HYBRIDOMAVolume 15, Number 1, 1996Mary Ann Liebert, Inc.

Characterization of a Series of Isotype Switch Variants of a

New CD20 Monoclonal Antibody

ERIK HOOIJBERG, IOHAN J. SEIN, PAUL CM. van den BERK, and ANNEMARIE HEKMAN

ABSTRACT

A series of heavy chain switch variants has been isolated from a new B cell-specific monoclonal antibody be-longing in the CD20 cluster. The antibodies NKI-B20/1, NKI-B20/2b, and NKI-B20/2a (of isotype IgG,, IgG2b,and IgG2a, respectively) have been used to study the influence of isotype and of the target antigen on the ca-

pacity to mediate cytotoxicity with a number of effector mechanisms. Unlike many mouse MAbs, NKI-B20/2band NKI-B20/2a were cytolytic with human complement on human target cells that did not express the com-

plement regulatory factor HRF20. All 3 isotypes of NKI-B20 mediated antibody-dependent cell-mediated cy-totoxicity (ADCC) with rIL-2-activated NK cells from mouse spleen. Here the antigen density seemed the mostimportant factor in determining the level of cell kill. With mouse peritoneal macrophages as effector cellsagain all 3 isotypes of NKI-B20 mediated cytotoxicity. For the IgGi and IgG2D variants of NKI-B20 this is atvariance to what has been reported for MAbs of other specificities. Despite the high activity with murine ef-fector cells none of the NKI-B20 MAbs mediated ADCC with human peripheral blood NK cells, with or with-out stimulation with rIL-2, due to the lack of interaction of the murine MAbs with the human Fc receptor.The CD20 antigen appears to be a good target antigen for various forms of cytotoxicity, to which its relativelyhigh antigenic density, its resistance to antibody-induced modulation, and its unusual structure all contribute.

differentiation, all malignancies of mature B cells are CD20positive, but early pre-B ALL cells often lack CD20 antigen.°-5)The restriction of its expression to B lineage cells, together withthe lack of expression on progenitor cells, makes CD20 anti-gen an interesting target for (antibody-based) immunotherapyfor B cell lymphoid malignancies. Moreover, CD20 antigen isexpressed at relatively high density on most normal and ma-

lignant B cells and it is not susceptible to antibody inducedmodulation.'6-7' This paper reports the production of a set ofheavy chain switch variants of a CD20 MAb and their charac-terization with regard to in vitro properties that may be rele-vant for therapeutic applications.

MATERIALS AND METHODS

Cell lines and culture conditions

Sp2/0 mouse myeloma cells and hybridomas were culturedin Dulbecco's modified minimum essential medium (DMEM,

Division of Immunology, The Netherlands Cancer Institute/Antoni van Leeuwenhoek Huis, Plesmanlaan 121, 1066 CX Amsterdam, TheNetherlands.

INTRODUCTION

The antigen recognized by CD20 MAbs is a 35-37 kDanonglycosylated phosphoprotein,'" that may function as a

Ca2+ ion channel.'2' The protein structure deduced from thecloned DNA sequence suggests that the CD20 antigen is a trans-membrane protein that passes the membrane four times, withboth the C- and the N-terminus on the cytoplasmic side andonly a small part exposed on the cell surface.'" Expression ofCD20 antigen is specific for B lymphocytes and is present dur-ing most differentiation stages, beginning at the early pre-B cellstage and disappearing upon differentiation into plasma cells.'3'The CD20 molecule appears to be involved in the activationand cell cycle progress of B lymphocytes. Stimulating as wellas inhibitory effects in different phases of the cell cycle havebeen reported for different CD20 MAbs. Pokeweed mitogen-induced differentiation of B lymphocytes into Ig-producingcells is inhibited by CD20 MAbs.'4)

The CD20 antigen is also expressed on most malignant Bcells. In agreement with the expression during normal B cell

23

24 HOOIJBERG ET AL.

Gibco, Paisley, Scotland) with glutamine, supplemented with10% fetal calf serum (FCS) (Gibco), 20 pM 2-mercaptoethanol(Merck, Darmstadt, Germany), penicillin 100 U/ml (GistBrocades, Delft, The Netherlands), and kanamycin 100 pg/ml(Sigma, St Louis, MO). The Burkitt lymphoma B cell linesDaudi, Bjab, Raji, Ramos, and Jiyoye, the EBV lymphoblas-toid cell line JY, and the cell line JVM-3, derived byEpstein-Barr virus, (EBV) transformation from a B-prolym-phocytic leukemia, were cultured in Iscove's modified medium(Gibco) with glutamine, supplemented with 5% FCS, penicillin100 U/ml and streptomycin 100 pg/m\ (Boehringer Mannheim,Mannheim, Germany). 3T3 wild type cells were cultured inDMEM with 10% FCS and 3T3 mouse fibroblasts transfectedwith cDNA encoding human CD20 antigen (3T3-hCD20) in thesame medium with 1 mg/ml G418 (Gibco).

Mice

BALB/c mice were bred and maintained in the animal de-partment of the Netherlands Cancer Institute. Immunization was

started when mice were 6-8 weeks old.

Generation of 1H4 MAb

BALB/c mice were immunized with Daudi cells. Mice were

injected sc on day 0 with 10 X 106 cells in complete Freundsadjuvant (DIFCO, Detroit, MI), with 10 X 106 cells in incom-plete Freunds adjuvant ip on day 14 and 10 X 106 cells with-out adjuvant ip on day 21. A final boost of 5 X 106 cells was

given iv after a variable interval of several weeks. Four daysafter the last injection spleen cells were fused with Sp2/0myeloma cells using polyethylene glycol (PEG 4000, Serva,Heidelberg) following the method previously described.<8)Fused cells were resuspended in HAT medium (DMEM with10% NCTC medium, Flow, Irvine, Scotland), 20% FCS, 50 pM2-mercaptoethanol, 0.1 mM hypoxanthine (Sigma), 0.4 pMaminopterine (Serva), 16 pM thymidine (Sigma), and distrib-uted over 10 flat-bottomed 96-well culture plates (Costar,Cambridge, MA) with mouse peritoneal macrophages as feedercells (3000/well).

Supernatants of growing colonies were tested in indirect im-munofluorescence on Daudi cells, on 3T3-hCD20 and on wildtype 3T3 cells. Hybridomas potentially producing CD20 MAbswere cloned by limiting dilution in HAT medium supplementedwith 5-10% human endothelial cell supernatant, which containsinterleukin (IL)-6.

Isolation of isotype switch variants

From the hybridoma D2-1H4, producing IgGi k CD20 an-

tibodies, spontaneously occurring variant clones secretingheavy chain switch variant antibodies were isolated followingthe method developed by Boot et a/.'9' D2-1H4 cells were

seeded at 500 cells/200 /¿1/well, on 10 flat-bottomed microtiterplates (Costar). After 6 days individual supernatants were testedfor the presence of IgG2b antibodies in an isotype-specificELISA as described below. IgG2t,-positive supernatants were

tested in indirect immunofluorescence on Daudi cells, using iso-type-specific FITC conjugates. Cells producing IgG2b antibod-ies by both assays were plated at 50 cells/well and the selec-tion procedure was repeated. Positive cells were plated at 5 and

1 cell/well. An IgG2t,-positive clone from the 1 cell/well plat-ing was recloned (0.3 cells/well) until all subclones producedonly IgG2b antibodies. From the IgG2b variant an IgG2a-pro-ducing variant clone was isolated by similar methods.Antibodies were purified from ascites fluid using protein A-Sepharose CL-4B columns (Pharmacia, Uppsala, Sweden).

Isotype-specific ELISA

Microtiter plates (Immunopiate, Nunc, Roskilde, Denmark)were coated overnight at 4°C with a rat monoclonal antibodyagainst mouse IgG2b or IgG2a (RM162aM, RM162bM, CLB,Amsterdam, the Netherlands) 5 /xg/ml in phosphate-bufferedsaline (PBS). After washing the plates with PBS containing0.2% gelatine (w/v) (DIFCO) and 0.02% (v/v) Tween 20(Merck) (PGT), cell culture supernatants were incubated for 30min at room temperature. After washing with PGT, boundmouse immunoglobulin was detected with rat monoclonal an-

tibody against mouse k light chain labeled with horseradish per-oxidase (RM19ME, CLB) 2 pg/m\ in PGT. After 30 min in-cubation at room temperature and washing, the plates were

developed with 3,3',5,5'-tetramethyl benzidine (Sigma, 0.1mg/ml) + H202 in acetate buffer pH 4.0. The reaction was

stopped by the addition of 0.8 M H2S04, and extinction at 450nm read in a Titertek ELISA reader.

For incidental samples the isotype was determined by IsotypeKit (Sigma or Holland Biotechnology, Leiden, the Netherlands).

ImmunofluorescenceImmunofluorescent staining was performed by standard

methods and analyzed on a FACScan flow cytometer (BectonDickinson, Mountain View, CA). Unless stated otherwise theconjugate used for indirect staining was FITC labeled, humanIg absorbed, F(ab)2 fragments of goat antimouse Ig (GAM-FITC) (Tago, Burlingame, CA). The isotype-specific conju-gates used were FITC labeled rat antimouse IgGi, IgG2a» or

IgG2b (CLB). Immunofluorescence inhibition tests were per-formed by incubating cells with the unlabeled antibody, wash-ing, and incubating with the FITC-labeled antibody, both for30 min at 4°C.

Monoclonal antibodiesBl and FITC-labeled B1 (CD20) were obtained from Coulter

(Hialeah, FL), FITC-labeled antihuman Leul6 (CD20, cloneL27) from Becton Dickinson. CLB-CD19/1 and /2a are isotypeswitched CD19 MAbs that have been described previously.'10*The MAb K8 (IgG2a), used as a negative control, is directedagainst the idiotype of the surface immunoglobulin of the tu-mor cells of a patient with non-Hodgkin's lymphoma.'8' Therat MAb R24.3, directed against HLA class II and used as pos-itive control in ADCC tests, has been described previously.'11'Several CD20 MAbs were used in comparative experiments.The hybridoma BCA-B20 has been described'l2-13) and was ob-tained from Bioprobe, Amsterdam. The MAbs 1F5 and 2H7were kindly provided by Bristol-Myers Squibb PharmaceuticalResearch Institute (Seattle, WA) and the chimeric mouse/hu-man yl version of 2H7 by Xoma Corp. (Berkeley, CA).Chimeric mouse/human yl CD 19 MAb FMC 63 was kindlyprovided by Dr H. Zola and Dr. G.A. Pietersz, Flinders Medical

CD20 ISOTYPE SWITCH VARIANTS 25

Centre, Australia. CD 16 and mouse anti-hu IgM-FITC were ob-tained from CLB, CD56 (Leu 19), FITC-labeled CD1 la, FITC-labeled CD 18, and FITC-labeled control antibodies fromBecton Dickinson, FITC-labeled CD54 from Immunotech(Marseille, France), CD59 (clone MEM-43'14') from Sanbio(Uden, The Netherlands), CD3 was from our laboratory (SPV-T3b"5').

Antigenic modulation

Two methods were used to test the effect of MAb on anti-gen density. (1) Incubation in excess MAb. Cells were incu-bated with MAb (10-50 pg/m\) in medium for 30 min at 4°C,then for the stated time of the experiment at 37°C washed twicewith PBS with 0.5% (w/v) BSA and 0.02% (w/v) sodium azideand incubated for 30 min at 4°C with GAM-FITC. (2)Incubation with cell-bound antibody only. Cells were washedwith medium after the 30 min incubation with antibody at 4°C;the test was continued as above.

ImmunoprecipitationImmunoprecipitation and gel electrophoresis were performed

by the method described by Ossendorp et a/."6'

Complement-dependent cytotoxicityIn round-bottom microtiter plates 50 p\ antibody dilution in

medium and 50 p\5lCr (Amersham) labeled target cells"3' (105cells/ml) were incubated in triplicate on ice for 15 min. Afteraddition of 100 p\ of the serum used as source of complement,incubation was continued for 45 min at 37°C. The plates were

centrifuged for 1 min at 1200 rpm and 100 p\ supernatant was

collected for counting in a y counter (Packard). Specific 51Crrelease was calculated by the formula (T

—

S)/(M—

S) X

100%, where T is the 5lCr released in the test sample, S is thespontaneous release in medium, and M is the maximal re-

leasable label in 2% Triton X-100 + 0.5% SDS. Low-Tox Hrabbit complement was obtained from Cedarlane (Ontario,Canada). The source of human complement was freshly frozenAB or O serum from healthy donors stored at -70°C.

Antibody-dependent cell-mediated cytotoxicity (ADCC)ADCC with mouse spleen NK cells and peritoneal

macrophages as effectors was performed as described previ-ously.'13' Spleen cells from BALB/c nude mice were incubatedwith rhIL-2 (500 Cetus units/ml, EuroCetus, Amsterdam) for6-7 days, whereafter erythrocytes and dead cells were removedby centrifugation over Lymphoprep (Nycomed Pharma, Oslo)and viable cells used as effectors as described."3' Peritonealexúdate cells were obtained from BALB/c mice treated with 1ml thioglycolate (Brewers, DIFCO) ip 5-7 days before call har-vest. Harvested cells were seeded in 96-well microtiter platesat densities determined by the desired E/T ratios; the next daynonadherent cells were washed away and target cells( 1000/well) and MAbs were added. After 3 days of incubationthe remaining target cells were quantified by incorporation of(3H]thymidine (Dupont's, Hertogenbosch, the Netherlands).

ADCC with human NK cells as effector cells was performedby the method described previously."7' NK cells were enrichedby isolating large granular lymphocytes (LGL) by centrifugal

elutriation from blood of normal donors."8' For some experi-ments the LGL were incubated overnight with rIL-2 (EuroCetus, Amsterdam, 25 or 50 Cetus U/ml). Frozen LGL were

used after incubation in medium overnight at 37°C. 51Cr-la-beled target cells were mixed with antibody in the appropriatedilution and added to effector cells in round bottom microtiterplates in 6 E/T ratios ranging from 100:1 to 3:1. The assay was

continued as described."7'

Vector construction

The purified Xhol fragment from 7rH3-CD20 containing thecDNA encoding the human CD20 antigen'19' was ligated intothe Sail site just upstream of the SV40 promotor of the vector

pLJ.'20' The viral long terminal repeat sequences are utilized to

promote expression of the CD20 cDNA. pLJ-hCD20 also con-

tains the neomycin gene conferring resistance to G418.

Transfection and selection of 3T3 cells

Mouse 3T3 fibroblasts (BALB/c genetic background) were

transfected with pLJ-hCD20 using Lipofectin™ Reagent (BRL,Life Technologies, Inc., Gaithersburg, MD) according to themanufacturer's procedure for stable transformation of cells.'21'Selection medium containing 1 mg/ml G418 (Gibco) was added24 h after transfection of the cells. Stable transfectants, 3T3-hCD20, were used for screening hybridoma supernatants.

RESULTS AND DISCUSSION

Generation and characterizationof monoclonal antibodies

After fusion of spleen cells from a BALB/c mouse immu-nized with Daudi Burkitt lymphoma cells with Sp2/0 myelomacells, supernatants of growing hybridoma colonies were

screened for the presence of potential CD20 MAbs. Hybridomaswere selected for reactivity in indirect immunofluorescencewith the transfected 3T3-hCD20 cells that express the. CD20antigen, but not with wild-type 3T3 cells. To avoid possible ar-

tifacts of the transfection the selected supernatants were alsotested on Daudi cells. Following this strategy three hybridomas,D1-11A9, D1-12F6, and D2-1H4, were identified that were

likely to produce CD20 MAbs.This was confirmed in blocking experiments, in which these

MAbs were found to inhibit the binding to Daudi cells of twoknown CD20 MAbs, Bl (Table 1) and anti-Leu 16 (data notshown). No differences were seen between the levels of inhi-bition by D2-1H4 and the other CD20 MAbs tested (Bl, 1F5,2H7, BCA-B20). Functional assays have shown differences be-tween various CD20 MAbs. For instance, Bl inhibits cell cy-cle progression of stimulated B cells, whereas 1F5 promotesthe induction of proliferation.'4' A conformational change in theCD20 antigen induced by IL-4 was detected only with L27/anti-Leu 16.'22' If these functional differences are due to binding todifferent epitopes, these epitopes must be so closely adjacentthat they cannot be discriminated by inhibition of binding.Immunoprecipitation from a lysate of 127I-labeled Daudi cellswith D2-1H4 and with the reference CD20 MAb Bl yieldedthe same characteristic 35-37 kDa double band (data not

26 HOOIJBERG ET AL.

Table 1. Competitive Binding of CD20 MAb"

Staining MAb Competing MAb Mean fluorescenceBl-FITC

BlNKI-B20/lb

1F52H7

BCA-B20

2185337423848

"Daudi cells were incubated with unlabeled antibodies (100jug/ml) before incubation with FITC-labeled B1. Fluorescenceintensity was measured by flow cytometry.

bNKI-B20/l is the IgGi antibody produced by clone D2-1H4.

shown). Together these data show that D1 -11A9, D1 -12F6, andD2-1H4 detect the same antigen as known CD20 MAbs.

The MAb D2-1H4 was submitted to the 5th HumanLeukocyte Differentiation Antigens Workshop (Boston, 1993)and indeed clustered in the CD20 cluster.'23'

Because isotype determination by solid phase ELISA showedthat Dl-11A9 and D1-12F6 were IgM k and D2-1H4 was IgG,k the latter MAb was selected for the isolation of isotype switchvariants.

Isolation of isotype switch variants of D2-IH4

Spontaneously arising isotype switch variants producingIgG2b and IgG2a antibodies were isolated by sequential sublin-ing. Variant clones were selected by antigen-independent iso-type-specific ELISA in combination with isotype-specific de-tection by immunofluorescence of antibodies binding to B cells.In our hands selection by antigen-independent methods alonesometimes resulted in clones producing immunoglobulins of theproper isotype that had lost antibody specificity.

The isotype switch variants derived from D2-1H4 are des-ignated NKI-B20/1 (the original IgG, mAb), NKI-B20/2b(IgG2b), and NKI-B20/2a (IgG2a). Staining of Daudi cells witheach isotype, followed by isotype-specific FITC conjugates isshown in Figure 1. Purified MAbs of each isotype were titratedand tested for inhibition of the binding of Bl-FITC and Leu 16-FITC. As shown in Figure 2 the titration curves of the threeisotypes were nearly superimposable, indicating that no majorchanges in avidity of the antibodies had occurred during theswitching process.

Antigenic modulation

No consistent changes in the density of CD20 antigen were

observed by immunofluorescence after incubation of Daudicells with NKI-B20/2b (10-50 pglml) at 37°C for periods of18 to 72 h, or of Raji, JVM3 and JY cells for up to 48 h, as

compared to cells incubated with a control antibody. Becauseby incubation in excess MAb the measured antigen density isthe net result of possible modulation of the antigen and reex-

pression, we also followed the presence of cell-bound MAb dur-ing incubation at 37°C up to 48 h. The fluorescence signal var-

ied somewhat over time, but did not decrease by more than 20%(data not shown). These observations are in agreement with thedata obtained by Vervoordeldonk et al. in modulation assaysusing radiolabeled NKI-B20/2a.<7'

FITC conjugatesanti-m lgG1 anti-m lgG2b anti-m lgG2a

neg. control

NKI-B20/2b

NKI-B20/2a

Log fluorescence intensity

FIG. 1. Confirmation of isotype specificity of NKI-B20/1,NKI-B20/2b, and NKI-B20/2a. Daudi cells were incubated withNKI-B20/1, NKI-B20/2b, or NKI-B20/2a followed by FITC-labeled rat anti mouse IgGi, IgG2b. or IgG2a.

Complement-dependent cytotoxicityThe NKI-B20 MAbs of the 3 isotypes were tested for cyto-

toxic capacity in the presence of rabbit and of human comple-ment. As shown in Figure 3 the original IgGi antibody NKI-B20/1 did not activate rabbit complement, whereas in thepresence of 5% rabbit serum the variants NKI-B20/2a and NKI-B20/2b were cytotoxic for Daudi cells, as well as for two otherCD20+ B cell lines, Raji and Bjab. The same pattern of iso-type-dependent capacity to activate rabbit complement was alsoseen with MAbs directed against another B cell specific anti-gen, the isotype switched MAbs CLB-CD19/1 and CLB-CD19/2a. This conforms to data in the literature that mouse

MAbs of IgG | isotype hardly ever activate complement andthat IgG2a and IgGib isotypes sometimes but not always can

do so.'24'

Leu Id FITC/ no inhibitor

neu. eontrol

inhibiting moabs

—

NKJ-B20/INKl-B20/2aNKI-B:0/:b

01 .1 1 10 100mAb cone. i|j.g/ml I

FIG. 2. Competition of binding with Leu 16-FITC by theNKI-B20 isotype switch family. Daudi cells were incubatedwith serial dilutions of NK1-B20/1, NKI-B20/2a, or NKI-B20/2b, before incubation with Leu 16-FITC. The log of themean fluorescence intensity is plotted against the concentrationof the inhibiting antibody. Similar results were obtained usingBl-FITC.

CD20 ISOTYPE SWITCH VARIANTS 27

-•— NKJ-B20/I-*— NKI-B20/2a- — NKI-B20/2b-O— CLB-CD *— CLB-CD

mAb cone my/mli

FIG. 3. Cytotoxicity of NKI-B20 and CLB-CD 19 isotypevariants with rabbit complement. Daudi target cells were sen-sitized with varying concentrations of NKI-B20 (closed sym-bols) or CLB-CD 19 (open symbols) MAbs ofdifferent isotypesand incubated with rabbit serum as source of complement (fi-nal cone. 5%). Specific 51Cr release was determined in tripli-cate wells. Two other experiments gave similar results.

With human serum of three donors as source of complementand Daudi cells as target NKI-B20/2a and NKI-B20/2b had cy-totoxic activity comparable to that with rabbit complement,whereas NKI-B20/1 again was inactive (Fig. 4). However, inthis assay both the target antigen and the target cell affected thelytic effect. Among five human B cell lines (Daudi, Raji, Bjab,JVM-3, and JY) with moderate to high expression of CD20 onlyDaudi cells were lysed consistently to high levels by NKI-B20/2a. Raji cells were lysed to an intermediate level (30-50%specific "Cr release) when higher antibody concentrations

T- 60

1 10mAb cone (Ug/ml)

FIG. 4. Cytotoxicity of NKI-B20 and CLB-CD 19 isotypevariants with human complement. Daudi target cells were sen-sitized with varying concentrations of NKI-B20 (closed sym-bols) or CLB-CD 19 (open symbols) MAbs of different isotypesand incubated with human AB serum as source of complement(final cone. 10%). Specific 51Cr release was determined in trip-licate wells. Four experiments with serum from two donors gavesimilar results.

(25-50 /ng/ml) were used. No lytic activity was seen with CLB-CD 19/2a on Daudi cells or on any of the other CD19+ celllines.

The lytic effect of activated complement on homologouscells is regulated by a number of membrane complement reg-ulatory proteins. To investigate the large difference in sensi-tivity with respect to lysis by NKI-B20 and human complementbetween cell lines that were equally sensitive to NKI-B20 andrabbit complement, we determined the expression on these celllines ofhomologous restriction factor HRF20. which is the anti-gen detected by CD59 MAbs.'25' In agreement with its sensi-tivity to human complement-mediated lysis Daudi cells were

found not to express HRF20 as determined with the MAbMEM-43, as has also been observed by others.'14-25' Raji, Bjab,JY, and JVM-3 cells all were CD59 positive. Raji cells, whichhad some sensitivity to lysis, showed the lowest HRF20 ex-

pression (mean fluorescence intensity of 80 as compared to200-600 for the other cell lines). Even though Daudi may be a

particularly sensitive target cell due to its lack of HRF20, thestrong lytic effect of NKI-B20/2a and NKI-B20/2b with humancomplement is remarkable in view of the generally poor ca-

pacity of mouse MAbs to activate human complement. Asshown in Figures 3 and 4, the IgG2a MAb against the CD 19antigen had activity similar to NKI-B20/2a with rabbit com-

plement, but was completely inactive with human complement.Considering the suggestion by Bindon et a/.'26' that the lyticcapacity of MAbs may be determined by the configuration ofthe target antigen, it seems possible that the structure of CD20antigen makes it into a good "cytotoxic target antigen."

ADCC with mouse effector cells

The influence of the heavy chain Fc region on the capacityto mediate ADCC was tested with mouse effector cells thatcarry different Fc receptors: NK lymphocytes obtained as rlL-2 activated spleen cells and peritoneal macrophages.(27) Afterculture with rIL-2 for 6-7 days spleen cells from nu/nu BALB/cmice consisted of 60-80% of Thy 1 + Fey RIII+ activated NKcells. With these cells as effectors NKI-B20/1, NKI-B20/2a, or

NKI-B20/2b mediated similar levels of ADCC (Fig. 5). Thiscorresponds with the results we obtained with the CLB-CD 19isotype switch variants."" Denkers et al.aH) reported similarADCC activity of the IgG,, IgG2a, and IgG2b members of a

switch variant family with nonstimulated spleen cells. However,in general, nonstimulated spleen cells are poor ADCC effec-tors.'29-30'

ADCC with thioglycolate-activated peritoneal macrophagesas effector cells was performed in a longer term assay whereantibody-coated target cells are incubated with adherentmacrophages for 3 days and surviving target cells quantitatedby incorporation of [3H]thymidine."u3) Results are expressedas the percentage reduction of thymidine incorporation inducedby the test antibody as compared to an irrelevant control anti-body. As shown in Figure 6, all isotypes of NKI-B20 were ca-

pable of mediating ADCC with mouse macrophages. Althoughthere are small differences between the isotypes of NKI-B20,the reactivity of NKI-B20/1 and NKI-B20/2b in this assay isremarkable, since most data from the literature'29-31-32' and our

own results with the isotype switch variants of CLB-CD19"3-33'show that only the IgG2a isotype is capable of mediating ADCC

28 HOOIJBERG ET AL.

R24.3K8 (neu control)

0 100 1000mAb cone, (pg/ml)

FIG. 5. ADCC of NKI-B20 isotype variants with mouse ac-tivated NK cells as effector cells. Spleen cells from nudeBALB/c mice were cultured with rIL-2 (500 Cetus U/ml). Daudicells were sensitized with varying concentrations of NKI-B20/1,NKI-B20/2a, or NKI-B20/2b or the positive control MAb R24.3and exposed to activated spleen cells at an E/T ratio of 50.1.Specific 51Cr release was determined in triplicate wells. Thisgraph is representative of two experiments.

with macrophages. Therefore it seems that CD20 antigen can

put antibodies in a configuration conducive to interaction withFc receptors as well as to complement activation.

ADCC with human lymphocytes as effectorsPeripheral blood of normal donors was fractionated by cen-

trifugal elutriation"8' and the fraction containing the large lym-phocytes, which is enriched for NK cells carrying the FeyReceptor III (CD 16), was used for ADCC experiments. Thelarge lymphocyte fraction contained 25^0% CD3" CD56+

c 100

B- 80

ai-oh-i

60

20

.01 10 100 1000mAb cone (pg/ml)

FIG. 6. ADCC of NKI-B20 isotype variants with mouse

macrophages as effector cells. Daudi target cells were sensi-tized with varying concentrations of NKI-B20/1, NKI-B20/2a,or NKI-B20/2b or the positive control MAb R24.3 and culturedin the presence of mouse peritoneal macrophages for 3 days atan E/T ratio of 20:1. Proliferation of remaining target cells wasmeasured by incorporation of [3H]thymidine. Shown is themean percentage of specific inhibition (n = 6) calculated in re-

spect to a control MAb.

CD16+ lymphocytes and no more than 5% monocytes. As pos-itive control we used the MAb R24.3, which is a rat IgG2b MAbagainst human HLA-DR and which was previously shown tomediate ADCC with human as well as mouse effectorcells."U7) Among the B cell lines that were tested (Daudi,Ramos, Raji, Jiyoye, Bjab, and JVM-3), considerable differ-ences were found in sensitivity to NK, not antibody-mediated,killing. Daudi and Bjab cells were the most sensitive and JVM-3 and Jiyoye the least sensitive (data not shown). None of thethree isotypes of NKI-B20 mediated ADCC of any of these celllines with NK cells obtained from five donors (Fig. 7A); nei-ther did CLB-CD19/2a (data not shown). Stimulation of the ef-fector cells with rIL-2 (25-50 Cetus units/ml for 18 h) did in-crease the NK/LAK cytotoxicity without antibody, but did notinduce ADCC (above the cytotoxicity seen in the presence ofcontrol antibodies) with MAbs that were inactive with unstim-ulated cells. In addition to our murine NKI-B20 and CLB-CD 19MAbs we could also test in the same assay a chimeric mouse/hu-man yl CD20'34' and a chimeric mouse/human yl CD 19MAb,<35) kindly provided by Xoma and by Dr. H. Zola, re-

spectively. Whereas the murine IgG2b CD20 MAb 2H7 was in-active in ADCC with human NK cells like the NKI-B20 MAbs,the chimeric version of this MAb mediated up to 50% cell killof JVM-3 target cells at an E/T ratio of 100 (Fig. 7B). Thesedata agree with the published results of the murine and chimeric2H7 MAb.'34' On the other hand the chimeric CD 19 MAb FMC63 showed no activity (Fig. 7B), confirming the lack of ADCCactivity seen with this MAb with a different assay.'35' This dif-ference between chimeric MAbs with the same human yl Fcpart may, at least partly, be due to the lower antigen density ofCD 19 antigen on the JVM-3 target cell as compared with CD20antigen, because in ADCC experiments with Jiyoye target cells,which have a low expression of CD20 and CD 19, both chim2H7 and chim FMC63 were negative (data not shown).

Direct evidence for the importance of antigen density inADCC has been obtained by the use ofJY cells transfected withthe cDNA of the human CD 19 antigen. These cells have an in-creased surface expression of CD 19 antigen compared to wild-type JY cells. In ADCC assays with CLB-CD 19/2a and mouse

NK cells the level of cell kill of JY cells overexpressing CD 19antigen was considerably higher than of wild-type JY cells atthe same E/T ratios and MAb concentrations."3' Human NKcells have been identified as the cells responsible for ADCC inexperiments using an IgG3 mouse MAb.'36' From two series ofisotype switch variants, directed against HLA class I, the IgG2aMAb gave the best ADCC with human lymphocytes (the seriesdid not include an IgG^).'37' HLA class I has a very high anti-gen density on most cells.

Taken together, these data indicate that both the isotype de-pendency of the binding to the Fc receptor and the density andpossibly the structure of the target antigen determine the out-come of the effector cell/MAb/target cell interaction, and thateach of these factors can partially compensate for the other.

CONCLUDING REMARKS

The series of isotype switch variants of NKI-B20, with iden-tical fine specificity and avidity of antigen binding but differ-ing in heavy chain, has offered the opportunity to study the pa-

CD20 ISOTYPE SWITCH VARIANTS 29

B

esu

Uos

80

60H

40

20

NKI-B20/2aK8R24.3

801

20 40 60

chim 2H7chim FMC63m2H7K8R24.3

80 100E/T ratio

80 100

E/T ratio

FIG 7. ADCC with human lymphocytes as effector cells. JVM-3 target cells were incubated with MAbs (final cone 2 5 ue/ml)and large granular lymphocytes from blood of a normal donor at EAT ratios ranging from 100/1 to 3/1. Specific 51Cr release was

í,T¿,e™üPllCate r11S- (A) The P°SitiVe COntrol MAb R24-3' tne ne8ative contro1 MAb K8 and NKI-B20/2a. NKI-B20/1and NKI-B20/2b were also equal to K8; data are omitted for clarity. Data are representative for six experiments with cells fromfour donors. (B) The samegraphs, for R24.3 and K8 for scale, the murine IgG2b MAb m2H7 and its chimeric version ch2H7 andthe chimeric CD 19 MAb FMC 63. Data are representative for two experiments with cells from two donors

rameters involved in a number ofcytotoxic effector mechanismsthat may play a role in the therapeutical action of MAbs. TheCD20 MAbs were found to be unusually effective in killing tar-

get cells by two completely different effector mechanisms.NKI-B20/1 and NKI-B20/2b could engage mouse macrophagesin phagocytosis/cytotoxicity equally well as NKI-B20/2a,whereas this capacity is usually limited to IgG2a antibodies. Towhat extent this reflects the interaction with other Fc receptorscould be investigated with this set of MAbs. In addition, NKI-B20/2a and NKI-B20/2b were strongly cytolytic with humancomplement on suitable target cells. The reported lack of cyto-toxicity with human complement of many mouse MAbs mayin many instances be due to the presence on the target cells ofprotective complement regulatory factors, as we have shownhere for a number of B cell lines. Even so, on sensitive Daudicells NKI-B20/2a and NKI-B20/2b, as well as other CD20MAbs of IgG2a and IgG2b isotype (Bl, BCA-B20, 1F5, 2H7,data not shown) were cytolytic, whereas other MAbs of theseisotypes, such as CLB-CD 19/2a, were not. It is tempting tospeculate that the remarkable cytotoxic capacity ofCD20 MAbsis due to the peculiar structure of the CD20 molecule, whichmight allow close contact of the MAb with the cell surface, or

the induction of a particular configuration. Additional factorsmay be the relatively high antigen density on most B cells, andthe resistance to antigenic modulation. The lack of ADCC ac-

tivity with human NK cells is largely due to poor interaction ofthe mouse Fc regions with the human FcRIII, as shown con-

vincingly with the chimeric 2H7.The overall results obtained in vitro with the set of NKI-B20

MAbs indicate that the CD20 antigen fulfills several of the re-

quirements of a target for immunotherapy. The results of in vivoexperiments in a xenotransplantation model by us"3-38' and oth-ers'39' confirm this. Clinical trials by Press et al. using an un-

labeled'40' and radiolabeled'4" CD20 MAb and by Maloney et

alS42) using a chimeric mouse/human CD20 MAb, although stillin small groups of patients, have shown very promising results.

ACKNOWLEDGMENTS

We thank Dr. Ineke C. Slaper-Cortenbach and MarjoleinDaams, Central Laboratory of the Red Cross Blood TransfusionService, Amsterdam, for transfecting 3T3 cells with the cDNAencoding for human CD20 antigen, which was kindly providedby Dr. Brian Seed, Boston. We are grateful to Bristol-MyersSquibb Pharmaceutical Research Institute for samples of themAbs 1F5 and 2H7, to Xoma Corp. for the chimeric 2H7, andto Dr. H. Zola and Dr. G.A. Pietersz, Flinders Medical Centre,for the chimeric FMC63. We appreciated the stimulating dis-cussions with Dr. C.J.M. Melief, Leiden. E.H. and P.C.M.v.d.B.were supported by a grant from the Koningin Wilhelmina Fonds(The Dutch Cancer Society), Grant NKI 89-06.

REFERENCES

Tedder TF, and Engel P: CD20: A regulator of cell-cycle progres-sion of B lymphocytes. Immunol Today 1994;15:450-454.Bubien JK, Zhou LJ, Bell PD, Frizzell RA, and Tedder TF:Transfection of the CD20 cell surface molecule into ectopic celltypes generates a Ca2+ conductance found constitutively in B lym-phocytes. J Cell Biol 1993;121:1121-1132.Zola H: The surface antigens of human B lymphocytes. ImmunolToday 1987;8:308-315.Dörken B, Möller P. Pezzutto A, Schwartz-Albiez R, andMoldenhauer G: B cell antigens: CD20. In: Leucocyte Typing IV.White Cell Differentiation Antigens. Knapp W, Dörken B, GilksWR, Rieber EP, Schmidt RE, Stein H, and von dem Borne AEGKr(Eds.). Oxford University Press, Oxford, 1989, pp. 46-48.

30 HOOIJBERG ET AL.

5. Ryan DH: Phenotypic heterogeneity in acute leukemia. Clin ChimActa 1992;206:9-23.

6. Press OW, Farr AG, Borroz KI, Anderson SK, and Martin PJ:Endocytosis and degradation of monoclonal antibodies targetinghuman B-cell malignancies. Cancer Res 1989;49:4906-4912.

7. Vervoordeldonk SF, Merle PA, van Leeuwen EF, von dem BorneAE, and Slaper-Cortenbach IC: Preclinical studies with radiola-beled monoclonal antibodies for treatment of patients with B-cellmalignancies. Cancer 1994;73:1006-1011.

8. Rankin EM, and Hekman A: Mouse monoclonal antibodies againstthe idiotype of human B cell non-Hodgkin lymphomas: Production,characterization and use to monitor the progress of disease. Eur JImmunol 1984;14:1119-1126.

9. Boot JHA, Geerts MEJ, de Groot ER, and Aarden LA: Murinemonoclonal isotype switch variants. Detection with rat monoclonalantibodies in Elisa and isolation by sequential sublining. J ImmunolMethods 1988;106:195-202.

10. De Rie MA, Schumacher TN, van Schijndel GM, van Lier RA, andMiedema F: Regulatory role of CD 19 molecules in B-cell activa-tion and differentiation. Cell Immunol 1989;118:368-381.

11. Vuist WM, v Buitenen F, de Rie MA. Hekman A, Rümke P, andMelief CJ: Potentiation by interleukin 2 of Burkitt's lymphomatherapy with anti-pan B (anti-CD 19) monoclonal antibodies in a

mouse xenotransplantation model. Cancer Res 1989:49:3783-3788.

12. Burova GF, Miterev GI, Korshunov LI, and Bulycheva TI:Monoclonal antibodies BCA-B/20 against membrane antigens ofhuman B-cells. (Russian). Gematol Transfuziol 1990;35:28-29.

13. Hooijberg E. van den Berk PCM. Sein JJ, Wijdenes J. Hart AAM,de Boer RW, Melief CJM, and Hekman A: Enhanced antitumoreffects of CD20 over CD 19 monoclonal antibodies in a nude mouse

xenograft model. Cancer Res 1995;55:840-846.14. Stefanova I, Hilgert I, Kristofova H, Brown R, Low MG, and

Horejsi V: Characterization of a broadly expressed human leuco-cyte surface antigen MEM-43 anchored in the membrane throughphosphatidylinositol. Mol Immunol 1989;26:153-161.

15. Spits H, Keizer G, Borst J, Terhorst C, Hekman A, and de VriesJE: Characterization of monoclonal antibodies against cell surfacemolecules associated with cytotoxic activity of natural and acti-vated killer cells and cloned CTL lines. Hybridoma 1983;2:423-437.

16. Ossendorp F. Jacobs H. van der Horst G, de Vries E, Berns A. andBorst J: T cell receptor aß lacking the ß chain V domain can beexpressed at the cell surface but prohibits T cell maturation. JImmunol 1992;148:3714-3722.

17. Vuist WM, Visseren MJ, Otsen M, Bos K, Vyth-Dreese FA, FigdorCG, Melief CJ, and Hekman A: Enhancement of the antibody-de-pendent cellular cytotoxicity of human peripheral blood lympho-cytes with interleukin-2 and interferon alpha. Cancer ImmunolImmunother 1993;36:163-170.

18. Figdor CG, Bont WS, and de Vries JE: Isolation of large numbersof highly purified lymphocytes and monocytes with a modified cen-

trifugal elutriation procedure. J Immunol Methods 1981:40:275.19. Stamenkovic I, and Seed B: Analysis of two cDNA clones encod-

ing the B lymphocyte antigen CD20 (Bl, Bp35), a type III inte-gral membrane protein. J Exp Med 1988;167:1975-1980.

20. Korman AJ, Frantz JD, Strominger JL, and Mulligan RC:Expression of human class II major histocompatibility complexantigens using retrovirus vectors. Proc Nati Acad Sei USA1987;84:2150-2154.

21. Feigner PL, Gadek TR, Holm M, Roman R, Chan HW, Wenz M,Northrop JP, Ringold GM, and Danielsen M: Lipofection: A highlyefficient, lipid-mediated DNA-transfection procedure. Proc NatiAcad Sei USA 1987;84:7413-7417.

22. Dancescu M, Wu C, Rubio M, Delespesse G, and Sarfati M: IL-4

induces conformational change of CD20 antigen via a protein ki-nase C-independent pathway. Antagonistic effect of anti-CD40monoclonal antibody. J Immunol 1992;148:2411-2416.

23. Zhou LJ, and Tedder TF: CD20 workshop panel report. In:Leucocyte Typing V. White Cell Differentiation Antigens.Schlossman SF, Boumsell L, Gilks W, Harlan JM, Kishimoto T,Morimoto C, Ritz J, Shaw S, Silverstein RL, Springer TA, TedderTF, and Todd RF (eds.) Oxford University Press, Oxford, 1995,pp. 511-514.

24. Herlyn D. Herlyn M, Steplewski Z, and Koprowski H: Monoclonalanti human tumor antibodies of six isotypes in cytotoxic reactionswith human and murine effector cells. Cell Immunol 1985;92:105-114.

25. Hadam MR: Cluster report: CD59. In: Leucocyte Typing IV. WhiteCell Differentiation Antigens. Knapp W, Dörken B, Gilks WR,Rieber EP, Schmidt RE, Stein H, and von dem Borne AEGKr(Eds.). Oxford University Press, Oxford, 1989, pp. 720-722.

26. Bindon CI. Hale G, and Waldmann H: Importance of antigen speci-ficity for complement-mediated lysis by monoclonal antibodies.Eur J Immunol 1988;18:1507-1514.

27. Ravetch JV. and Kinet JP: Fc receptors (review). Annu RevImmunol 1991;9:457-492.

28. Denkers EY, Badger CC, Ledbetter JA, and Bernstein ID: Influenceof antibody isotype on passive serotherapy of lymphoma. JImmunol 1985;135:2183-2186.

29. Kaminski MS, Kitamura K. Maloney DG, Campbell MJ, and LevyR: Importance of antibody isotype in monoclonal anti-idiotype ther-apy of a murine B cell lymphoma. A study of hybridoma classswitch variants. J Immunol 1986:136:1123-1130.

30. Shiloni E. Eisenthal A. Sachs D. and Rosenberg SA: Antibody-de-pendent cellular cytotoxicity mediated by murine lymphocytesactivated in recombinant interleukin 2. J Immunol 1987; 138:1992-1998.

31. Ezekowitz RAB, Bampton M, and Gordon S: Macrophage activa-tion selectively enhances expression of Fc receptors for IgG2a. JExp Med 1983;157:807-812.

32. Adams DO, Hall T, Steplewski Z, and Koprowski H: Tumors un-

dergoing rejection induced by monoclonal antibodies of the IgG2aisotype contain increased numbers of macrophages activated for a

distinctive form of antibody-dependent cytolysis. Proc Nati AcadSei USA 1984;81:3506-3510.

33. Vuist WM, van Buitenen F. Hekman A, and Melief CJ: Two dis-tinct mechanisms of antitumor activity mediated by the combina-tion of interleukin 2 and monoclonal antibodies. Cancer Res1990;50:5767-5772.

34. Liu AY, Robinson RR, Murray ED Jr, Ledbetter JA, Hellstrom I,and Hellstrom KE: Production of a mouse-human chimeric mono-

clonal antibody to CD20 with potent Fc-dependent biologic activ-ity. J Immunol 1987;139:3521-3526.

35. Zola H, MacArdle PJ, Bradford T, Weedon H, Yasui H, andKurosawa Y: Preparation and characterization of a chimeric CD 19monoclonal antibody. Immunol Cell Biol 1991;69:411^122.

36. Ortaldo JR, Woodhouse C, Morgan AC, Herberman RB, ChereshDA, and Reisfeld R: Analysis of effector cells in human antibody-dependent cellular cytotoxicity with murine monoclonal antibod-ies. J Immunol 1987;138:3566-3572.

37. Kipps TJ, Parham P, Punt J, and Herzenberg LA: Importance ofimmunoglobulin isotype in human antibody-dependent, cell-medi-ated cytotoxicity directed by murine monoclonal antibodies. J ExpMed 1985;161:1-17.

38. Hooijberg E, Sein JJ, van den Berk PCM, Hart AAM, van der ValkMA, Kast WM, Melief CJM, and Hekman A: Eradication of largehuman B-cell tumors in nude mice with unconjugated CD20 mon-

oclonal antibodies and interleukin 2. Cancer Res 1995;55:2627-2634.

CD20 ISOTYPE SWITCH VARIANTS 31

39. Buchsbaum DJ, Wahl RL, Normolle DP, and Kaminski MS:Therapy with unlabeled and 1311-labeled pan-B-cell monoclonalantibodies in nude mice bearing Raji Burkitt's lymphomaxenografts. Cancer Res 1992;52:6476-6481.

40. Press OW, Appelbaum F, Ledbetter JA, Martin PJ, Zarling J, KiddP, and Thomas ED: Monoclonal antibody 1F5 (anti-CD20)serotherapy of human B cell lymphomas. Blood 1987:69:584-591.

41. Press OW, Eary JF, Appelbaum FR, Martin PJ, Badger CC, NelpWB, Glenn S, Butchko G, Fisher D, and Porter B: Radiolabeled-antibody therapy of B-cell lymphoma with autologous bone mar-

row support. N Engl J Med 1993;329:1219-1224.42. Maloney D, Liles TM, Czerwinski DK, Waldichuk C. Rosenberg

J, Grillo-Lopez A, and Levy R: Phase I clinical trial using esca-

lating single-dose infusion of chimeric anti-CD20 monoclonal an-

tibody (IDEC-C2B8) in patients with recurrent B-cell lymphoma.Blood 1994;84:2457-2466.

Address reprint requests to:Annemarie Hekman, Ph.D.

Division of ImmunologyThe Netherlands Cancer Institute

Plesmanlaan 1211066 CX Amsterdam, The Netherlands

Received for publication: September 8, 1995Accepted for publication: October 23, 1995