[CANCER RESEARCH 50, 4227-4232, July 15, 1990]

Targeting Human T-Lymphocytes with Bispecific Antibodies to React against

Human Ovarian Carcinoma Cells Growing in nu/nu Mice

Maria A. Garrido, Maria J. Valdayo, David F. Winkler, Julie A. Titus, Toby T. Hecht, Pilar Perez, David M. Segal,and John R. WunderlichExperimental Immunology Branch, National Cancer Institute, NIH, Bethesda, Maryland 20892 ¡M.A. G., M. J. V., D. F. W., J. A. T., P. P., D. M. S., J. R. W.], andUniversity of Maryland Baltimore County, Department of Biological Sciences, Catonsville, Maryland 21228 ¡T.T. H.]

ABSTRACT

In the present study we tested whether human T-cells from normaldonors can be targeted against human ovarian carcinoma cells and blocki.p. growth of an established tumor in immunodeficient mice. For targetingwe used chemically cross-linked bispecific monoclonal antibodies (mAbs)reacting with CD3 on the T-cells and with cell-surface antigens selectivelyexpressed by tumor cells. The tumor model consisted of mice given i.p.injections of a human ovarian carcinoma cell line, OVCAR-3, whosegrowth includes development of massive ascites. Peripheral blood lymphocytes from normal human donors were cultured overnight with 50-100 units/ml recombinant interleukin 2, coated with bispecific antibodies,and injected i.p. into mice 4-6 days after tumor inoculation, at whichtime tumor cells were established and growing in about 85% of the hosts.Tumor growth was assessed by the number of tumor cells, and in sometests by cell-free tumor antigen, recovered in peritoneal lavage fluidcollected 15 days after tumor priming. Treatment with lymphocytesretargeted with bispecific mAbs, prepared with anti-CD3 and threedifferent antitumor mAbs, IMI I, OVB-3, and MOvl9, gave highlysignificant increases in percentages of mice without detectable tumor.Controls showed that the antitumor activity of retargeted lymphocytesdid not result simply from antibody-dependent cellular cytotoxicity orfrom heteroconjugates reacting only with CD3 or with lymphocyte majorhistocompatibility complex determinants and tumor cells. These resultsshow that targeted T-lymphocytes can significantly decrease the growthof an established tumor in a fashion specific for antigens expressed bythe neoplastic cells.

INTRODUCTION

The antigen specificity of T-lymphocytes can be changed bycross-linking the TCR1 complex to target cell determinants (1-3). By engaging "triggering sites" on the lymphocytes, such asthe TCR complex, cross-linking results not only in conjugateformation between lymphocytes and target cells but also inbridging of the triggering sites and subsequent functional activation of the lymphocytes in a MHC-independent fashion.Thus, cytotoxic T-lymphocytes have been induced to lyse cellsin vitro that they normally would not lyse. Artificial binding ofeffector to target cells through TCR complexes signals thecytotoxic T-cell to deliver a "lethal hit."

The linkage of TCR to cells other than the natural target ofthe cytotoxic T-cell has been achieved in several different ways:(a) by chemically coupling a mAb against the TCR (e.g., anti-CD3) to the target cell (4, 5); (b) by using cells that expressmembrane bound anti-TCR (i.e., hybridoma cells) as targets(6-8); (c) by using targets with Fc receptors that will bind anti-TCR mAbs of the appropriate isotype (9); and (d) by using

Received 11/13/89; revised 3/9/90.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

' The abbreviations used are: TCR. T-cell receptor; MHC, major histocompatibility complex; mAb, monoclonal antibody; SDS-PAGE. sodium dodecylsulfate-polyacrylamide gel electrophoresis; SPDP, succinimidyl-3-(2-pyridyldi-thio)propionate; IL-2, interleukin 2; rIL-2, recombinant interleukin 2; PBS,phosphate-buffered saline; PBL, peripheral blood lymphocytes; ELISA, enzyme-linked immunosorbent assay: NK, natural killer; BSA, bovine serum albumin:ADCC, antibody-dependent cellular cytotoxicity; DNP. dinitrophenyl.

bispecific reagents that bind both the TCR complex and thetarget cell. Thus, mAbs that are hetero-cross-linked2 chemically(5, 10-13), bispecific antibodies that are produced by hybridhybridomas (13-16), and mAbs that are cross-linked to hormones that bind to the target cells (17) have each been usedsuccessfully to target3 T-lymphocyte cytotoxicity against se

lected tumors.An exciting possible application of targeted cytotoxicity is to

direct T-cells against selected tumors growing in vivo. Previousstudies have shown that human T-cells can be successfullytargeted with bispecific antibodies to react against human tumors in vitro (10, 12, 13, 15, 18), including ovarian carcinomacells (13, 15). In the present study we have targeted human T-lymphocytes from normal donors against a human ovariancarcinoma cell line, OVCAR-3, growing i.p. in immunodeficient mice. Nude mice given i.p. injections of the tumor developmassive ascites in a fashion similar to naturally occurringovarian cancers in humans. Using a 2-week assay for in vivotumor growth, we found that human T-lymphocytes, targetedagainst the tumor by chemically hetero-cross-linked mAbs reacting with both the tumor cells and CD3, blocked tumorgrowth.

MATERIALS AND METHODS

Mice. Pathogen-free adult female athymic nu/nu mice were obtainedfrom the NIH Frederick Cancer Research Center (Frederick, MD) andwere housed in laminar flow racks. They received autoclaved food andwater. Age-matched mice were used for each experiment; in the courseof the entire study, mouse ages at the start of experiments ranged from7 to 24 weeks.

Tumor Cells. OVCAR-3 cells were kindly provided by Drs. Ira Pastanand Thomas Hamilton (National Cancer Institute) and were passagedi.p. in nu/nu mice at the pathogen-free DCBD Animal Holding Facility(NCI-Frederick Cancer Research Facility, Frederick, MD). These cellswere originally derived from the malignant ascites of a patient withcommon epithelial ovarian cancer (19) and OVCAR-3 cells were selected from the original cell line in part for growth in nu/nu mice (20).Tumor-bearing mice from the Frederick facility were used as a sourceof tumor cells within 2 weeks of arriving at our facility. Because ofmarked clumping of ascites tumor cells, all cell counts were based onnumbers of nuclei following cell treatment with Coulter DiagnosticsZap-oglobin (Curtin Matheson Scientific, Inc., Houston, TX). Preliminary tests established that injection of at least 5 x IO6OVCAR-3 cells

i.p. into nude mice was required to induce progressive tumor growth inabout 85% of the hosts. In all experiments reported here, mice weregiven i.p. injections of 1-2 x IO7tumor cells.

Antibodies. The following mAbs were used in this study: OKT3(IgG2a; anti-CD3) (21), W6/32 (IgG2a; anti-MHC class I mono-morphic determinant) (22), 3G8 (IgGl; anti-Fc,RIII) (23), 113F1(IgG3; an anti-breast cancer antibody that cross-reacts with ovariancarcinoma cells, kindly provided by Dr. D. Ring, Cetus Corp., Emeryville, CA) (24), OVB-3 (IgG2b; an anti-ovarian carcinoma mAb, kindlyprovided by Dr. I. Pastan, National Cancer Institute) (25), and MOvl9(IgG2a; another anti-ovarian carcinoma mAb, kindly provided by Dr.

2Hetero-cross-link and heteroconjugate have been used interchangeably.3Target and retarget (redirect) have been used interchangeably.

4227

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TARGETING HUMAN T-CELLS AGAINST OVARIAN CARCINOMA CELLS

P. Dadonna, Centocor Corp., Malvern, PA) (26). OKT3, W6/32, and3G8 were purified from ascites as described (27). The antitumor mAbswere obtained in pure form. Affinity-purified rabbit anti-DNP antibodies and Fab fragments were prepared as described elsewhere (28, 29).The Fab preparations were free of undigested antibody by SDS-PAGEanalysis.

Antibody Heteroconjugates. Antibodies were cross-linked with SPDP(Pierce Chemicals, Rockford, IL), and fractionated as described (30,31). Briefly, monoclonal antibodies produced in nude mice were purifiedfrom ascities fluid by ammonium sulfate precipitation, followed by gelnitration and ion-exchange chromatography. A 4x molar excess ofSPDP in anhydrous ethanol was added to 5 mg of each antibody in0.01 M borate-buffered saline, pH 8.5. After 30 min the pH of one ofthe antibody solutions was lowered to pH 4.5 with l M sodium acetate,and bound 3-(2-pyridyldithio)propionate groups were reduced withdithiothreitol at a final concentration of 0.02 M. After 10 min thereduced protein was passed through a Pharmacia PD10 column, equilibrated with 0.1 M sodium phosphate, pH 7.4, and then immediatelyadded to the nonreduced antibody, resulting in a total volume of about2 ml. After 4 h at room temperature 2 mg of iodoacetamide was addedand cross-linked material was separated from monomeric antibodieson a 1.6- x 90-cm Ultrogel AcA34 column. The cross-linked antibodiesmainly dimers, trimers, and tetramers, were pooled as the source ofheteroconjugates. Heteroconjugates are designated as "antibody 1 xantibody 2," e.g., anti-CD3 x 113F1.

In Vitro Cytotoxicity Assays. OVCAR-3 tumor cells, freshly collectedfrom tumor-bearing nu/nu mice, were cultured for 3 days at 37°Cunder

5% COz-95% air in RPMI 1640 medium supplemented with glutamine,penicillin, streptomycin, and 5% heat-inactivated fetal bovine serum("cultured medium"). Nonadherent cells and medium were removed

and fresh medium added. After 2 additional days, adherent cells, whichwere in log-phase growth, were harvested with trypsin/EDTA (GIBCO,Grand Island, NY), labeled with 51Cr, and incubated with titered dosesof effector cells, in triplicate, for 4 h in a standard "Cr-release assay inU-bottomed microtiter wells with IO4 tumor cells each. The spontaneous release of "Cr from tumor cells incubated in medium aloneaveraged 13%. By using 2% Triton X-100 detergent lysis to determinemaximum "Cr release, percentage specific lysis was calculated as

100 xtest —spontaneous

maximum —gamma counter background

Standard errors were calculated to account for the variance of each ofthe factors used to calculate percentage of specific lysis.

A two-step "Cr release assay for complement-mediated lysis wascarried out by incubating the OVCAR-3 tumor cells with titered concentrations of purified antibodies or heteroconjugates for 45 min at4°Cin RPMI 1640 with 10% fetal bovine serum followed by washingand a 45-min incubation at 37°Cwith rabbit complement (Low-tox-M

Rabbit Complement, Cedarlane Laboratories, Hornby, Ontario, Canada) diluted '/io in RPMI 1640.

Isolation and Culturing of Human Peripheral Blood Lymphocytes.Leukocyte-enriched cells were collected by leukapheresis of healthyblood bank donors. Mononuclear cells were separated by Ficoll-Hy-paque density gradient centrifugation, and lymphocytes were depletedof monocytes by Sepracell-MN density gradient (32) (Sepratech Co.,Oklahoma City, OK), by adherence to plastic (45 min, 37°C)followed

by nylon wool adsorption (33), or by carbonyl iron (34). A yield ofabout 2x10' lymphocytes was obtained from each donor. Lymphocytes

were cultured overnight in standard culture medium (see above) containing 50-100 units/ml of human rIL-2 (Cetus). After overnightculture in rIL-2, lymphocytes were transferred to rIL-2-free mediumand cultured until needed (up to 36 h later). Whether exposure oflymphocytes to rIL-2 was necessary for targeted cytotoxicity was nottested. Lymphocytes from only one donor were used for each experiment.

In Vivo Antitumor Assay. On Day 0, groups of 5-10 mice wereinjected i.p. with 1-1.5 x IO7freshly explanted ovarian carcinoma cells

in 0.5 ml of PBS. Between days 4 and 6, mice were treated with PBL±heteroconjugate or control antibodies. They were given four equaldoses over the 2-day period, IO7PBL/dose in 0.5 ml of PBS. In 14 of

18 experiments mice were treated on days 5 and 6, and in the otherexperiments on days 4 and 5. The two treatment schedules yieldedequivalent results. PBL (5 x 10' cells per 0.25 ml in PBS) used to treat

mice were preincubated for 1 h on ice with or without 100 ^g ofbispecific antibody and diluted to 2 x IO7cells/ml in PBS. Under these

conditions, the heteroconjugate was saturating, as indicated by preliminary in vitro retargeting tests. Cells (1 x IO7, 0.5 ml of PBS) were

injected into mice without removing unbound antibody. On day 15mice were sacrificed, and the peritoneal cavities were lavaged with threesequential 5-ml washes with PBS (8-ml disposable, large bulb plastictransfer pipets; PGC Scientifics, Gaithersburg, MD). In preliminarytests this procedure removed about 75% of 10" tumor cells injected i.p.

into nu/nu mice 0.5 h earlier. In some experiments, the initial 0.5 mlrecovered with the first 5-ml peritoneal lavage was centrifuged andsupernatants were stored at —20°Cfor later ELISA tests for tumorantigen. The tumor burden was scored as "plus" or "minus" aftercentrifuging the lavage fluid in a 15-ml conical tube. Lavage fluid pelletsscored as negative had only traces of cells, which microscopically werenot tumor cells. Tumor-positive animals had large pellets ranging from0.2 to 1.0 ml. OVCAR-3 tumor cells were clumped and clearly largerthan mouse cells. The morphological difference was confirmed byfailure of the mouse cells to stain with fluoresceinated W6/32 (seeabove), which reacts with human MHC class I antigens on the OVCAR-3 cells. Half of the 18 experiments reported here were read in a formaldouble-blind fashion. Results from the two blocks of experiments wereequivalent. Mice did not show signs of graft versus host disease, perhapsbecause human anti-mouse T-cell responses use human but not mouseantigen processing cells.4

Data Analysis. Each immunotherapy experiment with nu/nu miceincluded groups of tumor-injected mice that were (a) untreated, (b)treated with lymphocytes alone, and (c) treated with lymphocytescombined with hetero-cross-linked mAbs or control antibody preparations.

Results from each experiment had to meet two predetermined criteriain order to be added to the data pool. First, to ensure adequacy of thetumor preparation used for mouse inoculations, the incidence of tumorgrowth in the tumor-alone group within a given experiment could notbe significantly less than in our historical control (86% tumor takes)with p < 0.05. Thus, in an experiment where 10 mice were given tumoralone, at least seven mice had to develop tumor. Only one of 24experiments was excluded on this basis. Second, the incidence of tumorgrowth in mice treated with lymphocytes alone (i.e., without bispecificantibody), could not be significantly less than in our untreated historicalcontrol (86% tumor takes) with p < 0.05. This criterion excludedexperiments in which the donor's lymphocytes had high levels of NK-

like activity that would obscure additional activity resulting from retargeting with hetero-cross-linked antibodies. The minimum acceptableincidence of tumor growth associated with lymphocyte treatment wasdetermined by a Fisher exact test relative to the historical incidence oftumor growth in untreated mice. For example, at least three of fivemice treated with lymphocytes alone had to develop tumor. Five of 23experiments were excluded on this basis, which may relate to culturingthe PBL overnight in medium containing 50-100 units/ml human rlL-2, prior to their use in vivo. Within the remaining experiments thatwere accepted, the cumulative incidence of tumor in a treatment group,such as mice injected with lymphocytes retargeted with hetero-cross-linked antibodies, was compared to the cumulative incidence of tumorin mice treated with lymphocytes alone. Comparisons were done by astandard \2 test.

ELISA Assay. Assays for cell-free ovarian carcinoma tumor antigen,CA-125, were carried out as follows. Fifty /¿'of purified OC-125 (10Mg/ml in PBS + 0.2% azide), a mAb against CA-125 (35), were addedto the wells of Immulon 2, flat-bottom microtiter plates (DynatechLabs, Inc., Chantilly, VA), which were then kept at 4°Cfor 24 h to 1

week. The OC-125 antibody was a kind gift from the Centocor Corp.Unbound antibody was removed, followed by addition of 5% BSA inPBS to block remaining adsorption sites on the plastic. After l h atroom temperature, plates were washed 3 times with 0.01 M Tris-HCl,pH 8.0, 0.05% Tween 20, and 0.01% azide. Dilutions of test samples

4 R. Gress, personal communication.

4228

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TARGETING HUMAN T-CELLS AGAINST OVARIAN CARCINOMA CELLS

were added in 50 Mlof 10 mM Tris-buffered saline-0.1% BSA-0.05%azide, pH 8.0 (dilution buffer), and plates were kept overnight at 4°C.

The following morning, plates were washed 4 times, and a predetermined dilution of OC-125-alkaline phosphatase conjugate was addedin 50 n\ dilution buffer. After 2 h at room temperature, plates werewashed 5 times and alkaline phosphatase substrate (p-nitrophenylphosphate. Sigma 104; Sigma Chemical Co., St. Louis, MO) was added.The intensity of the resulting yellow color was determined spectropho-tometrically after it developed to a readable level (36). Standard curveswere generated using 10-160 ng of partially purified tumor antigen,CA-125, kindly provided by Centocor.

OC-125-Alkaline Phosphatase Conjugates. OC-125 and alkalinephosphatase (Sigma) were cross-linked using SPDP as described elsewhere (31). Briefly, alkaline phosphatase and OC-125 in borate-buff-ered saline, pH 8.5, were each treated with a 4-fold molar excess ofSPDP. OC-125 was reduced with dithiothreitol, passed through aPharmacia PD10 column, and added to 3-(2-pyridyldithio)propionate-bearing alkaline phosphatase. After 4-h incubation at 37°C,the cross-

linking reaction was stopped with iodoacetamide. The cross-linkedmaterial was separated from the monomeric fraction on an UltrogelAcA 34 column. The protein-containing fractions eluting ahead ofmonomeric IgG were pooled, and after preliminary titration to identifythe optimal concentration, the stock was used at about a 1/100-folddilution in the ELISA assay.

RESULTS

Targeting T-Lymphocytes against Ovarian Carcinoma Cells inVitro. Bispecific antibody containing anti-CD3 (OKT3) cross-linked to the antitumor antibody 113F1, which reacts withOVCAR-3, was tested for its ability to target human PBLagainst OVCAR-3 cells in vitro. Cytotoxic activities of targetedlymphocytes from three donors tested in three independentexperiments are presented in Fig. 1. Lymphocytes withoutantibody caused significant amounts of cytotoxicity, presumably a result of NK-like activity. In each experiment targetinglymphocytes with the bispecific antibody markedly increasedthe level of cytotoxicity from that of lymphocytes alone. Theincreased level of cytotoxicity did not result simply fromADCC, as apparent from the effectiveness of hetero-cross-linked Fab preparations of OKT3 and 113F1 antibodies (Fig.\A) and from the failure of 113F1 alone or cross-linked to anti-MHC antibodies to enhance antitumor activity (Fig. l B). OKT3alone or cross-linked to an irrelevant antibody, anti-DNP, alsofailed to enhance antitumor activity (Fig. l, B and Q.

We also tested the heteroconjugates for complement-mediated lysis of OVCAR-3 target cells in a standard two-step51Cr release assay, using rabbit complement. The antitumor

antibody 113F1 and heteroconjugates between intact OK.T3and 113F1 antibodies mediated lysis, with half-maximal lysisoccurring between 0.1 and 1.0 Mg/ml. Hetero-cross-linked Fabpreparations of OKT3 and 113F1 antibodies did not promotecomplement-mediated lysis, testing up to 10 Mg/ml. By contrast,113F1 cross-linked to anti-MHC antibody (W6/32) had thehighest lytic titer among the preparations.

Growth of OVCAR-3 in nu/nu Mice. To determine the i.p.growth rate of OVCAR-3 tumor cells in nu/nu mice, we gavemice i.p. injections of 1 x IO7 tumor cells and assessed tumor

growth by peritoneal lavage and cell counts on subsequent daysthereafter. Three or four mice were used for each time point.As shown in Fig. 2, which is a composite of two independentexperiments, tumor growth was apparent by day 2 and clearlyestablished by day 4 with an average tumor recovery of 1.8 ±0.2 x IO7 cells at that time. By day 15, the average tumorrecovery was 2.9 ±0.3 X IO8 cells. ELISA tests of the initial

lavage fluid collected on day 15 showed a median level of >5000

60

40 .

20

60

40

20 .

80

60 .

40

20.

40

BT RATIOFig. 1. Retargeting T-lymphocytes against OVCAR-3 tumor cells in vitro. In

three separate experiments 25 jig of hetero-cross-linked mAbs were used topretreat 5 x IO7effector cells in 0.25 ml of PBS (1 h, 4'C). Effector cells werethen washed and added to "Cr-labeled target cells. A, no mAb (O); OKT3 x113F1, intact mAbs (A); OKT3 x 113FI, Fab mAbs (A). B, no mAb (O); OKT3x 113F1, intact mAbs (A); W6/32 x 113F1, intact mAbs (•);OKT3 + 113F1,intact mAbs. not cross-linked (O). C, no mAb (O); OKT3 x 113F1, intact mAbs(A); OKT3 x anti-DNP, intact mAbs (•).Bars, SEM.

O 10 20

DAYS

Fig. 2. Intraperitoneal growth of OVCAR-3 cells in nu/nu mice. Mice weregiven injections of I x III tumor cells on day 0 and subsequent growth wasassessed by peritoneal lavage and microscopic cell counts as described in "Materials and Methods" (in vivo antitumor assay). Although the average incidence ofmice without detectable tumor among those inoculated with 1 x 10' OVCAR-3

cells was 14% overall, only I of the 25 mice in this test was tumor free and thuswas excluded from the calculations.

ng/ml CA-125 among 27 tumor-bearing mice in four experiments. The lowest level was 4850 ng/ml. Serum CA-125 levelsof eight tumor-bearing mice on day 15 were much lower, (50-350 ng/ml) in accord with the results reported by Hamilton etal. (20).

To identify a rapid, reliable assay for OVCAR-3 cells inlavage fluid from the peritoneal cavity, we compared the size ofcell pellets following centrifugation, cytology of the cell pellets,and results of ELISA assays for the CA-125 tumor antigen inthe lavage fluid. Among 49 tumor-primed mice, which had beeneither treated or not treated with targeted lymphocytes, 30 weretumor positive by visibly obvious cell pellets (range, 0.2-1.0

4229

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TARGETING HUMAN T-CELLS AGAINST OVARIAN CARCINOMA CELLS

ml), positive cytology, and CA-125 levels of >4500 ng/ml.Nineteen mice were tumor negative by three criteria: negativecytology, only a trace of cells in the cell pellet, and a CA-125level of <225 ng/ml. In fact, all but one of these tumor-negativemice had CA-125 levels of <25 ng/ml. One mouse that wastumor negative by cytology and lack of a cell pellet had a lavagewash CA-125 level of 225 ng/ml. Thus, by three criteria, cellpellet, cytology, and CA-125 levels, the mice were clearly eithertumor-positive or tumor-negative. There were no mice thatwere borderline tumor positive. In the remaining tests we judgedthe presence of i.p. tumor by looking at the size of the cellpellet in centrifuged lavage fluid and by microscopic examination of the centrifuge sediment.

i.p. Life Span of Retargeted Lymphocytes. To test the functional life span of retargeted lymphocytes, aliquots of lymphocytes were pretreated simultaneously with two different heter-oconjugates: anti-CD3 (Fab) x anti-DNP (Fab) and anti-Fc^RIII (Fab) x anti-DNP (Fab), which retargeted T-cells (11)and NK/K-cells (37), respectively. Retargeted lymphocytes (1x IO8) were injected i.p. into normal nu/nu mice and were

recovered by peritoneal lavage at time points up to 3 daysthereafter. Different groups of mice were used for each timepoint. Data presented in Table 1 show that only 25% of theinjected PBL were recovered 30 min later. Using the 30-minrecovery of human cells as a baseline, the half-life of retrievablehuman lymphocytes was less than 24 h. Moreover, the abilityof recovered lymphocytes to lyse TNP:EL4 cells also decreasedwith a half-life of less than 24 h. This decrease in activityappeared to result from the loss of heteroconjugates from thePBL surfaces in the peritoneal cavity, because activity returnedto expected levels when heteroconjugates were added to the invitro assay or injected i.p. 30 min prior to the peritoneal lavageprocedure (Table 1). Interestingly, retargeted lymphocytesmaintained in culture did not lose their activity. We have notidentified the cause of in vivo loss of heteroconjugates fromPBL.

Because of the relatively short functional life span of retargeted human lymphocytes in the peritoneal cavity, we electedto treat tumor-bearing mice with multiple injections of relatively low numbers of retargeted effector cells rather than witha single injection of a large number of cells. We treated micewith four injections between days 4-6, the total number of cellsper injection being limited by the number of lymphocytes obtained by leukapheresis from a given donor.

Targeting T-Lymphocytes against Established Ovarian Carcinoma Cells in Vivo. A compilation of results from 18 experi

ments involving in vivo retargeting of human T-cells againstOVCAR-3 is presented in Table 2. The combined results of 18experiments with lymphocytes in the absence of heteroconjugates showed a borderline level of protection, 20% mice withoutdetectable tumor relative to 14% for untreated mice (p = 0.07).In four experiments, we tested the heteroconjugate, anti-CD3x 113F1, in the absence of lymphocytes, and it failed to protectagainst tumor growth.

By contrast, anti-CD3 x 113F1 and two other hetero-cross-linked mAbs (anti-CD3 x OVB-3 and anti-CD3 x MOvl9),were each able to provide significantly greater antitumor protection than lymphocytes alone. This observation held whetherthe lymphocyte alone control was the composite for all 18experiments or only those experiments involving the particularhetero-cross-linked mAbs under consideration. For example,

mice without detectable tumor constituted 56% of those treatedwith lymphocytes targeted with anti-CD3 x OVB-3 in sixexperiments, compared to 20% of mice treated with lymphocytes alone in all 18 experiments or 18% of mice in the sixexperiments involving the anti-CD3 x OVB-3 hetero-cross-linked mAbs.

A composite of controls for in vivo retargeting of lymphocytesis presented in Table 3. Hetero-cross-linked mAbs preparedfrom Fab antibodies enhanced the level of tumor protectionwith lymphocytes, increasing the percentage of mice withoutdetectable tumor from 22 to 67%. This finding established thatADCC is not necessary for in vivo retargeting with hetero-cross-linked mAbs. Further controls showed that significant antitu-mor activity did not occur if (a) heteroconjugates bound to CD3without binding to the tumor (anti-CD3 x anti-DNP), (b)heteroconjugates cross-linked lymphocytes to tumor cells without binding to CD3 (anti-MHC x 113F1), or (c) cells weretreated with a mixture of non-cross-linked anti-CD3 and anti-tumor antibodies.

DISCUSSION

The object of this study was to determine if retargeted humanT-cells can block the growth of an established human tumor inan in vivo model. The results presented here show for the firsttime that human T-lymphocytes from normal donors can beretargeted with hetero-cross-linked antibodies to abrogategrowth of established human ovarian carcinoma cells growingi.p. in nu/nu mice (Tables 2 and 3). Staerz and Bevan havepreviously reported that survival times of mice bearing a system-ically growing murine lymphoma, could be increased by admin-

Table I i.p. loss of cytotoxic activity of antibody-coated cellsCytoloxic activity retargetedcells*From

i.D.lavaee(h)0.54244872coveredx10~702.62.52.21.30.65Human cells recovered x10~'2.51.80.750.160.02—cross-link37

±253±311

±2C2±3"4±

1-Hcross-link47

±569±671

±160±235

±2From

culture-cross-link53

±447±359±252

±246±3-(-cross-link54

±553±579

±265±262±4

" 10" human PBL incubated with anti-T3 (Fab) x anti-DNP (Fab) and anti-Fc RIII (Fab) x anti-DNP (Fab) (in 0.5 ml with 10 ¡¡%of each antibody) were injectedtogether i.p. into nu/nu mice (two mice/time point) at time 0. At the indicated times, mice were sacrificed and lavaged with 5-ml washes (2x). Recovered cells werewashed and resuspended in culture medium, and total cells counts were taken. The percentage of human cells was determined by staining with fluoresceinisothiocyanate-W6/32.

* Cells were tested for the ability to lyse TNP-EL4 cells at an effectorttarget ratio of 25:1. Indicated values are percentage of specific lysis ±SEM. Samples fromi.p. lavage were tested in the absence or in the presence of additional cross-linked antibodies (1 HI; ml each) added to the assay. In parallel to the i.p. experiments,cells coated with heteroconjugates were washed, then cultured, and finally tested for the ability to lyse TNP-EL4 (an H-2b murine lymphoma) (42) cells in the presenceor absence of additional cross-linked antibodies.

' Cytotoxic activity increased from 11 to 44% lysis if 100 ¿igof each cross-linked antibody were injected i.p. 30 min prior to peritoneal lavage at 24 h.* Cytotoxic activity increased from 2 to 45% lysis if only anti-CD3 (Fab) x anti-DNP (Fab) (1 ng/ml) was added to the in vitro assay.

4230

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TARGETING HUMAN T-CELLS AGAINST OVARIAN CARCINOMA CELLS

Table 2 Inhibitory effects of retargeted human PBL on growth of OVCAR-3 cellsin nu/nu mice

% ofCross-linked No. of tumor-free Total no.

Group123456antibodyPBLNone

0Anti-CD3x0113F1None

+Anti-CD3x+113F1Anti-CD3

x+OVB-3Anti-CD3

x+MOV-19tests184181665mice14720595645of

mice18227112994029P0.2°0.07°<10"**10-'*0.005*

°Compare to Group 1.* Compare to Group 3.

Table 3 Controls for the effects of retargeted human PBL on growth of OVCAR-3cells in nu/nu mice

%of

Group1234567bodyPBLNone

+Anti-CD3(Fab) x+113F1

(Fab)None+Anti-CD3

x+113F1Anti-MHC

x+113F1''Anti-CD3

x ami-+DNP'Anti-CD3

+ +tests"3366556mice22672163302533of

mice18183839302839PControl0.007*Control0.0003r0.4C0.7C0.2C

°Test groups (e.g.. Groups 4-7) were included in the same experiments as

controls (e.g.. Group 3)."Compare to Group 1. In the same three experiments PBL tested with anti-

CD3 x 113F1 prepared with intact antibodies, resulted in 12 of 18 mice tumorfree.

' Compare to Group 3.¿Cross-linked antibody was active insofar as it promoted conjugate formation

between OVCAR-3 tumor cells and human lymphocytes precultured overnight inmedium with 50 units/ml rIL-2.

' Cross-linked antibody was active insofar as it mediated lysis of trinitrophenyl-

coated target cells by human lymphocytes precultured overnight in medium with50 units/ml rIL-2.

^ Non-cross-linked antibodies were used at the same concentrations present inthe cross-linked preparation.

¡stratumof alloactivated spleen cells and a hybrid-hybridomaantibody that recognized both murine TCR and the lymphomacells (1). Thus, in two different systems, retargeted T-cells canact against established tumors in vivo.

In the current study, three factors were associated with successful retargeting of T-lymphocytes (Table 3): (a) the antitu-mor and antilymphocyte antibodies had to be cross-linked. Theywere ineffective when added as separate antibodies; (b) hetero-cross-linked antibodies had to react with triggering sites on thelymphocytes. Those that reacted with the tumor cells but notwith triggering sites on the lymphocytes (e.g., anti-MHC xantitumor) were ineffective; (c) hetero-cross-linked antibodieshad to react with the tumor cells. Those that reacted withlymphocyte triggering sites but not with the tumor cells (e.g.,anti-CD3 x anti-DNP) were also ineffective. These observationsparallel earlier in vitro findings that only heteroconjugatescontaining antitarget antibody cross-linked to a triggering molecule on the cytotoxic cell would induce lysis (10, 13, 29).

PBL contain potent ADCC effector cells. Moreover, the Fabantibody heteroconjugates resulted in less lytic activity in vitrothan intact antibody conjugates (Fig. 1). We feel, however, thatthis difference resulted primarily from chemical preparation ofthe conjugates rather than from ADCC and that ADCC and/or complement-mediated cytotoxicity had a minor role in vivofor the following reasons. The observation that heteroconjugate

prepared from Fab fragments induced antitumor activity (Table3) clearly showed that much of the effectiveness of targetedcells was not due to ADCC. Antitumor activity of the heteroconjugate prepared from the Fab fragments could not be attributed to contaminating undigested antibody: the Fab heteroconjugate failed to activate complement-mediated lysis in vitro,and undigested antibodies were not detected by SDS-PAGEanalysis. Moreover, 113F1 antitumor antibody used withoutcross-linking or used as a heteroconjugate with anti-MHC classI antibody (W6/32) failed to protect against tumor growth(Table 3); of note, both of these preparations were cytotoxic forOVCAR-3 in a complement-mediated reaction. One reasonthat we did not see significant antitumor effects with theseantibodies, which would be expected to promote ADCC and/or complement-mediated lysis, might have been because weexcluded tests in which lymphocytes had detectable levels ofNK-like activity in vivo. NK cells were well known to be a majorsource of ADCC activity (38, 39).

In the current study, the cells initiating the antitumor response were most probably T-cells, since the response wasdependent upon heteroconjugates that had specificity for human CD3. Because few functional retargeted T-cells remainedwithin the peritoneal cavity for more than a day (Table 1), therewas little opportunity for those cells to either proliferate ordifferentiate prior to mediating their antitumor activity. As wewill show in a later paper,5 all of the cytotoxic activity in 5'Cr-release assays, caused by retargeted T-cells within human PBLthat have been briefly stimulated with IL-2, lies within a CD3+,CD8+ subset of large granular lymphocytes, a fraction repre

senting about 5% of the total lymphocyte population. It istherefore likely that this small subset of cells is responsible forthe antitumor effects seen in Tables 2 and 3. While these cellsmust have remarkably high activity on a per cell basis to givethe observed results, significant numbers of treated mice in ourexperiments still get tumor. How the survival rates of OVCAR-3-bearing mice treated with retargeted T-cells will comparewith the survival rates of mice treated with OVB-3/pseudo-monas toxin immunoconjugate (25) or with IL-2-activated NKand T-cells (40) remains to be established. However, largeamounts of untapped cytotoxic activity can be generated fromthe remaining 90% of T-cells in PBL by treating them forseveral days with a TCR cross-linking agent and IL-2. Inaddition, K cells (37), monocytes (41), and neutrophils (41)may all be targeted against OVCAR-3 by using anti-Fc,R-containing heteroconjugates. By bringing all of these targetableactivities to bear against established i.p. ovarian carcinoma, weexpect to be able to further enhance the effectiveness of thistreatment.

ACKNOWLEDGMENTS

We are grateful for support provided by staff of the NIH Departmentof Transfusion Medicine and by animal care personnel of the NationalCancer Institute Office of Laboratory' Animal Science, Laboratory

Animal Resources Section. We thank Dr. Ronald Cress for criticallyreviewing the manuscript. We also thank Drs. Peter Dadonna, IraPastan, and David Ring for generously providing antitumor monoclonalantibodies and for comments on the manuscript.

REFERENCES

1. Staerz, U. D., and Bevan, M. J. Use of anti-receptor antibodies to focus Tcell activity. Immunol. Today, 7: 241, 1986.

2. Segal. D. M., and Wunderlich, J. R. Targeting of cytotoxic cells with

5M. A. Garrido et al., manuscript submitted for publication.

4231

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

TARGETING HUMAN T-CELLS AGAINST OVARIAN CARCINOMA CELLS

heterocrosslinked antibodies. Cancer Invest., 6: 83-92. 1988.3. Segal. D. M.. and Snider. D. P. Targeting and activation of cytotoxic 23.

lymphocytes. Chem. Immunol., 47: 179-213, 1989.4. Staerz, U. D., Kanagawa. O.. and Bevan, M. J. Hybrid antibodies can target

sites for attack by T cells. Nature (Lond.). 314: 628-631. 1985. 24.5. Kranz, D. M., Tonegawa, S., and Eisen. H. N. Attachment of an anti-receptor

antibody to non-target cells renders them susceptible to lysis by a clone ofcytotoxic T lymphocytes. Proc. Nati. Acad. Sci. USA. 81: 7922-7926. 1984. 25.

6. Erti. H. C.. Greene. M. I.. Noseworthy. J. H.. Fields. B. N.. Nepom, J. T.,Spriggs. D. R.. and Finberg. R. W. Identification of idiotypic receptors onreovirus-specific cytolytic T cells. Proc. Nati. Acad. Sci. USA, 79: 7479-7483. 1982. 26.

7. Lancki. D. W., Ma, D. I.. Havran, W. L.. and Fitch. F. W. Cell surfacestructures involved in T cell activation. Immunol. Rev., 81: 65-94, 1984.

8. Hoffman. R. W., Bluestone. J. A.. Leo, O., and Shaw, S. Lysis of anti-T3-bearing murine hybridoma cells by human allospecific cytotoxic T cell clonesand inhibition ofthat lysis by anti-T3 and anti-LFA-1 antibodies. J. Immu- 27.noi., 135: 5-8, 1985.

9. Leeuwenberg. J. F., Spits, H., Tax, W. J., and Capel, P. J. Induction ofnonspecific cytotoxicity by monoclonal anti-T3 antibodies. J. Immunol.. 134: 28.3770-3775. 1985.

10. Liu, M. A., Kranz. D. M., Kurnick, J. T., Boyle, L. A., Levy, R., and Eisen,H. N. Heteroantibody duplexes target cells for lysis by cytotoxic T lympho- 29.cytes. Proc. Nati. Acad. Sci. USA, 82: 8648-8652. 1985.

11. Perez. P., Hoffman. R. W., Shaw, S., Bluestone. J. A., and Segal, D. M.Specific targeting of cytotoxic T cells by anti-T3 linked to anti-target cell 30.antibody. Nature (Lond.), 316: 354-356. 1985.

12. Jung, G.. Honsik, C. J., Reisfeld, R. A., and Muller-Eberhard, H. J. Activation of human peripheral blood mononuclear cells by ami M: killing oftumor target cells coated with anti-target-anti-T3 conjugates. Proc. Nati. 31.Acad. Sci. USA, 83: 4479-4483. 1986.

13. Mezzanzanica, D.. Canevari, S.. Menard. S.. Pupa, S. M.. Tagliabue. E..Lanzavecchia. A., and Colnaghi, M. 1. Human ovarian carcinoma lysis bycytotoxic T cells targeted by bispecific monoclonal antibodies: analysis of the 32.antibody components. Int. J. Cancer. 41: 609-615, 1988.

14. Staerz, U. D., and Bevan. M. J. Hybrid hybridoma producing a bispecificmonoclonal antibody that can focus effector T-cell activity. Proc. Nati. Acad. 33.Sci. USA. S3: 1453-1457. 1986.

15. Lanzavecchia. A., and Scheidegger. D. The use of hybrid hybridomas totarget human cytotoxic T lymphocytes. Eur. J. Immunol., 17:105-111. 1987. 34.

16. i.illil.md. L. K., Clark. M. R.. and Waldmann. H. Universal bispecificantibody for targeting tumor cells for destruction by cytotoxic T cells. Proc.Nati. Acad. Sci. USA. «5:7719-7723, 1988. 35.

17. Liu, M. A., Nussbaum, S. R.. and Eisen. H. N. Hormone conjugated withantibody to CD3 mediates cytotoxic T cell lysis of human melanoma cells.Science (Wash. DC). 239: 395-398. 1988.

18. Perez, P., Titus, J. A., Lotze, M. T., Cuttitta, F., Longo, D. L., Groves, E. 36.S., Rabin, H., Durda. P. J., and Segal, D. M. Specific lysis of human tumorcells by T cells coated with anti-T3 cross-linked to anti-tumor antibody. J. 37.Immunol.. 137: 2069-2072. 1986.

19. Hamilton. T. C.. Young. R. C, McKoy. W. M., Grotzinger, K. R., Green. J.A.. Chu, E. W., Whang-Peng, J., Rogan. A. M.. Green, W. R.. and Ozols.R. F. Characterization of a human ovarian carcinoma cell line (NIH:OVCAR- 38.3) with androgen and estrogen receptors. Cancer Res., 43: 5379-5389. 1983.

20. Hamilton, T. C., Young, R. C., Louie, K. G., Behrens, B. C., McKoy, W. 39.M., Grotzinger. K. R.. and Ozols. R. F. Characterization of a xenograftmodel of human ovarian carcinoma which produces ascites and intraabdom- 40.inal carcinomatosis in mice. Cancer Res., 44: 5286-5290, 1984.

21. Kung, P.. Goldstein. G., Reinherz. E. L., and Schlossman. S. F. Monoclonalantibodies defining distinctive human T cell surface antigens. Science (Wash. 41.DC), 206: 347-349, 1979.

22. Barnstable, C. J., Bodmer, W. F., Brown, G., Galfre, G., Milstein, C.,Williams, A. F., and Ziegler, A. Production of monoclonal antibodies to 42.group A erythrocytes. HLA and other human cell surface antigens-new tools

for genetic analysis. Cell, 14: 9-20, 1978.Fleit, H. B.. Wright, S. D., and Unkeless, J. C. Human neutrophil Fc gammareceptor distribution and structure. Proc. Nati. Acad. Sci. USA. 79: 3275-3279. 1982.Frankel. A. E.. Ring, D. B., Tringale. F., and Hsieh-Ma. S. T. Tissuedistribution of breast cancer-associated antigens defined by monoclonal antibodies. J. Biol. Response Modif., 4: 273-286, 1985.Willingham. M. C.. FitzGerald, D. J.. and Pastan. I. Pseudomonas exotoxincoupled to a monoclonal antibody against ovarian cancer inhibits the growthof human ovarian cancer cells in a mouse model. Proc. Nati. Acad. Sci. USA,84: 2474-2478. 1987.Miotti. S., Canevari, S.. Menard, S., Mezzanzanica. D.. Porro, G., Pupa. S.M.. Regazzoni, M., Tagliabue. E.. and Colnaghi, M. I. Characterization ofhuman ovarian carcinoma-associated antigens defined by novel monoclonalantibodies with tumor-restricted specificity. Int. J. Cancer, 39: 297-303,1987.Unkeless, J. C. Characterization of a monoclonal antibody directed againstmouse macrophage and lymphocyte Fc receptors. J. Exp. Mcd., 150: 580-596. 1979.Segal, D. M., and Hurwitz, E. Dimcrs and trimers of immunoglobulin Gcovalently cross-linked with a bivalent affinity label. Biochemistry, IS: 5253-5258. 1976.Perez. P.. Hoffman. R. W., Titus, J. A., and Segal. D. M. Specific targetingof human peripheral blood T cells by heteroaggregates containing anti-T3crosslinked to anti-target cell antibodies. J. Exp. Med., 163: 166-178, 1986.Karpovsky. B.. Titus, J. A.. Stephany, D. A., and Segal. D. M. Productionof target-specific effector cells using hetero-cross-linked aggregates containing anti-target cell and anti-Fc gamma receptor antibodies. J. Exp. Med.,760: 1686-1701. 1984.Titus, J. A., Garrido, M. A., Hecht, T. T.. Winkler. D. F.. Wunderlich, J.R., and Segal. D. M. Human T cells targeted with anti-T3 cross-linked toantitumor antibody prevent tumor growth in nude mice. J. Immunol.. 138:4018-4022. 1987.Vissers, M. C., Jester. S. A., and Fantone, J. C. Rapid purification of humanperipheral blood monocytes by centrifugation through Ficoll-Hypaque andSepracell-MN. J. Immunol. Methods.. 110: 203-207. 1988.Trizio, D.. and Cudkowicz. G. Separation of T and B lymphocytes by nylonwool columns: evaluation of efficacy by functional assays in vivo.J. Immunol..113: 1093-1097, 1974.Lohrmann. H. P.. Novikovs. L.. and Graw. R. G., Jr. Cellular interactionsin the proliferative response of human T and B lymphocytes to phytomitogensand allogeneic lymphocytes. J. Exp. Med.. 139: 1553-1567, 1974.Klug. T. L., Bast. R. C., Jr., Niloff. J. M., Knapp, R. C., and Zurawski, V.R., Jr. Monoclonal antibody immunoradiometric assay for an antigcnicdeterminant (CA 125) associated with human epithelial ovarian carcinomas.Cancer Res., 44: 1048-1053, 1984.Engvall. E. Enzyme immunoassay ELISA and EMIT. Methods Enzymol.,70:419-439. 1980.Titus, J. A., Perez. P.. Kaubisch. A.. Garrido. M. A., and Segal, D. M.Human K natural killer cells targeted with hetero-cross-linkcd antibodiesspecifically lyse tumor cells in vitro and prevent tumor growth in vivo. J.Immunol.. 139: 3153-3158. 1987.Herbcrman. R. B.. and Holden, H. T. Natural cell-mediated immunity. Adv.Cancer Res., 27: 305-377. 1978.Trinchieri. G. Biology of natural killer cells. Adv. Immunol.. 47: 187-376,1989.Ortaldo, J. R.. Porter, H. R.. Miller. P., Stevenson. H. C.. Ozols. R. F.. andHamilton, T. C. Adoptive cellular immunotherapy of human ovarian carcinoma xenografts in nude mice. Cancer Res.. 46: 4414-4419, 1986.Graziano, R. F.. and Fanger, M. W. Human monocyte-mediated cytotoxicity:the use of Ig-bearing hybridomas as target cells to detect trigger moleculeson the monocyte cell surface. J. Immunol., 138: 945-950, 1987.Shearer. G. M. Cell-mediated cytotoxicity to trinitrophenyl-modified synge-neic lymphocytes. Eur. J. Immunol.. 4: 527-533. 1974.

4232

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1990;50:4227-4232. Cancer Res   Maria A. Garrido, Maria J. Valdayo, David F. Winkler, et al.  

Micenu/nuReact against Human Ovarian Carcinoma Cells Growing in Targeting Human T-Lymphocytes with Bispecific Antibodies to

  Updated version

  http://cancerres.aacrjournals.org/content/50/14/4227

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  .permissions@aacr.orgDepartment at

To request permission to re-use all or part of this article, contact the AACR Publications

on June 3, 2015. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from