Role of Oxygen in Antibody-dependent Cytotoxicity Mediated by Monocytes and Neutrophils

The antibody-dependent cell-mediated cytoxicity (ADCC) by human monocytes and neutrophils was investigated by measuring the release of 51chromate from prelabeled erythrocytes coated with immunoglobulin G. ADCC was found to be positively correlated to phagocytosis of 51Cr-labeled erythrocytes and to the postphagocytic events of the effector cells, activation of the hexose monophosphate shunt, and degranulation. Exclusion of oxygen from the incubation media halved the ADCC by both cell types without affectijg phagocytosis or degranulation. Likewise, ADCC by cells from patients suffering from chronic granulomatous disease (CGD) was only half the intensity of ADCC by cells from normals. Inhibitors of mitochondrial respiration were without depressing effect of ADCC. Azide, which in addition to its blocking action on oxydative phosphorylation also inhibits catalase and myeloperoxidase, resulted in a approximately equal to 40% stimulation of ADCC by cells from normals but was without effect of ADCC by cells from CGD patients. The hydroxyl radical scavenger, mannitol, significantly depressed ADCC by cells from normals (P < 0.01) but was without effect on cells from CGD patients. Azide and mannitol also were without effect on ADCC by normal cells when oxygen was excluded. In a xanthine-xanthine oxidase system, erythrocytes were effectively lysed. This lysis was inhibited by catalase, superoxide dismutase, and mannitol. When comparable concentrations of glucose oxidase were used no lysis was observed. H2O2 either alone or in combination with azide did not lyse erythrocytes. It is suggested that ADCC by both monocytes and neutrophils is partly dependent on the generation of hydroxyl radicals by the effector cells.

A B S T R A C T The antibody-dependent cell-mediated cytoxicity (ADCC) by human monocytes and neutrophils was investigated by measuiriing the release of 'chromate from prelabeled erythrocytes coated with immunoglobulin G. ADCC was found to be positively correlated to phagocytosis of 51Cr-labeled erythrocytes and to the postphagocytic events of the effector cells, activation of the hexose monophosphate shunt, and degranulation.
Exclusion of oxygen from the incubatioin media halved the ADCC by both cell types without affecting phagocytosis or degranutilation.Likewise, ADCC by cells from patients suffering from chronic granuloma- tous disease (CGD) was only half the intensity of ADCC by cells from normals.Inhibitors of mito- chondrial respiration were without depressing effect on ADCC.Azide, which in addition to its blocking action on oxydative phosphorylation also inhibits catalase and myeloperoxidase, resulted in a -40% stimulation of ADCC by cells from normals but was without effect oIn ADCC by cells from CGD patients.The hydroxyl radical scavenger, mannitol, significantly depressed ADCC by cells from normals (P < 0.01) but was without effect on cells from CGD patients.Azide and mannitol also were without effect on ADCC by normal cells when oxygen was excluded.
In a xanthine-xanthiine oxidase system, erythrocytes were effectively lysed.This lysis was inhibited by catalase, superoxide dismutase, and mannitol.When comparable concentrations of gltucose oxidase were tused no lysis was observed.H202 either alone or in combination with azidle did not lyse erythrocytes.
It is suggested that ADCC by 1)oth monocytes and neutrophils is partly dependent on the generation of hydroxyl ra(licals by the effector cells.

INTRODUCTION
Antibody-dependent cell-mediated cytotoxicity (ADCC)' refers to the lysis of target cells sensitized with IgG antitarget cell antibody by effector cells.Using antibody-sensitized human erythrocytes as target cells, the process is mediated by human mono- cytes andi neutrophils (1)(2)(3).
In concert with the cytolysis, phagocytosis of the IgG-coated target cells takes place (4, 5).The mechanisms producing the cytotoxic effect are unclear.Stable soluble cytotoxic mediators are not involved (3, 4).However, cytotoxic mediators that might require high local concentrations in regions of intimate membrane apposition cannot be excluded (6).Recent observations by Clark and Klebanoff (7) and Katz et al. (8) indirectly implicate a role of reduced oxygen radicals generated by the effector cells because ADCC was fouind significantly depressed in effector cells from patients suffering from chronic granulomatous disease (CGD).CGD is characterized by the lack of ability of neutrophils to form reduced oxygen radicals upon challenge despite normal phagocytosis (9).However, Fleer et al. (10) foundl normal ADCC in cells from CGD patients and found no evidence of oxygen radicals participating in ADCC.
The present investigation was carried out to delineate the role of oxygen ra(licals in ADCC.In view of the fre(Ituent infections of CGD patients, normal cells were investigated uinder aerobic and anaerobic conlditions becatuse the sole use of CGD patients to in- vestigate the role of oxygen radicals might introduce a bias in the methodology.Cells from patients anid I Abbreviations used in this paper: ADCC, antibodydepen(lent cell-mediated dlisease; CGD, chronic granulomatotns disease; FCS, fetal calf seruim; HBSS, Hanks' balanced salt solutioni; HMPS, hexose monophosphate shunt; IgG-latex, IgG-coated latex particles; LDH, lactate dehydrogenase; p)atients wvere i(lenitifiedI by their neutr-ophlils having (a) abolished killing of staphylococci de spite i)ormail ph)agocytosis, (b ) abolished activation of oxygen costiuption and hexose monophiosphate shtint upon stimtilaitioni by PMIA alnd( IgG-coated latex particles, and (c ) abolished superoxide liberation upoin PMIA stimutilationi.Carriers wvere eithlel- mothers or sisters of affecte(d bovs ancd had at clearly reduticed but not abolished response in the above mentionecl paranmeters.

Chroiiiiuimi release was clctulatte(d according to the expressioii:
C cpn)1l x 2 x 100( % 51 Cr releaise = C cpnm + R cpiii For eachl reaction the specific release wlas oh)tainedl 1b subtracting the release iii the coiitrol tube's, wvhich never exceedeed 2%.In some experimentts, results xwere expresse(d as the niumber of target cells lysed per effector cell: Number of target cells Iysed = (number of target cells x specific release) (number of effector cells x 100).
Phagocytosis was expressed as the percentage oftarget cells protecte(d froni complement-mediatec lysis: Phagocytosis = (R eprn -P epm) x 100 R epm + P epiim Anaerobic conditions.Buffers xvere evacuatedl undler zero pressure overniight in a vacuum chamber.An atmosphere of nitrogen was then introdcuced to the vacuum chamber andc the bottles containing buffers were capped.Nitrogen that had percolated through a saturated dithionite solution was then bubbled through the buffers for 1 h.
Both target and effector cells were washed once in anaerobic mediium.The anaerobic state was mainitained by blowing nitrogen into the tubes when the caps were removed.There- after, the cells were put into a couveuse, 0.15 m3 vol, together with all the utensiles needed for the experiment.Nitrogen was then continuously blown into the couveuse anc after 30 min, when 25 m3N had passed through the couveuse, the experiment was started.
The 02 concentration in the buffers was measured after ending the experiments.It was 6.3 ±+2.3 ,uNM (mean±SEM for six experiments).
Degrranulation.Assay mixtures were incubated as de- scribed for the cytotoxicity assay.At time indicated, 200 ,ul of the suipernate was withdrawni for the assay of lysozyme and lactate dehy(lrogeniase (LDH).Total lysozyme and LDH were measured after ultrasonic disruption of unstimu-tlated cells in ().2% Triton X-100.Lysozyme was meassured turbidimetrically by the lysis of Micrococcus lysodeikticus (14).LDH was measured as described (15).Liberation was given as percentiage of total amount of enzyme in unlstimulated cells.
Hexose monophosphate shunt (HMPS).14CO2 liberation from [ 1_14C]-and [6-'4C]glucose durinig the ADCC reaction was measuired by incubating the cells in the preseinee of 1 mM labeled glucose, sp act 2.2 mCi/mmol.The reaction was termliinated by injecting 100 ,ul 1 N NaOH through the caps into the reaction mixture.After 1 h the contenit of the tubes was withdlrawn and transferred to conicatl flasks in which 8()() ,ul 1 N HCI had been placed.A rubber cap immediately sealed the flasks and the killed reatetion mixture and the 1 N HCI were mixedl to drive out the dissolved CO2.A center well of 1 cm Diaim containe(l 400 A(l 1 N NaOH for the absorption of CO2.After standing overnight on a shaking bath, the conteint of the center well was withdratvwi andn counted in 10 ml Instagel.02 consumption during phagocyto.si.s of latex particles.02 consuimption was meassured in at rapidly stirred thermo- statize(l 37°C chamber with a Clark-type oxygeni electrode built in under the stirrer (Rank Bros., Bottishamii, England).
Sunperoxide production.PMA stimulslationi of stiperoxide prOdltiction was measured as descril)edl ( 17) by following the increase in absorbance at 550 i1nm of aw cell suispenision con- tatining 0.2 mM cytochrome c.Statistical analysis.Statistical significance was assessed utsinig Wilcoxon's test for paired observations.A P valtue < 0.05 wats considered significant.

RESULTS
ADCC and phagocytosis.Normall moniocytes as well ats normal netutrophils were effective in lysing imm-inlInoglolbul in-coated erythlrocyte s, the neuitroph ils being slightly more effective thaln the monocytes (Fig. IA anid B).
[6-14C]gltucose and the specific liberation of lysozyme (lysozyme vs. LDH), respectively.The timle-couirses of lysis, HMPS activity, lysozyme liberation, aindl phagocytosis (Fig. IA and B) showedl thsat these plhenlomiiena are accurately correlated.The number of target cells lysedl per eff'ector cell dlepenids on the target/effector cell ratio; ats seeni in Fig. 2A and B, the increasing lysis by the effector cells was accompanied by increased phagocytosis, increased HMPS activity, and increased lysozyme liberation.Thus, it is apparent that ADCC in this system is correlated to phagocytosis and to the post phagocytic events (HMPS activation and degranulation)..10X103 .5x103 Effect of varying the effector/target cell ratio on cytolysis, phagocytosis, and postphagocvtic events.In A the effector cells were monocytes and in B neutrophils.The time of incubation was 30 min.The concentration of effector cells was 106 cells/ml.Cytolysis (0), phagoeytosis (-), couints per minute of 14CO2 from [1-'4C]glucose (*), lysozyme liberation (M).Each figure is the mean of six individual experiments.SEM: for cytolysis s 0.8%, for phagocytosis s 1.2%, for lvso- zyme s 2.3%, for '4CO2 s 200 epm.
ADCC and CGD.A study of the ADCC by m0o1ocytes aind polymorphs from patients ancd carrier-s of CGD was carried ouit. 02consumiiiptioni of PMA- stimulatedl neuitrophils and stuperoxide prodltctioin of PMA-stinitlated monocytes an(cl neuitroplhils wTere com11- pared with ADCC (Table I).It is observed that monocytes and neuitroplhils slhare the dlefective response in postphagocytic metabolism as seeni fromii °2 prodcuction, that ADCC is severely depressed l)ut not ab)olished in patienits stuffering from CGD, and(i that the degree of depression is coiTelated to the (lepression of oxygen constumption and superoxide producttioni.Degranuilatioin anCl phagocYtosis by effector cells from CGD patients and carriers were not (lepresseci (data not given), showinig that the dlefective ADCC 1y these cells is not secondary to defective target-cell effector-cell contact.
These findings indicate that the formcationi of reduced oxygen radicals, which normiially takes place duiring phagocytosis by imonocytes andl netutroplhils, i s of importance for the ADCC bY these cells.
This observation is substantiated IY the findiing thalt the ADCC bvy normiial monocytes and(I neuitrophils becomes almost as e(qually (lepressed und(ler anaerol)ic conditions as the effector cells from CGD paitients are under normal aerobic conditions (Table II).It was seeni that, after readmission of oxygen, the cvtotoxicity of monocytes and neuitrophils returnedl to normiial (cf.legendl to Table II).The finding that phagocytosis aind lysozyme liberation were uinchanged uinder anaerol)ic conditions indicates that the anaerobic dlepressioni of ADCC is not cauised by (lefective conitact l)etween target and effector cell or by a general (lepressioni of the energy metabolism of the effector cells.This is in agreement with the findinig (Table III) that inhibitors of the oxydative phosphorylation, cyanide, azide, a.ndl 2,4-dinitrophenol, did not inhibit ADCC by any of the cell types investigated.The slight inhibition of mlono- cyte-m1edliatedl ADCC by an1tim11ycin1 A wvas eq(Iually expresse(d uinder anaerobic as tinder aerol)ic condCIitionis (P > 0.10) and thus was not the resuilt of inhibition of mitochondrial oxvydative phosphorvlatioin.Neuitroplhils were not influeneedl 1v antimivcin A.
To identify the oxygen radlicals that are of' sig- nificanice for ADCC, the effect of' scavengers of superoxide anion, hydroxyl radicals, andcl hydrogen peroxide, and the effect of inhibitors of catalase aindl myeloperoxidase were investigated (Table III).
No effect of catalase, cvtoclhrome c, or superoxide dismuitase was observed in anyv of the parameters investigated.As seen, azide, which in addition to block- ing the mitochondlrial oxycdative phosphorylation also blocks the activity of catalase anid mveloperoxidase and, thus, allows H202 to accuminulate, stimultltedl the ADCC of 1)oth cell types investigated w%ithout affectinig phagocytosis or lysozyme liberation.The (latat f'or Oxygen-dependent Cytotoxicity by Montocytes antd Neutrophil.sX is a boy sufferinig from the X-linked form of CGD.§ A is a girl sufferinig fromii the autosomal recessive form of CGD. lysozyme liberation are not shown.The stimulating effect of azide was neither observed on normal cells under anaerobic conditions nor on cells from CGD patients (data not shown).The hydroxyl radical scavenger, mannitol, considerably reduced the ADCC without affecting phagocytosis or lysozyme liberation.This effect of mannitol was not observed tinder anaerobic conditions, nor did mannitol reduce the ADCC by CGD patients (data not shown).This shows that the effects of mannitol in the concentration used is not a result of depression of the energy metabolism of the effector cells.It was ascertaiined that mannitol in the concentratioin uised did not influence the rate of IgG-latex-induced oxygen constumption by normal monocytes or neutrophils (data not showin) and, thus, does pot interfere with the generation of reduced oxygen radicals by these cells when uindergoing phagocytosis.
Table IV shows that lysis of target cells is efficiently carried ouit by 0.025 U/ml xanthine oxidase.Omission of either xanthine or xanthine oxidlase abolished the lysis.It is seen that the lysis is almost totally inhibited Incubations were for 30 min.Effector and target cells were made anaerobic and usedl for both the aerobic and anaerobic experiments.For the aerobic experiment, cells were made aerobic by a buffer shift after standing under anaerobic conditions when the anaerobic experimenlt was carried out.by catalase ancd mannitol, aindl partly b)y stuperoxidle dismutiase.When suiperoxidle di smutitase a(In catalase, which hlad been heated to dlestroy enizymatic activity, were added, no inhibitory effect on cell lysis was oh- served.It wats ascertained that 50 mM miannitol did not Incubations were as in ADCC experimenits except that effector cells were replaced by 0.025 U/ml xanthine oxidase and 1 mM xanthinie, and that cenltrifugation of the reactioni mixtture before incubation was omitted.The time of inicubationi was 1 h.Data are mean±SEM, of three experiments.Boiled catalase anid boiled superoxide dismiiutase are enzymes that have been heated at 95°C for 5 min to destrov enzymatic activitv. interfere witlh oxygen coinsumption of the xainthliniexanthine oxidase system (data not shown).
A gluicose-glucose oxidase system, 0.05 U/mIl, was not able to lyse target cells either at pH 5.8 or 7.4 uinider the sanme conditions as uised in the xanthine oxidase system (cf.legend to Table IV).Hydrogen peroxide, 5 ,uM and 200 ,uM, either alone or in combination with 1 mM azide likewise did not result in lysis of target cells wheni incubated for 1 h.

DISCUSSION
The idea that oxygen rad(licals may be involved in ADCC by m11on1ocytes andnIeUtrophils dOeS not a priori seenm remote.Both monocytes and neuitrophils are known to re(dtice oxygen in a cyanide-insenisitive wavy for the productioni of superoxide and hydrogell peroxide when the cells are stimtulated to phagocytose (18-21), annd it is seen from our results that the process of ADCC, phagocytosis, and postphaigocytic events are intimately correlated, allthouigh no ca'usal relation mav be deduced friom these results.
From ouir observation that ADCC by humani mono- cytes anid neuitrophils is -50% reduced when oxygen is removed, whereas phagocytosis and degrantulation are uinaffected by this treatment, it seems safe to concltude that oxygen plays a role for ADCC other than what might be secondary to oxydative phosphorylation in the mitochondria.In falet, ouir finding that phagocytosis and degranulation are uinaffected by anaerobiosis and by the lack of effect of eyanide and dinitrophenol on ADCC, rules ouit atny vsignificant demanid of energy provided by oxydative phosphorylation for the ADCC and phagocytosis by both monocytes and neutrophils.This is confirmed by the finding that oxydation of carl)on in position six of glucose to CO2 is not enhanced during ADCC, which shows that the activity of the citric acid cycle is not increased during ADCC.The observation that ADCC by 1)oth monocytes and neutrophils from patients with CGD is halved com- pared with cells from normals gives further support to the idea that oxygen radicals play an essential role in ADCC.This observation has also been made by Clark and Klebanoff (7) investigating neutrophils and by Katz et al. (8) investigating both monocytes and neutrophils from CGD patients, although the (legree of depression observed in CGD patients in these studies was somewhat higher than those observed by us.The results are, however, contradictory to those of Fleer et al. (10), who found that ADCC was totally dependent on the release of lysosomal enzymiies.The basis for this divergence remains obscure.
Further support for the idea of redcuced oxygen species being of significance in ADCC is gained by the observation that mannitol inhibits ADCC in concentra- tions that do not affect phagocytosis or degranulation durinig the ADCC reaction.The finding that mannitol is without effect on ADCC by cells from CGD patients and during anaerobic incubation of cells from normals strongly implies that the effect of mannitol is a result of its action as a hydroxyl radical scavenger (22).It has been observed that hydroxyl radicals are formed by stimulated monocytes and neutrophils (23,24).In recent studies by Green et al. (25) and Rosen and Klebanoff (26) using the technii(lue of electron spin resonsance spectroscopy, evidence for significant gen- eration of hydroxyl radicals by phagocytosing neutro- phils was found.It was seen that azide stimulated the formation of hydroxyl radicals andl that the pro- dtiction of hydroxyl radicals was higher by cells deficient in myeloperoxidase than by normal cells.This corresponds well with our observation that azide stimulates ADCC and with the observation of Clark and Klebanoff (7) that ADCC by myeloperoxidase-deficient neutrophils is enhanced.SOD, which dramatically reduces the ESR response of hydroxyl radicals (25,26), did not influence ADCC in our system.This has also been observed by Clark and Klebanoff.Recently, Capsoni et al. ( 27) have reported some depression of SOD on ADCC by neutrophils, although the level reached was far higher than ADCC by neutrophils from CGD patients.The lack ofeffect of SOD on ADCC does not rule out the possibility that hydroxyl radicals actively participate in ADCC because ADCC is dependent on intimate membrane contact between effector cell and target cell (4); it is highly likely that SOD, cytochrome c, and catalase do not gain access to the site of formation and action of hydroxyl radicals.
The involvement of hydroxyl radicals in target cell lysis gains further support from the observation that target cells are effectively lysed by xanithine-xanthine oxidase.This shows that lysis can be obtained by the radicals produced in this system (°-, H202, an(l OH, 28) and, thus, that ly'sosomal enzymes delivered by effector cells are not a necessary prerequisite for lysis of erythrocytes.The observation that the lysis of target cells by xanthine oxidase is abolished b)y catalase rules out that °-is the direct effector molectule.On the other hand, the strong depression of lysis observed by super- oxide dismutase, which accelerates the formation of H202, shows that lysis is not causedl by H202 dlirectly.This is also in accordance with the lack of lysing effect of glucose oxidase and H202.These observations together with the strong inhibitory effect of mannitol indicates that it is the hydroxyl radicals generated by xanthine-xanthine oxidase that are effector molecules.
FIGURE 2 Effect of varying the effector/target cell ratio on cytolysis, phagocytosis, and postphagocvtic events.In A the effector cells were monocytes and in B neutrophils.The time of incubation was 30 min.The concentration of effector cells

TABLE I
Correlation between ADCC, Oxygen Consumption, and Superoxide Liberation tion of neutrophils (3 x 106 cells/miil) stimiutilate( wvith 5.0 /ug/mli PMA (values fromii unstimulated cells have been suibtracted).0°liberation ofneuitrophils (106 cells/ml) andc mono- cytes (3 x 106 cells/ml) both stimulated with 5.0 ,ug/ml PMA, measured by the reducttion of a 0.2-mM cytochrome c soltution.Cells were incubated for 30 min for the determiinlation of cytotoxicity and phagocytosis.Figures for normals are meain±SEM for eight individuals.Figures for the patients are mean of two closely agreeing measuremenits.

TABLE II Effect
of Aniaerobiosi.soni CyJtotoxicitij of Monioci tes' and Nentrophils