Structure and dynamics of a platinum(II) aminophosphine complex and its nucleobase adducts

Article abstract of DOI:10.1039/b007670h

The anti-cancer aminophosphine complex, cis-[Pt(Me₂N(CH₂)₂PPh₂-N,P)Cl(Me ₂N(CH₂)₂PPh₂-P)]Cl, 1, is shown, by ¹H and ³¹P NMR studies, to Undergo reversible chelate ring-closing and -opening in aqueous solution. The rate for this process (21 ± 2 s^-1 at pH* 8.5, 295 K), which is slow on the NMR timescale, varies significantly with pH values, chloride ion concentration and temperature. The chelate-ring-opened form is favoured at acidic pH values. The activation parameters which govern this process, have been determined via ³¹P-2-D EXSY NMR spectroscopy, and the mechanism of chelate ring-opening is discussed. The X-ray crystal structure of the ring-opened complex, cis-[Pt(Me₂N(CH₂)₂PPh₂-N,P)Cl(Me ₂NH(CH₂)₂PPh₂-P)](NO₃) ₂·1.5H₂O, 1a, shows that the Pt-N bond is relatively long (2.147 A). The nucleotide 5′-guanosine monophosphate (5′-GMP) binds via N7 and displaces the Cl ligand. Unusually for a platinum(II) amine complex, GMP binding is rapid and reversible. Two isomeric GMP adducts are formed in a 3: 1 ratio below pH* 6 (pH meter reading in D₂O). Above pH* 8.2, the predominant species are non N7-bound adducts. The on-rate for 5′-GMP binding to the ring-closed form cis-[Pt(Me₂N(CH₂)₂PPh₂-N,P)₂ ]²⁺, complex 2, to give the major adduct (pH* 8.5, 298 K) is 0.20 s^-1 and the off-rate is 0.018 s^-1. The 5′-GMP adducts were modelled by molecular mechanics calculations. These revealed possible hydrogen bonding between the dangling arm amino group and the 5′-phosphate. Supporting evidence for this came from solution studies with other GMP derivatives (3′, 5′-cyclic GMP, 3′-GMP and 9-ethylguanine). Our findings are discussed in terms of potential new methods for drug delivery and new approaches to drug design.

Full text of DOI:10.1039/b007670h

View Article Online / Journal Homepage / Table of Contents for this issue
Structure and dynamics of a platinum(II) aminophosphine complex
and its nucleobase adducts
Abraha Habtemariam,a John A. Parkinson,a Nicola Margiotta,a Trevor W. Hambley,b
Simon Parsonsa and Peter J. Sadler*a
a Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh, UK EH9 3JJ.
E-mail: P.J.Sadler@ed.ac.uk
b School of Chemistry, University of Sydney, NSW 2006, Australia
Received 21st September 2000, Accepted 1st December 2000
First published as an Advance Article on the web 25th January 2001
The anti-cancer aminophosphine complex, cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2N(CH2)2PPh2-P)]Cl, 1, is shown, by
1H and 31P NMR studies, to undergo reversible chelate ring-closing and -opening in aqueous solution. The rate for
this process (21 2 sϪ1 at pH* 8.5, 295 K), which is slow on the NMR timescale, varies significantly with pH values,
chloride ion concentration and temperature. The chelate-ring-opened form is favoured at acidic pH values. The
activation parameters which govern this process, have been determined via 31P-2-D EXSY NMR spectroscopy,
and the mechanism of chelate ring-opening is discussed. The X-ray crystal structure of the ring-opened complex,
cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ1.5H2O, 1a, shows that the Pt–N bond is relatively
long (2.147 Å). The nucleotide 5Ј-guanosine monophosphate (5Ј-GMP) binds via N7 and displaces the Cl ligand.
Unusually for a platinum() amine complex, GMP binding is rapid and reversible. Two isomeric GMP adducts are
formed in a 3:1 ratio below pH* 6 (pH meter reading in D2O). Above pH* 8.2, the predominant species are non
N7-bound adducts. The on-rate for 5Ј-GMP binding to the ring-closed form cis-[Pt(Me2N(CH2)2PPh2-N,P)2]2ϩ,
complex 2, to give the major adduct (pH* 8.5, 298 K) is 0.20 sϪ1 and the off-rate is 0.018 sϪ1. The 5Ј-GMP adducts
were modelled by molecular mechanics calculations. These revealed possible hydrogen bonding between the dangling
arm amino group and the 5Ј-phosphate. Supporting evidence for this came from solution studies with other GMP
derivatives (3Ј, 5Ј-cyclic GMP, 3Ј-GMP and 9-ethylguanine). Our findings are discussed in terms of potential new
methods for drug delivery and new approaches to drug design.
usually inert and have a high thermodynamic stability.5,6 We
Introduction
have used molecular mechanics calculations7 to shed light on
Our interest in platinum() aminophosphine complexes arises
the nature of the isomers formed in guanine nucleotide adducts.
from their potential anti-cancer activity, in particular their
ability to inhibit the growth of cisplatin-resistant cell lines.1
Chelated bis-aminophosphine metal complexes undergo ring-
opening, which can be controlled by factors such as the sub-
stituent on the amino nitrogen atom, the length of the N–P
linker, the state of protonation of the amino group and the
presence of a competing ligand such as chloride. Compounds
of this type can exist as a mixture of ring-opened and ring-
closed forms. One of our aims is to exploit this behaviour in
order to optimise their biological activity. Ring-closed forms
may be effective anti-mitochondrial agents since they can be
lipophilic cations,2 whereas ring-opened complexes may bind to
specific sites on DNA. There is potential for complexes of this
type to act as prodrugs and to be activated at the target site due
to local environmental factors such as pH and chloride ion
concentration.
Platinum() aminophosphine complexes bind to adenine
and guanine residues on DNA and can form cross-links
between these residues.3 They can also bind rapidly and
strongly to thymine under physiological conditions, in contrast
to platinum() am(m)ine anti-cancer complexes.4
Here we present a detailed study of the influence of temper-
ature, pH and chloride ion concentration on the structure and
dynamics of bis(1-N,N-dimethylamino-2-diphenylphosphino-
ethane)platinum() complexes in aqueous solution. We also
show that exchange of the nucleotide, 5Ј-guanosine monophos-
phate (5Ј-GMP) with 5Ј-GMP bound to the ring-opened com-
plex is relatively rapid on the NMR timescale, which is very
unusual for platinum() complexes since the Pt–N bonds are
Experimental
Materials and methods
9-Ethylguanine and the sodium salts of 5Ј-GMP and 3Ј,5Ј-
cyclic-GMP were purchased from Sigma. The complex cis-
[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)]Cl2,
1,
was prepared according to previously reported procedures.8
Reaction of 1 with AgNO3 and subsequent removal of
AgCl afforded complex 2, cis-[Pt(Me2N(CH2)2PPh2-N,P)2]-
(NO3)2.
Reaction of 1 with nucleotides. Typically, 0.3 ml of a fresh
solution of 40 mM 1 or 2 in D2O was mixed with 0.3 ml of
a 40 mM solution of the appropriate nucleotide in D2O in a
5 mm NMR tube. The pH* was adjusted to the desired value
with either DNO3 or NaOD.
pH Measurements. A Corning 145 pH meter, equipped with a
micro-combination electrode (Aldrich) calibrated with stand-
ard buffer solutions at pH 4 and 7, was used for all pH meas-
urements. The pH readings for D2O solutions are designated as
pH* values, and are uncorrected for the effect of deuterium on
the glass electrode.
Calculations of pK values. The pH titration curves were fitted
to the Henderson–Hasselbalch equation using the program
KALEIDAGRAPH.9
362
J. Chem. Soc., Dalton Trans., 2001, 362–372
This journal is © The Royal Society of Chemistry 2001
DOI: 10.1039/b007670h

View Article Online
Table 1 Crystallographic data for cis-[Pt(Me2N(CH2)2PPh2-N,P)-
Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ1,5H2O (1a)
1-D NMR spectra were acquired with 512–2048 transients
using inverse-gated decoupling into 16–32 K data points with a
relaxation delay of 0.2 s. The final digital resolution of the data
was 3–5 Hz/point. Particular attention was given to the acquisi-
tion of 31P-{1H} 1-D NMR data at 202 MHz under conditions
suitable for the accurate measurement of integrals. Data were
acquired over a spectral width of 40.65 kHz into 41662 data
points (acquisition time = 512 ms) using a 90Њ pulse (24 µs)
followed by a recycle delay of 10.0 s to allow for full recovery of
31P magnetization.
1a
Formula
M
C32H44ClN4O7.5P2Pt
897.19
Temperature/K
Crystal system
Space group
λ/Å
a/Å
b/Å
150
Monoclinic
P21/n
0.71073
12.009(3)
12.415(3)
24.278(5)
91.74(2)
3618.0(13)
4
1.647
4.092
1796
31P T1 Relaxation measurements. 31P T1 Relaxation times
were measured using the standard inversion-recovery sequence
(180Њ–t1–90Њ–acquire) in which t1 was varied from 100 ms to 10
s. Proton decoupling was applied during the acquisition period.
Data were Fourier-transformed into 2 K data points using a 2
Hz line broadening function. Relaxation times were calculated
using the T1 routine within Xwin-nmr (version 1.3, Bruker
Ltd.).
c/Å
β/Њ
U/Å3
Z
Dc/g cmϪ3
µ(Mo-Kα)/mmϪ1
F(000)
Crystal size/mm3
Data/restraints
Parameters
Goodness of fit on F2
Colour, habit
Reflections collected
wR2 indices (F2)
Conventional R[F > 4σ(F)]
ρmin, ρmax/e ÅϪ3
0.12 × 0.16 × 0.27
6392/30
424
1.041
Colourless, block
6393
0.08974
R1 = 0.0400 [4953 data]
Ϫ0.56, 1.24
Two-dimensional (2-D) NMR spectroscopy. For studies of
chelate ring-opening, 2-D 31P–31P EXSY NMR data sets were
1
acquired with H decoupling using the pulse sequence 90Њ–t1–
90Њ–mix–90Њ–acq where t1 is the incremental delay and mix is
the period during which chemical exchange can occur. All 2-D
data sets were acquired at 202 MHz in a phase-sensitive mode
using the TPPI scheme; 16 transients were acquired into 1024
data points for each of 128 t1 increments over ω1 and ω2 fre-
quency widths of 16.2 kHz. A recycle delay of 10.0 s allowed
for full recovery of 31P magnetization between transients. A
total of 64 dummy transients were included to achieve a steady
state prior to the acquisition of data. All 2-D data sets were
processed with identical parameters. Data in ω2 were apodized
using a 90Њ shifted sinebell-squared window function prior to
Fourier transformation. Data in ω1 were apodized using the
same window function and zero-filled once to 256 data points
prior to Fourier transformation. The transformed data were
phase- and base-plane-corrected prior to volume integration.
All 31P NMR data sets were referenced externally to 85% H3PO4
(in H2O) at 0 ppm. Both 1-D and 2-D NMR data sets were
acquired at various pH values, temperatures and mixing times
(see text for details). Probe temperatures were calibrated using
methanol and ethylene glycol samples. All NMR data acquisi-
tion and processing was carried out using Xwin-nmr (version
X-Ray crystallography
Crystals suitable for X-ray analysis were obtained from an
aqueous solution of complex 1 which gave rise to the nitrate
salt, cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)]-
(NO3)2ؒ1.5H2O, 1a, under low pH conditions (the pH of the
solution was lowered by addition of HNO3). Crystal data and
other experimental details are summarised in Table 1. The
crystal was handled under perfluoropolyether oil, and data were
collected on a Stoë Stadi-4 diffractometer equipped with an
Oxford Cryosystems low-temperature device. A numerical
absorption correction was applied by Gaussian integration
after refinement of the crystal dimensions against a set of
ψ-scans (transmission range 0.664–0.855).10 Data were meas-
ured to 2θmax = 50Њ in ω–θ mode with on-line profile fitting.11
The structure was solved by Patterson methods12 and com-
pleted by iterative cycles of least-squares refinement and differ-
ence Fourier syntheses.13 One of the nitrate anions is disordered
over two sites with equal occupancies (0.5); in one of these, the
anion appears to be H-bonded to a partial-weight water
molecule which occupies the site of the alternative anion
position. All major residual density lies in this region, suggest-
ing further unresolved disorder. The partial weight nitrates were
restrained to be geometrically similar to each other and to have
three-fold symmetry. All full-weight non-H atoms were refined
with anisotropic displacement parameters; H-atoms attached to
C or N were placed geometrically and allowed to ride on their
parent atoms.
1
1.3, Bruker Ltd.). 2-D H ge-COSY45, TOCSY and NOESY
NMR data sets were acquired at 600 MHz on cis-[Pt(5Ј-GMP-
N7)(Me2N(CH2)2PPh2-N,P)(Me2NH(CH2)2PPh2-P)]3ϩ (charge
on GMP is ignored in formulae in this paper) in the absence
of chloride ions. Magnitude mode 2-D ge-COSY45 data were
acquired with 8 transients into 4096 complex data points
(acquisition time = 410 ms) over a 5 kHz spectral width in both
dimensions for each of 1024 t1 increments (200 ms acquisition
time in ω1). Phase-sensitive 2-D TOCSY data were acquired
with 8 transients into 4096 complex data points (acquisition
time = 410 ms) over a 5 kHz spectral width in both dimensions
for each of 2 × 512 hypercomplex t1 increments. A mixing time
of 50 ms was applied using an MLEV17 spin-lock field. Phase-
sensitive 2-D NOESY data (mixing time 150 ms) were acquired
with 8 transients into 2048 complex data points (acquisition
time = 145 ms) over a 7 kHz spectral width in both dimensions
for each of 2 × 512 hypercomplex t1 increments, using a recycle
delay of 3.5 s. Data from NOESY experiments were used to
generate inter-proton distance constraints where appropriate.
CCDC reference number 186/2301.
See http://www.rsc.org/suppdata/dt/b0/b007670h/ for crystal-
lographic files in .cif format.
NMR spectroscopy
One-dimensional (1-D) NMR spectroscopy. Standard 1-D
NMR spectra were acquired at 298 K using Jeol GSX 270 (1H,
270 MHz; 31P, 109.67 MHz), Bruker DRX500 and DMX500
(1H, 500 MHz; 31P, 202 MHz) and Varian UNITYINOVA 600
(1H, 600 MHz) NMR spectrometers. Typically 1H 1-D
NMR spectra were acquired with 64–256 transients using a
pulse width of 45–60Њ into 16–32 K data points with a relax-
ation delay of 2–3 s. The final digital resolution of the data was
0.2 Hz/point. The residual HOD resonance from the solvent
was suppressed by presaturation. Similarly, standard 31P-{1H}
1
3J coupling constants were measured directly from 1-D H
1
and H-{31P} data sets and were used to define dihedral angle
constraints for modelling studies.
Adducts formed between 1 or 2 and the nucleotide 5Ј-GMP
1
were also studied by H–1H 2-D EXSY NMR spectroscopy at
600 MHz. For each of 2 × 512 hypercomplex t1 increments,14 32
transients were acquired into 2048 complex data points over a
J. Chem. Soc., Dalton Trans., 2001, 362–372
363

View Article Online
Table 2 Selected bond distances (Å) and bond angles (Њ) for
cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ
1.5H2O (1a)
spectral width of 6 kHz in both dimensions. Data sets were
acquired at temperatures of 288, 298, 308, 318 and 328 K. Mix-
ing times of 50, 100, 150, 200, 250, 300, 400, 500 and 1000 ms
were used to assess the build-up of exchange peaks for the pur-
poses of establishing suitable conditions for the calculation of
exchange rates. Relative integrals of diagonal and off-diagonal
peaks were measured by volume integration of the data. The
population matrices P for the various species were calculated
from integration of 1-D 1H NMR spectra. Rate constants were
extracted from the appropriate off-diagonal elements of the
computed kinetic matrix L. Data were processed using VNMR
(version 4.1, Varian Instruments Ltd.) or Xwin-nmr (version
1.3, Bruker Ltd.).
Pt(1)–N(4A)
Pt(1)–P(1A)
Pt(1)–P(1B)
Pt(1)–Cl(1)
N(4A)–Pt(1)–P(1A)
N(4A)–Pt(1)–P(1B)
P(1A)–Pt(1)–P(1B)
N(4A)–Pt(1)–Cl(1)
P(1A)–Pt(1)–Cl(1)
P(1B)–Pt(1)–Cl(1)
2.147(5)
2.2311(17)
2.2597(18)
2.3737(16)
84.54(15)
175.43(14)
99.63(6)
90.06(15)
172.90(6)
85.63(6)
Populations of complexes of 1 and 2 in solution were deter-
mined by integration of baseline-corrected 1-D 31P-{1H} NMR
spectra acquired under different conditions of pH and temper-
ature. Error limits were derived from the differences in integrals
of the two separate 31P signals arising from the ring-opened
form of the complex. Integrals from 2-D NMR spectra were
used for exchange rate calculations. Volume integrals were
measured only for 31P bound to I = 0 (central peak 66% of Pt).
Volumes from peaks on opposite sides of the diagonal were
averaged. Species populations for the nucleotide-complexes
were measured according to a similar procedure with data
acquired at 600 MHz using the H8 resonance from the nucleo-
tide base.
Analysis of chelate ring-opening dynamics. Exchange rates
were calculated using the program D2DNMR15 and two-,
three- or four-site models. Volume integrals and population
estimates were used as input parameters. Error limits were
based on variations in integrals and are compared with the out-
put from the exchange rate program. Variations in rates within
an exchange rate matrix were of the same order as errors calcu-
lated directly using D2DNMR. Bandshape analysis was carried
out using the program gNMR16 using a 31P-{1H} linewidth
of 7.0 Hz, a J(31P–31P) coupling constant of 17 Hz and by
variation of the exchange rate parameters.
2
Fig. 1 X-Ray crystal structure of the cation of cis-[Pt(Me2N-
(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ1.5H2O (1a) show-
ing one closed P,N chelate ring and one protonated dangling arm.
Calculation of activation parameters. Thermodynamic calcu-
lations were made on the basis of the exchange rate data using
the fitting program KALEIDAGRAPH.9 The data for com-
plexes 1 and 2 over the temperature range 276–301 K were fitted
to the Arrhenius relationship (eqn. 1):
Results
X-Ray crystal structure of complex 1a
Crystals of the protonated, ring-opened complex, cis-[Pt(Me2N-
(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ1.5H2O,
1a, suitable for X-ray analysis were obtained from aqueous
solution at pH 2.5. The X-ray crystal structure is shown in Fig.
1 and selected bond distances and angles are listed in Table 2.
Platinum() in the complex is square-planar and coordinated to
N, Cl and the two P atoms in a cis arrangement. The large
P(1A)–Pt(1)–P(1B) angle (99.63Њ) can be attributed to steric
interactions of the phenyl substituents on P. This is also
reflected in the reduction of the P(1A)–Pt–N(4A) angle to
84.5Њ. The Pt–Cl distance (2.374 Å) is as expected for Cl trans to
phosphorus.22 The two Pt–P distances are relatively short,23
with that trans to Cl (2.231 Å) being somewhat shorter than
that trans to N (2.260 Å). The Pt–N bond distance (2.147 Å) is
significantly longer than those observed for the related ring-
closed complex cis-[Pt(H2N(CH2)2PPh2-P,N)2]Cl2 (2.090, 2.111
Å),24 which does not have substituents on N.
Ea
ln kex = Ϫ
ϩ ln A
(1)
RT
where kex is exchange rate (sϪ1), Ea is activation energy (J
molϪ1), R = 8.31441 J KϪ1 molϪ1, T is absolute temperature (K),
and A is the frequency factor. Activation parameters ∆G‡, ∆H‡
and ∆S‡ derived for these exchange processes were calculated
using eqn. 2.17
kex
∆H‡
∆S‡
log10 ͩ ͪ= 10.318 Ϫ
ϩ
(2)
T
19.143T 19.143
Molecular modelling
Starting models were generated using the program Hyper-
Chem18 and energy-minimised using the MOMECSGI19 pro-
gram with a force field based on one described elsewhere.20 New
parameters were required for the phosphine donor groups and
were based on those for thioether donor groups and adjusted to
reproduce the geometry of 1. All models were energy minimised
using MOMECSGI until convergence was achieved. Energy
profiles associated with rotation about the Pt-N7 coordination
bond were generated by applying constraints21 to a torsion
angle about this bond and stepping through the angles corres-
ponding to conversion from one orientation to the other.
Dynamics of chelate ring-opening and -closing
1-D 31P-{1H} and 2-D 31P–31P–{1H} EXSY NMR spectra of an
aqueous solution containing a mixture of the ring-opened
complex 1 and ring-closed form 2 of the aminophosphine com-
plex (Scheme 1, processes not involving 5Ј-GMP) are shown in
Fig. 2. The 2-D NMR data are characterized by the following
features: (i) 3 peaks along the diagonal (a, b and c); (ii) off-
diagonal exchange cross-peaks (linked by solid lines in Fig. 2);
364
J. Chem. Soc., Dalton Trans., 2001, 362–372

View Article Online
Fig. 3 Variation of the rate of ring-closure of complex 1 (20 mM,
295 K) with pH* (Scheme 1).
cross-peaks, arise from the dynamic process of chelate ring-
opening and closing, which results in chemical exchange of the
31P atoms (vide infra).
Fig. 2 Top, 202 MHz 1-D 31P-{1H} NMR spectrum of a 40 mM solu-
tion of cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)]Cl2
1
The pH dependence of the rate of chelate ring-closure was
studied by titrating a 20 mM aqueous solution of complex 1
with DNO3 (pH* range 6.0–9.0; 295 K). Rates were determined
from 2-D 31P EXSY NMR data. A plot of chelate ring-closure
rate vs. pH* is shown in Fig. 3. At high pH (>8.5) the chelate
ring-closed form of the complex, 2, was dominant (74%). The
corresponding rate of ring-closure was relatively high (242 35
sϪ1 at 295 K, pH* 8.5). There was a reduction in the rate of
chelate ring-closure with decreasing pH* to the point where, at
pH* 5.5, this rate was negligibly small. In contrast, even at very
high pH, the chelate ring-closed species was never present to
more than 80% of the total amount of complex. The pH
dependences of the rates of chelate ring-opening and ring-
closure were also measured as a function of temperature over
the range 276–301 K. At pH* 8.5, the chelate ring-closing
process was dominant, especially at higher temperatures,
when the exchange rate for chelate ring-closure was an order
of magnitude higher than the rate of chelate ring-opening
(e.g. 242 35 sϪ1 compared with 21 2 sϪ1 for chelate ring-
opening at 295 K). In comparison, at pH* 7.1 the rate of
chelate ring-closure was much lower (59 4 sϪ1 at 295 K)
while the rate of chelate ring-opening showed a small increase
(31 2 sϪ1 at 295 K). At pH* 6.25 and 295 K, the rates of
chelate ring-opening (30 2 sϪ1) and chelate ring-closure
(22 2 sϪ1) were similar. Thermodynamic calculations were
made on the basis of the exchange rate data derived from 2-D
31P–31P EXSY NMR studies. The ∆G‡, ∆H‡ and ∆S‡ values
derived for the exchange processes shown in part of Scheme 1
are given in Table 3.
in D2O, pH* 7, 277 K, showing peaks a and c for the ring-opened
complex 1 as well as peak b for the ring-closed complex 2. 195Pt satellites
are marked with asterisks (*). See Scheme 1 for assignments. Bottom,
2-D 31P–31P-{1H}-EXSY NMR spectrum (mixing time 7.5 ms) of the
same solution. Solid lines link the exchange cross peaks.
The exchange processes between 1 and 2, shown in Scheme 1
influenced the line-shapes of peaks in 31P-{1H} NMR spectra
acquired at different temperatures. Raising the sample temper-
ature led to signal broadening. Simulations were carried out for
low pH conditions, when the exchange rates were slow enough
to allow the 2J(31P–31P) splitting (ca. 17 Hz) to be observed. The
results compare well with those derived from the 2-D NMR
studies described above. Good fits were obtained when the
exchange rate-constants derived from the D2DNMR analysis
were used for the band-shape analysis. This confirmed the
experimental results and cross-validated both methods used for
calculating the exchange rates of the chelate ring-opening and
-closing processes.
Scheme 1 X = N7 (3/3*), X = N1 (4).
(iii) sets of 195Pt–31P satellite signals (see Fig. 2). The diagonal
peaks a and c correspond to two 31P NMR resonances from the
chelate ring-opened form of the complex, 1, in which the 31P
atoms are non-equivalent. Resonance a is assigned to P in the
P,N chelate ring (Pa, Scheme 1) of the ring-opened complex, 1 (δ
36.3, 1J(195Pt–31P) = 3767 Hz). The high frequency shift of peak
a is consistent with its presence in a 5-membered chelate-ring25
1
and its associated J(195Pt–31P) coupling constant is typical of
31P trans to Cl.26 Resonance c is assigned to P of the pendant
arm ligand (Pc, Scheme 1) of complex 1 (δ Ϫ0.55, J(195Pt–
Chloride ion dependence. Reaction of 1 with AgNO3 afforded
complex 2, which was subsequently titrated with chloride (pH*
8.5, 295 K). 2-D 31P EXSY NMR data were acquired at differ-
ent stages of the chloride ion titration. These were used to
calculate the rates of chelate ring-opening. The results are
1
31P) = 3160 Hz). Its 1J(195Pt–31P) coupling constant is typical of
1
31P trans to N.26 The singlet b (δ 25.1, J(195Pt–31P) = 3292 Hz)
is assigned to the two equivalent 31P nuclei (Pb, Scheme 1) of
the chelate ring-closed complex, 2. The off-diagonal exchange
J. Chem. Soc., Dalton Trans., 2001, 362–372
365

View Article Online
Table 3 Activation parameters for chelate ring-opening and -closure processes
pH*
Processa
ln A
Ea /kJ molϪ1
‡
∆H‡/kJ molϪ1
∆S‡/J molϪ1 KϪ1
∆G‡/kJ molϪ1
8.5
C
O
C
O
C
O
17
25
2
1
33
49
5
3
46
31
3
1
Ϫ42 12
Ϫ114 17
59
64
6
5
7.1
21.4 0.1
18.5 0.3
16.7 0.4
16.8 0.4
42.6 0.2
37 0.6
33.5 0.7
32.7 0.8
40.23 0.15
34.6 0.6
31.05 0.75
30.4 0.8
Ϫ75
Ϫ99
Ϫ114
Ϫ113
1
3
3
3
62.3 0.4
63.2 1.5
63.9 1.6
63.1 1.7
6.25
a C = Chelate ring-closing, O = chelate ring-opening.
Table 4 δ (31P) and 1J(195Pt–31P) coupling constants (Hz) for nucleobase adducts of complexes 1 and 2
δ (31P)
Isomer-
ratio
Adduct
P,N Chelate-ring
Pendant-arm
Nucleotide-phosphate
1J(Pt,P)
2J(P,P)
pH*
5Ј-GMP
3Ј-GMP
9-ethylguanine
3Ј,5Ј-Cyclic GMP
26.41a
28.73
27.10b
29.00
Ϫ7.70a
Ϫ4.87
Ϫ6.67b
Ϫ5.15
2.28a
1.21
—
1.82b
1.08
—
3468,c 3203d
3495,c 3201d
3471,c 3142d
3483,c 3196d
20
20
21
21
3:1
5.06
5.00
5.60
5.10
1:1
e
30.98
Ϫ1.32
28.66a
28.94b
Ϫ4.78a
Ϫ5.06b
—
—
1.1:1
a Major adduct. b Minor adduct. c trans to N7. d trans to chelate-ring N. e Only one isomer observed.
Fig. 4 Dependence of chelate ring-closure (grey) and ring-opening
(black) rates on chloride concentration for complexes 1 and 2, respect-
ively. Conditions: 40 mM 1 ϩ 2, pH* 8.5, 298 K.
shown in Fig. 4. The rates of chelate ring-opening and closing
increased with increasing chloride ion concentration.
Binding of 1 to 5Ј-GMP
When an equimolar quantity of 5Ј-GMP was added to an
aqueous solution of 1 and monitored using both 1H and
31P-{1H} NMR spectroscopy (293 K, pH* 5.5), the reaction was
complete within 5 minutes. The H8 singlet 1H NMR resonance
of free 5Ј-GMP (δ 8.10) almost disappeared, while two new
singlets appeared to higher frequency at δ 9.31 and δ 9.12. The
1
Fig. 5 (a) pH* dependence of the chemical shift of the H8 H NMR
peak of 5Ј-GMP in complex 3 (ꢀ), 3* (᭜) and free 5Ј-GMP (᭹) at 298
K, showing the lack of protonation of N7 for 3/3* at low pH* and
deprotonation of N1 and phosphate at higher pH*. The corresponding
pKa1 values were 5.31 (3) and 5.49 (3*) (attributed to protonation of the
phosphate group of 5Ј-GMP) and 8.51 (3) and 8.43 (3*) (associated
with deprotonation of N1 of 5Ј-GMP). (b) Dependence of the chemical
shift of the P atom of the pendant arm with pH* for 3 (ꢀ) and 3* (᭜).
The corresponding pKa1 values were 5.29 (3) and 5.46 (3*), respectively.
Above pH* 9, data were incomplete and not fitted due to resonance
broadening.
new resonances suggested the formation of two 5Ј-GMP
adducts in a 3:1 ratio, corresponding to major and minor
adducts (3 and 3*), respectively. The high-frequency shift of
1
these H8 H NMR resonances compared with that of free
5Ј-GMP is typical of metal coordination through the N7 atom
of guanine.27 Above pH* 6.5, an additional H8 resonance
appeared (δ 8.10), which may be assignable to an N1 bound
adduct, 4 (vide infra). The 31P-{1H} NMR spectrum at pH* 5.5
also showed new sets of major and minor resonances corre-
sponding to the presence of 3 and 3*. These resonances
appeared in three groups: (a) two sharp doublets assignable to P
in P,N chelate rings; (b) two broad doublets assignable to P in
P-bound chelate ring-opened ligands; (c) two singlets arising
from the phosphate groups of coordinated 5Ј-GMP. Chemical
shift and coupling constant data associated with each of these
adducts are given in Table 4. The similarities of the NMR
parameters for 3 and 3* suggest that they are very similar
adducts, perhaps isomers.
The binding of 5Ј-GMP through N7 was further confirmed
by a pH titration of 3/3* (Scheme 1). For free 5Ј-GMP,
protonation at N1 (pKa = 9.7) and N7 (pKa = 2.4) is known to
shift the H8 1H NMR resonance to higher frequency. In
366
J. Chem. Soc., Dalton Trans., 2001, 362–372

View Article Online
Fig. 6 The relative populations of all the species formed from reaction
of equimolar amounts of 2 and 5Ј-GMP as a function of pH*: 3 (ꢀ), 3*
(᭜) and 4 (ᮀ).
Fig. 7 Displacement of 5Ј-GMP from complex 3/3* on addition of
various amounts of ClϪ as determined by 31P NMR. Conditions: 20
mM of complex (in D2O), pH* 8.5, 298 K.
contrast, phosphate protonation (pKa = 6.2 and ≤1) causes
chemical shift changes to low frequency.28,29 Over the pH*
range 2.4–11.5, 3 and 3* were found to be stable (Fig. 5a),
although the corresponding ratios of the major and minor
isomers varied (from 3:1 at pH* 5.5 to 1:1 at pH* 8.5; Fig. 6).
No change in the chemical shift of H8 was observed over the
pH* range 2–4, confirming the binding of platinum to N7 of
5Ј-GMP. In contrast, large variations were observed over the
pH* range 5–9. The pKa1 values calculated for the two isomers
are 5.31 for 3 and 5.49 for 3*, which can be attributed to
protonation of the phosphate group of 5Ј-GMP. The pKa2
values of 8.51 and 8.43 for 3 and 3*, respectively, can be associ-
ated with protonation of N1 of 5Ј-GMP. No variation of the
H8 resonance at δ 8.10 (described above) was observed over the
pH range 8–10, suggesting that 4 is an N1-bound adduct.
The pH titration of 3/3* was also followed by 31P NMR
spectroscopy. A plot of chemical shift (δ) against pH for both
chelate and dangling-arm phosphorus atoms showed identical
profiles (Fig. 5b), from which pKa1 values of 5.29 and 5.46 for 3
and 3*, respectively, were calculated. These can be associated
with the protonation of the phosphate group and agree well
with the values above, derived from the 1H NMR data.
chloride ions resulted in increasing amounts of the ring-opened
complex 1 (Fig. 7). With a 10-fold molar excess of chloride ions
(200 mM) present, 50% of the bound 5Ј-GMP was displaced.
These data suggest that 5Ј-GMP binds about 10 times more
strongly than chloride.
Effect of temperature. Variable temperature 1H and 31P-{1H}
NMR experiments were carried out on 3/3* at pH* 5.5 in order
to detect possible chemical exchange between the major isomer
1
3 and minor isomer 3*. No coalescence of the H8 H NMR
signals was observed, even at 353 K. However, resonance
broadening was significant, especially in the aromatic region of
1
the H NMR spectra. The ratio of the two isomers remained
relatively constant over the temperature range 278–353 K. A
significant increase in the amount of free 5Ј-GMP was observed
at the expense of both adducts 3 and 3*. Twice as much free
5Ј-GMP was observed at 328 K compared with 278 K. The
original ratios of free and bound forms of 5Ј-GMP were
restored on returning the sample to ambient temperature. The
corresponding 31P-{1H} NMR study revealed broadening of
resonances assigned to the P-bound ring-opened ligand (Pc)
with increasing temperature, whereas the resonances from the P
in the P, N chelate ring (Pa) remained sharp with well resolved
2J(31P–31P) couplings of 17 Hz. The amount of free 5Ј-GMP
and free complex increased with increasing temperature, there-
by confirming earlier observations that the dissociation of
5Ј-GMP was enhanced at elevated temperatures.
The 31P-{1H} NMR spectra for the pH titrations also showed
a number of additional peaks (collectively assigned to species
4), in the ranges of δ 28.0 to δ 25.0 and δ Ϫ4.0 to δ Ϫ9.0. These
correspond to the P atom within the chelate ring (Pa) and the P
atom of the dangling arm (Pc), respectively, above pH* 6.5.
Thus the 31P-{1H} NMR data are consistent with the 1H NMR
data described above, which indicated the formation of an
N1-bound adduct.
Exchange studies
A plot of the relative populations of all the species in solu-
tion, derived from the integration of the relevant peaks in the
31P-{1H} spectra, as a function of pH*, is shown in Fig. 6.
Complex 4 became the dominant adduct above pH* 8.5.
Species 4 appears to be derived from species 3, since the amount
of 3* appears to remain constant over the measured pH range.
Since the exchange processes were slow on the NMR timescale,
1H 2-D-EXSY NMR spectroscopy was used to determine
exchange rates. The effects of temperature (over the range 288–
318 K) and of pH (over the range of 5.4–8.5) on the dynamics
of the exchange process were studied under chloride-free condi-
1
tions. A H–1H 2-D-EXSY NMR spectrum acquired for an
equimolar mixture of 2 and 5Ј-GMP at pH* 5.4 (temper-
ature = 298 K) is shown in Fig. 8a. The cross-peaks c/cЈ result
from chemical exchange between the major and minor adducts,
3 and 3*. The presence of cross-peaks a/aЈ and b/bЈ indicate
exchange between isomer 3 and 5Ј-GMP, and isomer 3* and
5Ј-GMP. Spiking with excess 5Ј-GMP confirmed that the
resonance at δ 8.10 corresponded to that of the H8 resonance
of free 5Ј-GMP.
The exchange processes under these conditions (pH 5.4, 298
K) are described by Scheme 2 and rate constants for the
exchange processes were determined to be: k1 = 0.05 0.01 sϪ1;
k2 = 0.06 0.01 sϪ1; k3 = 0.07 0.01 sϪ1; k4 = 0.03 0.01 sϪ1;
k5 = 0.09 0.01 sϪ1; k6 = 0.20 0.01 sϪ1. The formation of 3
Effect of chloride ions. In all of the reactions of 1 with
5Ј-GMP, the presence of a small amount of unreacted 1 was
always noted, even after the addition of two mole equivalents
of 5Ј-GMP. The relative binding strengths of 5Ј-GMP and ClϪ
were studied by a reverse titration of the 5Ј-GMP complex 3/3*
with chloride ions. Complex 3/3* was prepared by precipitating
all the chloride from complex 1 with AgNO3 (to give complex 2)
prior to the addition of an equimolar amount of 5Ј-GMP. The
resulting 31P-{1H} NMR spectrum showed formation of the
two 5Ј-GMP adducts (3 and 3*), but no signals for the ring-
closed complex 2. This solution of 3/3* (20 mM, pH* 8.5) was
titrated with chloride. The addition of increasing quantities of
J. Chem. Soc., Dalton Trans., 2001, 362–372
367

View Article Online
Table 5 1H NMR assignments for the sugar protons of 5Ј-GMP and isomeric GMP adducts 3 and 3*, (pH* 5.5, 298 K)
δ (1H)
Adduct
H1Ј
H2Ј
H3Ј
H4Ј
H5Ј or H5Љ
H5Љ or H5Ј
3
5.930
5.954
5.933
5,028
4.795
4.716
4.586
4.498
4.528
4.498
4.335
4.446
4.169
4.050
4.184
4.120
4.050
4.081
3*
5Ј-GMP
between 3 and 4 or between 3* and 5Ј-GMP are absent. At
higher temperature, the 3 to 3* conversion becomes apparent,
the data revealing a four-site exchange process as represented in
Scheme 2. The interconversion rate between 3 and 3* varied
from 0.01 sϪ1 at 298 K to 0.55 sϪ1 at 328 K for the forward
process, compared with variations from 0.04 sϪ1 at 298 K to
2.29 sϪ1 at 328 K for the reverse process.
1H NMR assignments and nOes. In order to investigate fur-
ther the nature of the complexes formed between 2 and 5Ј-
GMP in solution, and to rationalise the observed exchange
characteristics, 2-D ge-COSY45 and TOCSY data sets were
1
acquired and analysed. The H NMR resonance assignments
are shown in Table 5. A number of through-space interactions
were identified from NOESY data. In particular, strong nOes
were observed between base H8 and sugar H2Ј protons in both
3 and 3* indicating a preferred sugar C2Ј-endo geometry. At 288
K, nOe and exchange cross-peaks were negative. As the tem-
perature was raised, the nOes reduced in intensity compared
with the exchange cross-peaks, which increased in size. At 308
K all nOes were close to zero. At temperatures above 308 K, the
nOe cross-peaks were positive, while exchange cross-peaks
remained negative. The nOes indicated a preferred anti orient-
ation for the base moiety with respect to the sugar ring in each
case.
Fig. 8 600 MHz 1H 2-D-EXSY NMR spectra of a solution of
equimolar 2 and 5Ј-GMP in the region containing resonances for H8 of
5Ј-GMP. (a) pH* 5.4, 298 K; (b) pH* 8.5, 288 K; (c) pH* 8.5, 318 K;
G = free 5Ј-GMP. 4 = N1 bound complex. Arrows in (c) indicate the
presence of new exchange cross-peaks, which appear on raising the
temperature of the sample at pH* 8.5.
Reaction of 1 with GMP analogues
Reactions were carried out between 1 and guanine derivatives
differing either in the position of the phosphate group (e.g. 3Ј,
5Ј-cyclic GMP or 3Ј-GMP) or in the nature of the group
attached to the base (e.g. 9-ethylguanine). The reactions were
followed by 1-D 31P-{1H} NMR spectroscopy and the results
are summarised in Table 4. Data for all of the adducts, recorded
at pH* ≈5.0, were consistent with P in a P, N chelate ring,
together with P trans to N7. The chemical shifts and coupling
constants were similar to those for the 5Ј-GMP adduct. The
major difference was that the 3:3* isomer ratios were close to
1:1 for the 3Ј-GMP and 3Ј, 5Ј-cyclic-GMP adducts. In the case
of 9-ethylguanine only a single adduct (complex 5) was formed.
Molecular modelling
The force field reported previously,20 and extended here to
include phosphine ligands, accurately reproduced the structure
of 1. However, given the small data set used to derive the
parameters for the phosphine moiety, the results should be
taken as being more qualitative than quantitative. To comple-
ment the NMR and chemical studies described in previous
sections, and to investigate the nature of the two isomers
formed, molecular models of cis-[Pt(9-ethylguanine-N7)(Me2N-
(CH2)2PPh2-N,P)(Me2NH(CH2)2PPh2-P)]3ϩ (5) and of cis-
[Pt(5Ј-GMP-N7)(Me2N(CH2)2PPh2-N,P)(Me2NH(CH2)2PPh2-
P)]3ϩ (3/3*) were built. Complex 1 is not chiral but when the
chloro-ligand is replaced by an unsymmetrical ligand, such as
9-ethylguanine, enantiomers are generated, a consequence of
the incoming ligand having a head-to-tail directionality. Thus
two isomers are possible for 5. One isomer has the guanine H8
and the ring-opened amino pendant arm of the ligand on the
same side of the coordination plane (5a in Scheme 3). In the
Scheme 2
from 2 and 5Ј-GMP therefore occurred at a similar rate to the
dissociation process whereas the formation of 3* from 2 and
5Ј-GMP occurred at twice the rate compared with dissociation
of 3* into 2 and 5Ј-GMP. The rate of conversion of 3* to 3 is
twice the rate of formation of 3* from 3.
The same mixture was also studied at pH* 8.5 over the tem-
perature range 288–328 K. Two 1H–1H 2-D-EXSY data sets are
presented for the sample at 288 K and 318 K (Fig. 8). The data
show the presence of 4 in exchange with 3/3*. At low temper-
ature (Fig. 8b) no conversion between 3 and 3* is observed, as
evidenced by the absence of an exchange cross-peak between
the H8 resonances corresponding to 3 and 3*. The dominant
exchange processes are between 3 and 5Ј-GMP and between 3*
and the new species 4. The data suggest that exchange processes
368
J. Chem. Soc., Dalton Trans., 2001, 362–372

View Article Online
Table 6 Strain energies
Strain energy/
Compound
Isomer
kJ molϪ1
5
5a
28.9
29.2
46.9
50.8
46.8
49.6
5b
3/3*
3a
3b
3a*
3b*
Scheme 3
ions), one of the Pt–N bonds in the ring-closed complex, 2, is
labile and susceptible to cleavage. In contrast, it is difficult to
ring-open the unsubstituted complex cis-[Pt(H2N(CH2)2PPh2-
P,N)2]Cl2 even at much lower pH* values.24 Therefore under
physiological conditions one of the 5-membered chelate rings
of 2 can open, thereby providing a site for reaction with
biomolecules such as DNA.
other isomer, the guanine H8 and the ring-opened amino pen-
dant arm of the ligand are on opposite sides of the coordin-
ation plane (5b in Scheme 3). Molecular mechanics calculations
show that isomer 5a is preferred by about 0.3 kJ molϪ1 over 5b
(Table 6).
When the incoming ligand is itself chiral, as is the case for
5Ј-GMP, diastereomers are generated and these give rise to the
separate peaks observed in the NMR spectra of 3/3*. Four
energetically distinct species are possible (Scheme 4), since there
are two diastereomeric forms of each rotational isomer. The
diastereomers have similar strain energies (Table 6), but in each
case the rotational isomers 3a and 3a* (Scheme 4) with the
guanine H8 located on the same side of the coordination plane
as the amino group of the ring-opened ligand are preferred by
3–4 kJ molϪ1 over 3b and 3b* (Table 6). Barriers to rotation
about Pt–P (ring-opened) and Pt–N7 (guanine) bonds were
estimated using constrained refinement to step through the
rotation process. For the former, the barrier is approximately 55
kJ molϪ1 and for the latter it is greater than 100 kJ molϪ1. Thus,
rotation about the Pt–P bond is a facile process that will occur
at a significant rate but rotation about Pt–N7 is impeded by
interactions between the guanine and the coordination plane
and will occur only at elevated temperatures.
Dynamics of chelate ring-opening and -closing
The assignments of 31P NMR signals for the chelate ring-closed
complex, 2, and ring-opened complex 1 were based on chemical
shifts and coupling constants. Under the experimental condi-
tions used, no aquated adducts were observed. Resonance
broadening observed in the 1-D 31P NMR spectra suggested the
existence of an exchange process between 1 and 2. 2-D 31P
EXSY NMR, variable temperature NMR studies and NMR
simulations, confirmed that this was the case. The equilibrium is
dependent on chloride ion concentration and pH. The dynamic
nature of this chelate ring-opening and -closing process was
investigated by 2-D 31P–31P EXSY NMR spectroscopy. Short
mixing times (<10 ms) and long relaxation delays (>10 s) were
required in order that consistent exchange rate data were
obtained from the EXSY data sets. This was a reflection of the
rates of exchange which were found to vary significantly, but
were never less than 9 sϪ1 for either opening or closing rates, and
often much faster. 2-D EXSY NMR has been extensively
applied to the analysis of dynamic equilibria, which are slow on
the NMR timescale.31,32 The extraction of rate constants from
such data, carried out either by initial rate, iterative or direct
matrix transformation analysis, provides details of the thermo-
dynamics governing dynamic equilibria.
Discussion
X-Ray structure of 1
The protonated, ring-opened complex cis-[Pt(Me2N(CH2)2-
PPh2-N,P)Cl(Me2NH(CH2)2PPh2-P)](NO3)2ؒ1.5H2O, 1a, has a
long Pt–N bond relative to those in previously reported
aminophosphine complexes, which incorporated less hindered
nitrogen centres.24 The weakening of the Pt–N bond in 1 can
therefore be ascribed partly to steric effects.30 Under conditions
of low pH or in the presence of nucleophiles (such as chloride
The pH dependence of the chelate ring-opening and -closing
process is significant in physiological terms. At high pH > 7.5
the rate of ring-closure is high whereas lowering the pH to less
Scheme 4
J. Chem. Soc., Dalton Trans., 2001, 362–372
369

View Article Online
than 5.5 causes the ring-closing rate to tend towards zero
(Fig. 3). Under such conditions the ring-opened form, 1,
dominates. The implication for tumor cells, which often experi-
ence low associated pH conditions,33 will be for the complex to
react more readily at tumor cell sites as a result of the available
binding site on the metal centre. The ring-closed form, present
at higher pH and being cationic and lipophilic, is likely to bind
to mitochondrial membranes and disrupt membrane poten-
tials.2 Chloride ions also significantly affect the position of the
equilibrium. The rate of ring-opening increases with increasing
chloride ion concentration, consistent with a bimolecular
process. The increase in the rate of ring-closure is consistent
with an increase in the amount of ring-opened complex as the
chloride concentration increases. The proportion of ring-closed
species is likely to increase inside cells where the chloride
concentration (ca. 4 mM) is usually lower than outside cells
(ca. 104 mM).
were present in the 31P NMR spectrum than can be accounted
for by binding at the N7 and N1 positions alone.
For 3/3*, four energetically distinct species are possible, since
there are two diastereomeric forms of each rotamer (Scheme 4).
Four distinct resonances would be anticipated from the 4
diastereomers, but only two isomers in a 3:1 ratio (pH* 5.5)
were observed experimentally, as was indicated by the appear-
ance of two H8 signals and two sets of resonances in the 1H and
31P NMR spectra. Two conditions could give rise to this obser-
vation: i) only one set of the two possible pairs of rotamers (3a
and 3b or 3a* and 3b*, Scheme 4) is favoured, or ii) rapid inter-
conversion between diastereomers in each set of rotational
isomers occurs, giving rise to two sets of averaged NMR signals
corresponding to 3 and 3*. The latter seems to be the case as
suggested by the molecular mechanics studies (vide infra). The
major isomer was most favoured in the pH range where the
phosphate group was largely deprotonated. The phosphate
group may be involved in H-bonding with the amine group of
the dangling arm, which could favour the stabilisation of the
major isomer. This was further elaborated by molecular model-
ling studies (see the section on molecular modelling). Studies
with both 3Ј-GMP and 3Ј, 5Ј-cyclic GMP showed changes in
the isomer ratio, with each being present in equal amounts in
each of these cases. This supports the assertion that the position
of the phosphate group on the chiral sugar moiety of GMP
plays a prominent role in influencing the orientation of the
purine ring in relation to the metal centre. The NMR spectrum
of the 9-ethylguanine adduct, 5 (which lacks the chiral sugar)
contained only a single H8 resonance in the 1H NMR spectrum,
and gave rise to only one set of 31P NMR resonances.
Activation parameters. Studies of the exchange rates at vari-
ous pH values and temperatures enabled activation parameters
to be obtained. The negative entropies of activation for this
equilibrium, ∆S‡ (Table 3) are features normally associated
with an intra-molecular process.34–36 This implies that the
two-site exchange process, involving chelate ring-opening and
ring-closing, is governed by an associative mechanism, involv-
ing a 5-coordinate intermediate. Formation of the Pt–Cl bond
and dissociation of the Pt–N bond would then occur in a con-
certed fashion, with the incoming nucleophile approaching
from above or below the Pt square plane. Examination of
models based on the X-ray crystal structure of 1a show that this
is the only feasible mode of approach for incoming nucleo-
philes. In the absence of any chloride ions, chelate ring-opening
occurs only at very low pH values (<2).24
Exchange rate analysis
At pH* 5.4 it was clear that a classical three-site exchange
process existed between 3, 3* and 2 ϩ 5Ј-GMP in the absence
of chloride ions. The exchange processes were shown to be slow
with conversion lifetimes in the range 5–50 s. At pH* 8.5 the
exchange system was altered by the presence of the additional
species 4. At 288 K, no interconversion occurred between 3 and
3*. However, 3 was in exchange specifically with 2 ϩ 5Ј-GMP
(denoted by the resonance G in Fig. 8c, corresponding to
free 5Ј-GMP) and 3* was in exchange specifically with 4.
With increasing temperature, exchange between 3 and 3* was
observed together with additional exchange processes, namely
3 to 4 and 3* to 2 ϩ 5Ј-GMP, processes which were absent at
lower temperatures.
The NMR data show that once conversion between 3 and 3*
becomes more facile, conversion between 3 and 4 and between
3* and 2 ϩ 5Ј-GMP can be observed, without the need for
conversion between 4 and 2 ϩ 5Ј-GMP. Thus it appears that
formation of 4 from 2 ϩ 5Ј-GMP cannot occur directly at this
pH, but must go via 3/3*. This suggests that pre-binding at N7
is required to lower the pKa value of N1 (pKa of N1 is lowered
by ca. 1.2 log units in 3/3*) so that formation of 4 becomes
favourable through this route.
Structure and dynamics of the 5Ј-GMP adducts of 2
The reaction of [Pt(Me2N(CH2)2PPh2-P,N)2]2ϩ (2) with 5Ј-
GMP was rapid, being essentially complete by the time the first
NMR spectrum had been acquired. This is in contrast to reac-
tions of platinum() ammine complexes with 5Ј-GMP, which
are slow, but similar in rate to palladium() complexes.37 The
high reactivity of platinum() in 2 can be attributed to the high
trans effect of the phosphorus atom. The binding of 5Ј-GMP
through N7 was established on the basis of the shift of the H8
NMR resonance of the bound 5Ј-GMP compared with the free
form, when studied over a wide range of pH* values (Fig. 6).
1J(P, Pt) coupling constants of 3203 and 3468 Hz, are typical of
31P trans to N,8,26 trans to N within the chelate ring and to N7
in the purine ring, respectively. The major site of attack by
platinum am(m)ine anti-cancer complexes is N7 of guanine,
which is readily accessible in the major groove of duplex DNA
and is the strongest electron donor of the four bases.38 The
primary binding site for 2 under acidic or neutral conditions is
the N7 position on 5Ј-GMP. However, under basic conditions
N1 appears to bind. Complexes containing Pt–N1 binding have
been prepared previously39 but under physiological conditions
N1 of guanine in duplex DNA is involved in Watson–Crick
base pairing. This position is therefore not available for metal
binding. Metal ion binding at the N7 position acidifies the
proton at N1 by about 1.5–2 log units,40 thus rendering the N1
site more accessible to platinum. The presence of an inter-
mediate binuclear species, in which platinum is bound simul-
taneously to both N1 and N7, is also possible. At pH* 5–6 a
migration of [Pt()(dien)]2ϩ from inosine N7 to N1 occurs via
slow build-up and loss of a binuclear intermediate.41 The N3
atom of guanine is not usually a strong Pt-binding site, because
it is in a sterically-crowded position. Simultaneous binding of
platinum() to three different sites (N7, N1, N3) of a guanine
nucleobase has been reported by Lippert et al.42 Thus binding at
the N3 site cannot be excluded in the present study. More peaks
Effect of chloride ions. In platinum() am(m)ine complexes
nucleobase binding is stable due to the inertness and high
thermodynamic stability of the Pt–N bonds.5 The facile dis-
placement of 5Ј-GMP in 3/3* by chloride is unusual in com-
parison with platinum() am(m)ine complexes. This is probably
due to the relatively strong trans effect of the coordinated phos-
phino group. Normally the displacement of nucleobases from
platinum requires strong nucleophiles such as CNϪ and sulfur
ligands.43 However, in a few cases migration of coordinated
platinum() from one nucleobase to another has been reported
in both single-stranded44,45 and double-stranded oligo-
nucleotides.45,46 The rates of displacement of guanine from
cis-[Pt(NH3)2(guanine-N7)2]2ϩ by thiourea and other sulfur-
containing nucleophiles have been studied by 13C NMR
370
J. Chem. Soc., Dalton Trans., 2001, 362–372

View Article Online
spectroscopy and were found to be relatively slow (the rate of
displacement at 316 K was 1.5 × 10Ϫ5 sϪ1 for thiourea).47 For
[Pt(dien)(guanine-N7)]2ϩ, the displacement of the coordinated
nucleobase follows an associative mechanism in both neutral
and acidic aqueous solution, as do reactions with thiourea in
neutral solutions.48
5-coordinate intermediate. 31P-2-D EXSY NMR spectroscopic
studies showed that the rate of ring-opening is accelerated at
low pH and high chloride ion concentration. The X-ray crystal
structure of cis-[Pt(Me2N(CH2)2PPh2-N,P)Cl(Me2NH(CH2)2-
PPh2-P)](NO3)2ؒ1.5H2O, 1a, was determined and the complex
has a relatively long Pt–N bond (2.147 Å). The nucleotide 5Ј-
GMP binds rapidly and reversibly to complex 2 via N7 which
induces aminophosphine chelate ring-opening. Above pH* 8.2,
the predominant species appear to be non N7-bound adducts.
The facile displacement of 5Ј-GMP in cis-[Pt(5Ј-GMP-N7)-
(Me2N(CH2)2PPh2-N,P)(Me2NH(CH2)2PPh2-P)]3ϩ (3/3*) by
chloride is unusual in comparison with platinum() am(m)ine
complexes. 1H 2D-EXSY NMR studies reveal a slow exchange
process between Pt-bound and free-5Ј-GMP together with the
existence of two isomers. The two isomers observed by NMR (3
and 3*) can be formulated as diastereomers, which essentially
differ in the orientation of the plane of the purine residue. In
one isomer the O6 of the guanine is above, and in the other
isomer, below the platinum plane. Molecular mechanics calcu-
lations suggest that these nucleotide adducts are stabilized by
H-bonding between the dangling arm amino group and the
5Ј-phosphate of the nucleotide.
The features described here have at least two significant bio-
logical consequences for the design and action of Pt-based
drugs. Firstly, the reversible binding of platinum() amino-
phosphine complexes to 5Ј-GMP with lifetimes in the range
5–50 s is in marked contrast to the behaviour of cisplatin bind-
ing. Binding of cisplatin to DNA is usually non-reversible and
may be kinetically driven. A platinum() aminophosphine
complex, in contrast, is likely to be mobile on DNA and to
establish an equilibrium in which it is bound to the most
thermodynamically favoured site. The second biological con-
sequence is a dual mode of action. The ring-closed form of the
aminophosphine complex is formed in environments of low
chloride concentration and therefore antimitochondrial activity
could form part of its mode of action. However, increasing the
bulk of the substituents on N can hinder ring closure, a situ-
ation which may favour direct coordination to DNA bases.24
Molecular modelling. The modelling studies using the
9-ethylguanine adduct of 1, cis-[Pt(9-ethylguanine-N7)(Me2N-
(CH2)2PPh2-N,P)(Me2NH(CH2)2PPh2-P)]3ϩ, 5, revealed that
isomer 5a is preferred over isomer 5b by ca. 0.3 kJ molϪ1.
Isomer 5a is arranged in such a way that the purine moiety is
orientated with the H8 of the guanine on the same side of the
coordination plane as the amino pendant arm of the ligand in
contrast to isomer 5b, in which the purine moiety has the
opposite orientation. This preference is in spite of the fact that
the exocyclic O6 (guanine) is capable of forming a hydrogen
bond with the amino group of the pendant arm. These isomers
can be inter-converted in two ways. The first way is by rotation
about the Pt–P (ring-opened) bond or by rotation about the
Pt–N7 (guanine) bond. The second way is by the dissociation of
the 9-ethylguanine ligand, followed by closure of the amino-
phosphine chelate ring, opening of the other chelate ring and
coordination of the 9-ethylguanine in the same orientation with
respect to the coordination plane. A further stabilising feature
could be π–π stacking interactions between the 6-membered
aromatic ring of the purine moiety and one of the phenyl
groups. In the case of the 5Ј-GMP adducts, which have a chiral
sugar, diastereomers are generated. The preference for the
formation of isomers 3a and 3a* by 3–4 kJ molϪ1 over 3b and
3b* is consistent with the result for the 9-ethylguanine adduct,
which revealed an inherent preference for 5a. Hydrogen bonds
are possible for both rotational isomers, these being between
the phosphate and amino groups in the case of 3a and 3a* and
between guanine O6 and the amino group for 3b and 3b*. In
modelling these hydrogen bonds, no account was taken of the
differences in their energies. However, the hydrogen bond
between the protonated dangling arm amino and phosphate
groups is likely to be more stable since both groups are charged.
Therefore, the preference for 3a and 3a* is likely to be under-
estimated. Intramolecular hydrogen bonding between the NH
of cis-PtA2 and the phosphate backbone of DNA is believed to
influence the structure of the Pt–DNA adduct and hence the
activity of the platinum complex.27,49 Recent X-ray crystallo-
graphic studies of a 16-base-pair DNA duplex containing a
site-specific cis-[Pt(NH3)2{d(GpG)-N7(1),-N7(2)}] have shown
that the NH3 group is within 3.2 Å of the phosphate oxygen
atom, which is close enough to participate in hydrogen bond-
ing.50 Furthermore, in a DNA dodecamer containing a 1,2-
intrastrand d(GpG)-cisplatin adduct, one of the ammine
ligands bound to platinum is H-bonded to a phosphate
oxygen.51 In BipPt(guanine-N7)2 (Bip = 2,2Ј-bipiperidine) the
rotamer distribution was found to be influenced by phosphate –
NH(Bip) hydrogen bonds.52
Acknowledgements
We thank the BBSRC and EPSRC for support and Dr. Keith
Orrell, University of Exeter for the D2DNMR software pack-
age. We have benefited from stimulating discussions with
members of the EC COST Action D8.
References
1 A. Habtemariam and P. J. Sadler, Chem. Commun., 1996, 1785.
2 S. J. Berners-Price, R. J. Bowen, M. J. McKeage, P. Galettis and
P. C. Healy, Coord. Chem. Rev., 1999, 185, 823 and references
therein; S. J. Berners-Price, R. J. Bowen, M. J. McKeage, P. Galettis,
L. Ding, B. C. Baguley and W. J. Brouwer, Inorg. Biochem., 1997, 67,
154.
3 K. Neplechová, J. Kasˇpárková, O. Vrána, O. Novákoaá, A.
Habtemariam, B. Watchman, P. J. Sadler and V. Brabec, Mol.
Pharmacol., 1999, 56, 20.
4 N. Margiotta, A. Habtemariam and P. J. Sadler, Angew. Chem., Int.
Ed. Engl., 1997, 36, 1185.
5 F. Basolo and R. G. Pearson, Mechanisms of Inorganic Reactions,
Wiley, New York, 1967, Ch. 5.
Inter-conversion between rotamers can occur via two
processes. The first of these is by simultaneous rotation about
the Pt–P (ring-opened) and Pt–N7 (guanine) bonds. The
second is by the dissociation of 5Ј-GMP, for inter-converting
the enantiomers of 3. The latter dissociative process seems to be
more likely in view of the positive ∆S‡ values obtained for the
inter-conversion between the major and minor isomers.
6 R. G. Wilkins, Kinetics and Mechanism of Reactions of Transition
Metal Complexes, VCH, Weinheim, 1991, Ch. 4.
7 T. W. Hambley, Inorg. Chem., 1988, 27, 1073.
8 G. K. Anderson and R. Kumar, Inorg. Chem., 1984, 23, 4064.
9 KALEIDAGRAPH, version 3.09, Synergy Software, Reading, PA,
1997.
Conclusions
10 X-SHAPE, Crystal Optimisation for Absorption Correction, Stoë
and Cie, Darmstadt, Germany, 1996.
11 W. Clegg, Acta Crystallogr., Sect. A, 1981, 37, 22.
12 P. T. Beurskens, G. Beurskens, W. P. Bosman, R. de Gelder,
S. Garcìa-Granda, R. O. Gould, R. Israël and J. M. M. Smits,
DIRDIF-96, Crystallography Laboratory, University of Nijmigen,
The Netherlands, 1996.
The aminophosphine complex, cis-[Pt(Me2N(CH2)2PPh2-
1
P,N)2]Cl2, 2, has been shown by H and 31P NMR studies to
undergo reversible chelate ring-opening reactions in aqueous
solution. The rate for this process varies significantly with pH,
chloride ion concentration and temperature and was found
to be governed by an associative mechanism, involving a
J. Chem. Soc., Dalton Trans., 2001, 362–372
371

View Article Online
13 G. M. Sheldrick, SHELXL-97, University of Göttingen, Germany,
1997.
14 D. J. States, R. A. Haberkorn and D. J. Ruben, J. Magn. Reson.,
1982, 48, 286.
15 E. W. Abel, T. P. J. Coston, K. G. Orrell, V. Sîk and D. Stephenson,
J. Magn. Reson., 1986, 70, 34.
16 gNMR, version 3.6, Cherwell Scientific Publishing Limited, Oxford,
1995.
Hursthouse, D. E. Hibbs, K. M. Abdul Malik and N. G. Vassilev,
Organomet. Chem., 1998, 555, 35.
35 P. Fisher and A. Fettig, Magn. Reson. Chem., 1997, 35, 839.
36 A. Gelling, D. R. Noble, K. G. Orrell, A. G. Osborne and V. Sik,
J. Chem. Soc., Dalton Trans., 1996, 3064.
ˇ
37 R. B. Martin, in Cisplatin-Chemistry and Biochemistry of a Leading
Anticancer Drug, ed. B. Lippert, Verlag, Zürich, 1999, pp. 183–
205.
17 J.-J. Delpuech, in Dynamics of Solutions and Fluid Mixtures by
NMR, ed. J.-J. Delpuech, John Wiley & Son, Chichester, 1995,
p. 112.
18 HYPERCHEM, Hypercube Inc., Ontario, Canada, 1995.
19 T. W. Hambley and P. Comba, MOMECSGI, A Strain Energy
Minimisation Package Adapted to HyperChem., University of
Sydney, 1994.
20 T. W. Hambley, Inorg. Chem., 1991, 30, 937; E. C. L. Ling, G. W.
Allen and T. W. Hambley, J. Chem. Soc., Dalton Trans., 1993, 3705;
S. J. Weiner, P. A. Kollman, D. A. Case, U. C. Singh, C. Ghio,
G. Alagona, S. Profeta Jr. and P. Weiner, J. Am. Chem. Soc., 1984,
106, 765; T. W. Hambley, Inorg. Chem., 1998, 37, 3767.
21 T. W. Hambley, J. Comput. Chem., 1988, 8, 651.
22 G. J. Palenik and T. J. Giordano, J. Chem. Soc., Dalton Trans., 1987,
1175.
38 A. Pullman and B. Pullman, Q. Rev. Biophys., 1981, 14, 289; A. L.
Pinto and S. J. Lippard, Biophys. Acta, 1985, 780, 167; I. Saito,
M. Takayama, H. Sugiyama, K. Nakatani, A. Tsuchida and M.
Yamamoto, J. Am. Chem. Soc., 1995, 117, 6406.
39 G. Frommer, I. Mutikainen, F. J. Pesch, E. C. Hillgeris, H. Preut and
B. Lippert, Inorg. Chem., 1992, 31, 2429.
40 K. H. Scheller, V. Scheller-Krattinger and R. B. Martin, J. Am.
Chem. Soc., 1981, 103, 6833; J. H. J. den Hartog, M. L. Salm and
J. Reedijk, Inorg. Chem., 1984, 23, 2001.
41 R. B. Martin, in Platinum, Gold and Other Metal Chemotherapeutic
Agents: Chemistry and Biochemistry, ed. S. J. Lippard, American
Chemical Society, Washington, DC, 1983, ACS Symposium Series
vol. 209, p. 231.
42 G. Raudaschl-Sieber, H. Schollhorn, U. Thewalt and B. Lippert,
J. Am. Chem. Soc., 1985, 107, 3591.
23 A. G. Orpen, L. Brammer, F. H. Allen, O. Kennard, D. J. Watson
and R. Taylor, J. Chem. Soc., Dalton Trans., 1987, S1.
24 A. Habtemarian, B. Watchman, R. Palmer, B. Potter, S. Parsons,
A. Parkin and P. J. Sadler, J. Chem. Soc., Dalton Trans., in press.
25 P. E. Garou, Chem. Rev., 1981, 81, 229.
26 P. S. Pregosin, in Phosphorus-31 NMR Spectroscopy in
Stereochemical Analysis, eds. J. G. Verkade and L. D. Louis, VCH,
New York, 1987, p. 465.
43 B. Lippert, in Progress in Inorganic Chemistry, ed. S. J. Lippard,
Wiley, New York, 1989, vol. 37, pp. 1–97.
44 S. L. Bruhn, J. H. Toney and S. J. Lippard, in Progress in Inorganic
Chemistry, ed. S. J. Lippard, Wiley, New York, 1990, Vol. 38,
pp. 477–516.
45 M. Boudvillain, R. Dalbiès and M. Leng, in Metal Ions in Biological
Systems, eds. A. Sigel and H. Sigel, Marcel Dekker, New York,
1996, vol. 33, pp. 87–103.
27 S. J. Berners-Price, U. Frey, J. D. Ranford and P. J. Sadler, J. Am.
Chem. Soc., 1993, 115, 8649.
28 K. H. Scheller, V. Scheller-Krattinger, R. B. Martin, J. Am. Chem.
Soc., 1981, 103, 6833.
29 F. J. Dijt, G. W. Canters, J. H. J. den Hartog, A. T. M. Marcelis and
J. Reedijk, J. Am. Chem. Soc., 1984, 106, 3644.
30 A. Tongi and L. M. Venanzi, Angew. Chem., Int. Ed. Engl., 1994, 33,
497.
46 D. Yang, S. S. G. E. van Boom, J. Reedijk, J. H. van Boom and A. H.
J. Wang, Biochemistry, 1995, 34, 12912.
47 J. A. Beaty and M. M. Jones, Inorg. Chem., 1992, 31, 2547.
48 M. Mikola, K. D. Kilka, A. Hakala and J. Arpalahti, Inorg. Chem.,
1999, 38, 571.
49 Z. Guo, P. J. Sadler and E. Zang, Chem. Commun., 1997, 27;
S. J. Berners-Price, J. D. Ranford and P. J. Sadler, Inorg. Chem.,
1994, 33, 5842; S. J. Berners-Price, T. A. Frenkiel, J. D. Ranford and
P. J. Sadler, J. Chem. Soc., Dalton Trans., 1992, 2137.
50 E. R. Jamieson and S. J. Lippard, Chem. Rev., 1999, 99, 2467.
51 P. M. Takahara, C. A. Frederick and S. J. Lippard, Nature, 1995,
377, 649; P. M. Takahara, C. A. Frederick and S. J. Lippard, J. Am.
Chem. Soc., 1996, 118, 12309.
31 K. G. Orrell, in Encyclopedia of Nuclear Magnetic Resonance,
eds. D. M. L. Grant and R. K. Harris, Wiley, Chichester, UK, 1996,
vol. 8, pp. 4850–4857.
32 V. Jacques and J. F. Desreux, Inorg. Chem., 1994, 33, 4048.
33 G. Lamm and G. R. Pack, Proc. Natl. Acad. Sci. USA, 1990, 87,
9033.
52 S. O. Ano, F. A. Intini, G. Natile and L. G. Marzilli, Inorg. Chem.,
1999, 38, 2989.
ˇ
34 K. G. Orrell, A. G. Osborne, V. Sik, M. Webba da Silva, M. B.
372
J. Chem. Soc., Dalton Trans., 2001, 362–372