Multiple spectroscopic and magnetic techniques show that chloroquine induces formation of the μ-oxo dimer of ferriprotoporphyrin IX

Article abstract of DOI:10.1016/j.jinorgbio.2014.01.002

Interaction of the antimalarial chloroquine (CQ) with ferriprotoporphyrin IX, Fe(III)PPIX, was investigated in aqueous solution (pH 7.4) and as a precipitate from aqueous medium at pH 5.0. In solution, spectrophotometric titrations indicated strong association (logK_obs 13.3 ± 0.2) and a Job plot gave a stoichiometry of 1:2 CQ:Fe(III)PPIX. UV-visible absorbance and magnetic circular dichroism spectra of the complex were compared to various Fe(III)PPIX species. Close similarity to the spectra of the μ-oxo dimer, μ-[Fe(III)PPIX]₂O, was revealed. The induction of this species by CQ was confirmed by magnetic susceptibility measurements using the Evans NMR method. The observed low-magnetic moment (2.25 ± 0.02 μ_B) could only be attributed to antiferromagnetically coupled Fe(III) centers. The value was comparable to that of μ-[Fe(III)PPIX]₂O (2.0 ± 0.1 μ_B). In the solid-state, mass spectrometry confirmed the presence of CQ in the complex. Dissolution of this solid in aqueous solution (pH 7.4) resulted in a solution with a UV-visible spectrum consistent with the same 1:2 stoichiometry observed in the Job plot. Magnetic susceptibility measurements made on the solid using an Evans balance produced a magnetic moment (2.3 ± 0.1 μ_B) consistent with that in solution. Diffusion coefficients of CQ and its complex with Fe(III)PPIX were measured in aqueous solution (3.3 ± 0.3 and 0.6 ± 0.2 × 10^{- 10} m ²·s^{- 1}, respectively). The latter was used in conjunction with an empirical relationship between diffusion coefficient and molar volume to estimate the degree of aggregation. The findings suggest the formation of a 2:4 CQ:Fe(III)PPIX complex in aqueous solution at pH 7.4.

Full text of DOI:10.1016/j.jinorgbio.2014.01.002

Journal of Inorganic Biochemistry 133 (2014) 40–49
Contents lists available at ScienceDirect
Journal of Inorganic Biochemistry
journal homepage: www.elsevier.com/locate/jinorgbio
Multiple spectroscopic and magnetic techniques show that chloroquine
induces formation of the μ-oxo dimer of ferriprotoporphyrin IX
⁎
David Kuter, Stefan J. Benjamin, Timothy J. Egan
Department of Chemistry, University of Cape Town, Private Bag, Rondebosch 7701, South Africa
a r t i c l e i n f o
a b s t r a c t
Article history:
Interaction of the antimalarial chloroquine (CQ) with ferriprotoporphyrin IX, Fe(III)PPIX, was investigated in
aqueous solution (pH 7.4) and as a precipitate from aqueous medium at pH 5.0. In solution, spectrophotometric
Received 9 September 2013
Received in revised form 3 January 2014
Accepted 7 January 2014
titrations indicated strong association (logKobs 13.3
0.2) and a Job plot gave a stoichiometry of 1:2 CQ:Fe(III)
PPIX. UV–visible absorbance and magnetic circular dichroism spectra of the complex were compared to various
Fe(III)PPIX species. Close similarity to the spectra of the μ-oxo dimer, μ-[Fe(III)PPIX]2O, was revealed. The induc-
tion of this species by CQ was confirmed by magnetic susceptibility measurements using the Evans NMR method.
Available online 16 January 2014
Keywords:
Chloroquine
Ferriprotoporphyrin IX
Heme
Magnetic circular dichroism
MCD
The observed low-magnetic moment (2.25
Fe(III) centers. The value was comparable to that of μ-[Fe(III)PPIX]2O (2.0
0.02 μB) could only be attributed to antiferromagnetically coupled
0.1 μB). In the solid-state, mass spec-
trometry confirmed the presence of CQ in the complex. Dissolution of this solid in aqueous solution (pH 7.4) re-
sulted in a solution with a UV–visible spectrum consistent with the same 1:2 stoichiometry observed in the Job
plot. Magnetic susceptibility measurements made on the solid using an Evans balance produced a magnetic mo-
ment (2.3
0.1 μB) consistent with that in solution. Diffusion coefficients of CQ and its complex with Fe(III)PPIX
were measured in aqueous solution (3.3
0.3 and 0.6
0.2 × 10−10 m2·s−1, respectively). The latter was
used in conjunction with an empirical relationship between diffusion coefficient and molar volume to estimate
the degree of aggregation. The findings suggest the formation of a 2:4 CQ:Fe(III)PPIX complex in aqueous solution
at pH 7.4.
© 2014 Elsevier Inc. All rights reserved.
1. Introduction
hemozoin crystal as a result of adsorption onto the surface [10], or by
formation of a CQ–Fe(III)PPIX complex in solution. Regardless of the
Chloroquine (CQ) is the archetypal hemozoin-inhibiting antimalari-
al and is still the focus of much research despite widespread resistance
to this drug. This is because there is no evidence of a change in the
drug target. Rather, a large body of evidence has shown that parasite re-
sistance arises from transport of CQ away from the acidic digestive vac-
uole (DV), its site of action, by the Plasmodium falciparum chloroquine
resistance transporter (PfCRT) [1,2]. Indeed, incubating CQ-resistant
parasites with verapamil, an inhibitor of CQ transport by PfCRT [3], re-
stores CQ-sensitivity [4]. Furthermore, a number of other clinical anti-
malarials including some recently introduced or under development
such as piperaquine and ferroquine likely act, at least in part against
the same target and do not exhibit cross-resistance with CQ [5,6].
CQ inhibits detoxification of ferriheme (Fe(III)PPIX) produced as a
by-product of host hemoglobin degradation by the malaria parasite in
the DV [7], reported to have a pH in the range 4.8–5.2 [8]. This detoxifi-
cation process involves the conversion of Fe(III)PPIX to hemozoin, an in-
soluble crystalline solid [9]. The mechanism by which CQ inhibits
hemozoin formation is still not completely understood. There are two
prominent views, namely inhibition of the fastest growing face of the
mechanism of inhibition, the resulting free Fe(III)PPIX likely exists as a
CQ–Fe(III)PPIX complex in solution, at least within the DV where con-
centrations of CQ are expected to be high as a result of pH trapping
[11,12]. For this reason, the molecular details of the complex remain
of great interest.
The interaction of CQ and Fe(III)PPIX has been previously investigat-
ed using a broad range of experimental techniques such as UV–visible
[13,14], infrared (IR) [15], Raman [16,17], EXAFS (extended X-ray
absorption fine structure) [18], NMR [19–21] and Mössbauer [22] spec-
troscopies as well as mass spectrometry [23] and isothermal titration
calorimetry (ITC) [24,25]. These investigations have been conducted in
both the solid- and solution-state, often spanning wide pH ranges and
incorporating various solvents such as methanol and DMSO or deter-
gents. The combination of varying experimental conditions between
studies as well as reliance in some investigations on only a single exper-
imental technique, has led to conflicting conclusions. For instance,
reported binding stoichiometries of the CQ–Fe(III)PPIX complex in solu-
tion have ranged from 1:1 to 1:8 [13,19,24,26]. Further controversy has
revolved around the identity of the Fe(III)PPIX species in the CQ–Fe(III)
PPIX complex. Some studies have concluded that Fe(III)PPIX is a mono-
mer while others have suggested that it is a μ-oxo dimer, μ-[Fe(III)
PPIX]2O [27]. These investigations, however, have been complicated
⁎
Corresponding author. Tel.: +27 21 650 2528; fax: +27 21 650 5195.
E-mail address: timothy.egan@uct.ac.za (T.J. Egan).
0162-0134/$ – see front matter © 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jinorgbio.2014.01.002

D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
41
by the fact that speciation of Fe(III)PPIX in solution is greatly influenced
by the solvent and pH [28,29]. Additional complication has arisen from
reports that the Fe(III)PPIX species is not in the μ-[Fe(III)PPIX]2O form
when the CQ–Fe(III)PPIX complex is precipitated at more biologically
relevant pH (~5) [15]. These authors have suggested that changes to
Fe(III)PPIX speciation in the solid-state could occur during the drying
process. Thus there is much confusion regarding the interaction of CQ
with Fe(III)PPIX in both solution and solid-state.
The present study comprehensively investigates the nature of the
CQ–Fe(III)PPIX complex. This has been approached using multiple ex-
perimental techniques in purely aqueous solution in order to avoid
solvent-induced changes to Fe(III)PPIX speciation and to better mimic
biological conditions. Interaction of CQ with Fe(III)PPIX was investigat-
ed in solution at pH 7.4 and in the solid-state as a precipitate obtained at
pH 5.0. The species of Fe(III)PPIX and the CQ:Fe(III)PPIX stoichiometry
of the complex in both solution and solid-state have been identified
using UV–visible and magnetic circular dichroism (MCD) spectroscopy,
as well as by magnetic susceptibility measurements using the Evans
NMR method and an Evans balance. We further report the association
constant and diffusion coefficient for the complex in aqueous solution.
The latter was used to estimate the extent of aggregation of the com-
plex. We conclude that the complex formed in aqueous solution at
pH 7.4 is identical to the precipitate obtained from aqueous solution
at pH 5.0 and that Fe(III)PPIX exists as μ-[Fe(III)PPIX]2O in both, with
a stoichiometry of 1:2 CQ:Fe(III)PPIX.
discarded and the precipitate washed by resuspension in water. This
suspension was centrifuged, the supernatant discarded and washing re-
peated once more. After the final washing, the precipitate was left to dry
in a desiccator over phosphorous pentoxide. The solid precipitate in the
presence of CQ was obtained following the same procedure except solid
CQ (free base) was added to the initial hematin solution to give a con-
centration of 58.5 mM. CQ was used in its free base form in order to
eliminate the possible presence of phosphate in the solid which can ob-
scure peaks in the IR spectrum of the CQ–Fe(III)PPIX complex. To obtain
a solid sample of μ-[Fe(III)PPIX]2O a tetrasodium salt was prepared ac-
cording to a procedure modified from Brown et al. [30]. Briefly, this in-
volved dissolving hematin (100 mg) in DMSO (2 mL) and adding 5 M
NaOH (0.5 mL) with warming. The solution was then cooled and a
large excess of acetone added to produce a precipitate. This was filtered,
washed with acetone, and air dried and finally traces of moisture were
removed at 100 °C. The high pH form of HO-Fe(III)PPIX was obtained
by lyophilization of an aqueous solution containing hemin (23 mM)
and NaOH (0.07 M). This produces Na2[HO-Fe(III)PPIX] and NaCl.
2.3. Spectrophotometric titrations
Titrations were conducted in aqueous solution buffered to pH 7.4
using 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES,
0.02 M). A 50 μL Hamilton syringe was used to make additions to the
working solution. Titrations were performed by adding aliquots of an
aqueous Fe(III)PPIX solution (1 mM in 0.02 M HEPES) to an aqueous
CQ working solution (0.02 mM in 0.02 M HEPES). The Fe(III)PPIX solu-
tion was made by diluting a hemin stock solution (20 mM in 0.1 M
NaOH) with 0.02 M aq. HEPES. The CQ working solution was made by
diluting a CQ stock solution (0.08 mM in 0.02 M HEPES) in 0.02 M aq.
HEPES. Spectrophotometric titration data were analyzed using the
HypSpec software package (Hyperquad) [31] according to a 1:2 CQ:
Fe(III)PPIX model as determined from Job plot experiments (see
Results).
2. Materials and methods
2.1. Materials and instrumentation
All materials were purchased from Sigma-Aldrich, with the excep-
tion of hemin (Cl-Fe(III)PPIX, Fluka) and D2O (Merck). Reagents were
of analytical grade and were used without further purification. Unless
otherwise stated, the CQ used was the diphosphate salt. The free base
form of CQ was prepared from the diphosphate salt by adding 2 M
NaOH to an aqueous CQ solution (0.2 M) and extracting the resultant
free base into dichloromethane. The organic layer was separated,
dried over MgSO4 and removed under vacuum. The resulting oil was
dried over phosphorous pentoxide, washed with a small quantity of
diethyl ether and dried once more under vacuum. This afforded CQ
free base as a white solid. All Fe(III)PPIX solutions were freshly prepared
and stored in the dark until use. To avoid buildup of adsorbed Fe(III)
PPIX, all apparatus that came into contact with Fe(III)PPIX were scrupu-
lously washed as previously described [28].
2.4. CQ–Fe(III)PPIX binding stoichiometry
Job plots were conducted in triplicate by varying the mole fraction of
Fe(III)PPIX and CQ in the same aqueous solvent system used for spectro-
photometric titrations, subject to the constraint that the total additive
concentration of Fe(III)PPIX and CQ remained 0.2 mM. The stock
Fe(III)PPIX solution (2 mM in 0.015 M NaOH) and CQ solution (2 mM
in 0.2 M HEPES) were used to make fifteen working solutions with
Fe(III)PPIX mole fractions of 1.00, 0.95, 0.90, 0.85, 0.80, 0.75, 0.70,
0.65, 0.60, 0.50, 0.40, 0.30, 0.20, 0.10 and 0.00. The working solutions
consisted of (i) a combined volume of CQ and Fe(III)PPIX solutions of
0.1 mL (individual volumes based on the mole fraction required); (ii)
an aliquot of HEPES buffer (0.2 M, pH 7.4) equal to the difference in
volume between 0.1 mL and the volume of CQ solution added; (iii) an
aliquot of NaOH (0.015 M) equal to the difference in volume between
0.1 mL and the volume of Fe(III)PPIX solution added; and (iv) 0.8 mL
water. Solutions gave a measured pH in the range 7.5–7.6. The UV–
visible spectrum of each solution was recorded in a 0.1 cm path-
UV–visible spectra were recorded on a Varian Cary 100 or Shimadzu
UV1800 UV–visible spectrophotometer. Spectrophotometric titrations
and Job plots were performed in quartz cuvettes with 1 cm and 0.1 cm
pathlengths, respectively. These were conducted at temperatures of
25.0 or 30.0
0.2 °C maintained with a thermostated water bath. MCD
spectra were recorded at room temperature on a Chirascan-Plus CD spec-
trophotometer (operating wavelength range 165–1100 nm) using a
MCD accessory calibrated at 0.977 T. IR spectra were recorded on a Perkin
Elmer Spectrum 100 FT-IR spectrophotometer using an ATR attachment.
NMR measurements were performed on a Bruker Ultrashield 400 Plus
NMR spectrometer at 30 °C. Mass spectrometry was performed on an
Agilent 6530 QTOF LCMS system coupled with an Agilent 290 Infinity
UHPLC equipped with an Eclipse Plus C18 column (50 × 2.1 mm,
1.8 μm). The pH of solutions was measured using a Crison MicropH
2000 or Jenway 3510 pH meter.
length quartz cuvette at 25.0 or 30.0
0.2 °C and analyzed at 370 nm.
2.5. MCD spectra
The MCD spectrum of the (imidazole)2-Fe(III)PPIX complex was re-
corded in 5.64 M aq. DMSO (0.02 M HEPES, pH 7.4). Imidazole stock
(0.2 M, pH 7.4) and hemin stock (1 mM in DMSO) solutions were dilut-
ed to 0.02 M and 10 μM, respectively, using aq. HEPES (0.02 M, pH 7.4).
The spectrum of π–π dimeric Fe(III)PPIX was recorded using a solution
made by diluting a stock hemin solution (1 mM in 0.1 M NaOH) with
HEPES (0.02 M, pH 7.4), giving a final concentration of π–π dimeric
Fe(III)PPIX of 7.5 μM dimer units. To obtain the CQ–Fe(III)PPIX spec-
trum at pH 7.4, a CQ diphosphate stock solution (1 mM in water) was
2.2. Preparation of Fe(III)PPIX and CQ–Fe(III)PPIX species
The solid precipitate of Fe(III)PPIX was obtained by slowly adding
50 μL aliquots of acetic acid (1.75 M) to a hematin solution (11.5 mM
prepared in 0.1 M NaOH) until a measured pH of 5.0 was obtained. The
precipitate was centrifuged at 4000 rpm for 20 min, the supernatant

42
D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
added to the same Fe(III)PPIX solution that was used to record the π–π
dimer spectrum at pH 7.4. The final concentration of CQ in this solution
was 15 μM. The spectrum of μ-[Fe(III)PPIX]2O was reported previously
[32].
a HEPES solution (0.02 M, pH 7.4). The diffusion coefficient was calcu-
lated using Eq. (2):
ꢀ
ꢁ
Ch 2 π
D ¼
ð2Þ
2.6. Magnetic susceptibility
C0
t
The magnetic susceptibility of Fe(III)PPIX species in solution was de-
termined using the Evans NMR method as previously described [32].
The solution used to determine the magnetic susceptibility of μ-[Fe(III)
PPIX]2O was made by dissolving hematin (3 mg) in 1.20 mL NaOD
(0.3 M in D2O), to which 0.8 mL d6-DMSO was added. The reference so-
lution was prepared in the same way except no Fe(III)PPIX was added.
The value was reported previously in the Supporting Information of ref-
erence [32]. The magnetic susceptibility of Fe(III)PPIX in aqueous solu-
tion (π–π dimer) was determined for a solution consisting of 1.89 mL
hematin solution (3 mg in 0.01 M NaOD) and 0.11 mL phosphoric
acid (0.1 M in D2O), the measured pD of which was 7.2. D2O was used
as the reference solvent. To determine the magnetic susceptibility of
the CQ–Fe(III)PPIX complex, a solution consisting of 0.525 mL hematin
(3 mg in 0.1 M NaOD) and 1.475 mL CQ (12.2 mg in D2O), was pre-
pared. The reference solution was made by adding 0.49 mL NaOD
(0.1 M) to 1.51 mL of the same CQ solution used to prepare the CQ–
Fe(III)PPIX complex. Both solutions had a measured pD of 7.5.
Magnetic susceptibilities of solid Fe(III)PPIX samples were deter-
mined using a homemade Evans balance in which the change in mass
of a rare earth magnet was related to the magnetic susceptibility of
the sample using Eq. (1):
where C is the concentration in the upper chamber, C0 is the concentra-
tion in the lower chamber, h is the height (m) of the upper chamber and
t is the time (s) over which the diffusion took place. The concentration
in the upper chamber was determined by measuring the absorbance
at 342 (CQ) or 385 nm (CQ–Fe(III)PPIX) using extinction coefficients
obtained from Beer's Law plots of the working solution (18,560 for CQ
and 42,760 M−1 cm−1 per Fe(III)PPIX for the complex). A total of four
replicates were recorded for CQ at 297 K, while eight replicates were re-
corded for the CQ–Fe(III)PPIX complex, four at 297 K and the remaining
four at 295 K. Temperature and viscosity corrections were made using
the Stokes–Einstein equation in order to normalize diffusion coefficients
to 298 K.
3. Results
3.1. Stoichiometry of the CQ–Fe(III)PPIX complex
The interaction of Fe(III)PPIX and CQ was investigated at pH 7.4 and
5.0. At the former pH, Fe(III)PPIX is soluble and thus a Job plot was con-
structed in order to obtain the binding stoichiometry of the CQ–Fe(III)
PPIX complex. In aqueous solution (in the absence of CQ), Fe(III)PPIX
has been shown to exist in an equilibrium between the monomeric
and π–π dimeric forms. This equilibrium is described by the conditional
CbalðR0−R1ÞL
χg
¼
ð1Þ
m ꢀ 109
dimerization constant (logKD = 6.82
0.06 at pH 7.4) previously re-
ported by de Villiers et al. [28]. To eliminate possible complications
caused by changes in the dimerization state of Fe(III)PPIX, concentra-
tions were chosen at which the π–π dimeric form is dominant. The low-
est concentration of Fe(III)PPIX used was 20 μM at which 94% is
dimerized (with at least 97% dimerized from the third lowest concen-
tration onwards). As seen from the Job plot in Fig. 1a, at 25 °C a binding
stoichiometry corresponding to two equivalents of Fe(III)PPIX per CQ
was observed. The same stoichiometry was obtained at 30.0 °C.
This ratio was used in conjunction with spectrophotometric titration
data to determine the association constant, Kobs, of CQ with Fe(III)PPIX.
Meaningful data could only be obtained when Fe(III)PPIX was titrated
into a solution of CQ. Data were fitted to the whole spectrum (300–
800 nm) using the HypSpec software package [31]. The reverse titration
proved mathematically and experimentally intractable. This observa-
tion is probably in part related to a previous report by Crespo et al.
showing that titration of CQ into a Fe(III)PPIX solution produced differ-
ent results to that obtained when the reverse titration was performed
[14]. The origin of this phenomenon is not understood. In this investiga-
tion, such a titration produced results that could not be fitted to the ob-
served stoichiometry obtained from the Job plot.
where χg is the mass magnetic susceptibility (cm3 g−1), Cbal is the cali-
bration constant of the balance, R0 is the mass of the magnet (g) when
no sample is present, R1 is the mass of the magnet when the sample is
present, L is the length of the sample in the tube (cm) and m is
the mass of the sample (g) [33]. The calibration was determined using
the magnetic susceptibility constant of a standard, mercury(II)
tetrathiocyanatocobaltate(II), as reported by Figgis and Nyholm
(χg = 16.44 × 10−6 cm3 g−1) [34]. The diamagnetic contributions of
Fe(III)PPIX and CQ were corrected using values of −5.09 × 10−7 and
−4.64 × 10−7 cm3 g−1. The former value was determined as de-
scribed previously [32], while the latter was calculated using Pascal's
constants [35]. The molar magnetic susceptibility (χm) was calculat-
ed by multiplying the corrected χg value by the molecular weight of
the sample. Each measurement was repeated five times at room
temperature.
2.7. Diffusion coefficients
The diffusion coefficients of CQ and the CQ–Fe(III)PPIX complex in
aqueous solution (pH 7.4) were determined using the diffusion cell
method first reported by Linder et al. [36], following a protocol previ-
ously described [28]. For the measurement of CQ diffusion, a CQ work-
ing solution (0.6 mM in 0.02 M HEPES, pH 7.4) was prepared by
diluting a stock solution (6 mM in 0.2 M HEPES, pH 7.4) ten-fold with
water. In the case of CQ–Fe(III)PPIX diffusion, a working solution was
prepared by combining 2.5 mL of a hematin stock solution (3 mM in
0.1 M NaOH) and 2.5 mL of the same CQ stock solution used to deter-
mine CQ diffusion, and diluting with water to 25 mL. The final concen-
trations of CQ and Fe(III)PPIX in this working solution were 0.6 mM and
0.3 mM, respectively. Working solutions were loaded into the lower
chamber of the diffusion cell and allowed to diffuse for 1 h (CQ)
or 2 h (CQ–Fe(III)PPIX) into the upper chamber which contained
To account for known processes occurring in solution, both forma-
tion of the complex (CQ⋅M2) and dimerization of monomeric Fe(III)
PPIX need to be considered according to Eqs. (3) to (6):
2M K⇌D M2
ð3Þ
½M2ꢁ
KD
¼
ð4Þ
ð5Þ
2
½Mꢁ
2M þ CQ K⇌obs CQ ꢂ M2

D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
43
{6.82
0.06}), n = 12. This value was in good agreement with that ob-
tained with the more complex model, but exhibited a smaller error.
Investigation of CQ–Fe(III)PPIX solution interactions at pH 5.0 is
hampered by the insolubility of Fe(III)PPIX. Rapid precipitation occurs
at this pH and so neither a Job plot nor spectrophotometric titration
could be performed. While precipitation of Fe(III)PPIX in the presence
of CQ was also observed at this pH, there was a detectible concentration
still present in the solution, indicating that the CQ–Fe(III)PPIX complex
is more soluble than Fe(III)PPIX alone. This finding supports of the
hypothesis of Ursos et al. that antimalarial drugs influence the pH
dependent solubility of Fe(III)PPIX [37]. The presence of CQ in the pre-
cipitate obtained was confirmed using mass spectrometry where a mo-
lecular ion was observed at 320.19 m/z which corresponds to CQH+
(320.88 g·mol−1). No corresponding peak was observed in the spec-
trum recorded for Fe(III)PPIX precipitated at pH 5.0 in the absence of
CQ. This showed that the precipitate obtained at pH 5.0 is a CQ–Fe(III)
PPIX complex and not simply Fe(III)PPIX alone. The CQ:Fe(III)PPIX stoi-
chiometry in the solid was probed by dissolving the precipitate in aque-
ous buffer (pH 7.4) and measuring the ratio of UV–visible absorbances
at 340 and 400 nm (A340/A400) and 370 and 400 nm (A370/A400), re-
spectively. These absorbance values were selected because they show
a large response at low and high XFe(III)PPIX, respectively. Values deter-
mined for the redissolved solid (1.05 and 0.96 for A340/A400 and A370
/
A400 respectively) were then compared to corresponding ratios obtain-
ed from spectra used to construct the Job plot at pH 7.4. From Fig. 1b, it
can be seen that the ratios most closely correspond to those expected
for a stoichiometry of 1:2 CQ:Fe(III)PPIX.
Fig. 1. (a) A typical Job plot obtained for CQ and Fe(III)PPIX in aqueous solution, pH 7.4 and
25 °C (circles). Dashes represent the expected line for dilution of Fe(III)PPIX (Beer's Law).
3.2. Identification of the Fe(III)PPIX species in the CQ–Fe(III)PPIX complex
The intersection of the two solid lines is at mole fraction XFe(III)PPIX = 0.68
0.02
(n = 5), indicating a 1:2 CQ:Fe(III)PPIX binding stoichiometry. (b) Absorbance ratios cal-
culated from spectra used to construct the Job plot shown as a function of XFe(III)PPIX. Open
and filled circles represent the ratios A340/A400 and A370/A400 respectively. The horizontal
gray line (A370/A400 = 0.96) and dashed black line (A340/A400 = 1.05) represent the ab-
sorbance ratios determined from the spectrum of the CQ–Fe(III)PPIX precipitate when dis-
solved in aqueous solution and refer to the right and left y-axes, respectively. The arrow
indicates the stoichiometry that most closely corresponds these absorbance ratios, namely
1:2 CQ:Fe(III)PPIX.
The spectrophotometric titration data were best fitted using the sim-
plified model presented in Eqs. (7) to (9). By contrast to previous stud-
ies [13,14], the whole spectrum was used to obtain the association
constant rather than single wavelengths. The parameters obtained
from this fitting procedure include not only the association constant
but also the predicted spectra of the unbound reactants and bound
product. It should be noted that the three fitted spectra in Fig. 2a–c,
amount to average spectra that, when added together at ratios deter-
mined by the optimized value of K′, are able to reproduce the observed
spectroscopic envelope at each point in the titration within experimen-
tal error. Furthermore, while the spectra in Fig. 2b and c are directly ob-
served at the beginning and end of the titration, respectively, that
shown in Fig. 2a cannot be seen at any point in the titration. As shown
in Fig. 2b and a, best-fit spectra of the unbound species corresponded
closely to the independently recorded spectra of CQ and the π–π
dimer of Fe(III)PPIX, respectively. The fitted spectrum of the product is
in agreement with the experimental spectrum of Fe(III)PPIX in the pres-
ence of excess CQ (see Fig. 2c). Comparison of the fitted spectrum of the
product with that of the π–π Fe(III)PPIX dimer shows poor agreement,
while there is much closer agreement with that of μ-[Fe(III)PPIX]2O al-
beit with an apparent 20% hypochromism (see Fig. 2d). The fitted spec-
trum of the CQ complex was identical to that obtained using the more
complex model given by Eqs. (3) to (6).
Recently, we reported the use of MCD spectroscopy as a useful tool
for the identification of free Fe(III)PPIX species in solution [32]. MCD
provides better resolution of underlying bands because it contains
both positive and negative Gaussian peaks as well as Gaussian first
derivative-shaped peaks. On the other hand, bands making up peaks
in UV–visible spectra consist only of positive Gaussian-shaped features.
In order to obtain the MCD spectrum of the Fe(III)PPIX species in the
CQ–Fe(III)PPIX complex, a spectrum was recorded at pH 7.4. The MCD
spectra of a low-spin, monomeric (imidazole)2-Fe(III)PPIX complex, as
well as dimeric μ-[Fe(III)PPIX]2O and π–π Fe(III)PPIX species were also
recorded for comparison. All recorded MCD spectra are displayed with
their corresponding UV–visible spectra in Fig. 3.
½CQ ꢂ M2ꢁ
Kobs
¼
:
ð6Þ
2
½Mꢁ ½CQꢁ
Attempts to fit the data to this model without constraints were un-
successful owing to the number of parameters. However, since the spec-
tra of CQ, Fe(III)PPIX π–π dimer (M2) and KD are independently known,
they could be constrained as constants in the fit. This approach resulted
in a fitted log Kobs of 14.1
0.4 (n = 12). The fit showed that mono-
meric Fe(III)PPIX (M) never exceeded 2.5% of total Fe(III)PPIX concen-
tration at any point in the titration and averaged 1.4%. Consequently,
we made a simplification to the model according to Eqs. (7) to (9):
K′
M2 þ CQ ⇌ CQ ꢂ M2
ð7Þ
ð8Þ
0
½CQ ꢂ M2ꢁ
K ¼
½M2ꢁ½CQꢁ
K0 ꢀ KD
¼
ꢀ
¼
¼ Kobs
:
ð9Þ
½CQ ꢂ M2ꢁ ½M2ꢁ ½CQ ꢂ M2ꢁ
2
2
½M2ꢁ½CQꢁ
½Mꢁ
½CQꢁ½Mꢁ
Fitting the data to the equilibrium described by Eq. (7) ignores the
very small fraction of monomer and permitted free fitting of the smaller
number of parameters. These consisted of the spectra of CQ, Fe(III)PPIX
dimer and the complex with CQ as well as K′. The value of log Kobs ob-
tained in this way was 13.3
0.2 (=logK′ + logKD = {6.5
0.2} +

44
D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
Fig. 2. Experimental and predicted spectra obtained from spectrophotometric titration of CQ with Fe(III)PPIX at pH 7.4 (0.02 M HEPES) and 25 °C according to the simplified equilibrium
described by Eq. 7. (a) Predicted spectrum (black) of the free Fe(III)PPIX species compared to the experimental spectrum of π–π dimeric Fe(III)PPIX prepared independently (gray). (b)
Predicted spectrum of free CQ (black) compared to that of the initial solution of CQ with no Fe(III)PPIX added (gray). (c) Predicted spectrum of the CQ–Fe(III)PPIX complex (black) com-
pared to that of Fe(III)PPIX in the presence of a CQ concentration twice that of Fe(III)PPIX (gray). (d) Comparison of the predicted spectrum of the CQ–Fe(III)PPIX complex (black) with μ-
[Fe(III)PPIX]2O (dark gray) and Fe(III)PPIX π–π dimer (dashed light gray). The extinction coefficients of the complex and π–π dimer refer the left axis while that of μ-[Fe(III)PPIX]2O refers
to the right axis. All extinction coefficients are given per mole Fe(III)PPIX.
The UV–visible spectrum of the (imidazole)2-Fe(III)PPIX species is
diagnostic of a low-spin complex with two prominent low energy
peaks around 533 and 563 nm, and a Soret peak shifted to 411 nm
[38,39]. The corresponding MCD spectrum consists of just two promi-
nent derivative-shaped features; one very intense feature centered
around 410 nm and the other much less intense centered at 562 nm.
For the π–π dimer, as previously noted, the absorbance spectrum ex-
hibits a relatively blunt Soret band at about 386 nm with longer wave-
length peaks similar to monomeric H2O-/HO-Fe(III)PPIX centered
around 495, 530 and 614 nm [28]. The corresponding MCD spectrum
at long wavelength resembles that previously reported for monomeric
H2O-Fe(III)PPIX with a series of similar features between 450 and
700 nm [32]. Most notably, a fairly broad positive peak close to
480 nm with a less prominent shoulder around 450 nm, and two
derivative-shaped features centered around 530 and 620 nm. The latter
occur at longer wavelength in monomeric H2O-Fe(III)PPIX. On the
other hand, the Soret region differs markedly from that of monomeric
H2O-Fe(III)PPIX with a much less intense and apparently reversed
derivative-shaped feature. The characteristic UV–visible spectrum of
μ-[Fe(III)PPIX]2O consists of a sharper Soret peak than the π–π dimer
with a prominent high-energy shoulder, very wide low-energy shoul-
der and a broad peak centered at 590 nm. The MCD spectrum is charac-
terized by three intense Gaussian-shaped peaks of alternating sign
centered on 365 nm, an intense derivative-shaped feature around
606 nm and four low intensity peaks between these two main regions.
The spectra of the CQ–Fe(III)PPIX complex most closely resemble those
of μ-[Fe(III)PPIX]2O. In the UV–visible spectrum, the high-energy side of
the Soret band is obscured by peaks arising from the CQ molecule, but
the low energy shoulder is remarkably similar to that of μ-[Fe(III)
PPIX]2O and bears no resemblance to the low-spin imidazole complex,
π–π dimer or previously reported H2O-Fe(III)PPIX monomer [32].
The single broad peak around 600 nm is similar but not identical to
the peak of μ-[Fe(III)PPIX]2O at 590 nm. Thus, as already alluded to
above, the UV–visible spectrum suggests that Fe(III)PPIX may occur as
μ-[Fe(III)PPIX]2O in the CQ complex. This is further supported by a com-
parison of MCD spectra which exhibit strikingly similar features includ-
ing the three Gaussians of alternating sign in the Soret region, the
intense derivative-shaped feature in the long wavelength region and
the four less intense peaks between the two. Notably, the MCD spec-
trum does not resemble those of either the low-spin (imidazole)2-
Fe(III)PPIX species or the π–π Fe(III)PPIX dimer.
The induction of μ-[Fe(III)PPIX]2O formation by CQ in aqueous solu-
tion is further supported by magnetic susceptibility measurements de-
termined using the Evans NMR method (see Table 1 and Fig. 5). The
magnetic moment of the CQ–Fe(III)PPIX complex in aqueous solution
(2.25
for the π–π dimeric species in the absence of CQ under the same condi-
tions (4.8 0.1 μB). Rather, it is in good agreement with that measured
for μ-[Fe(III)PPIX]2O (2.0 0.1 μB) [32].
0.02 μB) was found to differ markedly from the value obtained
A similar multi-technique approach was taken to identifying the
Fe(III)PPIX species in the CQ–Fe(III)PPIX complex precipitated at
pH 5.0. In this case however, IR spectroscopy and magnetic susceptibil-
ity measurements using an Evans balance were employed. IR spectra of
Fe(III)PPIX obtained at low and high pH and the CQ–Fe(III)PPIX com-
plex precipitated at pH 5.0 are shown in Fig. 4. For comparison, the
spectra of CQ (free base), μ-[Fe(III)PPIX]2O and hemin have also been
included.
At first glance, the IR spectrum of the CQ–Fe(III)PPIX complex ap-
pears to be a simple combination of the spectra obtained for CQ and
Fe(III)PPIX precipitated at pH 5.0. Neither the spectrum of the complex
nor Fe(III)PPIX resembles that obtained for μ-[Fe(III)PPIX]2O, which has
a prominent peak near 1410 cm−1 and an intense characteristic υ(Fe–
O–Fe) feature at approx. 880 cm−1. Based solely on these results, one
may be tempted to conclude that the Fe(III)PPIX species in the precipi-
tated CQ–Fe(III)PPIX complex is not μ-[Fe(III)PPIX]2O but rather is the
same species as obtained in the absence of CQ at pH 5.0. However,
care needs to be exercised in making such simple qualitative compari-
sons. The μ-[Fe(III)PPIX]2O species was isolated from strongly alkaline
solution as a sodium salt. Under such conditions the propionate groups
are fully deprotonated. On the other hand, the CQ–Fe(III)PPIX complex
was precipitated at low pH. CQ is doubly protonated at this pH, so in
order to produce a neutral CQ–(μ-[Fe(III)PPIX]2O) complex the μ-
[Fe(III)PPIX]2O moiety must possess one protonated propionic acid
group and one deprotonated propionate group per Fe(III)PPIX molecule.

D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
45
Fig. 3. UV–visible and MCD spectra of Fe(III)PPIX species. (a and c) The (imidazole)2-Fe(III)PPIX complex in 5.64 M aq. DMSO solution, pH 7.4. (b and d) π–π Fe(III)PPIX dimer in aqueous
solution, pH 7.4. (e and g) μ-[Fe(III)PPIX]2O in 5.64 M aq. DMSO solution, pH 10. (f and h) CQ–Fe(III)PPIX complex in aqueous solution, pH 7.4. Asterisks in the UV–visible spectrum cor-
respond to CQ peaks. All MCD spectra have been corrected for natural CD.
In this respect, it is similar to H2O-Fe(III)PPIX obtained at pH 5 in the
Table 1
absence of CQ. The propionate υ(COO−) stretch likely occurs at about
Magnetic moments per Fe(III) center of Fe(III)PPIX species (μB) as determined in solution
by the Evans NMR method and in the solid state using an Evans balance.
1410 cm−1 in μ-[Fe(III)PPIX]2O [32]. The spectrum of HO-Fe(III)PPIX
obtained from alkaline aqueous solution by lyophilization also exhibits
Fe(III)PPIX species
μB
a υ(COO−) stretching peak at this position (Fig. 4). By contrast, the IR
spectrum of hemin, which possesses only propionic acid groups,
displays no such band, but rather a carboxylic acid υ(C_O) stretching
peak at approximately 1700 cm−1 (Fig. 4). H2O-Fe(III)PPIX isolated
at pH 5 exhibits both signals, as expected. The same is true for the
CQ–Fe(III)PPIX complex. As previously reported, [32] H2O-Fe(III)PPIX
Solution-state
Solid-state
Fe(III)PPIXa,b
4.8
2.0
2.25
6.2
4.0
2.3
0.1
0.1
0.02
0.2
0.2
0.1
μ-[Fe(III)PPIX]2Oc
CQ–Fe(III)PPIXd,b
Cl–Fe(III)PPIX
Fe(III)PPIXe,f
CQ–Fe(III)PPIXd,f
a
π–π dimer, axial ligand H2O/HO−
Aqueous, pH 7.4.
5.64 M aqueous DMSO, pH 10 from reference [32].
1:2 CQ:Fe(III)PPIX.
π–π dimer, axial ligand H2O.
.
displays a characteristic H-O-H bending mode around 1620 cm−1
.
b
c
d
e
f
This peak is absent from μ-[Fe(III)PPIX]2O, but unfortunately cannot
be used as a fingerprint in the CQ–Fe(III)PPIX complex because of an
overlapping strong CQ peak at about 1613 cm−1. Between 1000 and
650 cm−1 the spectroscopic bands of the CQ–Fe(III)PPIX complex are
Precipitated from aqueous solution at pH 5.

46
D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
Fig. 4. IR spectra of Fe(III)PPIX species, CQ and CQ–Fe(III)PPIX complex. (a) Fe(III)PPIX precipitated from aqueous solution in the presence of CQ (free base) at pH 5.0. (b) Fe(III)PPIX pre-
cipitated from aqueous solution at pH 5.0. (c) Fe(III)PPIX obtained from alkaline aqueous solution. (d) CQ (free base). (e) μ-[Fe(III)PPIX]2O precipitated from an alkaline DMSO solution.
(f) Hemin. The vertical dashed line in (a), (d) and (e) marks 880 cm−1, the position of the υ(Fe–O–Fe) peak in μ-[Fe(III)PPIX]2O.
fairly broad. There is no prominent peak around 880 cm−1, but there is
a noticeable shoulder at this position. Regrettably, this is inconclusive
for identifying the Fe(III)PPIX species present, because CQ exhibits an
intense overlapping peak centered at about 873 cm−1. Indeed, with
the exception of three peaks that are exclusive to CQ, all of the features
in this part of the spectrum are either common to all Fe(III)PPIX species,
or overlap with CQ bands. Thus, the use of IR spectroscopy to identify
the nature of Fe(III)PPIX in the complex is inconclusive.
Magnetic susceptibility measurements of these solids however,
are more definitive (see Table 1). Magnetic moments of the solid
samples reproduce the findings observed in aqueous solution at
pH 7.4, namely that the Fe(III)PPIX species in the CQ–Fe(III)PPIX com-
plex has a low magnetic moment which is similar to that observed for
μ-[Fe(III)PPIX]2O. As a control, the magnetic susceptibility of high-spin
(S = 5/2) hemin was also measured. The magnetic moment obtained
for this compound (6.2
0.2 μB) was close to the expected spin-only
Fig. 5. Predicted temperature dependence of the magnetic moment (μB) of μ-[Fe(III)
PPIX]2O (solid line) calculated with a J-coupling constant of −110
15 cm−1 using mag-
value of 5.92 μB. Using SQUID magnetometry, Stanek and Dziedzic-
Kocurek have determined the J-coupling constant for μ-[Fe(III)PPIX]2O
netic susceptibilities obtained from Eq. 14 in reference [40]. Dashed lines represent two
standard deviations. The magnetic moments of the solid CQ–Fe(III)PPIX complex at
298 K, the complex in aqueous solution and μ-[Fe(III)PPIX]2O at 303 K determined in
our study are represent by the filled circle, open square and cross, respectively. Error
bars represent one standard deviation (where there is no error bar, it is smaller than the
symbol). The gray dotted line is the expected spin-only value for a low-spin complex
with S = 1/2 (1.73 μB).
to be −110
15 cm−1 [40]. The magnetic moment as a function
of temperature can be predicted using this J-value and is shown in
Fig. 5. The value obtained for the CQ–Fe(III)PPIX complex as well as
for μ-[Fe(III)PPIX]2O itself agree within reported error with those pre-
dicted at the experimental temperature used.

D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
47
(P b 0.0001, r2 = 0.95) which is shown in Fig. 6 as a solid line alongside
the correlation previously reported (dashed line). Incorporating the
data for the π–π and μ-oxo dimeric Fe(III)PPIX species into this plot as
test molecules, shows that the extrapolated value from the correlation
represented by the solid line in Fig. 6 is in close agreement with expec-
tation. Thus this empirical relationship was used to estimate the degree
of aggregation of the CQ–Fe(III)PPIX complex.
To position the CQ–Fe(III)PPIX data point on the plot in Fig. 6, the
molar volume was required. This was estimated for monomeric (1:2),
dimeric (2:4), trimeric (3:6) and tetrameric (4:8) complexes by divid-
ing the molar mass (1567, 3134, 4701 and 6268 g mol−1, respectively)
by the predicted density of the complex. The latter was estimated using
a weighted average of the density of CQ (predicted using ACD
ChemSketch [43]) and Fe(III)PPIX (taken as 1.3753 g cm−3 [28]). It
can be seen from Fig. 6 that the data point describing the dimeric 2:4
unit is closest to the regression line. This suggests that the CQ–Fe(III)
PPIX complex in solution consists of two CQ molecules and two μ-
[Fe(III)PPIX]2O dimers, with predicted formula (CQ–μ-[Fe(III)PPIX]2O)2.
Table 2
Diffusion coefficients (D) normalized to 298 K and calculated molar volumes, V.
D (10−10 m2 s−1
)
V (cm3 mol−1
)
(CN)2-Fe(III)PPIXa
HO-Fe(III)PPIXa,b
μ-[Fe(III)PPIX]2Oa
CQ–Fe(III)PPIXc
2.2
1.4
1.6
0.6
0.2
0.1
0.1
0.2
485
920
907
1217 (1:2)
2434 (2:4)
3652 (3:6)
4869 (4:8)
288d
CQc
3.3
0.3
a
Values reported by de Villiers et al. [28].
π–π dimer, pH 10.
Determined in this work.
b
c
d
Predicted using ACD ChemSketch [43].
3.3. Determination of the aggregation state of the CQ–Fe(III)PPIX complex
While the findings described above strongly support the formation
of a CQ–μ-[Fe(III)PPIX]2O complex in aqueous solution, they cannot dis-
tinguish between monomers, dimers or higher aggregates of this com-
plex. In an attempt to determine the aggregation state, the diffusion
coefficient of the complex was measured in aqueous solution using a
diffusion cell. The diffusion coefficient of the CQ–Fe(III)PPIX complex
is listed in Table 2 along with the value determined for CQ in this
work. For comparison, previously reported values for various Fe(III)
PPIX species obtained using the same method are also shown [28]. All
values have been normalized to 298 K.
As detailed previously [28], an empirical linear relationship between
the logarithm of diffusion coefficient, logD, and molar volume, logV, re-
ported by Gustafson and Dikhut for polyatomic aromatic hydrocarbons
[41], has been shown to extend to the monomeric bis-cyano Fe(III)PPIX
species, (CN)2-Fe(III)PPIX, as well as the π–π dimeric species in aqueous
solution. To increase the number of points in the plot, we have included
the diffusion coefficients for several simple substituted aromatic com-
pounds reported by Neisner and Heintz [42]. In addition, the diffusion
coefficient of CQ in the absence of Fe(III)PPIX was also determined
and incorporated into the plot (Fig. 6). Fitting of this enlarged data set
using linear regression produced a statistically significant correlation
4. Discussion
Magnetic susceptibility measurements of the CQ–Fe(III)PPIX com-
plex both in solution and the solid-state unequivocally demonstrated
that the complex possesses a low magnetic moment of 2.3 μB. This is
close to the value determined for μ-[Fe(III)PPIX]2O (2.0
0.1 μB) [32].
A Job plot constructed in solution close to pH 7.5, as well as UV–visible
spectroscopic measurements obtained by dissolving the solid that had
been precipitated at pH 5.0, definitively showed that the mole fraction
of Fe(III)PPIX in the complex was 0.68
value of 0.67 expected for a 1:2 CQ:Fe(III)PPIX complex. The complex
is strong, with a logKobs of 13.3 0.2 at pH 7.4.
0.02. This is near to the
There are only two possible explanations for the low magnetic mo-
ment in the CQ–Fe(III)PPIX complex. Either it is low-spin with one
unpaired electron, or there are antiferromagnetically coupled iron
centers. Low-spin Fe(III)PPIX complexes are observed when two
strong-field ligands coordinate to the Fe(III) center, for example in
(imidazole)2-Fe(III)PPIX [38,39]. However, this type of complex can be
discounted on two grounds. Firstly, the mole fraction of Fe(III)PPIX in
such a complex would have to be 0.33, inconsistent with the observed
value of 0.68. Secondly, both the UV–visible and MCD spectra of low-
spin complexes have characteristic features which are not present in
the spectra of CQ–Fe(III)PPIX (see Fig. 3a and c vs. f and h). This leaves
strong antiferromagnetic coupling via a bridging ligand linking two
high-spin Fe(III) centers as the only explanation for the low magnetic
moment. In principle, a number of bridging ligands could be considered.
Chloride can be discounted on the basis of its deliberate exclusion from
the magnetic susceptibility experiments (which made use of hematin
rather than hemin). Phosphate cannot be the bridging group between
the paramagnetic centers because it cannot account for the observed
strong antiferromagnetic coupling. Direct M–L–M orbital overlap is
needed for an efficient superexchange mechanism [44]. Furthermore, it
was absent from the solid sample. In our system, this leaves only oxide
and hydroxide as feasible bridging ligands. Hydroxo-bridged iron(III)
porphyrin complexes are rare, but not unknown [45–47]. J-coupling
constants tend to be substantially weaker than those observed for μ-
oxo dimers [46,47], in agreement with a theoretical investigation
which has shown that protonation of μ-oxo bridging ligands significant-
ly reduces the J-coupling constant between iron(III) centers [48]. Thus,
the antiferromagnetic coupling of μ-hydroxo-bridged species would
be too weak to account for the low magnetic moment observed in
CQ–Fe(III)PPIX. This leaves the μ-oxo dimer of Fe(III)PPIX as the only
candidate able to account for the low magnetic moment observed.
Such a conclusion is further supported by the close agreement between
the observed magnetic moment and the value predicted at experimen-
tal temperature using the J-coupling constant of μ-[Fe(III)PPIX]2O (see
Fig. 5). This is further substantiated by the close resemblance of the
Fig. 6. Dependence of logD (at 298 K) on logV. Values for aromatic molecules were obtain-
ed from Gustafson and Dikhut [41], and Neisner and Heintz [42] (unfilled squares). Those
for (CN)2-Fe(III)PPIX, π–π Fe(III)PPIX dimer and μ-[Fe(III)PPIX]2O (unfilled triangle, filled
square and filled diamond respectively) were from de Villiers et al. [28]. Values for CQ (un-
filled circle) and CQ–Fe(III)PPIX complex were determined in this study. Various models of
the aggregation state of the CQ–Fe(III)PPIX complex are represented by filled triangles.
These are (a) 1:2, (b) 2:4, (c) 3:6 and (d) 4:8. The solid line represents the linear regression
fitted to the entire aromatic molecule data set together with CQ and (CN)2-Fe(III)PPIX,
while the dashed line represents the linear regression previously reported which incorpo-
rated only the Gustafson and Dikhut data set together with (CN)2-Fe(III)PPIX. Filled data
points were not used in the regression analysis. The models for the CQ–Fe(III)PPIX com-
plex that did not fall on the solid regression line are shown in gray.

48
D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
UV–visible and MCD spectra of CQ–Fe(III)PPIX and μ-[Fe(III)PPIX]2O
(see Figs. 2d and 3e and g vs. f and h). In addition, this would also ex-
plain the previously reported remarkable similarity between the
Mössbauer spectra of CQ–Fe(III)PPIX and μ-[Fe(III)PPIX]2O [22]. A com-
plex of the type CQ–(μ-[Fe(III)PPIX]2O) is thus consistent with both the
observed magnetic properties and stoichiometry.
similar IR absorptions in some μ-oxo manganese(III)phthalocyanins
[54]. These authors put forward no hypotheses to explain the absence
of the band. While a similar phenomenon may occur here, we suspect
that it is more likely that the peak is present, but that it is a shoulder.
Furthermore, owing to overlap with a CQ band at a similar position it
is poorly resolved.
The observation of a 1:2 CQ:Fe(III)PPIX stoichiometry is in agree-
ment with a number of previous studies in aqueous solution. These
include Chou et al. who first reported this ratio on the basis of equilibri-
um dialysis experiments in 50 mM phosphate buffer (pH 7.4) [49];
Constantinidis and Satterlee who performed a UV–visible Job plot at
pH 6 using urohemin [50]; NMR relaxation studies at pH 6.5 by Leed
et al. and Schwedhelm et al. [19,23]; and most recently a UV–visible
Job plot by Crespo et al. in 10 mM phosphate buffer (pH 7.4) [14].
It is notable that studies reporting higher ratios (1:4–1:8) were all
performed using ITC, usually with 0.25 M phosphate buffer and
often with relatively high concentrations of salts (e.g. 0.15 M KCl)
[24,25,51]. Elevated ionic strengths have previously been reported to
be able to induce higher aggregation of μ-[Fe(III)PPIX]2O [29]. Regard-
less of the reason for this discrepancy, it is clear from the Job plot
(Fig. 1a) and even more strikingly from the absorbance ratios of the
redissolved CQ–Fe(III)PPIX precipitate (Fig. 1b), that stoichiometries of
1:4 and higher fall well outside our experimental error and can there-
fore be excluded. A number of authors have suggested that the CQ–
Fe(III)PPIX complex in aqueous solution involves μ-[Fe(III)PPIX]2O
[19,20,23,24,26,52]. While some substantiated this claim with magnetic
moment measurements [20,21,23], in most cases it was based on the
supposition that μ-[Fe(III)PPIX]2O is the predominant species in aque-
ous solution. Several recent studies have shown that this is not the
case, but rather that it is predominantly a π–π dimer [14,28,29]. This is
further supported by the MCD spectrum shown in Fig. 3d which dra-
matically differs from that of μ-[Fe(III)PPIX]2O (Fig. 3g). Regardless of
this recent evidence, there is still a widespread misconception that μ-
[Fe(III)PPIX]2O is the major species in aqueous solution [27]. This is an
important point because the current findings show that CQ induces for-
mation of μ-[Fe(III)PPIX]2O, rather than simply binding to preformed μ-
[Fe(III)PPIX]2O. Thus in the equilibrium described by Eq. 7 above, M2 on
the left-hand side represents Fe(III)PPIX π–π dimer, while on the right-
hand side it represents μ-[Fe(III)PPIX]2O. Indeed, in the fitting of com-
plete spectra to the spectrophotometric titration data, it was quite evi-
dent that the free Fe(III)PPIX species is the π–π dimer (Fig. 2a). A
second misconception that needs to be dispelled is the idea that the
low magnetic moment of μ-[Fe(III)PPIX]2O arises from the Fe(III) cen-
ters existing in a low-spin state with one unpaired electron (S = 1/2).
This is not the case. Rather, the low magnetic moment arises from quan-
tum mechanical coupling between the five unpaired electrons on each
iron center in a temperature-dependent manner (see prediction in
Fig. 5). Such a system would be diamagnetic at 0 K, but becomes ther-
mally decoupled with increasing temperature [40]. In the case of μ-
[Fe(III)PPIX]2O, its J-coupling constant gives rise to a magnetic moment
at 298 K coincidentally similar, but not identical to that of a low-spin
complex (μB = 2.06 versus 1.73, see Fig. 5 solid black line vs. gray dot-
ted line).
Although both the Job plot and UV–visible spectroscopic measure-
ments on redissolved precipitate show that the CQ:Fe(III)PPIX ratio in
the complex is 1:2, they do not provide information on its aggregation
state. In other words, these techniques are not able to distinguish be-
tween 1:2, 2:4 and 3:6 complexes or even higher aggregates. There
have been differing opinions regarding this matter. For example, Mo-
reau et al. proposed that the CQ–Fe(III)PPIX complex stacks in very
large aggregates consisting of alternating CQ and μ-[Fe(III)PPIX]2O mol-
ecules [52], while Schwedhelm et al., proposed a 2:4 complex [23]. This
problem is difficult to investigate because it is not easily addressed by
spectroscopic techniques. Therefore, we have attempted to probe the
aggregation state using diffusion measurements. The Stokes–Einstein
relationship is sometimes used to relate diffusion coefficients to molec-
ular size. However, this equation was developed to describe the diffu-
sion of macroscopic colloids and does not strictly apply to the
molecular scale, especially in strongly solvating systems such as water.
Empirical relationships such as the Othmer–Thakar, Hayduk–Laudie,
Wilke–Chang and Scheibel equations have been proposed to predict
molecular diffusion coefficients [55–58]. More recent correlations ob-
served by Gustafson and Dikhut led to improved predictions for planar
extended aromatic systems [41]. We have previously shown this to be
of value for Fe(III)PPIX species [28]. Using this approach, with the addi-
tion of further empirical data for both aromatic systems and Fe(III)PPIX
species, we found that the 2:4 complex best fitted the correlation
line (see Fig. 6). We therefore tentatively identify the complex as
(CQ–μ-[Fe(III)PPIX]2O)2. It must, however, be recognized that the em-
pirical nature of this approach together with fairly large experimental
errors makes it impossible to completely exclude either the 1:2 or 3:6
aggregates. Large aggregates can however be discounted.
As a final point, we note that our proposed complex, (CQ–μ-[Fe(III)
PPIX]2O)2, is similar to that suggested by Schwedhelm et al. [23].
However, the structure proposed by these authors involving two adja-
cent μ-[Fe(III)PPIX]2O linked by reciprocal propionate-Fe(III) coordina-
tion bonds and capped with CQ, is inconsistent with IR evidence. As in
hemozoin/β-hematin, coordination of propionate groups to Fe(III)
should give rise to intense bands at around 1660 and 1210 cm−1. In-
deed, they are striking markers of such an interaction in hemozoin
[59], but are clearly completely absent from the spectrum of the CQ–
Fe(III)PPIX complex (Fig. 4b). In fact, Schwedhelm et al. proposed such
a structure to account for the low magnetic moment of the complex,
which they erroneously ascribed to a low-spin state which they claimed
required entirely six-coordinate Fe(III) centers. As we have noted above,
the low magnetic moment in fact arises from antiferromagnetic cou-
pling of high-spin Fe(III) centers. Moreover, six-coordinate Fe(III)PPIX
is only low-spin with strong-field ligands and would not be likely with
weak-field oxide and propionate π-donor groups.
Recent uncertainty about whether μ-[Fe(III)PPIX]2O is present in the
CQ–Fe(III)PPIX complex can be ascribed to the reported absence of the
Fe–O–Fe antisymmetric stretching band in the IR spectrum around
880 cm−1 [15]. Indeed, in the present study, we confirm the absence
of a prominent band at this position (Fig. 4a). However, given the incon-
trovertible magnetic evidence obtained from the very same sample for
the presence of μ-[Fe(III)PPIX]2O, we can only conclude that this peak
is not a definitive marker. There is precedence for the apparent disap-
pearance of this peak. Ercolani et al. demonstrated the existence of
two interconvertible forms of an iron(III)phthalocyanin μ-oxo dimer,
one of which lacked the characteristic υ(Fe–O–Fe) frequency despite
definitive magnetic and compositional evidence that it was indeed a
μ-oxo dimer [53]. Lever and coworkers also reported the absence of
5. Conclusions
Magnetic susceptibility measurements together with UV–visible and
MCD observations convincingly demonstrate that CQ induces formation
of the μ-oxo dimer of Fe(III)PPIX, forming a CQ–Fe(III)PPIX complex
with a 1:2 stoichiometry. This occurs both in solution at pH 7.5 and in
the solid precipitated at pH 5.0 (at least under the preparation condi-
tions used in this study). The IR band near 880 cm−1 that has been pro-
posed to be characteristic of the μ-oxo dimer species is not a definitive
marker in this system. It is not resolved in the sample of CQ–Fe(III)
PPIX which magnetic and spectroscopic measurements showed could
only be antiferromagnetically coupled. Determination of the aggrega-
tion state of the complex is less conclusive, but diffusion data suggest

D. Kuter et al. / Journal of Inorganic Biochemistry 133 (2014) 40–49
49
[13] T.J. Egan, W.W. Mavuso, D.C. Ross, H.M. Marques, J. Inorg. Biochem. 68 (1997)
137–145.
that the best model is (CQ–μ-[Fe(III)PPIX]2O)2. At this stage, we are not
offering a structural hypothesis for the complex. Rather, based on the
current findings, this is being investigated in a detailed molecular dy-
namics study supported by EXAFS. The results will be reported in the
near future.
[14] M.P. Crespo, L. Tilley, N. Klonis, J. Biol. Inorg. Chem. 15 (2010) 1009–1022.
[15] M. Asghari-Khiavi, J. Vongsvivut, I. Perepichka, A. Mechler, B.R. Wood, D.
McNaughton, D.S. Bohle, J. Inorg. Biochem. 105 (2011) 1662–1669.
[16] T. Frosch, B. Küstner, S. Schlücker, A. Szeghalmi, M. Schmitt, W. Kiefer, J. Popp,
J. Raman Spectrosc. 35 (2004) 819–821.
Finally, we would like to stress that we do not propose that the com-
plex described here is necessarily directly responsible for hemozoin in-
hibition. Indeed, recent studies from our group have lent support to the
proposal that CQ acts by inhibiting the fast-growing face of the
hemozoin crystal [7,60]. Nonetheless, it is likely that the complex will
form as a consequence of inhibition of hemozoin formation. It may
well then play an important role in the subsequent distribution of
Fe(III)PPIX in the malaria cell and hence influence its toxicity to the
parasite.
[17] G.T. Webster, D. McNaughton, B.R. Wood, J. Phys. Chem. B 113 (2009)
6910–6916.
[18] M.S. Walczak, K. Lawniczak-Jablonska, A. Wolska, A. Sienkiewicz, L. Suárez, A.J.
Kosar, D.S. Bohle, J. Phys. Chem. B 115 (2011) 1145–1150.
[19] A. Leed, K. DuBay, L.M.B. Ursos, D. Sears, A.C. de Dios, P.D. Roepe, Biochemistry 41
(2002) 10245–10255.
[20] L.B. Casabianca, D. An, J.K. Natarajan, J.N. Alumasa, P.D. Roepe, C. Wolf, A.C. de Dios,
Inorg. Chem. 47 (2008) 6077–6081.
[21] L.B. Casabianca, J.B. Kallgren, J.K. Natarajan, J.N. Alumasa, P.D. Roepe, C. Wolf, A.C. de
Dios, J. Inorg. Biochem. 103 (2009) 745–748.
[22] P.A. Adams, P.A.M. Berman, T.J. Egan, P.J. Marsh, J. Silver, J. Inorg. Biochem. 63 (1996)
69–77.
[23] K.F. Schwedhelm, M. Horstmann, J.H. Faber, Y. Reichert, M. Büchner, G. Bringmann,
C. Faber, Open Spec. J. 2 (2008) 10–18.
[24] A. Dorn, S.R. Vippagunta, H. Matile, C. Jaquet, J.L. Vennerstrom, R.G. Ridley, Biochem.
Pharmacol. 55 (1998) 727–736.
[25] K. Bachhawat, C.J. Thomas, N. Surolia, A. Surolia, Biochem. Biophys. Res. Commun.
276 (2000) 1075–1079.
[26] S.R. Vippagunta, A. Dorn, R.G. Ridley, J.L. Vennerstrom, Biochim. Biophys. Acta 1475
(2000) 133–140.
Abbreviations
CQ
chloroquine
EXAFS
extended X-ray absorption fine structure
Fe(III)PPIX ferriprotoporphyrin IX
MCD
ITC
magnetic circular dichroism
isothermal titration calorimetry
[27] A.P. Gorka, A.C. De Dios, P.D. Roepe, J. Med. Chem. 56 (2013) 5231–5246.
[28] K.A. de Villiers, C.H. Kaschula, T.J. Egan, H.M. Marques, J. Biol. Inorg. Chem. 12 (2007)
101–117.
[29] C. Asher, K.A. de Villiers, T.J. Egan, Inorg. Chem. 48 (2009) 7994–8003.
[30] S.B. Brown, P. Jones, I.R. Lantzke, Nature 223 (1969) 960–961.
[31] P. Gans, A. Sabatini, A. Vacca, Talanta 43 (1996) 1739–1753.
[32] D. Kuter, G.A. Venter, K.J. Naidoo, T.J. Egan, Inorg. Chem. 51 (2012) 10233–10250.
[33] D.F. Evans, J. Phys. E Sci. Instrum. 7 (1974) 247–249.
[34] B.N. Figgis, R.S. Nyholm, J. Chem. Soc. (1958) 4190–4191.
[35] G.A. Bain, J.F. Berry, J. Chem. Educ. 85 (2008) 532–536.
[36] P.W. Linder, L.R. Nassimbeni, A. Polson, A.L. Rodgers, J. Chem. Educ. 53 (1976)
330–332.
[37] L.M.B. Ursos, K.F. DuBay, P.D. Roepe, Mol. Biochem. Parasitol. 112 (2001) 11–17.
[38] H.M. Marques, O.Q. Munro, M.L. Crawcour, Inorg. Chim. Acta 196 (1992) 221–229.
[39] D. Kuter, K. Chibale, T.J. Egan, J. Inorg. Biochem. 105 (2011) 684–692.
[40] J. Stanek, K. Dziedzic-Kocurek, J. Magn. Magn. Mater. 322 (2010) 999–1003.
[41] K.E. Gustafson, R.M. Dickhut, J. Chem. Eng. Data 39 (1994) 281–285.
[42] R. Niesner, A. Heintz, J. Chem. Eng. Data 45 (2000) 1121–1124.
[43] Advanced Chemistry Development Inc., Toronto, ON, Canada, 2013.
[44] J.B. Goodenough, Magnetism and the Chemical Bond, Wiley, New York, 1963,
pp. 170–172.
Acknowledgments
This material is based upon the work supported in part by the Na-
tional Research Foundation of South Africa (NRF). Any opinion, findings
and conclusions or recommendations expressed in this material are
those of the authors and do not necessarily reflect the views of the Na-
tional Research Foundation. We also acknowledge the support of the
Medical Research Council of South Africa and the University of Cape
Town (UCT). DK acknowledges the NRF, UCT and Kevin J. Naidoo
(through the South African Research Chairs Initiative of the Department
of Science and Technology and NRF) for financial support.
References
[45] W.R. Scheidt, B. Cheng, M.K. Safo, F. Cukiernik, J.-C. Marchon, P.G. Debrunner, J. Am.
Chem. Soc. 114 (1992) 4420–4421.
[1] R.L. Summers, M.N. Nash, R.E. Martin, Cell. Mol. Life Sci. 69 (2012) 1967–1995.
[2] A. Ecker, A.M. Lehane, C. Clain, D.A. Fidock, Trends Parasitol. 28 (2012) 504–514.
[3] R.E. Martin, R.V. Marchetti, A.I. Cowan, S.M. Howitt, S. Bröer, K. Kirk, Science 325
(2009) 1680–1682.
[4] D. van Schalkwyk, T.J. Egan, Drug Resist. Updat. 9 (2006) 211–226.
[5] D.C. Warhurst, J.C. Craig, I.S. Adagu, R.K. Guy, P.B. Madrid, Q.L. Fivelman, Biochem.
Pharmacol. 73 (2007) 1910–1926.
[6] F. Dubar, T.J. Egan, B. Pradines, D. Kuter, K.K. Ncokazi, D. Forge, J.-F. Paul, C. Pierrot, H.
Kalamou, J. Khalife, E. Buisine, C. Rogier, H. Vezin, I. Forfar, C. Slomianny, X. Trivelli, S.
Kapishnikov, L. Leiserowitz, D. Dive, C. Biot, ACS Chem. Biol. 6 (2011) 275–287.
[7] J.M. Combrinck, T.E. Mabotha, K.K. Ncokazi, M.A. Ambele, D. Taylor, P.J. Smith, H.C.
Hoppe, T.J. Egan, ACS Chem. Biol. 8 (2013) 133–137.
[46] D.R. Evans, R.S. Mathur, K. Heerwegh, C.A. Reed, Z. Xie, Angew. Chem. Int. Ed. 36
(1997) 1335–1337.
[47] S.K. Ghosh, S.P. Rath, J. Am. Chem. Soc. 132 (2010) 17983–17985.
[48] Z. Chen, Z. Xu, L. Zhang, F. Yan, Z. Lin, J. Phys. Chem. A 105 (2001) 9710–9716.
[49] A.C. Chou, R. Chevli, C.D. Fitch, Biochemistry 19 (1980) 1543–1549.
[50] I. Constantinidis, J.D. Satterlee, J. Am. Chem. Soc. 110 (1988) 4391–4395.
[51] S.R. Cheruku, S. Maiti, A. Dorn, B. Scorneaux, A.K. Bhattacharjee, W.Y. Ellis, J.L.
Vennerstrom, J. Med. Chem. 46 (2003) 3166–3169.
[52] S. Moreau, B. Perly, C. Chachaty, C. Deleuze, Biochim. Biophys. Acta 840 (1985)
107–116.
[53] C. Ercolani, M. Gardini, F. Monacelli, G. Pennesi, G. Rossi, Inorg. Chem. 22 (1983)
2584–2589.
[8] R. Hayward, K.J. Saliba, K. Kirk, J. Cell Sci. 119 (2006) 1016–1025.
[9] T.J. Egan, J.M. Combrinck, J. Egan, G.R. Hearne, H.M. Marques, S. Ntenteni, B.T. Sewell,
P.J. Smith, D. Taylor, D.A. van Schalkwyk, J.C. Walden, Biochem. J. 365 (2002) 343–347.
[10] R. Buller, M.L. Peterson, Ö. Almarsson, L. Leiserowitz, Cryst. Growth Des. 2 (2002)
553–562.
[11] C.H. Kaschula, T.J. Egan, R. Hunter, N. Basilico, S. Parapini, D. Taramelli, E. Pasini, D.
Monti, J. Med. Chem. 45 (2002) 3531–3539.
[12] F. Dubar, S. Bohic, C. Slomianny, J.-C. Morin, P. Thomas, H. Kalamou, Y. Guérardel, P.
Cloetens, J. Khalife, C. Biot, Chem. Commun. 48 (2012) 910–912.
[54] G.W. Rayner Canham, A.B..P. Lever, Inorg. Nucl. Chem. Lett. 9 (1973) 513–517.
[55] D.F. Othmer, M.S. Thakar, Ind. Eng. Chem. 45 (1953) 589–593.
[56] W. Hayduk, H. Laudie, AIChE J 20 (1974) 611–615.
[57] C.R. Wilke, P. Chang, AIChE J 1 (1955) 264–270.
[58] E.G. Scheibel, Ind. Eng. Chem. 46 (1954) 2007–2008.
[59] A.F.G. Slater, W.J. Swiggard, B.R. Orton, W.D. Flitter, D.E. Goldberg, A. Cerami, G.B.
Henderson, Proc. Natl. Acad. Sci. U. S. A. 88 (1991) 325–329.
[60] J. Gildenhuys, T. le Roex, T.J. Egan, K.A. de Villiers, J. Am. Chem. Soc. 135 (2013)
1037–1047.