The Role of Superoxide Anion Generation

in Phagocytic Bactericidal Activity

STUDIES WITH NORMALANDCHRONIC

GRANULOMATOUSDISEASE LEUKOCYTES

RICHARDB. JOHNSTON,JR., BERNARDB. KEELE, JR., HARAP. MISRA,JOYCEE. LEHMEYER,LAWRENCES. WEBB, ROBERTL. BAEHNER,andK. V. RAJAGOPALAN

From the Departments of Pediatrics and Microbiology, Institute of DentalResearch, and Comprehensive Cancer Center, the University of Alabama inBirmingham, Birmingham, Alabama 35294; the Department of Pediatrics,Division of Pediatric Hematology-Oncology, Indiana University School ofMedicine and Riley Hospital for Children, Indianapolis, Indiana 46202; andthe Department of Biochemistry, Duke University Medical Center,Durham, North Carolina 27710

A B S T R A C T The capacity of human phagocytes togenerate superoxide anion (02-), a free radical of oxy-gen, and a possible role for this radical or its deriva-tives in the killing of phagocytized bacteria were ex-plored using leukocytes from normal individuals andpatients with chronic granulomatous disease (CGD).Superoxide dismutase, which removes 02-, consistentlyinhibited phagocytosis-associated nitroblue tetrazolium(NBT) reduction indicating the involvement of O2 inthis process. Similarly, superoxide dismutase inhibitedthe luminescence that occurs with phagocytosis, impli-cating 0- in this phenomenon, perhaps through itsspontaneous dismutation into singlet oxygen. Subcel-lular fractions from homogenates of both normal andCGD leukocytes generated 02- effectively in the pres-ence of NADHas substrate. However, 02- generationby intact cells during phagocytosis was markedly di-minished in nine patients with CGD. Leukocytes frommothers determined to be carriers of X-linked recessiveCGD by intermediate phagocytic reduction of NBTelaborated 02- to an intermediate extent, further demon-

This work was presented in part at the 65th AnnualMeeting of The American Society for Clinical Investiga-tion, Atlantic City, N. J., 29 April 1973 (J. Clin. Invest.52: 44a, 1973).

Received for publication 24 July 1974 and in revised form18 Janutary 1975.

strating the interrelationship between NBT reductionand 02 generation in phagocytizing cells.

Activity of superoxide dismutase, the enzyme re-sponsible for protecting the cell from the damagingeffects of 02-, was approximately equal in homogenatesof normal and CGDgranulocytes. Polyacrylamide elec-trophoresis separated this activity into a minor bandthat appeared to be the manganese-containing super-oxide dismutase associated with mitochondria and amore concentrated, cyanide-sensitive, cytosol form ofthe enzyme with electrophoretic mobility that corre-sponded to that of erythrocyte cuprozinc superoxidedismutase.

Superoxide dismutase inhibited the phagocytic killingof Escherichia coli, Staphylococcus aureus, and Strep-tococcus viridans. A similar inhibitory effect was notedwith catalase which removes hydrogen peroxide. Neitherenzyme inhibited the ingestion of bacteria. Peroxide and02- are believed to interact to generate the potent oxi-dant, hydroxyl radical (-OH). A requirement for -OHin the phagocytic bactericidal event might explain theapparent requirement for both 02- and H202 for suchactivity. In agreement with this possibility, benzoateand mannitol, scavengers of - OH, inhibited phago-cytic bactericidal activity. Generation of singlet oxy-gen from 02- and - OHalso might explain these findings.

The Journal of Clinical Investigation Volume 55 June 1975 1357-1372 1 X-)7

It would seem clear from these and other studies thatthe granulocyte elaborates 02 as a concomitant of therespiratory burst that occurs with phagocytosis. Towhat extent the energy inherent in 02- is translatedinto microbial death through 02- itself, hydrogen per-oxide, OH, singlet oxygen, or some other agent re-mains to be clearly defined.

INTRODUCTION

Recognition that superoxide anion (0O-),' a free-radicalform of oxygen, can be generated in biological systemshas emerged primarily from study of the aerobic oxida-tion of xanthine by the enzyme xanthine oxidase (1),according to the reaction:

xanthine oxidasexanthine + H20 + 02 * uric

acid + 0r + 2H+.

The radical can be detected by its capacity to reducecompounds such as ferricytochrome c (2) or nitrobluetetrazolium (NBT) (3, 4) or to oxidize epinephrine toadrenochrome (5). Auto-oxidative processes and en-zymes capable of generating this highly reactive anionhave been identified in various mammalian tissues, bac-teria, and fungi (reviewed in reference one). However,it has not been clear whether or not Or serves a use-ful biological purpose. On the contrary, its presenceshould pose a potential threat to those cells in whichit is elaborated.

Greater understanding of the physiological control ofOr2 developed after the demonstration by McCord andFridovich (6) of enzymic activity which catalyzes thedismutation of Or in the reaction:

02- +O±+ 2H+-> 02 + H202.

The enzyme, superoxide dismutase (SOD), is nowknown to be identical to "cuprein," the copper proteinwithout known function previously isolated from mam-malian erythrocytes, liver, brain, and other tissues (6).The presence of SOD in aerobic bacteria and its ab-sence from strict anaerobes (7), and the capacity ofincreased intracellular concentrations of SOD to pro-tect Escherichia coli (8) and Saccharoniyces cerevisiae(9) against oxygen toxicity have led to the assumptionthat this enzyme serves to protect oxygen-metabolizingcells from the potentially detrimental effects of 02-.

1 Abbrcviationis used in this paper: BSA, bovine serumalbumin; CGD, chronic granulomatous disease; KRP-Dbuffer, Krebs-Ringer phosphate buffer with dextrose butno albumin; KRP-DA, Krebs-Ringer phosphate buffer withdextrose and albumin; NBT, nitroblue tetrazolium; '02,singlet oxygen; 02, superoxide anion; -OH, hydroxylradical; PBS, phosphate-buffered solution; SOD, super-oxide dismutase.

Normal human neutrophils have the capacity to re-duce NBT to formazan during phagocytosis; neutro-phils from patients with chronic granulomatous disease(CGD), which can ingest bacteria normally but can-not kill them, lack this capacity (10). The ability ofOr2 to reduce NBT, as well as its potent reactivity andthe apparent ubiquity of its production in biologicalsystems, suggested to use the possibility that this freeradical might be generated by leukocytes and involvedin the killing of bacteria by these cells. Indeed, Babior,Kipnes, and Curnutte have demonstrated recently thatnormal leukocytes generate Or2 during phagocytosis(11) and that CGDphagocytes do not (12). We de-scribe here evidence to substantiate 02- production bythe phagocyte during ingestion and involvement of thisfree radical in phagocytosis-associated NBT reductionand chemiluminescence and in the killing of ingestedbacteria. A partial deficiency in O2 generation by leu-kocytes from carriers of X-linked CGDis also reported.

METHODSEnzxymlies, -OH scavengers, anid their controls. SODwas

purified from bovine erythrocytes by the method of McCordand Fridovich (6). A portion of each preparation was auto-claved for 10 min at 124°C for use as the "heated SOD"control. Apoenzyme was prepared by dialysis for 40 hagainst acetate-EDTA buffer (6). The autoclaved enzymeno longer inhibited the reduction of cytochrome c in theaerobic xanthine-xanthine oxidase system (2) ; the apoen-zyme lost approximately 90% of its original activity in thisassay. Bovine liver catalase and bovine serum albumin(BSA) were purchased from Sigma Chemical Co., St.Louis, Mo. SOD contaminating the catalase was removedby filtration over Sephadex G-75 (Pharmacia Fine Chemi-cals Inc., Piscataway, N. J.) using 0.05 M potassium phos-phate buffer with 0.1 M KCl, pH 7.8. The proteins weredissolved in saline (0.15 M NaCl) to a concentration of 1mg/ml and sterilized by filtration (Millipore Corp., Bed-ford, Mass., 0.22 ,um diameter pore size). Sodium benzoate,D-mannitol, and D-glucose (Fisher Chemical Co., Fair Lawn,N. J.) were made 0.2 M in water.

Souirce of leiikocytes. Venous blood was obtained fromhealthy adult volunteers, patients with CGD, and mothersand maternal grandmothers of CGDpatients. The diagnosisof CGDwas established in each case by demonstration thatthe patient's phagocytes could ingest but not effectively killStaphylococcus aureus, and in most cases E. coli also, andcould not reduce NBT normally. The X-linked carrier stateof CGD was determined in mothers and maternal grand-mothers of three of the male patients by the finding of anintermediate value in the quantitative NBT assay (10) anda family history of CGDor possible CGDonly in boys onthe maternal side. The non-X-linked state was determinedin five mothers of six patients by a normal result in theNBT assay on at least two occasions. The first diagnosisof CGD or suspected CGDin these families was made inthe patients studied; five of the six were boys.

Preparation of leukocytes for functional assays. For de-termination of phagocytosis-associated NBT and cytochromec reduction, chemiluminescence, bactericidal activity, andbacterial uptake, leukocytes were separated from heparinizedblood by sedimentation of erythrocytes at room tempera-

1358 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

ture for 20 min in an equal vol of 3% dextran (mol wvt250,000 or 500,000, Pharmacia Fine Chemicals, Inc., Pisca-taway, N. J.) in saline. The leukocyte-rich plasma wascentrifuged at 250 g for 10 min, and the cells were resus-pended in 0.2% NaCi for 20 s to lyse erythrocytes; iso-tonicity was restored with an equal vol of 1.6% NaCl. Leu-kocytes were washed twice more and suspended in Hank'sbalanced salt solution (Grand Island Biological Co., GrandIsland, N. Y.) containing 0.2% glucose, or in Krebs-Ringerphosphate buffer, pH 7.35, containing 0.2% glucose and0.2% BSA (KRP-DA buffer). Final preparations routinelycontained 80-90%o granulocytes.

Preparation of grani ulocytes for homogenates. Peripheralblood was defibrinated with glass beads and diluted with3 vol of 0.01 M phosphate-buffered saline, pH 7.4 (PBS).3 vol of diluted blood were layered over 1 vol of a solutioncontaining 6.3% Ficoll (Pharmacia Fine Chemicals Inc.)and 9.9%o Hypaque (Winthrop Laboratories, New York),and the mixture was centrifuged at 300 g for 30 min at200 C. The supernatant fluid containing lymphocytes andmonocytes was removed. Sedimented erythrocytes and gran-ulocytes were resuspended in PBS to approximately theoriginal volume of defibrinated blood and mixed with anequal vol of 3% dextran in saline for sedimentation oferythrocytes. Remaining erythrocytes were removed by lysiswith hypotonic saline, leaving preparations of 96-99% neu-trophils, 1-3% lymphocytes, and no detectable platelets onexamination of Wright's stained smears or counting-cham-ber preparations.

Preparation of clectron micrographs.2 The phagocyticbactericidal assay mixture, including latex particles, wasprepared as described below. Latex spherules, 0.1 ml, werepreincubated with horseradish peroxidase (Sigma ChemicalCo.), 1 mg/ml, for 20 min at room temperature; 100 ,ugBSA, 0.1 ml serum, 5 X 106 E. coli, and 2.5 X 106 phago-cytes were added, and the mixture was incubated on a ro-tating wheel at 37°C for 20 min. The reaction was stoppedby adding chilled saline; the cells were centrifuged at 180 gfor 5 min and washed gently in KRP buffer with dextrosebut no albumin (KRP-D buffer), osmolarity adjusted to290 mosmol with NaCl. The cells were suspended in 1-mlserum and centrifuged at 15,000 g for 1 min in an Eppen-dorf microcentrifuge (Brinkmann Instruments, Inc., West-bury, N. Y.). The cell button was resuspended in 2% glu-taraldehyde in KRP-D buffer, final osmolarity 290 mosmol.After 1 h at 25°C, the cell pellet was removed, rinsed inbuffer, diced into cubes, and reacted with diaminobenzidinefor 1 h by the method of Graham and Karnovsky (13).After several buffer rinses, the specimen was fixed for 1 hat 4°C with 1%o osmium tetroxide in KRP buffer, dehy-drated in ethanol, and embedded in Araldite (Ciba ProductsCo., Summit, N. J.). Sections were stained with uranylacetate or lead citrate, or both, and examined with a Phil-ips 200 electron microscope (Philips Electronic Instruments,Mount Vernon, N. Y.).

Polyacrylam ide c1ectrophoresis of granulocvte homtoge-nates. Purified granulocyte preparations were frozen andthawed three times, homogenized in 0.05 M potassiumphosphate buffer, pH 7.8, with a motor-driven Teflon pestlein a Duall tube (Kontes Glass Co., Vineland, N. J.),cleared by centrifugation at 800 g for 15 min, and stored

2Electron micrographs were prepared by Dr. A. W. At-kinson, Jr. in the laboratory of Dr. Sidney P. Kent, De-partment of Pathology, University of Alabama in Birming-hanm.

at - 700C. The supernatant fluid was thawed within 40 hand tested for protein content (14); 0.2-0.4 mg protein involumes up to 200 Ad was placed on polyacrylamide forelectrophoretic separation by the method of Davis (15).Gels were stained for SOD activity by the technique ofWeisiger and Fridovich (16). Subcellular fractions frompurified granulocytes homogenized in sucrose, obtained asdescribed below, were also analyzed. Cyanide sensitivity ofthe electrophoresed enzyme was tested by adding 1 mMKCN to the gel with the photoreduction system.

AssaysPhagoctosis-associated NBT reductioni. The NBT re-

duction accompanying phagocytosis was studied by a previ-ously described technique (17) with the following modifi-cations: Zymosan (Nutritional Biochemicals Corp., Cleve-land, Ohio) was prepared by the method of Wardlaw andPillemer (18) and suspended in saline to a concentration of50 mg/ml. 1 vol of zymosan was incubated with 4 vol ofnormal human serum (processed and frozen at - 70'C topreserve complement activity) for 30 min at 370 C. Theopsonized zymosan was centrifuged and resuspended insaline to a concentration of 10 mg zymosan/ml. Duplicatereaction mixtures consisted of 0.1-ml opsonized zymosanparticles, 0.2-ml (200 ,Lg) SOD or its controls, 0.4 ml of0.2% NBT in saline (17), and 2.5 X 106 phagocytes (granu-locytes and monocytes) in 0.2-ml Hank's buffer. The par-ticle: phagocyte ratio was approximately 50: 1. Reactiontime was 20 min at 370C.

NBT reduction by grannlocvte homtogcenates. The reduc-tion of NBT by fractions from homogenates of humangranulocytes was also studied. In one group of experimentsapproximately 109 cells were frozen and thawed three timesand homogenized for 30 s in 0.05 M potassium phosphatebuffer, pH 7.8, using an iced Duall glass tube and Teflonpestle. After homogenization no intact cells were seen onmicroscopic examination. A "particulate fraction" (primarilycell membranes and nuclear debris, as identified by elec-tron microscopy) was pelleted by centrifugation at 650 gfor 10 min; the supernatant fluid ("supernatant fraction")was removed for further processing and for study of itsSOD activity (see below). The pellet was washed in alarge volume of buffer then resuspended to 2 ml in phos-phate buffer or in 0.154 M KCl containing 3.2 X 10' MKHCO3 (alkaline isotonic KCl). Particulate fractions weresubjected to sonication using a Branson W185 Sonifier withmicrotip (Heat Systems-Ultrasonics, Inc., Plainview, N. Y.),power setting 55-60 W. The suspensions were placed inan iced bath and sonicated three times for 3 min each with2-min intervals inbetween. The scnicates were cleared bycentrifugation at 10,000 g for 15 min.

In a second group of experiments, 106-100 granulocyteswere suspended in 2 ml of 0.34 M sucrose and homogenizedas described above. The homogenate was centrifuged at 200 gfor 10 min to remove any intact cells that remained. Thepellet was washed once in 1-ml sucrose (200 g for 10 min),and the supernatant fluids were pooled and centrifuged at1,000 g for 10 min. The resulting pellet contained nucleardebris and cell membranes by electron microscopy and wasdesignated the "particulate fraction." This pellet was re-suspended in 1-ml sucrose and centrifuged at 1,000 g for10 min. Recentrifugation of the pooled supernates at 30,000gfor 20 min pelleted a "granular fraction" containing pri-marily lysosomal granules by electron microscopy. Thispellet was washed in sucrose, and the pooled supernateswere designated the "soluble fraction." The pellets were re-

Superoxide Generation and Phagocytic Bactericidal Activity 1359,.C

suspended in 1-ml sucrose, and all fractions were sonicatedfour times in an iced bath for 1 min each with coolinginbetween. Particulate and granular fractions were clearedby centrifugation before use. Each of the fractions wasstudied for its ability to reduce NBT in a modificationof the assay of Beauchamp and Fridovich (4), using 25'Cincubation temperature, 2.5 X 10' M NBT, 0.05 M sodiumcarbonate buffer, pH 10.2, 1 X 10-4 M EDTA, and 5 X 10-5M NADHor 2 X 10-' to 5 X 10- M NADPHas substratefor the reaction, in a total vol of 3 ml. The reaction wasinitiated by addition of NBT, and the reduction of NBT wasfollowed continuously by monitoring the increase in ab-sorbance of light at 560 nm using a Gilford model 2,000recording spectrophotometer (Gilford Instrument Company,Oberlin, Ohio). Involvement of 02- in the reaction wasdemonstrated by a decrease in the rate of NBT reductionwhen 15 ltg SOD was added before the addition of NBT.Duplicate assays of each cell preparation were run with andwithout 1 mM KCN and with and without NADH orNADPH. Certain preparations were studied at pH 7.0 orpH 7.4. Some of the starting leukocyte preparations werenot sedimented through Ficoll-Hypaque and contained 85-95% granulocytes and the remainder lymphocytes. Resultswith these preparations and those obtained with homoge-nates of 96-99% granulocytes were almost identical andwere therefore pooled.

Cytochrorne c reduction. A modification of the assay ofBabior et al. (11) was used. Phagocytes, 2.5 X 106 in 1 mlKRP-DA buffer, were preincubated for 5 min at 37'C inpolypropylene tubes (Falcon Plastics Co., Division of Bio-Quests, Oxnard, Calif.) with KRP-DA buffer and 30 ,ugSOD, as appropriate, to a total vol of 1.3 ml. 190 ml eachof ferricytochrome c, 1.2 mM (Type III, Sigma ChemicalCo.), and opsonized zymosan, 10 mg/ml prepared as above,were added to begin the reaction. After incubation for 10min the reaction was stopped by returning the tubes to aniced bath, and the contents of the tubes were promptly cen-trifuged at 4°C, 200 g for 10 min. Absorbance of the super-nates at 550 nm was determined in a spectrophotometer,and the results with duplicate or triplicate tubes were aver-aged. Results were converted to nanomoles cytochrome creduced using the extinction coefficient Erso nm =2.1 X 104M-1 cm-' (19). Tubes containing only buffer and cyto-chrome c were incubated as above and served as blanks.

SOD activity. Leukocytes from patients with CGD andpaired controls were lysed and homogenized as describedabove for the assay of NBT reduction. Supernatant fluidsafter the initial centrifugation at 650 g were cleared bycentrifugation at 30,000 g and assayed in multiple dilutionsfor their ability to inhibit reduction of ferricytochrome cby 02- generated by the aerobic xanthine-xanthine oxidasesystem (6). In this assay a unit of SODactivity is definedas the amount of enzyme required to inhibit by 50% astandard rate of ferricytochrome c reduction (0.025 ab-sorbance unit per minute) (6) ; our results are expressedas units of activity per milligram of cell protein.

Phagocytosis-associated chemilurminescence. Zymosan par-ticles were prepared and opsonized as described above butresuspended in KRP-DA buffer to a concentration of 4-mgzymosan/ml. The 4-ml reaction mixture consisted of 1.0-mlopsonized zymosan suspension; 50 ,ug SOD or BSA; 1.95ml KRP-DA buffer; and 10' phagocytes in 1.0 ml KRP-DA. The leukocytes were kept at 4°C until added to beginthe reaction; all of the other components were at ambienttemperature. The reaction was carried out in siliconizedliquid scintillation counting vials which had been stored

overnight in the dark and exposed only to red light duringthe experiment (20). Phagocytes were added at 1-min in-tervals, the contents of the vial were mixed by gentle swirl-ing, and the vial was immediately placed in the chamber forthe initial (time 0) 1-min reading, then remixed beforeeach subsequent count. Luminescence measurements weremade with a Beckman LS-150 ambient-temperature liquidscintillation counter (Beckman Instruments, Inc., Fullerton,Calif.) in the tritium region, with only the front photomulti-plier tube activated (20, 21). Empty vials were counted be-fore the experiment, and those giving the lowest counts(1,200-2,000 per min) were selected for use.

Phagocytic bactericidal activity. A modification of themethod of Quie, White, Holmes, and Good (22) was used,as previously described (23). The 1-ml reaction mixtureconsisted of 0.1 ml fresh human serum, approximately 2.5X 106 or 5 X 10' bacteria in 0.3 ml KRP-DA buffer, 50-200,gl of enzyme, benzoate or control solutions, and 2.5 X 106phagocytes in KRP-DA buffer. In experiments in whichlatex particles were included, 0.1 ml latex spherules, 0.81Atm diameter (23), were preincubated in the reaction tubewith SOD, catalase, or their controls for 20 min at roomtemperature. The tubes were then chilled in an iced bath,and serum, bacteria, and phagocytes were added in thatorder. Samples were removed at timed intervals, the leuko-cytes were disrupted in water, and the diluted suspensionwas sampled to make agar pour plates for colony counting(23). To determine the "representative" experiments graphedin Figs. 2, 4, 6, 7, and 8, differences in colony counts at 60and 120 min in the presence of the inhibitor (enzyme orscavenger) and its control were added for each experimentand then ranked. The representative experiment was selectedfrom the middle one or two experiments of each set.

Phagocytosis. To determine the effect of enzymes andscavengers on the ingestion of particles in the assays de-scribed above, particle uptake was measured in each assayusing the reaction mixture and conditions of that assay.For the assay of phagocytic NBT reduction, saline wassubstituted for NBT and duplicate tubes were preparedcontaining 200 jug SOD or BSA. Smears made of the re-action mixture after incubation for 20 min were stainedwith Wright's stain and coded. The number of zymosanparticles in 200 phagocytes was determined by light micros-copy for each of the four tubes without knowledge of thecontents of the tube from which individual smears weremade. To study uptake of zymosan in the chemiluminescenceassay, smears were made after incubation for 14 min, andthe number of particles ingested by 400 phagocytes was de-termined. The uptake of 1"5I-labeled zymosan was also usedto study phagocytosis in this system, as previously de-scribed (20).

The uptake of radiolabeled E. coli and S. aureus by pha-gocytes was studied in three to nine replicate tubes usingthe bactericidal assay system, except that the volume ofeach ingredient was doubled and the bacteria were culti-vated overnight in 5-ml trypticase soy broth or minimalmedium (24) to which 10 uCi ["C] amino acid mixture(New England Nuclear, Boston, Mass.) had been added.After incubation for 8-40 min (as noted), the reaction wasstopped with iced saline, and phagocytes containing radio-labeled bacteria were separated from uningested bacteria bycentrifugation at 150 g for 5 min at 4°C. The cell pellet waswashed and trapped on membrane filters, and radioactivitywas determined in a liquid scintillation counter as detailedpreviously (25). Radioactivity of pelleted material in dupli-cate tubes in which the reaction was stopped immediately

1360 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

after adding chilled phagocytes (time 0) or in duplicatetubes that lacked phagocytes was consistently less than 15%and usually less than 10% of that at the peak of bacterialuptake.

Phagocytic ingestion of bacteria was studied microscopi-cally by making smears from duplicate assay tubes after25 min incubation, staining them with Gram's stain, codingthem, and determining the percentage of 400 phagocyticcells containing at least one bacterium and the number ofbacteria per phagocyte, without knowledge of whether theoriginal reaction tube contained SOD or its control, BSA.

RESULTS

Effect of SOD on phagocytic NBT rcduiction. Thereduction of NBT by phagocytes ingesting opsonizedzymosan particles was studied in the presence of SODor an equal concentration (200 ,ug/ml) of BSA, heat-denatured SOD, or apoenzyme as controls. The extentof dye reduction achieved in the presence of SODwasconsistently approximately two-thirds of that achievedin the presence of BSA or inactivated SOD (Table I).Ingestion of zymosan in the presence of active enzyme(2.33 particles/cell) was not significantly differentfrom ingestion in the presence of BSA (2.37 particles/cell). We have previously reported that SOD did notinhibit ingestion of '"I-labeled zymosan at this sameparticle to phagocyte ratio (20).

Effect of SOD on NBT reduction by leukocyte ho-tiogenates. In further attempts to determine if phago-

cytes have the capacity to generate 02, homogenatesof purified granulocytes in phosphate buffer wereseparated into a particulate fraction containing cellmembranes and nuclei and a supernatant fraction con-taining the remaining cell constituents. These fractionswere tested for their ability to reduce NBT in the pres-ence or absence of SOD. The reaction was studied atpH 10.2 in order to maximize the one-electron reductionof oxygen to generate 02- (5, 26). As summarized inTable IIA, the rate of NBT reduction with either frac-tion was increased approximately threefold when NADHwas added as substrate for the reaction. SOD inhibitedthe rate achieved with NADHby particulate and su-pernatant fractions from 10 normal control homoge-nates by 61% and 57%, respectively. Specific activitywas greater in the particulate fraction both before andafter the addition of SOD. Boiling this fraction elimi-nated its SOD-inhibitable NBT reducing activity. Al-though the mean rate of NBT reduction by both frac-tions from patients with CGDwas approximately two-thirds the rate achieved with normal homogenatefractions, this difference wvas not significant (P > 0.2,\V7ilcoxon two-sample rank test, two-tailed [27]), andthe extent of inhibition by SODdid not differ appreci-ably in the two groups. There was no inhibition of NBTreduction by heated SODor in the presence of 1 mMKCN, which inactivates this enzyme (16).

TABLE I

Inhibition of Phagocytosis-Associated NBI' Reduction by SOD

NBTProtein added* reduction: Inhibition

ODi65 SoBSA 0.230i0.025 (7)SOD 0.158±0.021 (7) 31 (25-46)§SOD, heated 0.2274±0.021 (2)SOD, apoenzyme 0.228±0.041 (3)

* 200,ug/ml.t Mean ±SEMof averages of values for total dye reductionin duplicate tubes. The number of experiments is given inparentheses.§ The mean and range of values are given for seven experi-ments comparing results with SODand BSA.

At pH 7 or 7.4 the rates achieved by particulate frac-tions from three control homogenates in the presenceof NADHwere slower (mean 4.9 nmol/ min per mg),but the inhibitory effect of SOD remained appreciable(37%). NADPHwas much less effective as a substratefor the reaction. In three experiments at pH 10.2,NADPH concentrations as high as 40-fold that ofNADHpermitted only up to 60% as much NBT re-duction as achieved with NADH. NBT reduction inthe presence of NADHor NADPHwas consistentlyincreased slightly by 1 mMKCN. NADHoxidase ac-tivity of the particulate fraction from normal cells wasdemonstrable at pH 10.2 and pH 7.4, as determined bydecrease in absorbance at 340 nm in the presence of10' MI NADH (10); SOD had no effect on this ac-tivity at either pH.

In additional experiments normal and CGDgranulo-cytes were homogenized in sucrose, and particulate,granular, and soluble (cytosol) fractions separated bydifferential centrifugation were tested at pH 10.2 for02i-generating activity. The mean total protein contentof the three fractions and the mean number of cellshomogenized were similar for control and CGDprepa-rations (14.3 mg and 5.8 X 10' cells for controls; 19.1mg and 6 X 108 cells for CGD homogenates). In thecontrol preparations specific activity in the presence ofNADHwas again highest in the particulate fraction(Table IIB). Based upon the total protein content ofeach fraction, total activity was greatest in the cytosol,which accounted for almost two-thirds of the total ac-tivity found in the three fractions (data not shown).The decreased activity at pH 7.4 expected in an O-generating system (5, 26) was also found with all threefractions from control homogenates. Although SODin-hibited NBT reduction in the presence of NADHap-proximately equally in fractions from control and CGDhomogenates, the rate of NBT reduction per milligram

Superoxide Generation and Phagocytic Bactericidal Activity 1.361

TABLE IISuperoxide-Dependent NBT Reduction by Fractions from Normal and CGDLeukocyte Homogenates

Rate of NBT reduction

Subcellular +NADHand InhibitionSource of homogenates* fraction base line +NADH SOD by SOD:

nmol/min/mg§ %A. P04 buffer

Controls (10) particulate 5.7±1.3 18.1±3.5 7.4±1.8 61supernate 3.4±-0.1 11.3 ± 1.6 4.8 i 1.3 57

CGDpatients (6) particulate 3.8±0.4 12.2±0.5 3.8±0.3 69supernate 7.8±0.8 3.1 ±0.5 60

B. Sucrose

Controls (4) particulate 18.6±7.5 38.7±4.8 16.4±0.8 58granular 21.7 ±4.3 8.5 ±1.8 61soluble 21.3 ± 1.7 23.8±4-1.6 12.4 i0.7 48

CGDpatients (3) particulate 10.8±0.7 19.2±3.3 8.7±1.6 55granular 9.0±0.1 18.5 ±t4.0 8.3 ±1.7 55soluble 9.2 +0.7 11.640.5 6.5 ±0.8 44

* Granulocytes were homogenized in phosphate buffer (A) or sucrose (B) and fractionated by differentialcentrifugation as described in Methods. The number of different homogenates tested is shown inparentheses.

rslwihNADH - result with NADHandSO 10%(result with result with NADH ) X 100%.

§ The mean ±SEMof the averages of two or three replicate determinations for each fraction are shown.The rate of increase in absorbance at 560 nm per mg of homogenate protein was determined as describedin Methods; conversion to nanomoles NBT reduced was made using E560 nm = 2.3 mM-' cm-', which wascalculated from a standard curve obtained with known amounts of formazan generated by reduction ofNBT with alkaline sodium ascorbate. The presence in the reaction mixture of 5 X 10-5 m NADH, orNADHand SOD, 5 ,g/ml, is noted. All values listed here were determined in the absence of KCN. Eachpreparation was also tested in the presence of 1 mMKCNand NADH(without SOD); mean values wereslightly higher than those shown. Preparation of particulate fractions from part A in alkaline isotonic KCIrather than phosphate buffer did not result in improved specific activity of the extracted material.

protein by particulate and cytosol fractions from pa-tients was about half that of these fractions from nor-mals (Table IIB). However, the difference was notsignificant (P = 0.057, Wilcoxon two-sample rank test,two-tail [27]).

Cytochrome c reduction by normal and CGD cells.The ability of 02- to reduce ferricytochrome c in thesurrounding milieu was used to determine the capacityof phagocytizing leukocytes to generate this free radi-cal (11). Cells from 23 controls, 9 patients with CGD,

TABLE IIIInhibition of Phagocytosis-Associated Ferricytochrome c Reduction by SOD

Ferricytochrome c reduction*

PhagocytizingSubjects No. Resting Phagocytizing and SOD

nmol/min/2.5 X 106 PhagocytesControls 23 0.18±0.03 3.4340.11 0.14±0.02CGDpatients 9 0.19±0.06 0.24±0.06 0.25±40.06CGDcarriers, XLRJ 5 0.13±0.02 1.69±0.29 0.15±0.04CGDcarriers, non-XLRI: 3 0.17±0.01 3.68±0.38 0.14±0.10

* Mean ±SEMof averages of two or three determinations.XLR stands for X-linked recessive (inheritance).

1362 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

C,-)

a)

0-

C-)0

C)0m

0)

Time (minutes)FIGURE 1 Inhibition of phagocytic bactericidal activity by SOD. The number of viable bac-teria in the 1-ml reaction mixture, determined as colony-forming units, is plotted as a functionof the time at which the mixture was sampled. In the experiment with E. coli the enzyme waspresent at a concentration of 100 ,ug/ml; with staphylococci the concentration was 200 ,g/ml.Heated enzyme at equivalent concentrations and saline were used as controls. 2.5 X 106 phago-cytes were used in all bactericidal assays. Selection of "representative" experiments is de-scribed in Methods.

and 8 mothers or grandmothers of children with CGDwere studied. As summarized in Table III, normal con-trol leukocytes reduced cytochrome c actively duringphagocytosis, and this reduction was inhibited completelyby SOD. However, there was no significant cytochromec reduction by patients' leukocytes. All five carriers ofthe X-linked form of the disease, as determined by thequantitative NBT assay and a compatible family his-tory. had values which fell between those of patientsand controls: 23%, 54%, 56%, 59%, and 67% of theextent of cytochrome c reduction achieved by phago-cytizing control cells in the same experiment. Cells fromthree of these carriers had activity equal to that of con-trol cells if incubation was prolonged to 60 min fromthe usual 10 min; cells from the other two neverachieved normal activity. Phagocytes from all threematernal carriers of the non-X-linked form of the dis-ease, as determined by normal phagocytic reduction ofNBT and compatible family history, had normal ac-tivity in this assay.

Effect of SOD on phagocytosis-associatcd cheiila-minescence. During the process of phagocytosis, hu-man leukocytes emit a burst of luminescence that can bequantitated in a liquid scintillation spectrometer (28).To obtain further evidence that 02- is generated byphagocytizing leukocytes and to determine if 02- gen-eration plays a part in phagocytosis-associated chemi-luminescence, we tested the capacity of SOD to inhibitthis phenomenon. In the presence of BSA as control,luminescence rose rapidly from a background level ofapproximately 5 X 103 cpm at time 0 to a maximum ofover 1.7 X 105 cpm at 13 min. Xhen BSA was replacedby an equal concentration of SOD (50 /Ag/ml), the rateof luminescence generation fell from 2.3 X 10' cpm/minto 0.66 X 10' cpm/min, a reduction of 71%; and peakluminescence was reduced by 64%. In 11 experiments,SOD at a concentration of 100 /Ag/ml inhibited peakluminescence by 65 to 76% (mean 70%) (20). Particleingestion in the presence of the enzyme was not signifi-cantly different than that in the presence of BSA (2.81

Superoxide Generation and Phagocytic Bactericidal Activity 1363

FIGURE 2 Electron micrograph of one granulocyte illus-trating E. coli and latex particles (L) within the samephagocytic vacuole. The phagocytic bactericidal assay sys-tem was used, except that latex particles were preincubatedwith peroxidase in an attempt to mark the location ofparticles by surrounding deposition of electron-dense mate-rial (13). The nucleus (N) is shown. MagnificationX 13,000.

and 2.87 particles/phagocyte, respectively). In addition,SOD in a concentration of 100 ig/ml did not influencethe phagocytosis of "20I-labeled zymosan in this system,uningested labeled particles being separated from phago-

cytes with ingested particles by velocity sedimentationin an isokinetic gradient (20).

Effect of SOD on phagocytic bactericidal activity.The effect of SODon the bactericidal capacity of humangranulocytes was studied using strains of bacteria ob-tained from infected humans. In the presence of 50-2501ig SODmodest inhibition was achieved, as shown fortwo representative experiments in Fig. 1. However,the effect was reproducible in that inhibition was dem-onstrated in six consecutive experiments each withE. coli and staphylococci. Specifically, in 12 experimentsin which colony counts were compared at 30, 60, and120 min (and at 90 min in 5 experiments), the colonycount in the presence of the enzyme preparation wasgreater than that with the control 35 of 41 times; ofthe six times the count was higher with the control,five were at the 30-min sampling when differenceswould be expected to be minimal. The probability thatthis might occur by chance is less than 1 in 10,000 (signtest, two-tailed [27]).

In an attempt to facilitate entry of SOD into thephagocytic vacuole, various concentrations of the en-zyme or its controls (heat-denatured enzyme or BSA)were fixed to latex particles by preincubation in thereaction tube with 0.1 ml of the particles. Approximately20% of the total enzyme activity in 100 ltg of prepara-tion incubated with the latex was associated with theparticles when they were separated by centrifugation;5% of the total activity remained after one wash withwater. Electron micrographs made of this system after20 min of phagocytosis indicated that particles and

E. colb6SAn

SOD cys

Soo

,----- as00 106-BS 5

30 60 90 120

S. aureus

/8S Heated SOD-- tIez~_)Pto-M ___Ptcagoc

,_ - -." SOOI

- - HoeatedSODBSA

30 60 90 120

Time (minutes)

S. viridans

0(~1 SA So NoYt@S~~1j 9-~He~~ Phagocytes

SOD

BSAHeated

105, ~~~~~~~SOC)30 60 90

FIGURE 3 Inhibition of phagocytic bactericidal activity by SOD in the presence of latexparticles. In the experiments shown here, 200 ,ug of either BSA, SOD, or heat-denatured SODwas present in an individual assay tube. Illustrated experiments are representative of 11 withE. coli, 5 with staphylococci, and 7 with streptococci done at various enzyme concentrationsand bacteria: cell ratios. Seven additional confirmatory experiments showing a similar inhibitoryeffect were performed with staphylococci in a second laboratory (R.L.B.).

1364 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

E io7-V)Q)

.c0.LCD0m

C3)>0

11

bacteria could commonly be found together within thesame vacuole (Fig. 2).

In the presence of SODand latex particles there wasconsistent, marked inhibition of the killing of E. coli,staphylococci, and streptococci by normal human leuko-cytes, as shown for three representative experiments inFig. 3. In 23 experiments in which bacterial colonycounts were compared at 30, 60, 90, and 120 min usinginactivated SOD or BSA as controls, at each of 92time points the colony count with SODwas higher thanthat with either control (P 0.2) by analysis of variance (27).t 200 ,ug/nil.

6-

0

E

C ,/4-/

0.~~~~

a)

t2 // .-.~~~SOD

8 16 24 32 40Time (minutes)

FIGURE 4 Uptake of radiolabeled E. coli by plagocytes inthe presence of SOD, 200 jug/ml, or an equal concentrationof BSA as control. The phagocytic bactericidal assay sys-tem was used, including latex particles, except that volumesof all ingredients were doubled. The reaction was stoppedat 8-min intervals, and leukocytes were separated from un-ingested bacteria by lowspeed centrifugation. Means ofleukocyte-associated radioactivity in triplicate tubes areplotted as a function of the length of incubation. The barsrepresent the SE of each mean.

the absence of latex particles there was slight but con-sistent inhibition of the phagocytic killing of E. coliand staphylococci by catalase (Fig. 5). However, inthe presence of latex, catalase inhibited the killing ofall three bacterial strains as effectively as (lid SOD(Fig. 6). There was no enhancement of bacterial growthin the presence of the enzyme when leukiocytes wereomitted.

Inasmuch as bactericidal activity Was markedly in-hibited by enzymes that removed either Or or H202,we considered the possibility that optimal killing re-quired the combined effect of both these agents. In-teraction of 02- and H202 has been said to lead toproduction of hydroxyl radical (-OH) (29). To testthe possibility that -OH is generated by phagocytizingleukocytes and that this radical is required for thebactericidal event, the phagocytic bactericidal assaywas performed in the presence of sodium benzoate andmannitol, scavengers of - OH (30-32). As shown inFig. 7, there was modest but definite inhibition of phag-ocytic killing by benzoate concentrations that had no di-rect effect on the bacteria. At 50 of 52 time points sampledin 10 consecutive experiments, the colony count washigher in the presence of either 0.01 AI or 0.02 AI ben-zoate than the saline control (P

E. coli

107

30 60 90

S. aureus

Catalase

2- BSA

30 60 90 120

Time (minutes)FIGURE 5 Inhibition of phagocytic bactericidal activity by catalase, 200 /Ag/ml, in the absenceof latex particles. An equal concentration of BSA was used as control. These results arerepresentative of the effect achieved in two experiments with E. coli and five with staphylococci.At 27 of 27 sampling times in these experiments the colony count was higher in the presenceof catalase than in the presence of BSA (P

E-,

0)

00

C)

0'3)

0

~I.w5

Time (minutes)

FIGURE 7 Inhibition of phagocytic bactericidal activity by sodium benzoate at a final concen-tration of 0.01 M or 0.02 M compared to 0.15 M NaCl, and by 0.04 M mannitol compared to0.15 M NaCl and 0.04 M glucose. Latex particles were not present. The results with benzoatethat are plotted are representative of those achieved in six experiments with E. coli, threewith staphylococci, and one with streptococci. The presence of latex particles in several ofthese experiments had no influence on the inhibition achieved by benzoate. The results withmannitol shown are representative of those in six experiments with 0.03 M-0.05 M mannitoland E. coli. The osmolalities of the final reaction mixtures containing the saline, benzoate,mannitol, and glucose solutions were approximately equal.

and the addition of latex to the reaction mixture didnot increase the extent of inhibition by benzoate.

Mannitol inhibited the phagocytic killing of E. colislightly but consistently at concentrations of 0.03 M-0.05 M: at all 22 sampling points in six consecutiveexperiments colony counts were higher with mannitolthan with glucose or saline (P < 0.0001). Mannitol hadno direct effect on E. coli but inhibited slightly thegrowth of the two staphylococcal strains tested, pre-venting our use of this bacterium for accurate evalua-tion of mannitol effects on the neutrophil. Wewere un-able to demonstrate consistent inhibition of phagocyticbactericidal activity by ethanol, a less reactive scav-enger of -OH than benzoate (30, 31), in concentra-tions lower than those that inhibited ingestion slightly(0.08 M). There was no inhibition of the uptake ofradiolabeled organisms by catalase, benzoate, or man-nitol, as summarized in Table V.

Granulocyte SOD. Lysates of purified granulocyteswere placed on polyacrylamide for electrophoretic sepa-ration. The location of SOD after electrophoresis wasdemonstrated by its capacity to inhibit the reductionof NBT by an 02-producing system incorporated intothe gel (4, 16). Two distinct achromatic bands wereseen in all tested lysates from 12 normal individuals, 4

TABLE V

Uptake of 'IC-Labeled Bacteria by Phagocytes in the Presenceof Catalase, -OH Scavengers, or their Controls

Phagocyte-associatedCatalase, scavenger, radiolabeled bacteria

or controlpreparations Latex E. coli S. aureus

cpm* cpm*Catalaset absent 1,553 +56BSAJ absent 1,225+-81

Catalaset present 878 + 105 1,465 +704BSAt present 796-+i175 1,387 +670

Benzoate, 0.01 M absent 1,0204165Benzoate, 0.02 M absent 1,0894+185Saline absent 966+114

Mannitol, 0.04 M absent 2,312 +268Mannitol, 0.05 M absent 2,372+362Glucose, 0.05 M absent 2,389+473

* Mean +SD of values from 6-12 replicates in two experi-ments. Square root transformations of catalase, benzoate, andmannitol values were not significantly different from those oftheir controls (P > 0.2) by analysis of variance (27).1 200 ,ug/ml.

Superoxide Generation and Phagocytic Bactericidal Activity 1367

patients with chronic myelogenous leukemia, and 3 pa-tients with CGD. A minor, slower band could be foundslightly toward the cathode from electrophoresed man-ganese-containing SODpurified from E. coli (33). Themajor band corresponded in electrophoretic mobility tothe most prominent of three closely associated bandsfrom lysates of human erythrocytes; this band wasslightly anodal to that of purified bovine SOD. Themajor granulocyte band (but not the minor one) andall three erythrocyte bands were removed by cyanidetreatment of the lysates, a result expected with cupricenzyme (16, 34). Purified granulocytes were homoge-nized in 0.34 M sucrose, and subcellular fractions wereseparated by differential centrifugation. The super-natant fluid after centrifugation at 105,000 g (cytosolfraction) contained large amounts of the major, cya-nide-sensitive SODby this technique.

Supernatant fluids after centrifugation at 650 g of ho-mogenates prepared in phosphate buffer were assayedfor their ability to inhibit reduction of ferricytochromec by 02 generated by the xanthine-xanthine oxidasesystem. Granulocyte homogenates from four normalcontrols had 2.4±0.5 (mean±+-SEM) U of SODactivityper milligram protein; honmogenates from four patientswith CGDhad 2.4±0.2 U of activity per milligram pro-tein. Treatment with 1 mMKCNreduced SODactivityin both normal and CGD homogenates by 74-100%(mean 87%); heating to 100'C for 15 min eliminatedactivity. Particulate fractions from these homogenateshad negligible SOD activity.

DISCUSSION

Babior and colleagues have presented evidence that O-fis generated by normal granulocytes during phagocy-tosis (11) ; and Curnutte, Whitten, and Babior haveshown defective O2- generation by phagocytes from twopatients with CGD (12). Wehave confirmed defectiveOr2 generation by phagocytizing leukocytes from threepatients with the X-linked recessive form of CGDandsix patients with the non-X-linked (probably autosomalrecessive) form of this disease. The carrier state wasdetectable as an intermediate value in this assay in thethree mothers and two maternal grandmothers whosecells reduced NBT to an intermediate extent duringphagocytosis. CGD in these families occurred only inboys and in maternal relatives, consistent with X-linkedtransmission of the disease. The inhibition by SOD ofthe NBT reduction that accompanies phagocytosis fur-ther confirms the capacity of normal leukocytes to gen-erate 0-. Moreover, this finding, in conjunction withthe failure of NBT and cytochrome c reduction by CGDgranulocytes and the occurrence of a partial defect inthese two metabolic events in mothers and maternal

grandmothers of male patients, strongly implicates 02-in NBT reduction.

Subcellular fractions from granulocyte homogenatesexhibited the capacity to generate 02- in the presenceof NADHas substrate, the greatest activity per milli-gram protein being found in the fraction containingcell membranes and nuclear debris. Subcellular frac-tions from homogenates of CGDgranulocytes had one-half to two-thirds the 02-generating capacity that nor-mal homogenates did, but this difference was not sta-tistically significant and was slight compared to themarked deficiency of CGDcells in phagocytosis-associ-ated Or2 generation. Superoxide generation was not di-minished by 1 mMcyanide treatment of either the nor-mal or CGDhomogenates. This cyanide-insensitive uti-lization of NADHas substrate for 02- formation andthe demonstration of NADH oxidase activity in theparticulate fraction suggest that 02- can be generatedby NADHoxidase activity. Whether the 02--generatingactivity in the membrane-rich particulate fraction isactually associated with the cell membrane must bedetermined by study of purer cell fractions. Approxi-mately normal NBT reduction by homogenates of CGDleukocytes has been previously shown (35); our exten-sion of these findings further interrelates O2- genera-tion and NBT reduction. Demonstration of markedlydeficient 02- generation by intact CGD cells but onlya slight deficiency in 02- generation by CGD cell ly-sates again raises the possibility that the enzyme or en-zymes controlling NBT reduction, and perhaps thephagocytosis-associated respiratory burst, are presentwithin CGDleukocytes but not strategically placed or,for some other reason, not fully activated by ingestion.

Elaboration of luminescence by phagocytizing leuko-cytes appears to depend in large part on the generationof 02 during ingestion in that the rate of generationand the maximal extent of chemiluminescence achievedby phagocytizing granulocytes were significantly in-hibited by SOD. That 02- is the actual luminescingagent has not been shown, however. Allen, Stjernholm,and Steele have suggested that singlet oxygen (102)could be that agent (28). This electronically excitedform of oxygen appears to participate in ordinary chem-ical reactions involving molecular oxygen (36, 37)where its presence has been monitored by its generationof luminescence (36-42). To our knowledge no directproof that this species is generated by cells has beenpresented.

Several laboratories have reported that the aerobicxanthine-xanthine oxidase reaction, which generatesOf, can generate chemiluminescence (38-42). Stauff,Schmidkunz, and Hartmann (39) have attributed lumi-nescence in this system to generation of singlet oxygen

1.368 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

from the basic 027 dismutation reaction:

0_- + 02- 02'+ 102.42H+H202

It would appear that SOD has no direct effect on 102once this agent has been formed (42). However, theenzyme-catalyzed dismutation does not give rise to'02 (41, 42), which could explain the inhibition ofchemiluminescence by SOD.

The involvement of 0'7- in the NBT reduction, cyto-chromle c reduction, and chemiluminescence exhibited byphagocytizing granulocytes strongly supports the con-cept that 027 is a significant by-product of metabolismin aerobic cells. The possibility that this phenomenonmight serve some useful purpose would seem most likelyfor phagocytic cells. Results of the experiments re-ported here suggest that 02- generation is involved inthe killing of ingested bacteria by such cells. SOD inthe reaction mixture consistently diminished the killingof E. coli, staphylococci, and streptococci by phago-cytes, compared to an equal concentration of heat-de-natured enzyme or BSA. The effect was far morestriking when latex particles were also present, presuma-bly because the enzyme, shown to adhere to the par-ticles, could be introduced more effectively into thephagocytic vacuole with the bacteria. Electron micro-graphs confirmed the coexistence of latex and bacteriawithin the same vacuole in this system. The enzymedid not inhibit ingestion of bacteria in the presence orabsence of latex particles. The failure of SOD to in-hibit phagocytosis of paraffin oil-E. coli lipopolysaccha-ride emulsions has also been shown (43).

The inhibition of phagocytic bactericidal activity bySODdoes not necessarily indicate that 07- is itself thelethal agent. In fact, results obtained in this systemusing catalase, whiclh removes H202, suggest otherwise.Evidence that peroxide is involved in the killing ofphagocytized bacteria is well documented, as reviewedby Klebanoff and Hamon (44). The inhibition of pha-gocytic bactericidal activity by catalase reported hereand previously by others (45, 46) supports that con-tention. However, inasmuch as the predicted effect ofSOD would be to increase the rate of H202 formationfromn 02-, one might expect more H202 to be formedwithin the phagocytic vacuole in the presence of theenzyme and, therefore, bactericidal activity to be in-creased. That more H202 is formed under these co-nli-tions does in fact appear to be the case (47).' Onepossible explanation for our findings miight be the re-(uirement for both H202 and 02- for optimal phago-c\-tic bactericidal activity. Interaction of these agentshas been shown to occur in vitro and to lead to pro-duction of the potent oxidizing agent, - OH (31, 32,

48) by the cycle of Haber and Weiss (29), as repre-sented by the reaction:

02- +H0202 *OH + OH- + O2.The inhibition of phagocytic bactericidal activ ity abysodium benzoate and mannitol, scavengers of 01-1 butnot of 027 or H202 (30, 32), is consistent with thepossibility that -OH is formed during phagocytosisand is required for the bactericidal event. However, in-asmuch as the activity of benzoate and mannitol uponphagocytes may not be limited to removal of *011,it must be stated that their antibacterial effect is comi-patible with but does not prove the importance of ^.OHin phagocytic killing.

A role for '02 in the killing of phagocytized bacteriahas been suggested by the finding that Sarcinat lltcacontaining carotenoid, a scavenger of 102, were pro-tected against phagocytic bactericidal activity comparedto a colorless mutant (49). A second possible explana-tion for the apparent requirement for both 02- andH202 in the phagocytic bactericidal event, as well asfor the relationship between 02- and clhemiluminescence,might involve '02, as suggested by 'the experiments ofArneson. Using the aerobic xanthine-xanthine oxidasesystem to generate 027, he noted that the addition ofH202 greatly increased and the addition of catalase de-creased chemiluminescence (40). He suggested that*OH was being generated in the Haber-W'eiss reac-tions and that the reaction responsible for luminescencein this systemAwould be:

02- + * OH- OH- +'O 2.

Thus, SOD wsithin the phagocvtic vactiole cuold con-ceivablv inhibit bactericidal activity (and luminescence)through its suppression of '02 formation from 0-1 orfrom the spontaneous 02 dismutation reactiony, as welas through suppression of formation of *OH itself.

It seems highly likely that protection of a cell fromautogenous 02- and perhaps, thereby, other high-energyintermediates of oxidative metabolism, depends upon thepresence of SODwithin that cell. A copper- and zinc-containing form of this enzyme is located in the cyto-sol of eukaryotic cells which also possess smalleramounts of a manganese SOD in the matrix of theirmitochondria (16). Aerobic bacteria contain an intra-cellular mianganoenzynme (33) as well as a periplasiiciron ellnz-ye (50). Exvidence sup)p)orting a l)rotecti\erole for SOD inclndes the linding by MAlcCord, Keele,and 1ri(loxiclh of SOD activity in all eighlt species ofaerobic bacteria analyzed, vet in none of the six speciesof strict anaerobes studied (7). Additional experi-nments leave shown that increased concentrations of in-tracellular SOD could be induced in E. coli B or theeukaryote Saccharonzyces crervisiac by growth in 100%

Superoxide Generation and Phagocytic Bactericidal Activity 1369

oxygen; such preconditioning increased the resistanceof these organisms to growth in 20 atm oxygen (8, 9).Cultivation in an aerated, iron-poor medium resulted inbacteria with high levels of the manganoenzyme butlow levels of the iron enzyme and decreased resistanceto death from exogenous 02- generated by a photo-chemical or enzymatic system (8). In experiments notreported here we have found decreased resistance toexogenous 02- in a mutant E. coli lacking the periplas-mic enzyme.'

Beckman, Lundgren, and Tdrnvik have found SODwith the electrophoretic mobility of cuprozinc enzymein all human cells studied except peripheral blood gran-ulocytes, which apparently contained only the mito-chondrial enzyme (51). However, in our experiments,in agreement with other recent reports (52, 53), lysatesof 96-99% pure neutrophils from normals and patientswith CGDor chronic myelogenous leukemia have con-sistently shown two distinct bands on electrophoresisin polyacrylamide gels. A minor, slower band corre-sponds approximately in position to electrophoresedmanganese-containing SOD, purified from E. coli; themajor band corresponds in electrophoretic mobility toerythrocyte cupric enzyme and is removed by cyanidetreatment, a result expected with cupric enzyme.4

The results of the studies described here, placed inthe perspective of current understanding of granulo-cyte physiology, have led us to formulate the followingconceptual model of the process of phagocytosis: Stim-ulated by our own and reported observations (54-57)we would hypothesize that the enzyme system respon-sible for the respiratory burst of phagocytosis providesthe source of 02- and that this system is associated withthe cell membrane. Inasmuch as the same membraneforms the boundaries of the phagocytic vacuole, themechanism for converting molecular oxygen to super-oxide, then to peroxide, would be located between thecaptured bacteria and the cytoplasm of the cell. Be-cause 02 and H202 can diffuse even to the outside ofthe cell (11, 58-60), these agents would accumulate inthe vacuole where they would be free to act on thecaptured micro-organisms or to interact to generateother oxidants such as -OH or '02. Within the vacuole

'Miles, S. K., B. B. Keele, Jr., H. P. Misra, and R. B.Johnston, Jr. Manuscript in preparation.

4DeChatelet, McCall, McPhail, and Johnston originallyreported cyanide insensitivity of neutrophil homogenateSOD assayed by its interference with 02-mediated autoxi-dation of adrenalin (52). It would appear that this assay isinappropriate for study of the cyanide sensitivity of homoge-nate SODbecause such sensitivity can be demonstrated withthe same homogenates when interference with reduction ofcytochrome c or NBT, in fluid or gel systems, is used asthe assay of enzyme activity (Johnston and DeChatelet,unpublished observations).

they would have interference from neither cytoplasmiccatalase (61) nor, probably, cytoplasmic SOD; yetthese enzymes would act to protect critical cellularcomponents from the destructive effects of these agents.Lysosomal degranulation would introduce digestive hy-drolytic enzymes, cationic proteins, and myeloperoxi-dase into the vacuole. The last might stimulate free-radical (57, 62) or 102 (49, 63) formation from H202,or H202 and hypohalite anions, respectively, as a meansof enhancing the bactericidal effect of this agent (44,64).

It would seem clear from these and other studiesthat the respiratory burst coincident with granulocytephagocytosis is associated with significant productionof 02. Through this process the oxygen molecule isliterally charged with new energy. The inhibition ofphagocytic bactercidal activity, as well as chemilumi-nescence, by SODsuggests that this energy is involvedin the killing of ingested bacteria. Exactly how thisenergy is translated into microbial death remains to beelucidated.

ACKNOWLEDGMENTSWeare grateful to Doctors A. W. Atkinson, Jr. and SidneyP. Kent for the electron microscopy, John A. Burdeshawand Dr. Edwin A. Bradley for guidance in and perform-ance of the statistical analyses, Doctors Irwin Fridovichand Dale Kessler for constructive comments, Doctors BlairBatson, Richard Hortman, Paul Bianco, Steven L. Shore,and Rebecca Buckley for referral of patients with CGD,and to Dr. Alexander R. Lawton for critical review of themanuscript. Weare particularly appreciative of the diligenceand industry of Shannon Skalley, Peggy Morris, LindaGuthrie, Susan Murrmann, and Jackie Broadley who per-formed many of the experiments reported here.

This work was supported by U. S. Public Health ServiceGrants AI-10286, CA-13148, DE-02670, RR-00032, AI-10892, and GM-00091 and a grant from the Riley MemorialAssociation.

REFERENCES

1. Fridovich, I. 1972. Superoxide radical and superoxidedismutase. Accounts Chem. Res. 5: 321-326.

2. McCord, J. M., and I. Fridovich. 1968. The reductionof cytochrome c by milk xanthine oxidase. J. Biol.Chem. 243: 5753-5760.

3. Rajagopalan, K. V., and P. Handler. 1964. Hepatic alde-hyde oxidase. II. Differential inhibition of electrontransfer to various electron acceptors. J. Biol. Chem.239: 2022-2026.

4. Beauchamp, C., and I. Fridovich. 1971. Superoxide dis-mutase: improved assays and an assay applicable toacrylamide gels. Anal. Biochem. 44: 276-287.

5. Misra, H. P., and I. Fridovich. 1972. The role of super-oxide anion in the autoxidation of epinephrine and asimple assay for superoxide dismutase. J. Biol. Chem.247: 3170-3175.

1370 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan

6. McCord, J. M., and I. Fridovich. 1969. Superoxidedismutase. An enzymic function for erythrocuprein(hemocuprein). J. Biol. C/cnt. 244: 6049-6055.

7. McCord, J. M., B. B. Keele, Jr., and I. Fridovich. 1971.An enzyme-based theory of obligate anaerobiosis: thephysiological function of superoxide dismutase. Proc.Natl. Acad. Sci. U. S. A. 68: 1024-1027.

8. Gregory, E. M., F. J. Yost, Jr., and I. Fridovich. 1973.Superoxide dismutases of Eschlerichia coli: intracellularlocalization and functions. J. Bacteriol. 115: 987-991.

9. Gregory, E. M., S. A. Goscin, and I. Fridovich. 1974.Superoxide dismutase and oxygen toxicity in a eukary-ote. J. Bacteriol. 117: 456-460.

10. Baehner, R. L., and D. G. Nathan. 1968. Quantitativenitroblue tetrazolium test in chronic granulomatous dis-ease. N. Entgl. J. Med. 278: 971-976.

11. Babior, B. M., R. S. Kipnes, and J. T. Curnutte. 1973.Biological defense mechanisms. The production by leu-kocytes of superoxide, a potential bactericidal agent. J.Clini. Invest. 52: 741-744.

12. Curnutte, J. T., D. M. Whitten, and B. M. Babior.1974. Defective superoxide production by granulocytesfrom patients with chronic granulomatous disease. N.Enigl. J. Mcd. 290: 593-597.

13. Graham, R. C., Jr., and M. J. Karnovsky. 1966. Theearly stages of absorption of injected horseradish per-oxidase in the proximal tubules of mouse kidney: ultra-structural cytochemistry by a new technique. J. Histo-chcm. Cytoc/tein. 14: 291-302.

14. Ellman, G. L. 1962. The biuret reaction: changes inthe ultraviolet absorption spectra and its application tothe determination of peptide bonds. Ana/. Biochert. 3:40-48.

15. Davis, B. J. 1964. Disc electrophoresis. II. Method andapplication to human serum proteins. Ann. N. Y. Acad.Sci. 121: 404-427.

16. Weisiger, R. A., and I. Fridovich. 1973. Superoxidedismutase. Organelle specificity. J. Biol. Chent. 248:3582-3592.

17. Johnston, R. B., Jr., M. R. Klemperer, C. A. Alper,and F. S. Rosen. 1969. The enhancement of bacterialphagocytosis by serum. The role of complement compo-nents and two cofactors. J. Exp. Mcd. 129: 1275-1290.

18. Wardlaw, A. C., and L. Pillemer. 1956. The properdinsystem and immunity. V. The bactericidal activity ofthe properdin system. J. Exp. Med. 103: 553-575.

19. Massey, V. 1959. The microestimation of succinate andthe extinction coefficient of cytochrome c. Biochiimn. Bio-phys. Acta. 34: 255-256.

20. Webb, L. S., B. B. Keele, Jr., and R. B. Johnston, Jr.1974. Inhibition of phagocytosis-associated chemilumi-nescence by superoxide dismutase. Inzfect. liininun. 9:1051-1056.

21. Stanley, P. E., and S. G. Williams. 1969. Use of theliquid scintillation spectrometer for determining adeno-sine triphosphate by the luciferase enzyme. Anal. Bio-ch/ern. 29: 381-392.

22. Quie, P. G., J. G. White, B. Holmes, and R. A. Good.1967. In vitro bactericidal capacity of human polymor-phonuclear leukocytes: diminished activity in chronicgranulomatous disease of childhood. J. Clin. Invest. 46:668-679.

23. Johnston, R. B., Jr., and R. L. Baehner. 1970. Improve-ment of leukocyte bactericidal activity in chronic granu-lomatous disease. Blood J. Hcmnatol. 35: 350-355.

24. Curtiss, R. 1965. Chromasomal aberrations associatedwith mutations to bacteriophage resistance in Escheri-cihia coli. J. Bacteriol. 89: 28-40.

25. Johnston, R. B., Jr., P. Anderson, and S. L. Newman.1973. Opsonization and phagocytosis of Hcino phil/ts ini-efituen.ae, type b. In Hemophilus Influenzae. S. H. W.Sell and D. T. Karzon, editors. Vanderbilt UniversityPress, Nashville. 99-112.

26. Fridovich, I. 1970. Quantitative aspects of the pro-duction of superoxide anion radical by milk xanthineoxidase. J. Biol. Chcm. 245: 4053-4057.

27. Snedecor, G. W., and W. G. Cochran. 1967. StatisticalMethods. Iowa State University Press, Ames. 6th edi-tion. 125-127, 130-131, 162-163.

28. Allen, R. C., R. L. Stjernholm, and R. H. Steele. 1972.Evidence for the generation of an electronic excitationstate (s) in human polymorphonuclear leukocytes andits participation in bactericidal activity. Biochcin. Bio-phys. Res. Coninun. 47: 679-684.

29. Haber, F., and J. Weiss. 1934. The catalytic decomposi-tion of hydrogen peroxide by iron salts. Proc. R. Soc.Lond. A. Math. Phivs. Sci. 147: 332-351.

30. Neta, P., and L. M. Dorfman. 1968. Pulse radiolysisstudies. XIII. Rate constants for the reaction of hy-droxyl radicals with aromatic compounds in aqueoussolutions. Adv. Chicni. Scr. 81: 222-230.

31. Beauchamp, C., and I. Fridovich. 1970. A mechanismfor the production of ethylene from methional. Thegeneration of the hydroxyl radical by xanthine oxidase.J. Biol. Chen. 245: 4641-4646.

32. Fong, K.-L., P. B. McCay, J. L. Poyer, B. B. Keele,and H. Misra. 1973. Evidence that peroxidation oflysosomal membranes is initiated by hydroxyl free radi-cals produced during flavin enzyme activity. J. Biol.Checr. 248: 7792-7797.

33. Keele, B. B., Jr., J. M. McCord, and I. Fridovich.1970. Superoxide dismutase from Escherichia coli B.A new manganese-containing enzyme. J. Biol. Chrem.245: 6176-6181.

34. Beauchamp, C. O., and I. Fridovich. 1973. Isozymesof superoxide dismutase from wheat germ. Biochim.Biophys. Acta. 317: 50-64.

35. Nathan, D. G., R. L. Baehner, and D. K. Weaver.1969. Failure of nitro blue tetrazolium reduction in thephagocytic vacuoles of leukocytes in chronic granulo-matous disease. J. C/in. Invcst. 48: 1895-1904.

36. Khan, A. U. 1970. Singlet molecular oxygen fromsuperoxide anion and sensitized fluorescence of organicmolecules. Scicnzce (Wash. D. C.). 168: 476-477.

37. Khan, A. U., and M. Kasha. 1970. Chemiluminescencearising from simultaneous transitions in pairs of singletoxygen molecules. J. Amt. Chient. Soc. 92: 3293-3300.

38. Greenlee, L., I. Fridovich, and P. Handler. 1962.Chemiluminescence induced by operation of iron-flavo-proteins. Biochemistry. 1: 779-783.

39. Stauff, J., H. Schmidkunz, and G. Hartmann. 1963.Weak chemiluminescence of oxidation reactions. Nature(Lond.). 198: 281-282.

40. Arneson, R. M. 1970. Substrate-induced chemilumines-cence of xanthine oxidase and aldehyde oxidase. Arch.Biochcut. Biophys. 136: 352-360.

41. Finazzi Agr6, A., C. Giovagnoli, P. De Sole, L. Cala-brese, G. Rotilio, and B. Mondovi. 1972. Erythrocupreinand singlet oxygen. FEBS (Fed. EFur. Biochent. Soc.)Lett. 21: 183-185.

Superoxide Generation and Phagocytic Bactericidal Activity 1371

42. Goda, K., J. Chu, T. Kimura, and A. P. Schaap. 1973.Cytochovne c enhancement of singlet molecular oxygenproduction by the NADPH-dependent adrenoxin reduc-tase-adrenoxin system: the role of singlet oxygen indamaging adrenal mitochondrial membranes. Biochem.Biophys. Res. Commun. 52: 1300-1306.

43. Curnutte, J. T., and B. M. Babior. 1974. Biologicaldefense mechanisms: the effect of bacteria and serumon. sul)eroxicle production by granulocytes. J. Clin. In-vest. 53: 1662-1672.

44. Klebanoff, S. J., and C. B. Hamon. 1972. Role ofmyeloperoxidase-mnediated antimicrobial systems in in-tact leukocytes. J. Reticuloendothel. Soc. 12: 170-196.

45. McRipley, R. J., and A. J. Sbarra. 1967. Role of thephagocyte in host-parasite interactions. XI. Relation-ship between stimulated oxidative metabolism and hydro-gen peroxide formation, and intracellular killing. J.Bacteriol. 94: 1417-1424.

46. Klebanoff, S. J. 1967. Iodination of bacteria: a bac-tericidal mechanism. J. Exp. Med. 126: 1063-1078 +plate 80.

47. Baehner, R. L., S. K. Murrmann, and R. B. Johnston,Jr. 1974. The role of superoxide anion and hydrogenperoxide in phiagocytosis-associated oxidative metabolicreactions of human polymorphonuclear leukocytes.Blood. J. llematol. 44: 926 (Abstr.).

48. Goscin, S. A., and I. Fridovich. 1972. The role ofsuperoxide radical in a nonenzymatic hydroxylation.Arch. Biochemt. Biop/zys. 153: 778-783.

49. Krinsky, N. I. 1974. Singlet excited oxygen as a medi-ator of the antibacterial action of leukocytes. Science(Wash. D. C.). 186: 363-365.

50. Yost, F. J., and I. Fridovich. 1973. An iron-containingsuperoxide dismutase from Escherichia coli. J. Biol.Chem. 248: 4905-4908.

51. Beckman, G., E. Lundgren, and A. Tirnvik. 1973.Superoxide dismutase isoenzymes in different humantissues, their genetic control, and intracellular localiza-tion. Huim. Hered. 23: 338-345.

52. DeChatelet, L. R., C. E. McCall, L. C. McPhail, andR. B. Johnston, Jr. 1974. Superoxide dismutase activityin leukocytes. J. Clin. Invest. 53: 1197-1201.

53. Salin, M. L., and J. M. McCord. 1974. Superoxidedismutases in polymorphonuclear leukocytes. J. C/in.Invest. 54: 1005-1009.

54. Karnovsky, M. L. 1975. Biochemical aspects of thefunctions of polymorphonuclear and mononuclear leu-kocytes. In The Phagocytic Cell in Host Resistance.J. A. Bellanti and D. H. Dayton, editors. Raven Press,New York. 25-43.

55. Graham, R. C., Jr., M. J. Karnovsky, A. W. Shafer,E. A. Glass, and M. L. Karnovsky. 1967. Metabolicand morphological observations on the effect of surface-active agents on leukocytes. J. Cell. Biol. 32: 629-647.

56. Goetzl, E. J., and K. F. Austen. 1974. Stimulation ofhuman neutrophil leukocyte aerobic glucose metabolismby purified chemotactic factors. J. Clin. Invest. 53: 591-599.

57. Odajima, T. 1971. Myeloperoxidase of the leukocyte ofnormal blood. II. The oxidation-reduction reactionmechanism of the myeloperoxidase system. Biochim.Biophys. Acta. 235: 52-60.

58. Paul, B., and A. J. Sbarra. 1968. The role of thephagocyte in host-parasite interactions. XIII. The di-rect quantitative estimation of H202 in phagocytizingcells. Biochim. Biophys. Acta. 156: 168-178.

59. Baehner, R. L., D. G. Nathan, and W. B. Castle. 1971.Oxidant injury of Caucasian glucose-6-phosphate de-hydrogenase-deficient red blood cells by phagocytosingleukocytes during infection. J. Clin. Invest. 50: 2466-2473.

60. Root, R. K., N. Oshino, and B. Chance. 1974. Deter-minants of H202 release by human granulocytes(PMN). Clin. Res. 22: 402A (Abstr.).

61. Stossel, T. P., T. D. Pollard, R. J. Mason, and M.Vaughn. 1971. Isolation and properties of phagocyticvesicles from polymorphonuclear leukocytes. J. Clin.Invest. 50: 1745-1757.

62. Yamazaki, I., and L. H. Piette. 1963. The mechanismof aerobic oxidase reaction catalyzed by peroxidase.Biochim. Biophys. Acta. 77: 47-64.

63. Allen, R. C., and R. H. Steele. 1973. Electronic ex-citation of phagocytes. Proceedings of the 166th Na-tional Meeting of the Am. Chem. Soc., Biol. Sect. 116.(Abstr.)

64. Klebanoff, S. J. 1974. Role of the superoxide anion inthe myeloperoxidase-mediated antimicrobial system. J.Biol. Chem. 249: 3724-3728.

1372 Johnston, Keele, Misra, Lehmeyer, Webb, Baehner, and Rajagopalan