Redox properties and thiol reactivity of geldanamycin and its analogues in aqueous solutions

Article abstract of DOI:10.1021/jp304206n

Geldanamycin (GM), a benzoquinone ansamycin antibiotic, is a natural product inhibitor of Hsp90 with potent and broad anticancer properties, but with unacceptable levels of hepatotoxicity. Less toxic C17-substituted analogues have been synthesizedincluding 17-(allylamino)-17-demethoxygeldanamycin (17-AAG) and the water-soluble 17-(dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG). Redox properties and thiol reactivity are central to the therapeutic and toxicologic effects of quinones, and the question arises as whether the extent of toxicity of GM, 17-AAG, and 17-DMAG is related to their redox potentials. Using pulse radiolysis, the one-electron redox potentials (vs NHE) at pH 7.0 of GM and 17-AAG have been determined to be -62 ± 7 mV and -273 ± 8 mV, respectively, whereas a value of -194 ± 6 mV has been previously published for 17-DMAG. The rate constants of the reaction of GM and its analogues with glutathione, cysteine, or dithiothreitol under anoxia at pH 7.4 followed the order GM > 17-DMAG > 17-AAG, which correlates with the order of the redox potential of the quinone/semiquinone couple. Thus, GM reacts much faster with thiols compared to the less toxic 17-DMAG and 17-AAG, and is also expected to be more readily reduced by reductases to the respective semiquinone radical, which either decomposes to yield the respective hydroquinone or reduces oxygen to superoxide. Because both redox cycling and thiol reactivity have been associated with quinone toxicity, it is concluded that the toxicity of benzoquinone ansamycins is directly related to the redox potential of the quinone/semiquinone couple.

Full text of DOI:10.1021/jp304206n

Article
pubs.acs.org/JPCB
Redox Properties and Thiol Reactivity of Geldanamycin and its
Analogues in Aqueous Solutions
Amram Samuni† and Sara Goldstein*,†,‡
†Department of Molecular Biology, Medical School, and ‡Chemistry Institute, The Accelerator Laboratory, The Hebrew University of
Jerusalem, Jerusalem 91904, Israel
S
* Supporting Information
ABSTRACT: Geldanamycin (GM), a benzoquinone ansamycin antibiotic, is a natural prod-
uct inhibitor of Hsp90 with potent and broad anticancer properties, but with unaccept-
able levels of hepatotoxicity. Less toxic C17-substituted analogues have been synthesized
including 17-(allylamino)-17-demethoxygeldanamycin (17-AAG) and the water-
soluble 17-(dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG).
Redox properties and thiol reactivity are central to the therapeutic and toxicologic
effects of quinones, and the question arises as whether the extent of toxicity of
GM, 17-AAG, and 17-DMAG is related to their redox potentials. Using pulse
radiolysis, the one-electron redox potentials (vs NHE) at pH 7.0 of GM and 17-AAG
have been determined to be −62 7 mV and −273 8 mV, respectively, whereas a
value of −194 6 mV has been previously published for 17-DMAG. The rate constants of the reaction of GM and its
analogues with glutathione, cysteine, or dithiothreitol under anoxia at pH 7.4 followed the order GM > 17-DMAG > 17-
AAG, which correlates with the order of the redox potential of the quinone/semiquinone couple. Thus, GM reacts much
faster with thiols compared to the less toxic 17-DMAG and 17-AAG, and is also expected to be more readily reduced by
reductases to the respective semiquinone radical, which either decomposes to yield the respective hydroquinone or reduces
oxygen to superoxide. Because both redox cycling and thiol reactivity have been associated with quinone toxicity, it is
concluded that the toxicity of benzoquinone ansamycins is directly related to the redox potential of the quinone/semiquinone
couple.
synthesized differing only in their 17-substituent including
17-(allylamino)-17-demethoxygeldanamycin (17-AAG) and
the water-soluble 17-(dimethylaminoethylamino)-17-deme-
thoxygeldanamycin (17-DMAG) (Scheme 1), which have
entered clinical trials.6−9
INTRODUCTION
■
Geldanamycin (GM, Scheme 1), a benzoquinone ansamycin
antibiotic, is a natural product with potent and broad anticancer
properties. GM interferes with the action of the heat shock
Cytotoxicity associated with exposure to quinones has
generally been associated with one-electron redox cycling
and interaction with cellular nucleophiles, such as protein
and nonprotein sulfhydryls (RSH). Redox cycling of
quinones may lead to oxidative stress due to superoxide
formation (eq 1) while addition of RS− to quinone yields
thioether−hydroquinone (Michael adduct) via reaction 2,
which is also capable of redox cycling and might contribute to
quinone-mediated carcinogenicity and neurotoxicity.10−12
The role of Michael adduct formation in quinone toxicity is not
well understood. It is difficult to separate the biological effects
caused by Michael adduct formation from that of reactive
oxygen generation since redox cycling is inherent to both
quinones and thioether−hydroquinones, which form different
quinones via autoxidation and cross oxidation.10−12
Scheme 1. Structures of GM (R = CH3O), 17-AAG (R = CH2
CHCH2NH), and 17-DMAG (R = (CH3)2NCH2CH2NH)
protein 90 (Hsp90) leading to its degradation,1,2 but it has
also been suggested that its antitumor effect is mediated
through the production of reactive oxygen species as is the
case with other quinones.3,4 The progression of GM to
clinical trials was halted due to unacceptable levels of
hepatotoxicity.5 Less toxic structural analogues have been
Received: May 1, 2012
Revised: May 16, 2012
Published: May 16, 2012
© 2012 American Chemical Society
6404
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
Oxygen was removed from the system by saturating all solutions
with N2 gas.
Pulse radiolysis experiments were carried out using a 5-MeV
Varian 7715 linear accelerator (0.1−0.3 μs electron pulses, 200 mA
current). A 200 W Xe lamp produced the analyzing light. A cutoff
filter at 340 nm was used to minimize photochemistry. Measure-
ments were done in 2 cm spectrosil cells with three light passes. The
dose was determined using the thiocyanate dosimeter.23 All
experiments were carried out at room temperature.
RESULTS
■
Production of Radicals. The delivery of an electron pulse to
N2O-saturated solutions ([N2O] = 24 mM) containing GM or
17-AAG (Q) and 1 − 2 M 2-propanol yields the respective GM·−
or AAG·−(Q·−) or their protonated forms through reactions
3 − 7.
The toxicity of GM has been ascribed to its bioreduction to the
respective semiquinone radical, which reduces oxygen to
superoxide radical,13−15 and also to its relatively fast reaction
with GSH compared to 17-AAG and 17-DMAG.16,17 In vivo
toxicity studies demonstrated that GM was markedly more toxic
than 17-AAG and 17-DMAG,18−20 and the question arises as to
whether the extent of the toxicity is related to the redox
potentials of these benzoquinone ansamycins.
In contrast to aprotic organic solvents, direct determination of
one-electron redox potentials of quinones in aqueous solutions
using routine electrochemical techniques is almost impossible
because of the instability of the semiquinone radicals and the
limited solubility of quinones in water. Under these circum-
stances, the determination of the redox potential of quinones in
aqueous solution requires much more complicated nondirect
methods, mostly pulse radiolysis. Recently, the redox potential of
the water-soluble 17-DMAG has been determined to be −194
6 mV at pH 7.0 by pulse radiolysis.21 In the present study, the
redox potentials of the water insoluble GM and 17-AAG have
been determined by pulse radiolysis, and the redox potentials of
GM and its analogues have been correlated with their reactivity
toward various thiols.
γ
H2O 〰→ e− (2.6), OH (2.7), H• (0.6), H3O+ (2.6),
•
aq
H2O2 (0.72)
(3)
In eq 3, the values in parentheses are G-values representing the
yields of the species (in 10−7 M Gy1−), which are lower by about 7%
than those formed in the presence of high solute concentrations.
e− + N2O + H2O → OH + N2 + OH−
•
aq
(4)
•OH/H• + (CH3)2CHOH → (CH3)2C•OH + H2O/H2
(5)
(CH3)2C•OH + Q → Q•− + (CH3)2CO + H+
•− + H+ ⇌ QH•
(6)
(7)
Q
In some experiments N2O was replaced by 1 M acetone (reaction 8).
e− + (CH3)2CO + H+ → (CH3)2C•OH
aq
(8)
The rate constant of reaction 6 was determined under limiting
concentrations of (CH3)2C•OH to be (1.7 0.2) × 109 and
(2.2 0.1) × 109 M−1 s−1 for GM and 17-AAG, respectively
(Figure 1S), similar to that previously reported for 17-DMAG.21
The spectra of GM•− and AAG•− observed in the present study
have maximum absorption around 450 nm while the absorption of
the protonated radical at this region is significantly lower (Figure 1,
MATERIALS AND METHODS
■
Chemicals. Water for solution preparation was purified using
a Milli-Q system. All chemicals were of the highest available
grade and were used as received. GM, 17-AAG, and 17-DMAG
were purchased from LC Laboratories (Woburn, MA). Benzyl
viologen (BV2+), 1,4-naphthoquinone-2-sulfonate (NQS), glu-
tathione (GSH), dithiothreitol (DTT), and L-cysteine (CysSH)
were purchased from Sigma-Aldrich. Stock solutions of GM and
its analogues were prepared in DMSO for studying their
reactions with thiols. For pulse radiolysis study stock solutions of
17-AAG were prepared in 2-propanol since the latter was used as
an •OH scavenger forming the alcohol radical, which reduced the
quinone moiety. The solubility of GM in alcohols is very low, and
therefore, its stock solutions were prepared in DMF since its
density and the rate constant of its reaction with •OH are lower
than those of DMSO.22 The solubility of GM and 17-AAG in
aqueous phosphate buffer (pH 7) containing 1−2 M 2-propanol
was about 80 and 150 μM, respectively, which was sufficient for
the pulse radiolysis experiments. Mass flow controllers (Tylan,
Torrance, CA) were used for preparing gas mixtures of N2O and
O2. The concentration of oxygen-saturated solutions was taken
as 1.2 mM at 22 °C and 690 mm Hg.
Kinetic measurements of thiol reactions with the drugs were
carried out using an HP 8452A diode array spectrophotometer
equipped with a thermostat (HP 89075C programmable
multicell transport). All samples contained 50 μM DTPA to
suppress possible catalysis by trace metal-ion contaminants.
Figure 1. Absorption spectra of the semiquinone radicals. Solutions
were saturated with N2O and contained 40 μM 17-AAG or GM and 1 M
2-propanol at pH 1.3 ( ) or in 4 mM PB at pH 7.0 ( , ). The spectra
were corrected for the bleaching of GM and 17-AAG, which absorb at
this region.
■
●
○
6405
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
Table 1), which is typical of benzosemiquinone radicals.24,25 The
absorption at 450 nm decreases upon decreasing the pH
and O2 for (CH3)2C•OH, and AAG•− decayed via a first-order
reaction (Figure 2, traces A and B), which was accelerated upon
Table 1. Chemical Properties of the Semiquinone Radicals
Derived from GM, 17-AAG, and 17-DMAG in Aqueous
Solutions Containing 1−1.65 M 2-Propanol
ε450(Q•−), M−1 cm−1
pKa(Q•−
)
E7(Q/Q•−), mV
GM
6600
6000
7200
4.0 0.1
3.4 0.1
3.2 0.1
−62
−273
−194
17-AAG
17-DMAG
a
a
Data from ref 21.
demonstrating pKa = 4.0 0.1 and 3.4 0.1 for GM and 17-
AAG, respectively (Figure 2S).
Redox Equilibria. Upon the addition of a standard redox
couple (S) at pH 7.0, reactions 9 and 10 may take place:
(CH3)2C•OH + S → S•− + (CH3)2CO + H+
(9)
Figure 2. Effect of added oxygen on the formation and decay of Q•−
.
Q
•− + S ⇌ S•− + Q
(10)
Kinetic traces A and B obtained upon pulse-irradiation (4.45 Gy/pulse)
of solutions containing 100 μM 17-AAG in the absence and presence of
136 μM O2. Kinetic traces C and D obtained upon pulse-irradiation
(2.84 Gy/pulse) of solutions containing 40 μM GM in the absence and
presence of 136 μM O2. All solutions contained 1 or 1.65 M 2-propanol,
1 M acetone, and 12 mM PB (pH 7.0). The optical path length
was 6.2 cm.
These reactions are followed by reduction of Q and/or S
depending on the relative rate constants k6 and k9 and the relative
concentrations of Q and S. If the rate of the decomposition of
Q•− (or S•−) is accelerated due to relaxation of reaction 10 (or
−10) toward equilibrium while [S]o and [Q]o are in excess over
[(CH3)2C•OH]0, K10 can be determined from the dependence of
k
obs (eq 11) or the residual absorption of Q•− (or S•−) (eq 12) on
[S]0 and [Q]0, and E7(Q/Q•−) is calculated according to eq 13.
increasing [O2]0 or [17-AAG]0. In contrast, GM was reduced by
•−
O2 under the conditions where (CH3)2C•OH reacts mainly
kobs = k10[S]0 + k−10[Q]0
(11)
with O2 (Figure 2, traces C and D). These results demonstrate
that eq 1 is shifted to the left in the case of 17-AAG and to the
[Q]0 [S•−
]
eq
right in the case of GM, i.e., E7(17-AAG/AAG•−) < E(O2/O2
)
•−
K10
=
=
=
[S]0 [Q•−
]
eq
< E7(GM/GM•−).
In aerated solutions containing 30−80 μM GM, 1 M 2-
propanol, and 1 M acetone, the yield of GM•− was independent
of [GM]0 whereas kobs was linearly dependent on [GM]0
resulting in a slope equal to k1 = (2.0 0.1) × 108 M−1 s−1
and an intercept close to zero (Figure 3S), i.e., k1 ≫ k−1. In the
presence of 30 μM GM and upon further increasing [O2]0 up to
saturation, the yield of GM•− decreased and kobs increased.
[Q•−]0 − [Q•−
]
eq
[Q]0
[S]0
×
×
[Q•−
]
eq
[S•−
]
eq
[Q]0
[S]0
[S]0 − [S•−
]
eq
(12)
(13)
E7(S/S•−) − E7(Q/Q•−) = 0.06 log K10
•−
Under our experimental conditions the half-life of O2
formation via reaction 15 is about 10 μs. Therefore, the
determination of K−1 from the effect of [O2] on the residual yield
of Q•− is more accurate than its effect on the rate of the relaxation
of reaction −1 toward equilibrium. Figure 3 demonstrates the
linear relationship between 1/[GM•−]eq and [O2]0/[GM]0
yielding a slope/intercept equal to K−1 = (9.6 1.5) × 10−3.
Hence, E7(GM/GM•−) = −59 4 mV and k−1 = (1.9 0.1) ×
106 M−1 s−1.
•
O2/O2 as a Standard Redox Couple. E(O2/O2 −) was
initially determined to be −160 mV (vs NHE, 1.0 M O2),26,27 but
a careful re-examination of this redox couple indicates that a
more accurate value is −180 mV.28,29 (CH3)2C•OH reaction
with O2 yields the respective peroxyl radical (reaction 14, from
ref 22), which decomposes to yield O2•− in a process catalyzed by
HPO42− (reactions 15, from ref 22).
•−
In solutions containing 100 μM 17-AAG and 32−136 μM O2,
(CH3)2C•OH + O2 → (CH3)2C(OH)OO•
= 4 × 109 M−1 s−1
k14
the decay of AAG•− obeyed first-order kinetics (Figure 2B), and
k
obs was linearly dependent on [O2]0 (Figure 4). The slope and
(14)
(15)
the intercept of the line in Figure 4 equal k−1 = (8.6 0.7) × 107
M−1 s−1 and k1 = (1.0 0.6) × 107 M−1 s−1, respectively, where
the latter value is inaccurate as is this kinetic approach for the
determination of K−1. The residual absorption of AAG•− in the
presence of 32 μM O2 was 10.4% (close to zero at higher
concentrations of O2, e.g., Figure 2B) corresponding to
K−1 = 27 and E7(17-AAG/AAG•−) = −266 mV. These results
require the use of other standard redox couples for accurate
determination of E7(17-AAG/AAG•−) and verification of
E7(GM/GM•−).
2−
•−
(CH3)2C(OH)OO• + HPO → O2 + (CH3)2C = O
4
+ H2PO −
k15 = 1.1 × 107 M−1 s−1
4
Pulse-irradiation of deaerated solutions containing GM or 17-
AAG, 1−1.65 M 2-propanol, 1 M acetone, and 12 mM phosphate
buffer (PB) at pH 7.0 exhibited different kinetic traces upon the
addition of oxygen. The initial absorption of AAG•− decreased
upon increasing [O2]0 due to a competition between 17-AAG
6406
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
Figure 3. Effect of [O2]0 on the residual yield of GM•−. Solutions
contained 30 μM GM, 1 M 2-propanol, 1 M acetone, 12 mM PB (pH
7.0), and 0, 0.24, 0.72, and 1.2 mM O2. The absorption was monitored at
450 nm. The dose was 4.68 Gy/pulse and the optical path length 6.2 cm.
Figure 5. Effect of [17-AAG]0 and [BV2+]0 on the residual absorption of
BV•+. N2O-saturated solutions containing 0.5−2 mM BV2+, 15−90 μM
17-AAG, and 1−1.5 M 2-propanol in 4 mM PB (pH 7.0) were pulse-
irradiated (2.4 Gy/pulse), and the absorption was monitored at 600 nm.
The optical path length was 6.2 cm.
where the slope equals k−16 = (9.0 0.6) × 108 M−1 s−1 and the
intercept equals k16 = (9.1
5.0) × 106 M−1 s−1, which is
inaccurate. Using K16 = 48 2 derived from Figure 5 and k−16
=
(9.0 0.6) × 108 M−1 s−1 derived from Figure 5S, one calculates
k16 = (1.9 0.2) × 107 M−1 s−1, which is within the error limit of the
value derived from the intercept of Figure 5S.
NQS/NQS•− as a Standard Redox Couple. E7(NQS/
NQS•−) has been previously determined to be −60 mV against
O2 using E(O2/O2•−) = −160 mV,30 and is now corrected to
−80 mV for E(O2/O2•−) = −180 mV. The absorption changes
were monitored at 390 nm where the absorption of NQS•− (ε390
∼12 000 M−1 cm−1)24,26,31 is significantly higher than that of
GM•− (Figure 1). Under the conditions where both NQS•− and
GM•− are produced by the radiation, the decay of the absorption
at 390 nm obeyed first-order kinetics and kobs increased upon
increasing [GM]0 or [NSQ]0, which is attributed to relaxation of
reaction −17 toward equilibrium.
Figure 4. Effect of [O2]0 on the first-order decay of AAG•−. Solutions
containing 100 μM 17-AAG, 1.5 M 2-propanol, and 1 M acetone in
12 mM PB (pH 7.0) were saturated with different mixtures of N2O and
O2, and the absorption was monitored at 450 nm. The dose was 4.45
Gy/pulse and the optical path length 6.2 cm.
GM•− + NQS ⇌ NQS•− + GM
(17)
Figure 6 shows the linear dependence of kobs/[NQS]0 on
[GM]0/[NQS]0 resulting in a slope equal to k17 = (1.5 0.3) ×
107 M−1 s−1 and an intercept equal to k−17 = (2.8 0.3) × 107
M−1 s−1. Hence, K17 = 1.87 0.71 and E7(GM/GM•−) = −64
9 mV, which is in good agreement with the value determined
against O2.
Thiol Reactions with GM and Its Analogues. The rate
constants of the reactions of GM, 17-DMAG, and 17-AAG with
the tested thiols have been determined at pH 7.4 and 37 °C
under anaerobic conditions to avoid complications due to
autoxidation of the hydroquinones formed in this system.10,32
Significant changes in the UV absorption spectrum of the
quinones were observed upon addition of an excess of GSH,
CysSH, or DTT. A typical result is shown in Figure 7 for the
reaction of 40 μM GM with 0.5 mM CysSH. The decay of the
absorption obeyed pseudo-first-order kinetics, and kobs was
linearly dependent on [RSH]0 (Table 2). The apparent rate
constants derived from the slopes of the lines of kobs versus
[RSH]0 at pH 7.4, and the k2-values calculated using pKa = 8.18
and 8.75 for CysSH and GSH, respectively,33 and 9.2 for the first
dissociation of DTT,34 are summarized in Table 2.
BV2+/BV•+ as a Standard Redox Couple. E7(BV2+/BV•+) =
−374 mV, and BV•+ absorbs highly at 600 nm where the absorption
of Q•− is negligible.28
Q
•− + BV2+ ⇌ BV•+ + Q
(16)
As expected, GM was reduced by BV•+ in N2O-saturated
solutions containing 1 mM BV2+, 20−80 μM GM, 1 M 2-
propanol, and 4 mM PB (pH 7.0). The decay of BV•+ to zero
obeyed first-order kinetics, and kobs was linearly dependent on
[GM]0 yielding a slope equal to k−16 = (1.7 0.1) × 109 M−1 s−1
and an intercept close to zero, i.e., k−16 ≫ k16.
In contrast to GM, [BV•+] did not decay to zero in the
presence of 17-AAG. Typical kinetic traces are given in Figure 4S
demonstrating the decrease in [BV•+]0 due to the competition of
17-AAG with BV2+ for (CH3)2C•OH and the relaxation of reac-
tion −16 toward equilibrium. Figure 5 demonstrates the linear
relationship between [BV2+]0/[BV•+]eq − 1 and [17-AAG]0/
[BV2+]0 yielding a slope equal to K16 = 48 2, and hence E7(17-
AAG/AAG•−) = −273 mV. The linear dependence of
k
obs/[BV2+]0 on [17-AAG]0/[BV2+]0 is given in Figure 5S,
6407
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
Accordingly, the different redox potentials of GM, 17-AAG,
and 17-DMAG are due to the difference between σp(OCH3) =
−0.27,37 σp(NHCH2CHCH2), and σp(NH(CH2)2N(CH3)2),
respectively (Scheme 1). The value of σp(NHCH2CHCH2)
has not been determined, but it is assumed to be −0.61 on the
basis of σp(NHCH2CH3) = −0.61.37 Thus, E7(GM/GM•−) −
E7(17-AAG/AAG·−) ≈ 0.21 V, which is in good agreement with
the experimental values. The redox potential of 17-DMAG is
higher than that of 17-AAG due to protonation of the terminal
dimethylamine function at pH 7. This could raise the σp(NH-
(CH2)2N+H(CH3)2) above −0.61 in spite of the two carbon
“insulation” between the strongly electron-withdrawing proto-
nated terminal amine moiety and the electron-donating ring
amino substituent.
The redox potential of other analogues of GM can be
calculated using eq 18 and E7(GM/GM•−) = −0.062 V. For
example, 17-AAG is metabolized to 17-aminogeldanamycin (17-
AG), which retains Hsp90 inhibitory activity.38 Thus, E7(GM/
GM•−) − E7(17-AG/AG•−) = 0.61(σp(OCH3) − σp(NH2)), and
using σp(NH2) = −0.66,37 one calculates E7(17-AG/AG•−) =
−0.30 V.
Figure 6. Dependence of the rate of approach to equilibrium on [GM]0
and [NQS]0. N2O-saturated solutions containing 1 M 2-propanol in
4 mM PB (pH 7.0) were pulse-irradiated (4.84 Gy/pulse), and the
absorption was monitored at 390 nm.
According to the correlation reported between the pKa values
of semiquinone radicals and the one-electron redox potentials of
the respective quinones, the pKa values of DMAG•− and AAG•−
should have been higher than that of GM•− and around 5.24,25
The relatively low pKa values of 3.2 for DMAG•− and 3.4 for
AAG•− might be attributed to stabilization of the radical by an
intramolecular hydrogen-bonding of the amino-linkage.
A good correlation has also been observed between the one-
electron redox potential of many quinones in aqueous buffer and
in aprotic organic solvents excluding those with hydroxyl-
substituents.39,40 The order of the redox potential of GM and its
analogues in aqueous buffer at pH 7.0 (GM > 17-DMAG > 17-
AAG) is different than that obtained previously in DMSO (17-
DMAG > 17-AAG > GM),15 which can be also attributed to
internal hydrogen-bonding as is the case with hydroxyl-
substituents.39−41
The addition of sulfur nucleophiles to quinones yields
thioether−hydroquinone adducts as primary molecular products
(reaction 2). The subsequent chemistry of these products
involves autoxidation, cross-oxidation, disproportionation, and
free radical interactions,32 and is beyond the scope of this study.
The rate constant for the reductive addition of RS− to quinone
depends on the nucleophility of the thiol and on the
electrophilicity of the quinone ring, which increases with the
increase in the redox potential of the quinone. Therefore, it is
expected that a quinone with a more positive redox potential will
react faster with a thiol as previously demonstrated for GSH
reaction with p-benzoqinone and methylated p-benzoqui-
nones.42 The present results (Table 2) demonstrate that the
rates of the reaction with the tested thiols follow the order GM >
17-DMAG > 17-AAG and are in agreement with the order of
E7(Q/Q•−) of these drugs. The same order has been previously
reported for the reaction in aerated solutions and presence of
5 mM16,43 and 10 mM GSH,16,43 suggesting that order of the
reactivity of ansamycins toward GSH is unaffected by oxygen. A
linear correlation is found between log(k2) and E7(Q/Q•−) for all
tested thiols demonstrating that only the absolute rates vary with
the thiol (Figure 8). The k2-value for the reaction of GSH with
GM is about 4 orders of magnitude lower than that derived from
the linear correlation between log(k2) and E7(Q/Q•−) of p-
benzoqinone and methylated p-benzoquinones42 suggesting that
the rate of reaction 2 is also influenced by steric factors.
Figure 7. Reaction of GM with CysSH. The solid line corresponds to
absorption spectrum of 40 μM GM and 50 μM DTPA in 20 mM PB (pH
7.4), and the bold line is the spectrum monitored 400 s after the addition
of 0.5 mM CysSH, when spectral changes were complete. The inset
contains the kinetic trace at 305 nm, and the solid line is the fit of the
experimental data to a first-order reaction, kobs = 0.016 s−1.
DISCUSSION
■
The one-electron redox potentials of GM, 17-DMAG, and 17-
AAG have been determined to be −62 7, −194 6, and −273
8 mV, respectively, by pulse radiolysis using several known
redox couples in aqueous buffer at pH 7.0. The redox potentials
of quinones are affected by the substituents, dissociation con-
stants, and solvents.32 Substitution into the ring by electron-
donating groups reduces the one-electron redox potential of the
quinone and raises the pKa of the semiquinone radical, whereas
the opposite is expected for electron-withdrawing groups.32 In
the case of p-benzoquinone (E7(BQ/BQ•−) = 0.09 V),35 the
redox potential of substituted p-benzoquinone can be predicted
according to eq 18,36 where σp is the Hammett para substituent
constant, which reflects the substituent as an electron-donating
or -withdrawing group.37
E7(Q/Q•−)(V vs NHE) = E7(BQ/BQ•−) + 0.61
σ
∑
p
(18)
6408
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
Table 2. Rate Constants for the Reactions of Q with DTT, CysSH, and GSH (pH 7.4) and with their Deprotonated Forms under
Anoxic Conditions and 37 °C
[DTT], mM
kobs, s−1
[CysSH], mM
kobs, s−1
[GSH], mM
12.3
kobs, s−1
≈3.5 × 10−5
17-AAG
2
4
8
2.4 × 10−4
3.9 × 10−4
7.8 × 10−4
2.5
5
6.5 × 10−5
1.4 × 10−4
2.75 × 10−4
10
kapp, M−1 s−1
k2
0.098 0.005 M−1 s−1
6.3 0.3 M−1 s−1
0.0275 0.0013 M−1 s−1
≈2.8 × 10−3 M−1 s−1
0.19 0.01 M−1 s−1
≈0.065 M−1 s−1
17-DMAG
0.5
5.6 × 10−4
1.2 × 10−3
2.3 × 10−3
1.83
4.4 × 10−4
7.9 × 10−4
1.47 × 10−3
3.06
1.14 × 10−4
2.24 × 10−4
4.3 × 10−4
1
3.67
6.12
2
7.32
12.26
kapp (pH 7.4)
k2, M−1 s−1
GM
1.16 0.06
0.205 0.009
0.036 0.002
74.4 3.7
1.44 0.07
0.84 0.04
0.5
0.021
0.04
0.5
1
0.016
0.031
0.051
4
0.0093
0.0135
0.0179
0.026
1
6
2
0.084
0.165
2
8
4
12
k
app, M−1 s−1
a
41.3 2.0
(2.65 0.13) × 103
26.9 1.4
189 10
2.2 0.1
51.5 2.3
k2, M−1 s−1
a
Calculated using kapp and pKa = 9.2, 8.18, and 8.75 for DTT, CysSH, and GSH, respectively.
ASSOCIATED CONTENT
* Supporting Information
■
S
Dependence of kobs of the formation of the absorption at 450 nm
on [17-AAG]0 and [GM]0, effect of pH on the absorption of the
semiquinone radicals, dependence of kobs of GM reduction by
superoxide radical on [GM]0, kinetic traces of the formation and
decay of BV•+ in the absence and presence of 17-AAG, effect of
[17-AAG]0 and [BV2+]0 on the rate of BV•+ decay. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Phone: 972-2-6586478. Fax: 972-2-6586925. E-mail: sarag@
vms.huji.ac.il.
■
Notes
The authors declare no competing financial interest.
Figure 8. Rate constants of the addition of RS− to Q as a function of
E7(Q/Q•−).
ACKNOWLEDGMENTS
■
This work has been supported by the Israel Science Foundation
(Grant 1477).
CONCLUSIONS
■
ABBREVIATIONS
■
The one-electron redox potentials of GM and its C17-substituted
analogues follow the order GM > 17-DMAG > 17-AAG, which
correlates with the rate constants of the reaction of these
ansamycins with various thiols. Thermodynamic considerations
imply that the rate of one-electron bioreduction of the quinone
increases upon increasing its redox potential. Thus, bioreduction
of GM is more efficient compared to 17-DMAG and 17-AAG,
where the respective semiquinone radicals either decompose to
yield the respective hydroquinones or reduce oxygen to
superoxide radical. The bioreduction is expected to be the rate-
determining step, and therefore, the extent of superoxide
formation should be the highest in the case of GM when
superoxide is efficiently scavenged by biological targets. Because
both one-electron redox cycling and thiol reactivity are
associated with quinone toxicity and GM has been previously
demonstrated to be more toxic relative to 17-DMAG and 17-
AAG, it is concluded that the toxicity of benzoquinone
ansamycins is directly related to the redox potential of the
quinone/semiquinone couple.
17-AAG, 17-(allylamino)-17-demethoxygeldanamycin; BV2+,
benzyl viologen; CysSH, L-cisteine; 17-DMAG, 17-(dimethyla-
minoethylamino)-17-demethoxygeldanamycin; DMF, dimethyl-
formamide; DMSO, dimethylsulfoxide; DTPA, diethylene
triamine pentaacetic acid; DTT, dithiothreitol; GM, geldanamy-
cin; GSH, glutathione; NQS, 1,4-naphthoquinone-2-sulfonate;
PB, phosphate buffer; Q, quinone; RSH, thiol
REFERENCES
■
(1) Whitesell, L.; Shifrin, S. D.; Schwab, G.; Neckers, L. M. Cancer Res.
1992, 52, 1721−1728.
(2) Whitesell, L.; Mimnaugh, E. G.; De Costa, B.; Myers, C. E.;
Neckers, L. M. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 8324−8328.
(3) Sreedhar, A. S. M., K.; Pato, B.; Schnaider, T.; Stetak, A.; Kis-Petik,
K.; Fidy, J.; Simonics, T.; Maraz, A.; Csermely, P. J. Biol. Chem. 2003,
278, 35231−35240.
(4) Fukuyo, Y.; Inoue, M.; Nakajima, T.; Higashikubo, R.; Horikoshi,
N. T.; Hunt, C.; Usheva, A.; Freeman, M. L.; Horikoshi, N. Cancer Res.
2008, 68, 6324−6330.
6409
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410

The Journal of Physical Chemistry B
Article
(5) Supko, J. G.; Hickman, R. L.; Grever, M. R.; Malspeis, L. Cancer
Chemother. Pharmacol. 1995, 36, 305−315.
(41) Ashnagar, A.; Bruce, J. M.; Dutton, P. L.; Prince, R. C. Biochim.
Biophys. Acta 1984, 801, 351−359.
(42) Butler, J.; Hoey, B. M. Free Radical Biol. Med. 1992, 12, 337−345.
(43) Guo, W. C.; Reigan, P.; Siegel, D.; Ross, D. Drug Metab. Dispos.
2008, 36, 2050−2057.
(6) Banerji, U.; O’Donnell, A.; Scurr, M.; Pacey, S.; Stapleton, S.; Asad,
Y.; Simmons, L.; Maloney, A.; Raynaud, F.; Campbell, M.; et al. J. Clin.
Oncol. 2005, 23, 4152−4161.
(7) Pacey, S. C.; Wilson, R. H.; Walton, M.; Martin, E.; Moreno-Farre,
J.; Arkenau, H. T.; Steinfeldt, H.; Raynaud, F.; Workman, P.; Judson, I.
Mol. Cancer Ther. 2007, 6, 3332S−3332S.
(8) Gutierrez, M. E.; Kummar, S.; Gardner, E. R.; Figg, W.; Chen, X.;
Zajac-Kaye, M.; Yancey, M. A.; Ivy, P.; Conley, B.; Melillo, G.; et al. Ann.
Oncol. 2008, 19, 24−24.
(9) Lancet, J. E.; Gojo, I.; Burton, M.; Quinn, M.; Tighe, S. M.; Kersey,
K.; Zhong, Z.; Albitar, M. X.; Bhalla, K.; Hannah, A. L.; et al. Leukemia
2010, 24, 699−705.
(10) O’Brien, P. J. Chem.-Biol. Interact. 1991, 80, 1−41.
(11) Monks, T. J.; Lau, S. S. Crit. Rev. Toxicol. 1992, 22, 243−270.
(12) Song, Y.; Buettner, G. R. Free Radicals Biol. Med. 2010, 49, 919−
962.
(13) Dikalov, S.; Landmesser, U.; Harrison, D. G. J. Biol. Chem. 2002,
277, 25480−25485.
(14) Billecke, S. S.; Bender, A. T.; Kanelakis, K. C.; Murphy, P. J. M.;
Lowe, E. R.; Kamada, Y.; Pratt, W. B.; Osawa, Y. J. Biol. Chem. 2002, 277,
20504−20509.
(15) Samuni, Y.; Samuni, U.; Ishii, H.; Krishna, M. C.; Goldstein, S.;
Mitchell, J. B. Free Radicals Biol. Med. 2010, 48, 1559−1563.
(16) Cysyk, R. L.; Parker, R. J.; Barchi, J. J.; Steeg, P. S.; Hartman, N. R.;
Strong, J. M. Chem. Res. Toxicol. 2006, 19, 376−381.
(17) Lang, W.; Caldwell, G. W.; Li, J.; Leo, G. C.; Jones, W. J.; Masucci,
J. A. Drug Metab. Dispos. 2007, 35, 21−29.
(18) Schulte, T. W. N.; L., M. Cancer Chemother. Pharmacol. 1998, 42,
273−279.
(19) Behrsing, H. P.; Amin, K.; Ip, C.; Jimenez, L.; Tyson, C. A. Toxicol.
In Vitro 2005, 19, 1079−1088.
(20) Glaze, E. R.; Lambert, A. L.; Smith, A. C.; Page, J. G.; Johnson,
W. D.; McCormick, D. L.; Brown, A. P.; Levine, B. S.; Covey, J. M.;
Egorin, M. J.; et al. Cancer Chemother. Pharmacol. 2005, 56, 637−647.
(21) Goldstein, S. J. Phys. Chem. A 2011, 115, 8928−8932.
(22) Mallard, W. G.; Ross, A. B.; Helman, W. P. NIST Standard
Reference Database. 40, Version 3.0; NIST: Gaithersburg, MD, 1998.
(23) Buxton, G. V.; Stuart, C. R. J. Chem. Soc., Faraday Trans. I 1995,
91, 279−281.
(24) Patel, K. B.; Willson, R. L. J. Chem. Soc., Faraday Trans. I 1973, 69,
814−825.
(25) Rao, P. S.; Hayon, E. Biochim. Biophys. Acta 1973, 292, 516−533.
(26) Meisel, D.; Czapski, G. J. Phys. Chem. 1975, 79, 1503−1509.
(27) Ilan, Y. A.; Meisel, D.; Czapski, G. Isr. J. Chem. 1974, 12, 891−895.
(28) Wardman, P. Free Radical Res. Commun. 1991, 14, 57−67.
(29) Koppenol, W. H.; Stanbury, D. M.; Bounds, P. L. Free Radical Biol.
Med. 2010, 49, 317−322.
(30) Ilan, Y. A.; Czapski, G.; Meisel, D. Biochim. Biophys. Acta 1976,
430, 209−224.
(31) Meisel, D.; Fessenden, R. W. J. Am. Chem. Soc. 1976, 98, 7505−
7510.
(32) Brunmark, A. C. E. Free Radical Biol. Med. 1989, 4, 435−477.
(33) Stability Constants of Metal Ion Complexes, Part B: Organic Ligands;
Perrin, D. D., Ed.; Pergamon Press: Oxford, U.K., 1979.
(34) Singh, R. W.; Whitesides., G. M. J. Org. Chem. 1991, 56, 2332−
2337.
(35) Wardman, P. J. Phys. Chem. Ref. Data 1989, 18, 1637−1755.
(36) Wardman, P. Prediction and measurement of redox properties of
drugs and biomolecules. In Selective Activation of Drugs by Redox
Processes; Adams, G. E., Breccia, A., Fielden, E. M., Wardman, P., Eds.;
Plenum Press: New York, 1990; pp 11−24.
(37) Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165−195.
(38) Egorin, M. J. R.; D, M.; Wolff, J. H.; Callery, P. S.; Musser, S. M.;
Eiseman, J. L. Cancer Res. 1998, 58, 2385−2396.
(39) Wardman, P. Free Radical Res. Commun. 1990, 8, 219−229.
(40) Roginsky, V. A.; Pisarenko, L. M.; Bors, W.; Michel, C. J. Chem.
Soc., Perkin Trans. 2 1999, 871−876.
6410
dx.doi.org/10.1021/jp304206n | J. Phys. Chem. B 2012, 116, 6404−6410