You are on page 1of 9

J. Phys. Chem.

B 2008, 112, 4401-4409 4401

The Hydrolysis Mechanism of the Anticancer Ruthenium Drugs NAMI-A and ICR
Investigated by DFT-PCM Calculations

Attilio V. Vargiu,,,,|, Arturo Robertazzi,,,, Alessandra Magistrato,*,,,


Paolo Ruggerone,,| and Paolo Carloni,,
SISSA (ISAS), Via Beirut 4, I-34014 Trieste, Italy, CNR-INFM Democritos National Simulation Center, Italian
Institute of Technology (IIT), Via Beirut 2-4, I-34014 Trieste, Italy, CNR-INFM SLACS and Dipartimento di
Fisica, UniVersita di Cagliari, S.P. Monserrato-Sestu Km 0.700, I-09042 Monserrato, Italy
ReceiVed: October 17, 2007; In Final Form: December 19, 2007

(ImH)[trans-RuCl4(DMSO-S)(Im)], (Im ) imidazole, DMSO-S ) S-bonded dimethylsulfoxide), NAMI-


A, is the first anticancer ruthenium compound that successfully completed Phase I clinical trials. NAMI-A
shows a remarkable activity against lung metastases of solid tumors, but is not effective in the reduction of
primary cancer. The structurally similar (ImH)[trans-RuCl4(Im)2], ICR (or KP418), and its indazole analog
(KP1019) are promising candidate drugs in the treatment of colorectal cancers, but have no antimetastatic
activity. Despite the pharmacological relevance of these compounds, no rationale has been furnished to explain
their markedly different activity. While the nature of the chemical species responsible for their antimetastatic/
anticancer activity has not been determined, it has been suggested that the difference between reduction
potentials of NAMI-A and ICR may be the key to the different biological responses they induce. In this
work, Density Functional Theory calculations were performed to investigate the hydrolysis of NAMI-A and
ICR in both RuIII and RuII oxidation states, up to the third aquation. In line with experimental findings, our
calculations provide a picture of the hydrolysis of NAMI-A and ICR mainly as a stepwise loss of chloride
ligands. While dissociation of Im is unlikely under neutral conditions, that of DMSO becomes competitive
with the loss of chloride ions as the hydrolysis proceeds. Redox properties of NAMI-A and ICR and of their
most relevant hydrolytic intermediates were also studied in order to monitor the effects of biological reductants
on the mechanism of action. Our findings may contribute to the identification of the active compounds that
interact with biological targets, and to explain the different biological activity of NAMI-A and ICR.

Introduction CHART 1
Ru-based complexes are today becoming one of the most
promising alternatives to platinum-containing anticancer drugs.1
Among this class of compounds, (ImH)[trans-RuCl4(DMSO-
S)(Im)] (Im ) imidazole, DMSO-S ) S-bonded dimethylsul-
foxide), NAMI-A (Chart 1), is a powerful antimetastatic agent
that has already passed phase I clinical trials.2 Despite its
promising pharmacological activity, the nature of NAMI-A
chemical species responsible for the antimetastatic activity3,4
and the mechanism of action1,5 have not been unveiled. In vitro
experiments have shown the drug to readily undergo hydrolysis
at physiological conditions (pH ) 7.4),6-8 suggesting that
NAMI-A in vivo may form a mixture of hydrolyzed species,
whose nature strongly depends on pH. NAMI-A disappears after tion of such polymeric species.10 NAMI-A metabolites have
15-20 min6-8 as a result of both Cl- and DMSO hydrolyses; been shown to bind strongly to several proteins, including
45 min are required to complete the first hydrolysis step, while integrins, transferrins, and human serum albumin.10-15 In
the second occurs within 2 h.7 Under acidic conditions (pH ) contrast to platinum drugs, NAMI-A binds to DNA weakly,16
3-6), Cl- dissociation is much slower, and only a partial DMSO indicating that its antimetastatic activity may be not related to
dissociation occurs.3,4,6,7,9 Hydrolysis eventually leads to the DNA damage.
formation of polymeric oxo- or hydroxo-bridged species,7,8 albeit The compound (ImH)[trans-RuCl4(Im)2], ICR (or KP418),
this phenomenon is less likely in vivo because of bioligands (Chart 1) and its indazole analogue KP1019,5,17-19 albeit
availability, e.g., the presence of albumin precludes the forma- structurally similar to NAMI-A, have promising anticancer
activity against colorectal cancers,15,18,19 but no relevant anti-
* Corresponding author. metastatic effects.17 ICR reacts with DNA following cisplatin-
SISSA (ISAS).
like kinetics, inhibiting its template-primer properties for nucleic
CNR-INFM Democritos National Simulation Center.
Italian Institute of Technology. acid synthesis.20 Experimental data indicate that in vitro ICR
| Universita di Cagliari. undergoes stepwise hydrolysis of chloride ions,18,21 but the
Equally contributed to this work. nature and oxidation state of the active ruthenium metabolites
10.1021/jp710078y CCC: $40.75 2008 American Chemical Society
Published on Web 03/19/2008
4402 J. Phys. Chem. B, Vol. 112, No. 14, 2008 Vargiu et al.

are still unknown. The hydrolysis is supposed to be a crucial (c) Single-point calculations with larger basis sets on opti-
step, as only aged ICR solutions show the drug binding to mized structures. Namely, 6-311+G(3d) on Cl and S atoms,
DNA.22 6-311+G(d,p) on all the other nonmetal atoms, and Lan2DZ(f)
The biological diversity between NAMI-A and ICR may be (f ) 1.235)39 on Ru were used in order to obtain an improved
partially due to their different redox properties: reduction value of the internal energy.32,33
potentials Em (vs normal hydrogen electrode, NHE) are 0.235 (d) Single-point calculations (same setup as in c) on optimized
V and -0.275 V, respectively.23,24 Such a difference is caused structures mimicking hydration effects via an implicit solvation
by the DMSO, which is a moderate -acceptor and stabilizes model (CPCM).30 We are aware that PCM calculations of small
RuII.4,7 Because of its positive Em, NAMI-A may be easily ions such as Cl- may be problematic. However, we noticed
reduced in vivo by biological reductants, even before undergoing very good agreement between experimental and our theoretical
dissociation. Indeed, the reduced form of NAMI-A and/or its solvation free energies of Cl- (not reported).30,40
metabolites has been demonstrated to be of crucial impor- Free energies were then calculated, adding to the internal
tance,3,4,6,7 having a robust antimetastatic activity on tumor energy the zero-point vibrational energy, the entropic term (using
models.7 In contrast, ICR would need stronger reductants, which Wertz correction),41,42 and the solvation free energy. Correction
may be more rare in a biological environment.3,21,25 Consistently, for the standard concentration was also considered.27,28
ICR has proved to be more inert than NAMI-A, and its Activation free energies are usually calculated as the differ-
hydrolysis path is less affected by reductants.26 ence between the free energy of the TS and that of the reaction
Recently, Chen et al. have reported a theoretical investigation intermediate (RI), i.e., the stable complex between the reactant
on NAMI-A and ICR hydrolyses,27,28 focusing on Cl- dissocia- and a water molecule (the same terminology is used also for
tion. Calculated structural properties and kinetic constants are the product intermediate, indicated by PI). However, Lau et al.
in good agreement with the available experimental data. have cast doubts on the physical basis of considering reaction
However, dissociations of DMSO and imidazole have not been intermediates RI and PI as reference states for calculating the
studied, despite the proved biological relevance of DMSO activation and reaction free energies of hydrolysis processes.33
(partial) hydrolysis.6,7 Furthermore, the hydrolysis of the reduced Indeed, they have shown that the free energy of an isolated
compounds has not been considered, even though it may be reactant and an isolated water molecule (named as R and P for
important for both drugs because of the high reducing power the reactant and product, respectively) significantly differs from
of solid tumor tissues.21,29 that of RI (and PI). This discrepancy is believed to arise from
In order to understand the mechanism of action of NAMI-A the failure of implicit solvent models to reproduce the correct
and ICR at the molecular level, an extensive investigation of solvation entropy in bimolecular reactions. To overcome this
the hydrolysis paths in both Ru oxidation states, along with an issue in ref 33, the entropy was corrected using Wertzs
evaluation of the reduction potentials of the two drugs and their approach,41,42 obtaining almost identical free energies for RI
metabolites, is needed. In this work we tackle this issue and and R. Unfortunately, the same procedure did not provide
report a systematic study of the hydrolyses of the two drugs up consistent results when applied to the PIs, whose free energies
to the third aquation, combining density functional theory are systematically higher than those of Ps. In particular, all
(DFT) calculations and an implicit description of the solvent hydrolysis steps turned out to be endothermic (up to 10
through the conductor-like polarizable continuum model kcal/mol) when comparing the free energy of PI to that of RI.
(CPCM).30 Reduction potentials of NAMI-A, ICR, and of their Considering the free energy differences between reactants R
most relevant hydrolyzed species were estimated within the same and products P led to reaction free energies in close agreement
setup, to explore the effects of biological reductants. with experiments.
This study is a first attempt to rationalize the large amount Therefore, we calculated the activation free energies by
of available experimental findings that, in some cases, led to referring each TS to the corresponding RI (which are likely to
discrepant conclusions.4,21 In addition, it confirms the ability be affected to the same extent by the inaccuracy of our
of computational studies to qualitatively predict the hydrolysis computational model), while reaction free energies were cal-
of different substituents from metal-containing compounds. culated by comparing P with R, as in ref 33. We believe that
this protocol is more suitable to correctly calculate TS barriers,
Computational Details and to minimize the uncertainty when comparing different
hydrolysis products (e.g., those resulting from Cl- or DMSO
The hydrolysis of NAMI-A and ICR in both RuIII and RuII dissociation). Notably, this scheme gave results in good agree-
oxidation states was investigated in the framework of the density ment with previous theoretical and experimental results (Tables
functional theory, employing the Gaussian 03 package.31 In order 1 and 2), Vide infra.23,27,28,43 We are thus confident that our setup
to consider all possible hydrolytic paths, the dissociation of each may predict thermodynamic and kinetic data not experimentally
ligand (namely, Cl-, DMSO-S and Im) was studied in the available, and may be useful to rationalize both NAMI-A and
presence of one explicit incoming water. To achieve a trade- ICR hydrolyses.
off between accuracy and computational costs, the following For the sake of clarity, products of different steps were labeled
methodology was adopted:32,33 as PNa,b,c, where a,b,c refers to the ligand exchanged during
(a) Full in vacuo optimization without symmetry constraints hydrolysis, with N indicating the hydrolysis step. Similarly,
of reactants (R), products (P) and transition states (TS) using transition state structures were reported as TSNa, where a is
the B3LYP exchange-correlation functional,34,35 with 6-31G- the leaving group, with N indicating the hydrolysis step.
(d,p) and Lanl2DZ36-38 basis sets on N, S, C, O, H, Cl, and Ru We would like to stress that reactions involving small groups
atoms, respectively. or atoms are routinely characterized, while those in which the
(b) Harmonic frequency calculations at the same level of leaving groups are as large as Im (or DMSO) present many
theory, (i) to confirm that optimized structures were minima or issues. In particular, because of the large number of degrees of
transition states, (ii) to estimate zero-point vibrational energies freedom, the search for their TS may be awkward. We indeed
and thermal and entropic corrections. found accurate TS structures corresponding to Cl- and DMSO
Hydrolysis Mechanism of NAMI-A and ICR J. Phys. Chem. B, Vol. 112, No. 14, 2008 4403

TABLE 1: Selected Bond Lengths (d, ) and Electron Densities (G, a.u.) of NAMI-A and ICR in Both Oxidation Statesa
RuII RuIII
NAMI-A ICR NAMI-A ICR
d F d F d F d F
Ru-Cl 2.55 0.0604 2.58 0.0529 2.43(2.34) 0.0627 2.44(2.36) 0.0609
Ru-S 2.32 0.0576 2.40(2.30) 0.0520
Ru-N 2.10 0.0701 2.10 0.0772 2.10(2.08) 0.0854 2.10(2.08) 0.0761
a
Experimental values (in brackets) of NAMI23 (the analogous sodium salt of NAMI-A) and ICR48 bond lengths are displayed (Ru-Cl is reported
as an average).

TABLE 2: Calculated and Experimental Reduction Molecular properties were calculated using Bader et al.s
Potentials (Em vs NHE-V) of NAMI-A and ICR Metabolites theory of atoms in molecules (AIM),50,51 in order to analyze
NAMI-A ICR the electron density (F) at bond critical points (BCP). Indeed,
calc. expt. calc. expt. the electron density at such critical points correlates well with
the strength of chemical bonds and interactions.52-55 Topologies
R 0.17 0.235a -0.18 -0.275b
PICl 0.34 0.337a,c -0.01
were built using AIMPAC series of programs.56 Charges were
PIDMSO(Im) -0.13 -0.13 calculated with the full natural bond orbital (NBO) analysis.57
PIIcis
2Cl
0.67 0.12 Both NBO and AIM analyses were carried out considering the
trans
PII2Cl 0.57 0.07 solvent effects. Following previous works,58,59 reduction po-
PIICl,DMSO(Im) -0.17 -0.17 tentials (Em) at standard conditions were estimated from the
PIII3Cl 1.02 0.53 following equation:
PIII2Cl,DMSO(Im) 0.09 0.09
a From ref 23. b From ref 4. c Experimental reduction potential refers Em ) G/nF (1)
to mer-RuCl3(DMSO)2(Im); see text for further details.
where F ) 96485 C (the Faraday constant), Em is the reduction
dissociation. Unfortunately, such a satisfactory accuracy was potential, and n is the number of electrons involved in the redox
not achieved for Im dissociation. Because of these difficulties, process. To compare reduction potentials with experimental data,
we considered as a (pseudo) TS for Im dissociation the structure Em values were referred to the NHE. As suggested by previous
corresponding to the highest energy while breaking the benchmark studies,59 to reproduce redox potential is not a trivial
Ru-NIm bond in the presence of one explicit water. This should task, medium-large basis sets, zero-point-energy and entropic
allow a qualitative comparison of the Im dissociation barrier corrections should be employed to provide a semiquantitative
with activation free energies of Cl- and DMSO. Indeed, our agreement with experimental results.
results suggest, in line with experimental findings,6 that Im
dissociation is not likely under neutral conditions. Results and Discussion
A final comment on the reactant concentrations is necessary. Energetic, structural, and electronic properties of the most
It has been shown that cell chloride ion concentration strongly relevant metabolites of NAMI-A and ICR in both RuIII and RuII
affects the hydrolysis of other metal-based drugs, such as oxidation states (hereafter termed as NAMI-AIII/II, ICRIII/II) are
cisplatin.44-46 This may be expected to occur also for NAMI-A reported. As NAMI-A is likely to be reduced in vivo, and
and ICR, but our theoretical approach cannot consider such an NAMI-AII metabolites have shown antimetastatic ef-
important aspect. Nevertheless, Reedijk and co-workers6 have fects,3,4,7,18,19,25,60,61 the chemical behavior of reduced species
shown that Cl- concentration affects both Cl- and DMSO is crucial to understand the biological activity of this drug.
dissociations from NAMI-AIII in a similar manner, suggesting Reduction of ICR in biological environments is less likely, but
that other hydrolysis pathways may also be similarly affected may take place in cancerous tissues. For these reasons, and since
in vivo. theoretical studies on NAMI-AIII and ICRIII have recently been
reported,27,28 the discussion of RuII species is more detailed and
All possible hydrolytic paths47 were characterized employing
precedes that of RuIII ones.
the approach discussed above, yet, for the sake of clarity, we
Hydrolysis Mechanism of NAMI-A. Figure 1 displays the
only reported energetics, along with selected structural and most likely hydrolysis paths of NAMI-A in both ruthenium
electronic properties of the most relevant complexes (for more oxidation states, up to the third hydrolysis step (Figure S2a,b
details, see Supporting Information). Remarkably, structures of and Figure S3, Supporting Information).
NAMI-A and ICR obtained with our computational setup are Dissociation of Im typically requires an activation free energy
very similar to those reported in previous theoretical studies27,28 (G) larger than that of Cl- and DMSO, and products of Im
and in fair agreement with experimental results (Table 1).23,48 hydrolysis have the largest endothermic character. Therefore,
As suggested recently,27,28 discrepancies may be caused by the Im-water exchange is unlikely under neutral conditions in both
intrinsic inaccuracy of the computational approach as well as Ru oxidation states, in agreement with experimental results.6
by crystal packing forces present in the X-ray structures. To This may be partly due to the strength of the coordination bonds,
further test our computational setup, we performed calculations e.g., the Ru-N bond of NAMI-AII is somewhat stronger than
with larger basis sets with and without the inclusion of solvent the Ru-S bond (dRu-N ) 2.10 and FRu-N ) 0.0701 a.u.; dRu-S
during optimization. As shown in the Supporting Information ) 2.32 and FRu-S ) 0.0576 a.u., where F is the density at
(Table S1), geometrical parameters hardly changed. Further- the critical point, Table 1). Thus, the following discussion is
more, the structural and energetic properties of NAMI-AII were limited to Cl- and DMSO hydrolyses.
calculated at the MP2 level49 (Table S2). As shown in the NAMI-AII. Step I. In line with previous theoretical studies27
Supporting Information, a satisfactory agreement was found and experimental findings,4 our data suggest that Cl- dissocia-
between DFT and MP2 results. tion is the most favorable process. The activation free energies
4404 J. Phys. Chem. B, Vol. 112, No. 14, 2008 Vargiu et al.

Figure 1. (a) Proposed reaction paths for the hydrolysis of NAMI-A. Reduction potentials are calculated as reported in the Computational Details
(see Table 2 and text for further details); experimental reduction potentials (vs NHE) are in brackets. (b) Free energy profile (see Computational
Details) associated with the most likely hydrolysis path of NAMI-AII, up to the third substitution. The most favorable transition states and products
are represented by solid lines, while dotted lines are used for unlikely transition state and intermediates. Activation energies (kcal/mol) are displayed
in the squares along the reaction path. Reaction free energies are calculated with respect to R. (c) Free energy profile associated with the most likely
hydrolysis path of NAMI-AIII. Black and red lines correspond to Cl- and DMSO hydrolysis paths.

of DMSO and Cl- dissociations are comparable (Figure S1a), 10.7 kcal/mol). Indeed, the NBO analysis confirmed that after
but thermodynamics largely favors chloride ion hydrolysis (PICl DMSO dissociation, the drug maintains the overall -2 charge,
is virtually isothermic to R, while PIDMSO is endothermic by losing a large group on which the electron density may
Hydrolysis Mechanism of NAMI-A and ICR J. Phys. Chem. B, Vol. 112, No. 14, 2008 4405

delocalize. In contrast, the leaving Cl- has a partial charge of This is also in line with NMR, spectroscopic, and X-ray data,
approximately -0.9 a.u., i.e., the overall negative charge on which indicate that DMSO-S binds more strongly to RuII than
the metal complex is reduced to -1, promoting the reaction. to RuIII.4,62
The larger stability of PICl as compared to PIDMSO may be Irrespective of the path followed, the product of the second
partly caused by an intramolecular O-HO-DMSO-S H- hydrolysis step is PIICl,DMSO: the most likely leaving group from
bond between the coordinated water molecule and the DMSO PICl is DMSO, with a G ) 17.9 kcal/mol, while the release
oxygen (dHO ) 1.96 and F ) 0.0189 a.u.), as well as an of Cl- from PIDMSO occurs with a kinetic barrier of 18.9
intramolecular O-HCl H-bond, (dHCl ) 2.30 and FHCl ) kcal/mol. This scenario is compatible with experimental findings
0.0148 a.u.). We verified that the isomer in which the Ru- of a partial DMSO dissociation occurring simultaneously to Cl-
OH2 bond is rotated by 180 (no H-bonding) is less stable by hydrolysis.6 The loss of a further Cl- from PIICl,DMSO is
6 kcal/mol, i.e., the preference for PICl over PIDMSO should kinetically more favored in cis than in trans, with the entire
also remain in case the intramolecular H-bond to DMSO-S is process being slightly endothermic.
broken. These results suggest that reduction significantly affects the
Thus, both kinetic and thermodynamic features suggest that hydrolysis of NAMI-A: NAMI-AII hydrolysis is overall exo-
chloride ion hydrolysis is the most favored at the first step. thermic with slightly lower activation free energies barriers
Step II. Reaction proceeds faster from step I to II, as the compared to those of NAMI-AIII, this difference increasing as
activation free energy of the second Cl--water exchange the hydrolysis proceeds. Importantly, our data confirm that
decreases by 6-8 kcal/mol. As in the previous step, this neither NAMI-A reduction nor DMSO dissociation can be
dissociation is kinetically competitive with that of DMSO, i.e., neglected a priori, since they may both play a fundamental role
both reactions may in principle occur (Figure S2a). Yet, relative for NAMI-A antimetastatic activity.
product stability favors dissociation of a second Cl-. Indeed, Hydrolysis Mechanism of ICR. In order to identify those
the product of DMSO hydrolysis is endothermic by 10.4 features responsible for the different pharmacological activity
kcal/mol, while PIIcis2Cl and PII2Cl sit respectively at -2.6 and
trans
of NAMI-A and ICR, a broad analysis of the ICR hydrolysis
-3.1 kcal/mol with respect to R. This would rule out the mechanism is proposed. In agreement with experimental stud-
possibility of DMSO dissociation at the second step, in line ies,6 imidazole dissociation is typically unlikely in both oxidation
with in vitro studies at neutral pH.4,9 In addition, our calculations states (Supporting Information).
suggest the formation of a mixture of cis and trans diaqua ICRII. The most likely group to dissociate from ICR is Cl-
isomers, supporting experimental hypotheses.4 (Figure 2b, Figure S4) with G ) 23.2 kcal/mol, PICl being
Step III. The analysis of the third hydrolysis reveals intriguing virtually isothermic to R. Interestingly, the activation free energy
features. In particular, DMSO dissociation becomes kinetically is similar to that of 24.8 kcal/mol estimated for TSICl of NAMI-
unlikely with respect to that of Cl-, yet the free energy AII, suggesting that the kinetics of the first Cl- dissociation are
difference between PIII3Cl and PIII2Cl,DMSO reduces as compared virtually unaffected by the presence of DMSO-S or Im ligands.
to the first and second steps. A simple electrostatic analysis At the second hydrolysis step, PIItrans cis
2Cl , PII2Cl may originate
may provide a rationale for this behavior. During the first two with the following energetics: G ) 21.6 and G ) 0.7

hydrolysis steps, the metal complex bears an overall negative kcal/mol, G ) 16.1 and G ) -1.8 kcal/mol. Importantly,
charge of -2 and -1, respectively, i.e., dissociation of the our calculations strongly suggest that the formation of cis diaqua
negatively charged chloride is favored with respect to the neutral complexes is kinetically preferred to that of trans isomers. This
DMSO-S ligand. After the second Cl- dissociation, PII2Cl may be due to an intricate interplay of several factors, including
is neutral, with DMSO dissociation becoming more favor- trans influence and H-bonding. First, as consequence of the trans
able. influence, Ru-Cl in trans to water is the strongest in PICl, and
Thus, the hydrolysis of NAMI-AII mainly consists of a trans
Cltrans therefore is the least likely to leave (dRu-Cl ) 2.46
stepwise Cl--water exchange: all Cl- dissociations are exo- and FRu-Cl ) 0.0666 a.u., as compared to dRu-Cl ) 2.56 and
trans cis
thermic with decreasing activation free energy barriers as the FRu-Cl
cis
) 0.0560 a.u., averaged over the other two Ru-Cl
reaction proceeds. Findings of the first, second, and third bonds). Because of a strong H-bonding (Table 3), both entering
hydrolysis steps are in qualitative agreement with experiments water and leaving Cl- interact more strongly with the metal
by Alessio and co-workers, who extensively characterized the
center compared to TSIItrans Cl . These effects may provide the
hydrolytic behavior of NAMI-AII at pH ) 7.4.24
extra stabilization of the transition state that leads to the cis
NAMI-AIII Vs NAMI-AII Hydrolysis. At the first hydrolysis diaqua isomer. The kinetic preference for PIIcis 2Cl with respect to
step, the activation free energy of Cl- dissociation is larger (by
PIItrans is remarkable, as may be related to the different
6 kcal/mol) than that of DMSO (Figure 1). Notably, the 2Cl
pharmacological activities of NAMI-A and ICR. Yet, calcula-
calculated barrier for Cl- dissociation (G ) 24.5 kcal/mol)
tions that include dynamical properties of explicit solvent
is in good agreement with experimental measurements (G )
molecules are needed to confirm this point beyond doubt.
22.8 kcal/mol) and recent, more accurate and computationally
Nevertheless, we would like to remark that strong intramolecular
expensive, theoretical predictions (G ) 23.2 kcal/mol).23,27
H-bonds should be less affected by the presence of explicit water
Since PICl and PIDMSO are virtually isothermic, DMSO substitu-
molecules.53,63 In agreement with experimental findings,22
tion, thermodynamically hindered for NAMI-AII, is very likely
thermodynamics of cis and trans diaqua complexes are virtual-
to take place during NAMI-AIII hydrolysis. Indeed, geometrical
ly equivalent, i.e., on a long time scale both complexes are
properties and electron density analysis confirm that the Ru-S
formed.
bond is stronger in NAMI-AII than in NAMI-AIII, i.e., DMSO-S
is a moderate -acceptor, and the component of the RuII-S Similarly to NAMI-AII hydrolysis, the third Cl- substitution
bond is expected to be stronger when the complex has a negative is exothermic and most likely to take place from TSIItrans Cl .
charge (see Table 1).7 Consistent with these results, experiments ICR Vs ICR Hydrolysis. As suggested by the experiments
III II

by Sava et al.7 showed that a slow DMSO hydrolysis occurs. of Mestroni et al.,26 a change in the Ru oxidation state affects
4406 J. Phys. Chem. B, Vol. 112, No. 14, 2008 Vargiu et al.

Figure 2. (a) Proposed reaction paths for the hydrolysis of ICR. Reduction potentials are calculated as reported in the Computational Details (see
Table 2 and text for further details); experimental reduction potentials (vs NHE) are in parenthesis. (b) Free energy profile (see Computational
Details) associated with the most likely hydrolysis path of ICRII, up to the third substitution. The most favorable transition states and products are
represented by solid lines, while dotted lines are used for unlikely transition states and reaction intermediates. Activation energies (kcal/mol) are
displayed in the squares along the reaction path. Free energies are calculated with respect to R. (c) Free energy profile associated with the most
likely hydrolysis path of ICRIII. Only cis diaqua complexes are shown to illustrate the ICR kinetic preference for these isomers.

ICR hydrolysis to a minor extent (Figure 2c, Figure S4). Indeed, manding calculations (24.0 kcal/mol),28 and to that measured
the ICRIII hydrolysis consists of a stepwise Cl--water exchange experimentally (24.1 kcal/mol).43
with Im dissociation being unlikely at every step. The activation
free energy relative to the first hydrolysis is virtually insensitive As for the first step, Ru oxidation state hardly affects the
to the Ru oxidation state, but the process is thermodynamically second hydrolysis. Indeed, chloride ion dissociation is the most
more favored. Calculated activation free energy (G ) 22.7 likely event, kinetically preferred (by 4 kcal/mol) in cis rather
kcal/mol) is in good agreement with that recently found by a than in trans. The activation free energies, G ) 20.8 and
theoretical investigation based on more computationally de- 24.8 kcal/mol for TSIItrans
Cl and TSIItrans
Cl , respectively, are in
Hydrolysis Mechanism of NAMI-A and ICR J. Phys. Chem. B, Vol. 112, No. 14, 2008 4407

TABLE 3: Selected Geometrical () and Electron Density (in Brackets, a.u.) Properties of TS Leading to PIIcis
2Cl and PI
Itrans
2Cl Products of ICR Hydrolysis
Oa-HClb Ru-O-HClb RuClb RuaOH2
TSIIIcis
Cl
2.12 (0.0300) 1.95 (0.0437) 3.20 (0.0173) 2.84 (0.0189)

TSIIItrans
Cl
2.05 (0.0347) 3.28 (0.0135) 2.91 (0.0144)

a Entering water molecule. b Leaving Cl ion.

good agreement with the values of 20.0 and 25.4 kcal/mol metabolites, confirming the crucial role played by DMSO for
reported recently.28 In contrast, the thermodynamics of cis and NAMI-A pharmacological activity.4,7
trans diaqua complexes are similar, suggesting that both Reduction of ICR in vivo is unlikely, as the reduction
products may be found in solution.22 The third hydrolysis step potential is negative, yet this may occur in specific regions where
has a similar barrier to that of previous aquations, but, unlike strong biological reductants are available, such as proliferating
ICRII, the reaction is endothermic. cells.21 Alternatively, as the hydrolysis proceeds, Em increases,
Estimation of Reduction Potentials. As the electrochemical and the products of the second aquation have a positive reduction
behavior of NAMI-A and ICR is crucial for their biological potential. Thus, reduction may occur for PIIcis 2Cl, which has a
activity,3,7,21 reduction potentials of these drugs as well as those reduction potential similar to that of NAMI-A. These results
of their most relevant hydrolysis products were estimated (Table highlight the role of reduction in driving the hydrolysis of
2). Calculated and experimentally measured reduction potentials NAMI-A and ICR toward different chemical routes.
of NAMI-A and ICR are in good agreement, as they differ by ICR vs NAMI-A: Biological Implications. The analysis of
0.08 V on average (Table 2). Thus, the estimated difference activation and reaction free energies combined with the estima-
between reduction potentials of NAMI-A and ICR was found tion of reduction potentials reveals intriguing features of
to be close to the experimental value, Em 350 mV vs NAMI-A and ICR hydrolyses (Figures 1a and 2a).
m 500 mV. Unfortunately, data about redox potentials of
Eexp Under neutral conditions, NAMI-AIII hydrolysis leads to the
their hydrolysis products are not available in literature. However formation of compounds with one DMSO and one Cl-
electrochemical measurements on mer-RuCl3(DMSO)2(Im), exchanged by water molecules. In the presence of biological
which contains a Ru-ODMSO bond similar to the Ru-Owater of reductants, reduction does not take place for the products of
the first hydrolysis product of NAMI-A, are available.23 Indeed, DMSO dissociation. On the other hand, NAMI-A and PICl have
the experimental reduction potential (Em ) 0.337 V vs NHE), positive reduction potentials and undergo reduction more easily.
compares very well with our calculated value (0.34 V). These Therefore, the most abundant NAMI-A metabolites are expected
data confirm the reliability of our computational setup, allowing to be the RuII-diaqua and triaqua complexes bearing the
us to predict reduction potentials for NAMI-A and ICR DMSO-S ligand. However, we cannot exclude that monoaqua
metabolites possibly with a similar accuracy. complexes may play a role in its biological activity. Moreover,
As the hydrolysis of Cl- proceeds to PI, PII, and PIII, in vitro experiments have shown that poly-oxo species are
reduction potentials increase to 0.3, 0.7, and 1.0 V and to rapidly formed already during the second hydrolysis, whereas,
0.0, 0.1, and 0.5 V for products of NAMI-A and ICR, in vivo, the interaction with N- and S- biological donors would
respectively (Figure 3). This is in qualitative agreement with prevent polymerization.4 Either way, products of the third
experimental findings showing that, for each Cl--water hydrolysis should be less relevant for the biological activity of
exchange, a constant increase of the reduction potential oc- NAMI-A.4 Thus, our data stress the relevance of the oxidation
curs.25,64 In contrast, reduction potential decreases upon dis- state for the hydrolysis path, and support the hypothesis that
sociation of DMSO (or Im for ICR), to -0.13 V and -0.17 V the reduced form of NAMI-A or its hydrolytic products are
for PIDMSO(Im) and PIICl,DMSO(Im) (Table 2). This suggests that relevant for the antimetastatic activity.4,7,13,25 Furthermore, from
DMSO dissociation may preclude NAMI-A reduction, slowing the analysis of reduction potentials clearly emerged the crucial
down the hydrolysis and possibly the formation of the active role of DMSO as it stabilizes RuII species, promoting NAMI-A
reduction.
Reduction of ICR is less likely than that of NAMI-A, unless
strong biological reductants are available.21 However, as the
hydrolysis proceeds, the reduction potential of metabolites
increases, i.e., the products of the second hydrolysis have
positive reduction potentials. Thus, reduction may require a
relatively long time, perhaps long enough for ICRIII to undergo
hydrolysis, with the most likely products being a mixture of
mono- and diaqua complexes (Figure 2a). Once these species
Figure 3. Estimated reduction potentials Em (vs NHE, V) of NAMI- are formed, reduction may easily occur, particularly for the cis
A, ICR, and the products of Cl- dissociation. Experimental reduction isomer. If reduction occurs, the most abundant metabolites are
potentials are from ref 23 and ref 4. See text for further details. PIIcis
2Cl and PIII3Cl, holding for the latter similar considerations
4408 J. Phys. Chem. B, Vol. 112, No. 14, 2008 Vargiu et al.

as for NAMI-A. All these findings are compatible with cofunded by the Italian Ministry of University and Research
experimental results, suggesting that ICR binding to biological (MIUR) within the Programma Operativo Nazionale 2000-2006
targets occurs in hours and that only aged ICR solutions can Ricerca Scientifica, Sviluppo Tecnologico, Alta Formazione
bind to DNA.22 per le Regioni Italiane dellObiettivo 1 (Campania, Calabria,
In addition, the analysis of the second hydrolysis kinetics Puglia, Basilicata, Sicilia, Sardegna) Asse II, Misura II.2 Societa
reveals important differences between ICR and NAMI-A. In dellInformazione, Azione a Sistemi di calcolo e simulazione
particular, ICR (in both RuIII and RuII oxidation states) shows ad alte prestazioni. More information is available at http://
a strong kinetic preference for cis diaqua complexes, while both www.cybersar.it.
NAMI-A isomers have similar kinetics and thermodynamics.
In vivo, such a kinetic preference may affect the way of binding Supporting Information Available: Test calculations of
and also lead the drugs to different biological targets: DNA NAMI-A and ICR structures (Tables S1-S2); test calculations
for ICR, and proteins or receptors for NAMI-A (see for example of the Id/Ia hydrolysis mechanisms (Figure S1); details of
ref 65 and references therein). NAMI-A and ICR hydrolytic paths (Figures S3-S5); Cartesian
coordinates of selected metabolites. This material is available
Conclusions free of charge via the Internet at http://pubs.acs.org.
Extensive DFT calculations, with estimation of solvation and References and Notes
entropic effects, were performed to investigate the hydrolysis
(1) Sava, G.; Zorzet, S.; Turrin, C.; Vita, F.; Soranzo, M.; Zabucchi,
mechanisms of two promising anticancer drugs: NAMI-A and G.; Cocchietto, M.; Bergamo, A.; DiGiovine, S.; Pezzoni, G.; Sartor, L.;
ICR. Garbisa, S. Clin. Cancer Res. 2003, 9, 1898.
The computational set up has been validated by comparing (2) Rademaker-Lakhai, J. M.; van den Bongard, D.; Pluim, D.; Beijnen,
J. H.; Schellens, J. H. M. Clin. Cancer Res. 2004, 10, 3717.
available experimental and theoretical data with our findings. (3) Alessio, E.; Mestroni, G.; Bergamo, A.; Sava, G. Curr. Top. Med.
Predicted and measured reduction potentials of NAMI-A and Chem. 2004, 4, 1525.
ICR differ by 0.08 V on average, while calculated activation (4) Alessio, E.; Mestroni, G.; Bergamo, A.; Sava, G. Met. Ions Biol.
free energies for the first hydrolysis are within 2 kcal/mol from Syst. 2004, 42, 323.
(5) Dyson, P. J.; Sava, G. Dalton Trans. 2006, 1929.
those determined experimentally. A broad qualitative agreement (6) Bacac, M.; Hotze, A. C. G.; Van Der Schilden, K.; Haasnoot, J.
with experimental findings of NAMI-A and ICR hydrolysis at G.; Pacor, S.; Alessio, E.; Sava, G.; Reedijk, J. J. Inorg. Biochem. 2004,
physiological conditions was achieved. 98, 402.
(7) Sava, G.; Bergamo, A.; Zorzet, S.; Gava, B.; Casarsa, C.;
Our calculations confirmed that NAMI-A is easily reduced Cocchietto, M.; Furlani, A.; Scarcia, V.; Serli, B.; Iengo, E.; Alessio, E.;
in vivo and that reduction strongly affects the hydrolysis. We Mestroni, G. Eur. J. Cancer 2002, 38, 427.
suggest cis and trans RuII-diaqua metabolites to be the most (8) Groessl, M.; Reisner, E.; Hartinger, C. G.; Eichinger, R.; Semenova,
abundant and perhaps the most relevant for the biological O.; Timerbaev, A. R.; Jakupec, M. A.; Arion, V. B.; Keppler, B. K. J.
Med. Chem. 2007, 50, 2185.
activity of NAMI-A. However, we cannot exclude that monoaqua (9) Bouma, M.; Nuijen, B.; Jansen, M. T.; Sava, G.; Flaibani, A.; Bult,
complexes may also be involved. A.; Beijnen, J. H. Int. J. Pharm. 2002, 248, 239.
Reduction of ICR is less likely than that of NAMI-A, yet, if (10) Messori, L.; Orioli, P.; Vullo, D.; Alessio, E.; Iengo, E. Eur. J.
Biochem. 2000, 267, 1206.
occurring, would hardly affect the hydrolytic path. The most (11) Bergamo, A.; Messori, L.; Piccioli, F.; Cocchietto, M.; Sava, G.
abundant metabolites are the mono- and diaqua complexes, with InVest. New Drugs 2003, V21, 401.
a kinetic preference for the cis isomer in both oxidation states. (12) Frausin, F.; Scarcia, V.; Cocchietto, M.; Furlani, A.; Serli, B.;
Alessio, E.; Sava, G. J. Pharmacol. Exp. Ther. 2005, 313, 227.
Reduction may occur for the products of the second hydrolysis (13) Ravera, M.; Baracco, S.; Cassino, C.; Colangelo, D.; Bagni, G.;
step since reduction potentials increase as the hydrolysis Sava, G.; Osella, D. J. Inorg. Biochem. 2004, 98, 984.
proceeds. (14) Sava, G.; Frausin, F.; Cocchietto, M.; Vita, F.; Podda, E.; Spessotto,
The most relevant NAMI-A and ICR metabolites would retain P.; Furlani, A.; Scarcia, V.; Zabucchi, G. Eur. J. Cancer 2004, 40, 1383.
(15) Kapitza, S.; Pongratz, M.; Jakupec, M. A.; Heffeter, P.; Berger,
the main structural and chemical differences of parent com- W.; Lackinger, L.; Keppler, B. K.; Marian, B. J. Cancer Res. Clin. Oncol.
pounds: the DMSO-S ligand for the former, and the Im ligand 2005, 131, 101.
for the latter. In addition, the kinetic preference for the cis diaqua (16) Brabec, V.; Novakova, O. Drug Resist. Updates 2006, 9, 111.
(17) Zorzet, S.; Bergamo, A.; Cocchietto, M.; Sorc, A.; Gava, B.;
ICR compounds may affect the binding to biomolecules and Alessio, E.; Iengo, E.; Sava, G. J. Pharmacol. Exp. Ther. 2000, 295, 927.
lead to a different biological behavior. Such features may be of (18) Dhudhghaill, O. M.; Hagen, W. R.; Keppler, B. K.; Lipponer, K.;
primary importance in determining the diverse pharmacological Sadler, P. J. Dalton Trans. 1994, 3305.
activity of NAMI-A and ICR anticancer drugs. (19) Andersons, C.; Beauchamp, A. L. Can. J. Chem. 1995, 73, 471.
(20) Holler, E.; Schaller, W.; Keppler, B. Arzneimittelforschung 1991,
41, 1065.
Acknowledgment. The authors would like to thank Prof. (21) Ravera, M.; Cassino, C.; Baracco, S.; Osella, D. Eur. J. Inorg.
G. Sava, Prof. E. Alessio, Dr. A. Bergamo, and V. Leone Chem. 2006, 740.
(22) Kung, A.; Pieper, T.; Wissiack, R.; Rosenberg, E.; Keppler, B. K.
Alvarez for their fruitful help during the preparation of the J. Biol. Inorg. Chem. 2001, 6, 292.
manuscript. This work was carried out in the frame of the project (23) Alessio, E.; Balducci, G.; Lutman, A.; Mestroni, G.; Calligaris, M.;
Nuove terapie e farmaci antitumorali: antimetastatici inor- Attia, W. M. Inorg. Chim. Acta 1993, 203, 205.
ganici, terapia genica e nanobiotecnologie (Art. 11 L.R. 11/ (24) Sava, G.; Alessio, E.; Bergamo, A.; Mestroni, G. Topics in
Biological Inorganic Chemistry: Metallo-pharmaceuticals; Springer: Berlin,
2003, Art. 7 del Regolamento per la concessione di contributi 1999; Vol. 1.
per la realizzazione di progetti di ricerca scientifica e applicata (25) Ravera, M.; Baracco, S.; Cassino, C.; Zanello, P.; Osella, D. Dalton
e di iniziative di trasferimento e di diffusione dei risultati della Trans. 2004, 2347.
(26) Mestroni, G.; Alessio, E.; Sava, G.; Pacor, S.; Coluccia, M.;
ricerca). Computational resources were granted by CINECA Boccarelli, A. Met.-Based Drugs 1994, 1, 41.
(INFM grant) and CASPUR (SLACS collaboration). This (27) Chen, J.; Chen, L.; Liao, S.; Zheng, K.; Ji, L. J. Phys. Chem. B
project represents a scientific collaboration between the Trieste 2007, 111, 7862.
and Cagliari units of the CNR-INFM Democritos Modeling (28) Chen, J.; Chen, L.; Liao, S.; Zheng, K.; Ji, L. Dalton Trans. 2007,
3507.
Center. This work makes use of results produced by the Cybersar (29) Schafer, F. Q.; Buettner, G. R. Free Radical Biol. Med. 2001, 30,
Project managed by the Consorzio COSMOLAB, a project 1191.
Hydrolysis Mechanism of NAMI-A and ICR J. Phys. Chem. B, Vol. 112, No. 14, 2008 4409

(30) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 1995. (45) Jamieson, E. R.; Lippard, S. J. Chem. ReV. 1999, 99, 2467.
(31) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, (46) Kozelka, J.; Legendre, F.; Reeder, F.; Chottard, J. C. Coord. Chem.
M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. ReV. 1999, 192, 61.
N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; (47) For example, about 15 geometry optimizations were performed only
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; for the first hydrolysis step of NAMI-AIII, or more than 400 calculations
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; for describing the entire hydrolysis paths for NAMI-A and ICR in both
Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, oxidation states.
X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; (48) Keppler, B. K.; Rupp, W.; Juhl, U. M.; Endres, H.; Niebl, R.; Balzer,
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; W. Inorg. Chem. 1987, 26, 4366.
Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; (49) Moller, C.; Plesset, M. S. Phys. ReV. 1934, 46, 618.
Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, (50) Bader, R. F. W. Chem. ReV. 1991, 91, 893.
A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; (51) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford
Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; University Press: Oxford, 1990.
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, (52) Howard, S. T.; Lamarche, O. J. Phys. Org. Chem. 2003, 16, 133.
P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; (53) Rybarczyk-Pirek, A.; Dubis, A. T.; Grabowski, S. J.; Nawrot-
Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, Modranka, J. Chem. Phys. 2006, 320, 247.
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03; (54) Robertazzi, A.; Platts, J. A. Inorg. Chem. 2005, 44, 267.
Gaussian Inc.: Pittsburgh, PA, 2003. (55) Waller, M. P.; Robertazzi, A.; Platts, J. A.; Hibbs, D. E.; Williams,
(32) Deubel, D. V.; Lau, J. K.-C. Chem. Commun. 2006, 2451. P. A. J. Comput. Chem. 2006, 27, 491.
(33) Lau, J. K. C.; Deubel, D. V. J. Chem. Theory Comput. 2006, 2, (56) Friedrich Biegler-Konig, J. S. J. Comput. Chem. 2002, 23, 1489.
103. (57) Carpenter, J. E.; Weinhold, F. J. Mol. Struct. (THEOCHEM) 1988,
(34) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785. 169, 41.
(35) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (58) Blasco, S.; Demachy, I.; Jean, Y.; Lledos, A. New. J. Chem. 2001,
(36) Hay, P. J.; Wadt, W. R. J. Chem. Phys.1985, 82, 270. 25, 611.
(37) Wadt, W. R.; Hay, P. J. J. Chem. Phys. 1985, 82, 284. (59) Baik, M. H.; Friesner, R. A. J. Phys. Chem. A 2002, 106, 7407.
(38) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299. (60) Clarke, M. J. Met. Ions Biol. Syst. 1980, 231-283.
(39) Ehlers, A. W.; Bohme, M.; Dapprich, S.; Gobbi, A.; Hollwarth, (61) Mestroni, G.; Alessio, E.; Sava, G.; Pacor, S.; Coluccia, M. Metal
A.; Jonas, V.; Kohler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G. Complexes in Cancer Chemotherapy; VCH Verlagsgesellschaft: Weinheim,
Chem. Phys. Lett. 1993, 208, 111. Germany, 1993.
(40) Barone, V.; Cossi, M. J. Chem. Phys. 1997, 100, 16385. (62) Alessio, E. Chem. ReV. 2004, 104, 4203.
(41) Abraham, M. H. J. Am. Chem. Soc. 1981, 103, 6742. (63) Exarchou, V.; Troganis, A.; Gerothanassis, I. P.; Tsimidou, M.;
(42) Wertz, D. H. J. Am. Chem. Soc. 1980, 102, 5316. Boskou, D. Tetrahedron 2002, 58, 7423.
(43) Chatlas, J.; van Eldik, R.; Keppler, B. K. Inorg. Chim. Acta 1995, (64) Costa, G.; Balducci, G.; Alessio, E.; Tavagnacco, C.; Mestroni, G.
233, 59. J. Electroanal. Chem. 1990, 296, 57.
(44) Burda, J. V.; Zeizinger, M.; Leszczynski, J. J. Comput. Chem. 2005, (65) Bergamo, A.; Stocco, G.; Casarsa, C.; Cocchietto, M.; Alessio, E.;
26, 907. Serli, B.; Zorzet, S.; Sava, G. Int. J. Oncol. 2004, 24, 373.