Original Contribution

IMMUNOCHEMICAL DETECTION OF HEMOGLOBIN-DERIVED RADICALSFORMED BY REACTION WITH HYDROGEN PEROXIDE: INVOLVEMENT

OF A PROTEIN-TYROSYL RADICAL

DARIO C. RAMIREZ, YEONG-RENN CHEN, and RONALD P. MASON

Laboratory of Pharmacology and Chemistry, National Institute of Environmental Health Sciences, National Institutes of Health,Research Triangle Park, NC, USA

(Received 23 August 2002;Accepted 23 December 2002)

Abstract—To investigate the involvement of a hemoglobin radical in the human oxyhemoglobin (oxyHb) or metHb/H2O2 system, we have used a new approach called “immuno-spin trapping,” which combines the specificity andsensitivity of both spin trapping and antigen:antibody interactions. Previously, a novel rabbit polyclonal anti-DMPOnitrone adduct antiserum, which specifically recognizes protein radical-derived nitrone adducts, was developed andvalidated in our laboratory. In the present study, the formation of nitrone adducts on hemoglobin was shown to dependon the oxidation state of the iron heme, the concentrations of H2O2 and DMPO, and time as determined byenzyme-linked immunosorbent assay (ELISA) and by Western blotting. The presence of reduced glutathione orL-ascorbate significantly decreased the level of nitrone adducts on metHb in a dose-dependent manner. To confirm theELISA results, Western blotting analysis showed that only the complete system (oxy- or metHb/DMPO/H2O2) generatesepitopes recognized by the antiserum. The specific modification of tyrosine residues on metHb by iodination nearlyabolished antibody binding, while the thiylation of cysteine residues caused a small but reproducible decrease in theamount of nitrone adducts. These findings strongly suggest that tyrosine residues are the site of formation of theimmunochemically detectable hemoglobin radical-derived nitrone adducts. In addition, we were able to demonstratethe presence of hemoglobin radical-derived nitrone adducts inside red blood cells exposed to H2O2 and DMPO. Inconclusion, our new approach showed several advantages over EPR spin trapping with the anti-DMPO nitrone adductantiserum by demonstrating the formation of tyrosyl radical-derived nitrone adduct(s) in human oxyHb/metHb at muchlower concentrations than was possible with EPR and detecting radicals inside RBC exposed to H2O2. Published byElsevier Science Inc.

Keywords—Hemoglobin, Hydrogen peroxide, Immuno-spin trapping, Red blood cells, Free radicals

INTRODUCTION

Considerable attention has been focused on the forma-tion, stability, and reactions of hemoprotein-centeredfree radicals in biological systems [1,2]. Particular bio-chemical studies have been focused on hemoprotein-centered radicals [3–5] such as the prostaglandin H syn-thase-tyrosyl radical [6], the mitochondrial cytochromecoxidase-thiyl radical [7], the cytochromec peroxidase-tryptophanyl radical [8], the cytochromec-tyrosyl radical[9], the horse metmyoglobin-tyrosyl radical [10], and the

hemoglobin-thiyl [11–13] and methemoglobin(metHb)-tyrosyl [14,15] radicals.

Hemoprotein-mediated redox reactions have beensuggested to contribute to tissue damage and/or organdysfunction that occurs in some pathological states, char-acterized by the release of myoglobin or hemoglobin intothe extracellular environment as with cerebral hemor-rhage, ischemia/reperfusion, rhabdomyolysis, intravas-cular hemolytic anemia, atherosclerosis, and the infusionof hemoglobin-based blood substitutes [2,16,17]. Thereaction of hydrogen peroxide (H2O2) with hemoproteinshas been known for decades [5], and the development ofspecific and sensitive methods to characterize and deter-mine their intermediates is an important concern in freeradical biochemistry. Indeed, the general interest in the

Address correspondence to: Dr. Dario C. Ramirez, NIEHS/NIH, 111T. W. Alexander Drive, Room F048 - MD F0-02, Research TrianglePark, NC 27713, USA; Tel: (919) 541-3866; Fax: (919) 541-1043;E-Mail: ramirez1@niehs.nih.gov.

Free Radical Biology & Medicine, Vol. 34, No. 7, pp. 830–839, 2003Published by Elsevier Science Inc.

Printed in the USA. All rights reserved0891-5849/03/$–see front matter

doi:10.1016/S0891-5849(02)01437-5

830

reaction arises from the capability of metHb and otherhemoproteins to oxidize biologically important sub-strates in the presence of H2O2 [2,3,18–21].

In the case of hemoglobin, this H2O2-dependent spe-cies is thought to be two oxidizing equivalents abovemetHb. One oxidizing equivalent is retained in the ferrylmoiety where the iron has a valence state of �4 [22], andanother oxidizing equivalent is retained in the form of aprotein-derived radical [11,14]. The mechanism bywhich this hemoglobin-derived radical is generated, itslocalization in the protein, and the mechanisms by whichit is dissipated are poorly understood. Furthermore, ferrylhemoglobin [23] and hemoglobin-derived radical forma-tions in whole blood at liquid nitrogen temperature havealso been detected [24], but any dependence of the rad-ical on H2O2 has not been demonstrated.

The formation of the ferryl heme state can be studiedby optical spectroscopy, and the hemoglobin-derivedfree radical by direct electron paramagnetic resonance(EPR) spectroscopy [5,14] and EPR spin trapping[14,15,25]. In the latter technique, the spin trap convertsthe short-lived protein radical into a longer-lived nitrox-ide free radical. The most commonly employed spintraps are nitroso compounds (like 2-methyl-2-nitrosopro-pane, MNP), nitrones (like, �-phenyl-N-t-butylnitrone,PBN), and cyclic nitrones (like 5,5-dimethyl-1-pyrrolineN-oxide, DMPO) [26]. The nitrone DMPO was shown toform radical adducts with tyrosine-103 of sperm whale[27] and human metmyoglobin [28,29]. The direct six-line EPR spectrum (g � 2.0048) of the protein radicalobserved in H2O2-treated metHb was attributed to atyrosyl radical, based on its similarities to the metmyo-globin-tyrosyl radical [14].

Previously, an antiserum directed against a nitronespin trap coupled to octanoic acid (OA), DMPO-OA,was developed and validated [30] using as a model thehorse heart metmyoglobin/H2O2 system where theDMPO spin trap is known to react with the tyrosyl-103radical [10,27]. Our new immuno-spin trapping ap-proach, which combines the specificity of spin trappingand antigen:antibody interactions, could constitute apowerful tool in free radical research. The antiserumanti-DMPO-OA specifically recognizes protein radical-derived nitrone adducts (see Scheme 1) and could be acomplementary technique to EPR spin trapping in thedetection of protein-derived free radicals in biologicalsystems. The reaction between hemoglobin and H2O2 inthe presence of DMPO produces hemoglobin-derivedradicals that form paramagnetic radical adducts withDMPO (Hb•-DMPO radical) detectable at high hemoglo-bin concentrations (i.e., mM) by EPR spin trapping[14,15]. Presumably, it is necessary that the radical ad-duct be oxidized to the corresponding nitrone to be

detected by the antiserum anti-DMPO-OA (Ab inScheme 1).

Considering the importance of the detection of lowconcentrations of free radical in the pathophysiologicalpathway of hemoglobin/H2O2-driven free radical biolog-ical damage, we used immuno-spin trapping to investi-gate radical formation in the human hemoglobin/H2O2

system. In addition, the EPR detection of radical adductsinside cells has rarely been reported; however, we reporthere for the first time the detection of intracellular ni-trone adducts with immuno-spin trapping in red bloodcells (RBC) exposed to low concentrations of H2O2.

EXPERIMENTAL PROCEDURES

Materials

Human metHb (M-4257) and oxyferrousHb (oxyHb,H-0267, � 90% ferrous hemoglobin), diethylenetriaminepentaacetic acid (DTPA), reduced glutathione (GSH),L-ascorbic acid, and 5,5'-dithiobis(2-nitrobenzoic acid)(DTNB) were purchased from Sigma Chemical Co. (St.Louis, MO, USA). Beef liver catalase was purchasedfrom Roche Molecular Biochemicals (Indianapolis, IN,USA) and used as received. The spin trap DMPO waspurchased from Alexis Biochemicals (San Diego, CA,USA). The DMPO concentration was measured at 228nm, assuming a molar absorption coefficient of 7800M�1cm�1 [31]. Reagent grade 30% H2O2 was obtainedfrom Fisher Scientific Co. (Fair Lawn, NJ, USA). TheH2O2 concentration was verified using UV absorption at

Scheme 1.

831Immuno-spin trapping of a hemoglobin-derived tyrosyl radical

240 nm (� � 43.6 M�1cm�1). All buffers were storedover Chelex 100 (Bio-Rad Laboratories, Hercules, CA,USA) at 4°C for 24 h, followed by the addition of 100�M DTPA to avoid possible transition metal-catalyzedreactions.

Hemoglobin preparation

Purified human metHb or oxyHb (oxyferrous form)was dissolved in 0.1 M phosphate buffer (pH 7.4) treatedwith Chelex 100 and containing DTPA (100 �M). Be-fore use, the preparations of oxyHb and metHb werepassed through a prepacked Sephadex G-25 column (PD-10, Amersham Pharmacia Biotech, Piscataway, NJ,USA) equilibrated with 0.1 M phosphate buffer, and thecontribution of each oxyHb and metHb form in therespective preparations was determined to be higher than95% [32]. In this paper, the concentration of oxyHb andmetHb is always expressed in terms of heme concentra-tion (�M) determined by measuring the absorbance ofdeoxyhemoglobin, after the addition of excess sodiumdithionite, at 430 nm and 555 nm (�430 � 133mM�1cm�1 and �555 � 12.5 mM�1cm�1, respectively)at pH 7.4.

Red blood cells (RBC)

The RBC were obtained from human fresh bloodcollected via venipuncture from a 40 year old maleresearch volunteer in apparently good health at the Na-tional Institute of Environmental Health Science, accord-ing to guidelines for use of biochemical samples fromhuman research volunteers. A 19-gauge butterfly needlewas used to collect fresh blood in order to preventshear-induced hemolysis. Blood was collected in a 10 mltube using heparin as an anticoagulant. After removal ofplasma and buffy coat, RBC were washed three timeswith physiological saline (NaCl, 0.85% w/v) and sus-pended to 1% in Chelex-treated 0.1 M phosphate buffercontaining 100 �M DTPA. The RBC were counted in ahemocytometer and resuspended to yield a 0.1% RBCsuspension (approximately 6 � 106 RBC/ml) which wasmixed with different concentrations of H2O2 and 100mM DMPO in a total volume of 1 ml.

After incubating for 1 h at room temperature, the RBCsuspension was centrifuged at 500 � g for 10 min toseparate intact RBC (pellet) from the reaction medium(supernatant). After the reaction, the presence of hemo-globin radical-derived nitrone adducts was determinedby ELISA and Western blotting (see below) in: (i) totalRBC suspension (RBC and reaction medium); (ii) thereaction medium (medium without RBC); (iii) total he-molyzate (included hemoglobin and RBC debris) ob-tained from washed, packed RBC resuspended to 1 mlwith 0.1 M phosphate buffer, and then lyzed (3 cycles of

freeze-thawing); and, (iv) washed RBC debris (pellet)and membrane-free hemolyzate (supernatant) obtainedby centrifugation of total hemolyzate at 10,000 � g for10 min at 4°C. The washed RBC debris from totalhemolyzate was resuspended in 1 ml 0.1 M phosphatebuffer before the analysis.

Antiserum anti-spin trap

A rabbit polyclonal antiserum was obtained and val-idated in our laboratory, as described by Detweiler et al.[30]. Briefly, 5,5-dimethyl-2-(8-octanoic acid)-1-pyrro-line N-oxide (SRI International Co., Menlo Park, CA,USA) was reacted with 1-ethyl-3-(3-dimethylaminopro-pyl) carboiimide HCl and N-hydroxysuccinimide for 20min at room temperature in MES-buffered saline (pH6.5) and then conjugated to ovalbumin for 4 h at roomtemperature to produce DMPO-OA-ovalbumin (i.e., theimmunogen). Immunizations were performed on NewZealand White rabbits by Covance Antisera Services(Denver, CO, USA). After three boosters, the antiserumwas collected by exsanguination on day 106. This anti-serum was used in immuno-spin trapping assays to detecthuman hemoglobin radical-derived nitrone adducts.

Enzyme-linked immunosorbent assay (ELISA)

The amount of hemoglobin radical-derived nitroneadduct was determined using a standard ELISA in 96-well plates (Greiner Labortechnik, Frickenhausen, Ger-many). Three hundred microliters of the adduct solution(1 �g of protein) in coating buffer (100 mM sodiumbiocarbonate, pH 9.6) was incubated for 90 min at 25°C.The plates were washed twice with washing buffer(0.05% Tween-20 and 0.05% casein/BSA, 1:1) andblocked with coating buffer (carbonate/bicarbonate, 100mM, pH 9.6) containing 2.5% casein and 2.5% BSA for90 min at room temperature. Thereafter, the rabbit anti-serum (1:5000) in washing buffer was added and incu-bated for 60 min at room temperature. After beingwashed twice, a secondary antibody, antirabbit IgG-al-kaline phosphatase (1:5000 in washing buffer; PierceChemical Co., Rockford, IL, USA), was added and in-cubated for 60 min at room temperature. After washingtwice, the antigen-antibody complexes were developedby using a chemiluminescence system (CDP-Star, RocheMolecular Biochemicals), and the light emitted was re-corded as arbitrary units using Xfluor Software (TecanUS, Research Triangle Park, NC, USA).

Western blotting analysis

After reacting hemoglobin or RBC, H2O2, andDMPO, the sample was mixed with 4X loading buffer.The proteins (1.2 �g/lane) were separated on reducing4–12% Bis-Tris NuPAGE (Invitrogen, Carlsbad, CA,

832 D. C. RAMIREZ et al.

USA), and rainbow-colored protein molecular massmarkers (14.3–200 kDa, SeeBlue Plus2, Invitrogen) werealways loaded on each gel. The separated proteins wereelectroblotted onto a nitrocellulose membrane. Themembrane was blocked with a 2.5% BSA/casein solutionin 100 mM bicarbonate buffer (pH 9.6) for 90 min. Afterthree washes with washing buffer (TBS containing0.05% Tween-20 and 0.5% BSA/casein), the membranewas incubated for 90 min with the first antibody (rabbitanti-DMPO-OA antiserum, 1:5000 in washing buffer).After that, the membrane was washed as before andincubated for 60 min with the second antibody (goatantirabbit IgG conjugated with alkaline phosphatase,Pierce Chemical Co.) at a dilution of 1:5000 in washingbuffer. After two washes, antibody-antigen complexeswere detected by using enhanced chemiluminescence(Nitro Block II, Tropix, Bradford, MA, USA; CDP-StarII, Roche Molecular Biochemicals) and exposed to ra-diographic film.

Chemical modification of methemoglobin by iodinationof tyrosine residues of metHb

The reaction mixture contained 1 ml aliquots ofmetHb (200 �M) in 50 mM sodium phosphate buffer, pH7.4, and NaI (final concentration of 40 mM). Iodinationof tyrosine residues of metHb was initialized by theaddition of two N-chloro-benzenesulfonamide immobi-lized beads (IODO-BEADS, Pierce Chemical Co.). Thereaction was allowed to proceed with shaking at 23°C for15 min. To stop the reaction, the solution was removedfrom the vessel and passed through a prepacked Seph-adex G-25 column (PD-10, Amersham Pharmacia Bio-tech) to remove excess NaI. The iodinated metHb frac-tion was collected and its concentration was quantifiedby adding an excess of sodium dithionite and using theabsorbance of deoxyhemoglobin at 430 nm (�430 � 133mM�1cm�1).

Blocking of thiols groups in metHb

A suspension of 5,5'-dithiobis(2-nitrobenzoic acid)(DTNB, 0.2 M) was prepared in 50 mM sodium phos-phate buffer and mixed with the metHb at a molar ratioof 100:1. The reaction was allowed to proceed for 10 minat room temperature. The reaction mix was then dialyzed(Slide-A-Lyzer, Pierce Chemical Co.) overnight against50 mM sodium phosphate buffer (pH 7.4) with onechange of buffer.

RESULTS

We used immuno-spin trapping to demonstrate theformation of hemoglobin-tyrosyl radical-derived nitroneadducts after the reaction of purified hemoglobin prepa-ration (oxyHb and metHb) or RBC with H2O2 in the

presence of the spin trap DMPO. To facilitate a compar-ative analysis between hemoprotein systems, the hemo-globin concentration in the present study was alwaysexpressed as heme concentration. The typical heme con-centration in ELISA experiments was 1 �M and inWestern blotting experiments was 10 �M.

ELISA for the detection of hemoglobin radical nitroneadducts

In the present study, an ELISA was used to measurehemoglobin radical-derived nitrone adducts resultingfrom the reaction between human metHb or oxyHb andH2O2 in the presence of DMPO. After 1 h of reaction atroom temperature, the hemoglobin radical-derived ni-trone adducts generated were measured. Figures 1A and1B show a near linear dependence between the antibodybinding to oxyHb or metHb and H2O2 concentration. The

Fig. 1. Enzyme-linked immunosorbent assay (ELISA) for the detectionof protein-derived nitrone adducts from the reaction of human hemo-globin (met- or oxyHb) with H2O2. (A) MetHb or (B) oxyHb (1 �M)was reacted with different final concentrations of H2O2 in the absence(�) or in the presence of 1 (■ ), 5 (E), 10 (●), or 50 (*) mM DMPO.After 1 h of reaction, the hemoglobin radical-derived nitrone adductswere quantified by ELISA as described in Experimental Procedures.All data shown are mean values � SE from three independent deter-minations using fresh met/oxyHb preparations.

833Immuno-spin trapping of a hemoglobin-derived tyrosyl radical

omission of either hemoglobin (metHb or oxyHb),DMPO (Figs. 1A and 1B, open squares), or H2O2 did notshow antibody binding (i.e., signal) in the ELISA. In thepresence of 1 mM DMPO (Figs. 1A and 1B, closedsquares), we did not detect adducts by ELISA even whenusing the highest H2O2 concentration (i.e., 10 �M). With5 mM DMPO (open circles), a significant signal relativeto the control (open squares) was obtained using anH2O2:metHb molar ratio as low as 3:1.

Moreover, nitrone adduct formation increased withDMPO concentration up to 50 mM (Figs. 1A and 1B).The replacement of metHb with equimolar oxyHb undersimilar conditions led to a significant decrease in thehemoglobin radical-derived nitrone adducts as detectedby ELISA (compare Figs. 1A and 1B). Additionally, itwas possible to detect hemoglobin radical-derived ni-trone adducts under substoichiometric conditions byELISA when using 100 �M metHb and an H2O2/hemeratio as low as 0.5 (data not shown).

Western blotting for the detection of hemoglobinradical-derived nitrone adducts

To support our ELISA results, we separated the reac-tion mixture with reducing Bis-Tris NuPAGE and de-tected the adduct formation by Western blotting usingthe same antiserum as for the ELISA. In order to find thebest experimental conditions, we probed different con-centrations of metHb, H2O2, and DMPO. As shown inFigs. 2A and 2B, lanes 1–3, omission of H2O2 and/orDMPO resulted in no immunostaining at 14–16 kDa.

With 1 �M metHb in the presence of 1 mM DMPO,a relatively high concentration of 50 �M H2O2 wasnecessary to induce a signal different from the back-ground (Fig. 2A, lanes 4–6). When we used 10 mMDMPO, a concentration of H2O2 as low as 5 �M wasenough to produce an observable 14–16 kDa band (Fig.2A, lane 7). Higher concentrations of H2O2 produced aconcentration-dependent increase in the 14–16 kDa bandintensity (Fig. 2A, lanes 7–9). From the analysis of lanes10–12, it is evident that a metHb concentration of 10 �M(lane 11) was enough to ensure a high signal-to-back-ground ratio. When we used 50 mM DMPO, a concen-tration of H2O2 as low as 5 �M was enough to observea clearly evident 14–16 kDa from 10 �M metHb (Fig.2A, lane 10). The comparison among the triplets, lanes4–6, 7–9, and 10–12 (Figs. 2A and 2B), demonstratedthat the level of adducts is dependent on the DMPOconcentration, as observed in our ELISAs (see Figs. 1Aand 1B).

The replacement of metHb by equimolar oxyHb(compare Figs. 2A and 2B) produced lower adduct for-mation under the same experimental conditions. It isevident, in accordance with our ELISA results (compare

Figs. 1A and 1B), that the ferric form of hemoglobin, i.e.,metHb, generated more radical-derived nitrone adductsthan oxyHb under the same experimental conditions,consistent with EPR spin-trapping studies [14].

The effect of glutathione and L-ascorbate in theformation of hemoglobin radical-derived nitroneadducts

Glutathione and L-ascorbate are among the most im-portant antioxidants protecting hemoglobin against au-toxidation inside the RBC [33]. To probe their protectiverole in the system of metHb/H2O2, we studied their effecton the formation of hemoglobin radical-derived nitroneadducts. We analyzed the following chemical reaction byELISA: 1 �M metHb � 10 �M H2O2 in the presence of10 mM DMPO. We added the antioxidants immediatelybefore the H2O2. We observed an inverse relation be-tween the concentration of antioxidants and the amountof hemoglobin radical-derived nitrone adducts (Figs. 3Aand 3B). When a reaction mixture containing 10 �MmetHb was exposed to 10 �M H2O2 in the presence of50 mM DMPO and then analyzed by Western blotting,we observed a similar inhibitory effect for both antioxi-dants (data not shown).

Fig. 2. Western blotting assay for the detection of hemoglobin-derivednitrone adducts; (A) metHb and (B) oxyHb. Lane 1, met/oxyHb (10�M heme); lane 2, met/oxyHb � 50 mM DMPO; lane 3, met/oxyHb �50 �M H2O2; lanes 4–6, 1 �M met/oxyHb � 5, 10, and 50 �M H2O2,respectively, in the presence of 1 mM DMPO; lanes 7–9, 5 �Mmet/oxyHb � 5, 10, and 50 �M H2O2, respectively, in the presence of10 mM DMPO; lanes 10–12, 10 �M met/oxyHb � 5, 10, and 50 �MH2O2, respectively, in the presence of 50 mM DMPO. All reactions (1h) were carried out at room temperature in 0.1 M phosphate buffer, pH7.4, containing 100 �M DTPA. The hemoglobin-derived nitrone ad-ducts were investigated by Western blotting, as described in Experi-mental Procedures. Data show a representative experiment from threeexperiments using fresh hemoglobin preparations.

834 D. C. RAMIREZ et al.

Time- and concentration-dependent generation ofhemoglobin radical-derived nitrone adducts

Our preliminary ELISA and Western blotting exper-iments demonstrated that the addition of 5 IU/ml ofcatalase before 10 �M H2O2 prevented the formation ofhemoglobin radical-derived nitrone adducts. Neithercatalase (Fig. 4A, lane 1), catalase � 10 �M H2O2 (Fig.4A, lane 2), nor both in the presence of 50 mM DMPO(Fig. 4A, lane 3) produced an observable Western blot-ting 14–16 kDa band after incubation for 2 h at roomtemperature. Based on these observations, we used cata-lase as a reaction quencher to investigate the time de-pendence of the production of hemoglobin radical-de-rived nitrone adducts in the metHb/DMPO/H2O2 system(Fig. 4A).

When catalase was added at different times after theaddition of H2O2, followed by Western blotting, theintensity of the 14–16 kDa band increased significantlywith time for incubation times of up to 2 h (Fig. 4A).Longer incubation times did not significantly modify theintensity of the 14–16 kDa band (data no shown). Whenthe concentration of H2O2 was varied with incubationtime fixed at 1 h before the addition of catalase, theadduct formation increased with concentration (Fig. 4B).After the addition of catalase, the samples could bestored at 4°C for several days without a significant loss inantibody binding and specificity, as assessed by ELISAand Western blotting (data not shown).

Involvement of a protein-tyrosyl radical in theformation of hemoglobin radical-derived nitroneadducts

To determine the role of cysteine and tyrosine resi-dues in the formation of hemoglobin radical-derivednitrone adducts, we modified them by thiylation with

Fig. 3. Effect of antioxidants on the amount of hemoglobin-derivednitrone adduct generated from the metHb/DMPO/H2O2 system. MetHb(1 �M) was exposed to 10 �M H2O2 in the presence of 10 mM DMPOin a medium (0.1 M phosphate buffer, pH 7.4, containing 100 �MDTPA) containing different concentrations of either (A) reduced glu-tathione or (B) L-ascorbate. After 1 h of reaction, the hemoglobinradical-derived nitrone adducts were quantified in 1 �g of protein byELISA, as described in Experimental Procedures. All data shown aremean values � SE of three independent determinations using freshmetHb preparations.

Fig. 4. Concentration- and time-dependent effects of hydrogen perox-ide on generation of hemoglobin-derived nitrone adducts. (A) Lane 1,5 IU/ml catalase; lane 2, catalase � 10 �M H2O2; lane 3, catalase �10 �M H2O2 � 50 mM DMPO. The other lanes contained 10 �Mheme (metHb) � 50 mM DMPO � 10 �M H2O2, and the reaction wasstopped at different times (min) after the addition of 10 �M H2O2. (B)MetHb (10 �M) and 50 mM DMPO were incubated with differentconcentrations of H2O2 for 1 h, after which any remaining H2O2 wasremoved by the addition of 5 IU/ml catalase. The hemoglobin radical-derived nitrone adducts were investigated by Western blotting, asdescribed in Experimental Procedures. The figure shows data fromthree representative experiments.

835Immuno-spin trapping of a hemoglobin-derived tyrosyl radical

DTNB (DTNB-metHb) or iodination (Iodo-metHb), re-spectively. After incubating 10 �M hemoglobin (eithermetHb, DTNB-metHb, or Iodo-metHb) with 10 �M H2O2

in the presence of 50 mM DMPO, we determined theproduction of hemoglobin radical-derived nitrone adductsby ELISA and Western blotting (Figs. 5A and 5B). Block-ing cysteines (DTNB-metHb) caused a small but reproduc-ible decrease in the amount of nitrone adducts. In contrast,when tyrosine residues were modified (Iodo-metHb), weobserved a sharp decrease (approximately 90%) in proteinradical-derived nitrone adducts (Figs. 5A and 5B). Theseresults strongly suggest that tyrosine residues are the site offormation of the immunochemically detectable methemo-globin (met) hemoglobin radical-derived nitrone adducts.

Hemoglobin-derived nitrone adducts in human RBCexposed to H2O2

Normally, the amount of metHb in RBC does notexceed 1% of the total hemoglobin [33]. This low con-

centration of metHb is maintained by enzymes such ascatalase, superoxide dismutase, glutathione peroxidase,glutathione transferase, methemoglobin reductase, gluta-thione reductase, and glucose-6-phosphate dehydroge-nase, as well as by native antioxidants like L-ascorbateand GSH [33]. We observed an H2O2 concentrationdependence on the amount of hemoglobin radical-de-rived nitrone adducts formed inside RBC in the presenceof 100 mM DMPO, as detected by ELISA (Fig. 6A) andWestern blotting (14–16 kDa band intensity, Fig. 6B).Hemoglobin radical-derived nitrone adducts could be

Fig. 5. Effects of chemical modification of tyrosines and cysteines onformation of hemoglobin-derived nitrone adducts. (A) MetHb, DTNB-metHb, or iodo-metHb (1 �M) was incubated without (open bars) or with(closed bars) 10 �M H2O2 in the presence of 10 mM DMPO. After 1 h ofincubation, the reaction was stopped by the addition of 5 IU/ml catalase,and the hemoglobin radical-derived nitrone adducts were determined byELISA. (B) MetHb, DTNB-metHb, or iodo-metHb (10 �M) � 50 mMDMPO � 10 �M H2O2. After 1 h of incubation, the reaction was stopped,as in (A), and nitrone adducts were investigated by Western blotting. Allthe experiments were carried out in 0.1 M phosphate buffer, pH 7.4,containing 100 �M DTPA, as described in Experimental Procedures. Datashown are the mean values � SE or a representative experiment from threeexperiments carried out separately using fresh preparations of metHb or itschemical modification products.

Fig. 6. Immuno-spin trapping detection of hemoglobin-derived nitroneadducts in red blood cells exposed to H2O2. (A) ELISA to determineadducts in the reaction medium and in the membrane-free hemolyzate(2 �g protein/well); (B) Western blotting to investigate the presence ofadducts in the membrane-free hemolyzate (5 �g protein/lane): lane 1,10 �M heme (oxyHb) � 100 mM DMPO � 10 �M H2O2; lane 2,membrane-free hemolyzate (from 6 � 106 RBC) incubated 1 h at roomtemperature in 1 ml 0.1 M phosphate buffer; lane 3, as lane 2 but theincubation buffer contained 100 mM DMPO; lane 4, as lane 2 butcontaining 1000 �M H2O2; and, lanes 5–8, membrane-free hemolyzateincubated in phosphate buffer containing 100 mM DMPO and increas-ing concentrations of H2O2 (10, 50, 100, and 1000 �M, respectively).Before the analysis, the reaction medium was separated from RBC bycentrifugation. Then the RBC were washed three times with 0.1 Mphosphate buffer, resuspended to 1 ml with phosphate buffer, lyzed bythree cycles of freeze-thawing, and centrifuged at 10,000 � g for 10min at 4°C to separate the RBC debris. The supernatant (membrane-free hemolyzate, mainly hemoglobin) and RBC debris were taken andused to analyze for the presence of adducts by using immuno-spintrapping assays, as described in Experimental Procedures. Data shownare the mean values � SE or a representative Western blotting fromthree experiments using fresh RBC preparations.

836 D. C. RAMIREZ et al.

detected with 50 �M H2O2 using ELISA and 100 �MH2O2 using Western blotting (Fig. 6A and 6B).

The presence of adducts as detected by ELISA andWestern blotting in the total reaction mixture (RBC �supernatant), washed RBC ghosts, and membrane-freehemolyzate was studied. In addition, we made the fol-lowing observations in the RBC/DMPO/H2O2 system: (i)absence of adducts in the reaction medium by ELISA(Fig. 6A, open bars) and Western blotting (data notshown); (ii) absence of adducts in the washed RBCdebris (i.e., RBC membrane ghosts); and, (iii) presenceof adducts in the total suspension (RBC � reactionmedium), in the total hemolyzate, and in membrane-freehemolyzate (Fig. 6A, closed bars and Fig. 6B). Collec-tively, these results suggested that the formation andlocalization of the hemoglobin radical-derived nitroneadducts was inside the RBC. The higher concentration ofH2O2 needed to observe a significant immunosignal inRBC, in relation to purified oxyHb, is presumably due inpart to the high activity of antioxidant enzymes inside theRBC (e.g., catalase and glutathione peroxidase), whichconsume most of the H2O2 before it can react withoxyHb.

DISCUSSION

The extracellular concentration of hemoglobin inplasma is 3–6 mg/l and is usually eliminated by renalfiltration or hepatic capture [33]. Plasma hemoglobinconcentration can increase considerably as a result ofhemolysis induced by drugs and environmental hazardsor in patients given hemoglobin-based blood substitutes,which might saturate these systems [16]. Under theseconditions, hemoglobin can react with H2O2 to generatetissue damage [17,34]. Normally, tissue and plasmaH2O2 levels are quite low, but it has been estimated thatthe production of H2O2 by activated macrophages canreach concentrations as high as 100–600 �M [35]. In thebloodstream, H2O2 can be generated by autoxidation ofoxyHb [36,37], platelet activation [38], macrophage ac-tivation, or xenobiotic metabolism [39].

In this study, we used an antiserum that specificallyrecognizes protein radical-derived nitrone adducts to de-tect hemoglobin radical-derived nitrone adducts formedfrom purified human hemoglobin. We also demonstratedthat immuno-spin trapping is a sensitive and specificapproach to detect the formation of hemoglobin radical-derived nitrone adducts in a complex system such asRBC exposed to H2O2. Although a hemoglobin-derivedradical has been detected in blood with EPR, this spec-trum was not identified, is only detectable at liquidnitrogen temperature, and is thought to arise as a result ofthe freezing [24].

We observed higher hemoglobin-derived nitrone ad-duct formation in the metHb/DMPO/H2O2 system thanin the oxyHb/DMPO/H2O2 system (compare Figs. 1Aand 1B with 2A and 2B). This observation indicated thatmetHb is an intermediate in the formation of the globinradical, as is consistent with EPR spin trapping studies[14]. Moreover, we observed H2O2 concentration- andtime-dependent formation of hemoglobin radical-derivednitrone adducts by ELISA and Western blotting. Afterthe addition of catalase, the new epitope generated by thehemoglobin radical-derived nitrone adduct was shown tobe immunologically stable at 4°C for several days. Theimmunoreactivity of the hemoglobin radical-derived ni-trone adduct persists for a long time after the EPR signalhas decayed [40]. We found this novel approach to behighly sensitive and specific in the detection of hemo-globin radical-derived nitrone adducts.

The sensitivity of immuno-spin trapping is high-lighted by the recent report that the rate constant ofmetHb reaction with H2O2 is only 8.8 M�1s�1 [41],which is even slower than the Fenton reaction. Thisrate-limiting step forms ferryl hemoglobin and, after anintramolecular electron transfer to compound I, globinradical(s). After trapping of the globin radical(s) byDMPO, the radical adduct decays unimolecularly with ahalf-life of 97 s [40]. This decay rate is increased onlymarginally by H2O2. The product(s) of this decay is(are)unknown, but the nitrone epitope must be a major prod-uct.

Mao et al. [42] reported that the addition of DMPO toferrylHb reduces it to metHb. They proposed that DMPOis oxidized in the reaction. Since the oxidation potentialof DMPO is �1.63 V [43], whereas that of ferrylHb isonly �0.9 V [44–46], any oxidation of DMPO adductsby ferrylHb must be very slow. In contrast, the oxidationof the radical adduct nitroxide to the corresponding ni-trone by ferrylHb would be much faster. Such an in-tramolecular oxidation of Hb•-DMPO by ferrylHb to thecorresponding nitrone would also account for the first-order decay of Hb•-DMPO and the modest effect ofH2O2 on this decay [40].

In addition, immuno-spin trapping has been shown tobe useful in the study of the effect of antioxidants such asGSH and L-ascorbate on the formation of hemoglobinradical-derived nitrone adducts. GSH and L-ascorbatepreviously have been shown to quench free tyrosyl rad-icals [47] and protein-tyrosyl radicals [14,48]. In thisstudy, the L-ascorbate and GSH concentrations used arein the physiological range: L-ascorbate concentrations inRBC range from 40 to 70 �M and in plasma approxi-mately 45 to 85 �M [33]; although GSH levels in plasmaare low (5 �M) [33], intracellular levels of GSH aremuch higher (ranging from 2.5 to 10 mM). We suggestthat normal concentrations of L-ascorbate and GSH

837Immuno-spin trapping of a hemoglobin-derived tyrosyl radical

quench much of the hemoglobin-tyrosyl radical formedin human blood and might account for much of thedifficulty in the detection of protein-derived radicals invivo. Certainly, the effect of these antioxidants on thedetection of hemoglobin radical-derived nitrone adductsmay be due to a number of other factors including: (i) thereduction of nitroxide adducts to immunosilent hydroxy-lamines or (ii) an effect on the heme oxidation state[14,15].

To determine the nature of the amino acid residueinvolved in the hemoglobin-derived radical, we chemi-cally modified tyrosine and cysteine residues in metHb.These residues are frequently the source of the protein-derived free radicals generated when hemoproteins areexposed to H2O2 [4,10,13,29,49]. In bovine metHb, ty-rosine-�42 has been postulated to be the center of thehemoglobin-derived radical [14]. In addition, the globinmoiety contains other oxidizable groups such as thereadily accessible sulfhydryl group in a cysteine; thisresidue may be oxidized to a thiyl radical [12] by otherradicals, but probably not by ferryl-heme iron itself ow-ing to its limited reactivity toward sulfhydryl groups[17]. The definitive blocking of nitrone adduct formationby iodination of the metHb in conjunction with the smalleffect of the thiol reagent DTNB strongly suggests ty-rosine as the site of nitrone adduct formation. Appar-ently, the new epitope is generated by covalent bindingbetween a tyrosyl radical in the hemoglobin-derived rad-ical and DMPO.

In summary, in comparison with either direct EPR orEPR spin-trapping detection of the tyrosine-centered,protein-derived free radical in biological systems, im-muno-spin trapping has the following advantages: (i)exhibits higher sensitivity, i.e., the corresponding EPRexperiments used 0.83 mM metHb and 3.3 mM H2O2

[14]; (ii) requires smaller samples; (iii) gives the approx-imate molecular weight of the protein radical-derivednitrone adducts; (iv) permits the assay of many samplesat the same time; and, (v) does not require expensiveEPR instruments to detect the radical adducts. In addi-tion, the EPR detection of radical adducts inside cells hasrarely been reported, whereas we report here for the firsttime the detection of intracellular nitrone adducts usingimmuno-spin trapping.

Acknowledgements — The authors are grateful to Charles D. Detweilerwho obtained the antibody used in the study, to Jean T. Corbett for herexcellent technical assistance, and to Dr. Ann Motten and Ms. Mary J.Mason for their valuable help in the preparation of this manuscript.

REFERENCES

[1] Frey, P. A. Radical mechanisms of enzymatic catalysis. Annu.Rev. Biochem. 70:121–148; 2001.

[2] Alayash, A. I.; Patel, R. P.; Cashon, R. E. Redox reactions of

hemoglobin and myoglobin: biological and toxicological impli-cations. Antioxid. Redox Signal. 3:313–327; 2001.

[3] Ostdal, H.; Andersen, H. J.; Davies, M. J. Formation of long-livedradicals on proteins by radical transfer from heme enzymes—acommon process? Arch. Biochem. Biophys. 362:105–112; 1999.

[4] Giulivi, C.; Cadenas, E. Heme protein radicals: formation, fate,and biological consequences. Free Radic. Biol. Med. 24:269–279; 1998.

[5] Svistunenko, D. A. An EPR study of the peroxyl radicals inducedby hydrogen peroxide in the heam proteins. Biochim. Biophys.Acta 1546:365–378; 2001.

[6] DeGray, J. A.; Lassmann, G.; Curtis, J. F.; Kennedy, T. A.;Marnett, L. J.; Eling, T. E.; Mason, R. P. Spectral analysis of theprotein-derived tyrosyl radicals from prostaglandin H synthase.J. Biol. Chem. 267:23583–23588; 1992.

[7] Chen, Y.-R.; Gunther, M. R.; Mason, R. P. An electron spinresonance spin-trapping investigation of the free radicals formedby the reaction of mitochondrial cytochrome c oxidase with H2O2.J. Biol. Chem. 274:3308–3314; 1999.

[8] Sivaraja, M.; Goodin, D. B.; Smith, M.; Hoffman, B. M. Identi-fication by ENDOR of Trp191 as the free radical site in cyto-chrome c peroxidase compound ES. Science 245:738–740; 1989.

[9] Barr, D. P.; Gunther, M. R.; Deterding, L. J.; Tomer, K. B.;Mason, R. P. ESR spin-trapping of a protein-derived tyrosylradical from the reaction of cytochrome c with hydrogen perox-ide. J. Biol. Chem. 271:15498–15503; 1996.

[10] Gunther, M. R.; Sturgeon, B. E.; Mason, R. P. A long-livedtyrosyl radical from the reaction between horse metmyoglobinand hydrogen peroxide. Free Radic. Biol. Med. 28:709–719;2000.

[11] Kelman, D. J.; Mason, R. P. The myoglobin-derived radicalformed on reaction of metmyoglobin with hydrogen peroxide isnot a tyrosine peroxyl radical. Free Radic. Res. Commun. 16:27–33; 1992.

[12] Maples, K. R.; Jordan, S. J.; Mason, R. P. In vivo rat hemoglobinthiyl free radical formation following administration of phenyl-hydrazine and hydrazine-based drugs. Drug Metab. Dispos. 16:799–803; 1988.

[13] Maples, K. R.; Kennedy, C. H.; Jordan, S. J.; Mason, R. P. In vivothiyl free radical formation from hemoglobin following adminis-tration of hydroperoxides. Arch. Biochem. Biophys. 277:402–409; 1990.

[14] McArthur, K. M.; Davies, M. J. Detection and reactions of theglobin radical in haemoglobin. Biochim. Biophys. Acta 1202:173–181; 1993.

[15] Minetti, M.; Scorza, G.; Pietraforte, D. Peroxynitrite induceslong-lived tyrosyl radical(s) in oxyhemoglobin of red blood cellsthrough a reaction involving CO2 and a ferryl species. Biochem-istry 38:2078–2087; 1999.

[16] Alayash, A. I. Hemoglobin-based blood substitutes: oxygen car-riers, pressor agents, or oxidants? Nat. Biotechnol. 17:545–549;1999.

[17] Everse, J.; Hsia, N. The toxicities of native and modified hemo-globins. Free Radic. Biol. Med. 22:1075–1099; 1997.

[18] Irwin, J. A.; Ostdal, H.; Davies, M. J. Myoglobin-induced oxida-tive damage: evidence for radical transfer from oxidized myoglo-bin to other proteins and antioxidants. Arch. Biochem. Biophys.362:94–104; 1999.

[19] Kelman, D. J.; DeGray, J.; Mason, R. P. Reaction of myoglobinwith hydrogen peroxide forms peroxyl radical which oxidizessubstrates. J. Biol. Chem. 269:7458–7463; 1994.

[20] Giulivi, C.; Cadenas, E. Ferrylmyoglobin: formation and chemi-cal reactivity toward electron-donating compounds. Methods En-zymol. 233:189–202; 1994.

[21] Shiga, T.; Imaizumi, K. Electron spin resonance study on perox-idase- and oxidase-reactions of horse radish peroxidase and met-hemoglobin. Arch. Biochem. Biophys. 167:469–479; 1975.

[22] Patel, R. P.; Svistunenko, D. A.; Darley-Usmar, V. M.; Symons,M. C. R.; Wilson, M. T. Redox cycling of human methaemoglo-bin by H2O2 yields persistent ferryl iron and protein based radi-cals. Free Radic. Res. 25:117–123; 1996.

838 D. C. RAMIREZ et al.

[23] Giulivi, C.; Davies, K. J. A. Hydrogen peroxide-mediated ferryl-hemoglobin generation in vitro and in red blood cells. MethodsEnzymol. 231:490–496; 1994.

[24] Svistunenko, D. A.; Patel, R. P.; Voloshchenko, S. V.; Wilson,M. T. The globin-based free radical of ferryl hemoglobin isdetected in normal human blood. J. Biol. Chem. 272:7114–7121;1997.

[25] Davies, M. J.; Gilbert, B. C.; Haywood, R. M. Radical-induceddamage to proteins: E.S.R. spin-trapping studies. Free Radic. Res.Commun. 15:111–127; 1991.

[26] Mason, R. P. In vivo spin trapping. In: Rhodes, C. J., ed. Toxi-cology of the human environment. Liverpool, England: Taylor &Francis; 2000:49–70.

[27] Gunther, M. R.; Tschirret-Guth, R. A.; Witkowska, H. E.; Fann,Y. C.; Barr, D. P.; Ortiz De Montellano, P. R.; Mason, R. P.Site-specific spin trapping of tyrosine radicals in the oxidation ofmetmyoglobin by hydrogen peroxide. Biochem. J. 330:1293–1299; 1998.

[28] Witting, P. K.; Douglas, D. J.; Mauk, A. G. Reaction of humanmyoglobin and H2O2. Involvement of a thiyl radical produced atcysteine 110. J. Biol. Chem. 275:20391–20398; 2000.

[29] Witting, P. K.; Mauk, A. G. Reaction of human myoglobin andH2O2. Electron transfer between tyrosine 103 phenoxyl radicaland cysteine 110 yields a protein-thiyl radical. J. Biol. Chem.276:16540–16547; 2001.

[30] Detweiler, C. D.; Deterding, L. J.; Tomer, K. B.; Chignell, C. F.;Germolec, D.; Mason, R. P. Immunological identification of theheart myoglobin radical formed by hydrogen peroxide. FreeRadic. Biol. Med. 33:364–369; 2002.

[31] Buettner, G. R. On the reaction of superoxide with DMPO/•OOH.Free Radic. Res. Commun. 10:11–15; 1990.

[32] Winterbourn, C. C. Oxidative reactions of hemoglobin. MethodsEnzymol. 186:265–272; 1990.

[33] Faivre, B.; Menu, P.; Labrude, P.; Vigneron, C. Hemoglobinautooxidation/oxidation mechanisms and methemoglobin preven-tion or reduction processes in the bloodstream. Literature reviewand outline of autooxidation reaction. Art. Cells Blood Subs.Immob. Biotech. 26:17–26; 1998.

[34] Yoo, Y.-M.; Kim, K.-M.; Kim, S.-S.; Han, J.-A.; Lea, H.-Z.; Kim,Y.-M. Hemoglobin toxicity in experimental bacterial peritonitis isdue to production of reactive oxygen species. Clin. Diagn. Lab.Immunol. 6:938–945; 1999.

[35] Grisham, M. B.; Gaginella, T. S.; von Ritter, C.; Tamai, H.; Be,R. M.; Granger, D. N. Effects of neutrophil-derived oxidants onintestinal permeability, electrolyte transport, and epithelial cellviability. Inflammation 14:531–542; 1990.

[36] Misra, H. P.; Fridovich, I. The generation of superoxide radicalduring the autoxidation of hemoglobin. J. Biol. Chem. 247:6960–6962; 1972.

[37] Giulivi, C.; Hochstein, P.; Davies, K. J. A. Hydrogen peroxideproduction by red blood cells. Free Radic. Biol. Med. 16:123–129; 1994.

[38] Iuliano, L.; Violi, F.; Pedersen, J. Z.; Pratico, D.; Rotilio, G.;Balsano, F. Free radical-mediated platelet activation by hemoglo-bin released from red blood cells. Arch. Biochem. Biophys. 299:220–224; 1992.

[39] Kindt, J. T.; Woods, A.; Martin, B. M.; Cotter, R. J.; Osawa, Y.Covalent alteration of the prosthetic heme of human hemoglobinby BrCCl3. Cross-linking of heme to cysteine residue 93. J. Biol.Chem. 267:8739–8743; 1992.

[40] Kim, Y.-M.; Jeong, S.-H.; Yamazaki, I.; Piette, L. H.; Han, S.;Hong, S.-J. Decay studies of DMPO-spin adducts of free radicalsproduced by reactions of metmyoglobin and methemoglobin withhydrogen peroxide. Free Radic. Res. 22:11–21; 1995.

[41] Nagababu, E.; Ramasamy, S.; Rifkind, J. M.; Jia, Y.; Alayash,A. I. Site-specific cross-linking of human and bovine hemoglo-bins differentially alters oxygen binding and redox side reactionsproducing rhombic heme and heme degradation. Biochemistry41:7407–7415; 2002.

[42] Mao, G. D.; Thomas, P. D.; Poznansky, M. J. Oxidation of spintrap 5,5-dimethyl-1-pyrroline-1-oxide in an electron paramag-netic resonance study of the reaction of methemoglobin withhydrogen peroxide. Free Radic. Biol. Med. 16:493–500; 1994.

[43] McIntire, G. L.; Blount, H. N.; Stronks, H. J.; Shetty, R. V.;Janzen, E. G. Spin trapping in electrochemistry. 2. Aqueous andnonaqueous electrochemical characterizations of spin traps. J.Phys. Chem. 84:916–921; 1980.

[44] George, P.; Irvine, D. H. A possible structure for the higheroxidation state of metmyoglobin. Biochem. J. 60:596–604; 1955.

[45] Yamazaki, I.; Tamura, M.; Nakajima, R. Horseradish peroxidaseC. Mol. Cell. Biochem. 40:143–153; 1981.

[46] Koppenol, W. H.; Liebman, J. F. The oxidizing nature of thehydroxyl radical. A comparison with the ferryl ion (FeO2�). J.Phys. Chem. 88:99–101; 1984.

[47] Sturgeon, B. E.; Sipe, H. J. Jr.; Barr, D. P.; Corbett, J. T.;Martinez, J. G.; Mason, R. P. The fate of the oxidizing tyrosylradical in the presence of glutathione and ascorbate. Implicationsfor the radical sink hypothesis. J. Biol. Chem. 273:30116–30121;1998.

[48] Ostdal, H.; Skibsted, L. H.; Andersen, H. J. Formation of long-lived protein radicals in the reaction between H2O2-activatedmetmyoglobin and other proteins. Free. Radic. Biol. Med. 23:754–761; 1997.

[49] Hawkins, C. L.; Davies, M. J. Generation and propagation ofradical reactions on proteins. Biochim. Biophys. Acta 1504:196–219; 2001.

ABBREVIATIONS

DMPO—5,5-dimethyl-1-pyrroline N-oxideDTNB—5,5'-dithiobis(2-nitrobenzoic acid)DTPA—diethylenetriamine-pentaacetic acidELISA—enzyme-linked immunosorbent assayEPR—electron paramagnetic resonanceGSH—reduced glutathionemetHb—methemoglobinoxyHb—oxyhemoglobinRBC—red blood cells

839Immuno-spin trapping of a hemoglobin-derived tyrosyl radical