James D. Rush and Willem H. Koppenol. Department of Chemistry ...... G. W. Winston, D. E. Feierman, and A. I. Cederbaum, Arch. Biochem. Biophys. 232, 377 ...
The Reaction between Ferrous Polyaminocarboxylate Complexes and Hydrogen Peroxide: An Investigation of the Reaction Intermediates by Stopped Flow Spectrophotometry
James D. Rush and Willem H. Koppenol Department
of Chemistry, University of Maryland Baltimore County
ABSTRACT The reactions of Fe(II)EDTA, have been investigated. II,
Fe(II)DTPA, and Fe(II)HEDTA with hydrogen peroxide near neutral pH All these reactions have been assumed to proceed through an active intermediate,
Fe(II)pac + H202 2 I, I,+ Fe(II)pac “5 2Fe(III)pac + 20Hwhere pat is one of the three polyaminocarboxylates mentioned above. 11, whether *OH radical or an iron complex, reacts with ethanol, formate, and other scavengers at rates relative to /rz that, with the exception of t-butanol and benzoate, are similar, but not identical, to those expected for the *OH radical. In contrast, at pH 3, in the absence of ligands the reaction of 1, with Fe 2+ was inhibited by ethanol and t-butanol and the reactivity of I1 towards these two scavengers relative to ferrous ion is identical to that exhibited by the hydroxyl radical. When pat = HEDTA, the intermediate of the first reaction reacts with formate ion to form the ferrous HEDTA ligand radical complex, which is characterized by absorption maxima at 295 nm (E = 2,640 M-l cm-‘) and 420 nm (e = 620 M-l cm- ‘). For the reaction of Fe(II)HEDTA with H202,
Address reprint requests to Dr. W. H. Koppenol, Department of Chemistry, University Baltimore County, 5401 Wilkens Avenue, Catonsville, Maryland 21228, U.S.A. Journal of Inorganic 0
Biochemistry 29, 199-215 (1987) 1987 Elsevier Science Publishing Co., Inc., 52 Vanderbilt Ave., New York, NY 10017
of Maryland,
199 0162-0134/87/$3.50
200
J. D. Rush and W. H. Koppenol
the following mechanism
is proposed:
Fe(II)HEDTA Fe(II)(H202)HEDTA
+ H20Z “2 Fe(II)(H,Oz)HEDTA _
+ Fe(II)HEDTA -“%?
[Fe(III)HEDTA(OH)]12
(Fe(III)HEDTA(OH)]22- “2 2Fe(III)HEDTA(OH) where k,, = 4.2
x
lo4 M-’
set’
and
k19 = 5 -t 0.2 set-‘.
INTRODUCTION The oxidation of ferrous complexes by hydrogen peroxide has been cited as a source of hydroxyl radical that can cause damage in living cells [l-4]. However, the evidence that hydroxyl radicals are formed under conditions that exist biologically is indirect. In model systems where the decomposition of hydrogen peroxide is catalyzed by ferric EDTA or other such polyaminocarboxylate (pat) complexes, the ferric species is reduced by a biological reductant, e.g., 02- or ascorbate, and then reoxidized by H202 to form either *OH radicals or another oxidizing species frequently identified as ferryl, Fe(IV)02+, or ferrous-hydrogen peroxide. FeO”+ as an alternative to the hydroxyl radical was first proposed by Bray and Gorin [5]. Evidence for this species in acetonitrile was obtained by Groves and co-workers [6]. Ferry1 porphyrins are formed during the catalytic cycles of catalase and peroxidases [7]. Recently, a higher oxidation state of iron has been identified in the reaction of hydrogen peroxide with a water-soluble ferric porphyrin [8]. All possible intermediates are here referred to as I,, the product of reaction (1): Fe(II)pac + Hz02 + I, I, + Fe(II)pac
-+ 2Fe(III)pac
+ 20H-
I, + RH -+ Fe(III)pac + Hz0 + OH- + R *
(1) (2) (3)
In reaction (3), RH is a scavenging molecule, which under conditions where reaction (2) is unlikely to occur, such as in biological conditions. is the substrate for I,. Ethanol, I-butanol, and formate are commonly used as scavengers in biochemical model systems. In the present study the effects of various .OH radical scavengers on the overall stoichiometry of Fe(II1) production expressed by reactions (l)-(3) are measured for the HEDTA, EDTA, and DTPA complexes of iron. The values of krel = kJk2 thus obtained are compared with that expected for the hydroxyl radical. The mechanism of the oxidation of Fe(II)HEDTA by HZ02, which gives direct evidence of a nonhydroxyl radical intermediate, is considered. Recently we have reported that the reaction between Fe(Il)EDTA and peroxide induces the oxidation of ferrocytochrome c [9] and that this reaction is not inhibited by t-butanol [lo]. It was suggested that a higher oxidation state of iron rather than the -OH radical is the oxidizing intermediate. If formed, it could be more selective in its reactions. It has been estimated that the reduction potential of a ferryl/ferric complex is about 1 V [ll], 1.3 V lower than the couple .0H/H20 at pH 7 [12, 131. Recently it
FENTON
REACTION
has been shown that hypervalent iron complexed appreciable lifetimes in aqueous solution.
EXPERIMENTAL
by OH-
INTERMEDIATES
[14] and EDTA’
201
have
SECTION
Apparatus Stopped-flow experiments were performed with an apparatus purchased from Kinetic Instruments, Inc. The optical setup and microcomputer-based data acquisition were designed by On-Line Instruments Systems, Inc. Kinetic traces were analyzed using nonlinear least squares fitting routines, also by OLIS, Inc. Routine optical spectra were obtained on a Beckman DU-7 HS spectrophotometer. Gamma radiolysis experiments were performed with a Gammator B ‘37Cs radiation source (Kewaunee Scientific Co.). The dose rate was determined by Fricke dosimetry. The yield of Fe3+ was taken to be 15.8 ions per 100 eV absorbed, i.e., G(Fe(II1)) = 15.8. Scavenging
and Kinetic
Experiments
The effects of scavengers on the stoichiometries of the Fe(B)-pac/H202 systems were measured by stopped-flow spectrophotometry. Iron solutions were prepared under nitrogen from Fe(II)(NH&(S0&.6aq and a 1.3:1 excess of ligand (EDTA, DTPA, HEDTA). For standard scavenging experiments the concentration of Fe(II)pac was 1.5 X 10e3 M and [H202] = 7.5 x lo-’ M. The peroxide/scavenger solutions were also thoroughly deaerated before mixing and transferred to the flow machine in syringes with gas-tight fittings. The rate of Fe(II1) formation and overall absorbance changes were monitored at a convenient wavelength between 300 and 350 nm. All measurements were the mean of four determinations, which generally were reproducible to within ? 5%. Iron(B) solutions contained not more than - 5% ferric impurities. All experiments were performed at room temperature, 25 -t 2°C. The scavengers used were ethanol (Pharmco, loo%), t-butanol (twice recrystallized from Baker analyzed reagent quality), sodium formate, isopropanol , dimethylsulfoxide (all Baker Reagents), and imidazole (Sigma, 98 + %). H202 stock solutions were prepared from a 30% solution (Baker) and standardized by KMn04 titration. In some experiments NaC104 was used to vary the ionic strength. The reactions of *OH radicals with phosphate ions, k.0HIH2m4- = 2.2 x lo6 M-l see-’ 1151; excess ligand, kFC:OH+EDTA = 4 x lo8 M-’ set-’ [16]; Fe(III)pac: k.OH+Fe~Injan~A = 5.2 X lo8 M-l see-l [17], and hydrogen peroxide, k.0H+H202 = 2 x 10’ M-l see-’ [151, are not expected to influence the measurements. Stoichiometric data were obtained by measuring the magnitude of the optical density change as a function of scavenger concentration, [RH], at constant [HzOZ]. In these experiments less than 10% of the initial concentration of ferrous ions was converted to Fe@). In unscavenged solutions the stoichiometry, A[Fe(III)]/A[H202], was found to be 2. Both Fe(II)pac and the scavengers were present in large excess during scavenging experiments and side reactions were calculated to be of small importance (assuming that the most reactive intermediate was the hydroxyl radical) and were always compensated for with appropriate blanks. Values of k,l = kJkz were obtained from
’ J. D. Rush and B. H. J. Bielski, unpublished.
202
J. D. Rush and W. H. Koppenol
- AOD,) vs. [RH]-i, where AODa and AOD, plots of (AODe - AOD,)/(AOD represent the total absorbance changes in unscavenged and completely scavenged solutions, respectively. In addition to equations (l)-(3), the following reactions can take place, irrespective of whether I, is ‘OH + Fe(III)pac, FeO(II)pac or a Fe(II)HzOzpac complex: Fe(III)pac + R * -+ Fe(II)pac + R +
(4)
Fe(II)pac + R - + H + --+ Fe(III)pac + RH
(5)
I, + R. --t Fe(III)pac + 20HR. + R. --+ products
+ R’
(6) (7)
In the following, reaction (6) is neglected as well as the slow decomposition of hydrogen peroxide catalyzed by Fe(III)pac [ 181. With (Y as an abbreviation for k2[Fe(II)pac]lks[RH] one derives the following ratios of Fe(III)pac formed per H202 consumed, using the steady-state approximation for Ii and R.: a. R* is unreactive
towards Fe(II)-
or Fe(III)pac,
A[Fe(III)pac]
=1+a l+a
MH2Oz I
(8)
b. R* is reducing, A[Fe(III)pacJ
UWzl
=-
2a l+cY
(9)
c. R 9 is oxidizing, A[Fe(III)pac]
WzW
=2
(10)
Were Ii to oxidize RH by a two-electron transfer (for instance, from ethanol to acetaldehyde), one would still obtain equation (8). Thus this analysis does not allow one to characterize I, as an opportunistic two-electron oxidant. Gamma Radiolysis In order to determine the reactivity of certain organic radicals towards the Fe(I1) or Fe(II1) complexes, solutions of Fe(III)EDTA (3 x 1O-4 M) at pH = 7, and containing an excess of scavenger (t-butanol, benzoate, formate, acetate), were degassed with N2 and irradiated for 30 min and then analyzed spectrophotometrically for reduction of Fe(III)EDTA to Fe(II)EDTA. Under these conditions the aquated and .OH radicals form the corresponding electron, e,,- , reduces Fe(III)EDTA The amount of scavenger radicals (. CH2(CH3)2COH, *C02-, and +CH&OO-). Fe(II1) reduced corresponded to G( - Fe(III))rormate = 6, G( - Fe(III))t.i,,lanOl E 2.5, and G( - Fe(III)),,,,, = 0, indicating that formate radicals reduce Fe(II)EDTA, tbutyl radicals are unreactive towards Fe(II)EDTA and Fe(III)EDTA, and acetate radicals oxidize Fe(II)EDTA. The hydrated electron reacts nearly as quickly with benzoate as the hydroxyl radical. Nevertheless, we found that G( - Fe(III)EDTA) was 2.2, which indicates that the electron is transferred from benzoate to Fe(III)EDTA
FENTON
REACTION
quite efficiently and that the benzoate radical, hydroxyl radical, does not oxidize Fe@)EDTA.
formed
INTERMEDIATES
from the reaction
203
with the
RESULTS Scavenging of the Intermediates Fe(II)HEDTA- , Fe(II)EDTA’-
Produced in the Reactions of Fe2+, , and Fe(II)DTPA3 - with Hydrogen Peroxide
Preliminary to scavenging experiments with the pat complexes, the effects of added tbutanol and ethanol on the stoichiometry of the reaction of Fe2+ (1.5 x 1O-3 M) and Hz02 (7.5 x lo-’ M) were measured in acidic sulfate medium (pH = 2.65, [sulfate] = 0.2 M). This reaction, which proceeds with a rate constant of 43 M-i set-i under those circumstances [ 193, is generally considered to * involve a hydroxyl radical intermediate. The overall formation of FeS04+ (630~ = 2200 M- ’ cm- I) is inhibited by these scavengers in a manner consistent with reactions (1 l)-( 14), which are written for the ethanol system: Fe2+ +H,O,
+ Fe3+ + *OH+OH-
*OH+Fe2+
+ Fe3+ +OH-
. OH + C2HSOH + CH3CHOH + H20 Fe3+ + CH3CHOH
+ CH3COH + Fe2+
(11) (12) (13) (14)
For ethanol, k,i = k13/k12 = 6.3, which yields kt3 = 1.9 (kO.1) x lO’M_’ set-’ when k12 = 3 x lo* M-i see- ’ [20]. The corresponding value of ki3 for t-butanol (which does not react as in equation (14)) is (3 f 1) x lo* M-i see-i and is also in good agreement with literature values [ 151 for these reactions. A plot of the measured and calculated stoichiometries for reactions (1 l)-(14) as a function of [ethanol] is shown in Figure 1. Values of krel and related data for scavenging systems are summarized in Table 1. The rate of formation of FeS04+ is a pseudo-first-order process that is independent of the ethanol concentration. Traces of oxygen when present in the peroxide component caused a small, fast increase of absorbance, which was easily separated from the reaction of Fe2+ with Hz02 by measuring AOD at a time when this prereaction was complete, approximately 10% of one half-life after mixing. Experiments with added oxygen indicated that this process is most likely initiated by the reaction between an alcohol radical and 02 leading to ROi, which reacts with Fe2+ to produce some FeS04+. When corrected for as described, oxygen contaminations of the order of 5 PM do not affect stoichiometry measurements. A similar reaction due to trace oxygen was observed in ethanol scavenging experiments with Fe(II)EDTA and Fe(II)DTPA.
The Effects of Ligands
on Scavenging
Behavior
The variation of the ligand on the iron complex is expected to have consequences for the reactivity of I, if it is an iron complex. On the other hand, a hydroxyl radical intermediate will be unaffected. Hydroxyl radicals oxidize the Fe(II)EDTA and Fe(II)NTA (NTA = nitrilotriacetic acid) complexes with equal rates; k = 5.0 x log
204
J. D. Rush and W. H. Koppenol
^s s 2.0 -
a 1 0”
0 0
1.6-
0
-8
1.2 -
9
g
0
/:
” $
0/
0.0 -
z! N /A
0.4 0 1.4
I 1.8
I 2.2
I 3.0
I 2.6 - log
, 3.4
I 3.8
, 4.2
I 4.6
5.0
[Ethanol]
FIGURE 1. Experimental scavenging curve of the Fe2+/Hz02 reaction at pH = 2.65,0.2 M sulfate. The solid line calculated for k,el = 6.33 yields k~OH+ethanolj= 1.9 x lo9 M-’ set-I.
TABLE
1. Relative Rates of the Reactions between Scavengers H202 Compared with Hydroxyl Radical
[+$$I
System
Scavenger
k&expY
Fe*+/H *0 2
I-Butanol Ethanol
1.2 f 0.2 6.3
FeEDTA’-/H202
Ethanol Propanol Formate Imidazole DMSO
0.18 -0.2 0.7 1.9 1.36
t-Butanol Benzoate
0.66
1.8 1.75
FeDTPA3-/H202
Ethanol Ethanol” Formate FormateP t-Butanol
0.24 0.24 0.6 0.8 -
0 0 0 0 1.8
FeHEDTA-/H,OZ ~____.___
Ethanol ..~~
0.124 -~-~~--___~~~...~~~
1 0 0 0 0 0.3 1.5
0
and Intermediates of FeWY
/cl(litjb
[~I
1.33 6.3 0.38 0.42 0.54 1.8 1.3 -0.08 1.2 c c c c c c
b
1 0 0 0 0 -
..-
1 1 0 0 0 0 1 0 ___
’ &(exp) is kl/k~ as experimentally determined. * k&i0 is kdk2 as expected for .OH.k, o,,+Fez+j = 3 x lo* M-l set- I, k~.oH+Fe~,,,pac~~ = 5 x IO9 M-j set-‘. Rate constants for /q.o,,+s) are from Ref. 15. ’ Values of k,, are likely to be the same as for FeEDTA but literature data are not available (see text). ’ [Fe(II)DTPA] = 4 x 1O-4 M. ’ Ionic strength raised to 0.1 M with NaClO+
FENTON REACTION INTERMEDIATES
205
M- ’ see- ’ [ 161by direct attack on the metal ion and it is assumed here that this rate is the same for the related HEDTA and DTPA systems. This mechanism is in contrast to the Mn(II)EDTA and NTA complexes, which are oxidized to the corresponding Mn(II)pac ligand radical complexes by radiolyticahy generated - OH radicals [ 161 that probably abstract hydrogen from a methylene group of the ligand. Under stopped-flow conditions with excess Fe(II)pac, the reaction traces are essentially first order and did not give direct evidence of an intermediate. The overall rate constants are given in Table 2B for reactions (15) at pH = 7.4; pat = HEDTA, EDTA, DTPA. 2Fe(II)pac + H202 --) 2Fe(III)pac(OH)
(15)
The relative scavenging reactivities of Ii (HEDTA, EDTA, DTPA) towards ethanol, formate, and i-butanol were measured. Both ethanol and formate acted as two-equivalent scavengers, causing A[Fe(III)]/A[H202] to approach zero at sufficiently high [RH], i.e., they followed equation (8). The effect of varying the ligand on kðanol) is indicated by the scavenging plots of Figure 2. The computed values correspond to k,, = 0.124, 0.18, and 0.24 for the HEDTA, EDTA, and DTPA systems, respectively. A value of k,,r = 0.38 is expected if I1 = -OH assuming k(.oH+re(njpacj= 5.0 x log M-’ set-l and k(.OH+EIOH) = 1.9 x log M-’ set-’ [15]. No intermediates were detected in the presence of ethanol. Formate ion scavenges both I,(EDTA) and I,(DTPA) with comparable efficiencies. The values of k,,(formate) = 0.7 f 0.2 for the two complexes were not different within experimental error and close to the value of - 0.5 expected for the -OH radical. In the presence of formate ion, the Fe(II)HEDTA/H202 reaction results in the formation of a transient complex. This reaction will be discussed below. Dependence k,i on Ionic Strength and [Fe(II)pac] The relative rate constants listed in Table 1 were obtained at [Fe(B)] = 1.5 x 10v3 M and are assumed to be independent of [Fe(n)]. This was verified for the scavenging of I,(DTPA) by ethanol. A fourfold dilution of [Fe(II)DTPA] did not change the value of k,, = 0.24 + 0.04. The ionic strength of the scavenging media was normally determined by the phosphate buffer (5 mM, pH = 7.4). Addition of NaC104 to a final concentration of 0.1 M had a negligible effect on k,, in the DTPA/ethanol system, which suggests an uncharged intermediate. However, k,, in the DTPA/formate system increased by - 60% relative to the value obtained at lower ionic strength. Scavenging Behavior of Imidazole, DMSO, and Isopropanol in the Fe(II)EDTA/H202 Reaction Imidazole and dimethylsulfoxide exhibited relative scavenging rates that are similar to the expected ones for I, = hydroxyl radical, whereas isopropanol showed a reactivity similar to ethanol. Fractional limiting stoichiometries were obtained for imidazole and DMSO, perhaps indicating that they are oxidized by I, only partially to intermediates which can then be oxidized by Fe(III)EDTA. The stoichiometric data obtained for all the reactive scavenger systems employed with Fe(II)EDTA/H202 are shown in Figure 3. Values of k,, are proportional to the slopes of the plotted data and given in Table 1. The Scavenging Behavior of t-Butanol, Sodium Benzoate, and Sodium Acetate Tertiary butanol and benzoate react with the hydroxyl radical (k = 3 X 10s M- ’ set- ‘) to form radicals that are incapable of reacting with either Fe(II)EDTA or
206
J. D. Rush and W. H. Koppenol
TABLE 2. A. Reaction of Fe(II)HEDTA
with H202’
[Fe(II)HEDTA]
kslowsec-
1.5 x IO-3 5.7 x IO-4 2.9 + .2 x 1O-4
B. Bimolecular Ligand EDTA DTPA HEDTA
(M)
kfdsec
- ‘1
63 + 2 26 f 2 9~2
Rate Constants
’
5.1 5.0 5.0
for Fe(II)pac
+ H202’
k mmpac +H*O*1 7 * 1 x lo3 Mm1 set-I 5.1 * 0.2 x lo2 M-’ sec.’ 4.2 k 0.2 x lo4 M-’ set-’
a All runs at 0.005 M phosphate, 2°C.
pH = 7.4 and T = 25 +
Fe(III)EDTA (see Experimental Section). Thus, as in the Fe*+/H?Oz system, a limiting value of A[Fe(III)]/A[H202J = 1 at high [t-butanol] and [benzoate] is expected. The effect of a high concentration of t-butanol on iron(II1) formation in the EDTA system is shown in Figure 4. The unscavenged trace exhibits a small deviation from pseudo-first-order behavior, which, in the presence of 0.3 M 5 [t-butanol] 5 0.002 M is removed and seems to account for - 15 % of total iron(II1) formation. This small component of the reaction is restored when [t-butanol] s 0.001 M. Added tbutanol had no effect on the absorption spectra of Fe(III)EDTA or Fe(III)DTPA and
FIGURE 2. Scavenging curves of ethanol with the three iron complexes Fe(II)HEDTA (e), Fe@)EDTA (O), and Fe(II)DTPA (A ionic strength 0.1 M and (A) at -0.01 M). All other data were obtained at pH = 7.4 f 0.1, p = 0.01 M, 5 mM phosphate buffer: The concentration of Fe(II)pac was 1.5 x 10m3 M and solid curves are calculated on the bases of values of k,,, shown in Table 3.
-log
[ethanol
]
FENTON
REACTION
INTERMEDIATES
207
6-
o”“““( 2
4
I/([Scovenger]
6
a
x IO3 1, M-’
FIGURE 3. Plots showing the relative reactivity of various scavengers toward the intermediate of the Fe(II)EDTA/H202 reaction. (0) Imidazole. (A) dimethyl sulfoxide, (m) formate, (0) ethanol, (0) i-propanol. All points (some are not shown) were obtained at [Fe(II)EDTA],, = 1.5 x 1O-3 A4, pH = 7.2 f 0.2. The slopes of these lines are equal to k,rl [Fe(II)EDTA] .
air was rigorously excluded from the reacting solutions. Iron@) formation was also inhibited in the Fe(II)DTPA/Hz02 system, but this effect was not linear in (t-butanol) and could not result from the simple competition scheme exhibited by the other scavengers. As shown in Figure 5, the maximum absorbance change due to Fe(III)DTPA formation is reduced by - 15% at the highest [t-butanol] compared to the unscavenged solution. The solid curve in Figure 5 was obtained by fitting the data to a function a/(1 + a) in which cx = N/[t-butanol]2. N is a parameter that depends FIGURE 4. Kinetic traces at 320 nm of the formation of Fe(II)EDTA by reaction of Fe(II)EDTA (1.5 x 10m3 M) and HzOz (7.5 x 10e5 M) in the absence of any scavenger (B) and in the presence of 0.3 M tbutanol (A). Trace B is unchanged for t-butanol concentrations between 0.3 M and -0.002 M. While the final ab-
0
75
150 Time
(ms
225
1
sorbance for Trace B is the top of the figure, that of Trace A is indicated by the dashed line.
208
J. D. Rush and W. H. Koppenol
E .94i ro
.
G
c .86:: b
2
.78-
a
0
n
.70-
0.4
I 0.8
1.6
1.2 -log
2.0
2.4
2.8
[t-butonol]
FIGURE 5. The effect of added t-butanol on Fe(III)DTPA formation in the Fe(II)DTPA/ H202 reaction; [Fe(II)DTPA] = 1.5 x lob3 M. The absorbance change at 320 nm is proportional to [Fe(III)DTPA]. The solid curve is a fit to the data by a functionf(cY) = a/( 1 + cu); CY= N/[t-butanol]‘, where N is an arbitrary fitting parameter.
on [Fe(II)DTPA] but which cannot be interpreted meaningfully since the scavenging mechanism is not known. Sodium benzoate behaved similarly in that concentrations up to 25 mM did not prevent additional Fe(III)EDTA formation (eq. (2)). A stoichiometry of 1.7 Fe(II)EDTA oxidized per hydrogen peroxide consumed was observed at high benzoate concentration. However, scavenging efficiency was first order in [benzoate] unlike tbutanol . Sodium acetate, even at concentrations of 1 M, did not reduce the yield of Fe(III)EDTA, Fe(III)DTPA, and Fe(III)HEDTA. However, since the acetate radical is itself capable of oxidizing Fe(II)EDTA, this latter result does not necessarily exclude the OH radical as a primary reaction intermediate, as follows from equation (10). The Reaction
of Fe(II)HEDTA
with HzOZ
The kinetics of the oxidation of Fe(II)HEDTA by H202 in 5 mM phosphate buffer at pH = 7.4 were studied over the wavelength range 300-400 nm. Fe(II)HEDTA concentrations were maintained in at least a 5: 1 stoichiometric excess over H202. The kinetic traces of Fe(III)HEDTA formation were biphasic. The rate of the fast component was pseudo-first-order in [Fe(II)HEDTA], from which a second-order rate constant of f%e(II)HEDTA + H202) = 4.2 x lo4 M- ’ sec.- ’ was obtained. The slower phase of increasing absorbance has a first-order rate constant of 5 .O + 0.2 set - ’ , and is independent of [Fe(II)HEDTA]. Rate data are summarized in Table 2A. The overall absorbance changes corresponded to the formation of Fe(III)HEDTA(OH)- [21], the major equilibrium species at this pH. The spectrum of the slowly decaying intermediate was calculated from the changes in absorbance during its decay, and these data are summarized in Table 3. The rate of the slow reaction is comparable to the rate of dissociation of the oxo-bridged dimer, [Fe(III)HEDTA1202-, which is approximately 4 set I near pH 7 [22]. Since log K {2Fe(III)HEDTA(OH) - F-?
FENTON
REACTION
INTERMEDIATES
209
TABLE 3. Absorbance Data for the Spectra of Monomeric and Fe(lII)HEDTA at pH = 7.4 and of the Intermediate Dimeric Complex Wavelength
FeHEDTA(OH)’ EX 10-3(M-1 cm-‘)
320
2.6
-
330 340 350 360 370 380 390 400
2 1.55 1.15 0.8 0.5 0.3 0.15 0.06
1.I5 1.33 0.96 0.70 0.44 0.26 0.13 0.04
[FeHEDTA(OH)J, *- b Ex 10-3(M-1 cm-‘)
@Spectrum of Fe(III)HEDTA in 0.01 M phosphate buffer, pH = 7.4. b Fe(II)HEDTA reacted with 7.5 x IO-’ M H,O*; AC = AOD12[H,O,]; e = E(FeHEDTA(OH)-) - Ac(rxn 19).
[Fe(III)HEDTA]202+ H20} = 2.38 [21], dimeric iron@) would dissociate completely at the low total ferric concentration formed in these experiments. However, the spectrum of the intermediate is not similar to that of [Fe(III)HEDTA]202[23], nor can this process be attributed to the reaction of Fe(III)HEDTA(OH)- with phosphate ion since complexation was found to be negligible under these conditions. In order to account for this intermediate, we must assume that it is formed directly from the reaction of I, with excess Fe(II)HEDTA,. yielding an iron(III) species that is not at equilibrium. This excludes reaction (16): -OH + Fe(II)HEDTA
+ Fe(III)HEDTA(OH)-
(16)
The absorption spectrum of the intermediate and’its decay rate seem consistent with reactions (17)-(19), where ki7 = 4.2 x lo4 M-i set-’ and k,9 = 5.0 + 0.2 set-‘: H
_-0 Fe(II)HEDTA-
1
+ H202 ‘%HEDTAFe2+:zz_
or I,(HEDTA)
(17)
-0 H H I
.O.., I,(HEDTA)
+ Fe(II)HEDTA-
?l(fast) [HEDTAFe3C’Xt:
1 ,: Fe3+HEDTA12.o” H
(18)
210
J. D. Rush and W. H. Koppenol
H
xx, [HEDTAFe3+::
- 2Fe(III)HEDTA(OH) klg
,A ,,Fe3+HEDTA12,
-
(1%
I
H The p-hydroxy bridged dimer is not a major component of Fe(III)HEDTA solutions since the oxo-bridged species is more stable. However, the rates of dissociation of known hydroxy-bridged ferric dimers are comparable to oxo-bridged dimers [24]. The absorption spectrum of a CL-hydroxy bridged system would be expected to be similar to the sum of its component monomers as in the present system. Though the structure of I,(HEDTA) is speculative, the reaction between an iron peroxo intermediate such as It with Fe(II)HEDTAin equation (18) to form two hydroxyl bridges simultaneously could explain why the normal oxo-bridged iron(II1) dimer is not formed.
The Reaction between Il(EDTA) and Formate Ion It was suggested above that Fe(II)HEDTA is oxidized by H202 via an intermediate, is Ii(HEDTA), which does not decompose to *OH radicals. This intermediate scavenged by ethanol at a rate comparable to the DTPA and EDTA systems. Sodium formate reacts with I,(HEDTA) to form an unstable transient, which decays by second-order kinetics to a nonabsorbing product as shown by the stopped-flow trace in
FIGURE 6. Upper: the first-order formation and second-order decay of the Fe(II)HEDTA. ligand radical at 320 tnn in 0.02 M formate. Lower: second-order plot of the decay of the radical which yields a rate constant; 2kS = 5.7 x lo5 M-t set-’ at pH = 7.2 in the presence of 1.5 x 10e3 A4 Fe(II)HEDTA.
1
I
.04
I
I
I
.I2 TIME
I
.20 , seconds
I
I
.28
FENTON
REACTION
INTERMEDIATES
211
TABLE 4. Formation of Fe(II)HEDTA* as a Function of [Formate] AOD,
[Formate]
max”, 320 nm
AOD, max”
(MI
(experimental)
(calculated)
0.1 0.05 0.25 0.01 0.005 0.0025 0.00125
0.22 0.18 0.135 0.105 0.051 0.023 0.0125
0.214 0.18 0.14 0.08 0.05 0.027 0.015
ti The absorbances were measured at maximum formation of the transient with [Fe(lI)HEDTA],, = 1.5 x 10m3 A4 and [HZO& = 7.5 x 10V5M. Optical path length = 2 cm. b These values were calculated for AOD = 0.26 X K[HCO,-]I(1 + flHC02-]) where K = 47 M-l and 0.26 = tjzO x 7.5 x 10-5M X 2 cm; c320= 1.73 x 103M- cm-‘.
Figure 6. The amount of transient formed has a limiting dependence on [HCOZ-] and is proportional to [Hz02], i.e., [I,(HEDTA)]. The data in Table 4 show the formate dependence from which extinction coefficients of the transient and the equilibrium constant for its formation were calculated. The spectrum of the complex is shown in Figure 7. Reactions (17) and (20)-(22)
FKXJRE
7.
Spectrum of the Fe(II)HEDTA*
ligand radical complex
270
430
24
2c
-i E
16
” T =
12
7E! x W
8
4
0 310
350
390 X, nm
470
510
at
pH = 7.2.
212
J. D. Rush and W. H. Koppenol
represent the reactions of Fe(II)HEDTA with Hz02 in the presence of formate ion. The rate of reaction (17) is not dependent on formate concentration. I,(HEDTA)+HCO,I,(HC02 -)HEDTA 2Fe(II)HEDTA*
KsIl(HC02-)~~~~~ %!?Z!) Fe(II)HEDTA
“2 Fe(II)HEDTA
(20)
* + *CO?-
(21)
+ Fe(II)HEDTA(ox)
(22)
where KzO = 47 M-‘, and 2kz2 = 5.7 x IO5 M-’ set-’ at pH = 7.4. The transient is identified as the Fe(II)HEDTA* ligand radical on the basis of its decay rate and its absorption spectrum, which closely resembles that of the Mn(II)EDTA * ligand radical complex: X,,,(Mn(II)EDTA *) = 290 nm and 430 nm with emax = 2800 and 820 M- ’ cm-‘, respectively [16] compared with &,,,,,(Fe(II)HEDTA~) = 295 nm and 420 nm, Emax = 2640 and 620 M- ’ cm- ’ for the two peaks. Ferrous and manganous polyamino carboxylate complexes have weak d-d electronic transitions and charge transfer bands are below 220 nm. The strong absorptions of the ligand radical complexes are primarily due to the radical and depend only weakly on the central transition metal ion [16]. The manganous complex disproportionates or dimerizes with a rate constant 2k = 5 x lo5 M-’ set-’ at pH = 7.4 [16]. In both cases no M(III) is formed, although these oxidation states are stable with EDTA or HEDTA ligands. The ligand is presumed to be oxidized as a result of reaction (22). Equilibrium with formate ion, equation (20), must logically be prior to formation of the ligand radical complex and it is necessary to postulate an intermediate, Ii[HC02-IHEDTA in equilibrium with I, (HEDTA), which contains the iron complex, formate and retains both oxidizing equivalents of hydrogen peroxide. The decomposition of this species to the Fe(R) ligand radical in equation (20) is faster than reaction (18) and it is possible that hydroxyl radical is released in reaction (21) rather than COa- . Preassociation between the anionic Fe(II)HEDTAcomplex and HCO;!- seems an unlikely explanation of the formate dependence owing to the high value of K 20, which is of the order exhibited by formate towards M2+ ions [25].
DISCUSSION The existence of a highly oxidizing iron intermediate, e.g., ferry1 (Fe02+ or is suggested by some of our results. Fe(OH)22+), or ferrous peroxide, Fe(H20#+, The Fe(II)HEDTA reaction particularly seems to require such an intermediate. The values of k,, exhibited by scavengers in the Fe(II)EDTA and Fe(II)DTPA systems with the exception of t-butanol and benzoate do not clearly distinguish Ii from *OH but rather suggest that for many organic molecules it behaves similarly to *OH radical. EPR studies have shown that alcohol radicals, including the I-butyl radical, are formed in reacting Fe(II)EDTA/H202 solutions [26-281. These observations do not distinguish actual amounts formed or the pathways of their formation. It is evident from our observations that I, is significantly less reactive towards ethanol than is the -OH radical and does not react quantitatively with t-butanol. Yet, in general its pattern of reactivity towards other scavengers is similar to the -OH radical, and for the case of Fe2+ in acid, it appears that *OH is formed. The value of k,, = 6.3 is in very good agreement with that obtained by Walling et al. [29]. The pattern of t-butanol scavenging in the EDTA and DTPA systems provides the best evidence against simple *OH radical formation, although the reason for the apparent [t-butanol12 dependence
FENTON REACTION INTERMEDIATES
213
is unclear. Formaldehyde and acetone are produced when t-butanol is oxidized in Fe(III)EDTA-H202-ascorbate solutions, which suggests a complex process 1301. Inhibition by t-butanol of Fe(III)EDTA catalyzed chain oxidations has been observed 1281.However, under catalytic conditions where Fe(II)EDTA is present in steady-state amounts, reaction (2) is negligible and possibly I, is converted to a species that reacts quantitatively with t-butanol in these systems. A chain-termination process in the Fe(III)EDTA catalyzed oxidation of formate by hydrogen peroxide has been attributed to a possible Fe(IV) intermediate that reacts more slowly with formate than does the OH radical [31]. The formation of dimeric iron@) produced from ozonolysis of excess Fe2+ has been cited as evidence of the reaction between an iron(N) complex and iron in acidic medium [32]. The structure of the peroxo intermediate in reaction (17) is speculative, but similar to that favored for the complex of Fe(III)EDTA and H202 [33]. Resonance between structures A and B could account for the apparently strong affinity of peroxide towards Fe@)
,O
/H -
I
(HEDTA)Fe(IV)
‘0
‘H
/PH
, ; OH
lB
A
This interaction is clearly favored by the availability of a coordination site on the Fe(II)pac complex since Fe(II)HEDTA reacts about six times more rapidly than Fe(II)EDTA while having a reduction potential similar to the that of the EDTA complex [34]. On the basis of ‘aO’SO isotopic fractionation studies Cahill and Taube [35] concluded that peroxide bond cleavage in the Fe2+ - H202 transition complex was consistent with a minor pathway involving the two-electron oxidation of Fe2+ in acid. Thermodynamic calculations suggest that Fe(W) is more stable at neutral pH with EDTA-like ligands [ 1I], and this oxidation state may constitute the major pathway under biological conditions. Although the mechanism is uncertain, the formation of two carbon-centered radicals from hydrogen peroxide is thermodynamically feasible. The sum of reactions (16), (19), and (20),’ Fe@)HEDTA-
+ H20Z + HC02- + Fe@)HEDTA* + 2H20 + COZ-
(23)
is energetically possible at pH 7: HC02- + C02- +H+ +e-,
nP(V)=
- 1.64
(24)
as follows from Gibbs energies of formation of formate and carbon dioxide [36], and EO(CO*ICO*-) = -2.0 v [37]; Hz02+2H+ +2e-
+ 2H20,
nE”‘(V)=2.64
(25)
214
.I. D. Rush and W. H. Koppenol
and equating Fe(II)HEDTA-
-+ Fe(II)HEDTA
* -t Hi + e _
(26)
with CH3CH20H
-+ CH,C*HOH+H+
+e-,
n/Z”(V)=
-0.981341
(27)
Equations (24), (25), and (27) yield equation (23) with AGO’ = OV. Since estimates for E”‘(COzIC02-) vary between - 1 [38] and -2 V [37], the Gibbs energy change of 0 kcal for reaction (23) represents an upper limit and is likely to be more negative. The intermediate of the Fe(II)HEDTA/H202 reaction apparently retains both oxidizing equivalents of peroxide. Its interaction with formate, which results in transfer of one equivalent to the ligand without oxidizing the central metal, might arise from several factors: (1) the availability of a metal coordination site for binding by HC02-; (2) the availability of an or-hydrogen on formate that is readily accessible to coordinated *OH radical or H,O,; and (3) the formation of a reducing radical when formate ion is oxidized by the second equivalent of H20Z. The latter two factors are suggested by the failure of acetate to cause this reaction. Coordinated formate ion may stabilize a transient peroxo-Fe(I1) or Fe(OH)22+ system long enough to permit a complex intramolecular oxidation to occur. The result seems to be an example of a site-specific reaction in which the oxidizing equivalents of H202 are directed to the immediate environment of the Fe(II)pac/HZOz reaction, probably without producing free hydroxyl radicals. In nonaqueous solvents where Fe 3+ is less well solvated than in water and therefore less stable, the site directed hydroxylations of cyclohexanol 1391 and other compounds [40] by Fe(II)/H202 appear to involve only two electron changes in the formal oxidation state of iron, i.e., Fe(II)/Fe(IV). The implication for biological systems is that Fe(I1) peroxo or ferry1 complexes may cause oxidative damage at the site of binding rather than via a short-lived and indiscriminate hydroxyl radical intermediate. It is also important to recognize that the coordination environment of iron may greatly affect this potential. Supported by National Instifutes of Health grant No. GM 33883. We thank Dr. S. Rosenberg for the use of the Gammator B.
REFERENCES 1. S. D. Aust, L. A. Morehouse, and C. E. Thomas, .I. Free Radicals Biol. Med. 1, 3 (1985) and references therein. 2. G. W. Winston, D. E. Feierman, and A. I. Cederbaum, Arch. Biochem. Biophys.
232, 377 (1984). 3. 4. 5. 6.
B. Halliwell and G. M. J. Gutteridge, Arch. Biochem. Biophys. 246, 501 (1986). C. Walling, Act. Chem. Res. 8, 125 (1975). W. C. Bray and M. H. Gorin, J. Am. Chem. Sot. 54, 2 124 (1932). J. T. Groves, in Metal Zon Activation of Dioxygen, T. G. Spiro, Ed., Wiley, New York, 1980, pp. 125-162. 7. B. Chance, H. Sies, and A. Boveris, Physiol. Rev. 59, 527 (1979). 8. T. C. Bruice, M. F. Zipplies, and W. A. Lee, Proc. Natl. Acad. Sci. USA 83, 4646
(1986).
FENTON REACTION INTERMEDIATES
9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.
20. 21.
22. 23. 24.
25. 26.
27. 28. 29.
215
W. H. Koppenol, J. Free Radical Biol. Med. 1, 281 (1986). J. D. Rush and W. H. Koppenol, J. Biol. Chem. 261, 6730 (1986). W. H. Koppenol and J. F. Liebman, J. Phys. Chem. 88, 99 (1984). H. A. Schwarz and R. W. Dodson, J. Phys. Chem. 88, 3643 (1984). U. K. Kllning, K. Sehested, and J. Holcman, J. Phys. Chem. 89, 760 (1985). J. D. Rush and B. H. J. Bielski, J. Am. Chem. Sot. 108, 523 (1986). Farhataziz and A. B. Ross, Nat/. Stand. Ref. Data Ser., Natl. Bur. Stand., No. 59 (1977). J. Lati and D. Meyerstein, J. Chem. Sot. Dalton 1105 (1978). K. P. Kundu and N. Matsuura, ht. J. Radiat. Phys. Chem. 3, 1 (1971). K. C. Francis, D. Cummins, and J. Oakes, J. Chem. Sot. Dalton Trans., 493 (1985). T. J. Hardwick, Can. J. Chem. 35, 428 (1957). Z. Stuglik and Z. P. Zagorski, Radiat. Phys. Chem. 170, 229 (1981). R. L. Gustafson and A. E. Martell, J. Phys. Chem. 67, 576 (1963). R. G. Wilkins and R. E. Yelin, Inorg. Chem. 8, 1470 (1969). H. J. Schugar, C. Walling, R. B. Jones, and H. B. Gray, J. Am. Chem. Sot. 89, 3712 (1967). A. D. Gilmour, J. K. Cassidy, and A. McAuley, J. Chem. Sot. (A) 2847 (1970). L. G. Sillen and A. E. Martell, “Stability Constants of Metal-ion Complexes,” Special Pub. No. 17, The Chemical Society, London, 1964, p. 101. T. Shiga, J. Phys. Chem. 69, 3805 (1965). M. Fitchell, B. C. Gilbert, and M. Jeff, Phil. Trans. R. Sot. London B311, 517 (1975). c. Walling, R. E. Partch, and T. Weill, Proc. Natl. Acad. Sci. USA 72, 140 (1975). C. Walling, G. M. El-Taliawi, and R. A. Johnson, J. Am. Chem. SOC. 96, 133
(1974). 30. A. I. Cederbaum,
A. Qureshi, and G. Cohen, Biochem. Pharmac. 32, 3517 (1983). 31. H. C. Sutton and C. Winterbourn, Arch. Biochem. Biophys. 235, 106 (1984). 32. T. J. Connocchioli, E. J. Hamilton, Jr., and N. Sutin, J. Am. Chem. SOC. 87, 926 (1965). 33. C. Bull, G. McClure, and J. A. Fee, J, Am. Chem. Sot. 105, 5290 (1983). 34. W. H. Koppenol and J. Butler, Adv. Free Radical Biol. Med. 1, 91 (1985). 35. A. E. Cahill and H. Taube, J. Am. Chem. Sot. 74, 2312 (1952). 36. P. D. Wagman, W. H. Evans, V. B. Parker, R. H. Schumm, I. Halow, S. M. Bailey, and K. L. Chumey, J. Phys. Chem. Ref. Data 11, Suppl. 2 (1982). 37. M. Breitenkamp, A. Henglein, and J. Lilie, Ber. Bunsenges. Phys. Chem. 80, 973 (1976). 38. J. Lilie, G. Beck, and A. Henglein, Ber. Bunsenges. Phys. Chem. 75, 458 (1971). 39. J. T. Groves and M. Van Der Puy, J. Am. Chem. Sot. 98, 5290 (1974). 40. H. Sugimoto and D. T. Sawyer, J. Am. Chem. Sot. 1064298 (1985).
Received July 23, 1986; accepted October 9, 1986
