Antibody-Dependent Cellular Cytotoxicity inPrimary Immunodeficiency Diseases and withNormal Leukocyte Subpopulations

IMPORTANCEOF THE TYPE OF TARGET

S. OZDENSANALand REBECCAH. BUCKLEY, Departments of Pediatrics andMicrobiology and Immunology, the Duke University School of Medicine,Durham, North Carolina 27710

A B S T R AC T To gain insight into a possible role forantibody-dependent cell-mediated cytotoxicity in vivo,we examined the ability of leukocytes from 28 pa-tients with primary immunodeficiency and from 20normal controls to lyse three different types of anti-body-coated targets in vitro. Mean cytotoxic indices± 1 SD elicited by unfractionated mononuclear cellsfrom normal controls were 28.74±13.26 for humanHLA antibody-coated lymphocyte targets, 42.79±8.27for rabbit IgG antibody-coated chicken erythrocytetargets, and 47.58±10.34 for human anti-CD (Ripley)-coated O+ erythrocyte targets. Significantly (P= <0.05) lower than normal mean cytotoxic indicesagainst lymphocyte targets were seen with effectorcells from 10 patients with X-linked agammaglobuline-mia (3.7±4.33), in 10 with common variable agamma-globulinemia (16.05±7.74), in 3 with immunode-ficiency with hyper IgM (18.41±4.88), and in 2 withsevere combined immunodeficiency (3.94±0.3). Anti-body-dependent cytotoxicity against chicken erythro-cytes was significantly (P = <0.05) lower than normalonly in the common variable agammaglobulinemicgroup (33.33±12.3) and against human erythrocytesonly in the common variable (34.36±9.59) and hyperIgM (27.54±0.66) groups. Rosette and anti-F(ab')2depletion studies with normal leukocytes indicatedthat a nonadherent, nonphagocytic, non-Ig-bearing,non-C receptor-bearing, Fc receptor-bearing lympho-cyte was the only effector capable of lysing HLAantibody-coated lymphocyte targets. Patients with in-

Dr. Sanal's present address is Hacettepe University,Ankara, Turkey. Dr. Buckley is the recipient of AllergicDiseases Academic Award Al 70830-04.

Received for publication 6 June 1977 and in revised form.22 August 1977.

fantile X-linked agammaglobulinemia and severe com-bined immunodeficiency appear to have a markeddeficiency in this type of effector cell function.

INTRODUCTION

Antibody-dependent cell-mediated cytotoxicity isdefined as the lysis of IgG antibody-coated targetcells by nonimmune leukocytes bearing membranereceptors for the Fc portion of IgG (1). Althoughthere is evidence that the reaction can be enhancedby the presence of C3b on the target cells (2), thisphenomenon can occur in the absence of complement(3). A variety of leukocytes bear receptors for the Fcportion of IgG, including B (4), K (4, 5), and T (6)lymphocytes, monocytes (7), and polymorphonuclearcells (8); several of these also have membrane re-ceptors for C3b, namely bone marrow-derived (B)lymphocytes (9), monocytes (9), and polymorpho-nuclear cells (8). Considerable controversy exists overthe capability of these various cell types to serveas effectors in antibody-dependent cellular cyto-toxicity. Recent studies suggest that effector cells aredifferent for different types of targets (10, 11).

The role of antibody-dependent cell-mediated cyto-toxicity in vivo is uncertain, but it is thought to be ofpossible importance in graft (12) and tumor (13) rejec-tion, in certain autoimmune reactions (14), and inhemolytic disease of the newborn (15). There is littleor no information as to its role in host defense,and conflicting results have been reported regardingthe capacity of leukocytes from patients with immuno-deficiency diseases to serve as effectors in this type ofcell lysis (16-18).

In the studies reported here, we attempted to gainfurther information on the possible role of antibody-

The Journal of Clinical Investigation Volume 61 January 1978 -1 -10 1

dependent cellular cytotoxicity in vivo by examiningthis function in patients with various primary immuno-deficiency diseases against three different types of anti-body-coated targets. In addition, by using isolatedleukocyte subpopulations from normal hosts, wesought to define the types of cells capable of servingas effectors against these three targets.

METHODS

Patient population. This consisted of 28 patients withwell-defined primary immunodeficiency (19), including 10with X-linked agammaglobulinemia, 10 with commonvariableor B-lymphocyte agammaglobulinemia, 3 with X-linkedimmunodeficiency with hyper IgM, 2 with partial DiGeorgesyndrome, 2 with severe combined immunodeficiencydisease, and 1 with chronic mucocutaneous candidiasis.The clinical, immunologic, and lymphocyte membrane andfunctional characteristics of many of these patients have beenreported (20, 21); similar criteria were used to definethose not described previously. Normal controls consisted of20 healthy adults.

Isolation of mononuclear cells. Lymphocytes and mono-cytes were separated from heparinized venous blood by amodification of the method of Boyum (22). The blood wasdiluted fourfold in 0.95% sodium chloride, underlayeredwith 10 ml of Ficoll-Hypaque, having a specific gravity of1.078, in 50-ml sterile conical plastic centrifuge tubes (Corn-ing Glass Works, Science Products Div., Corning, N. Y.),and centrifuged for 30 min at 400 g at room temperature.The cells recovered at the interface represented greaterthan 80% of the mononuclear cells in the whole blood andconsisted of lymphocytes predominantly, with 4-29% mono-cytes and less than 1% granulocytes.

Identification of lymphocyte subpopulations. Fc recep-tor-bearing cells were detected by their ability to formrosettes with human erythrocytes sensitized with eitherhuman IgG antibodies derived from Ripley anti-CD serumor rabbit IgG antibodies to human type A erythrocytes.Unless otherwise indicated, Ripley-coated cells were usedthroughout the study as the antibody-coated erythrocyte indi-cator, but rabbit IgG-coated human type A erythrocytes wereused for this purpose in rosette-depletion studies. 1 vol ofpacked human 0+ erythrocytes in 8 vol of saline wasincubated at 37°C for 2 h with 1 vol of Ripley serum, washedfour times, and then adjusted to 0.5% with phosphate-buffered saline containing 2% fetal calf serum. Rabbit IgG-coated cells were prepared by mixing equal volumes of a 2.5%suspension of human type A erythrocytes and a subag-glutinating dilution of rabbit IgG anti-A antibody, incubat-ing them for 30 min at 370C, washing them four times,and then resuspending them to 0.5% with phosphate-bufferedsaline containing 2% fetal calf serum.

Lymphocytes with receptors for complement, those withsurface IgM and (or) IgD, and those forming spontaneousrosettes with sheep erythrocytes were identified andenumerated as described previously (20). In addition,neuraminidase-treated sheep erythrocytes were employed todetect thymus-derived (T) lymphocytes and were preparedas follows: 2 Al of neuraminidase (Behring Diagnostics,American Hoechst Corp., Somerville, N. J.) was added to 1 mlof a 1% sheep erythrocyte suspension in 4 mMbicarbonate-buffered saline containing 3 mMCaCl,/liter and incubatedfor 20 min at 37°C. The cells were then washed twice andadjusted to 0.5% for the rosetting studies.

After the above reagents were prepared, they were each

mixed with an equal volume (2 drops) of various mono-nuclear cell suspensions ranging in concentration from 2 to4 x 106/ml in the wells of round-bottom microtiter plates andcentrifuged for 6 min at 800 rpm at room temperature.Rosettes formed with erythrocytes coated with antibody orantibody plus complement were read in 0-6 h and sheeperythrocyte rosettes in 1-6 h (20). All studies were done intriplicate, and at least 400 cells were counted per well.

Identification and elimination of monocytes. Monocyteswere identified by staining for myeloperoxidase activity, usingthe method of Kaplow (23), and by their ability to ingestlatex particles (24). For the latex ingestion studies, 10-20x 106 mononuclear cells in 1 ml of phosphate-buffered salinecontaining 20% fetal calf serum were incubated with 10 ,ulof latex particles (size 0.3 gm, Dow Chemical Co., Midland,Mich.) for 20 min at 37°C on a rotating tumbler.

To eliminate monocytes, heparinized blood was eitherdripped undiluted through a nylon column (0.7 g of nylonfiber packed in a 6-in Pasteur pipette) at a rate of ap-proximately 1 drop/s or diluted to contain approximately3 x 106 leukocytes/ml and incubated with carbonyl iron at aconcentration of 4 mg/ml at 37°C for 20-30 min on a rotatingtumbler. After a few minutes were allowed to elapse forsedimentation of the carbonyl iron, the blood was thenlayered on Ficoll-Hypaque and the iron-laden monocyteswere separated by centrifugation.

Depletion of rosette-forming cells by gradient centrifuga-tion. Rosettes of the various types were prepared asdescribed above, except that equal volumes of mononuclearcells (2-4 x 106/ml) and 1% suspensions of the various typesof erythrocytes were mixed in 15- or 50-ml centrifuge tubesand then subjected to Ficoll-Hypaque gradient centrifugation.Lymphocytes were removed from the interface and washed.The sheep erythrocyte rosette pellets were incubated withrabbit anti-sheep erythrocyte hemolysin (Difco Laboratories,Detroit, Mich.) and fresh normal human serum as a sourceof complement to lyse the red cells; rabbit IgG-coatedhuman A cell rosette pellets were incubated with pony anti-rabbit IgG and human complement at 37°C for '/2 h tolyse the erythrocytes.

Removal of immunoglobulin-bearing cells by anti-F(ab'),column depletion. This procedure was performed accordingto the method of Chess et al. (25) with rabbit anti-F(ab')2antibody.

Targets for antibody-dependent cell-mediated cytotoxicity.For these studies, three different types of targets werelabeled with Na2Cr5"04. The first of these was HLA antibody-coated normal human lymphocytes. Platelets and adherentcells were removed from 10 ml of heparinized blood bypassing the blood through a nylon column as describedabove; the lymphocytes were then obtained by Ficoll-Hypaque gradient centrifugation and treated with 2 ml ofTris-NH4Cl to lyse any remaining erythrocytes. A suspensionof 6-8 x 106 lymphocytes in 0.5 ml of RPMI 1640 mediacontaining 20% fetal calf serum was labeled with 100 ACiof Na2Cr5"04 at 37°C for 2 h in a humidified 5% CO2 at-mosphere with frequent shaking. After being washed fourtimes, the cells were adjusted to a concentration of 1 x 106/ml in RPMI 1640 containing 20% fetal calf serum.

The second type of target was rabbit IgG antibody-coatedchicken erythrocytes. To prepare the labeled cells, a sus-pension of 20 x 106 fresh, thrice-washed chicken erythro-cytes in 0.5 ml of RPMI 1640 containing 20% fetal calf serumwas labeled with 100 dxCi Na2Cr5"04 at 37°C for 1 h withfrequent shaking. The cells were then washed four times andadjusted to 5 x 106/ml.

The final type of target was IgG-coated human erythrocytes.Fresh type 0 Rh+ erythrocytes were washed three times and

2 S. 0. Sanal and R. H. Buckley

labeled with 100 ACi of Na2Cr5'O4 in the same manner as forchicken erythrocytes. One-half of the labeled erythrocytesuspension was sensitized with Ripley anti-CD IgG antibody,as described for preparing the rosette indicator; the cells werethen washed and adjusted to 5 x 106/ml. The other half ofthe suspension served as the non-antibody-coated controltarget.

For all targets, the last washes were done just before usein the cytotoxicity assays.

Antibody-dependent cell-mediated cytotoxicity assay.The general features of the assay were modified fromTrinchieri et al. (26). When lymphocytes were used as targets,10 al of either an optimally diluted heat-inactivated humananti-HLA antiserum or RPMI 1640 containing 20% fetal calfserum were put together with 10 ,lp of the 5'Cr-labeled cellsuspension containing 104 lymphocytes into each of thetriplicate wells of a round-bottom microtiter (Linbro) plate.The HLA specificities of the targets were always selected tobe different from the specificities of the effector cells.When chicken erythrocytes were used as targets, 10 ,ulof heat-inactivated rabbit anti-chicken erythrocyte antiserumor medium was added to 10 ,ul of the 51Cr-labeled cellsuspension containing 5 x 104 erythrocytes in each oftriplicate wells. The mixtures were incubated for 30 minat 4°C. For human erythrocyte targets, 5 x 104 Ripleysensitized or unsensitized 5'Cr-labeled erythrocytes wereadded to the wells with no incubation. Effector cellswere then added to the various wells in 100-,ul aliquots,each containing 106 cells. Thus, the effector to target cellratios were 1:100 for the lymphocyte targets and 1:20 for thetwo erythrocyte targets. These ratios were found to be optimal

in preliminary studies with normal control leukocytes. Controlwells for all experiments, again in triplicate, included thosewithout effector cells and those without antiserum. The plateswere incubated for 4 h at 37°C in a humidified 5% CO2 at-mosphere; then 30 ,ul of phosphate-buffered saline wereadded to each well. The plates were then centrifuged at 400 gat 4°C for 10 min. Supernatant radioactivity was then deter-mined after carefully removing 50 ,ul from the side of eachwell with an Eppendorf pipette and disposable tip (Brink-mann Instruments, Inc., Westbury, N. Y.) and counting thealiquots in a gammaspectrometer (Beckman Instruments Inc.,Fullerton, Calif.). The counts per minute for each wellwere multipled by 3 to obtain total release. The mean countsper minute, standard deviation, and standard error were de-termined for each triplicate set and the percentage ofspecifically released 5'Cr was calculated as follows:

Experimental release - spontaneous releaseTotal releasable cpm - spontaneous release

= Cytotoxic Index

Only cytotoxic indices of 10% or greater were consideredpositive.

RESULTS

Antibody-dependent cytotoxicity by normal unfrac-tionated mononuclear cells. The mean cytotoxic in-dices obtained when normal non-monocyte-depletedFicoll-Hypaque-purified mononuclear cells were

TABLE IMean Antibody-Dependent Cellular Cytotoxicity Indices* for Normal Controls

and Immunodeficient Patients with Three Different Types of Target Cells

Lympht CRBCt HRBC$

Subjects nit I n t CIt nt CIt

Normals 10 28.74 14 42.79 11 47.58±13.26 +8.27 ±13.26

X-linked 10 3.70§ 9 44.17 9 41.15agammaglobulinemia ±4.33 ± 10.56 ± 10.34

Commonvariable 10 16.05§ 7 33.33§ 9 34.36§agammaglobulinemia ±7.74 ± 12.3 ±9.59

X-linked hyper IgM 3 18.41§ 3 42.92 3 27.545±4.88 +8.18 ±0.66

DiGeorge 2 41.43§ 1 45.3 1 33.60±5.43

Severe combined 2 3.94§ 1 52.36 1 25.59immunodeficiency ±.30

Chronic mucocuta- 1 23.94 1 1neous candidiasis

* ±1 SD.t Lymph = HLA antibody-coated lymphocyte target; CRBC= chicken erythrocytescoated with rabbit IgG; HRBC= human erythrocytes coated with Ripley IgG;n = number; CI = cytotoxic indices.§ Value significantly different (P = < 0.05) from normal mean when compared inStudent's t test.

Antibody-Dependent Cellular Cytotoxicity in Immunodeficiency 3

tested against the three different types of target cellsare presented in Table I and shown in Fig. 1. Whennormal rabbit or human sera were used instead of thevarious target-specific antisera, no significant 51Cr re-lease was observed. The coefficients of variation forday-to-day variability in the cytotoxic indices for thethree types of targets, as determined using a singledonor's effector cells on 14, 6, and 8 days againstlymphocyte, chicken erythrocyte, and human erythro-cyte targets, respectively, were 22.2, 21, and 31.5%.

Antibody-dependent cytotoxicity by mononuclearcells from immunodeficient patients. Whenantibody-dependent cytotoxicity was examined in 10 patientswith X-linked agammaglobulinemia and in 10 withcommon variable agammaglobulinemia, unfractionatedmononuclear cells from both patient groups lysed thetwo erythrocyte targets to a significant degree (TableI and Fig. 1). In contrast, mononuclear cells from theX-linked agammaglobulinemics did not cause signifi-cant 51Cr release from HLA antibody-coated lympho-cytes, whereas cells from patients with the commonvariable form did. It should be noted that cells from anumber of individual common variable agamma-globulinemic patients had normal cytotoxic activityagainst all three types of targets, whereas only one ofthe X-linked patients had a cytotoxic index (11.28)greater than 10%against lymphocyte targets. Whenthegroup means were compared with those from the nor-mal subjects by Student's t test, however, the meansfor the common variable group were significantly (P

80 Lymph CRBC . HRBC

70

60-

50-

40-'4

20'-10

Normals XAy CVAr Normols XAr CVAY Normols XAr CVAYln=io}(n0lfl (nl=0 (n=14) (n= (n=7) (nfl (ll n=)I (n=f

FIGURE 1 Comparison of antibody-dependent cellular cyto-toxicity against three different antibody-coated targets bymononuclear cells from X-linked (XA-y) and commonvariable(CVAy) agammaglobulinemic patients and normal controls.Lymph refers to human HLA antibody-coated lymphocytetargets, CRBC to rabbit antibody-coated chicken erythro-cytes, and HRBC to Ripley anti-CD-coated human 0+erythrocyte targets. The dots represent the cytotoxic indicesfor individual patients, the solid horizontal lines the groupmeans, and the dashed lines +1 SD. n = the number ineach group.

= <0.05) lower than normal for all three targets (TableI). Only the mean cytotoxic index against lympho-cyte targets differed significantly from the normalgroup mean for the X-linked agammaglobulinemicgroup.

Results of studies with effector cells from two in-fants with severe combined immunodeficiency areshown in Fig. 2, and the cytotoxic indices against thethree different types of targets are presented in TableI. No significant 51Cr release was obtained when un-fractionated mononuclear cells from one infant weretested against HLA antibody-coated lymphocyte tar-gets in five separate experiments, nor did cellsfrom the other patient cause significant release in theonly experiment in which his cells were tested.On the other hand, mononuclear cells from one pa-tient caused significant release of 51Cr from both typesof red cell targets. Previous studies from this labora-tory had shown that patient 1 had 4% and patient2 had 3% Fc receptor-bearing cells in their Ficoll-Hypaque-purified, macrophage-depleted mononu-clear cells. Comparison of the mean cytotoxic indexagainst lymphocyte targets for the infants with severecombined immunodeficiency with that of the normalgroup by Student's t test revealed it to be signifi-cantly lower (P = <0.05) than normal.

Mean cytotoxic indices for the three patients withhyper IgM, the two with DiGeorge syndrome, and theone with chronic mucocutaneous candidiasis are pre-sented in Table I. Student t tests comparing the meansrevealed the mean cytotoxic indices for the threehyper IgM patients against the lymphocyte and humanerythrocyte targets to be significantly (P = <0.05)lower than normal and the mean cytotoxic index for thetwo DiGeorge patients against lymphocyte targets tobe significantly (P <0.05) higher than normal.

90_ EFFECTORS80-0 Normal Control

° 70 _ :g g SCID

cr-60 1

E,p. Exp.2 Exp.3 Eap.4 Ep.5 Eap.6Lymph CRBC HRBC

TARGETS

FIGURE 2 Comparison of antibody-dependent cellular cyto-toxicity by mononuclear cells from two infants with severecombined immunodeficiency (SCID) against the same threetargets shown in Fig. 1. Patient 1 was studied over a 10-mo period against a variety of different HLA antibody-coated lymphocyte targets but on only one occasion againstthe erythrocyte targets. Patient 2 was studied only one andonly against lymphocyte targets at the age of 10 mo.

4 S. 0. Sanal and R. H. Buckley

Effect of depletion of adherent or phagocytic cellson antibody-dependent cell lysis by mononuclear cells.Because Ficoll-Hypaque-purified mononuclear cellsuspensions from both normal and immunodeficientsubjects were noted to vary in their monocyte content,and because some of the X-linked agammaglobuline-mic patients' mononuclear cell suspensions were ob-served to have very high (up to 63%) percentages ofcells staining with myeloperoxidase, a series ofstudies with both normal and immunodeficient sub-jects' cells was designed to determine what effect anexcessive number of monocytes might have on effectorcell function. Mean percentages of myeloperoxidase-

TABLE IIEffect of Adherent Cell Depletion on Percentage of Fc

Receptor Cells and Cytotoxicity with MononuclearCell Suspensions from Immunodeficiency Patients

Percentage of cells Cytotoxicitypositive for* index against

lymphocyteType of donor POX HEA target

Normal control 1Untreated mononuclear

cells 14.5 11.2 37.36Column treated cells 0.5 8.9 50.28

X-linked agammaglobulin-emic 1

Untreated mononuclearcells 20 17 7.01

Column treated cells 0.4 4.05 10.34

X-linked agammaglobulin-emic 2

Untreated mononuclearcells 9 7.7 5.92

Column treated cells 0.66 - 8.65

Commonvariable agamma-globulinemic 1

Untreated mononuclearcells 15.4 18 10.02

Column treated mononu-clear cells 0.3 11.5 18.42

Hyper IgM 1Untreated mononuclear

cells 23.7 20.24Column treated mononu-

clear cells 5.9 37.47

Hyper IgM 2Untreated mononuclear

cells 17.4 29 22.11Column treated cells 31.74

$ POX= myeloperoxidase positive cells; HEA= cells roset-ting with human O+ erythrocytes coated with Ripley IgGanti-CD antibody.

60

50-

40-

30

20-

I0

nag9 n2 no4 n3 nf3 nf3Lymph CRBC HRBC

TARGETS

FIGURE 3 The effect of depletion of adherent or phagocyticcells on antibody-dependent cellular cytotoxicity against vari-ous targets by normal mononuclear (MN) cells. Refer to legendto Fig. 1 for description of targets. n refers to the number ofexperiments in which the activity of the particular type ofeffector cell population was evaluated. The vertical linesrepresent + 1 SD of the mean % 'tCr released (repre-sented by the top of each bar) in all of the experiments donewith that particular type of effector.

positive cells detected in Ficoll-Hypaque-purifiedcells from 14 normal controls, 9 X-linked agamma-globulinemics, and 10 common variable agamma-globulinemics were 7.54+1.07, 20.96+16.69, and 10.32+3.26, respectively. Although the mean is higher forthe X-linked agammaglobulinemic group than for theother two, the difference is not statistically significantbecause of the large standard deviation for that group.

Neither nylon column nor carbonyl iron treatmentsignificantly lowered the percentage of cells rosettingwith Ripley IgG-coated human erythrocytes, yet iteliminated a majority of the monocytes (Table II).Moreover, antibody-dependent cellular cytotoxicityincreased significantly (P = <0.01) against lympho-cyte targets after both adherent and phagocytic cellremoval and remained essentially unchanged againstthe two types of erythrocyte targets (Fig. 3). Thesestudies indicate that Fc receptor-bearing lymphocytescan serve as effector cells against all three types oftargets. As shown in Table II, however, removal ofessentially all myeloperoxidase-positive cells from Fi-coll-Hypaque-purified cells of two patients with X-linked agammaglobulinemia by nylon column treat-ment did not significantly improve the cytotoxic ac-tivity of those cells against lymphocyte targets,whereas removal of myeloperoxidase-positive cellsfrom mononuclear cells of one patient with commonvariable agammaglobulinemia and two with hyperIgM increased this activity in the remaining cells,

Antibody-Dependent Cellular Cytotoxicity in Immunodeficiency 5

just as it did in normals. These results suggest thatX-linked agammaglobulinemic mononuclear cells arefunctionally deficient in a subpopulation of lympho-cytes capable of lysing antibody-coated lymphocytebut not antibody-coated erythrocyte targets. The im-paired effector cell function against lymphocyte tar-gets does not appear to be due to a lack of Fc receptor-bearing lymphocytes, since cells forming rosetteswith Ripley IgG-coated human erythrocytes have beenfound in normal or near normal quantity in all X-linkedagammaglobulinemia monocyte-depleted mononuclearcell preparations examined thus far.

Effect of sheep erythrocyte rosette depletion andfunction of sheep erythrocyte rosette-forming cellsin antibody-dependent cellular cytotoxicity. In afurther effort to characterize the cytotoxic effectorcells for the three types of antibody-coated targets em-ployed in these studies, sheep erythrocyte rosette-depletion studies were performed. As shown in Fig. 4,when monocytes were depleted from normal mono-nuclear cells before rosetting them with sheep erythro-cytes, lymphocytes forming such rosettes did notcause significant 51Cr release from any of the threetypes of targets. In contrast, if sheep erythrocyterosette-forming cells were prepared from unfrac-tionated mononuclear cells, the pelleted cells wereable to effect significant lysis of both red celltargets; such cells were shown by myeloperoxidasestaining to contain significant numbers of monocytes(around 5%). In both situations, the non-rosette-forming cells were able to serve as effectors againsteach of the three types of targets, and this activitywas often greatly enhanced in the depleted population(Fig. 4).

Effect of depletion of cells bearing the comple-ment receptor on antibody-dependent cellular cyto-toxicity. The results of experiments in which mono-

70

60

50

, 400:

a 30

20

10

UEp.1 Etp.2

Lymph

I 1IEap.1 Esp.2 Exp.l Exp.2

CRBC HRBCTARGETS

Effectors'O Monocyte-epleted LymphsOMonocyte-Depleted E Lymphs

Non-E-Roseffing CellsUnfroctionoted MNCells

1 E Lymphs + Mono's

CRBC HRBC

FIGURE 4 Effect of sheep erythrocyte (E) rosette-depletionand function of E-rosetting cells in cytotoxicity experimentsemploying the same three types of antibody-coated targetsdescribed in the legend to Fig. 1. The source of the mono-nuclear cells for these experiments was normal subjects.

nuclear cells forming rosettes with erythrocytes coatedwith IgM antibody and complement were depletedand the remaining cells tested against all three typesof targets are presented in Fig. 5. The activity inthe population depleted of complement receptor-bear-ing cells was comparable to that of the unfractionatedmononuclear cell suspension against all targets.

Effect of depletion of immunoglobulin-bearing cellson antibody-dependent cellular cytotoxicity. Asshown in Table III, depletion of a majority of immuno-globulin-bearing cells by passing Ficoll-Hypaquemononuclear cells over an anti-F(ab')2 column hadlittle effect on the cytotoxic activity of the remainingcells, regardless of the type of antibody-coated target,except that it was increased for the lymphocyte target.This was true for both normal and DiGeorge cells. Thedepleted populations still had o2% immunoglobulin-bearing cells, however, and contained from 2.3 to 11.9%monocytes and from 6.8 to 16% cells forming rosetteswith Ripley IgG-coated human erythrocytes.

Effect of depletion of cells bearing the Fc receptor.In contrast to the results of all other types of deple-tion studies, removal of cells forming rosettes withIgG-coated erythrocytes had a profound effect on thecytotoxic activity of the remaining mononuclear cellsagainst lymphocyte targets (Fig. 6). Specific 51Crrelease ranged only from 4.3 to 11% when non-rosette-forming cells were tested against HLA anti-body-coated lymphocyte targets in four separate ex-periments. Cytotoxic activity was detected with thesecells against both erythrocyte targets, however, butthis could have been due to residual monocytesfound in concentrations of 3.5 to 6% in all depletedpopulations tested. Cells rosetting with IgG-coatederythrocytes were able to effect lysis against all threetargets, but the activity of the cells against the

Effectors:60 O Unfroctionoted MNCells

ED Non-HEAC Cells50-

40-

-Y 30-

20-

10

Ep.I Ehp.2Lymph CRBC HRBC

FIGURE 5 Effect of depletion of complement receptor-bear-ing lymphocytes on antibody-dependent cellular cytotoxicityagainst various targets. See legend to Fig. 1 for descriptionof targets. HEAC, human erythrocytes coated with rabbitIgM antibody and mouse complement; MN, mononuclearcells.

6 S. 0. Sanal and R. H. Buckley

|

TABLE IIICharacteristics of Mononuclear Cell Populations and ADCCActivity

after Anti-F(ab')2 Column Treatment

Percentages of cellspositive for Cytotoxicity indices

MononuclearDonor cell population POX* E* HEA* SIg* Lymph* CRBC*

Normal Unfractionated 15 57.7 7.9 25.9 47.68mononuclear cells

Normal SIg- 6.5 69.6 11.1 - 43.15 50.67

Normal SIg- 2.3 68.4 6.8 2 37.03 50.18

DiGeorge SIg- 11.9 53.5 16 2 57.47 63.4

* POX= myeloperoxidase staining; E = sheep erythrocyte rosetting; HEA= rosettingwith human erythrocytes coated with Ripley IgG; SIg = surface immunoglobulin;Lymph = HLA antibody-coated lymphocyte targets; CRBC= chicken erythrocytescoated with rabbit IgG antibody.

lymphocyte targets was not as great as that of the un-fractionated cells. This was possibly due to the factthat red cells were not successfully removed in thelysis procedure.

DISCUSSION

We undertook the evaluation of antibody-dependentcellular cytotoxicity in patients with well-defined hostdeficits, not only to obtain base-line information on thisfunction in various immunodeficiency diseases, butalso to gain insight into the possible relevance ofthis phenomenon in vivo. Different conclusions werereached regarding this function in immunodeficientpatients in the limited number of published studies

70

60

5C

40

30

20

10

Eap.

Effectors:O Unfroctionoted MNCellsQ Non-HEA Cells0 HEA-Cells (Partiol Lysis of RBCs)

n n nr

Esp.

4 Exp. Exp.2CRBC

TARGETS

IEzp. 2 Exp. 3

Lymph HRBC

FIGURE 6 Effect of depletion of lymphocytes formingrosettes with human erythrocytes coated with IgG antibody(HEA) on antibody-dependent cellular cytotoxicity againstvarious targets. See legend to Fig. 1 for description oftargets. The rosette-forming cells still had some erythrocytesattached at the time of the cytotoxicity studies, becauseonly partial lysis was achieved with the pony anti-rabbit IgGand human complement.

to date (16-18). Blaese et al. (16), using normalhuman lymphocytes coated with HLA antibody as tar-gets, found four Wiskott-Aldrich patients' mean anti-body-dependent cellular cytotoxicity to be 95% of thenormal mean, whereas the mean cytotoxicity of eighthypogammaglobulinemic patients was only 50% (range25-95%) of normal. The authors speculated, primarilyon the basis of these observations and their find-ing that cancer and T-cell-deficient intestinal lym-phangiectasia patients also had normal or increasedantibody-dependent cytotoxicity function, that cyto-toxicity against that type of antibody-coated targetwas probably mediated by a B or some other non-T cell.Rachelefsky et al. (18) also found low cytotoxic activitywith five agammaglobulinemic patients' lymphocyteswhen they were tested against HLA antibody-coatednormal human lymphocytes and concluded thatdeficient antibody-dependent cellular cytotoxicity cor-related with decreased numbers of B cells. In contrast,Wisloff and Froland (17), in studying the cytotoxic ac-tivity of agammaglobulinemic patients' lymphocytesagainst antibody-coated chicken erythrocyte targets,found normal activity and concluded that such func-tion was independent of the presence or number ofB cells. The studies presented here point up thecrucial role of the type of target in assessing antibody-dependent cellular cytotoxicity (10, 11) and help ex-plain conflicting findings in various published reportsas to effector cell identity. Few studies have com-pared effector activity of mononuclear cells from eitherpatients or normals against more than one type of target,and we could find none that evaluated all three ofthe types of targets used in the present investigation.Nelson et al. (11), in studies comparing normal effectorcell function against rabbit antibody-coated Changliver cells and guinea pig antibody-coated chicken

Antibody-Dependent Cellular Cytotoxicity in Immunodeficiency

I

-i

7

erythrocytes, found that cytotoxicity was mediatedagainst Chang cells only by surface immunoglobulin-negative lymphocytes, but that lysis of chicken erythro-cyte targets was effected by polymorphonuclear leuko-cytes, macrophages, and immunoglobulin-negativelymphocyte subpopulations. The findings for chickenerythrocyte targets are in keeping with those in thisreport, and their results with Chang cells parallel thosewith HLA antibody-coated lymphocytes in the presentstudy.

Although cytotoxic activity was significantly (P= <0.05) lower than normal against lymphoid targetsfor most of the agammaglobulinemic patients in thepresent study, the results appear to differentiate pa-tients with X-linked agammaglobulinemia from thosewith common variable agammaglobulinemia by themore severe degree of impairment in K-cell functionin the former. X-linked agammaglobulinemic patientslack immunoglobulin-bearing B lymphocytes, whereascommon variable agammaglobulinemic patients usu-ally have a normal or near normal number of such cells(20). These observations plus the significantly higherthan normal mean cytotoxic index of the two DiGeorgepatients against lymphocyte targets tend to support thepostulate of Blaese et al. (16) and Rachelefsky et al.(18) that antibody-dependent cellular cytotoxic func-tion correlates with the number of B cells. This hy-pothesis fails, however, in the case of the two in-fants with severe combined immunodeficiency, one ofwhom had 69% and the other 12% surface immuno-globulin in M- and D-bearing cells (21) but neither ofwhose cells effected significant lysis of antibody-coated lymphocyte targets. These findings are ofinterest in view of current differences of opinionas to the nature of the K lymphocyte. As in the pres-ent study, Brier et al. (27) found the human lympho-cyte responsible for lysis of antibody-coated lympho-cyte targets in a subpopulation of nonadherent cellsexpressing neither the receptor for sheep erythro-cytes nor surface immunoglobulin. Later work fromthat laboratory, however, indicated that when thatpopulation of cells was held in culture for 6 days,immunoglobulin was synthesized by the cells, sug-gesting that they were B-cell precursors (28). Theiradditional finding that antibody-dependent cellularcytotoxicity could be abrogated by treatment of sucha cell population with an anti-Ia antiserum appearedto provide further evidence for a B-cell origin of theeffector lymphocyte in antibody-dependent lysis ofnucleated target cells (28). Since interaction of wholeanti-Ia antibodies or la antibodies complexed withantigen with Fc receptors on K lymphocytes couldhave led to abrogation of antibody-dependent cyto-toxicity, however, this conclusion remains open toquestion. In this regard, Nelson et al. (29) have re-

cently reported that Fc receptor-bearing, surface im-munoglobulin-negative lymphocytes which mediatecytotoxicity against antibody-coated Chang liver cellslack B-cell alloantigens.

The subpopulation of normal lymphocytes mediatingcytotoxicity against HLA antibody-coated lymphocytetargets in this study had surface membrane char-acteristics identical with those of the subpopulationfound to effect lysis of Chang targets by Nelson et al.(11). The rosette-depletion studies provide strongevidence that the antibody-dependent cytotoxicity ef-fector cell for HLA antibody-coated lymphocyte tar-gets is an Fc receptor-bearing lymphocyte. The presentstudy, as well as the earlier studies of Wisloffand Froland (17) and the more recent ones of Papa-michael and Temple (30), Nelson et al. (11), andPape et al. (31), demonstrate that there is no require-ment for immunoglobulin-positive lymphocytes to ob-tain lysis of either nucleated or nonnucleated anti-body-coated targets. Work by others (32) has shownthat B lymphocytes have Fc receptors for IgG, butthese can usually be demonstrated only by fluoro-sceinated aggregated IgG or immune complexes. Incontrast, Ripley-coated human O+ erythrocytes appearto form rosettes only with lymphocytes having high-avidity Fc receptors and lacking conventional T- andB-cell markers (5). Results of depletion of this latterpopulation in the present investigation, as well as instudies of Wisloff et al. (5) and Pape et al. (31),confirm that the presence of this receptor is a pre-requisite for antibody-dependent cellular cytotoxicityeffector activity. The present investigation also con-firmed the findings of Nelson et al. (11) that lympho-cytes bearing the complement receptor are not re-quired for lysis of antibody-coated nucleated targets,and of Wisloff et al. (5) and Greenberg et al. (33)that the lymphocyte mediating this function does notbear T-cell markers. Moreover, like Nelson et al. (11),we found that purified E-rosetting cells are incapableof effecting cell lysis. While no purified preparationsof monocytes were used in these studies, populationsof mononuclear cells depleted of Ripley rosettinglymphocytes still contained up to 5% myeloperoxi-dase-positive cells; such suspensions were unable toeffect greater than 10% 51Cr release from lymphocytetargets but still caused significant lysis of both red celltargets. Depletion of monocytes increased the cyto-toxic function of the remaining cells against lympho-cyte targets but not against the two erythrocyte targets,confirming that K lymphocytes can lyse all three typesof targets.

The absence of cytotoxic activity against nucleatedtargets, despite the presence of cells which formedrosettes with Ripley IgG-coated human erythrocytesamong macrophage-depleted lymphocyte populations

8 S. 0. Sanal and R. H. Buckley

from the patients with X-linked agammaglobulinemiaand severe combined immunodeficiency in the presentstudy (20, 21) and from patients with chronic lympho-cytic leukemia evaluated by Gale et al. (34), is evidencethat the mere presence of Fc receptor-bearing lympho-cytes is insufficient for antibody-dependent cellularcytotoxicity to occur. Unfortunately, the finding ofseverely impaired K-lymphocyte function in infantswith severe combined immunodeficiency does not addmuch to the understanding of either the nature of theK cell, since all lymphocyte function is lacking insuch patients, or of the relevance of antibody-dependent cellular cytotoxicity to host defense, sincethis defect is invariably fatal unless corrected. Thefinding of an equally severe K-lymphocyte impairmentin patients with X-linked agammaglobulinemia whohave normal T-cell function but lack surface immuno-globulin-bearing B lymphocytes may be a clue thatantibody-dependent cellular cytotoxicity does have anin vivo protective role. Patients with X-linked agam-maglobulinemia have recently been shown to have ahigh propensity to develop persistent enterovirusmeningoencephalitis, ultimately leading to death, de-spite adequate humoral replacement therapy (35). Thishas not been observed thus far in comparably treatedpatients with common variable agammaglobulinemia.

ACKNOWLEDGMENTS

Weare indebted to Dr. Euripedes Ferreira for performingthe antibody-dependent cellular cytotoxicity studies with thesevere combined immunodeficiency disease lymphocytes andfor his early guidance in the technical aspects of this as-say. We are also grateful to Mrs. Lora Whitfield for herexcellent secretarial assistance with the manuscript.

This work was supported in part by a fellowship grantto Dr. Sanal from the Department of Pediatrics, Hacet-tepe University, Ankara, Turkey, and a grant from the GeneralClinical Research Centers Program (RR-30) of the Divisionof Research Resources, National Institutes of Health.

REFERENCES1. Moller, E. 1965. Contact-induced cytotoxicity by lym-

phoid cells containing foreign isoantigens. Science (Wash.D. C.). 147: 873-879.

2. Perlmann, P., H. Perlmann, and H. J. Muller-Eberhard.1975. Cytolytic lymphocytic cells with complement re-ceptor in human blood. Induction of cytolysis by IgGantibody but not by target cell-bound C3. J. Exp. Med.141: 287-296.

3. Perlmann, P., and H. Perlmann. 1970. Contactual lysisof antibody-coated chicken erythrocytes by purifiedlymphocytes. Cell. Immunol. 1: 300-315.

4. Winchester, R. J., S. M. Fu, T. Hoffman, and H. G.Kunkel. 1975. IgG on lymphocyte surfaces; technicalproblems and the significance of a third cell population.

J. Immunol. 114: 1210-1212.5. Wisloff, F., S. S. Froland, and T. E. Michaelsen. 1974.

Antibody-dependent cytotoxicity mediated by human Fc-

receptor bearing cells lacking markers for B and Tlymphocytes. Int. Arch. Allerg. Appl. Immunol. 47:139-154.

6. Ferrarini, M., L. Moretta, R. Abrile, and M. L. Durante.1975. Receptors for IgG molecules on human lympho-cytes forming spontaneous rosettes with sheep red cells.Eur. J. Immunol. 5: 70-72.

7. Huber, H., and H. H. Fudenberg. 1968. Receptor sitesof human monocytes for IgG. Int. Arch. Allerg. Appl.Immunol. 34: 18-31.

8. Wong, L., and J. D. Wilson. 1975. The identification ofFc and C3 receptors on human neutrophils. J. Immunol.Methods. 7: 69-76.

9. Jondal, M., H. Wigzell, and F. Auiti. 1973. Human lym-phocyte subpopulations: classification according to sur-face markers and/or functional characteristics. Trans-plant. Rev. 16: 163-195.

10. Kovithavongs, T., G. Rice, K. L. Thong, and J. B.Dossetor. 1975. Effector cell activity in antibody mediatedcell dependent immune lysis. II. Evidence for differentpopulations of effector cells for different targets. Cell.Immunol. 18: 167-175.

11. Nelson, D. L., B. M. Bundy, H. E. Pitchon, R. M.Blaese, and W. Strober. 1976. The effector cells in humanperipheral blood mediating mitogen-induced cellularcytotoxicity and antibody-dependent cellular cytotoxicity.

J. Immunol. 117: 1472-1481.12. Ting, A., and P. I. Terasaki. 1974. Influence of lympho-

cyte-dependent antibodies on human kidney transplants.Transplantation (Baltimore). 18: 371-373.

13. Pollack, S., G. Heppner, R. J. Brawn, and K. Nelson. 1972.Specific killing of tumor cells in vitro in the presenceof normal lymphoid cells and sera from hosts immune tothe tumor antigens. Int. J. Cancer. 9: 316-323.

14. Calder, E. A., W. J. Penhale, D. McLeman, E. W.Barnes, and W. J. Irvine. 1973. Lymphocyte-dependentantibody-mediated cytotoxicity in Hashimoto's thyroidi-tis. Clin. Exp. Immunol. 14: 153-158.

15. Zawodnik, S. A., G. D. Bonnard, and E. Gautier. 1976.Antibody-dependent cell-mediated destruction of humanerythrocytes sensitized in ABOand Rhesus fetal-maternalincompatibilities. Pediatr. Res. 10: 791-796.

16. Blaese, R. M., E. Rosenberg, and J. Wunderlich. 1972.Evidence of two functionally distinct lymphocyte popula-tions in man. In Microenvironmental Aspects of Im-munity. B. D. Jankovic and K. Isakovic, editors. PlenumPress, New York. 315-320.

17. Wisloff, F., and S. S. Froland. 1973. Antibody-dependentlymphocyte-mediated cytotoxicity in man: no require-ment for lymphocytes with membrane-bound immuno-globulin. Scand. J. Immunol. 2: 151-157.

18. Rachelefsky, G. S., P. R. McConnachie, A. J. Ammann,P. I. Terasaki, and E. R. Stiehm. 1975. Antibody de-pendent lymphocyte killer function in human immuno-deficiency diseases. Clin. Exp. Immunol. 19: 1-9.

19. Cooper, M. D., W. P. Faulk, H. H. Fudenberg, R. A. Good,W. Hitzig, H. Kunkel, F. S. Rosen, M. Seligmann, J.Soothill, and R. J. Wedgwood. 1973. Classification ofprimary immunodeficiencies. N. Engl. J. Med. 288:966-967.

20. Schiff, R. I., R. H. Buckley, R. B. Gilbertsen, and R. S.Metzgar. 1974. Membrane receptors and in vitro respon-siveniess of lymphocytes in human immunodeficiency. J.Immunol. 112: 376-386.

21. Buckley, R. H., R. B. Gilbertsen, R. I. Schiff, E. Ferreira,S. 0. Sanal, and T. A. Waldmann. 1976. Heterogeneity oflymlphocyte subpopulations in severe combined immuno-

Antibody-Dependent Cellular Cytotoxicity in Immunodeficiency 9

deficiency. Evidence against a stem cell defect. J. Clin.Invest. 58: 130-136.

22. Boyum, A. 1968. Isolation of mononuclear cells andgranulocytes from human blood. Scand. J. Clin. Lab. In-vest. 21 (Suppl. 97): 77-89.

23. Kaplow, L. S. 1965. Simplified myeloperoxidase stainusing benzidine dihydrochloride Blood. 26: 215-219.

24. Zucker-Franklin, D. 1974. The percentage of monocytesamong "mononuclear" cell fractions obtained fromhuman blood. J. Immunol. 112: 234-240.

25. Chess, L., R. P. MacDermott, and S. F. Schlossman. 1974.Immunologic functions of isolated human lymphocytesubpopulations. J. Immunol. 113: 1113-1121.

26. Trinchieri, G., M. DeMarchi, W. Mayr, M. Savi, and R.Ceppellini. 1973. Lymphocyte antibody lymphocytolyticinteraction (LALI) with special emphasis on HL-A.Transplant. Proc. 5: 1631-1649.

27. Brier, A. M., L. Chess, and S. F. Schlossman. 1975.Human antibody-dependent cellular cytotoxicity. Isola-tion and identification of a sub-population of peripheralblood lymphocytes which kill antibody-coated autologoustarget cells. J. Clin. Invest. 56: 1580-1586.

28. Chess, L., R. Evans, R. E. Humphreys, J. L. Strominger,and S. F. Schlossman. 1976. Inhibition of antibody-dependent cellular cytotoxicity and immunoglobulinsynthesis by an antiserum prepared against a human Bcell Ia-like molecule. J. Exp. Med. 144: 113-122.

29. Nelson, D. L., W. Strober, L. D. Abelson, B. M. Bundy,

and D. L. Mann. 1977. Distribution of alloantigens onhuman Fc receptor-bearing lymphocytes: the presence ofB cell alloantigens on sIg-positive but not slg-negativelymphocytes. J. Immunol. 118: 943-946.

30. Papamichael, M., and A. Temple. 1975. Characterizationof the human effector cell causing antibody-mediatedcytotoxicity. Clin. Exp. Immunol. 20: 459-467.

31. Pape, G. R., M. Troye, and P. Perlmann. 1977. Char-acterization of cytolytic effector cells in peripheral bloodof healthy individuals and cancer patients. J. Immunol.118: 1919-1924.

32. Dickler, H. B., and H. G. Kunkel. 1972. Interaction ofaggregated y-globulin with B lymphocytes. J. Exp. Med.136: 191-196.

33. Greenberg, A. H., L. Hudson, L. Shen, and I. M. Roitt.1973. Antibody-dependent cell-mediated cytotoxicitydue to a "null" lymphoid cell. Nat. New Biol. 242:111-113.

34. Gale, R. R., J. Zighelboim, R. C. Ossorio, and J. L.Fahey. 1975. A comparison of human lymphoid cells inantibody-dependent cellular cytotoxicity (ADCC). Clin.Immunol. Immunopathol. 3: 377-384.

35. Wilfert, C. M., R. H. Buckley, T. Mohannakumar, J. F.Griffith, S. L. Katz, J. K. Whisnant, P. A. Eggleston,M. Moore, E. Treadwell, M. N. Oxman, and F. S.Rosen. 1977. Persistent and fatal Echovirus infection inpatients with agammaglobulinemia. N. Engl. J. Med. 296:1485-1489.

10 S. 0. Sanal and R. H. Buckley