,
complexes which contain η6-tetra- or di-hydroanthracene rings
(THA, DHA) or biphenyl (Bip) have much higher activities than
compounds containing a single ring such as η6-benzene (Ben)
or η6-cymene (Cym).10,11 A potential target for these organo-
metallic Ru(II) complexes is DNA, and the preferred binding
site appears to be guanine.11 We have studied the structures of
guanine adducts of THA, DHA, and Bip complexes and found
stereospecific hydrogen-bonding of en NH with G O6, and
strong arene-nucleobase stacking, both in the solid state by X-ray
crystallography, and in solution by NMR methods.12,13 These
studies have revealed several DNA recognition features and
suggested a potential DNA-binding mode, involving simulta-
neous intercalation and covalent coordination. The diamine NH
group and hydrophobic arene group as well as the Cl leaving
group may therefore all play important roles in the anticancer
activity.
Chart 2. Arene Complexes Studied in This Work
It is now important to understand how stereospecific hydrogen-
bonding can lead to specific DNA site-recognition and whether
arene-purine hydrophobic interactions influence the kinetics
of DNA binding. Therefore, we have studied both the thermo-
dynamics and kinetics of the binding of [(η6-arene)Ru(en)X]n+
complexes (arene ) tetra-, di-hydroanthracene (THA, DHA),
biphenyl (Bip), p-cymene (Cym) and benzene (Ben); X ) Cl
or H2O) to mononucleotides and mononucleosides (G, I, A, C,
and T) using 1H, 31P, 15N (15N-en) NMR spectroscopy, together
with pH titrations. These studies show that these novel orga-
nometallic ruthenium ethylenediamine anticancer complexes are
highly selective in their recognition of nucleic acid bases and
that the nature of the coordinated arene has a dramatic effect
on the kinetics of binding. Such insights can direct new synthetic
work for optimization of drug activity.
(5) Kikuta, E.; Murata, M.; Katsube, N.; Koike, T.; Kimura, E. J. Am. Chem.
Soc. 1999, 121, 5426-5436.
(6) Kikuta, E.; Matsubara, R.; Katsube, N.; Koike, T.; Kimura, E. J. Inorg.
Biochem. 2000, 82, 239-249.
(7) Bowler, B. E.; Ahmed, K. J.; Sundquist, W. I.; Hollis, L. S.; Whang, E.
E.; Lippard, S. J. J. Am. Chem. Soc. 1989, 111, 1299-1306.
(8) Sundquist, W. I.; Lippard, S. J. Coord. Chem. ReV. 1990, 100, 293-322.
(9) (a) Farrell, N.; Ha, T. T. B.; Souchard, J.-P.; Wimmer, F. L.; Cros, S.;
Johnson, N. P. J. Med. Chem. 1989, 32, 2240-2241. (b) Farrell, N. Met.
Ions Biol. Syst. 1996, 32, 603-639. (c) Bierbach, U.; Qu, Y.; Hambley, T.
W.; Peroutka, J.; Nguyen, H. L.; Doedee, M.; Farrell, N. Inorg. Chem.
1999, 38, 3535-3542. (d) Za´kovska´, A.; Nova´kova´, O.; Balcarova´, Z.;
Bierbach, U.; Farrell, N.; Brabec, V. Eur. J. Biochem. 1998, 254, 547-
557.
(10) Aird, R. E.; Cummings, J.; Ritchie, A. A.; Muir, M.; Morris, R. E.; Chen,
H.; Sadler, P. J.; Jodrell, D. I. Brit. J. Cancer 2002, 86, 1652-1657.
(11) Morris, R. E.; Aird, R. E.; Murdoch, P. del S.; Chen. H.; Cummings, J.;
Hughes, N. D.; Parsons, S.; Parkin, A.; Boyd, G.; Jodrell, D. I.; Sadler, P.
J. J. Med. Chem. 2001, 44, 3616-3621.
(12) Chen, H.; Parkinson, J. A.; Parsons, S.; Coxall, R. A.; Gould, R. O.; Sadler,
P. J. J. Am. Chem. Soc. 2002, 124, 3064-3082.
(13) A similar G C6O‚‚‚HN H-bond is present in the X-ray structure of
[(η6-benzene)Ru(L-Ala)(9EtG)]Cl‚2H2O: Sheldrick, W. S.; Heeb, S. Inorg.
Chim. Acta 1990, 168, 93-100.
Experimental Section
Materials. Chloro complexes [(η6-arene)Ru(en)Cl][PF6] (arene )
Bip, THA, DHA, Cym, and Ben, Chart 2) were synthesized as described
previously.11,12 The mono-nucleotides 5′-GMP, 3′-GMP, 3′, 5′-cyclic
GMP, 5′-IMP, 5′-AMP, 3′, 5′-cyclic AMP, 5′-CMP, and 5′-TMP (all
as sodium salts), and the nucleosides guanosine, inosine, adenosine,
cytidine, and thymidine were purchased from Sigma or Aldrich. The
1
purity was checked by H and 31P NMR spectroscopy. D2O (99.98%)
was obtained from Aldrich. All NMR experiments were performed in
10% D2O/90% H2O.
9
174 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
Scheme 1. Reaction of [(η6-Bip)Ru(en)Cl]+ with 5′-GMP (5 mM,
1:1) in Aqueous Solution at pH ) 7.15, 298 Ka
For reactions of [(η6-Bip)Ru(en)Cl]+ with 5′-AMP, 10 parallel
experiments were performed in ten NMR tubes under the similar
conditions with various pH values over the range of 3-10. For each
sample, the mixture of [(η6-Bip)Ru(en)Cl]+ and 5′-AMP (1:1, 5 mM)
with an adjusted pH was incubated at 310 K for 24 h, after which the
pH value was determined again and used as the equilibrium pH value.
Subsequently, 1H and 31P NMR spectra were recorded. The amount of
phosphate-bound adduct at each of the 10 equilibrium pH values was
1
determined by integration of either H or 31P NMR peaks.
Competitive Reactions of 5′-GMP, 5′-AMP, 5′-CMP, and 5′-TMP
with [(η6-Bip)Ru(en)Cl]+. 5′-GMP was mixed with one of three other
nucleotides in a 1:1 mol ratio (5 mM, GMP + AMP, GMP + CMP, or
GMP + TMP). Each of the combinations was then added to an aqueous
solution of [(η6-Bip)Ru(en)Cl]+ (final concentration 5 mM). The
1
reaction mixtures were monitored immediately by H and 31P NMR
spectroscopy at 298 or 310 K. The amounts of free and Ru-bound
nucleotides were determined by peak integration.
Kinetic Studies. The reactions between chloro or aqua Ru(arene)
complexes and nucleotides (5′-GMP, cGMP, 5′-IMP, 5′-AMP, cAMP,
5′-CMP, or 5′-TMP), or nucleosides (guanosine, inosine, adenine,
cytosine, or thymidine), were carried out at a concentration of 5 mM
Ru and 5 mM nucleobase in 10% D2O/90% H2O in NMR tubes. Most
solutions were unbuffered to avoid interaction of buffer components
with Ru(II). The solution pH was determined at the commencement of
the reaction and after the establishment of equilibrium. The kinetic
studies of various chloro or aqua Ru(arene) complexes with cGMP
were carried out in 100 mM NaClO4 at 298 K, pH 7.0. Due to the
limited aqueous solubility of chloro Ru(THA) and Ru(DHA), the
kinetics of reactions of various chloro Ru(arene) complexes with cGMP
a Species distribution (%) is shown after 55 min reaction, and is based
1
on 15NH2-en H NMR intensities (except for the dinuclear species which
was based on H8 intensities). The intermediate Ru-O-O(PO3)GMP (peak
b) formed during the reaction mixing time, slowly rearranged into Ru-
N7GMP (peak c), and then eventually disappeared after 22 h (see Figure
1A). The intermediate Ru-O(PO3)GMPN7-Ru (peak d) was evident after
2 h, and eventually disappeared after 22 h; it persisted if Ru was present in
excess. When more Ru-Cl (50%) was added to the final mixture, more of
species d (30%) formed after 24 h at 310 K.
were performed at a concentration of 2 mM for both reactants. 1H, 31
P
NMR, or 15N NMR data were recorded at appropriate time intervals.
The first spectrum was acquired 10-15 min after mixing of the
reactants, to allow time for temperature equilibration of the sample in
the probe. Quantitation of the reactions was achieved by integration of
the peaks for the base protons H8 (G), H8 and H2 (I and A), H6/H5
(C), and C(5)H3 (T), or by integration of 31P NMR peaks, or [1H, 15N]
cross-peaks.
Preparation of Aqua Ru(arene) Complexes. An aliquot of a 100
mM aqueous solution of AgNO3 (0.98 mol equiv) was added to a
solution of [(η6-arene)Ru(en)Cl][PF6], giving a yellow-orange solution
and an immediate precipitate of AgCl. This mixture was kept at ambient
temperature overnight (protected from light) and was then filtered to
remove AgCl. The resulting yellow solutions14 were used freshly and
directly for reactions with 5′-GMP and cGMP.
Methods and Instruments. NMR Spectroscopy. 1H NMR spectra
were acquired on a Bruker DMX 500 (1H ) 500 MHz) NMR
spectrometer using TBI [1H, 13C, X] or triple resonance [1H, 13C, 15N]
probe-heads, respectively, and equipped with z-field gradients.
pH Titrations. For the reaction of [(η6-Bip)Ru(en)Cl]+ (12 mM)
1D 1H NMR spectra were typically acquired with 64 transients into
32 K data points over a spectral width of 6.0 kHz. Water suppression
was achieved by presaturation. 31P NMR spectra were obtained on a
Bruker DMX 500 (31P ) 202 MHz) NMR spectrometer at 298 K, using
inverse-gated 1H decoupling. Typically 1D 31P-{1H} NMR spectra were
acquired with 512 transients into 42 K data points over a spectral width
of 40.7 kHz (200.8 ppm) using a relaxation delay of 1.3 s. Both 1D
with 5′-GMP (12 mM), 1H and 31P NMR pH titrations were performed
1
at 298 K within the first 6 h, to record H and 31P NMR resonances
for the intermediates Ru-O(PO3)GMP and Ru-O(PO3)GMPN7-Ru
(Scheme 1).
For the reaction of [(η6-Bip)Ru(en)Cl]+ (8 mM) with cGMP (9 mM),
the mixture was incubated at 298 K for 24 h. This sample was then
used directly for recording 1H and 31P NMR spectra at various pH
values.
1
15N-edited H NMR spectra and 2D [1H, 15N] HSQC NMR spectra
1
(optimized for J(NH) ) 73 Hz) were acquired using the sequence of
For the reaction of [(η6-Bip)Ru(en)Cl]+ (12 mM) with 5′-IMP (14
mM), the mixture was incubated at 298 K for 24 h, followed by addition
of a 14% molar excess of Ru. This mixture was used for a pH titration,
Stonehouse et al.15 A GARP decoupling sequence was used for 15N
decoupling during acquisition. Typically 1D 1H-{15N} NMR data were
acquired with 16 transients into 1024 data points over a spectral width
of 2.5 kHz (5.0 ppm) with a relaxation delay of 2.0 s. 2D [1H, 15N]
HSQC NMR data were acquired with 4 transients into 1024 data points
over a 1H (F2) spectral width of 2.5 kHz (5 ppm, centered at the water
resonance) for each of 256 t1 increments (TPPI), and over a 15N (F1)
spectral width of 0.76 kHz (15 ppm, centered at -28 ppm) using a
1
which was completed within a further 4 h. The resultant H and 31P
NMR spectra were recorded. The H8 and H2 chemical shifts of all six
IMP species were subsequently plotted as a function of pH.
For the reaction of [(η6-Bip)Ru(en)Cl]+ (5 mM) with inosine (6 mM),
1
the mixture was incubated at 298 K for 12 h. H NMR spectra were
then recorded at various pH values.
1
relaxation delay of 1.4 s between transients. Typically, 1D H-{15N}
For reactions of [(η6-Bip)Ru(en)Cl]+ (10 mM) with 5′-CMP, 5′-
TMP, or thymidine (10 mM), the mixtures were incubated at 310 K
NMR spectra were acquired over a 10 min period and 2D [1H, 15N]
HSQC NMR spectra were acquired over a 30 min period.
1
for 48 h, and H and 31P NMR spectra were then recorded at various
All data processing was carried out using XWIN NMR version 2.0
(Bruker U.K. Ltd.). 1H NMR chemical shifts were internally referenced
pH values.
1
(14) The purity of the aqua Ru(arene) complexes was confirmed by H NMR.
X-ray structures of the aqua adducts have been determined and will be
reported elsewhere.
(15) Stonehouse, J.; Shaw, G. L.; Keller, J.; Laue, E. D. J. Magn. Reson. Ser.
A, 1994, 107, 178-184.
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 175

A R T I C L E S
Chen et al.
to 1,4-dioxane (3.77 ppm) or to the methyl singlet of TSP (0 ppm) in
aqueous solutions. 31P resonances were referenced to 85% H3PO4
(external) at 0 ppm, and 15N resonances to 1 M 15NH4Cl in 1.5 M HCl
(external) at 0 ppm.
pH Measurements. The pH values of NMR samples in 10% D2O/
90% H2O were measured at 298 K directly in the NMR tube, before
and after recording NMR spectra, using a Corning 240 pH meter
equipped with an Aldrich micro combination electrode calibrated with
Aldrich buffer solutions at pH 4, 7, and 10. The pH values were adjusted
with dilute HClO4 and NaOH. No correction has been applied for the
effect of deuterium on the glass electrode.
Calculation of pKa Values. The pH titration curves were fitted to
the Henderson-Hasselbalch equation using the program KALEIDA-
GRAGH,16 with the assumption that the observed chemical shifts are
weighted averages according to the populations of the protonated and
deprotonated species.
Determination of Rate Constants. The reactions of aqua Ru(arene)
complexes with cGMP (1:1, 5 mM) (Ru-H2O + cGMP f Ru-
N7cGMP) were analyzed as second-order processes (see Figure S8).
The second-order rate constants (k) were determined by fitting kinetic
curves of concentration versus reaction time for formation Ru-N7cGMP
using the appropriate equation and the program Scientist (see Figure
S9).
Results
Figure 1. (A) 1H, and (B) 31P NMR spectra for the reaction of [(η6-Bip)-
Ru(en)Cl]+ with 5′-GMP (1:1, 5 mM, pH 7.15, 298 K) at various reaction
times. Plots of (C) H8 1H NMR chemical shift, and (D) 31P NMR chemical
shift vs pH for 5′-GMP species a-d. The curves are computer-fitted with
the pKa values listed in Table 1. The pH titration was performed for a similar
reaction of [(η6-Bip)Ru(en)Cl]+ (12 mM) with 5′-GMP (12 mM). Peak
assignments: a, free 5′-GMP; b, Ru-O(PO3)GMP; c, Ru-N7GMP and d,
Ru-O(PO3)GMPN7-Ru. See Scheme 1 for structures.
To investigate the specificity of the binding of {(η6-arene)-
Ru(en)}2+ to nucleobases, we have studied reactions of the
anticancer complex [(η6-Bip)Ru(en)Cl]+ with various nucleo-
1
sides and nucleotides (see Chart 1) in aqueous solution by H,
15N, and 31P NMR spectroscopy. The binding sites for {(η6-
Bip)Ru(en)}2+ on nucleobase derivatives were identified by 1H
NMR and 31P NMR pH titrations, and each adduct was
characterized by pKa values derived from the pH titration curves.
Some competition experiments were also performed to study
the relative binding preferences of {(η6-Bip)Ru(en)}2+ toward
mononucleotides. We have also investigated the influence of
pH, chloride, and phosphate on the rate of binding of {(η6-
arene)Ru(en)X}n+ (arene ) Bip, THA, DHA, Cym, or Ben, and
Table 1. Selected Chemical Shifts and pKa Values for 5′-GMP
and cGMP and Their {(η6-Bip)Ru(en)}2+ Adducts at 298 K
speciesa (peak)b
pK (group)
a
δ (H8)c
δ (31P)c
5′-GMP (a)
9.81 ( 0.10 (N1H)
6.49 ( 0.18 (-PO3H)
2.57 ( 0.01 (N7)
9.51 ( 0.05 (N1H)
8.39 ( 0.06 (N1H)
5.77 ( 0.04 (-PO3H)
8.175
3.899
Ru-O(PO3)GMP (b)
Ru-N7GMP (c)
8.027
8.733
9.017
4.582
1
X ) Cl or H2O) complexes to 5′-GMP and cGMP by H and
31P NMR. The detailed kinetics for the reactions with cGMP
were analyzed.
Ru-O(PO3)GMPN7-Ru (d) 8.07 ( 0.04 (N1H)
cGMP (a)
8.453
7.884 -0.979
9.237
9.67 ( 0.24 (N1H)
2.08 ( 0.02 (N7)
8.13 ( 0.03 (N1H)
Reactions of [(η6-Bip)Ru(en)Cl]+ with Nucleotides and
Nucleosides. Guanine Forms only N7 Adducts. Phosphate-
bound Intermediates Detected for 5′-GMP but not for
cGMP. The H8 region of the NMR spectrum of 5′-GMP after
reaction with [(η6-Bip)Ru(en)Cl]+ (1:1, 5 mM, pH 7.15, 298
K) for 0.25-22 h is shown in Figure 1A. Three new H8 signals
were observed: peak b (δ 8.027) appeared rapidly (during the
reaction mixing time) and was shifted to low frequency relative
to free 5′-GMP (peak a, δ 8.175), but disappeared at later times;
peak d at higher frequency (δ 8.453) was evident after 2 h and
finally disappeared as well; peak c appeared at higher frequency
(δ 8.733), increased in intensity with time, and was the only
final product observed for this reaction. An identical reaction,
which was followed by 31P NMR (Figure 1B), revealed corre-
sponding changes in intensities of 31P NMR peaks (a-d) which
Ru-N7cGMP (b)
8.221 -1.022
a For structures see Schemes 1 and 2. b For peak labels see Figures 1
and 2. c For 5′-GMP, pH 7.20 and for cGMP, pH 7.28.
To identify the intermediates and products, a pH titration on
a similar solution containing 12 mM Ru and 12 mM 5′-GMP
was carried out during the first 6 h of reaction. Peaks b and d
disappeared at pH values < 4, but the other peaks could be
followed over the pH range 2-12. The 1H and 31P NMR titration
curves are shown in Figure 1C and 1D, and associated pKa
values were determined by computer fits to the Henderson-
Hasselbalch equation. The characteristic pKa values associated
with peaks a, b, c, and d, allow the intermediates to be assigned
as Ru-O(PO3)GMP (b), Ru-O(PO3)GMPN7-Ru (d), and final
product as Ru-N7GMP (c) (for structures see Scheme 1). Their
pKa values are listed in Table 1. Three pKa values were
determined for free 5′-GMP: 9.81, 6.49, and 2.57, attributable
to the deprotonation of N1H, PO3H, and N7, respectively. Ru-
O(PO3)GMP (peak b) undergoes deprotonation of N1H (pKa
of 9.51), but no deprotonation of PO3H. Ru-N7GMP (peak c)
undergoes deprotonation of PO3H (pKa of 5.77) and N1H (pKa
were consistent with the H NMR results. However, for 31P,
1
there were large shifts in peaks b (δ 9.017) and d (δ 9.237) to
high frequency relative to the resonance for free 5′-GMP (peak
a, δ 3.899), compared with the slight high-frequency shift of
peak c (δ 4.482). The reactions were complete within ca. 22 h.
(16) KALEIDAGRAPH, version 3.09; Synergy Software: Reading, PA, 1997.
9
176 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
Scheme 2. Scheme for Reaction of [(η6-Bip)Ru(en)Cl]+ with
cGMP in Aqueous Solution at pH ) 7.28, 298 K
for peak a (free cGMP), and a single pKa value of 8.13 for the
1
product peak b, as derived from the H8 H curves (Figure 2C
and Table 1). Peak b can be assigned as Ru-N7cGMP (see
Scheme 2). pKa values of 9.67 (peak a) and 8.13 (peak b) are
attributable to deprotonation of N1H. The pKa value of 2.08
(peak a) is attributable to protonation of N7. Notable is the
lowering of the pKa for N1 by 1.54 units relative to free cGMP
(peak b). During the reaction, no phosphate-bound adducts were
detected. The reaction course is shown in Scheme 2. After 10
h, >95% of the cGMP was present as Ru-N7cGMP.
Figure 2. (A) 1H, and (B) 31P NMR spectra for the reaction of [(η6-Bip)-
Ru(en)Cl]+ with cGMP (1:1, 5 mM, pH 7.28, 298 K) at various reaction
times. Peak assignments: a, free cGMP; b, Ru-N7cGMP. See Scheme 2
The reaction of guanosine (Guo) with [(η6-Bip)Ru(en)Cl]+
1
1
(1:1, 5 mM, 310 K, pH 7.20) was followed by H NMR for 1
for structures. (C) Plots of H8 H NMR chemical shifts of cGMP species
d. Only one new peak assignable12 to Ru-N7Guo appeared at
high frequency (δ 8.22) relative to free Guo (δ 7.9), and after
14 h, > 97% of Guo was present as Ru-N7Guo. A similar
reaction was performed at pH 9.29, and only the Ru-N7Guo
adduct was present after 14 h (Table S1).
a and b vs pH for a similar reaction with 8 mM Ru and 9 mM cGMP. The
curves are computer best fits giving the pKa values listed in Table 1.
of 8.39), but no deprotonation of N7. Ru-O(PO3)GMPN7-
Ru (peak d) undergoes deprotonation of N1H, but no depro-
tonation of PO3H and N7. Notable is the lowering of the
pKa(N1) value by 1.42 units for Ru-N7GMP (peak c) and by
1.74 units for Ru-O(PO3)GMPN7-Ru (peak d) compared to
free GMP (peak a).
A similar reaction with 5′-GMP was also monitored by 2D
[1H, 15N] HSQC NMR by use of 15N-labeled [(η6-Bip)Ru(en)-
Cl]+ (Figure S1). Four species labeled Ru-Cl, Ru-H2O, Ru-
O(PO3)GMP, and Ru-N7GMP were detected during the time
course of the reaction (Figures S1 and S2). This result, together
with the 1H and 31P NMR time courses, allowed reaction
pathways to be drawn up, as shown in Scheme 1. First, the
chloro complex Ru-Cl undergoes hydrolysis in water to give
the reactive aqua complex Ru-H2O, followed by rapid reaction
with the 5′-phosphate of 5′-GMP, and then a slow displacement
of bound phosphate by N7. The species distribution determined
after 55 min of reaction is shown in Scheme 1.
Ru(II) Binds to N7 and N1 of Inosine, and Phosphate-
Bound Intermediates are Formed with 5′-IMP. The reaction
between 5′-IMP and [(η6-Bip)Ru(en)Cl]+ (1:1, 5 mM, pH 7.26)
1
was monitored by H and 31P NMR spectroscopy. During the
reaction, the pH value dropped from 7.26 (when mixed) to 6.55
1
(after 3 d at 310 K). The H8 and H2 regions of the H NMR
spectra of 5′-IMP at different reaction times are shown in Figures
3A-C. 31P NMR peak assignments are given in Table 2. During
the course of the reaction, 5 new sets of H8 and H2 peaks (peaks
b, c, d, e, and f) were seen. Of these, 3 sets (peaks c, d, and e)
are due to intermediates: their intensities decreased at later
reaction times and almost disappeared after 3 d at 310 K. The
1
intermediates and products were identified by a H NMR pH
titration on a similar solution (Figure 3D). It can be seen that
peaks c, d, and e disappeared at lower pH (<5.8). From 1H H8
shifts, the pKa values associated with all 6 sets of IMP peaks
were determined and are listed in Table 2. For free 5′-IMP (a),
pKa values of 9.22, 6.36, and 1.35 were determined for N1H,
PO3H, and N7, respectively, whereas peak b has two associated
pKa values of 7.74 and 5.71 assignable to N1H and PO3H, but
no pKa assignable to N7. Peak b can therefore be assigned as
Ru-N7IMP. Peak c has an N1H pKa value of 9.04 and no
associated phosphate pKa, and is assigned to Ru-O(PO3)IMP.
Peak d has a PO3H pKa of 6.35 but no N1H pKa, and is assigned
to Ru-N1IMP. Peak e has no measurable pKa values at all and
is assignable to Ru-O(PO3)IMPN1-Ru. Peak f has a PO3H
The reaction of 3′, 5′-cyclic GMP (cGMP) with [(η6-Bip)-
Ru(en)Cl]+ (1:1, 5 mM, pH 7.28, 298 K) was monitored by 1H
and 31P NMR for a period of ca. 12 h. After 11 min, a new H8
1H NMR peak (b, δ 8.221) was observed, shifted to high
frequency by 0.337 ppm compared to free cGMP (peak a, δ
7.884), Figure 2A. This was the only new H8 signal observed,
and after 12 h was the only H8 peak in the spectrum. In the 31
P
spectrum (Figure 2B), only one new peak appeared (b, δ
-1.022). This peak was shifted to low frequency by only 0.043
ppm relative to that for cGMP (peak a, δ -0.979). A pH titration
of a similar reaction solution gave pKa values of 9.67 and 2.08
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 177

A R T I C L E S
Chen et al.
Scheme 3. Scheme for Reaction of [(η6-Bip)Ru(en)Cl]+ with
5′-IMP (1:1, 5 mM) in Aqueous Solution at pH ) 7.28, 298 K.
Species Distribution (%) is Shown after 24 h Reaction at 310 K,
Based on H8 and H2 Intensities
Table 2. Selected Chemical Shifts and pKa Values for 5′-IMP and
Inosine (Ino) and Their {(η6-Bip)Ru(en)}2+ Adducts at 298 K
speciesa (peak)b
pK (group)c
δ (H8)d
δ (H2)d
a
5′-IMP (a)
9.22 ( 0.05 (N1H)
6.36 ( 0.05 (PO3H)
1.35 ( 0.01 (N7)
7.74 ( 0.03 (N1H)
5.71 ( 0.03 (PO3H)
9.04 ( 0.03 (N1H)
6.35 ( 0.05 (PO3H)
8.497
8.169
Ru-N7IMP (b)
9.053
8.131
Ru-O(PO3)IMP (c)
Ru-N1IMP (d)
Ru-O(PO3)IMPN1-Ru (e)
Ru-N7IMPN1-Ru (f)
Ino (a)
8.404
8.339
8.330
8.856
8.344
7.909
8.016
7.99
8.155
8.225
5.73 ( 0.03 (PO3H)
8.79 ( 0.02 (N1H)
0.909 ( 0.003 (N7)
7.11 ( 0.04 (N1H)
Ru-N7Ino (b)
8.745
8.135
8.624
8.201
7.977
8.165
Ru-N1Ino (c)
Ru-N7Ino N1-Ru (e)
a For structures, see Schemes 3 and 4. b For peak labels, see Figures 3
and 5. c On the basis of H8 chemical shifts. d pH 6.31 for 5′-IMP and pH
6.19 for Ino. For 5′-IMP, the corresponding 31P NMR resonances appeared
at δ 3.278 for peak a, δ 8.983 for c, and δ 8.272 for e, δ 3.410, 4.393, and
1.612 for peaks b, d, and f, but specific assignments were not made.
6.46 after 24 h at 310 K was determined by H8 and H2 peak
integration and is shown in Scheme 3.
The reaction of inosine (Ino, 5.5 mM) with [(η6-Bip)Ru(en)-
1
Cl]+ (5 mM) was followed by H NMR for 2 d at 310 K with
an initial pH of 7.22. After 1 d, the pH value of the solution
had decreased to 6.19. A representative 1H NMR spectrum for
the H8 and H2 region is shown in Figure 4A. Three products
are formed corresponding to 3 new sets of H8 and H2 peaks:
b, d and c. The products were identified by 1H NMR pH titration
(Figure 4B, note that product c disappeared at pH < 4.0). The
pKa values are derived from H8 chemical shifts17 and are listed
in Table 2. Peak a has two associated pKa values of 8.79 and
0.91, attributable to protonation of N1H and N7 of free inosine.
Peak b has one pKa value of 7.11, attributable to the protonation
of N1H of Ru-N7Ino. After 24 h at 310 K, the species
distribution was determined to be: Ru-N7Ino: 59.4%, Ru-
N1Ino: 23.0%, Ru-N7InoN1-Ru: 3.7% (Scheme 4). Species
distributions and N(7)/N(1) binding ratios for reactions of Ino
with [(η6-Bip)Ru(en)Cl]+ at various pH values (5.5-10.3) are
listed in Table S1. With increase in pH, the N(7)/N(1) binding
ratio decreased significantly from 7.72 (pH 5.5) to 0.22 (pH
10.3).
Figure 3. H8 and H2 region of 1H NMR spectra for the reaction of 5′-
IMP with [(η6-Bip)Ru(en)Cl]+ (1:1, 5 mM, initial pH 7.26) at 298 K after
(A) 13 min (pH 7.15), (B) 3.7 h (pH 6.74), and (C) after a further 3 d at
310 K (pH 6.55). (D) Plots showing the variation with pH of H8 (solid
symbols) and H2 (open symbols) 1H NMR chemical shifts for species a-f..
pKa values were determined from H8 shifts. Assignments: a, H8 (9) and
H2 (0), free 5′-IMP; b, H8 (b) and H2 (O), Ru-N7IMP; c, H8 (1) and
H2 (3), Ru-O(PO3)IMP; d, H8 ([) and H2 ()), Ru-N1IMP; e, H8 (ø)
and H2 ("), Ru-O(PO3)IMPN1-Ru; and f, H8 (2) and H2 (4), Ru-
N7IMPN1-Ru. For structures see Scheme 3. Note that d, e, and f are N1-
bound species.
Little Reaction with cAMP or Adenosine, and Only 5′-
Phosphate Binding for 5′-AMP. The reactions of 5′-AMP with
[(η6-Bip)Ru(en)Cl]+ (1:1, 5 mM, 310 K) at various pH values
1
were monitored for 24 h by H and 31P NMR spectroscopy
(Figures S5A and S5B). A rapid reaction was observed (within
(17) Similar pKa values were determined from H2 chemical shifts, e.g., pKa
values for N1H deprotonation were determined to be 8.78 for free Ino (a)
and 7.14 for Ru-N7Ino (b), close to the values of 8.79 (a) and 7.11 (b)
derived from H8 chemical shifts.
pKa of 5.73, but no N7 and N1 pKa values and is assigned to
Ru-N7IMPN1-Ru. The distribution of 5′-IMP species at pH
9
178 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
Table 3. Selected Chemical Shifts and pKa Values for 5′-AMP,
5′-CMP, 5′-TMP, Thymidine, and Their {(η6-Bip)Ru(en)}2+ Adducts
at 298 K
species (peak)a
pK b(group)
δ (1H)c
δ (31P)
a
5′-AMP (a)
6.41 ( 0.07 (PO3H) 8.555
4.05 ( 0.08 (N1H)
8.402
6.43 ( 0.04 (PO3H) 8.091
4.36 ( 0.03 (N3H)
7.875
3.557
Ru-O(PO3)AMP (b)
5′-CMP (a)
9.125
3.696
Ru-O(PO3)CMP (b)
Ru-N3CMP (c)
5′-TMP (a)
8.960
3.917
1.95
6.12 ( 0.08 (PO3H)
9.99 ( 0.03 (N3H)
6.65 ( 0.07 (PO3H)
d
1.914
Ru-N3TMP (d)
Ru-O(PO3)TMPN3-Ru
(b, c and e)
6.32 ( 0.02 (PO3H) 1.694
1.781, 1.754,
2.05
e
1.667
Thymidine (a)
Ru-N3Thymidine
(b and c)
9.76 ( 0.12 (N3H)
1.900
1.755, 1.677
a For peak labels see Figures S5 and S6. b On the basis of H8 or H6 or
T-CH3 chemical shift. c H8 and pH 6.8 for 5′-AMP, H6 and pH 7.10 for
5′-CMP, CH3 and pH 5.81 for 5′-TMP, and CH3 and pH 5.37 for thymidine.
d Peak c of H6 is overlapped with phenyl resonances in the 1H NMR spectra.
e 31P NMR resonances at ca. δ 9.0 for Ru-O(PO3) adducts, but specific
assignments were not made for peaks b, c, and e due to overlap.
time scale. Similarly, one new signal (peak b, δ 9.125) was
observed in the 31P NMR spectrum (Figure S5B) shifted to high
frequency by 5.568 ppm (pH 6.8). The variation in intensity of
31P NMR peak b with pH is shown in Figure S5C. It can been
seen that it reached its maximum intensity at pH ca. 7.2 and
decreased in intensity at either lower or higher pH; at pH 3.0,
only free 5′-AMP peaks were seen, and at pH 9.6, > 90% of
Figure 4. (A) The H8 and H2 region of the1H NMR spectrum of inosine
(5.5 mM) after reaction with [(η6-Bip)Ru(en)Cl]+ (5 mM) for 24 h at 310
K with initial pH 7.22. (B) Plots of H8 and H2 chemical shifts vs pH for
the five species a-d detected during the above reaction. Peaks labels: H8,
solid symbols; H2, open symbols, i.e., a, H8 (9) and H2 (0); b, H8 (b)
and H2 (O); c, H8 ([) and H2 ()); d, H8 (2) and H2 (4).
1
5′-AMP was free. The variations of H (H8 and H2) and 31P
Scheme 4. Scheme for Reaction of [(η6-Bip)Ru(en)Cl]+ (5 mM)
with Inosine (5.5 mM) in Aqueous Solution at pH ) 7.22. Species
Distribution (%) is Shown after 24 h Reaction at 310 K, Based on
H8 and H2 Intensities
NMR chemical shifts with pH are shown in Figure S7, and pKa
values determined from H8 H NMR shifts are listed in Table
1
3. Free 5′-AMP (peak a) has two associated pKa values of 6.41
and 4.05, attributable to the deprotonation of PO3H and N1H,
respectively. Peak b can be assigned to the phosphate-bound
adduct Ru-O(PO3)AMP and has no measurable pKa value
within the pH range 4-10.
Reactions of cAMP and adenosine (Ado) with [(η6-Bip)Ru-
1
(en)Cl]+ (1:1, 5 mM, 310 K) were monitored by H NMR at
different pH values (4.0, 7.0, and 9.0). Only small amounts
(<5%) of products were formed for both cAMP and Ado, even
after 3 d at 310 K (Table 4). These are possibly N7/N1 adducts
(based on H8 and H2 chemical shifts). pH titrations were not
attempted due to the low concentrations of products.
5′-CMP Forms 5′-Phosphate Adduct and a Minor N3
Adduct. Reactions of 5′-CMP with [(η6-Bip)Ru(en)Cl]+ (1:1,
1
5 mM, pH 7.23, 310 K) were followed for up to 3 d, and H
and 31P NMR spectra are shown in Figures S6A and B. A rapid
reaction was observed and gave rise to a new H6 1H NMR peak
shifted to low frequency (b, δ 7.875), which accounted for ca.
56% of the total 5′-CMP after 1 d. There was a corresponding
new 31P NMR peak (b, δ 8.960) shifted to high frequency by
5.264 ppm. In the 31P NMR spectrum, an additional minor peak
c appeared at 3.917 ppm. The corresponding H6 signal for peak
c was not detected, perhaps due to peak overlap with the phenyl
resonances. The 1H and 31P NMR pH titration curves are shown
in Figure S7, parts C and D (note that peak b disappears at
pH < 4.2). pKa values were determined from H6 1H NMR shifts
and are listed in Table 3. Free 5′-CMP (peak a) has pKa values
of 6.43 and 4.36, attributable to the deprotonation of PO3H and
N3H, respectively. Peak b can be assigned to Ru-O(PO3)CMP,
the time of mixing) which gave rise to one new set of H2 and
H8 H NMR signals (peak b, H2 δ 8.006 and H8 δ 8.402, pH
6.8), shifted to low frequency relative to free 5′-AMP (peak a,
H2 δ 8.240 and H8 δ 8.555) and in slow exchange on the NMR
1
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 179

A R T I C L E S
Chen et al.
Table 4. Reactions of Nucleosides and Nucleotides with [(η6-Bip)Ru(en)Cl]+
% reactiona
pH
nucleoside/nucleotide
species
1 d
3 d
0
1 d
3 d
guanosine (Guo)
Ru-N7Guo
Ru-N1Guo
Ru-N7Ino
100
0
59.4
23.0
3.7
100
0
54.3
22.8
7.7
7.20
7.25
6.19
7.37
inosine (Ino)
7.22
5.9
Ru-N1Ino
Ru-N7InoN1-Ru
Ru-N3Thymidine
Ru-N3Cytidine
Ru-N7/N1 Ado
Ru-N7GMP
Ru-O(PO3)GMP
Ru-N7cGMP
Ru-N7IMP
thymidine
cytidine
adenosine (Ado)
5′-GMP
30.6
11.6
<3
34.7
13.5
<5
7.20
7.24
7.24
7.10
5.19
7.19
7.28
7.22
5.37
7.33
7.30
7.26
>95
100b
0
< 5
cGMP
5′-IMP
100
7.28
7.28
7.26
6.46
82.0
79.7
6.55
Ru-N1IMP
4.4
8.7
1.6
3.1
12.5
0.7
0.5
Ru-N7IMPN1-Ru
Ru-O(PO3)IMPN1-Ru
Ru-O(PO3)IMP
1.4
5′-TMP
5′-CMP
5′-AMP
cAMP
Ru-N3 adducts (b, c, d, e)
Ru-O(PO3) adducts (b, c and e)
Ru-N3CMP
Ru-O(PO3)CMP
Ru-N7/N1 products
Ru-O(PO3)AMP
Ru-N7/N1 products
58.0
46.8
8.6
56.1
3.1
68.6
<5
56.8 (2 d)
47.2 (2 d)
7.4
7.23
7.23
7.04
7.30
6.03
7.10
6.78
7.27
6.23 (2 d)
7.10
40.7
3.5 (2 d)
63.5 (2 d)
<5
6.90 (2 d)
7.10
1
a On the basis of total nucleoside/nucleotide and determined by integration of H or 31P NMR resonances. Reaction conditions: Ru: nucleobase 5 mM:
5 mM (except for inosine: 5 mM: 5.5 mM), and at 310 K (298 K for cGMP). b To this sample NaCl (100 mM) was added and the solution was incubated
at 310 K; reversal of 5′-GMP binding was very slow and amounted to only ca. 10% after 1 week.
and has no measurable pKa value within the pH range 4-11.
Peak c has an associated pKa value of 6.12 (based on 31P NMR
titration), attributable to the deprotonation of PO3H and can be
assigned to the Ru-N3CMP product.
of the solution decreased from 7.20 (initial) to 5.37 after 3 d,
compatible with N3 coordination to Ru with concomitant N3H
deprotonation. Two new thymidine CH3 1H NMR peaks (b δ
1.755 and c δ 1.677) appeared at lower frequency compared to
that for free thymidine (peak a, δ 1.90; Figure S6D). The pH
titration curves for peaks a, b, and c are shown in Figure S7F,
and pKa values for free thymidine and its adducts are listed in
Table 3. For free thymidine (peak a), the pKa value of 9.76 is
attributed to deprotonation of N3H. Peaks b and c both have
no associated deprotonation of N3H and disappeared at pH <
3, and can therefore be assigned to two isomeric forms of Ru-
N3Thymidine adducts.19 After 3 d, N3-bound adducts accounted
for 34.7% of the total thymidine. In contrast to the sharp doublet
signal for the T CH3 protons of free thymidine (peak a), signals
b and c were broad, perhaps due to an exchange process.
Only 5′-GMP Forms Adducts with [(η6-Bip)Ru(en)Cl]+
in Competition with 5′-AMP, 5′-CMP, or 5′-TMP. Reactions
of [(η6-Bip)Ru(en)Cl]+ with 5′-GMP in competition with either
5′-AMP, 5′-CMP or 5′-TMP in a 1:1 molar ratio (5 mM) at pH
7.2 were monitored by 1H and 31P NMR at 310 K. In each case,
after 2 d, the only new peaks observed corresponded to those
for Ru-N7GMP (data not shown).
The reaction of cytidine with [(η6-Bip)Ru(en)Cl]+ (1:1, 5
mM, pH 7.2) was monitored by 1H NMR. On the basis of new
biphenyl 1H resonances, only 13% of cytidine reacted even after
3 d at 310 K (Table 4), possibly via N3 binding. A pH titration
was not attempted due to the low concentration of the product.
Ru(II) binds to Phosphate and N3 of 5′-TMP. Reactions
of 5′-TMP with [(η6-Bip)Ru(en)Cl]+ (1:1, 5 mM, pH 7.2, 298
1
K) were followed by H and 31P NMR spectroscopy for up to
3 d. Due to the overlap of H5 and H6 and H1′ signals of TMP
with those for biphenyl protons, the analysis for this reaction
was based on the T-CH3 signal. A rapid reaction was observed
within 10 min and gave rise to one new T CH3 peak c at δ
1.754 shifted to low frequency relative to that for free TMP
(peak a, δ 1.914). After 2 h, peaks b (δ 1.781), d (δ 1.694),
and e (δ 1.667) appeared and increased in intensity with time,
with a decrease in intensity of peak c. The pH value of the
solution decreased steadily during the reaction from 7.3 (initial)
to 6.20 (1 d). The reaction reached equilibrium after 1.5 d, at
which time the relative intensities of 5′-TMP CH3 peaks were
a, 39%; b, 24%; c, 15%; d, 12%; and e, 10%. A representative
1H NMR spectrum of the T CH3 region is shown in Figure S6C,
and 1H NMR pH titration curves are shown in Figure S7E. pKa
values were determined for peaks a and d (Table 3). Free 5′-
TMP has pKa values of 9.99 and 6.65 for N3H and PO3H,
respectively. Peak d shows one pKa value of 6.32 for PO3H,
and the absence of deprotonation of N3H, and is assigned to
Ru-N3TMP. Peaks b, c, and d all have no associated depro-
tonation for PO3H and N3H and can be assigned to adducts of
the type Ru-O(PO3)TMPN3-Ru, possibly isomers.18
Kinetics Studies. Phosphate, Chloride and Hydroxide
Decrease the Rate of GMP N7-Binding. The kinetics of
reactions of chloro and aqua Ru(Bip) with 5′-GMP and cGMP
in various solutions (different NaCl and phosphate concentra-
tions) and at various pH values were monitored by 1H and 31
P
NMR spectroscopy. The observed half-lives (t1/2) for N7-binding
are listed in Table 5. The kinetics of reaction of chloro Ru-
(Bip) with Na2HPO4 (5 mM, 1:1, 298 K, initial pH 7.26) were
1
also monitored by H and 31P NMR spectroscopy (Figure S4).
(18) (a) Renn, O.; Lippert, B.; Schollhorn, H.; Thewalt, U. Inorg. Chim. Acta
1990, 167, 123-130. (b) Pfab, R.; Jandik, P.; Lippert, B. Inorg. Chim.
Acta 1982, 66, 193-204.
The reaction of [(η6-Bip)Ru(en)Cl]+ with thymidine (1:1, 5
(19) At pH < 3, the competition for N3 binding by H+ is strong and Ru-N3
complexation is not favoured at equilibrium.
1
mM, pH 7.20) was monitored by H NMR at 310 K. The pH
9
180 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
Table 5. Half-Lives (t1/2) for Formation of Ru-N7 Bound Adducts
of 5′-GMP and cGMP with [(η6-Bip)Ru(II)(en)Cl]+ (Ru-Cl) and
[(η6-Bip)Ru(II)(en)(H2O)]2+ (Ru-H2O)
b
Ru−Cl/H O
G
conditionsa
pH
t
1/2 (h)c
2
Ru-Cl
5′-GMP
100 mM NaCl
5 mM phosphate
50 mM phosphate
H2O
7.25, 7.25
7.24, 7.30
7.22, 7.18
5.37, 5.41
7.22, 7.26
9.40, 9.13
7.34, 7.59
9.11, 9.40
7.30, 7.60
5.60, 5.88
7.16, 7.69
9.22, 9.64
7.0
5.0
15.0
1.1
3.1
5.0
1.8
4.0
1.0
0.5
0.8
4.3
Ru-H2O
5′-GMP
H2O
Ru-Cl
Ru-H2O
cGMP
cGMP
H2O
H2O
c
a Ru: G ) 1:1, 5 mM, 298 K. b pH (start), pH (equilibrium). t1/2 from
concentration versus time plots; rate constants were not determined.
The reaction reached equilibrium after 2 h, and the equilibrium
constant (K) for formation of Ru-O(PO3) was determined to be
3.2.
For the reactions of 5′-GMP with chloro Ru(Bip) in 100 mM
NaCl, the t1/2 value for formation of Ru-N7GMP was 7 h, with
89% of the 5′-GMP bound as Ru-N7GMP after two weeks. In
5 mM phosphate buffer, the t1/2 was 5 h, with 95% of GMP
bound as Ru-N7GMP after 4 d (Figure S3). In 50 mM phosphate
buffer, the t1/2 was more than 15 h, with only 50% Ru-N7GMP
formed even after one week. Under similar conditions at pH
ca. 7.3 in H2O, the reaction of 5′-GMP with aqua Ru(Bip) was
1.7 times faster than that with chloro Ru(Bip), and the reaction
of Ru(Bip) with cGMP was 3 times faster than with 5′-GMP.
No phosphate binding was detected for cGMP reactions, see
Figure 2 and Scheme 2. Therefore chloride, dianionic phosphate
ions and 5′-phosphate groups compete with N7 for binding to
Ru(II).
The reaction rate was also dependent on pH (Table 5). For
the reactions with 5′-GMP, t1/2 for formation of Ru-N7GMP
increased from 1.1 to 5.0 h when the pH was raised from 5.37
to 9.40. Similarly for cGMP, the reactions were much faster at
low pH. The t1/2 for formation of Ru-N7cGMP increased from
0.5 to 4.3 h when the pH was increased from 5.60 to 9.22.
Because the pKa of H2O in [(η6-Bip)Ru(en)(H2O)]2+ is 7.7, the
hydroxo complex Ru-OH is the predominant species in solution
at high pH,20 and thus slows down Ru-N7 binding.
Arene-purine Hydrophobic Interactions Greatly Enhance
N7-binding to GMP. Reactions of five Ru(arene) (arene ) Bip,
THA, DHA, Ben, and Cym) complexes with cGMP were
investigated by 1H and 31P NMR under physiologically relevant
conditions. Fitted plots of concentration vs time for the reaction
of aqua Ru(arene) with cGMP (1:1, 5 mM, pH 7.0) are shown
in Figure 6A, and the second-order rate constants (k)21 are listed
in Table 6. Reactions of cGMP with aqua Ru(THA), Ru(Bip)
and Ru(DHA) complexes were more than 3 times faster (k )
14.7, 8.1, and 7.1 × 10-2 M-1‚s-1, respectively) than those for
aqua Ru(Cym) and Ru(Ben) complexes (k ) 2.5 and 1.1 ×
10-2 M-1‚s-1, respectively). For the reactions of chloro Ru-
Figure 5. Regio- and stereo- specificities: H-bonds and nonbonding
repulsions for the interaction of {(η6-arene)Ru(en)}2+ with nucleobases, as
detected in nucleoside and nucleotide reactions. For clarity, the arene and
the CH2CH2 of the en ring have been omitted, except for the G adduct. For
the purine adducts, the diagrams indicate in a general way the reactions
with N7 and N1 and are not intended to apply specifically to the dinuclear
adducts.
(arene) complexes with cGMP (1:1, 2 mM), there were two
steps: Ru-Cl hydrolysis followed by Ru-N7 binding (see
Scheme 2). The observed t1/2 values for formation of Ru-N7
adducts are listed in Table 6. No attempt was made to determine
the rate constants. Plots of concentration against time are shown
in Figure 6B. The sequence of t1/2 values for the reactions of
chloro Ru(arene) is consistent with that for analogous reactions
of the aqua Ru(arene) complexes: Ru(THA) < Ru(Bip) < Ru-
(DHA) , Ru(Cym) < Ru(Ben) (Table 6). For the THA, Bip,
and DHA complexes, 100% of the N7-bound product was
observed after 2 d, whereas only ca. 80% of the N7-bound
product formed for Cym and Ben complexes after 4 d.
Discussion
Identification of Binding Sites. Purines exhibit a dichotomy
between binding at N1 and N7. The proton favors N1. In neutral
solution, N7 is usually the predominant binding site for transition
metal ions on N9-substituted purines such as adenosine and
guanosine. However, in adenine derivatives, N1 is only slightly
less favorable than N7.22 N1-substituted pyrimidines (cytidine,
thymidine, uridine) offer only one favorable ring binding site,
N3.23 Reactions of purine and pyrimidine nucleosides and
nucleotides with Pt(II) and Pd(II) complexes, for example, have
been investigated previously, usually by 1H and 31P NMR
(20) Wang, F.; Chen, H.; Parsons, S.; Oswald, I.; Davidson, J.; Sadler, P. J.,
unpublished results.
(21) Half-lives were determined for the reactions of [(η6-Bip)Ru(en)(H2O)]2+
with cGMP at various initial concentrations under the same conditions (ratio
1:1, 298 K, pH 7.0 and 100 mM NaClO4): 2 mM, t1/2 ) 105 min; 5 mM,
t1/2 ) 41 min; 10 mM, t1/2 ) 17 min. These data confirm that the reaction
is second order: t1/2 ) 1/CA0 k (k: M-1‚s-1, A ) B).
(22) Martin, R. B.; Mariam, Y. H. Metal Ions Biol. Syst. 1979, 8, 57-124.
(23) (a) Lim, M. C.; Martin, R. B. J. Inorg. Nucl. Chem. 1976, 38, 1915-1918.
(b) Scho¨llhorn, H.; Thewalt, U.; Lippert, B. J. Am. Chem. Soc. 1989, 111,
7213-7221.
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 181

A R T I C L E S
Chen et al.
ion is both diamagnetic and, on the NMR time scale, in slow
exchange between the two sites. Similarly, in this study, the
diamagnetic nature of Ru(II), together with slow ligand ex-
change, allow the acid-base properties of the products to be
individually assessed by pH titrations, and the populations of
all species in solution can be determined through measurement
of NMR peak areas. 31P NMR was also used to reveal directly
the interaction between Ru(II) and the phosphate group.26c,27
Thus, we have been able to assess the selectivity of {(η6-arene)-
Ru(en)}2+ complexes toward binding sites on purine and
pyrimidine nucleic acid derivatives.
Adducts of {(η6-arene)Ru(en)}2+ with nucleosides and nu-
cleotides exhibited characteristic 1H and 31P NMR signals, and
acid-base properties which allow their structures to be assigned
(Figures 1-4 and S6-7, and Tables 1-3). In the cases of 5′-
GMP, cGMP, Guo, 5′-IMP, and Ino, metalation at N7 gave Ru-
N7 products with high-frequency shifts of 0.3-0.6 ppm for H8
resonances, but slight low-frequency shifts of 0.02-0.04 ppm
for H2 resonances, compared with free ligands. Metalation at
N1 (only found in the reactions with 5′-IMP and Ino) gave Ru-
N1 products with low-frequency shifts of 0.16-0.21 ppm for
H8 and 0.22-0.26 ppm for H2 resonances. A similar chemical
shift behavior for H8 and H2 protons upon metalation at N7 or
N1 sites of purines has been reported for other metal ions.25a
For all nucleotides, the 31P NMR resonances exhibit large
downfield shifts of ca. 5 ppm when the phosphate group
coordinates to the Ru center (Tables 1-3).
The pH dependences of 1H and 31P NMR chemical shifts of
nucleobase derivatives were studied in detail. Protonation of
N7, N1, N3, or 5′-phosphate produced significant low- or high-
frequency shifts of 1H or 31P NMR resonances (Figures 1-4 and
S7). Ru(II) binding to N7, N1, N3, or the 5′-phosphate group
prevents the protonation of that site, and therefore, no change
in chemical shift due to protonation is expected. Therefore, the
Ru(II) binding sites can be identified, as shown in Tables 1-3.
It is notable that small amounts of dinuclear adducts28 of the
type Ru-N7BN1-Ru and Ru-O(PO3)BN1(N7)-Ru (B )
nucleosides and nucleotides) were detected in solution. Binding
site identification was supported by the following. (1) The
significant pH decrease (from 7.2 to ca. 6) during reactions with
I and T derivatives, is indicative of N1-coordination to Ru for
5′-IMP or inosine, and N3-coordination for 5′-TMP and
thymidine, with concomitant deprotonation. This was not
observed for reactions with G and A and C derivatives for which
almost no N1 or N3-binding occurred (Table 4). (2) Ru-N7
adducts have lowered pKa values of 1.4-1.7 units for N1, and
of 0.6-0.7 units for 5′-phosphate groups, but Ru-O(PO3)
binding causes only a small lowering of pKa values of 0.18-
0.30 for N1, and Ru-N1 binding has little effect on the pKa
values of 5′-phosphate groups, relative to the free ligands (Tables
Figure 6. Plots showing the variation of the concentration of (η6-arene)-
Ru-N7cGMP products with time during the reaction of (A) [(η6-arene)Ru-
(en)(H2O)]2+ (5 mM), (B) [(η6-arene)Ru(en)Cl]+ (2 mM) with cGMP (1:1,
pH 7.0), 100 mM NaClO4, 298 K, based on the integration of G H8 1H
NMR peaks. The curves in (A) are computer-fits to second-order kinetics
giving the rate constants listed in Table 6. These reactions were all followed
for periods of between 24 h and 4 days (see Figures S8 and S9). The
reactions of the chloro complexes in (B) were followed for up to 1 week,
and the half-lives for the formation of Ru-N7 adducts are shown in Table
6. No attempt was made to fit these data and determine the rate constants.
(C) Posssible π-π stacking between arene and purine in cGMP adducts
with Ru(THA), Ru(Bip), and Ru(DHA).
Table 6. Rate Constants (k) and Half-lives (t1/2) for Reactions of
Ru(arene) Complexes with cGMPa
rate constant
aqua Ru(arene)b
(× 10-2 M-1s-1
)
t1/2 (h)c
chloro Ru(arene)b
t
1/2 (h)d
Ru(THA)(H2O)
Ru(Bip)(H2O)
Ru(DHA)(H2O)
Ru(Cym)(H2O)
Ru(Ben)(H2O)
14.72 ( 0.69 0.38 (0.02
8.06 ( 0.15 0.69 (0.01
7.08 ( 0.20 0.78 ( 0.02
2.49 ( 0.03 2.23 ( 0.03
1.13 ( 0.02 4.94 ( 0.09
Ru(THA)Cl
Ru(Bip)Cl
Ru(DTA)Cl
Ru(Cym)Cl
Ru(Ben)Cl
1.1
2.0
3.6
7.1
13
a Conditions: Ru: G ) 1:1, 100 mM NaClO4, 298 K, 5 mM for aqua
Ru(arene) and 2 mM for chloro Ru(arene); pH was adjusted to 7.0 at reaction
commencement and found to be ca. 7.2 at equilibrium for each reaction.
b For the THA, Bip and DHA complexes, 100% of the N7-bound products
observed after 2 d, whereas only ca. 80% of the N7-bound products formed
for Cym and Ben complexes after 4 d. c Calculated from t1/2 ) 1/kC°,
d
C° ) 5.0 mM. t1/2 from concentration versus time plots; rate constants
were not determined.
spectroscopy.24-26 Resolution of the N7-N1 dichotomy is
unequivocal in the cases of Pt(II) and Pd(II) because the metal
(24) (a) Martin, R. B. In Cisplatin-Chemistry and Biochemistry of a Leading
Anticancer Drug; Lippert, B., Ed.; Wiley-VCH: Weinheim, Germany 1999;
pp 183-205. (b) Martin, R. B. In Platinum, Gold, and Other Metal
Chemotherapeutic Agents: Chemistry and Biochemistry; Lippard, S. J.,
Ed.; American Chemical Society, Washington, DC, 1983, ACS Symp. Ser.
209, pp 231-244.
(27) (a) Alessio, E.; Xu, Y.; Cauci, S.; Mestroni, G.; Quadrifoglio, F.; Viglino,
P.; Marzilli, L. G. J. Am. Chem. Soc. 1989, 111, 7068-7071. (b) Reily,
M. D.; Marzilli, L. G. J. Am. Chem. Soc. 1986, 108, 8299-8300.
(28) Small amounts of the dinuclear species Ru-N7IMPN1-Ru (8.8%, pH 6.4)
for 5′-IMP and Ru-N7InoN1-Ru (3.7%, pH 6.24) for inosine, were
detected at equilibrium. Both exhibited some stability at pH < 3 (Figures
4 and 5). Similar types of dinuclear M-N7BN1-M complexes have been
detected previously in reactions of {dienPd}2+ with all three purine 5′-
nucleoside monophosphates GMP, IMP, and AMP at equilibrium: Pd-
N7GMPN1-Pd (15%, pH 7.1), Pd-N7AMPN1-Pd (14%, pH 6.1), Pd-
N7IMP-N1-Pd (14%, pH 6.4) and Pd-N7InoN1-Pd (15%, pH 6.2), see
ref 25 Metalation at N7 promotes the acidity of the proton at N1 by up to
ca. 1.5 units, and therefore the binding of Ru to Ru-N7BN1H can occur
at the lower pH values compared to the free ligand BN1H.
(25) (a) Scheller, K. H.; Scheller-Krattiger, V.; Martin, R. B. E. J. Am. Chem.
Soc. 1981, 103, 6833-6839. (b) Vestues, P. I.; Martin, R. B. J. Am. Chem.
Soc. 1981, 103, 806-809. (c) Sovago, I.; Martin, R. B. Inorg. Chem. 1980,
19, 2868-2871.
(26) (a) Dijt, F. J.; Canters, G. W.; den Hartog, J. H. J.; Marcelis, A. T. M.;
Reedijk, J. J. Am. Chem. Soc. 1984, 106, 3644-3647. (b) Miller, S. K.;
Marzilli, L. G. Inorg. Chem. 1985, 24, 2421-2425. (c) Reily, M. D.;
Hambley, T. W.; Marzilli, L. G. J. Am. Chem. Soc. 1988, 110, 2999-
3007.
9
182 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
1-3). These data are consistent with metalation at N7 which
usually acidifies the proton at N1,29-31 and consistent with the
presence of an outer-sphere interaction involving H-bonding of
the 5′-phosphate group to an en NH proton in Ru-N7 adducts
(but not in Ru-N1 adducts) which can contribute to the larger
lowering of the pKa value of 5′-PO3H for Ru-N7 adducts
compared to Ru-N1 adducts.25b,32,33 (3) Ru-O(PO3), Ru-N1
and Ru-N3 adducts disappeared at pH < 5.0 (Figures 1, 3, 4
and S7), due to protonation of N1 and the 5′-phosphate group.34
Selective Recognition of Nucleobases. A summary of the
extent of reaction of various nucleosides and nucleotides with
[(η6-Bip)Ru(en)Cl]+ is given in Table 4. Nucleotides and their
corresponding nucleosides have similar affinities for Ru, except
for a significant amount of 5′-phosphate binding in the former
case. A schematic representation of the possible interactions
between the nucleobases and the en NH2 groups is given in
Figure 5. H-bonding and nonbonding repulsive interactions, in
addition to the electronic properties of the various nucleobase
coordination sites,35 appear to play key roles in these site-
selective reactions.3b For G, N7 is the favored site for Ru(arene)
binding. Our previous studies of Ru-N7 guanine adducts have
revealed a strong H-bonding interaction between one en NH
(pointing toward the G base) and G C6O with N‚‚‚O distance
2.8 Å and N-H‚‚‚O angle 163°, which is present in the X-ray
crystal structures of [(η6-Bip)Ru(en)(9EtG)]2+ and [(η6-Bip)-
Ru(en)(Guo)]2+. Such high stereospecific H-bonding has also
been detected in aqueous solution by NMR.12 Thus, coordination
at N7 is stabilized by H-bonding of en NH2 with G C6O and
coordination at N1 is unfavorable due to the repulsive interaction
of en NH2 with G2 NH2. This is consistent with the data for
reactions of Ru(Bip) with Guo, 5′-GMP and cGMP: only Ru-
N7 adducts were detected (and no Ru-N1 adducts) over the
pH range 2-12 (Tables 1 and S1). Inosine (I), in contrast to G,
has no NH2 group. Coordination at either N7 or N1 can be
stabilized by H-bonding of en NH with C6O, and significant
amounts of Ru-N7 and Ru-N1 adducts were detected for
reactions of inosine and 5′-IMP with Ru(Bip) (Table 4). Binding
at N1 is promoted at high pH with the concomitant deprotona-
tion of N1H (Table S1). Adenine (A) has a C6 NH2 group and
coordination at either N7 or N1 of A is weakened by repulsive
interactions with the exocylic amino group: no base-binding
was observed for reactions of Ru(Bip) with Ado, cAMP or 5′-
AMP over the range pH 3.1-9.6 (Table 4, Figure S5). For
thymine (T), significant amounts of Ru-N3 adducts were
formed in reactions of Ru(Bip) with both thymidine and 5′-
TMP at pH ca. 6.0 (Table 4). Coordination at N3 may be favored
by H-bonds between en NH and the exocyclic oxygens at C2
and C4,36 For C, only small amounts of Ru-N3 adducts (ca.
10%) were found for reactions of cytidine and 5′-CMP (Table
4). Coordination at N3 is weak due to the repulsive interaction
of en NH with the C4 NH2 group.
On the basis of the above nucleoside and nucleotide studies,
the reactivity of the various binding sites of nucleobases37 toward
Ru(Bip) at neutral pH decreases in the order G(N7) > I(N7) >
I(N1), T(N3) > C(N3) > A(N7), A(N1). In competitive binding
experiments with 5′-GMP versus either 5′-AMP or 5′-CMP or
5′-TMP, the only final adduct was [(η6-Bip)Ru(II)(en)(N7-
GMP)]. Also in enzymatic digestion experiments, it was found
that the binding of [(η6-p-cymene)Ru(en)Cl]+ to a 14-mer
oligonucleotide occurred specifically at guanine sites.11 Experi-
ments on the inhibition of RNA synthesis show that the
termination sites produced by {(η6-arene)Ru(en)}2+ adducts of
DNA are at guanine residues.38 Such a high specificity for
guanine-N7 does not appear to have been reported for other
metal-based anticancer agents. Many other Pt and Ru anticancer
complexes also bind covalently to adenine. The complex [Ru-
(II)(NH3)5Cl]+ reacts with many nitrogen bases,31b preferentially
with the N7 of guanine,31a N1 of adenine39 and N4 of
cytosine.31b For cis-[Ru(II)(DMSO)4Cl2], the reaction sites in
double-stranded polynucleotides are thought to be N7 of both
guanine and adenine.40 The trifunctional organometallic complex
[(η6-Ben)Ru(H2O)3]2+ binds to both N7 and N1 of adenine.41
Cis-{(NH3)2Pt(II)}2+ binds to adenosine at both N7 (55%) and
N1 (45%),42 and the monofunctional complex {(dien)Pt(II)}+
also binds to adenosine at both N7 (50%) and N1 (50%).23a
Cis-{(NH3)2Pt(II)}2+ forms both GG and AG intrastrand cross-
links that account for about 65% and 25%, respectively, of DNA
platinations.43 Organometallic metallocene dichloride antitumor
agents of the type [Cp2MCl2] (M ) Mo, Ti) exhibit little, if
any, coordinative selectivity in nucleotide competition experi-
ments.44,45 The anti-HIV agent Zn(II)-[12]aneN4 shows a high
selectivity for N3 of thymine and uridine derivatives among all
of the nucleobases, with formation of cyclen NH H-bonds to
the C2 and C4 carbonyl oxygens of T or U.46
The ability of the NH proton of ethylenediamine to act as an
H-bond donor toward an exocyclic oxo group but not toward
an amino group appears to play an important role in controlling
site recognition of the nucleobases by these Ru(arene) anticancer
agents. When the en ligand is replaced by H-bond acceptor
ligands (such as acetylacetonate derivatives), the Ru(arene) can
also recognize adenine bases.47 Initial data suggest that there is
(36) The pKa(N3H) value of thymidine is 9.76 (Table 3). Protonation of N3
competes with Ru-N3 binding, and partially explains T did not fully react
with Ru(Bip) at lower pH (7.2-5.8).
(37) In Watson-Crick double-helical DNA, N3 of thymine and cytosine would
not be expected to be available for binding to Ru(II) because N3 is involved
in H-bonding in base-pairs. N3 binding would therefore be restricted to
regions of single-stranded DNA. Competitive experiments at pH 7.2 show
that N3-binding to T or C is weak compared to G N7.
(38) Chen, H.; Nova´kova´, O.; Zaludova´, R.; Brabec, V.; Sadler, P. J., unpublished
results.
(39) Clarke, M. J. J. Am. Chem. Soc. 1978, 100, 5068-5075.
(40) Cauci, S.; Alessio, E.; Mestroni, G. Inorg. Chim. Acta 1987, 137, 19-24.
(41) Korn, S.; Sheldrick, W. S. J. Chem. Soc., Dalton Trans. 1997, 2191-
2199.
(42) Arpalahti, J.; Lehikoinen, P. Inorg. Chim. Acta 1989, 159, 115-120.
(43) Fichtinger-Schepman, A.-M. J.; van der Veer, J. L.; den Hartog, J. H. J.;
Lohman, P. H. M.; Reedijk, J. Biochemistry 1985, 24, 707-713.
(44) Kuo, L. Y.; Kanatzidis, M. G.; Sabat, M.; Tipton, A. L.; Marks, T. J. J.
Am. Chem. Soc. 1991, 113, 9027-9045.
(45) Guo, M.; Guo, Z.; Sadler, P. J. J. Biol. Inorg. Chem. 2001, 6, 698-707.
(46) Shionoya, M.; Kimura, E.; Shiro, M. J. Am. Chem. Soc. 1993, 115, 6730-
6737.
(29) (a) Inagaki, K.; Kidani, Y.; J. Inorg. Biochem. 1979, 11, 39-47. (b)
Schro¨der, G.; Lippert, B.; Sabat, M.; Lock, C. J. L.; Faggiani, R.; Song,
B.; Sigel, H. J. Chem. Soc., Dalton Trans. 1995, 3767-3775.
(30) Guo, Z.; Sadler, P. J.; Zang, E. J. Chem. Soc., Chem. Commun. 1997, 27-
28.
(31) (a) Clarke, M. J.; Taube, H. J. Am. Chem. Soc. 1974, 96, 5413-5419. (b)
Clarke, M. J. Met. Ions Biol. Syst. 1980, 11, 231-283. pKa(N1) is lowered
by 0.8 for [Ru(II)(NH3)5(Guo)]2+ (pKa(N1) 8.7) and 2.2 for [Ru(III)(NH3)5-
(Guo)]3+ (pKa(N1) 7.36). The acidifying effect of Ru(II) is therefore less
than that of Ru(III).
(32) Berners-Price, S. J.; Frey, U.; Ranford, J. D.; Sadler, P. J. J. Am. Chem.
Soc. 1993, 115, 8649-8659.
(33) Reily, M. D.; Marzilli, L. G. J. Am. Chem. Soc. 1986, 108, 6785-6793.
(34) Small amounts of dinuclear Ru-N7BN1-Ru adducts (2%) still persist at
pH < 3. There is enhanced stability of Ru-N1 binding via formation of
dinuclear Ru-N7BN1-Ru adduct.
(35) Despite the similar acid-base properties of the nucleophilic sites of
guanosine (G) and inosine (I) (see Tables 1 and 2, and ref 3b), {(η6-Bip)-
Ru(en)}2+ coordinates to the N1 site of I but not of G. This suggests that
interactions between the Ru en NH2 groups and the exocyclic groups on
the bases play a major role in controlling site selectivity.
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 183

A R T I C L E S
Chen et al.
a correlation between the anticancer activity of Ru(arene)
complexes and the presence of an en NH H-bond donor group.
Activity is lost when the en NH2 hydrogens are substituted, e.g.,
with methyl groups.48 The choice of the specific chelating ligand
appears to be an effective way of controlling selectivity for
nucleobase recognition by octahedral Ru complexes. The steric
constraints for nucleobase binding to an octahedral site are much
more demanding compared to those for binding at a square-
planar site, e.g., Pt(II).
Phosphate Binding. The first stage in reactions of [(η6-Bip)-
Ru(en)Cl]+ with nucleotides (5′-GMP, 5′-IMP, 5′-AMP, 5′-
CMP, and 5′-TMP) involved rapid 5′-phosphate binding (Figures
1 and 3, Table 4).49 The phosphate adducts of 5′-GMP and 5′-
IMP rearranged into N7 and N1 adducts, but for adenine,
cytidine, and thymidine nucleotides, significant amounts of
phosphate complexes were still present at equilibrium (Table
4). For 5′-AMP reactions, it can been seen that Ru-O(phos-
phate) binding is sensitive to pH, reaching a maximium at pH
7.2 (Figure S5). Lower pH values promoted protonation of the
5′-phosphate group, and higher pH values promoted the
competitive binding of OH- and thus reduced the formation of
the 5′-phosphate adducts. Similarly, inorganic dianonic phos-
phate ions were also found to bind rapidly to {(η6-Bip)Ru(en)}2+
to form a Ru-O(PO3) product (Figure S4). Thus binding of
{(η6-Bip)Ru(en)}2+ to N7 of 5′-GMP was significantly retarded
by phosphate buffers at neutral pH (Figure S3 and Table 5). In
contrast, no binding of Ru(II) to the phosphodiester group of
cGMP or cAMP was detected (Figure 2).
At neutral pH, cis-[PtCl2(NH3)2] forms monodentate adducts
with the inorganic ortho-, pyro- and tri-phosphate ions,50 and
an N7, PO macrochelate cis-[Pt(NH3)2(5′-GMP-N7, PO)] with
the nucleotide 5′-GMP, but does not bind to the phosphate group
of the methyl phosphate diester of 5′-GMP (Me-5′-GMP).51
{Cp2Mo}2+ coordinates to nucleotides (5′-dAMP, 5′-dCMP and
5′-dTMP) through Mo-N7/N3 and Mo-O(phosphate) chelation,
but does not show direct phosphate coordination to the methyl-
phosphate diester of 5′-dGMP (Me-5′-dGMP)), diphenyl phos-
phate ((C6H5O)2P(O)OH), or diethyl phosphate ((CH3CH2O)2P-
(O)OH).52 Zn(II) cyclen also acts as a good monotopic receptor
for dianionic phosphate monoesters.53 1H NMR studies of
reactions of Zn(II)-bis(cyclen) with 5′-dTMP and 5′-dTDP
indicate that the terminal phosphate dianion interacts with one
of the Zn(II)-cyclens and the imido anion of dT binds to the
other Zn(II)-cyclen.54 Zinc-containing carboxylate-bridged het-
erodimetallic complexes can react with the phosphodiester ligand
diphenyl phosphate to form bis(phosphate) complexes.55 {(NH3)5
Ru(III)}3+ does not appear to coordinate directly to 5′-phosphate
groups of nucleotides,56 but the chelation of N7 and phosphate
group of 5′-(d)GMP with cis/trans-[Ru(II)Cl2(DMSO)],57 and
N7,O(P)-macrochelation of adenosine and guanosine 5′-mono-,
-di-, and -tri-phosphates41 with [(η6-Ben)Ru(II)(H2O)3]2+ have
both been detected by 1H and 31P NMR pH titrations. The latter
organometallic complex is also capable of coordinating to the
phospho- diester group of 5′-ADP.41
The {(η6-arene)Ru(II)(en)}2+ complexes studied here bind
to dianionic phosphate and phosphomonoesters, but very weakly
to phosphodiesters. This has potential biological implications.
The reversible binding to phosphate58 as well as transient binding
to membrane phospholipids may facilitate cellular uptake59 of
Ru(arene) species. Direct Ru(II) coordination with the backbone
phosphodiester groups of DNA will be weak, but electrostatic
interactions and H-bonding may be involved in the initial
recognition of [(η6-arene)Ru(II)(en)(H2O)]2+ prior to binding
to guanine N7, as proposed for some Pt complexes.60
Kinetic Studies. The rate of reaction of [(η6-Bip)Ru(en)X]n+
with N7 of GMP depends on whether Cl-, H2O, OH-, or
phosphate occupies the available coordination position (X).
Chloro Ru(arene) complexes rapidly hydrolyze in water to give
more reactive aqua Ru(arene) species (Scheme 1 and Figures
S1 and S2). Hydrolysis is suppressed in the presence of 100
mM chloride ions. The intracellular chloride concentration (ca.
4-23 mM) is significantly lower than the extracellular con-
centration (ca. 103 mM).61 Thus, reactive Ru-H2O species are
likely to predominate inside cells under physiological conditions.
The reactions of [(η6-Bip)Ru(en)(H2O)]2+ with both 5′-GMP
and cGMP are more than 3 times slower at pH 9 compared to
pH 5 to 7 (Table 5). This is attributable to the lower reactivity
of Ru-OH compared to Ru-OH2 (as is also the case for am-
(m)ino Pt(II) complexes), since the pKa values for [(η6-arene)-
Ru(en)(H2O)]2+ complexes studied here are all within the range
7.7-8.25 (Figure S10).20,62
Arene-purine Hydrophobic Interactions. There appear to
be few previous studies of the mechanisms of ligand substitution
reactions of (η6-arene)Ru complexes. Reactions of aqua Ru-
(arene) complexes with N7 of cGMP obey second-order kinetics
and therefore appear to proceed via an associative pathway
involving a seven-coordinate transition state.63 The rates of
reaction of [(η6-arene)Ru(II)(en)(H2O]2+ complexes (and chloro
complexes) with cGMP depends markedly on the nature of the
arene, decreasing by over an order of magnitude in the series:
THA > Bip > DHA . Cym > Ben (Figure 6 and Table 6).
(56) (a) Rodriguez-Bailey, V. M.; Clarke, M. J. Inorg. Chem. 1997, 36, 1611-
1618. (b) Rodriguez-Bailey, V. M.; LaChance-Galang, K. J.; Doan, P. E.;
Clarke, M. J. Inorg. Chem. 1997, 36, 1873-1883.
(57) (a) Alessio, E.; Xu, Y.; Cauci, S.; Mestroni, G.; Quadrifoglio, F.; Viglino,
P.; Marzilli, L. G. J. Am. Chem. Soc. 1989, 111, 7068-7071. (b) Tian, Y.;
Yang, P.; Li, Q.; Guo, M.; Zhao, M. Polyhedron 1997, 16, 1993-1998.
(58) Binding of Ru(II) to dianionic nucleotides mono-, di-, and tri-phosphates
may occur rapidly inside cells at pH 7.4, but phosphate is readily displaced
by G N7. Also the pH close to the surface of DNA is thought to be more
acidic (as low as pH 4.5): Lamm, G.; Pack, G. R. Proc. Natl. Acad. Sci.
U.S.A. 1990, 87, 9033-9036, and Ru(II) binding to mono-nucleotide
phosphates will be weak at this pH due to phosphate protonation.
(59) (a) Davies, M. S.; Thomas, D. S.; Hegmans, A.; Berners-Price, S. J.; Farrell,
N. Inorg. Chem. 2002, 41, 1101-1109. (b) Davies, M. S.; Berners-Price,
S. J.; Hambley, T. W. Inorg. Chem. 2000, 39, 5603-5613. (c) Frey, U.;
Ranford, J. D.; Sadler, P. J. Inorg. Chem. 1993, 32, 1333-1340. (d)
Yamamoto, T.; Moriwaki, Y.; Takahashi, S.; Tsutsumi, Z.; Yamakita, J.-
I.; Higashino, K. Metabolism 1997, 46, 1339-1342.
(47) Ferna´ndez, R.; Melchart, M.; Habtemariam, A.; Hanson, I.; Parsons, S.;
Sadler, P. J., unpublished results.
(48) Habtemariam, A.; Sadler, P. J.; Aird, R. E.; Cummings, J.; Jodrell, D. I.,
unpublished results.
(49) Additionally in some cases, we have detected dinuclear adducts formed
through Ru-N7 or -N1 or -N3 and Ru-O(PO3) chelation: Ru-O(PO3)-
GMPN7-Ru, Ru-O(PO3)IMPN1-Ru and Ru-O(PO3)TMPN3-Ru.
(50) (a) Bose, R. N.; Viola, R. E.; Cornelius, R. D. J. Am. Chem. Soc. 1984,
106, 3336-3343. (b) Bose, R. N.; Cornelius, R. D.; Viola, R. E. Inorg.
Chem. 1985, 24, 3989-3996.
(51) Reily, M. D.; Marzilli, L. G., J. Am. Chem. Soc. 1986, 108, 8299-8300.
(52) Kuo, L. Y.; Kanatzidis, M. G.; Sabat, M.; Tipton, A. L.; Marks, T. J. J.
Am. Chem. Soc. 1991, 113, 9027-9045.
(53) Koike, T.; Kimura, E. J. Am. Chem. Soc. 1991, 113, 8935-8941.
(54) Aoki, S.; Kimura, E. J. Am. Chem. Soc. 2000, 122, 4542-4548.
(55) Tanase, T.; Yun, J. W.; Lippard, S. J. Inorg. Chem. 1996, 35, 3585-3594.
(60) (a) Marcelis, A. T. M.; Erkelens, C.; Reedijk, J. Inorg. Chim. Acta. 1984,
91, 129-135. (b) Elmroth, S. K. C.; Lippard, S. J. J. Am. Chem. Soc. 1994,
116, 3633-3634. (c) Kozelka, J.; Barre, G. Chem. Eur. J. 1997, 3, 1405-
1409.
(61) Jennerwein, M.; Andrews, P. A. Drug Metab. Dispos. 1995, 23, 178-184.
(62) Hung, Y.; Kung, W.-J.; Taube, H. Inorg. Chem. 1981, 20, 457-463.
(63) Tobe, M. L.; Burgess, J. Inorganic Reaction Mechanisms, 1st ed.; Wesley:
New York, 1999; pp 37 and 584.
9
184 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003

Novel Recognition Mechanisms
A R T I C L E S
There is no correlation between the rates of cGMP N7-binding
and pKa values of aqua Ru(arene) complexes.64 The faster rates
of reaction for [(η6-arene)Ru(II)(en)(H2O)]2+ complexes con-
taining the large arene ligands THA, Bip and DHA65 imply
lower activation free energies (∆G‡) for formation of seven-
coordinate transition states. A significant contribution to ∆G‡
may arise from π-π stacking of the arene and purine ring12 in
the transition state (negative ∆H‡). Such an interaction is not
possible for the p-cymene and benzene complexes. Arene-purine
π-π stacking may also result in a positive contribution to ∆S‡
through desolvation of cGMP in the reaction transition state.
a repulsive interaction also occurs for G when Ru binds to N1,
but, if this NH2 group is absent, as in inosine derivatives, then
a significant amount of binding to N1 is observed. Similarly,
binding of Ru to N3 of C derivatives is weak due to the repulsive
interaction with the C4 NH2 group. Binding to N3 of T is
significant, but was not competitive with binding to G N7 at
neutral pH. It is clear that nucleobase site-selectivity is strongly
related to interactions with the en NH2 group which is attractive
toward exocyclic carbonyl oxygens but repulsive toward exo-
cyclic amino groups of nucleobases.
Reactions of the chloro Ru(II) arene complexes with nucleo-
tides proceed via hydrolysis followed by rapid binding to the
phosphate monoester group. In contrast, binding to the monoan-
ionic phosphodiester group of nucleotides is weak and was not
detected. Bound phosphate groups are then readily displaced
by base nitrogens. Studies of the kinetics of binding of [(η6-
arene)Ru(en)X]n+ (X ) H2O, Cl) to cGMP suggested that arene-
purine base-stacking plays a significant role in stabilizing the
transition state in the associative substitution reactions. Reactions
of complexes containing arenes (THA, Bip, and DHA) which
can take part in π-π stacking are up to an order of magnitude
faster than those containing arenes which cannot (Cym, Ben).
Such kinetic effects could play a role in the biological activity
of this class of complexes.
Hydrophobic interactions could produce a driving force for
DNA binding. A recent analysis of free energy contributions
to DNA binding suggests that intercalation of ethidium, pro-
pidium, daunorubicin, and adriamycin is driven by hydrophobic
effects and van der Waals contacts within the intercalation site.66
Studies of the interaction of thymine derivatives with zinc cyclen
containing a pendant acridine show that acridine-thymine ring-
stacking increases the affinity of Zn(II) for dT N3.67 For many
other reported metal agents, however, the presence of a rigid
aromatic rings, e.g., quinoline in trans-[Pt(quinoline)(NH3)-
(9EtG)Cl]+,68 has been found to significantly retard the rate of
N7-binding of 5′-GMP. In contrast, for the Ru(arene) complexes
studied here, the flexibility of large arene ligands reduces the
steric requirements and makes simultaneous arene-nucleobase
stacking and N7-covalent binding compatible. Our initial
investigations of reactions of calf-thymus DNA show that chloro
Ru THA, DHA and Bip complexes bind to DNA faster than
the Cym complex, and CD, LD experiments suggest that the
larger arene rings (THA, DHA and Bip) are oriented on DNA
in a manner compatible with intercalation.38 The rate of DNA
binding appears to be specifically enhanced by hydrophobic
interactions between the arene and the purine base.
These studies provide a basis for investigating the recognition
of duplex DNA by octahedral arene Ru(II) complexes. We
anticipate that it should be possible to enhance the discrimination
of DNA base recognition by appropriate modification of the
chelating ligand. By appropriate choice of arene it may also
possible to achieve a high degree of selectivity via kinetic effects
due to π-π arene-base stacking interactions (intercalation). Dual
mode G N7/intercalative binding may also induce novel
structural distortions in duplex DNA, and hence, the recognition
by DNA binding proteins and repair enzymes. Consideration
of all these factors allow optimization of the design of this series
of anticancer complexes.
Conclusions
We have shown here that anticancer complexes of the type
[(η6-arene)Ru(en)Cl]+ are highly selective in their recognition
of binding sites on nucleosides and nucleotides. This arises not
only from the differences in basicity between the binding sites,
but also from the demanding constraints imposed on the reactive
monofunctional site in these pseudo-octahedral “piano-stool”
Ru(II) arene complexes.
{(η6-arene)Ru(en)}2+ complexes exhibit a remarkably high
preference for binding to guanine N7. Such binding is stabilized
by H-bonding between the C6O of G and NH of the en ligand.12
The lack of binding to adenine can be attributed to unfavorable
interactions between the C6 NH2 group and Ru en NH2. Such
Acknowledgment. We thank the CVCP (ORS Award for
H.C.), EPSRC, ETF, Wolfson Foundation, Royal Society and
Wellcome Trust for their support for this work, and our
colleagues in EC COST D8 and D20, and in the Cancer
Research UK Oncology Unit, Western General Hospital
(Edinburgh), for stimulating discussions. We also thank Dr. Fuyi
Wang, Dr. Juan Mareque, Dr. Abraha Habtemariam and Dr.
Claudia Blindauer (University of Edinburgh) for helpful dis-
cussions and for comments on this script.
Supporting Information Available: Table S1 giving species
distribution and N(7)/N(1) ratios for inosine and guanosine
binding by {(η6-Bip)Ru(en)}2+ at various pH values. Figures
S1-S10 showing 2D [1H, 15N] HSQC NMR time course for
(64) Hydroxo Ru(arene) species (Ru-OH) appear to be much less reactive than
aqua Ru(arene) species (Ru-OH2), but the pKa values for these five aqua
Ru(arene) complexes (ref 20) lie within a narrow range: 8.01 (Ru-THA),
7.89 (Ru-DHA), 7.71 (Ru-Bip), 8.25 (Ru-Cym, see Figure S10) and
7.9 (Ru-Ben, ref 62). Thus, at pH 7.0, aqua complexes should predominate
for all these arene complexes.
(65) The slower rate of reaction of DHA compared to THA complexes, may
arise from the difference in hinge-bending of the arene. In both the chloro
and the aqua Ru(DHA) complexes, the tricyclic ring system of DHA is
bent down towards the Ru-Cl(O) bond by ca. 40° (refs 12 and 20), which
may impose a steric constraint on coordination of N7. For the Ru(THA)
complexes, such a steric constraint would be smaller because the ring system
is bent up away from Ru by ca. 7° (refs 12 and 20).
1
reaction of [(η6-Bip)Ru(15N-en)Cl]+ with 5′-GMP, H and 31P
NMR spectra for reaction of [(η6-Bip)Ru(en)Cl]+ with 5′-GMP
in phosphate buffer,31P NMR spectra for reaction of [(η6-Bip)-
Ru(en)Cl]+ with Na2HPO4, 1H and 31P NMR spectra for reaction
of [(η6-Bip)Ru(en)Cl]+ with 5′-AMP at various pH values and
1
percentage of Ru-O(PO3)AMP adduct as a function of pH, H
and 31P NMR spectra for reaction of [(η6-Bip)Ru(en)Cl]+ with
(66) Ren, J.; Jenkins, T. C.; Chaires, J. B. Biochemistry 2000, 39, 8439-8447.
(67) Shionoya, M.; Ikeda, T.; Kimura, E.; Shiro, M. J. Am. Chem. Soc. 1994,
116, 3848-3859.
1
5′-CMP, 5′-TMP and thymidine, H or 31P NMR pH titration
(68) Bierbach, U.; Farrell, N. Inorg. Chem. 1997, 36, 3657-3665.
curves for reactions of Ru(Bip) with 5′-AMP, 5′-CMP,
9
J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003 185

A R T I C L E S
Chen et al.
5′-TMP, and thymidine, Scientist model used to determine the
rate constants given in Table 6, time dependence plots of species
concentrations for reactions of aqua Ru(arene) with cGMP, 1H
NMR pH titration for determination of the pKa (H2O) value of
[(η6-Cym)Ru(en)(H2O)]2+. This material is available free of
charge via the Internet at http://pubs.acs.org.
JA027719M
9
186 J. AM. CHEM. SOC. VOL. 125, NO. 1, 2003