Aminoferrocene-based prodrugs activated by reactive oxygen species

Article abstract of DOI:10.1021/jm2014937

Cancer cells generally generate higher amounts of reactive oxygen species than normal cells. On the basis of this difference, prodrugs have been developed (e.g., hydroxyferrocifen), which remain inactive in normal cells, but become activated in cancer cel

Full text of DOI:10.1021/jm2014937

Article
pubs.acs.org/jmc
Aminoferrocene-Based Prodrugs Activated by Reactive Oxygen
Species
Helen Hagen,†,§ Paul Marzenell,†,§ Elmar Jentzsch,† Frederik Wenz,‡ Marlon R. Veldwijk,‡
and Andriy Mokhir*,†
†Institute of Inorganic Chemistry, Ruprecht-Karls-University of Heidelberg, Im Neuenheimer Feld 270, 69120 Heidelberg, Germany
‡Department of Radiation Oncology, University Medicine Mannheim, Theodor-Kutzer-Ufer 1-3, 68167 Mannheim, Germany
S
* Supporting Information
ABSTRACT: Cancer cells generally generate higher amounts
of reactive oxygen species than normal cells. On the basis of
this difference, prodrugs have been developed (e.g., hydrox-
yferrocifen), which remain inactive in normal cells, but become
activated in cancer cells. In this work we describe novel
aminoferrocene-based prodrugs, which, in contrast to hydrox-
yferrocifen, after activation form not only quinone methides
(QMs), but also catalysts (iron or ferrocenium ions). The
released products act in a concerted fashion. In particular,
QMs alkylate glutathione, thereby inhibiting the antioxidative system of the cell, whereas the iron species induce catalytic
generation of hydroxyl radicals. Since the catalysts are formed as products of the activation reaction, it proceeds autocatalytically.
The most potent prodrug described here is toxic toward cancer cells (human promyelocytic leukemia (HL-60), IC50 = 9 μM, and
human glioblastoma-astrocytoma (U373), IC50 = 25 μM), but not toxic (up to 100 μM) toward representative nonmalignant
cells (fibroblasts).
compounds are arsenic trioxide,4 buthionine sulfoximine,5
INTRODUCTION
Cancer is a group of diseases which is caused by abnormalities
in the genetic material of the transformed cells. One of the
most successful methods of cancer treatment is chemotherapy
■
procarbazine,5b β-phenylethyl isothiocyanate,6 NO-ASA,7 mo-
texafin gadolinium,5b,8 ferrocenium ion containing salts,9 and
[(η6-arene)Ru(azpy)I]+ complexes.10 Unfortunately, these ROS
regulating drugs not only kill cancer cells, but also increase the
with cytotoxic drugs that is often used in combination with
surgery and radiotherapy. Cisplatin, oxaliplatin,1 and 5-
ROS amount in normal cells. Though the increased ROS does
fluorouracil2 are representative examples of practically
not kill normal cells, it can stimulate their transformation, thus
potentially inducing secondary tumors.11
important anticancer drugs. Though these agents target rapidly
Prodrugs, which are converted to toxic species at cancer-
dividing cells, their cell specificity is usually low. For example,
specific conditions (e.g., high ROS), potentially lack this
healthy tissues with a quick replacement rate (e.g., intestinal
dangerous side effect. For example, Jaouen and co-workers have
lining) and rapidly dividing normal cells (e.g., cells of the
developed an anticancer drug, hydroxyferrocifen, a ferrocenyl
hematopoietic system) are especially strongly affected.
analogue of 4-hydroxytamoxifen (Scheme 1).12 This compound
The tumor microenvironment is different from that of
is converted in cells into toxic quinone methide (QM),7,13
normal tissues. For example, most cancer cells both in isolated
which alkylates glutathione (GSH), thus inhibiting the
form and in tissue exhibit enhanced reactive oxygen species
(ROS) production.3 ROS include 1O2, O2−, HO•, and H2O2. As
antioxidative system of cells. The ferrocene fragment in this
prodrug facilitates formation of the quinone methide rather
a consequence, they function at higher concentrations of ROS.
than acts itself as a catalyst of ROS generation. In particular,
For example, the maximal intracellular concentration of H2O2
([H2O2]in) in Jurkat T-cells was determined to be 7 μM.3d
Some cancer cells are known to resist 0.1−10 mM extracellular
Salganik and co-workers have observed that ferrocifen
analogues, which are unable to form quinone methides
H2O2 ([H2O2]out).3e,f Since [H2O2]in was found to be 7−10-
(tamoxifen−ferrocene, Scheme 1), do not exhibit high
cytotoxicity.14 Examples are known where the ferrocene-
containing drugs act by alternative mechanisms, neither ROS
nor p-quinone methide related. In particular, Jaouen and co-
workers have prepared ferrocenes carrying two aminoalkyl
chains.15 The authors suggested that antiproliferative properties
fold below [H2O2]out,3d one may estimate that [H2O2]in in
some cancer cells may reach 10−100 μM. In contrast, [H2O2]in
in normal cells varies between 0.001 and 0.7 μM, while
[H2O2]in = 1 μM is already toxic.3g Since they function with a
heightened basal level of ROS, cancer cells are more vulnerable
to oxidative stress than healthy cells.3 Therefore, exogenous
compounds inducing production of ROS or other radicals can
potentially be used as anticancer drugs. Examples of such
Received: November 7, 2011
Published: December 19, 2011
© 2011 American Chemical Society
924
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
Scheme 1. Structures of Hydroxyferrocifen and Its Active
Metabolite (Quinone Methide)12 and a Conjugate of
Tamoxifen with Ferrocene14
RESULTS AND DISCUSSION
■
Concept. The structure of prodrug 1 is shown in Scheme 2.
It is a derivative of aminoferrocene, which is linked to the
Scheme 2. Structure of Prodrug 1 and Its Activation in the
Presence of Hydrogen Peroxide with Formation of Cytotoxic
p-Quinone Methide 5, Ferrocenium Ions 7, and Iron Ions
of these compounds on hormone-independent MDA-MB-231
cells can be explained by their ability to bind Zn2+. Here, the
ferrocene fragment seems to play a purely structural role.
Herein we report on novel ferrocenes, which, as well as
hydroxyferrocifen, are activated in cancer cells by oxidation
with generation of quinone methide species. In addition to that,
an efficient catalyst for ROS production is formed during the
activation process (Figure 1). We demonstrated that these
Figure 1. Concept of an anticancer prodrug activated by H2O2: blue
half-circles, ligands binding iron; gray circle, a H2O2-sensitive
protecting group. The prodrug is converted in cancer cells (high
ROS) into I (aminoferrocene) and II (p-quinone methide).
Compound II alkylates GSH, and I is converted to the toxic
ferrocenium derivative (I+), which can be degraded to free iron ions.
Both I+ and iron ions generate ROS catalytically.
organometallic complexes exhibit anticancer activity in cellular
assays and target cancer cells selectively over normal cells. In
particular, we tested representative cancer cell lines (non-
adherent, human promyelocytic leukemia (HL-60), and
adherent, human glioblastoma-astrocytoma (U373)) and non-
malignant cells (fibroblasts). To the best of our knowledge,
they are the first prodrugs, which are able to simultaneously
inhibit the antioxidative cellular system and to induce catalytic
generation of ROS.
pinacol ester of 4-(hydroxymethyl)phenylboronic acid via a
carbamate linker. We envisioned that this prodrug would be
activated at cancer-specific conditions (high ROS, e.g., H2O2)
in accordance with the mechanism outlined in Scheme 2. In
particular, the B−C bond is first cleaved in the presence of
hydrogen peroxide with formation of phenol 2. The latter
compound exists in aqueous solution in equilibrium with
phenolate 3 (pKa ≈ 9), which undergoes spontaneous 1,6-
elimination15 with formation of compound 4 and p-quinone
methide 5. Compound 4 is unstable. In aqueous solution it is
cleaved, forming CO2 and aminoferrocene 6. This sequence of
the cleavage reactions shifts the phenol → phenolate
equilibrium to the right until all phenol is used up. Both
compounds 5 and 6 are expected to be cytotoxic and act in a
concerted fashion. In particular, aminoferrocene 6 is first
oxidized by H2O2 or O2, forming aminoferrocenium cation 7
and either hydroxyl radicals (HO•) or superoxide anion radical
(O2−). 7 is further decomposed in water with formation of iron
ions. The latter ions can also catalyze generation of toxic ROS
from less reactive H2O2 or O2.19 Furthermore, p-quinone
methide 5 reacts with ROS scavengers such as GSH with
formation of alkylated products 8,13 which are not efficient
ROS scavengers. It has been previously demonstrated that p-
quinone methide releasing compounds can affect cancer cells.
For example, Wijtmans and co-workers have described NO-
ASA drug, which is activated in the presence of esterases with
Cytotoxic Ru(II) complexes [(η6-arene)Ru(azpy)I]+ re-
ported by the group of Sadler also act by the dual mechanism.10
In particular, they first oxidize GSH to GSSG. The oxidant in
this case is the 2-(phenylazo)pyridine ligand coordinated to Ru.
Then the recovery of the initial complexes in the presence of
O2 occurs by generation of H2O2. In contrast to our metal
complexes, these Ru(II) complexes are not prodrugs. There-
fore, they can also potentially affect normal cells. Moreover, the
oxidized product (GSSG) can be recovered by intracellular
reductases, which should diminish the effect of the drug.
Among other prodrugs activated at disease-specific con-
ditions, pro-antioxidants have also been described. Examples
include metal binding ligands and matrix metalloprotease
(MMP) inhibitors, which are triggered in the presence of
H2O2,16 specific enzymes,17 and protons.18
925
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
generation of p-quinone methide.7 NO-ASA was found to be
toxic toward SW480 cells (human colon adenocarcinoma, IC50
= 85 μM) and HT29 cells (human colon adenocarcinoma, IC50
= 3−7 μM).20 In contrast to a monofunctional drug such as
NO-ASA, prodrug 1 not only inhibits the antioxidant system of
the cancer cell by generating p-quinone methide 5, but also
generates reactive ROS catalytically in the presence of the iron
complexes (6, 7, and iron ions). These processes are expected
to act synergistically.
Transformations of the Prodrugs in the Presence of
Hydrogen Peroxide. We studied the cleavage/activation of
the prodrugs (0.9 mM) in the presence of hydrogen peroxide
(9 mM) by using electrospray ionization (ESI) mass
spectrometry. The data for two representative compounds, 1a
and 1e, are shown in Figure 2. The experiments were
We prepared a series of prodrugs 1 with different
substituents R′ and R″ (Scheme 2). These substituents were
expected to modulate cell-membrane permeability of the
prodrugs and their reactivity toward H2O2.
Synthesis of Prodrugs. Starting materials 9a−c and 6c−e
were required for synthesis of compounds 1a−e (Scheme 3).
Scheme 3. Synthesis of the Produgs (A) and the
Corresponding Starting Materials, Which Were Not
a
Reported Before (B, C)
Figure 2. ESI mass spectra of mixtures consisting of hydrogen
peroxide (9 mM), triethylamine (0.9 mM), and either prodrug 1a (A)
or prodrug 1e (B) (concentration of the prodrugs 0.9 mM) in
CH3CN/water (10/1.1, v/v) solution. After addition of H2O2 to the
mixtures and their incubation for the specific time, the mass spectra
were acquired: (A) incubation times are 2 min (a), 5 min (b), and 10
min (c); (B) incubation times are 2 min (a) and 10 min (b). The
labeling scheme used is explained in Scheme 2; nonassigned peaks are
labeled with an asterisk.
a
Reagents and conditions: (a) triphosgene, toluene, 120 °C; (b) (1)
NHMe2, BuLi, −78 °C; (2) B(OiPr)3, −78 °C; (3) aqueous NaOH;
(c) pinacol, 22 °C; (d) R″C(O)H, Na[B(CN)H3].
9a was purchased from commercial sources, 9c was prepared by
esterification of 4-(hydroxymethyl)-3-methylphenylboronic
acid with pinacol, and 9b was obtained in accordance with
Scheme 3B. In particular, the boronic acid residue was first
introduced in place of the bromide in 4-bromo-3-fluorobenzyl
alcohol. The same protocol was applied earlier by the group of
Armstrong to prepare similar arylboronic acids.21 The resulting
4-(hydroxymethyl)-3-fluorophenylboronic acid was esterified
with pinacol to obtain 9b. Aminoferrocene 6c was prepared
according to the protocol reported by the group of Heinze.22
Its alkylated derivatives 6d and 6e were obtained by reductive
amination of 6c in the presence of correspondingly
acetaldehyde and benzaldehyde (Scheme 3B).
Prodrugs 1 (Scheme 3) were synthesized by coupling
aminoferrocenes 6c−e with pinacol esters of 4-
(hydroxymethyl)phenylboronic acids 9a−c in the presence of
triphosgene. Yields of the isolated products were in the range of
5−71%. The prodrugs based on the unsubsituted amino-
ferrocene (R′ = H, 1a−c) were obtained with higher yields
(24−71%) than those with R′ ≠ H (1d−e, <20%). The latter
fact is explained by the lower reactivity of the substituted
aminoferrocenes with triphosgene.
conducted in CH3CN/water (10/1.1, v/v) solvent mixtures
whose pH was adjusted to 7 by addition of NEt3. The peak
corresponding to the product of the B−C bond cleavage in
prodrug 1a (2a) was observed already 2 min after addition of
H2O2 (trace a, Figure 2A). As expected, the intensity of this
peak is decreased with increasing incubation time (traces b and
c). The peak corresponding to the product of phenolate
decomposition (aminoferrocene 6c) could be detected after 5
min of incubation (trace b). Another product of the latter
reaction should be p-quinone methide 5. This compound was
observed neither in positive nor in negative detection mode in
the spectra of the corresponding mixtures. However, we could
detect the product of scavenging of 5 by water by using thin-
layer chromatography (TLC) and the corresponding reference
compound 4-(hydroxymethyl)phenol (Figure S13, Supporting
Information). Moreover, we observed that both [N-(4-
hydroxybenzyl)amino]ferrocene (6c(QM)) and [N,N-bis(4-
hydroxybenzyl)amino]ferrocene (6c(QM)2) were formed in
prodrug 1/H2O2 mixtures after 10 min of incubation, which
926
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
confirms the formation of electrophilic p-quinone methide at
these conditions. These products were not formed at lower
concentrations of the prodrugs (≤100 μM, data not shown).
Since experiments with cells were conducted with prodrugs at
≤100 μM, one does not expect that 6c(QM)x (x = 1, 2)
adducts will be generated in cells. In accordance with literature
reports,13 in cells, 5 would rather alkylate more nucleophilic
glutathione, which is present in high concentrations, 5−10 mM.
Prodrug 1e is degraded in the presence of hydrogen peroxide
similarly to 1a (Figure 2B). In particular, after 10 min of
incubation with H2O2, compound 1e is fully decomposed with
formation of (N-benzylamino)ferrocene (6e) and its adduct
with p-quinone methide (6e(QM)).
Iron Release from Prodrugs 1a−e. The fate of
aminoferrocenes released from the prodrugs in the presence
of H2O2 was studied next by using ESI mass spectrometry and
UV−vis spectroscopy. In particular, we observed that in the
mass spectra of mixtures of 6c (0.9 mM) and H2O2 (90 mM)
the peak corresponding to 6c disappeared after 10 min of
incubation, whereas two new peaks appeared, at 162 m/z,
which corresponds to bis(aminocyclopentadiene) ((H2N-
CpH)2), and at 132 m/z, which corresponds to dicyclopenta-
diene ((CpH)2; Figure 3). These spectral changes indicate that
Figure 4. (A) Reaction scheme used for detection of iron ions released
from prodrugs 1a−e in the presence of H2O2. (B) Prodrug 1a (0.1
mM) dissolved in MOPS buffer (100 mM, pH 7.5) and treated (left
cuvette) or not treated (right cuvette) with H2O2 (10 mM) for 100
min and then with sodium dithionite (20 mM) and 2,2′-bipyridine
(300 μM). (C) Absorbance at 519 nm of mixtures containing iron
complexes indicated on the plot (100 μM) dissolved in MOPS buffer
(100 mM, pH 7.5) and treated with varying H2O2 concentrations for
100 min. Before the absorbance measurement the mixture was treated
with sodium dithionite (20 mM) and 2,2′-bipyridine (300 μM). (D)
Kinetics of iron release from prodrug 1a in the absence of H2O2 and
presence of H2O2 (0.1 and 1.0 mM). All conditions are the same as
described for (C).
control, (N-acetylamino)ferrocene (10), which does not
contain the cleavable group, was stable in the presence of H2O2.
It has previously been known that aminoferrocene 6c is not
stable under oxidative conditions.23 However, neither the rate
nor the mechanism of its decomposition in aqueous solution
has been reported yet. On the basis of our data, we suggested
the mechanism of oxidative decomposition of 6c, which is
outlined in Scheme 4. In the first step, an 18e− iron(II)
Figure 3. ESI mass spectra of mixtures consisting of hydrogen
peroxide (90 mM), triethylamine (0.9 mM), and either 6c (A) or 6e
(B) (concentration 0.9 mM) in CH3CN/water (10/1.1, v/v) solution.
After addition of H2O2 to the mixtures and their incubation for the
specific time, the mass spectra were acquired: (A) incubation times are
2 min (a), 5 min (b), and 10 min (c); (B) incubation times are 2 min
(a) and 20 min (b). (CpH)2 = dicyclapentadiene, and (H2N-CpH)2 =
bis(aminocyclopentadiene). Nonassigned peaks are labeled with an
asterisk.
Scheme 4. Possible Mechanism of Decomposition of
Aminoferrocene 6c under Oxidative Conditions
6c is decomposed in the presence of H2O2 with release of the
ligands. Another product obtained in this reaction should be
iron ions. Thus, H2O2-induced decomposition of prodrugs 1a−
c does not stop at the stage of aminoferrocene 6c. We
confirmed this conclusion by detecting iron ions in mixtures of
prodrugs 1a−c and H2O2 by using the chromogenic reaction
with 2,2′-bipyridine (bipy). This assay is based on the formation
of red [Fe(bipy)3]2+ complex in the presence of iron(II) ions
(Figure 4A,B). For example, we observed that H2O2 (1 mM)
induced conversion of 95% of prodrug 1a, 92% of 1b, and 74%
of 1c (0.1 mM) into iron ions within 100 min. We explain the
lower conversion of 1c by the +I effect of the 2-methyl group
(R′, Scheme 2). In particular, the electron donor group R′
disfavors formation of the phenolate 3c, leading to lowering of
the yield of iron ions released from prodrug 1c. We observed
that the amount of iron released from the prodrugs was
proportional to the reaction time and the concentration of
hydrogen peroxide (Figure 4C,D). As expected, a negative
complex, 6c, is oxidized with formation of a 17e− iron(III)
complex, 11c, and a 18e− radical iron(II) complex, 12c. The
latter compound is expected to be unstable due to its electron
deficiency. It is decomposed with formation of the ligands and
927
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
iron ions. Among the prodrugs based on alkylated amino-
ferrocenes (1d, 1e), only 1d generates in the presence of H2O2
(1 mM) iron ions in amounts comparable to those released
from prodrugs 1a−c at the same conditions, 72% in 100 min
(Table 1). In contrast, the H2O2-induced degradation of 1e is
6c−e is currently under way and will be published separately in
due course.
Amino substituents at the ferrocene periphery are strongly
activating groups for ferrocene oxidation with formation of
ferrocenium ions (Fc+). For example, the oxidation potential of
aminoferrocene 6c was found to be 0.37 V more negative than
that of ferrocene.25 We observed that the oxidizing potential of
(N-benzylamino)ferrocene (6e) was also substantially more
negative (0.34 V) than that of ferrocene. Correspondingly, one
could expect that Fc+ species will be formed from compound 6e
in the presence of oxidants such as H2O2 and O2. According to
the data reported by Sohn et al.,26 formation of Fc+ ions from
ferrocenes correlates with the increased light absorption in the
region between 250 and 350 nm. In this spectral region,
ferrocene derivatives exhibit intense charge transfer transitions
whereas ferrocenes have only weak transitions. We observed
that absorbance at 300 nm of solutions of prodrug 1e is
strongly increased upon addition of H2O2. The increase was
substantially weaker in the presence of 1c and 1d (Figure 5).
Table 1. Efficacy of Release of Iron Ions and ROS
Generation in Vitro from the Prodrugs and Control
Compounds in the Presence of H2O2
b
c
iron
complex
efficacy of iron release
(%)
efficacy of ROS generation
(%)
a
entry
1
2
3
4
5
6
7
FeSO4
10
100
0
100
4
1a
95
92
74
72
9
56
50
52
50
1b
1c
1d
d
1e
24 (53)
a
Structures of prodrugs 1a−e are given in Scheme 2. Compound 10 is
b
a negative control, (N-acetylamino)ferrocene. Iron release efficacy =
(A519 nm(prodrug) − A519 nm(buffer))/(A519 nm(FeSO4) −
A519 nm(buffer)), where A519 nm(prodrug) is the absorbance at 519
nm of a prodrug (100 μM) solution in MOPS buffer (100 mM, pH
7.5), which was treated first with H2O2 (1 mM) for 100 min and then
with Na2S2O4 (20 mM) and 2,2′-bipyridine (300 μM) and
A519 nm(buffer) and A519 nm(FeSO4) are the absorbances at 519 nm
of the MOPS buffer and iron sulfate (100 μM) dissolved in this buffer
treated similarly to the prodrug. The experiments were conducted at
least three times. The standard deviation is not higher than 11% of the
c
values given in the table. Efficacy of ROS release = (F(prodrug) −
F0)/(F(FeSO4) − F0), where F(prodrug) is the emission at 531 nm
(λex = 501 nm) of a solution containing MOPS buffer (100 mM, pH
7.5), EDTA (10 mM), glutathione (5 mM), 2′,7′-dichlorofluorescein
(10 μM), and H2O2 (10 mM), which was incubated with a prodrug
(100 μM) for 17 min, F(FeSO4) is the emission of a mixture
containing FeSO4 in place of the prodrug, and F0 is the emission of a
mixture containing no iron complex. The experiments were conducted
at least three times. The standard deviation is not higher than 10% of
Figure 5. Iron complexes (100 μM) were dissolved in dimethylforma-
mide (100 μL), and concentrated H2O2 solution (30% in water, 1 μL)
was added: the final H2O2 concentration was 9.8 mM. Light
absorbance at 300 nm was monitored as a function of time.
Thus, we can conclude that after formation of 6e from the
prodrug 1e (R″ = Bn) and H2O2 the former compound is
further converted to ferrocenium species [6e]+.
d
the values given in the table. Incubation time 37 min.
practically stopped at the stage of aminoferrocene 6e: 1e
generates only 9% of iron ions in the presence of 1 mM H2O2
within 100 min. This conclusion was corroborated by the mass
spectrometric study (Figure 3). In particular, the intense peak
of 6e (0.9 mM) was observed in the mass spectra during at least
20 min of its incubation with H2O2 (90 mM). For comparison,
6a was not detectable in the mass spectra already after 10 min
of its incubation with the same concentration of H2O2
(compare trace c in Figure 3A with trace b in Figure 3B).
On the basis of the above suggested mechanism of the
oxidative degradation of aminoferrocene (Scheme 4), the
higher stability of N-benzyl-substituted 6e than unsubstituted
6c or N-ethyl-substituted 6d can be explained by the stronger
electron acceptor effect (−I) of the benzyl group in comparison
to the proton or the ethyl substituent (R″ in Scheme 2). We
estimated the inductive effects by comparing the basicities of
anilines, which can be considered as analogues of amino-
ferrocenes (pKa): PhNHEt (5.12), PhNH2 (4.63), and
PhNHCH2Ph (3.89).24 The stronger electron acceptor effect
of the Bn group leads to the destabilization of the radical cation
12c. Consequently, less of this ion is formed and less iron ions
are released upon its decomposition. The detailed theoretical
study of the reaction of oxidative cleavage of aminoferrocenes
Not only hydrogen peroxide, but also other naturally
occurring reactive oxygen species can potentially induce
activation of our prodrugs. In particular, we have studied the
stability of the representative prodrug 1a in the presence of
hydroxyl radicals (HO•), singlet oxygen (1O2), and superoxide
anion radical (O2−; Figure S14, Supporting Information). We
observed that hydroxyl radicals activated 1a more efficiently
than hydrogen peroxide, whereas the reactivity of singlet
oxygen and superoxide anion radical was substantially lower.
Generation of Reactive Oxygen Species in the
Presence of Prodrugs 1a−e. In accordance with the data
described above, prodrugs 1a−d are decomposed in the
presence of H2O2 with formation of iron ions whereas prodrug
1e generates Fc+ species. All these products can act as catalysts
of conversion of less reactive H2O2 into highly reactive hydroxyl
radicals (HO•).9,19 We studied the reaction of generation of
hydroxyl radicals in mixtures containing prodrugs and H2O2 by
using a fluorogenic 2′,7′-dichlorofluorescein probe. As a positive
control we took FeSO4 salt and as a negative control (N-
acetylamino)ferrocene (10), which was found to be stable
toward H2O2 (Figure 4). The kinetics of HO• generation in the
presence of positive and negative controls, as well as of a
representative prodrug 1a, are shown in Figure 6.
928
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
monitoring the boron concentration in cells before and after
their incubation with the prodrugs. Boron was detected by
using a curcumin probe in accordance with a protocol described
elsewhere (Table 2).28
Table 2. Cytotoxicity, Cellular Membrane Permeability, and
ROS Generation Efficiency of Prodrugs 1a−e and Controls
a
Fe
IC50
membrane
permeability
efficacy of ROS
generation in cells
b
c
entry complex
(μM)
Figure 6. Monitoring changes of the fluorescence intensity (λex = 501
nm, λem = 531 nm) of solutions containing 2′,7′-dichlorofluorescein
(10 μM), iron complex (100 μM), MOPS buffer (100 mM, pH 7.5),
N,N,N′,N′-ethylenediaminetetraacetic acid (EDTA; 10 mM), gluta-
thione (5 mM), and H2O2 (10 mM). Iron complexes (100 μM) were
added at the time point indicated by an arrow. The fluorescence is
increased due to HO•-induced oxidation of nonfluorescent 2′,7′-
dichlorofluorescein to fluorescent 2′,7′-dichlorofluorescein.
1
Fe(8-
3
1
d
HQ)2
2
3
4
5
6
7
8
10
1a
1b
1c
1d
1e
6e
>200
1.1 0.2
52
55
40
24
9
3
4
3
2
1.0 0.2
0.3 0.1
1.3 0.2
3.1 0.3
1.8 0.3
17.5 1.4
e
12.3 0.8 (41)
20.0 1.1 (15.4)
25.7 2.1 (8.3)
28.8 2.3 (16)
2
>200
The rate of this reaction in the presence of 1a is increased
with time (Figure 7), which indicates that the reaction is
a
HL-60 cells were incubated with iron complexes over 48 h, and their
viability was determined by using the MTT assay (MTT = 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide). The cell
b
membrane permeability of prodrugs 1a−e was determined by
detecting the boron concentration in cells after their incubation with
the prodrugs. 1a was used as a reference: its permeability was set to be
c
1, and its absolute amount in cells corresponds to 4.4 fmol/cell. ROS
generation in cells was monitored by using 2′,7′-dichlorofluorescein
diacetate. Efficacy = F(prodrug)/F0, where F(prodrug) is the mean
fluorescence intensity of cells incubated with a prodrug and F0 is the
mean fluorescence intensity of cells which were not treated with any
d
iron complex. The prodrug concentration is 100 μM. Fe(8-HQ)2 is a
positive control which was used to load cells with iron ions; this
complex is formed from 1 equiv of FeCl3 and 2 equiv of 8-
e
hydroxyquinoline. In parentheses is given the relative ROS generation
efficacy corrected for the membrane permeability of prodrugs, which is
equal to [F(prodrug)/F0]/(relative membrane permeability).
Figure 7. Monitoring changes of the fluorescence intensity (λex = 501
nm, λem = 531 nm) of solutions containing prodrug 1a (100 μM), 2′,7′-
dichlorofluorescein (100 μM), MOPS buffer (100 mM, pH 7.5),
EDTA (10 mM), glutathione (5 mM), and variable amounts of H2O2
(concentrations are shown on the plot).
The permeabilty of prodrug 1a (R′ = R″ = H) was taken as a
reference. We observed that, relative to this compound, more
polar 1b (R′ = F, R″ = H) is ∼3 times less permeable (entry 4)
whereas less polar 1c (R′ = CH3, R″ = H), 1d (R′ = H, R″ =
CH2CH3), and 1e (R′ = H, R″ = CH2Ph) are 1.3−3.1-fold more
permeable (entries 5−7). The cell loading with the reference
compound 1a corresponds to 4.4 fmol/cell.
Next we studied the cytotoxicity of the prodrugs and control
compounds on the human promyelocytic leukemia cell line
(HL-60; Table 2). In these experiments an iron(III) complex
with 8-hydroxyquinoline (Fe(8-HQ)2) was applied as a positive
control. This coordination compound permeates the cell
membrane substantially more quickly than simple iron salts
do, which allows loading the cells with iron ions in a
reproducible fashion.29 We observed that Fe(8-HQ)2 is highly
autocatalytic.27 This is caused by formation of a highly active
catalyst (iron ions) as a product.
The efficacy of HO• generation was determined for all
prepared prodrugs 17 min after the addition of the iron
complex to a mixture of H2O2 and 2′,7′-dichlorofluorescein
(Table 1). As a reference (100%) we used the ROS amount
released in the presence of FeSO4. We observed that the
efficiency of ROS generation by prodrugs 1a−d correlates with
the efficiency of iron release from these compounds (Table 1).
It is noteworthy that, though prodrug 1e releases an 8.2−10.6-
fold lower amount of iron ions than 1a−d, it generates only ∼2
times less hydroxyl radicals (24%) than the corresponding
prodrugs (Table 1). These data indicate that both ferrocenium
ions released from 1e as a major product and iron ions (minor
product) act in this case as catalysts of ROS generation.
Though prodrug 1e generates less ROS than other prodrugs
within 17 min, when the reaction is allowed to proceed longer,
the amount of ROS released in the presence of this compound
is increased to the level of that of prodrugs 1a−d, 53% (in 37
min, Table 1). The lower reactivity can be considered as an
advantage of 1e over other studied prodrugs, since this property
may result in its higher selectivity toward cancer cells.
cytotoxic, IC50 = 3
1 μM (entry 1). In contrast, stable
ferrocene complex 10 (negative control) is not toxic at all, IC50
> 200 μM (entry 2). This confirms that simple ferrocenes are
not efficient catalysts of ROS generation in cells. The
cytotoxicity of prodrugs 1a−e is intermediate between those
of positive and negative controls (entries 3−7, Table 2). The
most efficient complex in this series is prodrug 1e, whose IC50
= 9 2 μM (entry 7). It is the only compound which releases
ferrocenium species rather than iron ions in the presence of
hydrogen peroxide (see the discussion above). These data
indicate that in cells Fc+ is more cytotoxic than iron ions. This
is a surprising fact, since in vitro iron ions are more potent
catalysts of ROS generation than Fc+ ions (Table 1). We
Cytotoxicity of Prodrugs 1a−e. The natural abundance
of boron in cells is rather low. Therefore, we could determine
the cell membrane permeabilty of prodrugs 1a−e by
929
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
suggest that the reactivity is reversed due to the ability of cells
to neutralize iron overload via efficient natural, biochemical
mechanisms. In contrast, Fc+ is not a natural compound.
Therefore, its metabolism is probably substantially slower.
Consequently, the lifetime of Fc+ in the cell should be longer.
We observed that the cytoxicity of all iron releasing prodrugs
1a−d correlates well with their membrane permeability (Figure
Figure 9. Cytotoxicity of prodrug 1e toward cancerous and
noncancerous cells: human promyelocytic leukemia (HL-60), human
glioblastoma-astrocytoma (U373), and fibroblasts.
By using 2′,7′-dichlorofluorescein diacetate as an intracellular
probe, we observed that the incubation of HL-60 cells with the
prodrugs 1a−e (100 μM) for 5 h leads to a significant increase
of the intracellular concentration of ROS (Table 2). In contrast,
stable ferrocene derivative 10 does not affect the intracellular
concentration of ROS at the same conditions. Interstingly, the
ROS amount present in normal cells (fibroblasts) was found to
be >10-fold lower than that in cancer cells (HL-60, Figure S15,
Supporting Information), which is in agreement with the low
toxicity of prodrug 1e toward normal cells (Figure 9). The
cytotoxicity of the prodrugs correlates with the efficacy of ROS
generation in cells loaded with these prodrugs (Figure 10).
These data confirm the mechanism of cytotoxicity of the
prodrugs, which is outlined in Figure 1 and Scheme 2.
Figure 8. Correlation of the cytotoxicity of prodrugs 1a−e and their
cell membrane permeabilities.
8). In contrast, the properties of prodrug 1e strongly deviate
from this correlation. These data also support our conclusion
that complexes 1a−d and 1e release different active species in
cells.
According to the data discussed above, prodrug 1e acts in the
following fashion: it permeates the cellular membrane, is
converted into toxic p-quinone methide (5a) and intermediate
6e, and finally 6e is oxidized with formation of another toxic
component, [6e]+ (Scheme 2). Therefore, providing it reaches
the intracellular space, intermediate 6e should also exhibit some
cytotoxicity due to the generation of toxic ferrocenium species.
Interestingly, we observed that 6e itself is not cytotoxic at all
(Table 2). This fact may indicate that 6e is not cell membrane
permeable. However, simple, uncharged ferrocenes usually
permeate the cellular membrane well. We explain these data in
the following way. Since compound 6e has an unusually low
oxidation potential (−0.34 V versus ferrocene), it is easily
oxidized in air with formation of [6e]+. Indeed, we observed
that already 1 h after its dissolution in the medium saturated
with air, compound 6e could not be detected by thin-layer
chromatography. In contrast, a colored, low-mobility spot was
detected which may correspond to ferrocenium species. The
ferrocenium complex [6e]+ is charged and can have low
membrane permeability. Thus, in contrast to 1e, compound 6e
is not applicable as a prodrug.
Figure 10. Correlation of the cytotoxicity of the prodrugs 1a−e and
the efficacy of ROS generation in cells loaded with these prodrugs.
The data used to plot this graph and the experimental conditions for
determination of IC50 values and efficacies of the ROS generation are
presented in Table 2 and its footnotes.
Furthermore, we tested the cytoxicity of our most active
prodrug 1e toward other cell types: human glioblastoma-
astrocytoma cells (U373) and fibroblasts (Figure 9). The
former ones are malignant cells, whereas the latter ones are
normal (nonmalignant) cells. We observed that as well as HL-
Since the medium is known to scavenge excess ROS, which is
released by cancer cells, the ROS concentration in cultivated
cells is expected to be lower than that in the same cells in vivo.
To model the microenvironment of cancer cells in vivo, we
preincubated the cells with nontoxic H2O2 concentrations for 1
h (Supporting Information). We found that cells prepared in
such a way were 1.4-fold (U373) to 3-fold (HL-60) more
sensitive to the representative prodrug 1a than the untreated
cells.
Interestingly, HL-60 cells overexpressing catalase (HL-60cat)
exhibit H2O2-independent sensitivity to prodrug 1a. These data
additionally confirm that H2O2 indeed acts as an intracellular
trigger of the activity of the prodrugs.
60 cells U373 cells were sensitive to prodrug 1e (IC50 = 25
2
μM) whereas normal cells (fibroblasts) turned out to be
resistant at a concentration of the prodrug of up to 100 μM
(Figure 9). In particular, after incubation of fibroblasts for 48 h
with 1e (100 μM), 78% of the cells (relative to the control cells,
which were incubated in the medium only) remained viable. In
contrast, only 4% of the HL-60 cells and 0% of the U373 cells
incubated with 1e (100 μM) for 48 h remained viable. These
data indicate that 1e is substantially more toxic to malignant
cells (IC50 = 9−25 μM) than to nonmalignant fibroblasts (IC50
> 100 μM).
930
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
1.35 (s, 12H), 4.71 (s, 2H), 7.37 (d, 2H, 3J = 8.2 Hz), 7.81 (d, 2H, 3J =
8.2 Hz).
CONCLUSIONS
■
We prepared five novel aminoferrocene-based prodrugs which
are protected with a 4-[(carbonyloxy)methyl]phenylboronic
acid pinacol ester residue. We demonstrated that four of these
prodrugs are activated at cancer-specific conditions (high
concentration of reactive oxygen species) with formation of
toxic p-quinone methide and iron ions. In contrast, one of the
prodrugs (4-[[[(N-ferrocenyl-N-benzylamino)carbonyl]oxy]-
methyl]phenylboronic acid pinacol ester, 1e) is activated at
the same conditions with formation of p-quinone methide and
(benzylamino)ferrocenium ([6e]+) ions. The released products
act in a concerted fashion. In particular, p-quinone methide
alkylates glutathione and inhibits the antioxidative system of the
cell, whereas iron or ferrocenium ions induce catalytic
generation of highly reactive ROS (hydroxyl radicals). The
activation reaction proceeds autocatalytically, which leads to
generation of large quantities of ROS in cancer cells. We
observed that among the studied compounds 1e exhibited the
highest cytotoxicity toward representative cancer cell lines
(nonadherent, human promyelocytic leukemia (HL-60), and
adherent, human glioblastoma-astrocytoma (U373)). Interest-
ingly, this prodrug was found to be not toxic toward
nonmalignant cells (fibroblasts). We observed that with the
exception of 1e the cytotoxicity of the prodrugs correlates with
their membrane permeability. Moreover, higher cytotoxicity is
observed for the prodrugs releasing ferrocenium ions rather
than iron ions. Both these trends provide us a rationale for
future improvements of the properties of the aminoferrocene-
based prodrugs.
4-[[[(Ferrocenylamino)carbonyl]oxy]methyl]phenylboronic Acid
Pinacol Ester (prodrug 1a). Aminoferrocene (1.20 g, 6.0 mmol)
and triphosgene (1.80 g, 6.0 mmol) were added to dry toluene (110
mL) and purged with argon. The mixture was heated at 120 °C for 1 h.
During this time all starting materials were dissolved. The solution
obtained was cooled to 22 °C, and 4-(hydroxymethyl)phenylboronic
acid pinacol ester (1.40 g, 6.0 mmol) dissolved in CH2Cl2 (150 mL)
was added dropwise over 80 min. The solution was left stirring
overnight at 22 °C. Then the solvent was removed in vacuum (10
mbar), and the product was purified by column chromatography on
silica gel using CH2Cl2 as the eluent. Yield: 1.97 g (71%). Rf = 0.3
(silica, eluent CH2Cl2). 1H NMR (400 MHz, include the list for
acetone-d6): δ (ppm) 1.33 (s, 12H), 3.95 (s, 2H), 4.12 (s, 5H), 4.59
(s, 2H), 5.17 (s, 2H), 7.42 (d, 2H), 7.76 (d, 2H). 13C NMR (100.55
MHz, acetone-d6): δ (ppm) 25.3, 61.1, 64.7, 66.5, 69.8, 84.6, 127.7,
127.8, 135.7, 141.4, 154.6. ESI TOF mass spectrometry (negative
mode): m/z calcd for [M − pinacol − H]− (C18H17BFeNO4) 378.07,
found 378.04. Anal. Calcd for C24H28BFeNO4·0.5H2O: C, 61.3; H, 6.2;
N, 3.0. Found: C, 61.3; H, 5.9; N, 3.1.
4-(Hydroxymethyl)-2-fluorophenylboronic Acid. This compound
was prepared according to the protocol described in ref 21 for
analogous derivatives. Dimethylamine (1.9 mL of 2 M solution in
THF, 3.7 mmol) was cooled to −78 °C. n-Butyllithium (1.7 mL of 1.6
M solution in hexane, 2.7 mmol) was added dropwise within 5 min.
Next 4-bromo-3-fluorobenzyl alcohol (0.56 g, 2.7 mmol) dissolved in
THF (1 mL) was slowly added while the temperature of the reaction
mixture was kept at −78 °C. After the addition of the latter reagent
and stirring for 5 min, the mixture was allowed to warm to room
temperature (22 °C), and the solvent was removed in vacuum (10
mbar). The rest was redissolved in THF (10 mL) and cooled back to
−78 °C. n-butyllithium (1.8 mL of 1.6 M solution, 2.8 mmol) was
added within 5 min, and the mixture was allowed to react for 1 h while
being stirred. Finally, triisopropyl borate (1.5 mL, 6.5 mmol) was
added dropwise, and after 1 h the unreacted compounds were
quenched with NaOH solution in water (5.1 mL, 1 M) and then with
water (5.1 mL). After that, the mixture was warmed to 22 °C, all
volatiles were removed in vacuum (10 mbar), and a mixture of water
and diethyl ether was added. The aqueous phase was separated and
acidified with concentrated HCl to pH 4. At these conditions the raw
product was precipitated. It was purified by column chromatography
on silica gel using CH2Cl2/EtOAc (1/1, v/v) as the eluent. Yield: 0.90
EXPERIMENTAL SECTION
■
General Information. Commercially available chemicals of the
best quality from Aldrich/Sigma/Fluka (Germany) were obtained and
used without purification. NMR spectra were acquired on a Bruker
Avance DRX 200, Bruker Avance II 400, or Bruker Avance III 600
spectrometer. ESI mass spectra were recorded on an ESI Q-tof Ultima
API mass spectrometer (Waters), FAB mass spectra on a Jeol JMS-700
instrument using p-nitrobenzyl alcohol as a matrix, and EI mass spectra
on a Finnigan MAT 8200 instrument. IR spectra were recorded on a
Biorad Excalibur FTS 3000 instrument. The samples were prepared as
KBr pellets. Cyclic voltamperometry was conducted on an EG@G
262A potentiostat/galvanometer. C, H, N analysis was performed in
the microanalytical laboratory of the chemical institute of the
University of Heidelberg. For analytical reversed-phase thin-layer
chromatography, Polygramm TLC plates (Macherey-Nagel) were
used. UV/vis spectra were acquired on a Varian Cary 100 Bio UV/vis
spectrophotometer by using 1 cm optical path black-wall absorption
semimicrocuvettes (Hellma GmbH, Germany) with a sample volume
of 0.7 mL. Fluorescence spectra were acquired on a Varian Cary
Eclipse fluorescence spectrophotometer by using black-wall fluo-
rescence semimicrocuvettes (Hellma GmbH) with a sample volume of
0.7 mL. The fluorescence of live HL-60 cells was quantified by using
an Accuri C6 flow cytometer. The data were processed by using the
CFLow Plus (Accuri) software package. The purity of the prodrugs
used in the biological tests was determined by C, H, N analysis and
thin-layer chromatography. According to these data, the prodrug
samples were ≥95% pure.
1
g (99%). Rf = 0.5 (silica, eluent CH2Cl2/EtOAc, 1/1, v/v). H NMR
(200 MHz, D2O): δ (ppm) 7.03 (m, 2 H), 7.49 (m, 1H).
4-(Hydroxymethyl)-2-fluorophenylboronic Acid Pinacol Ester. 4-
(Hydroxymethyl)-2-fluorophenylboronic acid (152 mg, 0.89 mmol)
and pinacol (104 mg, 0.90 mmol) in tetrahydrofuran (17.2 mL) were
refluxed for 3 h. The solvent was removed in vacuum (10 mbar), and
the crude product obtained was purified by column chromatography
on silica gel using CH2Cl2/EtOAc (1/1, v/v) as the eluent. Yield: 115
mg (51%). Rf = 0.7 (silica, eluent CH2Cl2/EtOAc, 1/1, v/v). 1H NMR
(200 MHz, CDCl3): δ (ppm) 1.36 (s, 12H), 4.71 (s, 2H), 7.08 (m,
3
2H), 7.71 (t, 1H, J = 7.4 Hz).
4-[[[(Ferrocenylamino)carbonyl]oxy]methyl]-2-fluorophenylbor-
onic Acid Pinacol Ester (Prodrug 1b). Aminoferrocene (82 mg, 0.4
mmol) and triphosgene (122 mg, 0.4 mmol) were added to dry
toluene (7.5 mL) and purged with argon. The mixture was heated to
120 °C and kept at this temperature until all starting materials were
dissolved. The solution obtained was cooled to 22 °C, and 4-
(hydroxymethyl)-2-fluorophenylboronic acid pinacol ester (103 mg,
0.4 mmol) dissolved in CH2Cl2 (10 mL) was added dropwise. The
solution was left stirring overnight at 22 °C. Then the solvent was
removed in vacuum (10 mbar), and the product was purified by
column chromatography on silica gel using hexane/EtOAc (8/2, v/v)
as the eluent. Yield: 48 mg (24%). Rf = 0.4 (silica, eluent hexane/
EtOAc, 8/2, v/v). 1H NMR (200 MHz, acetone-d6): δ (ppm) 1.34 (s,
12H), 3.94 (m, 2H), 4.11 (s, 5H), 4.56 (s, 2H), 5.19 (s, 2H), 7.13 (d,
1H), 7.23 (d, 1H), 7.71 (t, 1H). 13C NMR (100.55 MHz, acetone-d6):
δ (ppm) 25.2, 61.1 (two overlapping signals), 64.7, 65.6, 69.8, 84.7,
Synthesis. 4-(Hydroxymethyl)phenylboronic Acid Pinacol
Ester. A suspension of 4-(hydroxymethyl)boronic acid (1.00 g, 6.6
mmol) and pinacol (0.79 g, 6.7 mmol) in THF (40 mL) was refluxed
for 22 h. During this time the starting materials were completely
dissolved. The solvent was removed in vacuum (10 mbar) and the
residue redissolved in CH2Cl2/EtOAc and purified by column
chromatography on silica gel using a mixture of CH2Cl2/EtOAc (9/
1, v/v) as the eluent. Yield: 1.4 g (92%). Rf = 0.3 (silica, eluent
1
CH2Cl2/EtOAc, 9/1, v/v). H NMR (200 MHz, CDCl3): δ (ppm)
931
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
114.9, 123.4, 137.9 (two overlapping signals), 166.9, 169.4. 19F NMR
(376.3 MHz, acetone-d6): δ (ppm) −103.2. FAB MS: m/z calcd for
C24H27BFNO4Fe 479.1, found 479.1. Anal. Calcd for C24H27BFNO4Fe
+ acetone (one molecule of solvent): C, 60.8; H, 6.2; N, 2.6. Found:
C, 60.4; H, 6.2; N, 2.6.
converts Fe3+ into Fe2+. Finally, 2,2′-bipyridine solution (3 μL, 100
mM in DMF) was added to form a dark red [Fe(2,2′-bipyridine)3]2+
complex (λmax = 519 nm). FeSO4 sulfate (100 μM) was used as a
positive control to determine the absorbance intensity when the
ferrocene complexes were completely converted by H2O2 to free Fe2+.
The efficacy of iron release from a prodrug in the presence of H2O2
(the data are provided in Table 2) was determined by using the
following formula: iron release efficacy = (A519 nm(prodrug) −
A519 nm(buffer))/(A519 nm(FeSO4) − A519 nm(buffer)), where
A519 nm(prodrug) is the absorbance at 519 nm of a prodrug treated
with H2O2 (1 mM) for 100 min and then with 2,2′-bipyridine as
described above, A519 nm(FeSO4) is the absorbance at 519 nm of FeSO4
solution treated in a similar way, and A519 nm(buffer) is the absorbance
at 519 nm of the buffer treated analogously.
Reversed-Phase Thin-Layer Chromatography for Monitoring
Decomposition of 1a. Three probes were analyzed by reversed-
phase TLC using plates covered with C18-modified silica (ALUGRAM
RP-18W, Macherey-Nagel) and 1/1 (v/v) DMF/MOPS buffer (100
mM, pH 7.5) as the eluent. The first probe was prodrug 1a (10 mM)
dissolved in DMF containing 2% water. The second one was prodrug
1a (10 mM) dissolved in the same solvent treated with excess H2O2
(100 mM) for 24 h. The third probe was 4-(hydroxymethyl)phenol
(8a; 10 mM) also dissolved in DMF containing 2% water. At these
conditions Rf of phenol 8a was found to be 0.80, whereas that of
prodrug 1a was found to be 0.08. In the TLC of 1a treated with H2O2,
the spot corresponding to the intact prodrug was not observed
whereas the intense spot at Rf = 0.80 corresponding to phenol 8a was
detected (Figure S13, Supporting Information). In the TLC of 1a
treated with H2O2 for 30 min, another spot at Rf = 0.4 was also
observed (data not shown). According to ESI MS analysis, this
product could be identified as compound 2a, which is formed as a
result of cleavage of the B−C bond in 1a (Scheme 2).
4-((Ferrocenylaminocarbonyl)oxymethyl)-2-mehylphenylboronic
acid pinacol ester (prodrug 1c). Triphosgene (1.61 g, 5,44 mmol)
and aminoferrocen (1.09 g, 5.44 mmol) were added to toluene (98
mL) and purged with argon. The mixture was heated up to 120 °C and
kept at this temperature until all starting materials were dissolved (∼30
min). The solution obtained was cooled down to 22 °C and 4-
(hydroxymethyl)-2-methylphenylboronic acid pinacol ester (1.35 g,
5.44 mmol) dissolved in CH2Cl2 (132 mL) was added dropwise. The
solution was left stirring at 22 °C for 44 h. Then, the solvent was
removed in vacuum (10 mbar) and the product was purified by
column chromatography on silica gel using hexane/EtOAc (10/2, v/v)
as eluent. Yield: 0.83 g (32%). Rf = 0.33 (silica, eluent − CH2Cl2/
EtOAc, 7/2, v/v). 1H NMR (200 MHz, acetone-d6): δ (ppm) 7.72 (d,
1H), 7.21 (m, 2H), 5.12 (s, 2H), 4.56 (s, 2H), 4.11 (s, 5H), 3.93 (s,
2H), 2.52 (t, 3H),1.34 (s, 12H). 13C NMR (100.55 MHz, acetone-d6):
δ (ppm) 145.8, 141.0, 137.0, 129.7, 124.7, 84.3, 69.8, 66.5, 64.7, 61.1,
25.3, 22.5. FAB MS: calculated for C25H30BFeNO4 475.2, found 475.2
m/z. C, H, N analysis: calculated for C25H30BFeNO4 − C 63.2%; H
6.4%; N 3.0%; found - C 63.3%; H 6.6%; N 2.9%.
4-[[[(N-Ethyl-N-ferrocenylamino)carbonyl]oxy]methyl]-2-phenyl-
boronic Acid Pinacol Ester (Prodrug 1d). Aminoferrocene (0.40 g,
1.98 mmol) and acetaldehyde (0.1 mL, 1.98 mmol) were dissolved in
methanol (10 mL) and refluxed for 2 h. Then Na[B(CN)H3] (0.12 g,
1.98 mmol) dissolved in MeOH (10 mL) was slowly added. The
mixture obtained was acidified with HCl (2 mL, 1 M in water) and left
stirring for 30 min. Afterward the solvent was removed in vacuum
(0.01 mbar), and the rest was mixed with triphosgene (0.59 g, 1.98
mmol) in toluene (25 mL). The suspension obtained was refluxed for
1 h, cooled to 22 °C, and then mixed with a solution of 4-
(hydroxymethyl)phenylboronic acid pinacol ester (0.46 g, 1.98 mmol)
in toluene (10 mL). The resulting solution was heated to 120 °C and
stirred at these conditions for 6 h. Then the solvent was removed in
vacuum (0.01 mbar), and the crude product was purified by column
chromatography on silica gel using hexane containing 5% acetone as
the eluent. Yield: 0.20 g (20%). Rf = 0.5 (silica, eluent hexane/acetone,
Monitoring Generation of ROS by Using Fluorescence Spectros-
copy. 2′,7′-Dichlorofluorescein diacetate (DCFH-DA; 4.9 mg) was
dissolved in DMF (100 μL) and mixed with aqueous NaOH (0.1 M,
900 μL). The resulting mixture was incubated for 30 min at 22 °C in
the dark to obtain a stock solution of 2′,7′-dichlorofluorescein (DCFH;
10 mM). Next a solution (1 mL) containing DCFH (10 μM), MOPS
buffer (100 mM, pH 7.5), EDTA (10 mM), glutathione (5 mM), and
H2O2 (10 mM) was prepared. Monitoring of the fluorescence (λex
=
1
501 nm, λem = 531 nm) of this solution was started. After 5 min, an
iron complex (100 μM, ferrocene prodrugs 1a−e, positive or negative
controls) was added, and the fluorescence monitoring was continued
until the fluorescence signal growth was stalled. The efficacy of ROS
generation (the data are provided in Table 2) was determined using
the following formula: efficacy = (F(prodrug) − F0)/(F(FeSO4) − F0),
where F(prodrug) is the fluorescence of the DCFH mixture treated
with a prodrug for 17 min, F(FeSO4) is the fluorescence of the same
mixture treated with FeSO4 for 17 min, and F0 is the fluorescence of a
mixture containing no iron complex.
CV Measurements. The concentration of iron complexes tested was
1 mM. Measurements were conducted at 22 °C. Ag/AgCl was used as
a reference electrode. The supporting electrolyte in experiments
conducted in pure DMF was tetramethylammonium hexafluorophos-
phate (160 mM). The aqueous buffer was phosphate-buffered saline
(PBS) buffer (phosphate, 9.5 mM; NaCl, 137.9 mM; KCl, 2.7 mM)
containing 50% DMF (v/v).
Cellular Assays. Cells and Cell Culture. The human glioblasto-
ma-astrocytoma cell line (U373) was cultured in Eagle's minimum
essential medium (EMEM) supplemented with 10% fetal calf serum
(FCS), 1% L-glutamine, and 1% penicillin/streptomycin. The human
promyelocytic leukemia cell line (HL60) was cultured in Roswell Park
Memorial Institute (RPMI) 1640 medium supplemented with 10%
FCS and 5 μg/mL penicillin/streptomycin (all media and supplements
from Gibco Invitrogen Corp., Karlsruhe, Germany). Fibroblasts were
grown in DMEM (Dulbecco's modified Eagle's medium) supple-
mented with 10% FCS, 1% L-glutamine, and 1% penicillin/
streptomycin.
5/1, v/v). H NMR (400 MHz, acetone-d6): δ (ppm) 1.27 (t, 3H),
1.33 (s, 12H), 3.77 (q, 2H), 4.00 (s, 2H), 4.13 (s, 5H), 4.53 (m, 2H),
5.22 (s, 2H), 7.46 (d, 1H), 7.77 (d, 2H). 13C NMR (100.55 MHz,
acetone-d6): δ (ppm) 14.4, 25.3, 45.8, 62.8, 65.1, 66.9, 67.5, 69.8, 84.6,
127.8, 128.1, 135.7 (two overlapping peaks), 139.1, 141.2. EI MS: m/z
calcd for C26H32BNO4Fe 489.2, found 489.2. IR spectra (in KBr):
wavenumber (cm−1) 3101, 2973, 1696, 1623. Anal. Calcd for
C26H32BNO4Fe: C, 63.8; H, 6.6; N, 2.9%. Found: C, 63.8; H, 6.8;
N, 2.9.
4-[[[(N-Benzyl-N-ferrocenylamino)carbonyl]oxy]methyl]-2-phe-
nylboronic Acid Pinacol Ester (Prodrug 1e). Compound 1e was
obtained analogously to 1d except that benzaldehyde (0.2 mL, 1.98
mmol) was used in place of acetaldehyde. Yield: 60 mg (5%). Rf = 0.4
(silica, eluent hexane/acetone, 5/1, v/v). 1H NMR (400 MHz,
acetone-d6): δ (ppm) 1.33 (s, 12H), 3.97 (s, 2H), 4.11 (s, 5H), 4.45
(s, 2H), 5.02 (s, 2H), 5.52 (s, 2H), 7.34 (m, 7H), 7.72 (m, 2H). 13C
NMR (100.55 MHz, acetone-d6): δ (ppm) 25.3, 53.9, 63.4, 65.1, 67.6,
67.9, 69.8, 84.6, 127.3 (two overlapping peaks), 127.8, 127.9, 129.4,
135.7, 140.0, 141.0, 142.5. EI MS: m/z calcd for C31H34BNO4Fe 551.2,
found 551.2. IR spectra (in KBr): wavenumber (cm−1) 3070, 2973,
1700, 1623. Anal. Calcd for C31H34BNO4Fe: C, 67.5; H, 6.2; N, 2.5.
Found: C, 67.4; H, 6.5; N, 2.5.
In Vitro Assays. Monitoring Release of Iron Ions from the
Prodrugs and Control Compounds in the Presence of Hydrogen
Peroxide. A solution of a ferrocene complex (10 μL, 10 mM in DMF
containing 2% water and 1% sodium ascorbate) was diluted with
MOPS buffer (990 μL, 100 mM, pH 7.5). Such probes were incubated
with H2O2 (stock solutions of different concentrations (1 μL) were
added) at 22 °C for selected time periods. Then the reaction was
quenched by adding sodium dithionite (20 μL, 1 M in water), which
Assay for Determination of Cell Permeability of the Prodrugs
(Table 2) (According to the Protocol Described in Ref 28). HL60-
932
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
cells grown in RPMI 1640 medium supplemented with 10% FCS, 1%
glutamine, and 1% penicillin/streptomycin were centrifuged, and the
medium was replaced with RPMI 1640 medium (5% FCS, 1% L-
glutamine, 1% penicillin/streptomycin) to obtain suspensions
containing 106 cells/mL. Solutions of prodrugs (5 μL, solvent DMF
containing 2% water and 1% sodium ascorbate) were added to the
suspensions (500 μL) and incubated for 1 h. The final concentration
of the prodrugs in the suspensions was 100 μM. Then the cells were
washed three times with PBS buffer (3 × 500 μL) and treated with
concentrated H2O2 solution (200 μL, 1 M) for 30 min, and all volatiles
were removed by lyophilization. Dry, lysed cells were washed with
water (200 μL), and aqueous solution obtained was acidified with HCl
(400 μL, 0.1 M). Then this solution was extracted with 2-ethyl-1,3-
hexanediol (100 μL, 10% in CHCl3, v/v), and a portion of the organic
phase obtained (40 μL) was mixed with H2SO4/CH3CO2H (400 μL,
1/1, v/v). Curcumin solution in methyl isobutyl ketone (500 μL, 2 mg
in 1 mL of the solvent) was added and allowed to react for 24 h. The
reaction was quenched by addition of water (1 mL). Light absorbances
at 550 and 780 nm of the organic phase were measured. The value
A(550 nm) − A(780 nm) was proportional to the concentration of
boron in the mixture. The amount of boron released from prodrug 1a
was taken as a reference. All other values were determined relative to
this reference.
Estimation of Oxidative Stress in Live HL-60 Cells. An aliquot of
the HL-60 cells was taken from the cultivation medium (RPMI 1640
supplemented with 10% FCS, 1% L-glutamine, and 1% penicillin/
streptomycin). The medium was replaced with PBS buffer to obtain a
cell suspension with 106 cells/mL. DCFH-DA solution (1 μL, 10 mM
in DMSO) was added to the suspension (1 mL) and incubated in the
dark chamber filled with CO2 (5%) at 37 °C for 5 min. Then the cells
were washed with PBS, and iron complexes (10 μL of stock solutions,
solvent DMF containing 2% water and 1% sodium ascorbate, final
concentration of the complex 100 μM) in Opti-MEM medium were
added. After 30 min of incubation in the dark chamber filled with CO2
(5%) at 37 °C, the cells were washed with Opti-MEM medium and left
standing at 20 °C for 4.5 h. After that the fluorescence of live cells (λex
= 488 nm, λem = 530 nm) in the suspensions was determined by using
the flow cytometer. The relative efficacy of ROS generation of the
prodrugs (1a−e) as well as of negative control 10 was calculated using
the formula efficacy = F(prodrug)/F0, where F(prodrug) is the mean
fluorescence intensity of cells incubated with a prodrug and treated as
described above and F0 is the mean fluorescence intensity of cells
which were not treated with any iron complex.
Assay for Determination of the Viability of Adherent Cells (MTT
Assay). Adherent U373 cells were grown in EMEM supplemented
with 10% FCS, 1% L-glutamine, and 1% penicillin/streptomycin to
80−90% confluence. Then the medium was removed, and the cells
were washed two times with PBS buffer, trypsinated, and resuspended
in the RPMI 1640 medium containing 1% FCS, 1% L-glutamine, and
1% penicillin/streptomycin. This suspension was spread in the wells of
a 96-well microtiter plate (∼25 000 cells per well per 100 μL) and left
standing at 37 °C in the chamber filled with CO2 (5%) for 5 h. Stock
solutions of prodrugs of different concentrations (1 μL, solvent DMF
containing 2% water and 1% sodium ascorbate) were added to the
wells and incubated for specific periods of time. Four experiments
were conducted for each concentration of the prodrug. Finally, 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; 20 μL
of the solution prepared by dissolving MTT (5 mg) in PBS buffer (1
mL) was added to each well, incubated for 3 h, treated with sodium
dodecyl sulfate (SDS) solution (90 μL, 10% solution in 0.01 M
aqueous HCl), and incubated overnight. Then the intensity of
absorbance at 590 nm was measured (MTT is converted to blue dye
(λmax = 590 nm) in live cells). The absorbance at 690 nm was taken as
a baseline value. These data were applied to calculate the relative
number of viable cells.
streptomycin) to obtain suspensions containing 106 cells/mL. Stock
solutions of prodrugs of different concentrations (5 μL, solvent DMF
containing 2% water and 1% sodium ascorbate) were added to the
cellular suspensions (500 μL) and incubated for specific periods of
time. Four experiments were conducted for each concentration of the
prodrug. At the end of the incubation period, propidium iodide (PI; 1
μL, 1.5 mM in water) was added to label leaky cells, and in 5 min the
suspension of the cells was analyzed by using flow cytometry.
ASSOCIATED CONTENT
* Supporting Information
■
S
1H and 13C NMR spectra of new prodrugs, monitoring reaction
of prodrug 1a with H2O2 by using reversed-phase TLC, release
of iron ions from prodrug 1a in the presence of different
reactive oxygen species, evaluation of oxidative stress in HL-60
cells and fibroblasts in the absence and presence of prodrug 1e,
and description of cellular assays. This material is available free
of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: Andriy.Mokhir@urz.uni-heidelberg.de. Phone: +49-
6221-548441. Fax: 49-06221-548439.
■
Author Contributions
§These authors contributed equally to this work.
ACKNOWLEDGMENTS
■
We thank Prof. Dr. Peter Comba and Dr. Bodo Martin
(Ruprecht-Karls-University of Heidelberg) for discussions of
the mechanism of oxidative decomposition of aminoferrocenes.
The funding for this project was provided by Ruprecht-Karls-
University Heidelberg and Bernthsen-Stiftung.
ABBREVIATIONS USED
■
azpy, 2-(phenylazo)pyridine; Bipy, 2,2′-bipyridine; Cp, cyclo-
pentadienyl; EDTA, N,N,N′,N′-ethylenediaminetetraacetic acid;
Fc, ferrocene; Fc+, ferrocenium; GSH, reduced glutathione;
GSSG, oxidized glutathione; HL-60, human promyelocytic
leukemia cells; HL-60cat, human promyelocytic leukemia cells
overexpressing catalase; 8-HQ, 8-hydroxyquinoline; HT29,
human colon adenocarcinoma cells; MDA-MB-231, human
breast adenocarcinoma cells; MMP, matrix metalloprotease;
MOPS, 3-(N-morpholino)propanesulfonic acid; MTT, 3-(4,5-
dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; NO-
ASA, nitric oxide-donating aspirin; QM, quinone methide;
ROS, reactive oxygen species; SW480, human colon
adenocarcinoma cells; U373, human glioblastoma-astrocytoma
cells
REFERENCES
■
(1) Jung, Y.; Lippard, S. J. Direct cellular responses to platinum-
induced DNA damage. Chem. Rev. 2007, 107, 1387−1407.
(2) Galmarini, C. M.; Mackey, J. R.; Dumontet, G. Nucleoside
analogues and nucleobases in cancer treatment. Lancet Oncol. 2002, 3,
415−424.
(3) (a) Halliwell, B. Oxidative stress and cancer: have we moved
forward? Biochem. J. 2007, 401, 1−11. (b) Engel, R. H.; Evens, A. M.
Oxidative stress and apoptosis: a new treatment paradigm in cancer.
Front. Biosci. 2006, 11, 300−312. (c) Finkel, T. Oxidant signals and
oxidative stress. Curr. Opin. Cell Biol. 2003, 15, 247−254.
(d) Schumacker, P. T. Reactive oxygen species in cancer cells: live
by the sword, die by the sword. Cancer Cell 2006, 10, 175−176.
(d) Antunes, F.; Cadenas, R. Estimation of H2O2 gradients across
biomembranes. FEBS Lett. 2000, 475, 121−126. (e) Szatrowski, T. P.;
Assay for Determination of the Viability of Nonadherent Cell
Lines (PI Assay). HL-60 cells grown in RPMI 1640 medium
supplemented with 10% FCS, 1% L-glutamine, and 1% penicillin/
streptomycin were centrifuged, and the medium was replaced with
RPMI 1640 medium (5% FCS, 1% L-glutamine, 1% penicillin/
933
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934

Journal of Medicinal Chemistry
Article
Nathan, C. F. Production of large amounts of hydrogen peroxide by
human tumor cells. Cancer Res. 1991, 51, 794−798. (f) O’Donnell-
Tormey, J.; DeBoer, C. J.; Nathan, C. F. Resistance of human tumor
cells in vitro to oxidative cytolysis. J. Clin. Invest. 1985, 76, 80−86.
(g) Stone, J. R. An assessment of proposed mechanisms for sensing
hydrogen peroxide in mammalian systems. Arch. Biochem. Biophys.
2004, 422, 119−124.
(4) Davison, K.; Mann, K. K.; Miller, W. H. Arsenic trioxide:
mechanisms of action. Semin. Hematol. 2002, 39 (Suppl. I), 3−7.
(5) (a) Dorr, R. T.; Liddil, J. D.; Soble, M. J. Cytotoxic effects of
glutathione-synthesis inhibition by L-buthionine-(S,R)-sulfoximine on
human and murine tumor cells. Invest. New Drugs 1986, 4, 305−313.
(b) Renschler, M. F. The emerging role of reactive oxygen species in
cancer therapy. Eur. J. Cancer 2004, 40, 1934−1940.
(6) Trachootham, D.; Zhou, Y.; Zhang, H.; Demizu, Y.; Chen, Z.;
Pelicano, H.; Chiao, P. J.; Achanta, G.; Arlinghaus, R. B.; Liu, J.;
Huang, P. Selective killing of oncogenically transformed cells through a
ROS-mediated mechanism by β-phenylethyl isothiocyanate. Cancer
Cell 2006, 10, 241−252.
(7) Hulsman, N.; Medema, J. P.; Bos, C.; Jongejan, A.; Leurs, R.;
Smit, M. J.; de Esch, I. J. P.; Richel, D.; Wijtmans, M. Chemical
insights in the concept of hybrid drugs: the antitumor effect of nitric
oxide-donating aspirin involves a quinone methide but not nitric oxide
nor aspirin. J. Med. Chem. 2007, 50, 2424−2431.
radical formation. J. Am. Chem. Soc. 2006, 128, 12424−12425. (b) Wei,
Y.; Guo, M. Hydrogen peroxide triggered prochelator activation,
subsequent metal chelation, and attenuation of the Fenton reaction.
Angew. Chem., Int. Ed. 2007, 46, 4722−4725. (c) Major Jourden, J. L.;
Cohen, S. M. Hydrogen peroxide activated matrix metalloproteinase
inhibitors: a prodrug approach. Angew. Chem., Int. Ed. 2010, 49, 6795−
6797.
(17) (a) Folk, D. S.; Franz, K. J. A prochelator activated by β-
secretase inhibits Aβ aggregation and suppresses copper-induced
reactive oxygen species formation. J. Am. Chem. Soc. 2010, 132, 4994−
4995. (b) Schugar, H.; Green, D. E.; Bowen, M. L.; Scott, L. E.; Storr,
T.; Bohmerle, K.; Thomas, F.; Allen, D. D.; Lockman, P. R.; Merkel,
M.; Thompson, K. H.; Orvig, C. Combating Alzheimer’s disease with
multifunctional molecules designed for metal passivation. Angew.
Chem., Int. Ed. 2007, 46, 1716−1718.
(18) McDonnell, S. O.; Hall, M. J.; Allen, L. T.; Byrne, A.; Gallagher,
W. M.; O’Shea, D. F. Supramolecular photonic therapeutic agents. J.
Am. Chem. Soc. 2005, 127, 16360−16361.
(19) Goldstein, S.; Meyerstein, D.; Czapski, G. The Fenton reagents.
Free Radical Biol. Med. 1993, 15, 435−445.
(20) Trachootham, D.; Alexandre, J.; Huang, P. Targeting cancer
cells by ROS-mediated mechanisms: a radical therapeutic approach?
Nat. Rev. Drug Discovery 2009, 8, 579−591.
(21) Zheng, N.; Armstrong, J. D.; Eng, K. K.; Keller, J.; Liu, T.;
Purick, R.; Lynch, J.; Hartner, F. W.; Volante, R. P. A convergent
asymmetric synthesis of a growth hormone secretagogue. Tetrahedron:
Asymmetry 2003, 14, 3435−3446.
(8) Magda, D.; Lepp, C.; Gerasimchuk, N. N.; Lee, I.; Sessler, J. L.;
Lin, A.; Bioglow, J. E.; Miller, R. A. Redox cycling by motexafin
gadolinium enhances cellular response to ionizing radiation by forming
reactive oxygen species. Int. J. Radiat. Oncol., Biol., Phys. 2001, 51,
1025−1036.
(22) Heinze, K.; Schlenker, M. Main chain ferrocenyl amides from 1-
amino-1′-carboxyferrocene. Eur. J. Inorg. Chem. 2004, 2974−2988.
(23) Knox, G. R.; Pauson, P. L. 903. Ferrocene derivatives. Part XI.
Azoferrocenes. J. Chem. Soc. 1961, 4615−4618.
(24) The IUPAC Stability Constants Database; SCQuery Vn 4.06/
4.045, Academic Software: Sourby Old Farm, Timble, Otley, Yorks,
LS21 2PW, UK.
(25) Britton, W. E.; Kashyap, R.; El-Hashash, M.; El-Kady, M. The
anomalous electrochemistry of the ferrocenylamines. Organometallics
1986, 5, 1029−1031.
(26) Sohn, Y. S.; Hendrickson, D. N.; Grey, H. B. Electronic
structure of metallocenes. J. Am. Chem. Soc. 1971, 93, 3603−3612.
(27) Dutta, S.; Mokhir, A. An autocatalytic, chromogenic and
fluorogenic photochemical reaction controlled by nucleic acids. Chem.
Commun. 2011, 47 (4), 1243−1245.
(28) Wimmer, M. A.; Goldbach, H. E. A miniaturized curcumin
method for the determination of boron in solutions and biological
samples. J. Plant Nutr. Soil Sci. 1999, 162, 15−18.
(29) Ma, Y.; Liu, Z.; Hider, R. C; Petrat, F. Determination of the
labile iron pool of human lymphocytes using the fluorescent probe,
CP655. Anal. Chem. Insights 2007, 2, 61−67.
(9) (a) Kopf-Maier, P.; Kopf, H.; Neuse, E. W. J. Ferricenium
̈
̈
complexes: a new type of water-soluble antitumor agent. Cancer Res.
Clin. Oncol. 1984, 108, 336−340. (b) Kopf-Maier, P.; Kopf, H. Non-
̈
̈
platinum group metal antitumor agents. History, current status, and
perspectives. Chem. Rev. 1987, 1137−1152. (c) van Staveren, D. R.;
Metzler-Nolte, N. Bioorganometallic chemistry of ferrocene. Chem.
Rev. 2004, 104, 5931−5985.
(10) Dougan, S. J.; Habtemariam, A.; McHale, S. E.; Parsons, S.;
Sadler, P. J. Catalytic organometallic anticancer complexes. Proc. Natl.
Acad. Sci. U.S.A. 2008, 105, 11628−11633.
(11) (a) Totter, J. R. Spontaneous cancer and its possible relationship
to oxygen metabolism. Proc. Natl. Acad. Sci. U.S.A. 1980, 77, 1763−
1767. (b) Beckman, K. B.; Ames, B. N. Oxidative decay of DNA. J.
Biol. Chem. 1997, 272, 19633−19636.
(12) (a) Hamels, D.; Dansette, P. M.; Hillard, E. A.; Top, S.;
Vessieres, A.; Herson, P.; Jaouen, G.; Mansuy, D. Ferrocenyl quinone
methides as strong antiproliferative agents: formation by metabolic
and chemical oxidation of ferrocenyl phenols. Angew. Chem., Int. Ed.
2009, 48, 9124−9126. (b) Jaouen, G.; Top, S.; Vessieres, A.
Organometallics targeted to specific biological sites: the development
of new therapies. In Bioorganometallics; Jaouen, G., Ed.; Wiley-VCH:
Weinheim, Germany, 2006; p 65. (c) Hillard, E. A.; Vessieres, A.;
Thouin, L.; Jaouen, G.; Amatore, C. Ferrocene-mediated proton-
coupled electron transfer in a series of ferrocifen-type breast-cancer
drug candidates. Angew. Chem., Int. Ed. 2006, 45, 285−290.
(13) (a) Khosrow, K.; Rigas, B. The mechanism of action of nitric
oxide-donating aspirin. Biochem. Biophys. Res. Commun. 2007, 358,
1096−1101.
(14) Wlassoff, W. A.; Albright, C. D.; Sivashinski, M. S.; Ivanova, A.;
Appelbaum, J. G.; Salganik, R. I. Hydrogen peroxide overproduced in
breast cancer cells can serve as an anticancer prodrug generating
apoptosis-stimulating hydroxyl radicals under the effect of tamoxifen-
ferrocene conjugate. J. Pharm. Pharmacol. 2007, 59, 1549−1553.
̀ ̀
(15) Pigeon, P.; Top, S.; Vessieres, A.; Huche, M.; Gormen, M.; El
̈
Arbi, M.; Plamont, M.-A.; McGlinchey, M. J.; Jaouen, G. A new series
of ferrocifen derivatives, bearing two aminoalkyl chains, with strong
antiproliferative effects on breast cancer cells. New J. Chem. 2011, 35,
2212−2218.
(16) (a) Charkoudian, L. K.; Pham, D. M.; Franz, K. J. A pro-chelator
triggered by hydrogen peroxide inhibits iron-promoted hydroxyl
934
dx.doi.org/10.1021/jm2014937 | J. Med. Chem. 2012, 55, 924−934