Oxidative status of red blood cells, neutrophils,and platelets in paroxysmal nocturnal hemoglobinuria

Johnny Amer, Orly Zelig, and Eitan Fibach

Department of Hematology, Hadassah–Hebrew University Medical Center, Jerusalem, Israel

(Received 22 April 2007; revised 27 November 2007; accepted 5 December 2007)

Objective. Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired stem-cell disorderassociated with intravascular hemolysis and thrombosis. Hemolysis is caused by the hypersen-sitivity of PNH-red blood cells (RBC) to complement-mediated lysis due to deficiency in thesurface glycosyl phosphatidylinositol–anchored antigens, CD55 and CD59. Thrombosis maybe related to the platelet tendency to undergo hyperactivation. We previously suggestedthat hemolysis and thrombosis in other hemolytic anemias are related to oxidative stress.In the present study, we assessed the oxidative status of blood cells in PNH and tested thepotential protective effects of antioxidants.

Materials and Methods. Blood samples were obtained from 11 PNH patients and 11 normalcontrol donors. Flow cytometry was used to measure oxidative stress markers in conjunctionwith the PNH immunophenotype.

Results. Results indicated that abnormal, CD55/CD59-negative, RBC, neutrophils, and plate-lets are under oxidative stress. Their intracellular reactive oxygen species, membrane lipidperoxides, and external phosphatidylserine were higher and their reduced glutathione waslower than CD55/CD59-positive cells of the same patient or cells of normal controls. PNH-RBC were hypersensitive to an oxidative insult (e.g., hydrogen peroxide) and their oxidativestatus increased following interaction with complement, prior to hemolysis. Antioxidantsreduced this hemolysis as well as activation of PNH platelets.

Conclusion. We propose that oxidative stress mediates the symptoms of PNH and suggest thatantioxidants might be considered as a therapeutic modality. � 2008 ISEH - Society forHematology and Stem Cells. Published by Elsevier Inc.

Experimental Hematology 36 (2008) 369–377

Paroxysmal nocturnal hemoglobinuria (PNH) is a stem celldisorder caused by acquired somatic mutations predomi-nantly in the phosphatidylinositol glycan complementationclass A (PIG-A) gene located on the X-chromosome atXp22.1 [1,2]. The PIG-A gene produces a protein responsi-ble for the first step in production of the glycosyl phospha-tidylinositol (GPI)- anchor, by which various proteins areattached to the cell membrane. Consequently, there is a par-tial or complete deficiency of GPI-anchored proteins on thesurfaces of PNH hematopoietic stem cells and their progeny(for a recent review see [3]). The disease is characterized byintravascular hemolysis, frequent infections, bone marrowhypoplasia and cytopenia, and a high incidence of life-threatening venous thrombosis [4,5]. Despite intensive

Offprint requests to: Eitan Fibach, Ph.D., Department of Hematology,

Hadassah–Hebrew University Medical Center, Ein-Kerem, P.O. Box

12000, Jerusalem 91120, Israel; E-mail: Fibach@yahoo.com

study, the underlying mechanisms of many symptoms inPNH remain obscure [6]. The functions of GPI-linked pro-teins are extremely varied [7]. PNH-red blood cells (RBC)are abnormally sensitive to complement-mediated lysis [8],as demonstrated in vitro by the acidified serum lysis test(Ham’s test) [9], due to a deficiency in the GPI-anchoredcomplement inhibitor proteins, CD55 (decay acceleratingfactor) and CD59 (membrane inhibitor of reactive lysis).CD55 inhibits complement at the level of C3, whereasCD59 prevents terminal complement components (C5b-9)from forming the hemolytic membrane pore [10]. Theseproteins are also deficient on other blood cells, such as onpolymorphonuclear neutrophils (PMN), monocytes, lym-phocytes, and platelets [4,11].

We have previously shown that in chronic hemolytic ane-mia, such as thalassemia [12–16] and sickle cell disease[17], RBC, platelets, and PMN are under oxidative stress.This might be involved in the short RBC lifespan leading

0301-472X/08 $–see front matter. Copyright � 2008 ISEH - Society for Hematology and Stem Cells. Published by Elsevier Inc.

doi: 10.1016/j.exphem.2007.12.003

370 J. Amer et al. / Experimental Hematology 36 (2008) 369–377

to chronic anemia, increased platelet activation leading tohigh incidence of thrombotic events, and diminished capac-ity of PMN to respond by oxidative burstdleading to recur-rent bacterial infections in these patients. Because thesesymptoms are also present in PNH [5,18,19], in the presentstudy we used flow cytometry to analyze the oxidative statusof these cells in PNH compared with their normal counter-parts. Results indicated that the abnormal RBC, PMN andplatelets, lacking CD55/CD59, are under oxidative stress;their reactive oxygen species (ROS) were higher and theircontent of reduced glutathione (GSH) was lower thanCD55/CD59-positive cells of the same patient as well ascells derived from the blood of normal controls. We alsofound that complement-mediated hemolysis of PNH-RBCinvolves oxidative stress: in vitro treatment of PNH-RBCwith complement resulted in oxidative stress prior to anysigns of hemolysis, suggesting its involvement in the pro-cess; addition of antioxidants reduced the hemolysis. Ourresults propose that oxidative stress participates in thepathological consequences of PNH, in particularly anemiaand thrombosis, and suggest that antioxidants might be con-sidered as a potential therapeutic modality.

Materials and methods

Blood samplesPeripheral blood of 11 PNH patients and normal donors ofmatched age and gender was used after all diagnostic laboratorytests were completed. Blood samples were obtained in tubes con-taining ethylene diamine tetraacetic acid, unless otherwise stated.Patients were diagnosed based on their clinical symptoms and de-ficiency in the GPI-anchored proteins, CD55 and CD59, as mea-sured by DiaMed-ID Micro Typing System (Morat, Switzerland)and by flow cytometry on RBC and PMN. The proportions ofCD55– and CD59– populations in their RBC, platelet, and PMNpopulations ranged from 6% to 55%. Informed consent wasobtained in all cases.

RBC were studied at concentration of 1 to 5 � 106/mL (unlessotherwise indicated) by diluting the blood with Caþþ and Mgþþ

free phosphate-buffered saline (PBS; Biological Industries, Kib-butz Beit-HaEmek, Israel). Platelets were obtained by centrifugingwhole blood for 12 minutes at 800 rpm, collecting the supernatant(the platelet-rich plasma) and centrifuging again for 3 minutes at1200 rpm; the platelets-containing pellet was washed twice andresuspended in PBS. To obtain PMN-enriched fractions wholeblood was diluted (1:1) with PBS and mixed with equal volumeof 3% gelatin (Sigma, St Louis, MO, USA); left to stand for 30minutes at room temperature, and the supernatant, containingmostly PMN, was collected and the cells washed with PBS. Allexperiments started within 2 hours of blood withdrawal.

Treatment with oxidants and antioxidantsRBC and platelet suspensions were treated with or without theantioxidants N-acetylcysteine (NAC, Sigma), vitamin C (DolderLtd, Basle, Switzerland), tocotrinol (a vitamin E derivative)(TwinLab, Rankonkona, NY, USA) or the oxidant hydrogen per-oxide (H2O2) (Sigma) as specified in the Results section.

Measurements of oxidative stressROS assay. Cells were incubated with 20-70-dichlorofluoresceindiacetate (DCF; Sigma), dissolved in methanol (Bio Lab, Jerusa-lem, Israel), at final concentrations of 0.1 or 0.4 mM in PMN orRBC and platelets, respectively. After incubation at 37�C for 15minutes in a humidified atmosphere of 5% CO2 in air, the cellswere washed, resuspended in PBS to the original cell concentra-tion and analyzed by flow cytometry. In some experiments,1 mM H2O2 was added for 15 minutes before analysis.

GSH assay. Cells were diluted with PBS and spun down. Thepellet was suspended and incubated for 3 minutes at room temper-ature with 40 mM (final concentration) of mercury orange [1(4-chloromercuryphenyl-azo-2-naphthol)] (Sigma). A 100 mM stocksolution of mercury orange was made up in acetone and storedat 4�C. Cells were then washed and resuspended in PBS.

Lipid peroxidation assay. RBC suspensions in PBS were labeledwith 50 mM N- (fluorescein-5-thiocarbamoyl) 1,2-dihexadeca-noyl-sn-glycero-3-phosphoethanolamine, triethyl ammonium salt(fluor-DHPE; Molecular Probes Inc., Eugene, OR, USA) dissolvedin ethanol. The cells were incubated for 1 hour at 37�C in a humid-ified atmosphere of 5% CO2 in air with continuous agitation,centrifuged once to remove unbound label and resuspendedin PBS.

Phosphatidylserine (PS) assay. Cells resuspended in 100 mLcalcium binding buffer (10 mM HEPES, 140 mM NaCl, and 2.5mM CaCl2 [pH 7.4]) were stained for 20 minutes at roomtemperature with 5 mL Annexin-V–phycoerythrin.

Flow cytometryFollowing staining and washing with PBS, cells were analyzed bya flow cytometer (FACSCalibur; Becton-Dickinson, Immunofluor-ometry Systems, Mountain View, CA, USA). Cells were passed ata rate of about 1000 per second, using saline as the sheath fluid. A488-nm argon laser beam was used for excitation. RBC, PMN, andplatelets were gated based on their size (forward light scatter,FSC) and granularity (side light scatter, SSC) as described previ-ously [12–15,17]. The identity of each cell population was verifiedby staining with antibodies to glycophorin-A, CD41, and CD15specific for RBC, platelets, and PMN, respectively. To determinethe presence of GPI proteins, cells were labeled with a phycoery-thrin-conjugated anti-CD55 antibody and a fluorescein isothiocya-nate–conjugated anti-CD59 antibody (IQ Products, Groningen,The Netherlands). The mean fluorescence channel (MFC) of atleast 10,000 cells was calculated by the CellQuest software (Bec-ton-Dickinson). For each experiment, unstained cells served ascontrols; their MFC was in the range of 3 to 7. The MFC of cellsstained with DCF, mercury orange and Annexin-V is proportionalto generation of ROS, the content of GSH and extent of externalPS, respectively; while that of fluor-DHPE–stained cells isreversely proportional to their extent of lipid peroxidation [14].

RBC were counted by adding a known number of fluorescentplastic beads (FACSCount Kit); the RBC/bead mixtures wereanalyzed by the FACSCalibur. The absolute number of RBCwas calculated from the ratio of RBC to beads following themanufacturer’s instructions.

371J. Amer et al./ Experimental Hematology 36 (2008) 369–377

Statistical analysisResults of Figures 3 to 7 are expressed as the mean 6 SD andcompared by the two-sample Student’s t-test for differences inmeans.

Results

Oxidative status of PNH cellsTo determine the oxidative status of RBC, PMN, and plate-lets each population was gated based on its size (FSC) and

granularity (SSC) and its identity was confirmed by stainingwith specific antibodies as described previously [17]. Fig-ure 1 shows distribution histograms and the MFC of RBC(Fig. 1A–C) and platelets (Fig. 1D–F) derived from a nor-mal donor (white histograms) and a PNH donor (gray histo-grams) with respect to ROS, GSH, and lipid peroxidation orPS exposure. We found that both RBC and platelets fromthe PNH patient had higher ROS, and lower GSH than nor-mal RBC. The membrane lipid peroxidation of PNH-RBCand the external PS of PNH-platelets were higher than intheir normal counterparts.

Figure 1. Flow cytometry analysis of the oxidative status of red blood cells (RBC) and platelets in normal and paroxysmal nocturnal hemoglobinuria (PNH)

donors. On the day of the experiment, 23% of the patients RBC and 32% of his platelets were negative for CD55. RBC were assayed for reactive oxygen species

(ROS) (A) Reduced glutathione (GSH) (B) and lipid peroxidation (C), and platelets for ROS (D), GSH (E) and phosphatidylserine (PS) (F). ROS was measured

following exposure for 15 minutes to 1 mM H2O2. The fluorescence distribution histogram and the mean fluorescence channels (MFC) of each population

derived from the normal donor (white histograms) and the PNH donor (gray histograms) are shown. MFC of cells stained with 20-70-dichlorofluorescin diac-

etate, mercury orange and annexin V is proportional to generation of ROS, the content of GSH and extent of external PS, respectively; while that of fluor-

DHPE–stained cells is reversely proportional to their extent of lipid peroxidation.

372 J. Amer et al. / Experimental Hematology 36 (2008) 369–377

Figure 2. The relationship between the paroxysmal nocturnal hemoglobinuria (PNH) phenotype and oxidative status of red blood cells (RBC). RBC of

a PNH patient (with 20% to 27% CD55/CD59-negative cells) were stained with phycoerythrin (PE)-conjugated antibody to CD55 and 2-7-dichlorofluorescin

diacetate (DCF) for reactive oxygen species (ROS) (A,B) or with fluorescein isothiocyanate–conjugated (FITC) antibody to CD59 and PE-conjugated An-

nexin V for phosphatidylserine (PS) (C,D). Forward light scatter (FSC) vs CD55 (A) and FSC vs CD59 (C) dot plots are presented. Gates were set on positive

(R1), intermediate (R2) and negative (R3) CD55/CD59 expressing RBC, and the distribution histogram plus the mean fluorescence channel (MFC) of RBC in

each gate with respect to ROS (B) and PS (D) are shown. The results indicate a reverse relationship between the PNH phenotype (CD55/CD59 expression)

and oxidative stress (ROS generation/PS externalization).

Table 1 shows the MFC of RBC and platelets from 11PNH patients and 11 normal donors stained for ROS,GSH, and lipid peroxidation both at their basal level (unsti-mulated) and following stimulation with H2O2. Resultsindicate significant differences (p ! 0.005) with respectto these parameters between normal and PNH cells.

The blood cell populations in PNH patients constitutea mosaic of cells derived from normal stem cells, havingnormal phenotype, and a clonal population of cells derivedfrom abnormal stem cells with varying degrees of GPI pro-tein deficiency [20]. To determine the oxidative status ofeach of these populations present in the same patient, bloodcells of PNH patients (with 20% to 27% CD55/CD59-negative cells) and normal controls were stained simulta-neously for CD55 and ROS or for CD59 and PS (Figs. 2and 3). Gates were set on platelets, RBC and PMN, andeach cell population was analyzed for CD55 and ROS orCD59 and PS. A FSC vs CD55 dot plot of RBC fromone PNA patient is demonstrated in Figure 2A. Gates

were set on negative, intermediate and strongly CD55-expressing RBC, and the distribution of RBC in each gatewith respect to ROS generation is presented as a histogramin Figure 2B. The relationship between CD59 and PS exter-nalization is similarly presented in Figure 2C and D. Theresults indicate a reverse relationship between the PNHphenotype (CD55/CD59 expression) and oxidative stress(ROS generation/PS externalization). These results wereconfirmed for RBC, platelets, and PMN obtained fromfour PNH patients (Fig. 3). Figure 3A also shows thatCD55-positive cells from PNH patients have higher ROSlevels than the same cell types derived from normal donors,suggesting that extracellular factors in PNH patients mayaffect the phenotype of genotypically normal cells.

Susceptibility of PNH-RBC to oxidative insultTo compare their susceptibility to oxidative insult, RBCobtained from normal and PNH donors were incubated atroom temperature with different concentrations of H2O2

Figure 3. The oxidative status of CD55/CD59 negative and positive blood cells. Blood samples obtained from four normal donors and four paroxysmal

nocturnal hemoglobinuria (PNH) patients were stained for CD55 and rective oxygen species (ROS) or for CD59 and phosphatidylserine (PS). Gates

were set on platelets (PLT), RBC, and polymorphonuclear leukocytes (PMN), and the intensity of CD55 and ROS or CD59 and PS fluorescence was deter-

mined as described in Materials and Methods. The proportion of CD55– and CD59– cells in the patients’ blood ranged from 20% to 27%. Average 6 SD of

the mean fluorescence channel (MFC) of ROS (A) and PS (B) in cells obtained from normal donors and CD55– and CD55þ cells derived from PNH patients

are shown. The results with cells from PNH patients show higher ROS/PS levels in CD55/CD59-negative cells than CD55/CD59-positive cells (p ! 0.01).

CD55/CD59-positive cells from the PNH had higher ROS/PS levels than the same cell types derived from normal donors (p ! 0.01).

373J. Amer et al./ Experimental Hematology 36 (2008) 369–377

for 72 hours. RBC were then counted by flow cytometryfollowing addition of predetermined number of fluorescentbeads. The results (Fig. 4) indicated a dose-responsedecrease in cell number as a result of hemolysis; this effectwas more pronounced with PNH-RBC than with normalRBC. Spectrophotometric measurement of hemoglobin inthe hemolysate supported these findings (data not shown).

We next determined the effect of H2O2 treatment on thephenotype composition of the RBC population. Followingovernight treatment of PNH-RBC with 1 mM H2O2,CD55/CD59-negative cells decreased from 22% down to0.2%, indicating a selective lysis of these cells relative toCD55/CD59-positive cells.

Figure 4. Hydrogen peroxide-induced lysis of red blood cells (RBC).

RBC obtained from normal donors and paroxysmal nocturnal hemoglobin-

uria (PNH) patients (having 20–27% CD55/CD59-negative cells) were

washed and diluted in PBS and incubated with different concentrations

of H2O2 at room temperature for 72 hours. RBC were then counted by

flow cytometry as described in Materials and Methods. Results are

presented as the RBC count in normal and PNH samples treated with

H2O2 compared to untreated controls, normal and PNH RBC, respectively

(each taken as 100%). The mean 6 SD of four experiments carried out

with RBC from different normal and PNH patients are shown.

Involvement of oxidative stressin complement-mediated effect on PNH cellsTo investigate the association between of complement sen-sitivity of PNH-RBC and their oxidative status, RBC fromnormal and PNH donors were exposed at 37�C to freshnormal ABO-compatible serum. Complement-inactivatedserum (heated at 56�C for 30 minutes) was used as control.At various time points, RBC samples were assayed forROS. The results (Fig. 5) show a significant increase inROS generation in PNH-RBC, but not in normal RBC,treated with complement-containing serum during the firsthour of treatment, prior to any sign of hemolysis. Theseresults indicate that ROS are increased during the earlystages of complement-mediated lysis of PNH-RBC andsuggest the involvement of oxidative stress in the process.

Next, normal and PNH blood samples, obtained in hepa-rin, were incubated at 37�C in their plasma in the absence orpresence of different antioxidants: NAC, vitamin C, or toco-trinol (a vitamin E derivative). Counting total RBC beforeand after 16-hour incubation showed a 25% reduction, indi-cating cell lysis. This hemolysis was completely inhibitedby the antioxidants (Fig. 6). No hemolysis was observedin the normal samples (data not shown). To determinewhether the antioxidants protected the abnormal RBC,RBC obtained from PNH patients were phenotyped forCD55 before and after incubation in their plasma with orwithout the antioxidants. Results (Fig. 6) indicate that inthe absence of antioxidants such incubation caused a selec-tive elimination of CD55– RBC, decreasing from 25% to12% (with a concomitant increase in the percentage ofCD55þ RBC). This was inhibited by treatment with antiox-idants. The results suggest that hemolysis induced by plasmacomplement in abnormal PNH-RBC is mediated by oxida-tive stress and that it can be alleviated by antioxidants.

Effect of antioxidants on oxidative stress in plateletsThe effect of antioxidants on platelets was first determinedby incubating heparinized platelet-rich plasma from normal

374 J. Amer et al. / Experimental Hematology 36 (2008) 369–377

Figure 5. Effect of complement on reactive oxygen species (ROS) of red blood cells (RBC). Normal and paroxysmal nocturnal hemoglobinuria (PNH)-

RBC were incubated at 37�C with complement-containing serum (C0) or complement-inactivated serum (heated at 56�C for 30 minutes) (No C0). The sera

were ABO compatible with the treated RBC. At the indicated times, aliquots of RBC were assayed for ROS. Results are expressed as the average 6 SD

of the mean fluorescence channel of four experiments carried out with RBC from different normal donors and PNH patients (having 20–27% CD55/

CD59-negative cells).

donors or PNH patients overnight at 37�C in the absence orpresence of NAC or vitamin C, followed by measuring theplatelets’ ROS and PS. The results (Fig. 7A and B) indicatehigher ROS generation and PS exposure in PNH-plateletsthan in normal platelets following incubation in their autol-ogous plasma. Antioxidants inhibited these effects. We nextstudied the effect of autologous PNH plasma and heterolo-gous (ABO-compatible) normal plasma on platelets fromPNH patients. The results (Fig. 7C) indicate that PNHplatelets generated more ROS when incubated with autolo-

Figure 6. Effect of antioxidants on complement-induced lysis of paroxys-

mal nocturnal hemoglobinuria (PNH)-red blood cells (RBC). Blood sam-

ples obtained in heparin from four patients with PNH were diluted and

incubated for 16 hours in autologous plasma at 37�C with and without

the antioxidants N-acetylcysteine (NAC), vitamin C (Vit C) or tocotrinol

(1 mM each). Results show the percentages of total RBC after incubation

relative to the initial RBC inoculum (gray columns) and the percentage of

CD55– RBC in the total RBC population (black). The original RBC inoc-

ulum of the patients contained 20% to 27% CD55– RBC (mean: 25%).

Following incubation, total RBC decreased to 83% (17% lysis) while

CD55– RBC dropped from 25% to 12% (52% lysis). The results show

a decrease in hemolysis of total RBC and in of CD55– RBC by treatment

with the antioxidants.

gous plasma than with normal plasma. In both plasmas, in-activation of complement by heating at 56�C for 30 minutesdid not reduce their effect on ROS generation. NAC or vi-tamin C reduced ROS in both plasmas regardless the pres-ence of complement. The results suggest that plateletactivation and the effect of antioxidants were comple-ment-independent.

DiscussionIn PNH, due to a somatic mutation in the PIG-A gene, theGPI-anchor protein biosynthesis is defective in the hemato-poietic stem cells and their progeny. The resulting defi-ciency in the GPI-anchored complement inhibitorproteins, CD55 and CD59, are responsible for the increasedsensitivity of PNH-RBC to complement-mediated lysis[21]. Although the pathological consequences of GPI-protein deficiency on other blood lineages remain obscure,there is a high incidence of life-threatening venous throm-bosis and some patients suffer from frequent infections,indicating the role of platelets and possibly PMN in thedisease [4]. Oxidative stress in blood cells has beenreported to be associated with these pathologies in otherforms of hemolytic anemias [13–15,17,22,23]. In the pres-ent study, using flow cytometry methodology which permitssimultaneous analysis of various blood cell types, we foundthat in PNH, the RBC, PMN, and platelets are under oxida-tive stress: their ROS is higher and their GSH is lower thantheir normal counterparts. In addition, their membranelipids are highly peroxidized and many cells carry PS moi-eties on their outer surface. Lipid peroxidation and exter-nalization of PS are considered significant steps inpathogenic processes related to oxidative stress [24].

In PNH, both pathological and normal stem cellscoexist, giving rise to a mosaic of normal blood cells, hav-ing the CD55þCD59þ phenotype, and abnormal cells with

375J. Amer et al./ Experimental Hematology 36 (2008) 369–377

Figure 7. The effect of complement and antioxidants on platelets oxidative stress. (A,B). Heparinized platelet-rich plasma samples from normal donors or

paroxysmal nocturnal hemoglobinuria (PNH) patients (having 22–30% CD55/CD59-negative platelets) were incubated overnight at 37�C in the presence of 1

mM N-acetylcysteine (NAC), vitamin C (Vit C), or none. Reactive oxygen species (ROS), measured following exposure for 15 minutes to 1 mM H2O2 (A),

expressed as the average 6 SD of the average 6 SD of the mean fluorescence channel (MFC) and phosphatidylserine (PS) (B) expressed as mean 6 SD of

the percentage of positive platelets, respectively, are shown. (C) Platelets from heparinized blood of PNH patients were incubated overnight at 37�C in their

autologous plasma or heterologous (ABO-compatible) normal plasma in the presence of 1 mM of NAC, Vit C or none. Plasmas were either heated at 56�C for

30 minutes to inactivate complement (No C0) or left unheated (C0). ROS, measured following exposure for 15 minutes to 1 mM H2O2, expressed as the

average 6 SD of the mean fluorescence channel (MFC) is shown. The results indicate that platelets generated more ROS when incubated with autologous

plasma than normal plasma. In both plasmas, C0-inactivation did not reduce their effect on ROS generation. NAC and Vit C reduced ROS in both plasmas

regardless the presence of C0.

a CD55–CD59– phenotype [9]. In double-staining experi-ments with fluorescent antibodies to CD55/CD59 and oxi-dative state markers, we demonstrated a higher oxidativestatus in cells derived from the pathological clone com-pared with cells derived from normal clones in the samepatient.

We also studied the oxidative status of lymphoblastoidcell lines derived from three patients with PNH. To establishthese lines, peripheral blood mononuclear cells wereinfected with Epstein-Barr virus and cultured for about 3

weeks in fetal bovine serum supplemented medium.Labeling of the cells with anti-CD55 and anti-CD59 anti-bodies indicated that about 50% of the cells had the PNHphenotype and 50% had normal phenotype. Cells werethen stained simultaneously with these antibodies andDCF in order to measure ROS generation. The results indi-cated that both populations had similar levels of ROS. Cul-turing these lines for 3 days in the presence of eithercomplement-containing (ABO compatible) serum or com-plement-inactivated (56�C for 30 minutes) serum did not

Table 1. The oxidative status of normal and PNH cells

Unstimulated H2O2-Stimulated

ROS GSH LP ROS GSH LP

RBC

Normal 12 6 4 620 6 140 681 6 98 359 6 120 510 6 79 595 6 100

PNH 17 6 6 312 6 110 195 6 75 600 6 165 199 6 68 171 6 78

Platelets

Normal 7 6 6 385 6 112 ND 196 6 87 200 6 79 ND

PNH 13 6 4 220 6 100 ND 320 6 56 101 6 44 ND

Red blood cells (RBC) and platelets obtained from 11 normal and 11 paroxysmal nocturnal hemoglobinuria (PNH) patients were stained for reactive oxygen

species (ROS), reduced glutathione (GSH) and lipid peroxidation (LP) before and after 15-minute stimulation with 1 mM H2O2. Results are presented as the

average 6 SD of the mean fluorescence channel. The differences between normal and PNH in each category were statistically significant. ND 5 not done.

376 J. Amer et al. / Experimental Hematology 36 (2008) 369–377

change the ROS levels. These results show that in contrast toRBC and platelets derived directly from the peripheralblood of PNH patients, in vitro established lymphoblastoidcells lines were not under oxidative stress. This could bedue to the higher antioxidative capacity of these proliferatingcultured cells compared to RBC, platelets and neutrophils.

The higher oxidative status of blood cells in patientswith PNH could be the result of cellular abnormalitiesrelated to the basic PNH defect, e.g., the lack or reducedexpression of specific membrane components, such asCD55 and CD59, that affect complement fixation. Never-theless, among the patients studied we did not find a clearlinear correlation between the proportion of CD55/CD59-negative cells and oxidative stress. This finding points tothe influence of additional, extracellular, factors. These fac-tors could include complement-mediated destruction and/oran immune attack that is often associated with the diseaseand high levels of serum iron-containing compounds, e.g.,free heme and hemoglobin due to intravascular hemolysis.Iron is known to participate in biochemical reactions (e.g.,the Fenton reaction) that generate free radicals and thusinduces oxidative stress [25,26]. We have previously dem-onstrated that incubating normal RBC, PMN, or plateletswith iron-containing compounds increased their oxidativestatus [13–15] while iron-chelators reduced it [13]. Extra-cellular factors may affect cells derived from both normaland abnormal clones as well as RBC derived from trans-fused blood (in patients undergoing transfusion therapy).Indeed, in the present study, we noticed that normal cells(with the CD55/CD59-positive phenotype) in the blood ofPNH patients were at higher oxidative status than cellsfrom blood of normal donors (Fig. 3).

Complement-mediated hemolysis due to the deficiencyin CD55/CD59 is the major feature of PNH. Hemolysisin other forms of hemolytic anemia, such as in thalassemiaand sickle cell anemia has been reported to involve oxida-tive stress [13–15,17,22,23]. The results of the presentstudy suggest that oxidative stress is also involved in hemo-lysis in PNH. In agreement with a previous report [27], weshowed that PNH-RBC are more sensitive to oxidants, suchas hydrogen peroxide, than normal RBC. It has been previ-ously reported that oxidants activate complement [28]. Inthe present study, we showed that in vitro treatment ofPNH-RBC with complement resulted in oxidative stressprior to any signs of hemolysis and that anti-oxidantsreduced hemolysis, suggesting the involvement of oxidativestress as a mediator or an adjuvant.

Venous thrombotic events occur in up to half of PNHpatients and constitute a major cause of death [29]. Thereis evidence suggesting that platelets in PNH that lack theGPI-anchored proteins are susceptible to complement[30]. When this occurs, the platelets are not destroyed,but rather become hyperactivated [19] and produce procoa-gulants [6]. In the present study, we demonstrate that PNH-platelets are at oxidative stress and expose PS, a marker of

activation [31], on their outer membrane. The relationshipbetween oxidative stress and platelet activation has beenwell-documented: Platelet activators, e.g., thrombin,increase ROS [13,32,33], whereas oxidants, through pro-duction of ROS, cause platelet activation [13,34,35]. Ourresults suggest that platelet activation, as determined byROS generation and PS externalization (Fig. 7) and the pro-tective effect of antioxidants are not related to the activityof complement, as both were observed in the presence ofcomplement-inactivated plasma. Oxidative stress in PNHplatelets could be due to platelet abnormalities and/or toextraplatelet factors, i.e., iron-containing compounds, asdiscussed above for RBC. In addition, many types of cells,including RBC, respond to oxidative stress by externaliza-tion of their PS, which acts as a procoagulant that amplifiesthrombin generation and thus initiates platelet activation[36]. Taken together, our results suggest that, in additionto other mechanisms, oxidative stress is involved in plateletactivation that leads to thromboembolic complications inPNH.

Results of the present study suggest that oxidative stressparticipates in mediating the pathological consequences ofPNH. Thus, treatment with antioxidants might be consid-ered as a therapeutic modality in PNH. Eculizumab, a hu-manized monoclonal antibody that specifically targets thecomplement protein C5 and prevents its cleavage [37] hasbeen recently introduced for treatment of certain aspectsof PNH [38]. Antioxidants might serve as an inexpensiveadjuvant or alternative to this treatment.

References1. Takeda J, Miyata T, Kawagoe K, et al. Deficiency of the GPI anchor

caused by a somatic mutation of the PIG-A gene in paroxysmal

nocturnal hemoglobinuria. Cell. 1993;73:703–711.

2. Inoue N, Murakami Y, Kinoshita T. Molecular genetics of paroxysmal

nocturnal hemoglobinuria. Int J Hematol. 2003;77:107–112.

3. Parker CJ. The pathophysiology of paroxysmal nocturnal hemoglobin-

uria. Exp Hematol. 2007;35:523–533.

4. Meletis J, Terpos E. Recent insights into the pathophysiology of

paroxysmal nocturnal hemoglobinuria. Med Sci Monit. 2003;9:

RA161–RA172.

5. Smith LJ. Paroxysmal nocturnal hemoglobinuria. Clin Lab Sci. 2004;

17:172–177.

6. Rosse WF. New insights into paroxysmal nocturnal hemoglobinuria.

Curr Opin Hematol. 2001;8:61–67.

7. Hall C, Richards SJ, Hillmen P. The glycosylphosphatidylinositol

anchor and paroxysmal nocturnal haemoglobinuria/aplasia model.

Acta Haematol. 2002;108:219–230.

8. Holguin MH, Fredrick LR, Bernshaw NJ, Wilcox LA, Parker CJ. Iso-

lation and characterization of a membrane protein from normal human

erythrocytes that inhibits reactive lysis of the erythrocytes of paroxys-

mal nocturnal hemoglobinuria. J Clin Invest. 1989;84:7–17.

9. Marsh JC, Elebute MO. Stem cells in paroxysmal nocturnal haemoglo-

binuria and aplastic anaemia: increasing evidence for overlap of

haemopoietic defect. Transfus Med. 2003;13:377–386.

10. Walport MJ. Complement. First of two parts. N Engl J Med. 2001;344:

1058–1066.

11. Rosse WF. Phosphatidylinositol-linked proteins and paroxysmal

nocturnal hemoglobinuria. Blood. 1990;75:1595–1601.

377J. Amer et al./ Experimental Hematology 36 (2008) 369–377

12. Amer J, Goldfarb A, Fibach E. Flow cytometric measurement of reac-

tive oxygen species production by normal and thalassaemic red blood

cells. Eur J Haematol. 2003;70:84–90.

13. Amer J, Fibach E. Oxidative status of platelets in normal and thalas-

semic blood. Thromb Haemost. 2004;92:1052–1059.

14. Amer J, Goldfarb A, Fibach E. Flow cytometric analysis of the oxida-

tive status of normal and thalassemic red blood cells. Cytometry A.

2004;60:73–80.

15. Amer J, Fibach E. Chronic oxidative stress reduces the respiratory

burst response of neutrophils from beta-thalassaemia patients.

Br J Haematol. 2005;129:435–441.

16. Amer J, Etzion Z, Bookchin RM, Fibach E. Oxidative status of vali-

nomycin-resistant normal, beta-thalassemia and sickle red blood cells.

Biochim Biophys Acta. 2006;1760:793–799.

17. Amer J, Ghoti H, Rachmilewitz E, Koren A, Levin C, Fibach E. Red

blood cells, platelets and polymorphonuclear neutrophils of patients

with sickle cell disease exhibit oxidative stress that can be ameliorated

by antioxidants. Br J Haematol. 2006;132:108–113.

18. Gonzalez Moraleja J, Rubio Perez P, Cabello Carro J, Garcia-Casasola

G, Alvarez-Sala JL. [Recurrent infections, severe neutropenia and neu-

trophil chemotaxis defect in paroxysmal nocturnal hemoglobinuria].

Ann Med Intern. 1994;11:490–492.

19. Gralnick HR, Vail M, McKeown LP, et al. Activated platelets in

paroxysmal nocturnal haemoglobinuria. Br J Haematol. 1995;91:

697–702.

20. Fujioka S, Yamada T. Varying populations of CD59-negative, partly

positive, and normally positive blood cells in different cell lineages

in peripheral blood of paroxysmal nocturnal hemoglobinuria patients.

Am J Hematol. 1994;45:122–127.

21. Young NS. Paroxysmal nocturnal hemoglobinuria: current issues in

pathophysiology and treatment. Curr Hematol Rep. 2005;4:103–109.

22. Filosa A, Valgimigli L, Pedulli GF, et al. Quantitative evaluation of

oxidative stress status on peripheral blood in beta-thalassaemic

patients by means of electron paramagnetic resonance spectroscopy.

Br J Haematol. 2005;131:135–140.

23. Walter PB, Fung EB, Killilea DW, et al. Oxidative stress and inflam-

mation in iron-overloaded patients with beta-thalassaemia or sickle

cell disease. Br J Haematol. 2006;135:254–263.

24. Bird RP, Draper HH. Comparative studies on different methods of

malonaldehyde determination. Methods Enzymol. 1984;105:299–305.

25. Droge W. Free radicals in the physiological control of cell function.

Physiol Rev. 2002;82:47–95.

26. Kohen R, Nyska A. Oxidation of biological systems: oxidative stress

phenomena, antioxidants, redox reactions, and methods for their

quantification. Toxicol Pathol. 2002;30:620–650.

27. Mengel CE, Ebbert L, Stickney D, Essig L, Brubaker L. Biochemistry

of PNH cells: nature of the membrane defect. Ser Haematol. 1972;5:

88–100.

28. Shingu M, Nonaka S, Nishimukai H, Nobunaga M, Kitamura H,

Tomo-Oka K. Activation of complement in normal serum by hydrogen

peroxide and hydrogen peroxide-related oxygen radicals produced by

activated neutrophils. Clin Exp Immunol. 1992;90:72–78.

29. Socie G, Mary JY, de Gramont A, et al. Paroxysmal nocturnal haemo-

globinuria: long-term follow-up and prognostic factors. French

Society of Haematology. Lancet. 1996;348:573–577.

30. Wiedmer T, Hall SE, Ortel TL, Kane WH, Rosse WF, Sims PJ. Com-

plement-induced vesiculation and exposure of membrane prothrombi-

nase sites in platelets of paroxysmal nocturnal hemoglobinuria. Blood.

1993;82:1192–1196.

31. Eldor A, Rachmilewitz EA. The hypercoagulable state in thalassemia.

Blood. 2002;99:36–43.

32. Eidt JF, Allison P, Noble S, et al. Thrombin is an important mediator

of platelet aggregation in stenosed canine coronary arteries with

endothelial injury. J Clin Invest. 1989;84:18–27.

33. Hanson SR, Harker LA. Interruption of acute platelet-dependent

thrombosis by the synthetic antithrombin D-phenylalanyl-L-

prolyl-L-arginyl chloromethyl ketone. Proc Natl Acad Sci U S A.

1988;85:3184–3188.

34. Iuliano L, Violi F, Pedersen JZ, Pratico D, Rotilio G, Balsano F. Free

radical-mediated platelet activation by hemoglobin released from red

blood cells. Arch Biochem Biophys. 1992;299:220–224.

35. Blockmans D, Deckmyn H, Vermylen J. Platelet activation. Blood

Rev. 1995;9:143–156.

36. Ninomiya H, Kawashima Y, Hasegawa Y, Nagasawa T. Complement-

induced procoagulant alteration of red blood cell membranes with

microvesicle formation in paroxysmal nocturnal haemoglobinuria

(PNH): implication for thrombogenesis in PNH. Br J Haematol.

1999;106:224–231.

37. Thomas TC, Rollins SA. Rother et al. Inhibition of complement activ-

ity by humanized anti-C5 antibody and single-chain Fv. Mol Immunol.

1996;33:1389–1401.

38. Hillmen P, Hall C, Marsh JC, et al. Effect of eculizumab on hemolysis

and transfusion requirements in patients with paroxysmal nocturnal

hemoglobinuria. N Engl J Med. 2004;350:552–559.