Journal of Inorganic Biochemistry 140 (2014) 236–244

Contents lists available at ScienceDirect

Journal of Inorganic Biochemistry

journal homepage: www.elsevier.com/locate/jinorgbio

[Fe(CN)5(isoniazid)]3 −: An iron isoniazid complex with redox behavior

implicated in tuberculosis therapy
Eduardo Henrique Silva Sousa a,b,⁎, Francisca Gilmara de Mesquita Vieira a, Jennifer S. Butler b,
Luiz Augusto Basso c,d, Diógenes S. Santiago c,d, Izaura C.N. Diógenes a,
Luiz Gonzaga de França Lopes a,c, Peter J. Sadler b,⁎⁎
a
Laboratório de Bioinorgânica, Departamento de Química Orgânica e Inorgânica, Universidade Federal do Ceará, Cx. Postal 6021, Fortaleza 60440-900, Brazil
b
Department of Chemistry, University of Warwick, Coventry CV4 7AL, United Kingdom
c
Instituto Nacional de Ciência e Tecnologia em Tuberculose, Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, Brazil
d
Centro de Pesquisas em Biologia Molecular e Funcional, Pontifícia Universidade Católica do Rio ande do Sul, Av. Ipiranga, 6681, Tecnopuc, Prédio 92A Partenon, Porto Alegre 90619-900, RS, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: Tuberculosis has re-emerged as a worldwide threat, which has motivated the development of new drugs. The an-
Received 18 June 2014 tituberculosis complex Na3[Fe(CN)5(isoniazid)] (IQG607) in particular is of interest on account of its ability to
Received in revised form 4 August 2014 overcome resistance. IQG607 has the potential for redox-mediated-activation, in which an acylpyridine
Accepted 5 August 2014
(isonicotinoyl) radical could be generated without assistance from the mycobacterial KatG enzyme. Here, we
Available online 12 August 2014
have investigated the reactivity of IQG607 toward hydrogen peroxide and superoxide, well-known intracellular
Keywords:
oxidizing agents that could play a key role in the redox-mediated-activation of this compound. HPLC, NMR and
Isoniazid electronic spectroscopy studies showed a very fast oxidation rate for bound isoniazid, over 460-fold faster than
Metallodrugs free isoniazid oxidation. A series of EPR spin traps were used for detection of isonicotinoyl and derived radicals
Pentacyanoferrate bound to iron. This is the first report for an isonicotinoyl radical bound to a metal complex, supported by 14N
Tuberculosis and 1H hyperfine splittings for the POBN and PBN trapped radicals. POBN and PBN exhibited average hyperfine
EPR radical trapping coupling constants of aN = 15.6, aH = 2.8 and aN = 15.4, aH = 4.7, respectively, which are in close agreement
to the isonicotinoyl radical. Radical generation is thought to play a major role in the mechanism of action of iso-
niazid and this work provides strong evidence for its production within IQG607, which, along with biological and
chemical oxidation data, support a redox-mediated activation mechanism. More generally the concept of redox
activation of metallo prodrugs could be applied more widely for the design of therapeutic agents with novel
mechanisms of action.
© 2014 Elsevier Inc. All rights reserved.

1. Introduction
According to the World Health Organization (WHO), tuberculosis is
still the highest cause of death among infectious disease claiming about
1.4 millions of lives per year [1]. There is a lack of effective drugs for
treatment of tuberculosis, especially the increasing number of resistant
strains [2]. Several outbreaks of multi-drug-resistant strains (MDTR)
have been reported with over 600,000 deaths per year [1]. Some strains
are extensively drug-resistant (XTR), which makes TB an alarming
threat without borders. Another major concern is latent tuberculosis,
⁎ Correspondence to: E.H.S. Sousa, Federal University of Ceará, Department of Organic
and Inorganic Chemistry, Fortaleza 60440-900, Brazil. Tel.: +55 85 3366 9938; fax: +55
85 3366 9978.
⁎⁎ Corresponding author. Tel.: +44 2476 523818.
E-mail addresses: eduardohss@dqoi.ufc.br (E.H.S. Sousa), p.j.sadler@warwick.ac.uk
(P.J. Sadler).
found in about 1/3 of the world population, and challenging any attempt
to eradicate this disease [3].
Currently-used anti-tuberculosis (anti-TB) drugs were developed
long ago, the newest widely used drug (ethambutol) being over
40 years old. Fortunately, a major international effort is now directed
toward development of new compounds, some of which are in ad-
vanced clinical trials, with one being recently approved (bedaquiline),
but still with limited application, mainly for multidrug-resistant tuber-
culosis cases [4].
Over 60 years ago, it was observed that nicotinamide had anti-
tuberculosis activity. This prompted the investigation of a series of
structurally-related compounds, and isoniazid was identified as a highly
active and selective drug for Mtb therapy. Currently, isoniazid is a front-
line drug used against tuberculosis. This pyridinic acylhydrazide com-
pound is a prodrug, whose mechanism of action has been elucidated
during the last two decades [5–7]. Isoniazid requires an oxidative
electron-transfer process mediated by a catalase-peroxidase enzyme
KatG [8,9] (Scheme 1). This enzyme is a heme protein, which can

http://dx.doi.org/10.1016/j.jinorgbio.2014.08.002
0162-0134/© 2014 Elsevier Inc. All rights reserved.
 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244 237

(I)

(II)

Scheme 1. Isoniazid activation and the proposed IQG607 redox-mediated-activation mechanism. (I) isoniazid is enzymatically activated upon oxidation promoted by the catalase-
peroxidase KatG, giving the isonicotinoyl radical intermediate. The latter reacts with NAD+ producing the pharmacological adduct izd-NAD, which efficiently inhibits the enoyl reductase
enzyme InhA, blocking cell wall formation. Isoniazid-Mtb resistant strains mainly disable the KatG-mediated activation step, so one alternative to overcome this would be route II using a
metal complex with a redox-mediated-activation step. (II) IQG607 is an iron complex which upon reaction with biological oxidizing agents (e.g. hydrogen peroxide, superoxide) can
quickly generate iron(III) that promotes fast oxidation of bound isoniazid leading to isonicotinoyl radical production that can eventually react with NAD+ forming the izd-NAD adduct.
The generation of other radical species might also cause M. tuberculosis death and be part of the mechanism of action.

employ hydrogen peroxide or superoxide for isoniazid-catalyzed oxida- metal complexes can be rapidly oxidized by biological oxidizing agents
tion. During this oxidation a key radical intermediate (isonicotinoyl) is (e.g. hydrogen peroxide, superoxide), which is not the case for free
generated leading to the formation of active drug and other metabolites. isoniazid. Accordingly, if a redox-active metal complex and isoniazid
Isoniazid oxidation, induced either in vivo or in vitro using purified KatG, are coupled then there is a route for redox-mediated-activation within
produces a series of pyridinic metabolites, among which isonicotinic the mycobacteria (Scheme 1). Once activated without a requirement
acid and 4-pyridinecarboxyaldehyde were consistently found [10]. Dur- for KatG, this compound can potentially eliminate resistant strains
ing those studies another metabolite (NAD-IZD) was identified as a and, hopefully, show a broader spectrum of anti-bacterial activity. This
result of a reaction with NAD+. This adduct was found to be the phar- complex was shown to exhibit a KatG-independent inhibition of InhA,
macological agent that strongly inhibits an enoyl reductase enzyme the major biological target of isoniazid. Additionally, IQG607 was oxi-
called InhA. This enzyme is involved in mycolic acid biosynthesis, dized mainly by an intramolecular electron transfer process without
which is essential for proper cell wall formation and survival in KatG assistance [15]. However, as yet there are no in vitro or in vivo
Mycobacterium tuberculosis. Despite its remarkable success, isoniazid
has encountered major clinical drawbacks due to the emergence of
resistant strains. Interestingly, the major route for isoniazid resistance
relies on the disruption of the isoniazid activation step promoted by
KatG enzyme [11,12]. This problem has opened up new opportunities
for the development of anti-tuberculosis drugs.
Recently, exciting new strategies for tuberculosis therapy have been
proposed, such as using peptides targeting DevR, a protein involved in
M. tuberculosis (Mtb) latency [13], organic sulfur dioxide releasing com-
pounds [14], and metal-based compounds [15–17]. The latter strategy is
based on medicinal inorganic chemistry principles, and has become a
highly active field with several new compounds under development
for cardiovascular disorders, cancer, pathogenic diseases, among others
[18–20]. The design of [Fe(CN)5(isoniazid)]3− (IQG607) was originally
based on rational inorganic chemistry since the drug isoniazid itself
requires redox activation, which may involve iron. This compound is
currently a candidate for pre-clinical tuberculosis development in
Brazil [16,17,21,22] (Fig. 1).
IQG607 is an octahedral complex containing isoniazid bound to an
pentacyanoferrate(II) center through its pyridinic nitrogen (Fig. 1) [17,
21]. Being low-spin 3d6, the complex is expected to be relatively inert
and so stands a good chance of reaching its target site without ligand
substitution. The strategy for the design of IQG607 was originally pro-
posed by Sousa and Moreira [17] aiming to overcome isoniazid resis-
tance [15,16] by alteration of the activation pathway of isoniazid
which is a major factor in the resistance found in several clinical tuber-
culosis strains (Scheme 1). Deletion or mutation of the gene coding for
KatG-activating enzyme is correlated to drug resistance [11,12]. These
findings have motivated the development of a new strategy. Some Fig. 1. Structure of IQG607.
238 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244
data to support the hypothesis that the mechanism of activation of
IQG607 occurs through an electron-transfer-based pathway.
We have shown that IQG607 inhibits enzyme activity of both
isoniazid-sensitive wild-type (WT) InhA and I21V mutant InhA identified
in IZD-resistant (isoniazid-resistant) clinical isolates of M. tuberculosis
[21]. The IQG607 compound has shown promising anti-TB activity
against two isoniazid-resistant strains of M. tuberculosis harboring either
an inhA structural gene mutation (S94A) or an inhA promoter region mu-
tation [C(-15)T] [23]. IQG607 has shown anti-proliferative activity in
Mtb-infected macrophage and Mtb-infected mouse model [22], and dis-
rupts mycolic acid biosynthesis in isoniazid-sensitive M. tuberculosis
H37Rv strain [24]. Accordingly, IQG607 appears to be a promising candi-
date for anti-tuberculosis therapy. However, there is still a lack of detailed
studies of IQG607 stability in the presence of oxidizing agents and, partic-
ularly, its potential for generation of radicals, which might be of key im-
portance in the proposed redox-mediated-activation mechanism. Here,
we have investigated the reactivity of IQG607 toward the biological
oxidizing agents hydrogen peroxide and superoxide using HPLC, NMR
and electronic absorption spectroscopy, along with EPR and spin traps
to investigate possible production of key radicals during these reactions.
2. Experimental
2.1. Materials
Sodium nitroprusside, isoniazid, hypoxanthine, nitrotetrazolium blue
(NBT), isonicotinamide, isonicotinic acid, 4-pyridinecarboxyaldehyde,
ascorbic acid, dithiothreitol (DTT), trifluoroacetic acid (HTFA),
sodium hydroxide, α-(4-pyridyl-N-oxide)-N-tert-butylnitrone
(POBN), Na2H2edta,, Na2HPO4, NaH2PO4, and hydrogen peroxide
were purchased from Aldrich-Sigma. The radical traps 4-hydroxy-
5,5-dimethyl-2-trifluoromethylpyrroline-1-oxide(FDMPO), 5-
(diethoxyphosphoryl)-5-methyl-1-pyrroline-N-oxide (DEPMPO),
and N-tert-butyl-α-phenylnitrone (PBN) were purchased from
Enzo Life Sciences. Organic solvents are all of chormatographic
grade. Xanthine oxidase in ammonium sulfate suspension was
purchased from SERVA. Quantofix hydrogen peroxide strips were
from SEOH and IT-TK-25 cyanide water test kit from SenSafe.
Water used in the experiments and synthesis was ultra purified
using a Milli-Q system. Sodium amminepentacyanoferrato(II)
was prepared according to the procedure described by Brauer
with modifications [25,26]. Sodium (isoniazid)pentacyanoferrate(II)
(Na 3[Fe(CN)5 (isoniazid)]; IQG607) and (isonicotinamide)
pentacyanoferrate(II) (Na3[Fe(CN)5(isonicotinamide)]) were prepared
using amminepentacyanoferrato(II) and isoniazid or isonicotinamide,
respectively, as previously reported [21,27].
2.2. Reactions with hydrogen peroxide and superoxide
A stock solution of IQG607 was freshly prepared in 0.1 M phosphate
buffer pH 7.4 and immediately used for the oxidative reactions. IQG607
was reacted with hydrogen peroxide in a molar ratio of 1:0.5 up to
1:100. These reactions were stopped by addition of an excess of dithio-
threitol (DTT), followed by immediate incubation with 110 mM
dimethylsulfoxide (DMSO) for at least 1 h. This mixture was loaded
onto a C-18 reverse-phase column (250 mm × 4.6 mm, 5 μm particle
size, Sunfire from Waters) and separated by isocratic elution using
10% methanol/NaTFA (sodium trifluoroacetate) pH 3.5 and a Shimadzu
liquid chromatography (HPLC) instrument equipped with a model LC-
10 AD pump and a SPD-M20A UV-visible (UV–Vis) photodiode-array
detector with a CBM-10 AD interface. Samples of 20 μL were injected
along with standards to identify and quantify the products. Hydrogen
peroxide stock solutions were quantified using a molar absorptivity of
39.4 M-1 cm-1 at 240 nm [28], and IQG607 quantified at 436 nm by its
molar absorptivity of 4.0 × 103 M-1 cm-1 [15,17].
IQG607 was reacted with superoxide and the reaction was stopped
using DTT followed by incubation with DMSO similar to the assay de-
scribed above. Superoxide was generated using 0.3 or 0.6 mM hypoxan-
thine and 0.07 U of xanthine oxidase (1 U = one unit catalyzes the
oxidation of 1 μmol of xanthine per minute at 25 °C, pH 7.5), and
measured using NBT. Reactions without IQG607 or hypoxanthine
were carried out as controls. Electronic absorption spectroscopy
was also used to evaluate the progress of these reactions. Semi-
quantitative analysis of hydrogen peroxide using Quantofix test strips
was used to detect any hydrogen peroxide generated as a side product
of the hypoxanthine/xanthine oxidase reaction, but this was estimated
to be less than 30 μM from a 0.3 mM hypoxanthine reaction.
2.3. NMR studies
NMR experiments were carried out using 6 mM IQG607 in 40 mM
phosphate buffer pH 7.4, either with or without 6 m hydrogen peroxide,
were followed for a period of up to 2 days of incubation at ambient tem-
perature. A displacement experiment was carried out using an excess of
DMSO (110 mM) added directly into the NMR tube containing the
sample, after 2 days of reaction, and incubated for 1 h before the spec-
trum was recorded. All of these NMR measurements were recorded on
a Bruker AV111-400 spectrometer at 298 K using phosphate buffer in
D2O.
2.4. EPR measurements
All EPR spectra were recorded at ambient temperature (ca. 291 K) on
a Bruker EMX (X-band) spectrometer. A cylindrical Tm110 mode cavity
(Bruker 4103TM) was used. This cavity is particularly useful for studies
of aqueous or other samples exhibiting high dielectric loss. The samples
were transferred to spectrosil quartz tubes with an inner diameter (I.D.)
of 1.0 mm and an outer diameter (O.D.) of 1.2 mm (Wilmad Labglass)
filled with the sample and sealed with T-Blu Tack®. This tube was
placed inside a larger quartz tube (O.D. 2.0 mm) so that the sample
could be accurately and reproducibly positioned inside the resonator.
All samples were prepared in 40 mM phosphate buffer pH 7.4. The
first spectrum was recorded ca. 9 min after initiating the reactions, un-
less otherwise stated. Samples were also assayed with and without
1.5 mM Na2H2EDTA and in ultrapure Milli-Q grade water alone, along
with POBN organic radical traps to rule out any reaction promoted by
free metals. Hydrogen peroxide was added last to initiate the reactions
and a rapid color change was noted followed by a moderate release of
gas bubbles from all reactions with IQG607. The following EPR instru-
ment settings were used: modulation amplitude 2.0 G, microwave
power 0.63 mW, attenuation of 25 dB, receiver gain 1.0 × 104, sweep
gain 41.94 s, with repeated number of 5 X-scans and resolution in Y of
24.
3. Results
3.1. Iron complex promotes fast oxidation of bound isoniazid
Electronic absorption spectroscopy was used to monitor IQG607
oxidation, since the metal-to-ligand charge transfer (MLCT) band at
436 nm is sensitive to iron(II) oxidation, while isoniazid or its oxidized
products do not have bands in the visible range [29]. Attempts to quan-
tify this reaction were carried out using several molar ratios of IQG607:
H2O2, with 1:2 showing the largest spectroscopic change (Supporting
Fig. S1). Under such conditions the intensity of the MLCT band strongly
decreased reaching a constant value, which was unchanged upon
further addition of hydrogen peroxide (Supporting Fig. S1). Another at-
tempt to quantify isoniazid oxidation stoichiometry was conducted
using hexacyanoferrate(III) in basic medium (1 mM NaOH). This oxidiz-
ing agent can be used to quantify the electron transfer process via its
“d–d” transition at 416 nm (log ε = 3.01) [30], which disappears
 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244 239

A upon reduction to hexacyanoferrate(II). A quantitative titration was

carried out using isoniazid and provided evidence for an apparent
stoichiometry of 3 electrons from isoniazid (Supporting Fig. S2),
which would support a 4-electron process for IGQ607 oxidation.
Further oxidation reactions were studied using NMR. A mixture of
IQG607 and H2O2 in a 1:1 molar ratio in 40 mM phosphate buffer in
D2O gave new sharp peaks in the 1H-NMR spectrum (Fig. 2C). IQG607
gave peaks at 9.15 ppm (H-2,2′) and 7.48 ppm (H-3,3′), whereas
after the hydrogen peroxide reaction these signals were shifted to
9.07 (H-2,2′) and 7.51 ppm (H3,3′). Additionally, a reaction conducted
with a molar ratio of 1:2 (IQG607:H2O2) led to the disappearance of
both the sharp pyridinic hydrogen signals in the aromatic region of
the 1H NMR spectrum (Fig. 2C). A similar result was obtained using
the complex [Fe(CN)5(isonicotinamide)]3−, an analog of IQG607 but
lacking the redox reactive acyl hydrazide group (Supporting Fig. S3).
NMR spectra recorded 1, 3, 24 and 49 h after IQG607 oxidation were
all very similar suggesting that the reaction was complete within 1 h
(Supporting Fig. S4). Additionally, the displacement of the bound
ligand(s) using an excess of dimethylsulfoxide (DMSO) was investigat-
ed as described elsewhere [31]. DMSO-treated IQG607, which was
previously reacted with hydrogen peroxide in 1:1 molar ratio, showed
B major peaks at 8.68 ppm and 7.80 ppm, while a non hydrogen-
peroxide-treated sample of IQG607 exhibited peaks at 8.75 ppm and
7.77 ppm, before and after DMSO treatment (Supporting Figs. S5 and
S6). 1H NMR spectra were also recorded for free isoniazid (8.75 ppm
and 7.77 ppm), isonicotinic acid (8.68 ppm and 7.80 ppm) and
isonicotinamide (8.78 ppm and 7.83 ppm), and used to aid assignment
of peaks for the released ligands.
The oxidative reaction was also monitored by HPLC and stopped at
specific time points using DTT as a reducing agent followed by incuba-
tion with DMSO. The interaction of pentacyanoferrate complexes with
reverse phase columns is very weak due to the high polarity of these
species as previously described [15,17]. Therefore, the chromatographic
monitoring of the DMSO substitution reaction was performed following
separation of the released organic molecules on a C-18 column. This
strategy allowed quantification of the oxidized ligand and unreacted
isoniazid by comparison to the chromatographic peak (area) of the
standard ligands. IQG607 is very rapidly and efficiently oxidized by
H2O2 even at stoichiometric ratios with an initial reaction rate of 11.5%
C of oxidation per minute (at 1:1 mol ratio), whereas free isoniazid was
hardly oxidized even with a 100-fold excess and much longer incuba-
tion time (b 0.025% of oxidation per minute, over 460-fold slower) at
25 °C (Figs. 2A, B and 3B).

3.2. Superoxide can also promote efficient and fast oxidation of IQG607

Superoxide was enzymatically generated by using hypoxanthine

and xanthine oxidase. This superoxide production rate was measured
by the reaction of O−2 with nitroblue tetrazolium dye (NBT). This setup
generates a smaller initial concentration of oxidizing species in compar-
ison to the previous experiments with H2O2 that used a defined amount
of reagent. It was observed that IQG607 oxidation occurs with a de-
crease in intensity of the MLCT band at 436 nm, similar to the reaction
with H2O2. The rate of consumption of IQG607 was reasonably fast
even compared to H2O2, as indicated by UV–Vis spectroscopy and
HPLC with an initial reaction rate of 3% of isoniazid oxidation per minute

Fig. 2. Oxidation of IQG607 promoted by hydrogen peroxide. A. HPLC profile for reaction of
1 mM IQG607 with 2 mM of hydrogen peroxide in phosphate buffer, stopped immediately
after mixing (start), 1, 2, 5 and 10 min using DTT, followed by incubation with DMSO in
excess, 1 (isonicotinic acid), 2 (isoniazid). B. Comparison of reaction rates for 0.3 mM
IQG607 oxidation with 0.6 mM H2O2 measuring isonicotinic acid production (red squares)
or isoniazid consumption (blue triangles) followed by HPLC and iron(II) oxidation (black
circles) followed by electronic spectroscopy at 436 nm. (HPLC data were reported as nor-
malized area). C. 1H-NMR of 6 mM IQG607 without oxidation (black, bottom) and upon
oxidation using with 6 mM hydrogen peroxide (red, middle) and 12 mM hydrogen perox-
ide ((blue, top), after 49 h, in 40 mM phosphate buffer pH 7.4.
240 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244
Fig. 3. Reaction of 0.3 mM of IQG607 with superoxide. A) HPLC profile for the IQG607 re-
action at different time points, stopped using DTT followed by incubation with DMSO in
excess. (Superoxide was generated using 0.3 mM of hypoxanthine and xathine oxidase
in phosphate buffer pH 7.4), 1 (isonicotinic acid), 2 (isoniazid), *(xanthine and hypoxan-
thine peaks). B) Kinetic profile for the reaction of IQG607 or free isoniazid with hydrogen
peroxide and superoxide monitored by HPLC: red circles (0.3 mM IQG607 + 0.6 mM
H2O2), blue triangles (0.3 mM IQG607 + generated up to 0.6 mM of O− 2 ), black squares
(0.3 mM isoniazid + generated up to 0.6 mM of O− 2 ), and orange diamonds (0.3 mM
isoniazid + 0.6 mM H2O2. 0.3 mM was used hypoxanthine for superoxide generation.
(Fig. 3B). There is an efficient consumption of isoniazid from IQG607
with concominant production of isonicotinic acid as shown by the
chromatograms (Fig. 3A). On the other hand, free isoniazid was hardly
oxidized even during much longer incubation times (b 5% after 6 h).
So, these data indicated that O− 2 promotes IQG607 oxidation with a
reaction rate over 215-fold faster than for free isoniazid at 25°C.
3.3. IQG607 generates radicals based on isoniazid
The spin trap POBN was used to probe the reaction of IQG607 with a
stoichiometric amount of hydrogen peroxide. A six-line spectrum was
observed resulting from triplet-splitting of the radical signal by the
14
N and a further doublet splitting from 1H coupling, likely assignable
to an oxygen or carbon-based radical (Fig. 4A). This trapped radical
was somewhat unstable, and decayed with a half-life of 20 min
(Fig. 4B). The EPR signals exhibited some broadening and unequal signal
intensities. This trapped radical showed a mean g-value of 2.0126 and
apparent hyperfine splitting constants of aN = 15.84–15.44 G and
aH = 2.7–2.9 G. This reaction was further investigated by altering the
ratios of IQG607 and hydrogen peroxide, as well as the amount of
POBN; concentrations of 5 mM of IQG607, 5 mM of H2O2 and 20 mM
POBN were selected for further studies (Supporting Fig. S7).
Additionally, radicals trapped by PBN showed a similar EPR profile as
for POBN with apparent hyperfine splitting constants of aN = 15.2–15.6
and aH = 4.5–4.9 G (Fig. 4D). Addition of mannitol, a known hydroxyl
radical quencher, gave rise to a moderate decrease in EPR signal
(Supporting Fig. S8).
The analog of IQG607, (isonicotinamide)pentacyanoferrate(II))
([Fe(CN)5(isn)]3−), in which an amide group replaces the redox active
acyl hydrazide group (Fig. 4C) was also investigated. This control was
studied to provide evidence of the role of isoniazid in radical generation
upon reaction with H2O2. EPR spectra for [Fe(CN)5(isn)]3− did not show
any detectable radical peaks suggesting that isoniazid is essential for
generation of radical species.
Other EPR trap agents were used to further investigate this reaction.
DMPO did not give rise to any trapped radical signals even at high con-
centrations (e.g. 50 mM). On the other hand, a related trap with higher
stability, FDMPO, did give signals for trapped radicals. The EPR signal
with g = 2.0127 consisted of a triplet of 1:3:3:1 quartets also character-
istic of an oxygen- or carbon-based radical with 14N and 19F hyperfine
splittings of aN = 13.68 and aF = 2.74 G (Fig. 5A). This radical decayed
within 2 h (Fig. 5B). The radical signal was not sensitive to mannitol or
ethanol, discouraging hydroxyl radical assignment (Supporting Fig. S9).
DEPMPO was also used as a spin-trap to further investigate the nature
of these radicals. During the reaction of IQG607 with hydrogen peroxide,
DEPMPO was able to a trap radical with an 8-line EPR spectrum with
1:1.5:1.5:1 intensities, g = 2.0128 and hyperfine splitting constants of
aN = 13.7–14.0 G, aHβ = 13.5–14.1 G and aP = 47.2 G (Fig. 5C and D).
These values are a close match to those expected for trapped OH radicals
(literature values aN = 13.9–14.0 G, aHβ = 13.2–13.7 G and aP = 47.1–
47.3 G) [32], as confirmed by a simulation (Supporting Fig. S10). The
isonicotinamide complex [Fe(CN)5(isn)]3− did not give rise to an EPR sig-
nal when reacted with H2O2 in the presence of this spin trap, as observed
previously also for POBN trap, confirming that isoniazid is required for
radical generation (Supporting Fig. S11).
4. Discussion
Isoniazid is a pro-drug whose pharmacological activity relies mainly
on its activation by the catalase-peroxidase KatG enzyme. This enzyme
is a hemeprotein that can employ hydrogen peroxide, organic peroxides
and superoxide to promote oxidation of isoniazid [33–35]. Studies con-
ducted with KatG and analogous enzymes show that isoniazid oxidation
using H2O2 is neither as fast nor as quantitative as expected [36,37].
Additionally, there are data suggesting that Mn3+ could be involved in
isoniazid oxidation mediated by KatG, where a Mn3 + intermediate
would be generated from Mn2+ within KatG, promoting fast isoniazid
oxidation [38]. This is interesting experimental evidence, since our pro-
posed model for IQG607 incorporates a similar principle (Scheme 1).
The lack of more direct response to H2O2 levels has also been proposed
as one of the reasons for isoniazid insensitivity toward many types of
bacteria [10,39]. Indeed, free isoniazid is quite stable and resistant to-
ward oxidation by H2O2 at physiological pH. It is also true that under
very basic conditions isoniazid can be oxidized more rapidly even
when using only mild oxidizing agents (e.g. ferricyanide) [15]. Previous-
ly, we observed that H2O2 can oxidize IQG607 generating mainly
isonicotinic acid [15]. Here, we have investigated in more detail reac-
tions with both H2O2 and O− 2 .
A quantitative study of IQG607 oxidation using hydrogen peroxide
showed evidence for a stoichiometric reaction of 1 mol of IQG607
with 2 mol H2O2 (Supporting Fig. S1), resulting in fully oxidized isonia-
zid and iron(III). These data indicated an overall reaction involving 4
electrons donated by the complex to reduce H2O2 to H2O. Attempts to
detect superoxide species during this reaction were not successful, sug-
gesting that there is no generation of this potential 1-electron interme-
diate. Additionally, isoniazid alone was also quantitatively oxidized
 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244
Fig. 4. POBN and PBN spin traps in reaction with IQG607 and hydrogen peroxide. A) EPR signal for a mixture of 5 mM IQG607, 5 mM H2O2 with 20 mM POBN. B) Kinetic plot for EPR signal.
C) EPR signal for a mixture of 5 mM IQG607 (black dashed line) or 5 mM [Fe(CN)5(isn)]3− (red solid line) with 5 mM H2O2 and 20 mM POBN. D) EPR signal for a mixture of 5 mM IQG607
with 5 mM H2O2 and 20 mM POBN (black dashed line) or PBN (red solid line).
using hexacyanoferrate(III) (Supporting Fig. S2), the titration curves
showing a reaction of 1 mol of isoniazid with 3 mol hexacyanoferrate(III).
This mild monoelectronic oxidizing agent is a well-studied oxidant and
could replicate isoniazid oxidation within IQG607 [30], supporting a 4-
electron oxidation process for IQG607. Besides this, the lack of sharp
1
H-NMR signals for IQG607 after oxidation by 2 mol equivalent of H2O2
suggests the formation of paramagnetic Fe(III) species (Fig. 2C). A similar
behavior was observed for Na3[Fe(CN)5(isn)] oxidation using 1:1 molar
ratio (Supporting Fig. S3) and has also been described for other
pentacyanoferrate(III) complexes [40].
1
H NMR clearly showed IQG607 signals at 9.15 ppm and 7.48 ppm,
which could be distinguished from the main oxidation product with
chemical shifts at 9.07 ppm and 7.51 ppm (Fig. 2C). These data show
that a 2-electron oxidation process could still lead to an iron(II) com-
plex, while keeping a derivative pyridinic ligand bound, as evidenced
by the behavior of the 1H NMR signals over time and also the displace-
ment reaction using dmso (Supporting Figs. S5 and S6). The DMSO dis-
placement reaction provided clear evidence for isonicotinic acid
production during the reaction with major peaks at 8.68 and 7.80 ppm
that match the free ligand, while non-oxidized IQG607 under similar re-
action conditions yielded isoniazid as expected (Supporting Fig. S5).
IQG607 was kept in solution for over 49 h before the displacement
assay was conducted, and peaks for other pyridinic derivatives
appeared, most likely due to IQG607 oxidation in air.
In accordance with the NMR data, further studies using HPLC
showed that isoniazid is fully oxidized using equimolar H2O2 yielding
mainly isonicotinic acid (Fig. 2A, Supporting Fig. S6). Unfortunately, it
was not possible to use HPLC to investigate the intact complexes due
to their poor retention on reverse phase columns, so only the DMSO-
displaced ligands could be followed and quantified. However, a
241
relatively fast reaction rate for IQG607 oxidation was measured by
HPLC based on isoniazid disapearance or isonicotinic acid production
(Fig. 2B). Bound isoniazid was oxidized more than 460-fold faster than
free isoniazid. This is a remarkable enhancement in the activation of
this prodrug and could be of key importance for its activity within
M. tuberculosis as well as resistant strains.
Superoxide is another biological radical generated by cells as a de-
fense to eliminate pathogens, and could play an important role in acti-
vating IQG607. Our data showed that O− 2 promoted efficient oxidation
of IQG607 as evidenced by the rapid conversion of isoniazid to
isonicotinic acid (Fig. 3A). A kinetic plot of isoniazid consumption deter-
mined by HPLC showed that IQG607 oxidation was more than 215-fold
faster than free isoniazid (Fig. 3B). Again this result supports that
IQG607 could be activated within M. tuberculosis either by H2O2 or O− 2
species.
Our experiments suggest that IQG607 is reasonably stable in solu-
tion, and even the pyridinic product of oxidation remains bound to
iron (Fig. 2C, Supporting Fig. S4), which is an important property for
its potential therapeutical usage. Additionally, a semi-quantitative
strip test for free cyanide did not show evidence of any significant
cyanide release. These data are in agreement with the low toxicity
described for other related cyanoferrate complexes and along with our
own toxicological data for IQG607 in mice [22,41].
Isoniazid acts as a pro-drug that requires activation through an
electron-transfer reaction catalyzed by KatG. This leads to the produc-
tion of a key carbon-based acylpyridine radical (Scheme 1) [42]. Once
formed, this radical species reacts with NAD+ giving the adduct NAD-
IZD, which is associated with the major pharmacological action of
the drug [10,35]. Hence the lack of activation of isoniazid in resistant
strains leads to less radical formation, and therefore diminished
242 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244
Fig. 5. FDMPO and DEPMPO spin traps in reaction with IQG607 and hydrogen peroxide in phosphate buffer pH 7.4. A) Monitoring the reaction of IQG607 with hydrogen peroxide using FDMPO.
B) Kinetic plot for EPR signal of FDMPO over time. C) Monitoring the reaction of IQG607 with hydrogen peroxide using DEPMPO. D) Kinetic plot for EPR signal of DEPMPO over time.
pharmacological action [43]. On that basis, if IQG607 uses an enzyme-
independent activation pathway, the production of isoniazid-based
radicals would occur even in KatG-resistant strains.
Isoniazid-based radicals have been studied by several groups using
POBN and PBN spin traps [44–46]. However, the possible generation
of isonicotinoyl radicals bound to a metal complex has never been re-
ported. Here, a series of EPR spin traps was used to probe the formation
of isoniazid-based IQG607-generated radicals under oxidizing condi-
tions. POBN and PBN have been used previously for trapping free isoni-
azid radicals upon KatG activation, chemical oxidation and hepatocyte-
induced oxidation (Table 1) [44–46]. In those cases, spin-trapping
showed evidence for more than one type of radical. Our data have
clearly shown that reactions of IQG607 with hydrogen peroxide
give rise to radicals. The related isonicotinamide complex (Na3
[Fe(CN)5(isonicotinamide)]) which lacks the acylhydrazide redox-
active group did not produce any radicals, supporting an isoniazid-
dependent process. The hyperfine splittings for the POBN spin adduct
from IQG607 in reaction with H2O2 (aN = 15.84–15.44, aH = 2.7–
2.9 G), are significantly different from those expected for hydroxyl rad-
icals (aN = 14.97, aH = 1.68) [47], but close to carbon-based isoniazid
radicals (aN = 15.4, aH = 2.9) [45]. The PBN-trapped radical exhibited
a trend with hyperfine splitting constants (aN = 15.2–15.6, aH = 4.5–
4.9 G; Fig. 4), again far from those of hydroxyl radicals (aN = 15.3–
15.5, aH = 2.75) or even hydroperoxide radicals (aN = 14.8–15.0,
aH = 2.75–3.2), but closer to values reported for isonicotinoyl derived
radicals (aN = 15.0–16.5, aH = 3.3–3.6) [46]. The hyperfine coupling
constants might be expected to differ for an iron-bound trapped radical
compared to free isoniazid radicals. There are also reports that isoniazid
can produce different types of radicals including pyridyl, acylpyridine
and acylperoxopyridyl radicals, which might account for the observa-
tion of more than one species trapped by POBN and PBN [46].
DMPO is a widely-used spin trap; however, its adducts can exhibit
short life-times, and other derivatives with much longer life-times
have emerged, e.g. FDMPO. Isoniazid-based radicals have yet to be in-
vestigated using the FDMPO spin trap; however, IQG607 in reaction
with hydrogen peroxide gave rise to well-defined and stable triplet of
1:3:3:1 quartet spectrum with hyperfine splittings (aN = 13.68, aF =
2.74 G) closely matching those of hydroxyl radicals aN = 13.9 G, aF =
2.75 rather than some reported carbon-based radicals, e.g. CH3 at
aN = 14.9 G and aF = 2.05 G or CH2OH at aN = 14.75 G and aF =
2.4 G (Table 1) [48]. However, addition of the hydroxyl-radical scavenger
mannitol did not show any significant EPR-signal suppression discourag-
ing this radical assignment. Among other isoniazid-derived species,
acylperoxopyridyl might also be a match to an oxygen-bound radical.
Another radical trap used was DEPMPO, which has previously
shown useful applications, particularly for identifying oxygen-based
radicals. This radical trap exhibits three different hyperfine splittings
based on the nuclei 14N, 1H and 15P. So, which can provide a more defi-
nite assignment of some types of trapped radicals [49]. Here, IQG607 ox-
idation showed a radical with 8-lines for DEPMPO-trapped spins as a
doublet of 1:1.5:1.5:1 quartets with hyperfine couplings in close to
values reported for OH radicals, which was fitted to EPR simulations
(Table 1, Supporting Fig. S10) [32,50]. However, peroxyacylpyridine
species could emerge as well from isoniazid oxidation as described be-
fore, which bound to a metal moiety could lead to such radical trapped.
Indeed, a tertiary alkylperoxyl radical from AAPH (2,2′-azobis-2-
methylpropionamidine hydrochloride) shows hyperfine splitting con-
stants at aN = 13.7 G, aHβ = 12.9 G and aP = 46.1 G with better intensity
match of lines to IQG607 data at 1:1.77:1.77:1, while OH is 1:1.9:1.9:1
[32]. The complex [Fe(CN)5(isonicotinamide)]3− did not form radicals,
suggesting that an acylhydrazide group is required. Production of ROS
during isoniazid oxidation has been reported [45,51] and could
 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244 243

Table 1
Hyperfine coupling constants for radicals trapped during IQG607 oxidation and comparison with literature values for known radicals.

Reactant/radical
Hyperfine coupling constant (G)

Spin trap IQG607 Isoniazid(a) OH/OOH(b) CH3(c)

POBN aN = 15.44–15.84 aN = 15.4 OH aN = 15.8–16.1

aH = 2.7–2.9 aH = 2.9 aN = 14.97 aH = 2.16–2.77
aH = 1.68
PBN aN = 15.2–15.6 aN = 15–16.5 OH aN = 14.15–14.24
aH = 4.5–4.9 aH = 3.3–3.6 aN = 15.3–15.5 aH = 3.41–3.45
aH = 2.75
OOH
aN = 14.8–15.0
aH = 2.75–3.2
FDMPO aN = 13.68 – OH aN = 14.9
aF = 2.74 aN = 13.9 aF = 2.05
aF = 2.75
DEPMPO(d) aN = 13.7–14.0 – OH aN = 15.2
aH = 13.5–14.1 aN = 13.9–14 aH = 22.3
aP = 47.2 aH = 13.2–13.7 aP = 47.7
aP = 47.1–47.3
OOH
aN = 13.1–13.4
aH = 9.0–12.0
aP = 48.6–51.2
(a)
Refs [44–46].
(b)
Refs [32,47,48,50].
(c)
Refs [32,48].
(d)
DEPMPO usually gives ca. 1:1 mixture of cis–trans isomers.

therefore also occur during IQG607 oxidation. Indeed, there are reports
suggesting that besides isonicotinoyl radical production, ROS species
could also be produced and be responsible for some of the anti-
mycobacterial activity. Interestingly, the use of a series of organic radi-
cal traps (POBN, PBN, FDMPO, DEPMPO) in our work has shown that
IQG607 oxidation can produce more than one type of radical, both
carbon- and oxygen-based species, all of which involve bound isoniazid.
Enzymatic studies of the inhibition of the enoyl reductase enzyme
(InhA), the main target for isoniazid, by IQG607 have indicated direct
binding [52]. Additionally, molecular docking calculations have also
suggested that IQG607 can bind directly and reversibly to InhA [53].
These data were the only mechanistic pieces of information regarding
IQG607 action until now. However, other data point toward an alterna-
tive mechanism involving redox processes (Scheme 1). Two analogs of
isoniazid complexes, one positively charged ([Ru(NH3)5(izd)]2+, with
a low electrochemical potential, E1/2 = 356 mV vs normal hydrogen
electrode (NHE)) and another negatively-charged ([Ru(CN)5(izd)]3−,
with a high electrochemical potential, E1/2 ~ 1000 mV vs NHE) show a
different behavior toward InhA inhibition. Interestingly, the positively-
charged complex [Ru(NH3)5(izd)]2+, with lower electrochemical po-
tential than IQG607 (E1/2 = 578 mV vs NHE), did not inhibit InhA
[15]. This can be understood as a charge repulsion since the binding
site for IZD-NAD in InhA has positive pockets, where negatively charged
phosphates of izd-NAD adduct can interact. However, the negatively-
charged complex [Ru(CN)5(izd)]3 −, with a higher electrochemical
potential than IQG607, was also unable to inhibit InhA [15]. These
data suggest that direct binding may not fully explain IQG607 antituber-
culosis activity, which could be due to a combination of an achievable
electrochemical potential for redox activation, along with negative
charge as key properties.
5. Conclusions
IQG607 has emerged as a promising anti-tuberculosis candidate,
being effective against Mtb both in vitro and in a murine model [22].
This work has provided evidence of IQG607 stability including, upon ox-
idation, mediated by hydrogen peroxide. Indeed, we have shown that
hydrogen peroxide and superoxide promoted very efficient and fast
oxidation of bound isoniazid. This is a remarkable step toward activation
of isoniazid. There is no requirement for KatG to activate isoniazid in
IQG607; indeed, IQG607 oxidation is faster than isoniazid-KatG catalyzed
reaction [36,37]. Considering that Mtb resides mainly inside the macro-
phage, one of the most oxidizing environments found in living cells, a
route for redox-mediated-activation is likely to occur as proposed
(Scheme 1), which might even bring some cell selectivity for activation
as well. Hydrogen peroxide levels of at least 20 μM have been reported
[54], which could ensure enough for triggering IQG607. Furthermore, we
observed for the first time the trapped isonicotinoyl metal-complex radi-
cal produced by activation of IQG607 mediated by hydrogen peroxide.
These data support our proposal for a redox-mediated-activation
agent, IQG607, which has been found to inhibit InhA enzyme and is
active against isoniazid-resistant strains of M. tuberculosis harboring
an inhA structural gene mutation (S94A) and an inhA promoter region
mutation [C(-15)T] (Scheme 1). However, this is the first demonstration
that IQG607 can generate isoniazid-based radicals reinforcing its poten-
tial for involvement in in vivo activity.
Altogether, these data shed light on the potential mechanism of
action of IQG607 supporting a redox-mediated-activation route, which
may explain its promising in vivo activity. Accordingly, the results
reported here will aid the design of other metal-based agents, not
only for Mtb but for several other pathologies that could benefit from
this type of metal-based redox-regulated strategy.
Acknowledgments
The authors thank CAPES, CNPq, FUNCAP (no. SPU 11294757-3), the
National Institute of Science and Technology on Tuberculosis (Decit/
SCTIE/MS-MCT-CNPq-FNDTC-CAPES), the ERC (grant no. 247450) and
the Science City (AWM/ERDF) for the financial support. L G F Lopes, D.
S. Santos and L. A. Basso are research career awardees of the National
Council for Scientific and Technological Development of Brazil (CNPq).
Appendix A. supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.jinorgbio.2014.08.002.
244 E.H.S. Sousa et al. / Journal of Inorganic Biochemistry 140 (2014) 236–244
References
[1] WHO, World Health Organization Global Tuberculosis Report, 2012.
[2] S. Keshavjee, P.E. Farmer, N. Engl. J. Med. 367 (2012) 931–936.
[3] Y. Zhang, W.W. Yew, M.R. Barer, Antimicrob. Agents Chemother. 56 (2012)
2223–2230.
[4] J.C. Palomino, A. Martin, Future Microbiol 8 (2013) 1071–1080.
[5] Y. Zhang, Annu. Rev. Pharmacol. Toxicol. 45 (2005) 529–564.
[6] C. Vilcheze, W.R. Jacobs, Annu. Rev. Microbiol. 61 (2007) 35–50.
[7] O.J. Njuma, E.N. Ndontsa, D.C. Goodwin, Arch. Biochem. Biophys. 544 (2014) 27–39.
[8] L.A. Basso, R.J. Zheng, J.S. Blanchard, J. Am. Chem. Soc. 118 (1996) 11301–11302.
[9] D.A. Rozwarski, G.A. Grant, D.H.R. Barton, W.R. Jacobs, J.C. Sacchettini, Science 279
(1998) 98–102.
[10] R.A. Slayden, C.E. Barry III, Microbes Infect. 2 (2000) 659–669.
[11] S.V. Ramaswamy, R. Reich, S.J. Dou, L. Jasperse, X. Pan, A. Wanger, T. Quitugua, E.A.
Graviss, Antimicrob. Agents Chemother. 47 (2003) 1241–1250.
[12] H.J. Marttila, H. Soini, E. Eerola, E. Vyshnevskaya, B.I. Vyshnevskiy, T.F. Otten, A.V.
Vasilyef, M.K. Viljanen, Antimicrob. Agents Chemother. 42 (1998) 2443–2445.
[13] S. Dhingra, K. Kaur, N.K. Taneja, J.S. Tyagi, FEMS Microbiol. Lett. 330 (2012) 66–71.
[14] S.R. Malwal, D. Sriram, P. Yogeeswari, H. Chakrapani, Bioorg. Med. Chem. Lett. 22
(2012) 3603–3606.
[15] E.H.S. Sousa, L.A. Basso, D.S. Santos, I.C.N. Diogenes, E. Longhinotti, L.G.D. Lopes, I.D.
Moreira, J. Biol. Inorg. Chem. 17 (2012) 275–283.
[16] E.H.S. Sousa, D.L. Pontes, I.C.N. Diogenes, L.G.F. Lopes, J.S. Oliveira, L.A. Basso, D.S.
Santos, I.S. Moreira, J. Inorg. Biochem. 99 (2005) 368–375.
[17] E.H.S. Sousa, Reactivity of Inorganic Compounds that Inhibit 2-trans-enoil-ACP(CoA)
Reductase, a tuberculosis target, (PhD Thesis) Federal University of Ceara, Fortaleza,
2003.
[18] N.P.E. Barry, P.J. Sadler, Chem. Commun. 49 (2013) 5106–5131.
[19] M. Navarro, C. Gabbiani, L. Messori, D. Gambino, Drug Discov. Today 15 (2010)
1070–1078.
[20] T.W. Hambley, Dalton Trans. (2007) 4929–4937.
[21] J.S. Oliveira, E.H.S. Sousa, L.A. Basso, M. Palaci, R. Dietze, D.S. Santos, I.S. Moreira,
Chem. Commun. (2004) 312–313.
[22] V.S. Rodrigues, A. dos Santos, A.J. Santos, C.Z. Schneider, J.B. Calixto, E.H.S. Sousa, L.G.
D. Lopes, A.A. Souto, L.A. Basso, D.S. Santos, M.M. Campos, Int. J. Antimicrob. Agents
40 (2012) 182–185.
[23] L.A. Basso, C.Z. Schneider, A.J.A.B. dos Santos, A.A. dos Santos, M.M. Campos, A.A.
Souto, D.S. Santos, J. Braz. Chem. Soc. 21 (2010) 1384–1389.
[24] V.S. Rodrigues-Junior, A.A. dos Santos Junior, A.D. Villela, J.M. Belardinelli, H.R.
Morbidoni, L.A. Basso, M.M. Campos, D.S. Santos, Int. J. Antimicrob. Agents 43
(2014) 82–85.
[25] J.H. Espenson, S.G. Wolenuk, Inorg. Chem. 11 (1972) 2034–2041.
[26] G. Bauer, Handbook of Preparative Inorganic Chemistry, Academic Press, New York,
1965.
[27] H.E. Toma, J.M. Malin, Inorg. Chem. 12 (1973) 2080–2083.
[28] D.P. Nelson, L.A. Kiesow, Anal. Biochem. 49 (1972) 474–478.
[29] H.E. Toma, J.M. Malin, J. Am. Chem. Soc. 97 (1975) 288–293.
[30] J.M. Leal, B. Garcia, P.L. Domingo, Coord. Chem. Rev. 173 (1998) 79–131.
[31] H.E. Toma, J.M. Martins, E. Giesbrecht, J. Chem. Soc. Dalton Trans. (1978) 1610–1617.
[32] H. Karoui, F. Chalier, J.P. Finet, P. Tordo, Org. Biomol. Chem. 9 (2011) 2473–2480.
[33] N.L. Wengenack, H.M. Hoard, F. Rusnak, J. Am. Chem. Soc. 121 (1999) 9748–9749.
[34] R.A. Ghiladi, D.E. Cabelli, P.R.O. de Montellano, J. Am. Chem. Soc. 126 (2004)
4772–4773.
[35] B. Wiseman, X. Carpena, M. Feliz, L.J. Donald, M. Pons, I. Fita, P.C. Loewen, J. Biol.
Chem. 285 (2010) 26662–26673.
[36] J. Bodiguel, J.M. Nagy, K.A. Brown, B. Jamart-Gregoire, J. Am. Chem. Soc. 123 (2001)
3832–3833.
[37] R.S. Magliozzo, J.A. Marcinkeviciene, J. Am. Chem. Soc. 118 (1996) 11303–11304.
[38] R.F. Zabinski, J.S. Blanchard, J. Am. Chem. Soc. 119 (1997) 2331–2332.
[39] J.L. Rosner, Antimicrob. Agents Chemother. 37 (1993) 2251–2253.
[40] F.J. Wu, D.M. Kurtz, J. Am. Chem. Soc. 111 (1989) 6563–6572.
[41] P. Nielsen, B. Dresow, R. Fischer, H.C. Heinrich, Arch. Toxicol. 64 (1990) 420–422.
[42] R.I. Amos, B.S. Gourlay, C.H. Schiesser, J.A. Smith, B.F. Yates, Chem. Commun. (2008)
1695–1697.
[43] R.A. Ghiladi, K.F. Medzihradszky, F.M. Rusnak, P.R. de Montellano, J. Am. Chem. Soc.
127 (2005) 13428–13442.
[44] E. Albano, A. Tomasi, Biochem. Pharmacol. 36 (1987) 2913–2920.
[45] K. Ito, K. Yamamoto, S. Kawanishi, Biochemistry 31 (1992) 11606–11613.
[46] N.L. Wengenack, F. Rusnak, Biochemistry 40 (2001) 8990–8996.
[47] G.R. Buettner, Free Radic. Biol. Med. 3 (1987) 259–303.
[48] V.V. Khramtsov, V.A. Reznikov, L.J. Berliner, A.K. Litkin, I.A. Grigorev, T.L. Clanton,
Free Radic. Biol. Med. 30 (2001) 1099–1107.
[49] C.L. Hawkins, M.J. Davies, Biochim. Biophys. Acta Gen. Subj. 1840 (2014) 708–721.
[50] M. Mojovic, M. Vuletic, G.G. Bacic, in: Z.D. Vucinic, B, S.S. Stojilkovic, P.R. Andjus (Eds.),
Biophysics from Molecules to Brain: In Memory of Radoslav K. Andjus, 1048, Annals of
the New York Academy of Sciences, Belgrade, 2005, pp. 471–475.
[51] H.A. Shoeb, B.U. Bowman, A.C. Ottolenghi, A.J. Merola, Antimicrob. Agents
Chemother. 27 (1985) 404–407.
[52] J.S. Oliveira, E.H.S. de Sousa, O.N. de Souza, I.S. Moreira, D.S. Santos, L.A. Basso, Curr.
Pharm. Des. 12 (2006) 2409–2424.
[53] A.L. da Costa, I. Pauli, M. Dorn, E.K. Schroeder, C.G. Zhan, O.N. de Souza, J. Mol. Model.
18 (2012) 1779–1790.
[54] F.R.F. Nascimento, D. Rodriguez, E. Gomes, E.C. Fernvik, M. Russo, Braz. J. Med. Biol.
Res. 36 (2003) 1221–1226.