Proc. Nat. Acad. Sci. USAVol. 70, No. 8, pp. 2406-2410, August 1973

NAD(P) Glycohydrolase Deficiency in Human Erythrocytes and Alterationof Cytosol NADH-Methemoglobin Diaphorase by MembraneNAD-Glycohydrolase Activity

(polyacrylamide-gel electrophoresis/glutathione reductase/dichlorophenolindophenol)

HENRI FRISCHER*t, RICHARD NELSONt, CLAUDIA NOYESt, PAUL E. CARSON*t, JAMES E.BOWMANt§, KARL H. RIECKMANN*t, AND FRANCO AJMARI* Department of Medicine, Rush University 60612; t Departments of Medicine and § Pathology, University of Chicago 60637;t University of Illinois, Chicago, Ill. 60637; and I University of Genoa, Italy

Communicated by Leon 0. Jacobson, February 26, 1973

ABSTRACT Erythrocytic NADH methemoglobin dia-phorase acquires NADH-dichlorophenolindophenol dia-phorase activity when enzyme-associated NAD is removed.This transformation is reversible and can be mediated bymembrane NAD glycohydrolase (EC 3.2.2.5) in hemolysatesas well as in intact cells exposed to hydrogen peroxide. Itis abolished either in NADH methemoglobin diaphorasedeficiency or in NAD(P) glycohydrolase (EC 3.2.2.6) de-ficiency which is common in Afro-American but not inEuropean-American adults. Activities of erythrocyticNADP glycohydrolase and NAD glycohydrolase appear todepend on a single membrane enzyme.

This report characterizes a previously unsuspected modifica-tion of the NADH-methemoglobin diaphorase of humanerythrocytes by membrane NAD glycohydrolase (NAD-GHase). This interaction was found during investigation ofthe catalytic properties of erythrocytic NAD(P)H-oxidizedglutathione reductase (GSSG-R; EC 1.6.4.2.) (1,2), and ofmembrane-induced alterations of the electrophoretic isozymepattern of this enzyme (3), when a sharply localized band ofNADH-dichlorophenolindophenol diaphorase activity ap-peared in hemolysates that had been incubated with mem-branes before electrophoresis (4). The data indicate that amembrane enzyme whose physiological function is as yet un-known, and the deficiency of which is frequent in Afro-Americans, can interact both in hemolysates and in intacterythrocytes stressed by hydrogen peroxide with the mainenzyme responsible for methemoglobin reduction. Further-more, the usual methods for staining methemoglobin diaphor-ase actually demonstrate a form of this enzyme modified byNAD removal as well as other diaphorases.

METHODSBlood samples obtained in acid-citrate-dextrose (USP Solu-tion B) were centrifuged for 10 min at 3000 X g (40), and theplasma and buffy coat were aspirated; the cells, washed threetimes in 0.15 M Tris * HC1 buffer (pH 7.4), were hemolyzed byfreeze-thawing in a dry ice-acetone mixture. The hemolysatewas diluted with three volumes of buffer and an aliquot waskept at 40; the other portion was centrifuged for 1 hr at 45,000

Abbreviations: NAD-GHase, nicotinamide-adenine dinucleotide(NAD) nucleosidase (NAD glycohydrolase, EC 3.2.2.5); NADP-GHase, nicotinamide-adenine dinucleotide phosphate (NADP)nucleosidase (NADP glycohydrolase, EC 3.2.2.6); NDD, NADHdichlorophenolindophenol diaphorase; GSSG, oxidized gluta-thione; GSSG-R, oxidized glutathione reductase; G6PD, glucose6-phosphate dehydrogenase; 6PGD, phosphogluconate dehydro-genase.

X g (40) and the supernate was passed twice through 0.8-,umMillipore filters. In addition to hemolysates with unremovedmembranes, reconstituted membrane-containing hemolysateswere prepared by addition of one volume of purified mem-branes (5, 6) to one volume of previously destromatizedhemolysates. Hemolysates containing or lacking membraneswere incubated with or without additives for 1 hr at 370 or450, in a covered water bath. At the end of the incubation andbefore electrophoresis, hemolysates lacking membranes werekept at 40, while the membranes from the other samples wereremoved by centrifugation and Millipore filtration. In someexperiments, destromatized hemolysates were also dialyzedfor 90 min against a 0.15 M Tris buffer (pH 7.4; ZeinehDialyser, Chicago) or passed through a Sephadex G-25 (Phar-macia) column equilibrated and eluted with 1 M NaCl in167 mM Tris buffer (pH 7.4); dialysis conditions were estab-lished initially by monitoring the disappearance of hemo-lysate glutathione and lactate.

In studies with intact erythrocytes exposed to hydrogenperoxide (7), the cells, washed three times with 0.162 N NaCl(40), were suspended at hematocrits approximating 20 inisosmolar NaCl-phosphate buffer of pH 7.4 (147 mM NaCl-2.1 mM KH2PO4-8.9 mM Na2HPO4); the suspensions wereincubated for 3 hr at 370 with or without D-amino-acid oxi-dase (EC 1.4.3.3, Sigma or Worthington) and D-alanine inconcentrations generating 11.2 ,umol of hydrogen peroxide in2.5 hr at an average rate of 27 nmol min-' ml-'. The cells werethen washed five times in buffer and hemolyzed by freeze-thawing; before electrophoresis, the samples were destroma-tized by centrifugation and Millipore filtration.

Vertical acrylamide electrophoresis was performed in dis-continuous Tris-borate buffer systems at 0-40 in 3-mm thickgel slabs (7% cyanogum-41 Fisher or Biorad polymerized byN,N,N',N'-tetramethylethylenediamine Biorad, with am-monium persulfate as catalyst). Two nonpreparative elec-trophoresis systems were used; in system A, the samples (40,ul/1-cm slot) were subjected to electrophoresis for 17 hr inthe cold room (40) in Smithie's vertical starch gel apparatus(Buchler Co.), at a gradient of 10 V/cm of gel (280-300 V);gel buffer: 45 mM Tris HCl-38.4 mM borate (pH 8.3); anode(lower chamber) buffer: 180 mM Tris HCl-154 mM borate(pH 8.3); cathode buffer: 129 mM Tris HCl, 110 mM borate(pH 8.3). Subsequently and more conveniently (system B),electrophoresis was also performed in a vertical acrylamide-gel apparatus (EC Corp., Philadelphia), fitted with a constantpulsed power supply (Ortec); gel buffer: 380 mM Tris HCl(pH 8.9); electrode buffers: 5 mM Tris- HCI-4.4 mM borate

2406

Membrane NAD(P)ase and Methemoglobin Diaphorase 2407

(pH 8.5). The gels were loaded with 25,ul/1-cm slot (4 partsof hemolysate to 1 part of 40% sucrose) cooled by circulatingice water; electrophoresis was performed for 2.5 hr at 400 Vwith 300 pulses per sec and capacitance of 1 1uF.After electrophoresis, gels were washed in distilled water

and incubated for 15 min in the dark at 370 in a filtered,freshly prepared solution of 1 mM 3(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide and 40 uM dichloro-phenolindophenol in 0.15 M Tris HCI buffer (pH 8.2). Thegels were then stained for 1.75 hr in the dark at 370 after theaddition of either: (a) 1.2 mM NADH (for NADH-dichloro-phenolindophenol diaphorase); (b) 1.2 mM NADPH (forNADPH-dichlorophenolindophenol diaphorase); (c) 1.2 mMNADH and 2.3 mM GSSG (for NADH dichlorophenolindo-phenol diaphorase and NADH-GSSG-R); (d) 1.2mM NADPHand 2.3 mM GSSG (for NADPH-GSSG-R).Methemoglobin diaphorase was assayed with ferric hemo-

globin as substrate according to Hegesh (8). Assays for glu-cose-6-phosphate dehydrogenase (G6PD, EC 1.1.1.49) andphosphogluconate dehydrogenase (6PGD, EC 1.1.1.44) wereaccording to Glock and McLean (9), for GSSG-R accordingto Long and Carson (10) (modified to give a pH of 6.9).The activities of NADP-glycohydrolase (GHase) and of

NAD-GHase were assayed in intact erythrocytes as follows;the cells, washed three times in 0.162 N NaCl (40), were sus-pended to a monitored hematocrit of about 20, in a totalvolume of 4 ml, in isosmolar NaCl-phosphate buffer of pH 7.4containing either NADP (for NADP-GHase) or NAD (forNAD-GHase) at a final concentration of 0.7 mM. In ini-tial studies, comparing the activities of intact erythrocyteswith that of their isolated membranes (5), the suspensionswere adjusted to contain 106 particles per mm' as monitoredby a Coulter Counter B. The NAD(P) content of the cell-freesupernate was assayed before and after 2 hr of incubation at370 in neutralized protein-free filtrates [to avoid possible in-terference by liberated erythrocytic NAD(P)dehydrogenasesand hemoglobin]. Protein was precipitated with one volumeof cold 0.6M perchloric acid per volume of original supernate.Samples were neutralized with one-third volume of 2.2 MK2HPO4 per volume of protein-free supernate. NADP in 1.0ml of filtrate was determined in a total volume of 3.02 ml at340 nm (eM :6200) with excess G6PD (10 units) and G6P (11.8mM) in a Tris* HC1 (67 mM)-MgCl2 (10 mM) buffer of pH 8.0;NAD was determined with excess alcohol dehydrogenase (EC1.1.1.1, 10 units) and ethanol (118 mM) in 10 mM sodiumpyrophosphate-10 mM semicarbazide buffer (pH 8.8). Re-sults are expressed in units of /Amol of NAD (P) consumed hr'liter-I of erythrocytes (370, pH 7.4). No measurable NAD(P)was liberated in cell suspensions incubated without addedcoenzyme and no NAD(P) was consumed in the absence oferythrocytes or membranes; NAD(P)-GHase activity in-creased linearly with erythrocyte concentrations up to hemato-crits of 40 and after 1, 2, or 3 hr of incubation (370). Compari-sons of erythrocytic NADP-GHase activities of freshly drawnblood samples with those of blood samples stored in acid-citrate-dextrose solution at 40, did not reveal significant dif-ferences when tested repeatedly over a period of 3 weeks insamples from five individuals. Precision, as exemplified by 15replicate NADP-GHase assays during several months,in blood samples from one subject was 1450 d= 129 SD units.In all persons tested, NAD(P)-GHase activity of intacterythrocytes approximated closely those measured in their iso-lated membranes (P> 0.5; paired t-test; n = 10); in a given

person, NAD-GHase and NADP-GHase activities were highlycorrelated with a mean NADP-GHase/NAD-GHase ratioof 0.60/1; no isolated NAD-GHase or NADP-GHase de-ficiency was detected and all NADP-GHase-deficient (seebelow and Figs. 4 and 7) individuals also had NAD-GHasedeficiency in the range previously observed in cord and inadult blood samples (11, 12, 6; unpublished observations).NADP-GHase activity in erythrocytes from 33 apparentlyhealthy European-American men was 1186 i 175 SD units;in this group, the activities were distributed with a singlemode centered between 1200 and 1300 units, and no personwith an activity of less than 780 units was found (range 780-1581). In contrast, erythrocytic NADP-GHase activity wassignificantly lower in a group of 52 apparently healthy Afro-American men (843 i 295 SD, range: 92-1549 units); in thisgroup, at least two activity modes located at 850 and 350units were discernible and there was no significant differencebetween the NAD (P)-GHase activities of the 11 G6PD;de-ficient (GdA-) and of the 41 nondeficient individuals; 15 of 52Afro-Americans had activities less than the lowest found inEuropean Americans, and in six of these 15 (5 GdA+ and 1GdA-), designated as NAD(P)-GHase deficient, the NADP-GHase activities were less than 450 units (280 4± 131 SD;range 92-433). The difference between the two groups re-mained significant (P < 0.05) even when the six NAD(P)-GHase-deficient Afro-American subjects were excluded (916± 221 SD; range 536-1546; n = 46).

RESULTS

Fig. 1 shows that undialyzed hemolysates, which had beenincubated before electrophoresis with their membranes (I-2),contain a sharply localized band of NADH-dichloropheno-lindophenol diaphorase (NDD) activity undetectable in con-trol hemolysates incubated without their membranes (1-1).This band, with Rf of 0.75 relative to hemoglobin A, appearedrapidly, reached maximal intensity within 2 hr of staining, wasdemonstrable by stains containing NADH (Figs. 1-I; 2-Iand II), but not by stains containing NADPH (Figs. 1-II,2-111 and IV), and was not affected by the presence or ab-sence of GSSG (Fig. 2-I and II). In contrast, the slower elec-trophoretic bands with mean Rf of 0.33 with respect to HbA (Figs. 1-I and II, 2, 5, 6-I, and 7), stained with NADPHas well as with NADH (Fig. 2) and were markedly intensi-fied in the presence of GSSG (Fig. 1-I and II). Detection of

I rr

, @US* * GSSG-R

NDD

t 0 % _4_ _HbA

2 2 2 2

FIG 1. Formation of NADH-dichlorophenolindophenol dia-phorase activated by membranes. Nondialyzed hemolysates wereincubated before acrylamide gel electrophoresis (system B) for1 hr at 45° without (8sots 1) or with (slots 2) their membranes.A portion of the gel (I) was stained for NADH-dichlorophenolin-dophenol diaphorase activity with dichlorophenolindophenoland 3(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromideand without GSSG; another portion (II) was stained withNADPH and GSSG for NADPH-GSSG-R. HbA, hemoglobin A.

Proc. Nat. Acad. Sci. USA 70 (1973)

2408 Medical Sciences: Frischer et al.

I JI mii =5 S ~~GSSG-R

NDD_ HbA

2 1 2 2 1 2

FIG 2. NDD induction in destromatized hemolysates byremoval of NAD, suppression by addition of NAD, NADHspecificity and differentiation from NAD(P)H GSSG-R. De-stromnatized hemolysate was passed through a Sephadex 3-25column equilibrated and eluted with 1 N NaCl in 157 mM Trisbuffer (pH 7.4); slots I were loaded with aliquots of this hemoly-sate to which no NAD was added and slots 2 with aliquots towhich 0.1 mM NAD was added after passage through the column.Acrylamide-grl electrophoresis, system B, was performed withborate in the electrode buffer raised to 5.7 mM. Gel portions Iand II were stained for NADH diaphorase without (I) or with(II) GSSG; gel portions III and IV were stained with NADPHinstead ofNaDH either without (III) or with (IV) GSSG.

these slower bands in dialyzed hemolysates required GSSG inthe stain (Figs. 2, 3, and 6-II). Further work with fractionseluted from gels, inhibitors, and activators of GSSG-R andwith GSSG-R-deficient hemolysates, confirmed that theslower bands were isozymes of NAD(P)H-GSSG-R appearingas NAD(P)H diaphorases in undialyzed hemolysates. Incontrast to these GSSG-R bands, the single faster (Rf 0.75)stromal-induced NDD was present in dialyzed and undialyzedhemolysates incubated with complete membranes as well as in

IM lw; Wit. NDDe~~~~~~~nc

w>_ ^ _ ~~~~~HbA

2 3 4 5 6 7 8 9 10

FIG. 3. Competitive inhibition of membrane-induced NDDby 0.1 M nicotinamide, failure of nicotinamide to reverse NDD,and formation of NDD in destromatized hemolysates by expo-sure to Neurospora NAD-CIHase and to Norit. Destromatizedhemolysate, dialysis against 0.15 M Tris buffer (40, pH 7.4).Slots I and 2 were loaded with hemolysates treated with Noritand NeurTspora NAD-GHase, respectively (see text). All otherhemolysates were incubated for 1 hr at 450; slot 3, with purifiedmembranes alone; slot 4, without membranes; slot 6, withoutmembranes but with nicotinamide; slots 6 and 7, initial incuba-tion without membranes or nicotinamide then reintubationtion without membranes nor nicotinamide then reincubation(1 hr at 370) with or without nicotinamide, respectively; slots 8and 9, initial incubation with membranes but without nicotin-amide, followed by ghost removal and reincubation (1 hr at 37°)with or without nicotinamide, respectively; slot 10, incubationwith membranes and nicotinamide. Acrylamide-gel electrophore-sis (system A); NADH diaphorase stain. Destromatized hemoly-sates treated with Norit (1) or Neurospora NAD-GHase (2)develop NDD. All other hemolysates incubated without mem-branes lack NDD (4-7); rpembrane-induced NDD (3, 9) is ab-sent if nicotinamide is present during membrane exposure (10);unlike NAD, nicotinamide does not reverse previously formedNDD (8).

hemolysates dialyzed after incubation with these membranes.Formation of NDD required incubation of hemolysates withtheir unremoved membranes (Figs. 1-I and 6-I) or with puri-fied autologous or homologous membranes (Figs. 3, 4, and6-Ij) in some blood samples, NDD could be induced by amembrane-hemolysate contact as short as 15 min at roomtemperature. Heating the membranes to 80° for 15 minabolished their effect; Tris buffer eluates or Triton X extractsof membranes did not contain NDD, nor could such mem-brane eluates or extracts induce NDD. Incubation of hemoly-sates containing membranes with either nicotinamide (Fig. 3)or NAD, but not with NADP or flavin adenine dinucleotidephosphate, prevented NDD formation. Membrane-inducedNDD could be abolished by reincubation of destromatizedhemolysates with NAD; in contrast, reincubation withoutadditives or with NADP, nicotinamide (Fig. 3)j or flavinadenine dinucleotide phosphate did not remove NDD. NDDwas also induced without membranes, by the following treat-ments designed to remove bound NAD from destromatizedhemolysates: (a) incubation with Neurospora crassa NADase(Sigma D9880, 0.25 units ml-' hemolysate; 1 hr at 370; (Fig.3-2)]; (b) exposure to acid-washed, activated charcoal (Norit:20 mg ml-' hemolysate; 10 min at 250, Fig. 3-1); (c) dialysisagainst buffers of high ionic strength or passage through aSephadex G-25 (Pharmacia) column equilibrated with andeluted by a high-ionic-strength buffer [167 mM Tris HCl,(pH 7.3) with 1 M NaCl; Fig. 2-I and II, slots 11; (d) electro-phoresis in continuous Tris -borate buffer systems (Fig. 5-1).NDD induction by membranes or by high-ionic-strengthdialysis can be reversed by addition of NAD before electro-phoresis (Fig. 2-1 and II, slots 1), and the appearance of NDDafter electrophoresis of destromatized hemolysates in Tris-borate or other continuous buffer systems was also preventedby the addition of NAD (but not of NADP) to the gel andelectrode buffers. Fig. 4 shows that membranes from erythro-cytes deficient in NAD(P)-GHase fail to induce NDD in auto-logous or homologous normal hemolysates, whereas normalmembranes can induce NDD in hemolysates from erythro-

e

.., GSSG-R

_P NDD_ N~~bA_ _

2 3 4 5 6 7 8 9 10 11 12

FIG. 4. Membranes from erythrocytes deficient in NAD(P)-GHase fail to induce NDD in autologous or homologous hemoly-sates; normal membranes can induce NDD in hemolysates fromNAD(P)-GHase-deficient erythrocytes. Nondialyzed, destroma-tized hemolysates and hemoglobin-free membranes were pre-pared from blood samples of two subjects with normal erythro-cytic NADP-GHase (J:872 units; H:704 units), as well as fromerythrocytes deficient in NADI-GHase (A:92 units; see Methods).Four aliquots of hemolysates from each individual were incu-bated (1 hr at 370) without membranes, with their own mem-

branes, as well as with homologous membranes from each of thetwo other persons. Slots 1-4, 5-8, and 9-12 contain the hemoly-sates of J, H, and A, respectively. Acrylamide-gel electrophoresis(system B) was stained for NADH diaphorase without GSSG.No NDD was observed in hemolysates incubated without mem-branes (1, 5, 9) or incubated with membranes deficient in NAD-(P)GHase (3, 7, 1i); all other systems developed NDD.

Proc. Nat. Acadi. Sci. USA 70 (1973)

Membrane NAD(P)ase and Methemoglobin Diaphorase 2409

cytes deficient in NAD(P)-GHase. Membranes with normalNAD(P)-GHase activity from erythrocytes deficient inGSSG-R or G6PD, induced NDD in their hemolysates likethose of normal erythrocytes. Fig. 6-1 demonstrates that ahemolysate with homozygous NADH methemoglobin dia-phorase deficiency contains the slow GSSG-R bands and failsto develop NDD activity when incubated with their mem-branes; these membranes can induce NDD in normal hemoly-sates (Fig. 6-II). Fig. 7 indicates that NDD was induced inintact erythrocytes if the cells were exposed to H202, whereasno NDD band was detectable in control erythrocytes in-cubated in parallel without H202; the induction by H202 ofNDD in intact erythrocytes cannot be demonstrated if thecells are NAD (P)-GHase deficient.

DISCUSSIONHemolysates incubated with erythrocytic memebranes acquirea new electrophoretic band with NADH-specific dichloro-

w

I

b2

r'IC)rlLO

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30Eluted fractions

FIG. 5. NDD corresponds to NADH-methemoglobin dia-phorase. Hemolysate proteins were first separated by electro-phoresis: 6-mm preparative starch gel (Electrostarch 71 Con-naught: 55 g/550 ml of buffer); 0.2 ml of destromatized hemoly-sate per 1-cm slot; 10-V gradient per 1 cm of gel for 17 hr at 40;continuous Tris-borate buffer system [gel and electrode buffer:180 mM Tris *HCl-154 mM borate (pH 8.5)]. A strip of this gel,stained for NADH diaphorase, was used to locate the position ofNDD (I, first peak of curve 1) and of hemoglobin A (I, secondpeak of curve 1). The rest of the gel was rotated 900 with respectto the direction of the first electrophoresis. Proteins were elutedby electrodialysis into 31 fractions [elution convection cell, EC,Philadelphia; electrodialysis: 4 hr at 4°; 20 V, 340 mA in 72 mMTris -HCl-61 mM H3B03 (pH 8.5)]. The cleared gel was stainedto document the disappearance of NDD and of hemoglobin (I,curve 2), and all eluted fractions were assayed for NADH-methemoglobin diaphorase (II, curve 1) and for their hemoglobincontent, used as an additional position marker (II, curve 2). Theonly hemolysate fractions capable of reducing ferric hemoglobinwith NADH, are located in the NDD area. I, gel scans:

1,-, before elution; 2, after elution. II, fraction assays: 1,NADH methemoglobin diaphorase; 2, hemoglobin.

:II

HbA*o to

2 3 4 5 6

IO

NDD

IHbA

G)

7 8 9

FIG. 6. Hemolysate with homozygous NADH methemoglobindiaphorase deficiency fails to develop NDD when incubatedwith its membranes; these membranes can induce NDD in nor-mal hemolysates. In I, nondialyzed hemolysates with normalNADH methemoglobin diaphorase activity (F.A.: 1.4 units;slots 1 and 2), with homozygous NADH methemoglobin dia-phorase deficiency (A.C.: 0.1 units; slots 3 and 4), and with de-creased NADH methemoglobin diaphorase (G.C., mother ofA.C.: 0.9 units; slots 6 and 6), were incubated (1 hr at 450) with-out (slots 1, 3, and 5) or with their membranes (slots 2, 4, and 6).Acrylamide-gel electrophoresis (system B with borate raised to5.7 mM), stain for NADH-diaphorase in the presence of GSSG.NDD, present in the control normal hemolysate incubated withmembranes (2), is faint in the membrane-treated, heterozygouslydeficient hemolysate (6), and absent in the membrane-treated,homozygous deficient blood (4) as in all hemolysates incubatedwithout membranes (1, 3, 5). In II, erythrocytic membranesfrom A.C. were incubated with destromatized dialyzed hemoly-sate of normal NADH methemoglobin diaphorase activity (1.8units, slot 7); this hemolysate was also incubated with (slot 8)and without its own membranes (slot 9). Acrylamide-gel elec-trophoresis (system B); NADH diaphorase stain without GSSG.Normal hemolysate exposed to membranes from the proposituswith NADH methemoglobin diaphorase, develops NDD (slot 7).

phenolindophenol diaphorase activity (Figs 1 and 2). Thisactivity is not eluted from the membranes and requires theinteraction of a nondialyzable hemolysate component withoutNDD specificity, with a heat-labile constituent of autologousor homologous membranes.

Nicotinamide prevents NDD formation if incubated withmembranes and hemolysate; it does not abolish NDD whenreincubated with membrane-free hemolysate in which this

I II

_ a - 6 _] &~~~~~SSG-RNODHbA

2 3 4 S 6 7 8 12 3 4 5 6 7 8FIG. 7. Induction of NDD in intact erythrocytes not deficient

in NAD(P)-GHase but not in erythrocytes deficient in NAD(P)-GHase exposed to hydrogen peroxide. Washed erythrocytes withnormal membranes (I, NADP-GHase 1132 units) and withNAD(P)-GHase deficiency (II, NADP-GHase 21 units), wereexposed to H202. Nondialyzed hemolysates were prepared fromcells incubated without -aminoacid oxidase or D-alanine (slots 1),with n-alanine but without oxidase (slots 2), and with oxidasebut without D-alanine (slots 3). Slots 4, 5, and 6, respectively,were loaded with hemolysates prepared from cells incubated withD-aminoacid oxidase and 1-, 1.5-, or 2-times the standard amountof n-alanine. Control destromatized (slots 7) and nondestro-matized (slots 8) hemolysates from each person were incubatedwithout H202 (3 hr at 37°) before electrophoresis. Acrylamide-gelelectrophoresis (system B); NADK diaphorase stain.

Proc. Nat. Acad. Sci. USA 70 (1973)

2410 Medical Sciences: Frischer et al.

activity had been induced by prior exposure to membranes(Fig. 3). Since nicotinamide inhibits human erythrocyticmembrane NAD(P)-GHase (6, 13, 14), these observationssuggested that membranes acted through their NAD-GHaseand/or their NADP-GHase activity. NAD, like nicotinamide,prevents NDD formation if incubated with membranes andhemolysates; unlike nicotinamide, however, reincubation ofmembrane-free hemolysates with NAD, can also reversepreviously induced NDD; NADP neither inhibits nor reversesmembrane-induced NDD. Thus, membranes seem active byvirtue of their NAD-GHase activity. This suggestion wassupported further by the activation of NDD in membrane-free hemolysates by procedures designed to remove NAD andby the reversal of their effect with NAD but not with NADP(Figs. 3, 5, and Results). The requirement for membraneNAD-GHase was then established by the search for anddetection of persons deficient in erythrocytic membraneNAD(P)-GHase activity (see Methods), and by finding thattheir membranes could not induce NDD in hemolysates inwhich this activity could be formed by normal membranes(Fig. 4). Furthermore, NDD can also be induced in intactnormal but not in NAD (P)-GHase-deficient erythrocytesexposed to low amounts of H202 generation (Fig. 7).

Since NADH methemoglobin diaphorase is often assayedas a dichlorophenolindophenol diaphorase (15), it seemedpossible that this enzyme was the nondialyzable hemolysatecomponent altered by NAD-GHase. This possibility wasstrengthened by finding that the only protein with NADHmethemoglobin diaphorase activity (with ferric hemoglobinas substrate) elutable from preparative gels was located atthe position of NDD (Fig. 5). This conclusion was thenestablished when hemolysates with homozygous deficiency ofNADH methemoglobin diaphorase, exposed to their mem-branes, failed to develop NDD, although these membranescould induce NDD in control hemolysates (Fig. 6).These studies thus demonstrate a reversible modification of

NADH ferrihemoglobin diaphorase by NAD removal; thismodification can be mediated in vitro by the NAD-GHaseactivity of human erythrocytic membranes, in intact cellsstressed by exposure to hydrogen peroxide, as well as inhemolysates. The origin and mechanism of some previouslyreported NADH methemoglobin diaphorase "aging variants"(16-19) may now also be explained. Electrophoresis of NADHmethemoglobin diaphorase reveals an NAD-stripped formof the native enzyme and requires control of exposure to mem-branes; furthermore interpretation of such studies for geneticanalysis must also take into account that presently usedmethods for phenotyping NAD(P)H methemoglobin dia-phorase can include NAD(P)H GSSG-R isozymes even whenGSSG is omitted from the stain; GSSG-R appears then inundialyzed hemolysates as an NAD(P)H diaphorase, and itsprincipal isozyme should not be misinterpreted as a slowerNADH methemoglobin diaphorase band (and/or as a slowNADPH methemoglobin reductase band).

Studies of NAD-GHase and NADP-GHase activities inmembranes and in intact cells necessary to establish the natureof the membrane component interacting with methemoglobindiaphorase have confirmed the common occurrence of eryth-rocytic NAD-GHase deficiency in Afro-American adults

(12) and revealed that they are also deficient in erythrocyticNADP-GHase. Furthermore, erythrocytic NAD-GHase andNADP-GHase activities are highly correlated in a given in-dividual, and isolated erythrocytic NAD-GHase or NADP-GHase deficiency has not been detected. These findingsstrongly suggest, in contrast to previous reports (20, 21),that a single membrane enzyme determines the activities ofboth NADP-GHase and NAD-GHase. The physiological roleof membrane NAD(P)-GHase and the reason for the highfrequency of NAD(P)-GHase deficiency in Afro-Americansremain unknown. Further studies are thus necessary to de-termine whether erythrocytic NAD(P)-GHase deficiency canbe associated, in vivo and under appropriate stress, withmetabolic disturbances mediated by alterations of enzymesaffected in vitro byNAD (P)-GHase activity (14, 6, 3).

We thank Deatra Kinney, Edward C. Patterson, Sharon L.Ziegler, and Tanveer Ahmad for excellent technical assistance.This work was supported by U.S. Public Health Service Grant no.7-RO 1 HE 14859, by U.S. Army Contract DADA-17-71-C-1103, by U.S. Navy Contract NOOO 14-72-A-0288, and by agrant under the Mutual USA-Italy Educational and CulturalExchange Act.

1. Frischer, H., Carson, P. E., Bowman, J. E. & Rieckmann,K. H. (1973) J. Lab. Clin. Med. 81, 603-612.

2. Frischer, H., Bowman, J. E., Carson, P. E., Rieckmann,K. H., Willerson, W. D. & Colwell, E. (1973) J. Lab. Clin.Med. 81,613-624.

3. Frischer, H., Noyes, C. & Nelson, R. (1971) J. Clin. Invest.50, 34 abstr.

4. Frischer, H., Noyes, C., Nelson, R. & Kinney, D. (1971)Clin. Res. 19, 418.

5. Dodge, J. T., Mitchell, C. & Hanahan, D. J. (1963) Arch.Biochem. Biophys. 100, 119-130.

6. Ajmar, R., Scharrer, B., Hashimoto, F. & Carson, P. E.(1968) Proc. Nat. Acad. Sci. USA 59, 538-545.

7. Frischer, H., McNamara, J., Rieckmann, K. H., Stockert,T., Powell, R. & Carson, P. E. (1968) Clin. Res. 16, 303.

8. Hegesh, E., Calmanovia, N. & Avron, M. (1968) J. Lab.Clin. Med. 72, 339-344.

9. Glock, G. E. & McLean, P. (1953) Biochem. J. 55, 400-408.10. Long, W. K. & Carson, P. E. (1961) Biochem. Biophys. Res.

Commun. 5, 394-399.11. Bergren, W. R., Ng, W. G. & Donnell, G. N. (1967) Pedi-

atrics 40, 136-137.12. Ng, W., Donnell, G. N. & Bergren, R. (1968) Nature 217,

64-65.13. Webb, L. J. (1966) in Enzyme and Metabolic Inhibitors II

(Academic Press New York), pp. 485-500.14. Carson, P. E., Okita, G. T., Frischer, H., Hirasa, J., Long,

W. K. & Brewer, G. J. (1963) in Proceedings of the NinthCongress of European Societies of Hematology, Lisbon (S.Karger, Basel/New York), pp. 655-665.

15. Scott, E. M. & Griffith, I. V. (1959) Biochim. Biophys. Acta34,584-586.

16. Brewer, G. J., Eaton, J. W., Knutsen, C. S. & Beck, C. C.(1967) Biochem. Biophys. Res. Commun. 29, 198-204.

17. Bloom, G. E. & Zarkowsky, H. S. (1969) N. Engl. J. Med.281, 919-922.

18. Detter, J. C., Anderson, J. E. & Giblett, E. R. (1970) Amer.J. Human Genet. 22, 100-104.

19. Kaplan, J. C. (1971) in Red Cell Structure and Metabolism,ed. Ramot, B. (Academic Press, New York), pp. 125-133.

20. Gasiorowska, I. & Raczynska-Bojanowska, K. (1963) Bull.Acad. Pol. Sci. 11, 417-421.

21. Hofmann, E. & Noll, F. (1961) Acta Bid1. Med. Ger. 6,1-6.

Proc. Nat. Acad. Sci. USA 70 (1973)