Redox Properties and Activity of Iron–Citrate Complexes: Evidence for

Article pubs.acs.org/crt

Redox Properties and Activity of Iron−Citrate Complexes: Evidence for Redox Cycling Fatima I. Adam,† Patricia L. Bounds,‡ Reinhard Kissner, and Willem H. Koppenol* Institute of Inorganic Chemistry, Department of Chemistry and Applied Biosciences, ETH Zurich, Vladimir-Prelog-Weg 1, CH-8093 Zurich, Switzerland S Supporting Information *

ABSTRACT: Iron in iron overload disease is present as non-transferrin-bound iron, consisting of iron, citrate, and albumin. We investigated the redox properties of iron citrate by electrochemistry, by the kinetics of its reaction with ascorbate, by ESR, and by analyzing the products of reactions of ascorbate with iron citrate complexes in the presence of H2O2 with 4-hydroxybenzoic acid as a reporter molecule for hydroxylation. We report −0.03 V < E°′ > +0.01 V for the (Fe3+−cit/ Fe2+−cit) couple. The first step in the reaction of iron citrate with ascorbate is the rapid formation of mixed complexes of iron with citrate and ascorbate, followed by slow reduction to Fe2+−citrate with k = ca. 3 M−1 s−1. The ascorbyl radical is formed by iron citrate oxidation of Hasc− with k = ca. 0.02 M−1 s−1; the majority of the ascorbyl radical formed is sequestered by complexation with iron and remains EPR silent. The hydroxylation of 4-hydroxybenzoic acid driven by the Fenton reduction of iron citrate by ascorbate in the presence of H2O2 proceeds in three phases: the first phase, which is independent of the presence of O2, is revealed as a nonzero intercept that reflects the rapid reaction of accumulated Fe2+ with H2O2; the intermediate oxygen-dependent phase fits a first-order accumulation of product with k = 5 M−1 s−1 under aerobic and k = 13 M−1 s−1 under anaerobic conditions; the slope of the final linear phase is ca. k = 5 × 10−2 M−1 s−1 under both aerobic and anaerobic conditions. Product yields under aerobic conditions are greater than predicted from the initial concentration of iron, but they are less than predicted for continuous redox cycling in the presence of excess ascorbate. The ongoing formation of hydroxylated product supports slow redox cycling by iron citrate. Thus, when H2O2 is available, iron−citrate complexes may contribute to pathophysiological manifestations of iron overload diseases.

■

INTRODUCTION Iron in physiology is always present in complexed form, whether as functional iron in hemoglobin and myriad enzymes, as iron stored in ferritin and transferrin, or when shuttled between the various physiological iron pools. The iron pool in human tissues is supplied from the diet via intestinal enterocytes and from existing iron stores recovered mainly from erythrocytes via macrophages.1 When the amount of iron entering the circulatory system exceeds the capacity of transferrin to bind it, the iron is chelated in the form of nontransferrin-bound iron (NTBI),2 which readily enters hepatocytes, cardiac myocytes, and endocrine tissues,1 where excess levels of iron cause damage.3 Iron citrate has also been found in knee-joint synovial fluid drawn from patients with rheumatoid arthritis4 and in cerebrospinal fluid.5 Under pathophysiological conditions of iron overload, the iron content of the blood exceeds the capacity of the transferrin to bind it, and iron is present in the as yet poorly described NTBI. The precise chemical and biochemical nature of NTBI has been a subject of study for more than 2 decades, and it is generally accepted that NTBI consists largely of citrate complexes of iron. NTBI is composed of Fe, citrate, and albumin: Grootveld et al.6 found NMR and HPLC evidence to suggest that NTBI in blood fractions from hemochromatosis patients is largely iron citrate, with possible contributions iron− citrate−acetate and protein-bound iron/iron citrate. Serum © XXXX American Chemical Society

albumin has been suggested as a candidate for the protein component of NTBI.7,8 The capacity of human serum albumin to bind iron or iron citrate at low pH was demonstrated by Coddington and Perkins,9 and the interaction of bovine serum albumin with iron and iron citrate complexes was reported by Løvstad.10 Hider and co-workers11,12 have presented convincing evidence that NTBI is composed in part of serum albumin. It has been proposed that iron citrate redox activity is responsible for, or at least plays a role in, pathophysiological manifestations of iron overload diseases. The structure of an iron−citrate complex depends on the ionization state of the complex. The pKa values for the three carboxylic acid functions of citric acid are 3.13, 4.76, and 6.40.13 The pKa of the hydroxyl group was judged to be >11 by potentiometric titration,14 whereas Silva et al.15 reported 14.4 ± 0.3 on the basis of NMR measurements. The first ionization occurs at the central C3 carboxylic acid to form the symmetric monoionized carboxylate.16 Metal ions are complexed by citrate generally via the C1 and C3 carboxylates and the C3 hydroxyl group. Hamm et al.17 concluded on the basis of titration data that the iron(III)−citric acid complex behaves as a tetrabasic acid, whereas the corresponding iron(II)−citrate exhibits tribasic behavior; similar findings were reported by Francis Received: September 13, 2014

A

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology and Dodge.14 Conductometric titration of ferric sulfate with excess trisodium citrate suggest complexes with stoichiometries of Fe3+2(cit)3 and Fe3+5(cit)4, whereas titration with excess free citric acid indicates complexes of Fe3+3(cit)2; thermometric titrations support the formation of both Fe3+2(cit)3 and Fe3+3(cit)2.18 Timberlake19 reported monomeric (Fe3+cit) and dimeric Fe3+2(Hcit)2 species on the basis of potentiometric and spectrophotometric measurements. Generally, however, titrations and conductometric studies do not discriminate between single complexes and oligomers. Shweky et al.20 reported the synthesis and crystal structures of two different dinuclear iron(III)−citrate complexes. One of these has a core stoichiometry of Fe3+2(cit)2 where the citrate ions are fully ionized; in the other, the stoichiometry is Fe3+2(cit)3 with one terminal carboxylate protonated and not coordinated to the Fe3+ ions. Matzepetakis et al.21 reported the structure of the mononuclear [Fe(C6H4O7)2]5− anion in which all three acid functions of both citrate moieties are fully ionized, the central hydroxyl is deprotonated, and each citrate moiety contributes two ionized carboxylates and the hydroxylate to the iron(III) coordination sphere. This arrangement of citrate around the iron center is similar to that reported for polymeric structures of Fe2+−citrate crystallized from aqueous citrate solution, in which case the hydroxyl group of the citrates is protonated.22 Higher-order complex structures, with a core of Fe3+9(cit)823 or Fe3+8(cit)6,24 have also been reported. The five known modes of coordination reported for iron(III)−citrate complexes have been reviewed.25,26 The redox properties of iron in citrate media have been investigated since the 1940s. Lingane27 found, for ca. 2 mM iron in 0.5 M citrate in the presence of 0.005% gelatin, a single half-wave potential, corresponding to E1/2 = 0.426−0.108 V, at physiological and lower pH, attributed to a reversible reduction of Fe3+ to Fe2+. Meites28 reported multiple pH-dependent waves for 2 mM iron with 0.12 and 0.5 M citrate in the absence of gelatin; four half-wave potentials were observed at an iron− citrate ratio of 1:60, compared to three at 1:250. These workers speculated that the higher relative iron content gives rise to an additional species containing relatively more iron per complex. In vivo iron species in blood6 and synovial fluid4 have been identified as iron(III)−citrate in NMR experiments. Martin29 reasoned that the predominant iron(III)−citrate species at neutral pH is the 1:2 complex [Fe(Hcit)2]3−. An exacting investigation of iron citrate speciation under presumed physiological conditions was carried out by Evans et al.;11 these workers concluded that iron speciation is strongly dependent on the iron−citrate ratio and that, for expected thalassemia sera iron−citrate ratios of 1:10−1:100, the monoiron dicitrate species, [Fe(cit)2]5−, is likely to dominate, with contributions from other poorly defined oligomeric forms. The protein fraction of NTBI is serum albumin,30 the concentration of which in serum is estimated at 0.60 mM.31 The presence of serum albumin influences the in vivo iron− citrate speciation: Hershko et al.7 noted that 59FeCl3 added to thalassemic serum comigrates electrophoretically with albumin. The form of iron citrate that binds to human serum albumin (HSA) in vitro has been investigated: Coddington and Perkins9 reported on the basis of equilibrium dialysis results that, at low pH, HSA has 13 binding sites for Fe3+ in citrate complexes of composition [Fe2(C6H4O7)3]3−. These authors further demonstrated that acetylation of positively charged residues abolishes, whereas esterification of negatively charged residues enhances iron binding by HSA, and concluded that Fe binds to HSA in

the form of a citrate complex with net negative charge. In contrast, Silva and Hider12 reported an increase in iron binding to recombinant HSA after glycation of positively charged residues, but glycation with glucose or fructose introduces chelating functionalities that may bind iron directly. Interestingly, the iron binding capacity of native HSA at an iron−citrate ratio of 1:5 was half that at 1:1;9 this trend was reversed for esterified HSA, higher at 1:5 than at 1:1, findings consistent with binding of negatively charged iron−citrate complexes. Damage to DNA, lipids, and proteins caused by exposure to ferrous citrate has been reported as evidence of iron citrate redox activity. Gutteridge32 reported that ferric citrate prepared at acidic pH undergoes autoreduction to ferrous citrate upon standing, a reaction promoted by light.33,34 The resulting species rapidly autoxidizes at neutral pH to degrade deoxyribose, which is measured via thiobarbituric acid reactivity (TBAR) and benzoate hydroxylation; the degree of TBAR activity was modulated by the iron−citrate ratio and could be inhibited by inclusion of HO• and H2O2 scavengers. Baker and Gebicki35 demonstrated that Fe3+−citrate catalyzes conversion of O2•− to HO•. ESR evidence for in vivo formation of HO• in response to long-term dietary iron overload has been published.36 However, Sutton37 showed that production of CO2 from formate catalyzed by Fe2+−citrate is less wellprevented by isopropanol (100 mM) than CO2 formation catalyzed by other iron chelates, suggestive of a mechanism that does not necessarily involve free HO•. Minotti and Aust38 reported that citrate chelation of Fe2+ promotes autoxidation and subsequent lipid peroxidation in the presence of H2O2; Miller and Aust39 proposed that ascorbate exerts a pro-oxidant effect by promoting formation of a complex comprised of both Fe2+ and Fe3+ rather than by production of HO•. We set out to investigate the redox activity of iron citrate by electrochemistry, by the kinetics of its reaction with ascorbate (Hasc−), by ESR, and with Fenton chemistry by analyzing the products of reactions of Hasc− with iron citrate complexes in the presence of H2O2 and 4-hydroxybenzoate (4-HBA) as a reporter molecule for hydroxylation.40 We report a reduction potential of −0.03 V < E°′ > +0.01 V, i.e., ca. 0 V for the (Fe3+−cit/Fe2+−cit) couple. We find evidence for a rapid reorganization of ligands around Fe3+ as a first step in the reaction of Fe3+−citrate with Hasc−, followed by slow reduction to Fe2+−citrate. Mixed complexes of iron with citrate and ascorbate are likely formed, and the majority of ascorbyl radical (asc•−) formed by oxidation of Hasc− apparently remains sequestered and EPR silent by complexation with iron. The reduction of Fe3+−citrate by Hasc−, with H2O2 present, causes hydroxylation of 4-HBA that proceeds in three phases, the first phase of which is rapid and does not likely involve free HO•, the intermediate phase is slow and O2-dependent, and the final sluggish phase may be promoted by redox cycling of iron in mixed compexes with citrate, ascorbate, and aqua ligands.

■

METHODS

All solutions were prepared with Milli-Q (EMD Millipore, Darmstadt, DE) deionized water (18.2 MΩ). Stock solutions of iron(III) were prepared from Fe(Cl3)·6H2O (Merck, Darmstadt, DE) in 1% HCl; Fluka BioUltra grade L-ascorbic acid was obtained from Sigma-Aldrich AG (Buchs, CH). All other chemicals were analytical reagent grade and were used as supplied. Buffering agents that do not strongly coordinate iron(III) or iron(II) were selected: pyridine sulfonic acid (PSA) for pH 4.4, 2-(N-morpholino)ethanesulfonic acid (MES) for pH 5.4 and 6.4, and 3-(N-morpholino)propanesulfonic acid (MOPS) for pH 7.4. B

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology

from reactions performed under aerobic and anaerobic conditions to monitor the kinetics of hydroxylation. For the aerobic conditions, the yields of product are reported as the mean ± standard deviation from 2−5 measurements; for the reaction under anaerobic conditions, which was performed once only, an error of 11%, corresponding to the highest amount of error found in a data set under aerobic conditions, is reported. The effect of added HSA (300 μg/mL) was also qualitatively evaluated.

To prepare the iron(III)−citrate complex solutions, appropriate amounts of citric acid solution were added to 25 mM buffer with stirring. After stirring 5 min, FeCl3 was added from a stock solution, and the resulting pale yellow solution was allowed to equilibrate for 5 min. The pH was adjusted with CO2-free solutions of NaOH or HClO4, and the volume was adjusted with 25 mM buffer to give the desired final concentration of Fe3+. All preparations were carried out anaerobically in a glovebox in the dark; gastight syringes were used to transfer solutions; kinetics analyses were performed within 30 min of solution preparation unless otherwise noted. The optical spectra of the iron−citrate complexes were recorded with a SPECORD 200 (Analytik Jena, Jena DE) spectrophotometer in quartz cells with a 1 cm optical path. Cyclic Voltammetry. Electrochemical measurements were performed with an AMEL 2049 potentiostat and an AMEL 568 function generator (AMEL S.r.l., Milan IT) at ambient temperature and pressure; the auxiliary electrode was a glassy carbon electrode; the reference electrode was Ag/AgCl/KCl(3M) (Metrohm AG, Herisau CH). Voltammograms were obtained with gold disc (area 7 mm2), glassy carbon disc (area 3 mm2), and mercury hanging drop (screwdrive capillary, typical area 3 mm2) working electrodes from Metrohm. Solutions were prepared with 100 μM Fe3+ and 100 μM, 1 mM, or 10 mM citrate in 25 mM buffer (prepared 30 min prior to measurement) and purged with nitrogen under stirring for ca. 5 min before each scan. Potentials corrected for the Ag/AgCl/KCl(3M) electrode are reported as vs the normal hydrogen electrode (NHE). Kinetics. Rate constants for the oxidation of Hasc− by iron(III)− citrate were determined with an Applied Photophysics (Leatherhead, Surrey, UK) SX 17MV or SX 18MV stopped-flow spectrophotometer at pH 4.4, 5.4, 6.4, and 7.4 at 25 ± 0.1 °C and ambient pressure. The reaction was studied at Fe3+−citrate ratios 1:100 with Fe3+ at 50 μM in 25 mM MOPS at pH 7.4; Hasc−-induced reduction of the iron(III)− citrate complexes was followed at 325 nm. Reactions were started by rapidly mixing solutions of ascorbate with iron(III)−citrate solutions. Absorbance was recorded point-by-point by opening a shutter for 2.5 s, every 60 s, over a total reaction time of 995 s; this procedure was followed to minimize contributions from photoreduction.33 Reactions were performed in triplicate; pseudo-first-order rate constants were fit with Excel and are reported as the mean ± standard deviation. EPR. EPR spectroscopy was performed with a Bruker (Karlsruhe, Germany) EMX X-Band spectrometer; a quartz flat cell from WilmadLabGlass (Vineland, NJ, USA), 70 mm × 14 mm × 0.2 mm, was used, and the instrument was operated with ν = 9.76 GHz at T = 293 K. The yield of asc•− produced by addition of 4, 8, and 16 mM Hasc− to 50 μM Fe3+ and 10 mM citrate was determined in 25 mM MOPS at pH 7.4. After the reactants were mixed, spectra were recorded every 15 min for 60 min. Coarse quantification was obtained by measuring a calibration sample, provided by the spectrometer manufacturer, mounted, however, in a tube rather than in a flat cell. The instrument conditions, microwave power, receiver amplification, critical coupling of the resonator, and temperature, were the same during calibration as those used for the asc•− sample. Products of Reaction with 4-Hydroxybenzoic Acid. The capacity of iron citrate to promote hydroxylation via Fenton chemistry was tested with 4-HBA as a reporter molecule. Reactions were performed at room temperature under pseudo-first-order conditions with 22.5 μM Fe3+, 2 mM citrate, 0.2 mM Hasc−, and 2 mM 4-HBA in 1 mM bisTris buffer, pH 7.4, initiated by addition of H2O2 to 1 mM and quenched by adding desferrioxamine (dfo) to 0.2 mM to inhibit further redox reactivity of iron.41 Products were analyzed by reversedphase HPLC with a system comprised of Hewlett-Packard (Agilent) modules: HP-1050 autosampler, quaternary pump; HP-1100 diodearray detector; Agilent-1100 degasser. The column was from Macherey-Nagel: C-18ec, 150 mm × 4.5 mm, 5 μm particles; the mobile phase was 80% (v/v) 0.01 M H3PO4, 20% (v/v) CH3OH at 1 mL/min; detection was at 294 and 213 nm, and spectra from 200−400 nm were recorded for peaks of interest. Screening reactions (20 min) were performed under aerobic and anaerobic conditions, with Fe2+−cit and Fe3+−edta as positive controls and without Hasc− and H2O2 or with FeSO4 as negative controls. A series of timed aliquots was drawn

■

RESULTS Electrochemistry. Voltammograms recorded with the Au working electrode for 0.1 mM Fe3+ in 10 mM citrate (Fe3+−cit ratio 1:100) at pH 4.4, 5.4, 6.4, and 7.4 are shown in Figure 1A; voltammograms recorded with other Fe3+−cit ratios are shown in Figure S1 (Supporting Information). The major feature of the cathodic sweep is a continuous increase in current from ca.

Figure 1. Cyclic voltammetry of Fe3+−citrate; glassy carbon auxiliary electrode, Ag/AgCl/KCl(s) reference electrode; ambient temperature: (A) 100 μM Fe3+; Fe3+−citrate 1:100; Au working electrode; scan rate = 200 mVs−1; 25 mM MOPS (pH 7.4), MES (pH 6.4, 5.4), and PSA (pH 4.4); the trend shows poorer reversibility at lower pH; (B) Fe3+− citrate 1:1; C working electrode; scan rate = 100 mVs−1; 25 mM MOPS, pH 7.4; estimated upper limit of E°′ = 0.09 V vs NHE. 100 μM Fe3+; (C) Fe3+−citrate 1:1000; Hg hanging drop working electrode; scan rate =200 mVs−1; 25 mM MOPS, pH 7.4; estimated E°′ = −0.03 V vs NHE. C

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology −600 mV, with a peak that appears as a shoulder at ca. −1100 to −1500 mV; the anodic sweep is completely featureless at every pH. The findings are consistent with adsorption of polymeric Fe3+−cit at the electrode and irreversible reduction to Fe0. At pH 6.4 and 7.4, a weak second peak current at ca. −0.2 V might indicate the presence of some mononuclear [Fe3+(cit)2]5−, which, corrected for the Ag/AgCl/KCl (3M) electrode at 25°, is consistent with an electrode potential of ca. 0 V. The sloping shape of the voltammogram (Figure 1B) collected with a graphite electrode at pH 7.4 with a Fe3+−cit ratio of 1:1 is indicative of reduction of iron on the periphery of the polymeric form of Fe3+−cit, which is expected to be the major form present under conditions where the excess of citrate is +0.01 V). Kinetics. The dependence on [Hasc−1] of reduction of 3+ Fe −cit (1:100) at pH 4.4, 5.4, 6.4, and 7.4 is shown in Figure 2; additional data is shown in Figure S3. A transient blue color has been observed upon mixing Hasc− with Fe3+−citrate concentrations of 50 and 25 mM, respectively;44 as the concentrations used in our experiments were orders of magnitude lower, we did not observe any color development. The rate of disappearance of Fe3+−cit, measured at 325 nm, is proportional to the Hasc− concentration at the lower concentrations studied. At pH 4.4, there is pronounced inhibition of reduction at higher concentrations of Hasc−; this concentration-dependent inhibition is relaxed at pH 5.4 and 6.4 and apparently resolved at pH 7.4, where the dependence of the first-order rate constant on [Hasc−] is linear but does not pass through the origin. The second-order rate constants, ca. 2−8 M−1 s−1, from linear regression of all points in the pH 7.4 plot and the linear portions at low [Hasc−] from the other plots, are dependent on pH. The second-order rate constants are summarized in Table 1. The apparent initial absorbance values for Fe3+−cit, which are 5- to 10-fold lower than expected from recorded spectra, are collected in Table 2. EPR. Formation of asc•− produced by addition of 4, 8, and 16 mM Hasc− to 0.05 mM Fe3+ and 10 mM citrate at pH 7.4 upon reduction of Fe3+−citrate was observed by EPR (Figure 3). Samples were prepared and observed at 15 min intervals after mixing; the asc•− signal had reached 80% of the maximum level within 15 min; thus, we assume the steady-state condition. The apparent steady-state concentration of asc•− (inset) is ca. 30 nM, as follows from eq 1 d[asc•−] = 0 = k[Fe3 +:cit][Hasc−] − kD[asc•−]2 dt

Figure 2. Pseudo-first-order rate constants for the reaction of Fe3+− citrate 1:100, with 50 μM Fe3+ in 25 mM buffer, as a function of excess ascorbic acid/ascorbate ([Hasc−]) in (A) PSA pH 4.4, (B) MES pH 5.4, (C) MES pH 6.4, and (D) MOPS pH 7.4. Reactions were performed at room temperature, and rates were corrected for photoinduced autoreduction. Plots for pH 4.4, 5.4, and 6.4 indicate inhibition at higher [Hasc−].

Table 1. Second-Order Rate Constants k for the Linear Portions of Plots of Pseudo-First-Order Rate Constants kobs as a Function of [Hasc−] (Figure 2) for the Reaction of Fe3+−Citrate 1:100 ([Fe3+] = 50 μM) pH

k/M−1 s−1

4.4 5.4 6.4 7.4

7.8 5.9 5 3

± ± ± ±

0.1 0.2 2 3

where k is second-order rate constant for the reaction of Fe3+ with Hasc− and kD is the rate constant for disproportionation of asc•−, estimated as ca. 106 M−1 s−1 at neutral pH.45 On the basis of eq 1, we estimate the second-order rate constant k = 0.02 M−1 s−1 for reaction of Fe3+−cit with Hasc−; we note that this rate constant is 2 orders of magnitude lower than the rate of reduction of Fe3+ measured by stopped-flow spectroscopy. Spectra recorded in the absence of ascorbate show no signs of the characterisic two-line signal, and when Hasc− is measured in the absence of Fe3+ under aerobic conditions, we observe a steady signal for asc•− at ca. 0.5% the intensity of that obtained

(1) D

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology Table 2. Initial Absorbance at 325 nm (A0325) Expectedb and Observedc during Stopped-Flow Kinetics Experimentsa [Hasc−]/mM

A0325 c

pH

exp.

4.4 5.4 6.4 7.4

22.9 22.8 25.0 26.0

0.2

0.4

0.040 (2) 0.0314 (6) 0.0200 (8) 0.026 (2)

0.0415 (2) 0.032 (1) 0.0176 (2) 0.0251 (5)

0.8 0.0399 0.0323 0.0185 0.0241

1.6 (7) (2) (7) (2)

0.0368 (3) 0.021 (1) 0.022 (1) 0.0232 (1)

3.2 0.0335 0.0201 0.0204 0.0181

6.4 (1) (5) (4) (2)

0.0316 0.0169 0.0206 0.0163

(8) (1) (8) (3)

a

Uncertainty in the last digit (standard deviation) is given in parentheses. bTaken from recorded UV spectra. cDetermined by extrapolation of absorbance data to t = 0.

Figure 3. EPR spectra obtained upon mixing 0.05 mM Fe3+, 10 mM citrate with 0.4, 0.8, and 1.6 mM Hasc− at pH 7.4 showing formation of the ascorbyl radical (asc•−); the [asc•−] reaches a steady-state concentration of ca. 30 nM (inset).

in the presence of Fe3+. The spectra recorded with [Hasc−] < 0.4 mM were too noisy to be used for quantitative comparison. Products of Fenton Chemistry Reaction with 4Hydroxybenzoic Acid. The reduction of Fe3+−citrate by ascorbic acid was carried out in the presence of H2O2, presumably to form HO• by Fenton chemistry, with 4-HBA as a reporter molecule. 4-HBA reacts with HO•, or higher oxidation states of iron, to form a radical adduct that reacts further with Fe3+ or O2 to ultimately form a relatively stable hydroxylated initial product, 3,4-dihydroxybenzoic acid (3,4DHBA), which is further oxidized to 1,4-hydroquinone (1,4HQ);46 the results are similar to those found for Fenton oxidation of salicylate by Fe3+−edta.47 Preliminary screening experiments indicated that Fe3+−cit exhibits ca. 61% of the hydroxylating activity of Fe3+−edta. Products formed upon hydroxylation of 4-HBA formation in the presence of Fe3+−cit, Hasc−, and H2O2 were analyzed by HPLC. Concentrations of Fe3+ (22.5 μM) and citrate (2 mM) were chosen to ensure that [Fe(cit)2]5− was the predominant iron citrate species;43 0.2 mM Hasc− was present in ca. 10-fold excess over iron. Reactions were initiated by addition of 1 mM H2O2 and were quenched by adding 0.2 mM dfo, an effective chelator of iron that prevents redox-cycling.41 Single-time-point assays were performed to screen conditions for hydroxylation and for controls; addition of 300 μg/mL (ca. 5 μM) HSA was found to attenuate the hydroxylation to ca. half of the level found in the absence of HSA (Table S1). Chromatograms showing the analysis of aliquots over a time period of up to 19 h are shown in Figure 4A, together with spectra recorded during the separation. Figure 4B shows the time dependence of the 3,4-DHBA yield: hydroxylation of 4-HBA proceeds apparently in three phases: a rapid initial phase indicated by a nonzero y-axis intercept, an intermediate phase, and a slow

Figure 4. (A) Reversed-phase HPLC analysis of the aerobic hydroxylation of 2.23 mM 4-HBA (red) in the presence of 22.5 μM Fe3+, 2.23 mM citrate, and 0.23 mM Hasc−, at room temperature and in 1.12 mM bis-Tris at pH 7.4. The reaction was started by adding H2O2 to a final concentration of 1.12 mM and was stopped by quenching with 0.2 mM dfo. Chromatograms show time-dependent formation of 3,4-DHBA (blue) and HQ/BQ (green). (B) [3,4-DHBA] plotted as a function of time under aerobic and anaerobic conditions showing product formed in three phases: an initial O2-independent phase with nonzero intercept (shaded blue), an intermediate pseudofirst-order O2-dependent phase, and a latent O2-independent linear phase (shaded lilac, compressed scale). (C) Expanded view of the nonzero intercept of the initial O2-independent phase. (D) View of linear relationship of later time points.

linear phase during which hydroxylated product continues to be produced apparently indefinitely. Only the intermediate phase is sensitive to the presence of O2; the y-axis intercepts associated with the initial phase as well as the linear slopes of the slow phase are essentially the same whether the reaction is performed aerobically or anaerobically. The data points of the intermediate phase can be fit to a first-order rate law, corresponding to the formation of 3,4-DHBA, as shown by the solid lines; the corresponding first-order rate constants are 9.7 × 10−4 s−1 for the aerobic reaction and 2.7 × 10−3 s−1 for the anaerobic reaction. The rate constant under aerobic conditions is divided by the concentration of ascorbate (0.2 mM) to obtain an estimate (5 M−1 s−1) for the second-order rate constant, which is comparable to that obtained by stoppedE

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology flow at pH 7.4 (3 ± 3 M−1 s−1) for the reduction of Fe3+− citrate by Hasc−; k = 13 M−1 s−1 is estimated for the secondorder rate constant under anaerobic conditions. The yield of 3,4-DHBA during the intermediate phase under anaerobic conditions is 0.012 mM, equal to half of the initial concentration of Fe3+; the yield under aerobic conditions is 0.045 mM, twice the initial [Fe3+]. The early time points under both aerobic and anaerobic conditions have nonzero yintercepts; in both cases, the y-axis is intersected at ca. 0.01 mM product (pale blue zone in Figure 4B; shown expanded in Figure 4C). A linear reaction continues after the intermediate phase (pale violet zone in Figure 4B), under both aerobic and anaerobic conditions, at a rate of ca. 2.5 × 10−10 M s−1, from which a second-order rate constant k of 5 × 10−2 M−1 s−1 is estimated.

[OH−]. However, the shift of the cathodic peak to less negative potentials and the decrease in the peak maximum with increasing pH cannot be explained conventionally; if the peak corresponds to further reduction of Fe2+ formed at potentials above −1.1 V, then the deflection should be greater and it should migrate to more negative potentials with increasing pH, since Fe2+ is also stabilized by OH−. Also, if the iron species being reduced were freely diffusing in solution, then the slope of the current increase at very negative potentials would be expected to be less steep at higher pH, rather than steeper. The findings may be consistent with increased speciation of iron citrate as aggregates or polymers at higher pH,48 although at ≥20-fold excess citrate, most iron citrate is present as [Fe(cit)2]5− chelates.42,43 In the scan performed at pH 7.4 (Figure 1C), the small nonfaradaic signal observed at ca. −0.2 V is typical for irreversible adsorption49 and cannot be related to diffusion-limited electrolysis. The voltammogram in Figure 1B, obtained for Fe−cit 1:1 at pH 7.4 with a graphite electrode, is more suggestive of a reversible reaction and allows estimation of an upper limit for the electrode potential of the Fe3+−cit/ Fe2+−cit couple of +0.01 V vs NHE. At high excess citrate, the predominant species is the [Fe(cit)2]5− chelate,42,43 and the voltammogram in Figure 1C obtained for Fe−cit 1:1000 at pH 7.4 with a mercury electrode exhibits faradaic reversibility consistent with a freely diffusing species with an electrode potential of ca. −0.03 V. Thus, we can bracket the value for the electrode potential of low-molecular-weight iron citrate species at pH 7.4 at −0.03 V < E°′ < +0.01 V. Königsberger et al.50 reported calculated electrode potentials for the Fe3+−cit/Fe2+− cit couple based on titrations with a ternary Fe3+−Fe2+−cit system across the pH range 1−6; extrapolation of their reported values to pH 7.4 gives a E°′ of ca. −0.2 V. A value near 0 V for the electrode potential allows us to conclude that the reduction of iron citrate by ascorbate, which has an electrode potential E°′ (asc•−, H+/Hasc−) = +0.28 V51 at pH 7, is unfavorable by ca. 27 kJ/mol.51 The subsequent oxidation of Fe2+ citrate by H2O2 then, based on E°′ (H2O2, H+/HO•, H2O) = +0.39 V,52 has a Gibbs energy of ca. −38 kJ/mol. In early literature reports of Fe3+−citrate polarography,28 multiple half-waves and limiting currents that decrease with increasing pH at constant [Fe3+] were observed but not addressed by the authors; these observations indicate that a considerable fraction of Fe3+ was present in the form of large aggregates even at 250-fold citrate excess. The observed wave features can be explained by a diminished diffusion coefficient, which depends on particle size, i.e., degree of aggregation and viscosity. The viscosity is unlikely to change substantially under the experimental conditions described; signals like those in Figure 1A that are likely caused by polymeric species cannot be used to determine E°′ values. The polymeric Fe3+−citrate species observed in the in vitro polarographic experiments described here are, however, unlikely to be formed in vivo in the presence of a complex mixture of solutes. In stopped-flow studies of the reaction of ascorbate with iron citrate, we observe loss of ca. 80−90% of Fe3+−cit absorbance at 325 nm during the mixing time, followed by a slower absorbance decay; we ascribe both processes to reduction of Fe3+ to Fe2+. Keypour et al.53 reported that FeCl3 in pure water coordinates rapidly, within the mixing time of the instrument, with ascorbate at pH 6 to form mixed complexes of Fe3+ with Hasc− and solvent-derived ligands. These authors described decay of these complexes by two routes: (1) to a blue Fe2+− asc•− radical complex that rapidly decays via replacement of the

■

DISCUSSION In this study, we set out to describe a complex series of reactions involving iron citrate reduction by ascorbate as a model for redox activity of NTBI in diseases of iron overload. Each of the systems investigated, the electrochemistry of iron citrate, the reduction of Fe3+ by ascorbate, and Fenton chemistry driven hydroxylation of 4-HBA, considered separately has been scrutinized exhaustively in the literature and described as complex. Our attempt to elucidate the mechanisms governing the reactions at the confluence of these systems has yielded results no less complex. Our main findings include an estimate for the reduction potential of iron citrate under physiological conditions and evidence that a mixed iron− citrate−ascorbate charge-transfer complex reacts rapidly in the presence of H2O2 to hydroxylate organic substrates. Experiments were carried out under neutral to moderately acidic conditions; pH 7.4 was chosen for physiological relevance, and we avoided higher pH ranges to circumvent formation of inert iron(III) hydroxide compounds. As much of the earlier kinetics data had been collected under acid conditions where aqua Fe species are soluble, we chose to conduct experiments at pH 6.4, 5.4, and 4.4 to survey the pH range between physiological and acidic conditions. We chose to use Fe3+ at ca. 20 μM for relevance to the physiological range found in hemochromatosis; other components in the HPLC assay were adjusted accordingly to achieve 1:100 Fe−citrate and the necessary excess of reagents for pseudo-first-order conditions. Concentrations of iron used in other experiments, 50 μM for kinetics and EPR analyses and 100 μM for cyclic voltammetry, were adapted to be as low as possible for the given detection systems. Cyclic voltammograms of Fe3+−cit 1:100 obtained with the Au electrode at pH 4.4 to 7.4 (Figure 1A) exhibit variations consistent with changes in speciation. In the cathodic scan, the current increases continuously from ca. −0.6 V, with a shoulder that appears at ca. −1.3 (pH 7.4) to −1.6 V, depending on pH; the anodic scan is featureless. We deduce from the lack of anodic current that Fe3+ is completely and irreversibly reduced to Fe metal or to FeO, which likely forms a deposit on the Au electrode; Fe0 or FeO would be expected to reoxidize only at high overvoltage. Features of the reductive scan that vary as a function of pH are the onset potential and height of the shoulder peak, which corresponds to Fe3+ to Fe2+ reduction, and the steepness of the cathodic current at potentials lower than −1.6 V. The shift of the onset of the current in the cathodic scan to more negative potentials as a function of pH is expected, as Fe3+ is less susceptible to reduction at higher F

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology asc•− by solvent and (2) a relatively small fraction that remains Fe3+, coordinated to solvent at pH 6, or to neutral H2asc at low pH, that is reduced slowly to Fe2+. Similarly, Xu and Jordan54 reported that the reaction of ascorbic acid with aqueous Fe3+ under acidic conditions proceeds via a biphasic reaction, with the initial rapid phase (k = 5.5 × 103 M−1 s−1) assigned to formation of an iron ascorbate (Fe3+−Hasc−) complex upon substitution of Hasc− for water; Xu and Jordan54 also described inhibition by Fe2+ in aqueous solution. In our experiments, we observed pH-dependent inhibition by Hasc−. In Figure 5, we

Figure 6. Structures proposed for the components of a mixed chargetransfer complex of Fe3+ and Fe2+ with citrate and ascorbate.

detection of a low steady-state concentration of asc•− by EPR. Coordination of Hasc−, and, we assume, asc•−, to iron is monodentate,57 which may leave sites free for coordination of H2O2 to iron in the mixed complex. Xu and Jordan54 in their mechanism ascribe the second, slower phase of reaction of ascorbate with aqueous Fe3+ under acidic conditions to reversible electron transfer to form asc•−, which is subsequently further oxidized to dehydroascorbate (dhasc) by Fe(OH2)63+. The slower phase of the reaction is dependent on pH and [Hasc−] (Figure 2). At higher concentrations of ascorbate, there is apparent inhibition of Fe3+ reduction at pH 4.4, 5.4, and 6.4, with the greatest inhibition observed at the lowest pH; inhibition at pH 7.4 is not evident. The equilibria in Reactions 3 and 8 of Figure 5 would explain the observed [Hasc−]- and pH-dependent inhibition of the reduction reaction. Keypour et al.53 suggested that Fe3+ is stabilized by low pH and by the presence of higher concentrations of ascorbate. The linear parts of the plots of the pseudo-first-order rate constants determined by stopped-flow spectroscopy as a function of [Hasc−] (Figure 2) were used to estimate second-order rate constants of ca. 2−8 M−1 s−1 (Table 1) for reduction by Hasc− of Fe3+ to Fe2+. In Figure 2D, the plot of pseudo-first-order rate constants as a function of [Hasc−] at pH 7.4 is apparently linear with a nonzero intercept, which might be interpreted as reversible behavior. However, the kinetics data do not support the conclusion that the reaction is reversible. Also, the standard deviation associated with the linear plot for pH 7.4 is quite large, i.e., the data do not fit a simple linear relationship, which, we suspect, is due to differences in the mechanism at low vs high [Hasc−]. The very small amount of free asc•− detectable by ESR reaches a steady-state concentration of ca. 30 nM (Figure 3,

Figure 5. Mechanism proposed for the reduction of [Fe(cit)2]5− by Hasc− at pH 7.4.

propose a mechanism for reduction of Fe3+−cit based on our observations and informed by published work from others. The initial rapid loss of Fe3+−cit absorbance is indicative of substitution of Hasc− for citrate around Fe3+, Reaction 1, to form a mixed charge-transfer complex (Reaction 2) with both citrate and ascorbate as ligands around iron (Figure 6). Taqui Khan and Martell55 proposed a similar mechanism for formation of mixed Fe3+−ligand chelates in a pre-equilibrium step upon reaction with Hasc−, as did Martinez et al.56 for the oxidation of Hasc− by trisoxalatoferrate(III). The magnitude of the initial absorbance decrease at 325 nm, corresponding to reduction of 80−90% of the Fe3+ present, suggests to us that the electron being transferred (Reaction 2) resides more on iron than on ascorbate in the mixed complex, leading to an apparent lowering of the extinction coefficient. Keypour et al.53 also reported [Hasc−]-dependent effects on absorptivity of Fe−Hasc− complexes. We observe that the initial absorbance is decreased as a function of increasing [Hasc−] and pH (Table 1). The self-exchange reaction, as depicted in the boxed Reaction 2 of Figure 5, predicts the pHdependence of the initial [Fe3+]. Reaction 3 is the reduction of Fe3+ to Fe2+, for which we found a rate constant at pH 7.4 of ca. 3 M−1 s−1, we observed inhibition of Fe3+ reduction in the presence of relatively high concentrations of ascorbate. Reaction 5 corresponds to the G

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology inset). The absolute [asc•−] might be not entirely correct because of the difference in sample geometry between calibration standard and asc•− sample, but we believe that the order of magnitude is correct. The amount of radical detected in the EPR experiments, which is similar to levels determined in a rat model of oxidative stress,58 corresponds to asc•− that has escaped the inner coordination sphere of Fe (Figure 5, Reaction 5) and is a tiny fraction of the decrease in [Fe3+] detected by stopped-flow spectroscopy; the second-order rate constant k = 0.02 M−1 s−1 calculated for asc•− formation is ca. 200 times lower than that for reduction of Fe3+ measured by stopped-flow spectroscopy. Thus, the majority of asc•− formed remains sequestered and EPR silent in the inner coordination sphere of iron, a further indication that mixed iron−ascorbyl− citrate complexes are formed. The majority of ascorbate oxidation may be completed in a reaction between an Fe3+− ascorbate complex and a second Fe3+ complex, a quasidisproportionation, to form dehydroascorbate directly, in accord with the observations reported by Xu and Jordan.54 Box 4 of Figure 5 depicts the essentially equivalent pairs of ascorbyl radical species available for recombination by disproportionation. The relative stability constants reported for iron with citrate43,50 and ascorbate59 and the relative concentrations of the ligands would dictate that citrate displaces ascorbate/dehydroascorbate to form Fe2+(cit)2 (Figure 5, Reactions 6 and 7). However, equilibrium to form Fe2+(Hasc−)2 (Figure 5, Reaction 8) is also possible and may be a gateway to formation of relatively kinetically inert aqua iron species. When the reaction of iron citrate with Hasc− is carried out in the presence of H2O2 and the reporter molecule 4-HBA, Fenton chemistry (Figure 5, Reaction 9) leads to formation of HO•, and hydroxylation of 4-HBA proceeds according to a multistep mechanism essentially as described by Duesterberg et al.46 Addition of HO• to 4-HBA (Figure 5, Reaction 10) produces the carboxylated dihydroxycyclohexadienyl radical intermediate (HO-4-HBA•). Production of 3,4-DHBA can occur by disproportionation of the radical intermediate (Reaction 11), oxidation by a second equivalent of Fe3+47 (Reaction 12), or oxidation by O2 (Reaction 13). Our product analyses indicate that formation of the first stable hydroxylation product, 3,4-DHBA, occurs in three distinct phases (Figure 4). The first phase, detected as a nonzero y-intercept indicating formation of ca. 0.01 mM product (Figure 4C), corresponding to 44% yield relative to the initial amount of Fe3+, is independent of the presence of O2. As the reaction was initiated by addition of H2O2 to a premixture of Fe3+, citrate, and ascorbate, we may assume that essentially all iron is present as Fe2+, primed to react upon addition of H2O2, with the relatively rapid Fenton reaction being ratelimiting. Given the concentrations of the reactants and the published rate constants for reaction of iron(II) citrate with H2O2,60 we expect 90% of the HO•, formed stoichiometrically from Fe2+, to react with 4-HBA to form the HO-4-HBA• radical intermediate. This intermediate disproportionates to 3,4DHBA and 4-HBA, accounting for the initial yield of 44% 3,4-DHBA. The half-life of the Fenton reaction under the conditions described for Figure 4B is ca. 0.14 s; thus, this initial reaction was essentially over before the first aliquot was removed and quenched for analysis. Upon depletion of the initial pool of hydroxylating species, slow reduction of Fe3+−citrate becomes rate limiting, accounting for the second phase of product formation. This

intermediate phase is O2-dependent. The product yields plotted as a function of time (Figure 4B) for this phase under aerobic and anaerobic conditions can be fit to a first-order rate law with constants of 9.7 × 10−4 s−1 and 2.7 × 10−3 s−1, respectively. The corresponding second-order rate constant under aerobic conditions, 4.5 M−1 s−1, is the same order of magnitude as that determined by stopped-flow (3 M−1 s−1). The yield of product formed during this phase in the absence of O2 corresponds to half of the amount of iron in the system; in the presence of O2, the yield of product formed is twice the amount of iron in the system. The yields of 3,4-DHBA found during this phase are predicted by the mechanism proposed by Duesterberg et al.46 The intermediate phase continues for ca. 1 h and is followed by slow hydroxylation that continues apparently linearly under both aerobic and anaerobic conditions with a rate constant k of ca. 0.05 M−1 s−1; this very slow formation of 3,4-DHBA continues for at least 18 h. We suggest that this slow phase reflects exchange of citrate, ascorbate, and HO− ligands around iron and that redox reactions involving such mixed complexes may be sluggish. Duesterberg et al.46 did not observe an initial burst of hydroxylated product in their kinetics studies, which were performed under different conditions with Fe(ClO4)2 at pH 3; the mechanism of hydroxylation of 4-HBA proposed by these authors is complex and involves formation of many products in addition to the initial product, 3,4-DHBA. Further oxidation of 3,4-DHBA by O2 (Reaction 14) produces the decarboxylation product 1,4-hydroquinone (1,4-HQ), which may, in turn, be further oxidized to 1,4-benzoquinone (1,4-BQ) and ringopened products. Our HPLC assay (Figure 4A) indicates that at least one of these products, consistent with 1,4-HQ, is formed when O2 is present (Figure 5, Reaction 14), as are other minor products detected but not analyzed; 1,4-HQ and 1,4-BQ either coelute or interconvert under the conditions of the assay. According to Duesterberg et al.,46 1,4-HQ is formed together with 3,4-DHBA; in our timed assays, 1,4-HQ is formed subsequent to 3,4-DHBA. The formation of the quinone product reached 0.034 mM and was strictly dependent on the presence of O2. Since all secondary products are formed at the expense of 3,4-DHBA, a more accurate estimate of the yield of 3,4-DHBA at the end of the O2-dependent phase (6 h) is 0.085 mM, higher than the 200% yield relative to iron predicted by the Duesterberg et al. mechanism.46 We can calculate theoretical yields of hydroxylated product for limiting scenarios, where a minimum amount of product would be formed under conditions of no redox cycling by iron citrate and a maximum amount of product would be formed when redox cycling exhausts the supply of electrons from ascorbate. In the minimum case, asc•− does not reduce Fe3+− cit, and two HO-4-HBA• radicals disproportionate to 3,4dHBA and 4-HBA, which would lead to formation of 3,4-dHBA at a concentration corresponding to half that of the initial concentration of Fe3+−cit, ca. 11 μM, as observed at the end of the initial phase and for the intermediate phase under anaerobic conditions. In the maximum case, each equivalent of Hasc− yields 2 equiv of 3,4-dHBA, ca. 0.4 mM. Over a period of 19 h in the absence of O2, we find ca. 35 μM, and in presence the of O2, ca. 66 μM 3,4-DHBA, only ca. 16% of the maximum possible yield. We do not take into consideration the possibility of reduction of Fe3+−cit complexes by O2•− as evidence obtained by pulse radiolysis for this reaction (Merkofer, Nauser, H

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology

that, at lower concentrations, ascorbate promotes the reorganization of ligands around the iron center and initiates reduction of Fe3+ to Fe2+. The mixed Fe2+ complex with citrate and ascorbate promotes rapid hydroxylation of the reporter molecule; the subsequent slower hydroxylation reactions may also be physiologically important.

Koppenol, unpublished) indicates that this reaction is too slow at pH 7 to play a role. Since it is widely accepted that serum albumin comprises part of NTBI, we investigated the effect of its presence on the hydroxylation reaction. The yield of 3,4-DHBA with HSA present at ca. 5 μM was ca. half that without HSA, and no further reaction to form 1,4-HQ/BQ was observed. HSA may be acting to scavenge HO• or HO-4-HBA• intermediates or interacting with the iron−citrate complex in a way that blocks reaction with H2O2 or slows the release of product. The lack of formation of quinone products may indicate that HSA binds 3,4-DHBA to prevent further oxidation. Although iron is known to bind to serum albumin in the absence of citrate, and it is conceivable that Fe−HSA without citrate could react via Fenton chemistry to form hydroxylated products, no product was observed when citrate was excluded. We assume that the complexation of iron by citrate and albumin and the association of iron citrate with serum albumin are in dynamic flux. Although the concentration of albumin in plasma is ca. 0.5 mM, compared to citrate at ca. 0.1 mM, albumin probably does not play a protective role in vivo by out competing citrate binding. Furthermore, hypoalbuminemia is likely to be found in association with liver cirrhosis61 and inflammation.62 Taken together, the initial burst of product, the greater than predicted amount of hydroxylated product under aerobic conditions, and the continuing formation of hydroxylated product beyond the maximum yield expected based on the amount of iron present support an interpretation of redox activity and cycling by iron citrate upon reduction by ascorbate. Any of these reaction phases, however, is physiologically relevant only in cellular compartments where catalase is absent. The iron citrate of NTBI is, thus, capable of ascorbatepromoted redox cycling and may be responsible for cellular damage via hydroxylation of biomolecules in the presence of H2O2. The conditions used to produce the data in Figure 4 approximate physiological conditions in terms of concentrations of iron and citrate for a hemochromatosis patient (10 μM Fe(cit)2]5−) with a typical plasma concentration of 100 μM Hasc−, for which the O2-dependent rate of hydroxylation is expected to be on the order of nanomolar/second. The rate due to presumed redox cycling of the iron is an order of magnitude slower, but it was preceded by an apparent rapid hydroxylation that might be expected to produce micromolar levels of hydroxylated product under physiological conditions upon exposure to H2O2. There have been reports of oxidation of hepatic lipids and proteins in animal models of iron overload,63−69 resulting from ascorbate prophylaxis applied during obligatory ischemia− reperfusion in vascular surgery, attributed to increased peripheral concentrations of lipid hydroperoxides due to ironinduced oxidative lipid damage via a Fenton-type reaction. Others, however, found neither lipid nor protein oxidation in studies with human plasma or any other evidence to indicate pro-oxidant activity of ascorbate promoted by metal ions and H2O2 in plasma.70,71 The relative pro- and antioxidant properties of ascorbate have been discussed: 72,73 low concentrations of ascorbate are generally considered to be pro-oxidative, whereas antioxidant activity is observed with relatively high ascorbate concentrations, and it has been suggested that vitamin C supplementation of patients with iron overload disease could be detrimental. Harmful effects of ascorbate in iron overload disease may also be attributed to enhanced uptake of iron by ascorbate.74 Our evidence suggests

■

ASSOCIATED CONTENT

S Supporting Information *

Additional voltammograms and kinetics data are shown in Figures S1−S3, and product analyses is given in Table S1. This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Addresses †

(F.I.A.) GSK Consumer Healthcare, St. Georges Avenue, Weybridge, Surrey, KT13 0DE, United Kingdom. ‡ (P.L.B.) Foundation for Information Technology in Society (IT’IS), Zeughausstrasse 43, CH-8004 Zurich, Switzerland. Funding

This research was supported by the Schweizerische Nationalfonds (SNF) and the Swiss Federal Institute of Technology (ETH). Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Professor R. Hider is thanked for discussions. ABBREVIATIONS 1,4-BQ, 1,4-benzoquinone; 1,4-HQ, 1,4-hydroquinone; 3,4DHBA, 3,4-dihydroxybenzoic acid; 4-HBA, 4-hydroxybenzoate; asc•−, ascorbyl radical; cit, citrate; dhasc, dehydroascorbate; dfo, desferrioxamine; Hasc−, monohydrogenascorbate; HO-4HBA•, carboxylated dihydroxycyclohexadienyl radical intermediate; HSA, human serum albumin; NHE, normal hydrogen electrode; NTBI, nontransferrin-bound iron; MES, 2-(Nmorpholino)ethanesulfonic acid; MOPS, 3-(N-morpholino)propanesulfonic acid; PSA, pyridinesulfonic acid; TBAR, thiobarbituric acid reactivity

■

REFERENCES

(1) Ganz, T. (2012) Macrophages and systemic iron homeostasis. J. Innate Immun. 4, 446−453. (2) Brissot, P., Ropert, M., Le Lan, C., and Loréal, O. (2012) Nontransferrin bound iron: a key role in iron overload and iron toxicity. Biochim. Biophys. Acta, Gen. Subj. 1820, 403−410. (3) Sohn, Y. S., Ghoti, H., Breuer, W., Rachmilewitz, E., Attar, S., Weiss, G., and Cabantchik, Z. I. (2012) The role of endocytic pathways in cellular uptake of plasma non-transferrin iron. Haematologica 97, 670−678. (4) Parkes, H. G., Allen, R. E., Furst, A., Blake, D. R., and Grootveld, M. C. (1991) Speciation of non-transferrin-bound iron ions in synovial fluid from patients with rheumatoid arthritis by proton nuclear magnetic resonance spectroscopy. J. Pharm. Biomed. Anal. 9, 29−32. (5) Nischwitz, V., Berthele, A., and Michalke, B. (2008) Speciation analysis of selected metals and determination of their total contents in paired serum and cerebrospinal fluid samples: an approach to investigate the permeability of the human blood-cerebrospinal fluidbarrier. Anal. Chim. Acta 627, 258−269. I

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology (6) Grootveld, M., Bell, J. D., Halliwell, B., Aruoma, O. I., Bomford, A., and Sadler, P. J. (1989) Non-transferrin-bound iron in plasma or serum from patients with idiopathic hemochromatosis. J. Biol. Chem. 264, 4417−4429. (7) Hershko, C., Graham, G., Bates, G. W., and Rachmilewitz, E. A. (1978) Non-specific serum iron in thalassaemia: an abnormal serum iron fraction of potential toxicity. Br. J. Haematol. 40, 255−263. (8) Singh, S., Hider, R. C., and Porter, J. B. (1990) A direct method for quantification of non-transferrin-bound iron. Anal. Biochem. 186, 320−323. (9) Coddington, A., and Perkins, D. J. (1960) The binding of Fe+++ to native and chemically modified human serum albumin in the presence of sodium citrate. Biochim. Biophys. Acta 44, 361−363. (10) Løvstad, R. A. (1993) Interaction of serum albumin with the Fe(III)−citrate complex. Int. J. Biochem. 25, 1015−1017. (11) Evans, R. W., Rafique, R., Zarea, A., Rapisarda, C., Cammack, R., Evans, P. J., Porter, J. B., and Hider, R. C. (2008) Nature of nontransferrin-bound iron: studies on iron citrate complexes and thalassemic sera. J. Biol. Inorg. Chem. 13, 57−74. (12) Silva, A. M. N., and Hider, R. C. (2009) Influence of nonenzymatic post-translation modifications on the ability of human serum albumin to bind iron. Implications for non-transferrin-bound iron speciation. Biochim. Biophys. Acta 1794, 1449−1458. (13) Hastings, A. B., and Van Slyke, D. D. (1922) The determination of the three dissociation constants of citric acid. J. Biol. Chem. 53, 269− 276. (14) Francis, A. J., and Dodge, C. J. (1993) Influence of complex structure on the biodegradation of iron−citrate complexes. Appl. Environ. Microbiol. 59, 109−113. (15) Silva, A. M. N., Kong, X. L., and Hider, R. C. (2009) Determination of the pKa value of the hydroxyl group in the αhydroxycarboxylates citrate, malate and lactate by 13C NMR: implications for metal coordination in bilogical systems. BioMetals 22, 771−778. (16) Martin, R. B. (1961) A complete ionization scheme for citric acid. J. Phys. Chem. 65, 2053−2055. (17) Hamm, R. E., Shull, C. M., Jr., and Grant, D. M. (1954) Citrate complexes with iron(II) and iron(III). J. Am. Chem. Soc. 76, 2111− 2114. (18) Bobtelsky, M., and Jordan, J. (1947) The structure and behavior of ferric tartrate and citrate complexes in dilute solutions. J. Am. Chem. Soc. 69, 2286−2290. (19) Timberlake, C. F. (1964) Iron−malate and iron−citrate complexes. J. Chem. Soc., 5078−5085. (20) Shweky, I., Bino, A., Goldberg, D., and Lippard, S. J. (1994) Syntheses, structures and magnetic properties of two dinuclear iron(III) citrate complexes. Inorg. Chem. 33, 5161−5162. (21) Matzapetakis, M., Raptopoulou, C. P., Tsohos, A., Papaefthymiou, V., Moon, N., and Salifoglou, A. (2009) Synthesis, spectroscopic and structural characterization of the first mononuclear, water soluble iron−citrate complex, (NH4)5Fe(C6H4O7)2·2H2O. J. Am. Chem. Soc. 120, 13266−13267. (22) Strouse, J., Layten, S. W., and Strouse, C. E. (1977) Structural studies of transition metal complexes of triionized and tetraionized citrate. Models for the coordination of the citrate ion to transition metal ions in solution and at the active site of aconitase. J. Am. Chem. Soc. 99, 562−572. (23) Bino, A., Shweky, I., Cohen, S., Bauminger, E. R., and Lippard, S. J. (1998) A novel nonairon(III) citrate complex: a “ferric tripledecker”. Inorg. Chem. 37, 5168−5172. (24) Gautier-Luneau, I., Fouquard, C., Merle, C., Pierre, J. L., and Luneau, D. (2001) Synthesis, structure and magnetic properties of a novel octairon(III) citrate complex. J. Chem. Soc., Dalton Trans., 2127− 2131. (25) Pierre, J. L., and Gautier-Luneau, I. (2000) Iron and citric acid: a fuzzy chemistry of ubiquitous biological relevance. BioMetals 13, 91− 96. (26) Gautier-Luneau, I., Merle, C., Phanon, D., Lebrun, C., Biaso, F., Serratrice, G., and Pierre, J.-L. (2005) New trends in the chemistry of

iron(III) citrate complexes: correlations between X-ray structures and solution species probed by electrospray mass spectrometry and kinetics of iron uptake from citrate by iron chelators. Chem.−Eur. J. 11, 2207−2219. (27) Lingane, J. J. (1946) Polarographic investigation of oxalate, citrate and tartrate complexes of ferric and ferrous Iron. J. Am. Chem. Soc. 68, 2448−2453. (28) Meites, L. (1951) Polarographic studies of metal complexes. V. The cadmium(II), zinc(II) and iron(III) citrates. J. Am. Chem. Soc. 73, 3727−3731. (29) Martin, R. B. (1986) Citrate binding of Al3+ and Fe3+. J. Inorg. Biochem. 28, 181−187. (30) Graham, G., Bates, G. W., Rachmilewitz, E. A., and Hershko, C. (1979) Nonspecific serum iron in thalassemia: quantitation and chemical reactivity. Am. J. Hematol. 6, 207−217. (31) Hider, R. C., Silva, A. M. N., Podinovskaia, M., and Ma, Y. (2010) Monitoring the efficiency of iron chelation therapy: the potential of nontransferrin-bound iron. Ann. N.Y. Acad. Sci. 1202, 94− 99. (32) Gutteridge, J. M. C. (1991) Hydroxyl radical formation from the auto-reduction of a ferric citrate complex. Free Radical Biol. Med. 11, 401−406. (33) Abrahamson, H. B., Rezvani, A. B., and Brushmiller, J. G. (1994) Photochemical and spectroscopic studies of complexes, of iron(III) with citric acid and other carboxylic acids. Inorg. Chim. Acta 226, 117− 127. (34) Benrath, A. (1910) Ü ber reine und kombinierte photochemische Reaktionen. Z. Phys. Chem. 74, 115−124. (35) Baker, M. S., and Gebicki, J. M. (1986) The effect of pH on yields of hydroxyl radicals produced from superoxide by potential biological iron chelators. Arch. Biochem. Biophys. 246, 581−588. (36) Kadiiska, M. B., Burkitt, M. J., Xiang, Q.-H., and Mason, R. P. (1995) Iron supplementation generates hydroxyl radical in vivo. An ESR spin-trapping investigation. J. Clin. Invest. 96, 1653−1657. (37) Sutton, H. C. (1985) Efficiency of chelated iron compounds as catalysts for the Haber−Weiss reaction. J. Free Radicals Biol. Med. 1, 195−202. (38) Minotti, G., and Aust, S. D. (1987) An investigation into the mechanism of citrate−Fe2+ dependent lipid peroxidation. Free Radical Biol. Med. 3, 379−387. (39) Miller, D. M., and Aust, S. D. (1989) Studies of ascorbatedependent, iron-catalyzed lipid peroxidation. Arch. Biochem. Biophys. 271, 113−119. (40) Ste-Marie, L., Boismenu, D., Vachon, L., and Montgomery, J. (1996) Evaluation of sodium 4-hydroxybenzoate as an hydroxyl radical trap using gas chromatography mass spectrometry and high-performance liquid chromatography with electrochemical detection. Anal. Biochem. 241, 67−74. (41) Merkofer, M., Kissner, R., Hider, R. C., Brunk, U. T., and Koppenol, W. H. (2006) Fenton chemistry and iron chelation under physiological relevant conditions: electrochemistry and kinetics. Chem. Res. Toxicol. 19, 1263−1269. (42) Spiro, T. G., Bates, G., and Saltman, P. (1967) The hydrolytic polymerization of ferric citrate. II. The influence of excess citrate. J. Am. Chem. Soc. 89, 5559−5562. (43) Silva, A. M. N., Kong, X. L., Parkin, M. C., Cammack, R., and Hider, R. C. (2009) Iron(III) citrate speciation in aqueous solution. Dalton Trans., 8616−8625. (44) Laurence, G. S., and Ellis, K. J. (1972) The detection of a complex intermediate in the oxidation of ascorbic acid by ferric ion. J. Chem. Soc., Dalton Trans., 1667−1670. (45) Bielski, B. H. J., Allen, A. O., and Schwarz, H. A. (1981) Mechanism of disproportionation of ascorbate radicals. J. Am. Chem. Soc. 103, 3516−3518. (46) Duesterberg, C. K., and Waite, T. D. (2007) Kinetic modeling of the oxidation of p-hydroxybenzoic acid by Fenton’s reagent: implications of the role of quinones in the redox cycling of iron. Environ. Sci. Technol. 41, 4103−4110. J

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology (47) Maskos, Z., Rush, J. D., and Koppenol, W. H. (1990) The hydroxylation of the salicylate anion by a Fenton reaction and gammaradiolysis: a consideration of the respective mechanisms. Free Radical Biol. Med. 8, 153−162. (48) Spiro, T. G., Pape, L., and Saltman, P. (1967) The hydrolytic polymerization of ferric citrate. I. The chemistry of the polymer. J. Am. Chem. Soc. 89, 5555−5559. (49) Laviron, E. (1974) Adsorption, autoinhibition and autocatalysis in polarography and in linear potential sweep voltammetry. J. Electroanal. Chem. Interfacial Electrochem. 52, 355−393. (50) Königsberger, L.-C., Königsberger, E., May, P. M., and Hefter, G. T. (2000) Complexation of iron(III) and iron(II) by citrate. Implications for iron speciation in blood plasma. J. Inorg. Biochem. 78, 175−184. (51) Williams, N. H., and Yandell, J. K. (1982) Outer-sphere electron transfer reactions of ascorbate anions. Aust. J. Chem. 35, 1133−1144. (52) Koppenol, W. H., Stanbury, D. M., and Bounds, P. L. (2010) Electrode potentials of partially reduced oxygen species, from dioxygen to water. Free Radical Biol. Med. 49, 317−322. (53) Keypour, H., Silver, J., Wilson, M. T., and Hamed, M. Y. (1986) Studies on the reactions of ferric iron with ascorbic acid. A study of solution chemistry using Mössbauer spectroscopy and stopped-flow techniques. Inorg. Chim. Acta 125, 97−106. (54) Xu, J., and Jordan, R. B. (1990) Kinetics and mechanism of the reaction of aqueous iron(III) with ascorbic acid. Inorg. Chem. 29, 4180−4184. (55) Taqui Khan, M. M., and Martell, A. E. (1967) Metal ion and metal chelate catalyzed oxidation of ascorbic acid by molecular oxygen. II. Cupric and ferric chelate catalyzed oxidation. J. Am. Chem. Soc. 89, 7104−7111. (56) Martinez, P., Zuluaga, J., Uribe, D., and van Eldik, R. (1987) Oxidation of L-ascorbic acid by trisoxalatoferrate(III) in aqueous solution. Kinetic and spectroscopic evidence for the formation of an intermediate species. Inorg. Chim. Acta 136, 11−16. (57) Plug, C. M., Dekker, D., and Bult, A. (1984) Complex stability of ferrous ascorbate in aqueous solution and its significance for iron absorption. Pharm. Weekbl., Sci. Ed. 6, 245−248. (58) Galleano, M., Aimo, L., and Puntarulo, S. (2002) Ascorbyl radical/ascorbate ratio in plasma from iron overloaded rats as oxidative stress indicator. Toxicol. Lett. 133, 193−201. (59) Gorman, J. E., and Clydesdale, F. M. (1983) The behavior and stability of iron−ascorbate complexes in solution. J. Food Sci. 48, 1217−1220. (60) Rush, J. D., Maskos, Z., and Koppenol, W. H. (1990) Reactions of iron(II)nucleotide complexes with hydrogen peroxide. FEBS Lett. 261, 121−123. (61) Valberg, L. S., and Ghent, C. N. (1985) Diagnosis and management of hereditary hemochromatosis. Annu. Rev. Med. 36, 27− 37. (62) Walter, P. B., Fung, E. B., Killilea, D. W., Jiang, Q., Hudes, M., Madden, J., Porter, J., Evans, P., Vichinsky, E., and Harmatz, P. (2006) Oxidative stress and inflammation in iron-overloaded patients with βthalassaemia or sickle cell disease. Br. J. Haematol. 135, 254−263. (63) Bacon, B. R., Park, C. H., Brittenham, G. M., O’Neill, R., and Tavill, A. S. (1985) Hepatic mitochondrial oxidative metabolism in rats with chronic dietary iron overload. Hepatology 5, 789−797. (64) Bacon, B. R., O’Neill, R., and Park, C. H. (1986) Iron-induced peroxidative injury to isolated rat hepatic mitochondria. J. Free Radicals Biol. Med. 2, 339−347. (65) Britton, R. S., Ferrali, M., Magiera, C. J., Recknagel, R. O., and Bacon, B. R. (1990) Increased prooxidant action of hepatic cytosolic low-molecular-weight iron in experimental iron overload. Hepatology 11, 1038−1043. (66) Poli, G. (2000) Pathogenesis of liver fibrosis: role of oxidative stress. Mol. Aspects Med. 21, 49−98. (67) Sochaski, M. A., Bartfay, W. J., Thorpe, S. R., Baynes, J. W., Bartfay, E., Lehotay, D. C., and Liu, P. P. (2002) Lipid peroxidation and protein modification in a mouse model of chronic iron overload. Metabolism 51, 645−651.

(68) Petrak, J., Myslivcova, D., Man, P., Cmejla, R., Cmejlova, J., Vyoral, D., Elleder, M., and Vulpe, C. D. (2007) Proteomic analysis of hepatic iron overload in mice suggests dysregulation of urea cycle, impairment of fatty acid oxidation, and changes in the methylation cycle. Am. J. Physiol.: Gastrointest. Liver Physiol. 292, G1490−G1498. (69) Bailey, D. M., Raman, S., McEneny, J., Young, I. S., Parham, K. L., Hullin, D. A., Davies, B., McKeeman, G., McCord, J. M., and Lewis, M. H. (2006) Vitamin C prophylaxis promotes oxidative lipid damage during surgical ischemia−reperfusion. Free Radical Biol. Med. 40, 591− 600. (70) Berger, T. M., Polidori, M. C., Dabbagh, A., Evans, P. J., Halliwell, B., Morrow, J. D., Roberts, L. J., and Frei, B. (1997) Antioxidant activity of vitamin C in iron-overloaded human plasma. J. Biol. Chem. 272, 15656−15660. (71) Suh, J., Zhu, B. Z., and Frei, B. (2003) Ascorbate does not act as a pro-oxidant towards lipids and proteins in human plasma exposed to redox-active transition metal ions and hydrogen peroxide. Free Radical Biol. Med. 34, 1306−1314. (72) Buettner, G. R., and Jurkiewicz, B. A. (1996) Chemistry and biochemistry of ascorbic acid, in Handbook of Antioxidants (Cadenas, E., Packer, L., Eds.) Chapter 5, Marcel Dekker, Inc., New York. (73) Buettner, G. R., and Jurkiewicz, B. A. (1996) Catalytic metals, ascorbate and free radicals: combinations to avoid. Radiat. Res. 145, 532−541. (74) Gerster, H. (1999) High-dose vitamin C: a risk for persons with high iron stores? Int. J. Vitam. Nutr. Res. 69, 67−82.

K

DOI: 10.1021/tx500377b Chem. Res. Toxicol. XXXX, XXX, XXX−XXX