J Neurosurg 76:660-669, 1992

Human antiglioma monoclonal antibodies from patients with astrocytic tumors MICHAEL D. DAN, M.D., PH.D., CHRISTOPHER M. SCHLACHTA,M.D., JOHN GUY, M.D., RICHARD G. MCKENZIE, DELBERT R. DORSCHEID, B.S., ViCtOR A. SANDOR, M.D., JEAN-GUY VILLEMUREsM.D., AND GERALD B. PRICE, PH.D.

Department of Neurology and Neurosurgery, Montreal Neurological Institute, and Department of Medicine. McGill Cancer Centre, McGill University, Montreal, Quebec. Canada ~," The current management of malignant gliomas is unsatisfactory compared to that of other solid tumors; the expected median survival period is less than 1 year with the patient undergoing conventional surgery, radiotherapy, and chemotherapy treatment. Immunological reagents could be a useful adjunct. Human monoclonal antibodies derived from patients with astrocytic tumors might recognize subtle antigenic specificities that would differ from those recognized by xenogeneic (murine) systems. Five hybridomas, designated as BT27/1A2, BT27/2A3, BT32/A6, BT34/A5, and BT54/BS, were produced from the fusion of peripheral blood lymphocytes of four patients with astrocytic tumors to the human myeloma-like cell line TM-H2-SP2. This cell line has a 46, XX karotype and is negative for hypoxanthine guanine phosphoribosyltransferase. All five human monoclonal antibodies produced 2.4 to 44 #g/ml of immunaglobulin M, had a similar but not identical pattern of reactivity against a panel of human tumor cell lines, and failed to reacl with normal human astrocytes. Labeling of four neuroectodermal tumor explant cultures by BT27/2A3 was demonstrated by flow cytometry. Karyotyping of three of the five hybridomas demonstrated that two were pseudodiploid (2-3n) and one hypodiploid (< 2n). The monoclonality of the hybridomas was evaluated by Southern blot analysis of JH gene rearrangements, revealing two types of rearrangements for each hybridoma, both consistent with monoclonality. Preliminary antigen characterization indicated that at least four of the five human monoclonal antibodies were directed to cell-surface glycolipids. KEY WORDS immtmotherapy

9

astroeytoma

9 antigen

ALIGNANT gliomas are the most common of all primary brain tumors. They are also among the most difficult to treat. The combination of surgery and radiotherapy is seldom curative; with the best of care, the median survival period is less than 1 year) ~ Despite major efforts to introduce new adjunctive therapies, the prognosis for patients with a malignant glioma has remained essentially unchanged for the past 20 years. Many adjunctive therapies for malignant gliomas, including most forms of immunotherapy, are either ineffective 21 or carry a high risk of serious morbidityJ However, treatment involving tumor immunodiagnosis and immunotherapy with monoclonal antibodies has emerged as a promising area of investigation. As xenogeneic reagents, however, murine monoclonal antibodies may not be ideally suited for clinical applications in humans because of poor biological compatibility and increased clearance by the host reticuloendothelial system. We postulated that human monoclonal antibodies TM might offer: 1) more subtle and specific recognition

M

660

9 human monoelonal antibody

9

of tumor-associated antigens; and 2) improved compatibility with host cellular effector mechanisms than their murine counterparts. In this report, we describe five stable human antiglioma monoclonal antibodies of immunoglobulin (Ig)M isotype, derived from four patients with astrocytic tumors. Materials and Methods

Hybridoma Derivation Hybridomas were derived from peripheral blood lymphocytes (PBL's) of four patients with astrocytic tumors. BT-27 fusion was derived from lymphocytes of a 23-year-old man who presented with an 18-month history of seizures and was shown to have a frontal astrocytoma grade III. BT-32 fusion lymphocytes were obtained from a 14-year-old boy who presented with intraventricular hemorrhage due to a fibrillary astrocytoma located in the area of the right lateral ventricle. BT-34 fusion lymphocytes were derived from an l lyear-old girl who presented with an 8-year history of J. Neurosurg. / Volume 76/April, 1992

Human antiglioma monoclonal antibodies from astrocytic tumors headaches and was shown to have a fibrillary astrocytoma of the posterior fossa. BT-54 fusion lymphocytes were obtained from a 53-year-old man who presented after a single seizure, and was found to have a left frontal glioblastoma. Clinical Material Heparinized peripheral venous blood was obtained prior to the administration of any corticosteroids from patients with documented brain tumors. Lymphocytes were separated on a Ficoli-Hypaque density gradient? washed with sterile phosphate-buffered saline (PBS), and resuspended in 10 ml alpha-minimum essential medium (aMEM) supplemented with 10% fetal calf serum (FCS),* L-glutamine (292 mg/liter), L-asparagine (44 mg/liter), penicillin (100 vU/ml), and streptomycin (100 vg/ml). The lymphocytes were counted in a hemacytometer and either used immediately for a fusion procedure or stored at -70*C in 1 ml aMEM supplemented with 10% FCS and 10% dimethyl sulfoxide. Autologous tumor specimens (1 to 5 gm) were removed at operation, finely minced with sterile scissors, and agitated overnight at 4"C with 3 M KC1 (1 mi/gm wet tissue). The crude cell lysate was diluted 1:10 with distilled HzO and clarified by high-speed centrifugation at 12,000 G for 20 minutes at 4"C. The extract was then dialyzed and equilibrated using PBS and sterilized using a 0.22-~m millipore filter. The protein concentration was determined, J9 and dilutions with PBS were made to bring the final concentration to 100 ug/ml. Cell Lines The TM-H2-SP2 cell line is the immunoglobulin nonsecreting subline of the IgG (~) parental line TMH2, 3~ a hypoxanthine guanine phosphoribosyltransferase (EC 2.4.2.8)-deficient derivative of an unknown human myeloma-like line selected in 0.8% methylcellulose for its resistance to 6-thioguanine (6 #g/mi) and failure to grow in hypoxanthine-aminopterin-thymidine (HAT) medium. I7 The TM-H2-SP2 cell line has a 46, XX karyotype and is negative for Epstein-Barr nuclear antigen (EBNA). It was continuously maintained in ~MEM supplemented with 10% FCS, 6thioguanine (6 ~g/ml), penicillin (100 #U/ml), and streptomycin (100 gg/ml) at 37"C in a humidified chamber with 5% COz and 95% air. The human tumor cell lines used in our tests of human monoclonal antibody reactivity included: seven glioma cell lines (SK-MG-I, SKI-I, LN-215, LN-340, SK-MG-13, U-178, and U-373);t four colonic adenocarcinoma cell lines (HT-29, SW948, SW1166, and * Fetal calf serum supplied by Flow Laboratories, Rockville, Maryland. t Human glioma cell lines SK-MG-I and SKI-1 were supplied by L. Panasei, the Lady Davis Research Institute, Montreal, Quebec, Canada; glioma lines LN-215 and LN-340 by N. de Tribolet, Centre HospitalierUniversitaireVaudois, Lausanne, Switzerland; and glioma lines SK-MG-13, U-178, and U-373 by J. G. Caimcrass, London Regional Cancer Centre, London, Ontario, Canada. J. Neurosurg. / Volume 76/April, 1992

SWI417); three melanoma cell lines (M4, IGR-37, and IGR-39); three cervical carcinoma cell lines (HeLa, ME180, and C-33A); three hemopoietic cell lines (CEM-T (T-cell leukemia), K562 (chronic myelocytic leukemia), and HL-60 (acute promyelocytic leukemia)); two lymphoma cell lines (U937 (histiocytic lymphoma) and Raji (B-cell lymphoma)); two neuroblastoma cell lines (IMR-32 and SK-N-MC); and one retinoblastoma cell line (Y-76). All tumor cell lines were grown in aMEM supplemented with 10% FCS, L-glutamine (292 mg/ liter), anhydrous L-asparaglne (44 mg/liter), penicillin (i00 vU/ml), and streptomycin (100 ug/ml) at 37"C in a humidified chamber with 5% CO2 and 95% air. Fusion Procedure Twenty-four hours prior to fusion, the TM-H2-SP2 cell line was diluted 1:1 with fresh aMEM supplemented with 10% FCS and 6-thioguanine (6 ug/ml). The cells were counted by the dye-exclusion method, ~~ and directly combined with PBL's in a lymphocyte:myeloma cell ratio of 4:1. After the mixture was pelleted at 500 G for 5 minutes, the cells were fused under serum-free conditions by the gradual addition of 1 ml of 50% polyethylene glycol (MW 1450)~t diluted 1:1 in serum-free aMEM over a 1-minute period.25"27 This was followed by gradual dilution over a period of 4 to 5 minutes with 10 ml of serum-free aMEM. The fusion mixture was then repelleted as before and washed twice with 10 ml HAT medium supplemented by 10% FCS, 10-4 M hypoxanthine, 4 x 10 -7 M aminopterin, 1.6 x 10-s M thymidine, L-glutamine (292 rag/liter), L-asparagine (44 mg/liter), penicillin (100 uU/ml), and streptomycin (100 vg/ml). The cells were diluted in HAT medium to a concentration of 1 x 105 or 2.5 x 105 viable myeloma cells/ml, dispensed in 200ul aliquots into 96-microwell traysw (0.28 sq era/well), and incubated at 37"C in a humidified chamber with 5% CO2 and 95% air. Culture Conditions The hybridomas were fed with HAT medium once on Day 5 and with hypoxanthine-thymidine medium (HAT medium minus aminopterin) on Day 7. Further exchanges of hypoxanthine-thymidine medium were carded out every 5 days for a period of 4 to 6 weeks, at which time macroscopicaUy visible colonies appeared. Selected hybridomas were continuously propagated in standard culture flasks in hypoxanthine-thymidine medium. Tumor-Extract Assay The method used for the enzyme-linked immunosorbent assay (ELISA) of the tumor specimen was a modification of that described by Douillard and Hoffman? Briefly, 96-well polyvinyl chloride microtiter :~Polyethyleneglycol supplied by Sigma Chemical Co., St. Louis, Missouri. wMicrowell trays manufactured by Linbro Laboratories, Inc., McLean, Virginia. 661

M. D. Dan, et al. ELISA plates** were incubated for 12 to 18 hours at 4"C with 200 ul/well of 3 M KCI autologous tumor extract at a concentration of 5 ug protein/ml ( 1 ~g/well) in carbonate-bicarbonate buffer (pH 9.6; 0.8 gm Na~CO~, 1.47 gm NaHCO3, and 0.1 gm NaN~ combined with distilled H20 to a total volume of 500 ml). After three washings with PBS containing 0.05% Tween-20 (PBSTween),* 50 ul of the hybridoma supernatant was added in 50 ul PBS/well. Culture supernatant from the parental myeioma line (TM-H2) was used as a source of nonspecific control IgG, and supernatant from one of the hybridomas (BT27/2D2) was used as a source of nonspecific control IgM. After 2 hours of incubation at 20"C, the plates were washed three times with PBS-Tween. Alkaline phosphatase-conjugated goat anti-human K plus goat antihuman ~, Ig light chain, diluted 1:2000 with 1% bovine serum albumin (BSA), Fraction V, in PBS, was then added to each well at 200 ul/well.t After another 2 hours of incubation at room temperature, the plates were again washed three times with PBS-Tween, and developed with freshly prepared phosphatase substrate (p-nitrophenyl phosphate disodium),:~ 1 mg/ml in diethanolamine buffer (pH 9.8; 97 ml diethanolamine, 100 mg MgCI2.6H20, 0.2 gm NAN3, and distilled H20 to a total volume of 1 liter). Phosphatase activity was determined as absorbance at 4.5 nm using a microELISA autoreader.w A hybridoma supernatant was considered to be positive if phosphatase activity exceeded the mean background level of wells with control culture supernatants by more than two standard deviations.

Fixed Cells Assay Human tumor cell lines were seeded into flexible polyvinyl chloride U-shaped microwells precoated with poly-L-lysine, fixed with 0.1% glutaraldehyde in PBS for 3 minutes, and exposed to hybridoma supernatants in an ELISA reaction according to the method of Surer, et al) 4 Alkaline phosphatase-conjugated goat anti-human K plus goat anti-human ~ Ig light chain, each diluted 1:1000 with PBS containing 0.05 % Tween-20 and 0.1% gelatin 275 Bloom, II were added to each microwelI, and phosphatase activity was determined as in the tumor-extract ELISA. Flow Cytometry Confluent phase cultures of tumor cell lines were **Microtiter ELISA plates manufactured by Dynatech, Alexandria, Virginia. * Tween-20 manufactured by Sigma Chemical Co., St. Louis, Missouri. tGoat anti-human Ig supplied by TAGO, Budingame, California; bovine serum albumin supplied by Sigma Chemical Co., St. Louis, Missouri. ~:Sigma 104 phosphatase substrate manufactured by Sigma Chemical Co., St. Louis, Missouri. wMicroELISA Autoreader manufactured by Dynatech, Alexandria, Virginia. IIGelatin 275 Bloom manufactured by Fisher Scientific, Montreal, Quebec, Canada

662

scraped into ice-cold PBS containing 1% FCS, aliquoted into 10-ml test tubes, and pelleted by centrifugation for 5 minutes at 500 G. After 300 ~1 of hybridoma supernatant was added to each sample, the cells were resuspended and placed on ice for 60 minutes. Following three washes with 5 ml cold PBS supplemented with 1% FCS, 100 #1 of fluorescein isothiocyanate-conjugated goat anti-human IgM (IgG fraction, u chainspecific),* diluted 1:10 in sterile PBS, was added to each sample. The cells were then placed on ice for 60 minutes, and washed three times with PBS supplemented with 1% FCS. After the final wash, the cell pellet was resuspended in 500 ~1 of ice-cold PBS containing 1% FCS. Flow cytometry was performed using a fluorescence-activated cell sortert as previously described. 32

Hybridoma Kao,otyping Metaphase chromosomes were obtained by overnight culture in hypoxanthine-thymidine medium supplemented with 10% FCS and coleemid (0.06 ug/ml). After a wash in PBS, the cells were resuspended in 3 ml 0.07 M KC1 for 20 minutes at 20"C. Following three washes with a methanol:acetic acid solution (3:1 concentration), the cells were resuspended in 1 ml of the solution and dropped from a height of approximately 2 mm onto prewarmed glass slides. The cells were fixed in Giemsa stain for 3 minutes and 15 to 20 metaphase spreads were counted. Southern Blot Analysis Genomic deoxyribonucleic acid (DNA) was isolated from the TM-H2-SP2 cell line and the BT27/1A2, BT27/2A3, BT32/A6.5 (a subclone of BT32/A6), and BT27/2D2 hybridoma cells according to a method of Maniatis, et al.22 Ten ug of DNA was digested overnight at 37"C with two restriction enzymes: BamHl and HindllI) (_> 3 U/ug DNA).~ After digestion, the DNA was subjected to electrophoresis on a 0.8 % (w/v) agarose gel 22 and transferred to a Genescreen Plus membrane.w Following prehybridization with herring sperm DNA, the Southern blot was hybridized ~ overnight at 65"C with a ~2p-labeled JH probe (specific activity 5 to 6 x 10~ cpm/ug). II The probe includes the entire human Ja region, a total of 5.6 kilobase pairs? 8 The blot was exposed to x-ray film with intensifying screens at -70"C after washing it according to the manufacturer's specifications. * Goat anti-human IgM supplied by Cappel Laboratories, Cochranville, Pennsylvania. t Fluorescence-activated cell sorter, Model III, manufactured by Becton-Diekinson, Mountain View, California. BamHl and HindllI restriction enzymes supplied by Boehringer-Mannheim, Germany. wGenescreen Plus membrane supplied by New England Nuclear, Boston, Massachusetts. II Herring sperm DNA supplied by Boehringer-Mannheim, Germany; JH probe supplied by Oncor, Inc., Gaithersburg, Maryland.

J. Neurosurg. / Volume 76/April, 1992

Human antiglioma monoclonal antibodies from astrocytic tumors Determination of Immum,glol,ulin lsot.lTJe and Concentration The method used to determine the lg isotype and concentration was a modification of the tumor extract ELISA described above. Briefly, ELISA microtiter plates were coated with goat anti-human Ig (IgA + IgG + IgM),* diluted 1:4500 with carbonate-bicarbonate buffer. The plates were incubated with hybridoma supernatant in parallel with spent culture media from the parental line, TM-H2, as a source of Kand 3' chains and supernatant from the human B-lymphocyte line (RPMI 1788) ~2as a source for ~, and u chains. Individual plates were then labeled with alkaline phosphatase-conjugated goat anti-human x, ~, 3', or u chains at a dilution of 1:2000 in PBS supplemented with 1% BSA, and phosphatase activity was developed in the usual manner. Measurement of human IgM concentrations was performed by quantitative ELISA using affinity-purified human IgM'f in dilutions of 50, 5, 0.5, and 0.05 ug/ml in PBS.

seconds, the plates were blocked with 5% BSA in Trisbuffered saline (20 mM Tris-HC1, 0.5 M NaCI; pH 7.5) for 2 hours at 4'C. The chromatograms were exposed to hybridoma supernatants for an additional 2 hours at 4~ then washed six times with Tris-buffered saline. Peroxidase-conjugated goat anti-human IgG (heavy and light chain-specific) or anti-human IgM (Fc-specific),ll diluted 1:2000 with Tris-buffered saline containing 1% BSA, was added for 2 hours at 4"C. After further washes, the plates were developed with 1 mg/ml 3,3'-diaminobenzidine tetrahydrochloride dissolved in a solution of 0.05 M Tris-HC1 (pH 7.5) and 0.05% H20 and rinsed with distilled H,O.

Labeling of Normal Human Astrocytes Cultured normal human astrocytes* grown on glass coverslips were labeled according to the method used for flow cytometry described above with the exception that all washes were carried out using PBS supplemented with 1% FCS and 0.02% NAN,. Results

Tumor Explant Cultures Sterile surgical biopsy specimens were collected in ice-cold aMEM containing 20% FCS. Approximately 0.5 to 1 gm of specimen was finely minced with sterile scissors, and five to 10 fragments measuring approximately 1 cu mm were placed in T25 flasks containing 10 ml of culture medium. After 7 days of incubation under standard conditions, the medium was replenished and the cultures were returned to the incubator. After 2 to 3 weeks, the explants formed confluent monolayers that could be passaged and analyzed by flow cytometry, as outlined above.

lmmunofluorescence Microscopy Cytocentrifuge preparations of human cell lines that had been labeled for flow cytometry were mounted in PBS supplemented with 10% glycerol and examined with an epi-fluorescence microscope apparatus.

lmmunochromatography Total lipid extracts were prepared according to Svennerholm and Fredman. 35 Four ul of lipid extract was spotted onto high-performance thin-layer chromatography aluminum-backed silica gel 60 plates~ and dried under vacuum for at least 4 hours. The plates were then chromatographed in a chloroform:methanol:water (60:35:8) solution and dried under vacuum overnight for at least 16 hours. After coating with 0.05% polyisobutyl-methyl methacrylatew in hexane for 30 to 60 * Goat anti-human Ig supplied by Cappel Laboratories, Cechranville, Pennsylvania. t Affinity-purifiedhuman IgM supplied by Cappel Laboratories, Cochranville, Pennsylvania. Plates manufactured by Applied Analytical Industries, Wilmington, North Carolina. wPolyisobutyl-methyl methacrylate supplied by Polyscience, Inc., Warrington, Pennsylvania.

J. Neurosurg. / Volume 76/April. 1992

Fusion Procedures Four fusions from the four patients with astrocytic tumors described above yielded stable hybridomas. The number of PBL's available for fusion varied from 1.2 x 107 to 2.0 x 107. Typically, hybridoma colonies appeared between 4 and 6 weeks after fusion. Not all colonies could be succ~sfully propagated; however, there were no instances of hybridoma failure occurring later than 3 months postfusion.

Primary Screening by ELISA Hybridomas were initially screened against either extracts of autologous tumor or giutaraldehyde-fixed human glioma cell lines in an ELISA assay. Screening was carried out approximately 4 weeks after the fusion procedure, following at least four exchanges of culture medium to ensure that an Ig secreted by unfused lymphocytes in the original fusion mixture would be diluted at least 16-fold. Table 1 lists the ELISA data for the four fusions and indicates the number that exceeded the mean background binding by two to four standard deviations.

Secondary Screening by Flow Cytometry Nine of 61 potential hybridomas from the four fusions demonstrated robust growth in vitro and produced high concentrations of Ig. These were screened for reactivity with the human glioma line SK-MG-1 by flow cytometry. Five supernatants designated BT27/ 1A2, BT27/2A3, BT32/A6, BT34/A5, and BT54/BS, were found to label this particular glioma cell line. All five supernatants contained tumor-reactive IgM species. II Anti-human IgM supplied by BIO/CAN, Toronto, Ontario, Canada. * Normal human astrocytesprovided by J. Anlel, Montreal Neurological Institute, Montreal, Quebec, Canada. 663

M. D. Dan, et al. TABLE 1

Tl#nor-extract enzyme-linked immunosorbent assay Fusion

No. of Wells

BT-27

136

BT-32

82

BT-34 BT-54

82 43

Antigen* autnlogous SK-MG-1 LN-340 SK-MG-I LN-340 autologous BT-38

Level of Positivityw

Culture Supernatantt

Ig Chain~

2SD

3SD

>4SD

% TotallJ

TM-H2 TM-H2 TM-H2 TM-H2 TM-H2 TM-H2 BT27/2D2

~ + ~, ~+ h ~+ h ~+ h K+ h K+ h ~

4 2 1 6 10 17 1

1 I 0 4 3 9 0

0 2 3 12 8 4 3

3.7 3.7 2.9 26.8 25.6 36.6 9.3

* Antigens used in screening: autologous = 3 M KCI autologous tumor extract; LN-340; and SK-MG-1 -0.1% glutaraldehyde-fixedcells; BT38 = 3 M KCI allogeneic tumor extract. 'I"TM-H2 = parental myr line containing Kimmunoglobulin (Ig)G and used to establish a background level of nonspecific binding of Ig; BT27/2D2 = hybridoma used as a source of nonspecifie IgM (~t chains). :~A mixture of alkaline phosphatase-conjugated goat anti-human plus Ig light chain was used as an indicator of antigen-bound Ig; in some experiments, alkaline phosphatase-conjugated goat anti-human IgM (~t chain-sl~zifie) was used as second antibody. wNumber of microwells with a given standard deviation (SD) above mean background. II Proportion of mierowells reacting > 2 SD above mean background level.

TABLE 2

Hybridoma Karyotypes

Karyotypes of three human hybridomas and the parental cell line Hybridoma

No. of Metaphases

No. of Chromosomes Mean* Mode Range

BT27/1A2 20 61 ___15 66 BT27/2A3 17 41 ___2 40 BT32/A6 20 53 __.8 56 TM-H2-SP2? 19 46 _+2 46 * Values are expressed as mean + standard deviation. ~"Parental cell line.

42-88 37~16 43-69 42--49

C h r o m o s o m a l content for three o f the five hybridomas is summarized in Table 2. The parental cell line T M - H 2 - S P 2 was also studied and found to have a m e a n and modal n u m b e r of 46 chromosomes. All three hybridomas were pseudodiploid, with the BT27/1A2 and BT32/A6 antibodies having 2-3n c h r o m o s o m e s and the BT27/2A3 antibody slightly less than 2n.

Determination of Monoclonality Southern blot analysis (Fig. 1) revealed that the BT27/1A2, BT27/2A3, and BT32/A6.5 hybridomas each possess two rearranged bands bearing homology to the J . gene region. The B T 2 7 / 2 D 2 hybridoma appears to possess three such bands. The B cell fusion partner used in these experiments, TM-H2-SP2, has only one band and an apparent deletion. Furthermore, there is no evidence o f a T M - H 2 - S P 2 type of rearrangem e n t of the Jn region in any o f the hybridomas. As a control, normal PBL D N A , which is composed o f 60% to 70% T cell-derived genetic material, was found to yield a blot profile identical to that of placental (germline) DNA.

Immunoglobulin Isotypes and Production Levels

FIG. 1. Southern blot analysis of hybridoma deoxyribonucleic acid (DNA). Genomic DNA was isolated from hybridoma cells and digested with BamHl and HindllI restriction enzymes, then electrophoresed on a 0.8% agarose gel, blotted, and labeled with a Jn region probe. Lane a = placental DNA; Lane b = DNA from normal peripheral blood lymphocytes; Lane c = TM-H2 (the parental line); Lane d = BT32/A6; Lane e = BT27/1A2, Lane f = BT27/2A3, and Lane g = nonspecific control immunoglobulin (Ig)M hybridoma, BT27/ 2D2. These results indicate the presence of two J . rearrangements in each of the three hybridomas, consistent with monoclonality, kbp = kilobase pair. 664

All five hybridomas supernatants were found to contain u heavy chains. Only the BT34/A5 antibody contained X light chains; the other four m o n o c l o n a l antibodies contained ~ light chains. No 7 chains were detected in any of the hybridoma supernatants. After approximately 6 m o n t h s of maintenance in culture, the estimated IgM concentration was 5.0 ug/ml for the BT27/1A2 antibody, 44 ug/ml for the BT27/2A3 antibody, 3.5 ug/ml for the BT32/A6 antibody, 2.4 #g/ml for the BT34/A5 antibody, and 22.4 ug/ml for the BT54/B8 antibody. After an additional 6 months in continuous culture, IgM production levels were found to be comparable. J. Neurosurg. / Volume 76/April, 1992

Human antiglioma monoclonal antibodies from astrocytic tumors

FIG. 2. Two-dimensioaaJrepresentation of SK-MG-1 cells analyzed by flow cytometry. For purposes of comparison, a box of cross-hatched bars has been placed around the "small" ceils and a box of dot-dash bars has been placed around the "large" cells. A: Control (BT27/2D2) labeled cells. B: Cells labeled with the BT27/2A3 monoclonal antibody. There is a significant shift in fluorescenceaffectingonly the "small" cell subpopulation.

Flow Cytometric Analysis of Human Tumor Celt Lines Flow cytometric analysis of cultured human cell lines and strains was performed in parallel with established positive cell lines on two or more separate occasions. Several different classes of neuroectodermal and nonneuroectodermal tumors and tissues were tested. All five human monoclonal antibodies reacted with the SK-MG-1 and U-373 glioma cell lines. The BT27/1A2, BT27/2A3, and BT32/A6 antibodies were also found to react with the SK-MG- 13, SKI-l, and U- 178 glioma cell lines but not with the LN-215 or LN-340 glioma cell lines, the two neuroblastoma lines, or the retinoblastoma line. In addition, BT27/1A2, BT27/2A3, and BT32/A6 failed to react with an epithelial cervical carcinoma cell line (C-33A), a transitional bladder carcinoma cell line (J82), three melanoma cell lines, and four colonic adenocarcinoma cell lines. None of the five monoclonal antibodies reacted with the normal embryonic tung tibmblast ceil line (WI38~ or three hematological cell lines (T cell and myeloid); the BT34/ A5 and BT54/B8 antibodies also failed to react with two other hematological lymphoma cell lines. All five human monoclonal antibodies demonstrated a similar pattern of reactivity with the 30 cell lines that were studied, with a few notable exceptions. For example, unlike the other three, the BT34/A5 and BT54/ B8 antibodies both failed to react with the SKI- 1 glioma cell line, but labeled the M-4 melanoma cell line. Only the BT27/1A2 and BT27/2A3 antibodies reacted with the ME-180 epithelial cervix carcinoma cell line. In summary, l) the BT27/IA2 and BT27/2A3 antibodies exhibited one pattern of reactivity, reacting with only J. Neurosurg. / Volume 76/'April, 1992

glioma cell lines, and the HeLa and ME-180 epithelial cervical carcinoma ceil lines; 2) the BT32/A6 antibody had a singular pattern of reactivity and only differed from the BT27/1A2 and BT27/2A3 antibodies by not reacting with the ME-180 cell line; and 3) the BT34/ A5 and BT54/B8 antibodies had a slightly differenl pattern of reactivity, reacting with two of the three gliomas as did the other monoclonal antibodies and with the M-4 melanoma cell line, and not with any of the three epithelial cervical carcinoma lines. In many instances, two-dimensional flow cytometry revealed the presence of two discrete subpopulations of cells (Fig. 2). After the SK-MG-I cell line was exposed to any of the f~vehuman monoclonal antibodies, there appeared to be selective labeling of only the smaller of the two subpopulations. This was consistently observed over many experiments and in many other tumor cell lines that also contained two discrete subpopulations (for example, the HeLa cell line).

Flow Cytometric Analysis of Primary Tumor Explant Cultures Four tumor explant cultures were labeled with the BT27/2A3 antibody after brief exposure in vitro and studied by flow cytometry. The results are shown in Fig. 3 as a composite illustration. The greatest apparent proportion of labeling (46%) occurred with a recurrent meningioma. lmmunofluorescence Figure 4 is a composite picture of cytocentrifuge preparations of fluorescently labeled SK-MG-I cells. 665

M. D. Dan, et al.

FIG. 3. Histograms showing BT27/2A3 antibody labeling of four tumor explain cultures (broken line). Solid line = background labeling with the BT27/2D2 control hybridoma. A: Gfioblastoma multiforme. B: Recurrent meningioma. C: Medulloblastoma. D: Astrocytomagrade lI.

The BT27/IA2 and BT27/2A3 antibodies were found to label the cell surface in a contiguous fashion, whereas the BT32/A6 monoclonal antibody appears in a punctate manner on the cell surface. Linear areas of increased labeling may be observed for the BT27/1A2 and BT27/2A3 antibodies; these represent folds in the cell membrane and are likely a result of the method of cell preparation. Preliminary Antigen Characterization Numerous attempts at demonstration of hybridoma reactivity with Western blots of cell membrane vesicles from cultured neuroectodermal tumor cell lines were unsuccessful (results not shown). Figure 5 illustrates immunochromatography of a total lipid extract of the SK-MG-1 cell line developed with the BT34/A5 antibody. There is a single specific band (ratio of movement of the band to the front of the solvent (Rf) approximately 0.60) and a nonspecific band (Rf approximately 0.80) evident in both the BT34/A5 antibody and the control lanes. In separate experiments, similar results were obtained for the BT27/1A2 and BT27/2A3 antibodies. Immunochromatography with the BT54/B8 monoclonal antibody failed to reveal the presence of any specific banding pattern. These results suggest that human monoclonal antibodies recognize a determinant of a glycolipid or ganglioside that was not detected in association with glycoproteins. 666

FIG. 4. Fluorescence labeling of SK-MG-1 cells. All exposure times for fluorescencephotography are the same. A: Labeling with the BT27/1A2 human monoclonal antibody. B: Labeling with the BT27/2A3 antibody. C: Labeling wilh the BT32/A6 antibody. D: Labeling with the control immunoglobulin M antibody, BT27/2D2.

Reactivity with Normal Human Astrocytes None of three human monoclonal antibodies (BT27/ I A2, BT27/2A3, and BT32/A6) appeared to label cultured astrocytes from two individuals. The BT34/A5 and BT54/B8 antibodies also failed to label normal human astrocytes on at least one occasion. All reagents were checked for immune reactivity against an established positive human glioma cell line (U-373). Discussion In this report, we describe five human monoclonal antibodies from four patients with neurological tumors. All five were found to produce human IgM in concentrations of 2.4 to 44 ~g/ml and were shown to be monoclonal using Southern blot analysis of JH gene rearrangements. Karyotyping of three hybridomas demonstrated that two were pseudodiploid (2-3n) and one was hypodiploid (< 2n). When tested by quantitative flow cytometry against a panel of human tumor cell lines, all five monoclonal antibodies exhibited a similar but not identical pattern of reactivity. No labeling of cultured normal human astrocytes was demonstrated. Tumor explant cultures of four different primary neurological tumors reacted positively with the BT27/2A3 antibody. Preliminary antigen characterization suggested that at least four of the five human monoclonal antibodies recognize cell-surface glycolipids. J. Neurosurg. / Volume 76/April 1992

Human antiglioma monoclonal antibodies from astrocytic tumors

FIG. 5. Immunochromatogram of a lipid extract of SKMG-1 cells. O = origin of chromatogram. The plates were exposed to either the BT34/A5 antibody (a) or the TM-H2 cell line (b) supernatant. Following detection using an immunoperoxidasereaction, both specificand nonspecificlabeling was observed (arrows). There is a specific band ~th a ratio of movement of the band to the front of the solvent (Rf) of 0.61 in the BT34/A5 lane, representing the glycolipid identified by this antibody.

This is consistent with a monoclonal origin for each of the three hybridomas. Using a JH probe similar to the one used in these experiments, Ford, el al., 9 found two rearrangement bands among each of four monoclonal B-lineage cell lines: RPMI 1788, Bristol 7, REH, and NALM 6. Korsmeyer, et al.,~6 also found examples of heavy-chain double rearrangements among non-T cell and non-B cell leukemias, indicating that these cells are actually precursor cells already committed to B cell differentiation at the Ig gene level. Rudders, el a1.,29 also analyzed 52 tumors from patients with B cell-derived lymphomas for potential oligoclonality, using Bglll, BamHI, and EcoRI DNA digests probed with the JH gene. In 72% of all cases, rearrangements of both JH alleles could be detected following digestion with multiple restriction enzymes, suggesting that the process of heavy-chain rearrangement in human B-cell tumors is relatively inefficient, often requiring two attempts before rearrangement is productive. Failure to find a TM-H2-SP2 cell line heavy-chain rearrangement in any of the three hybridomas suggests either that none is carrying chromosome 14 from the TM-H2-SP2 cell line with the heavy-chain rearrangement or that a deletion on the chromosome has taken place involving the heavy-chain alleles.

Monoclonality of Hybridomas In the present study, proof of monoclonality was approached at the genotypic level. The Ig heavy-chain gene is located on chromosome 14 in humans and is composed of four distinct elements known as VH (variable), D . (diversity), J , (joining), and Cn (constant) regions. In germ-line DNA, these regions are separated by intervening sequences that are spliced out once the cell becomes committed to B cell differentiation.23 The heavy-chain probe used in these experiments spans the entire human germ-line JH region. In situations where there has been rearrangement of germ-line DNA, digestion with the BamHI and HindlII restriction enzymes will produce DNA fragments that differ in length from the 5.6-kilobase pair Ja probe. According to the principle of allelic exclusion, it was originally postulated that only one chromosome would undergo rearrangement while the other remained in the germ-line configuration. In the mouse heavy-chain gene, however, the second chromosome is essentially always rearranged?4 It was therefore proposed that for a given cell, there was a low probability of a functional rearrangement. S One would thus encounter the situation of two nonfunctional or one functional and one nonfunctional, but very rarely two functional heavychain rearrangements per cell. These observations are valid for the mouse, and also seem to be valid for higher mammalian species. 16 The finding of two such fragments in each hybridoma lane indicates that two heavychain rearrangements are detected for each hybridoma.

Other Human Monoclonal Antibodies to Astrocytomas Sikora, el aL, ~' were the first to report the production of human monoclonal antibodies reactive to 0.25% glutaraldehyde-fixed glioma cells. Intratumoral lymphocytes from 12 patients undergoing craniotomy for malignant glioma were fused with the LICR-LONHMy2 human lymphoblastoid cell line (positive for EBNA and IgG(h) and resistant to 8-azaguanine). A total of 71 hybridomas were obtained from five patients. No attempt was made to prove the monoclonality of the hybridomas; however, in a subsequent report,26 the same group identified a single heavy- and light-chain in the supernatant of one of their antiglioma human monoclonal antibodies, LGLI-IDr, using sodium dodecyl sulfate-polyacrylamide gel electrophoresis of 13q_ labeled Ig. Cross-reactivity of antiglioma human monoclonal antibodies with other tumor cell lines was reported by Sikora, et al? 1 Of nine human monoclonal antibodies studied with a panel of tumor cell lines, four reacted with the HT-29 colon carcinoma cell line, three with the MOR lung carcinoma cell line, and one with the CALU- 1 lung carcinoma cell line. This should be compared with only two of 19 nongliomatous tumor cell lines exhibiting cross-reactivity with the BT27/1A2 and BT27/2A3 antibodies, only one of 19 tumor cell lines cross-reacting with the BT32/A6 antibody, and only one of 12 tumor cell lines cross-reacting with the BT34/ A5 and BT54/B8 antibodies.

J. Neurosurg. / Volume 76/April, 1992

667

M. D. Dan, et aL Glioma-Associated Antigens Other human monoclonal antibodies cross-reactive with glioma-associated antigens have been produced and characterized. ~'t~-" Perhaps the most extensively studied is the IgM (K) molecule produced by the human B-lymphoblastoid cell line L72.13 This cell line was derived from a patient with a malignant melanoma due to Epstein-Barr transformation of PBL. It secretes IgM at a level of 9 ug/ml in the spent medium of 1 • l& cells within 4 to 5 days of subculture. The L72 antibody binds weakly in an immune adherence assay to four cultured human glioma lines (AS, LA-N-I, A172 MG, and SK-MG-3) but not to the SKMG-1 cell line? 4 The antigen recognized by the L72 antibody was determined to be the ganglioside GD2.' Given the lack of L72 cross-reactivity with the SK-MG1 cell line and the presence of cross-reactivity of all five human monoclonal antibodies discussed in the present study to the SK-MG-1 cell line, it seems unlikely that any of the antibodies recognize the GD2 ganglioside. Characteristics and Utility of Five Human Monoctonal Antibodies The five human monoclonal antibodies described in this report were selected from 61 potential hybridomas (based on ELISA screening data) for robustness of growth, high Ig production levels, and cell-surface reactivity. All five antibodies appear to react with cellsurface glycolipids, as determined by immunochromatography experiments. The fact that all five were of the IgM isotype may be significant. Previous work with the TM-H2 cell line has shown that this particular myeloma-like cell line fuses equally well with B lymphocytes of all isotypes. 33 At least three studies 32~ have shown that elevated preoperative levels of IgM are frequently detected in patients with glioblastomas, whereas overall Ig levels are normal in most other patients with intracranial tumors. Another explanation for finding only IgM antibodies may lie with glioma immunosuppressive mechanisms that might prevent the elaboration of a secondary (IgG) immune response in these patients. The fact that all five human monoclonal antibodies were of IgM isotype may simply be a reflection of the type of antigen that they recognize, namely a carbohydrate determinant. Complete characterization of the glycolipids identified by five human monoclonal antibodies may help to further elucidate the host antitumor immune response in patients with neurological tumors, as well as permit the generation of new monoclonal antibodies against the same or related substances. It may also be possible to immunize glioma patients with the gangliosides themselves, as others ~8 have done with melanoma patients using a mixture of the monosialoganglioside GM2 and Bacille bili6 de Calmette-Gurrin. Conclusions

We have described five human antiglioma monoclonal antibodies of IgM isotype that appear to label 668

cell-surface glycolipids of cultured human glioma cell lines and other neurological tumors. The monoclonal antibodies also label tumor explant cultures, but fail to react with normal human astrocytes. The hybridomas have been stable in culture for over 1 year, produce human IgM in concentrations of 2.4 to 44 #g/ml, and appear to be monoclonal by Southern blot analysis of JH gene rearrangements. These human monoclonal antibodies may be useful as reagents for tumor immunodiagnosis and possibly for immunotherapy. References

1. Barba D, Saris SC, Holder C, et al: Intratumoral LAK cell and intedeukin-2 therapy of human gliomas. J Neurosurg 70:175-182, 1989 2. Bhondeley MK, Mehra RD, Mehra NK, ct al: Imbalances in T cell subpopulations in human gliomas, d Neurosurg 68:589-593, 1988 3. Bryum A: Isolation of mononuclear cells and granulocyles from human blood. Isolation of monuclear cells by one centrifugation, and of granulocytes by combining centrifugation and sedimentation at 1 g. St'and J Clin Lab Invest 21 (Supp197):77-89, 1968 4. Cuban LD, Irie RF, Singh R, el al: Identification of a human neuroectodermal tumor antigen (OFA-I-2) as ganglioside GD2. Prac Natl Aead Sei USA 79:7629-7633. 1982 5. Coleclough C, Perry RP, Karjalainen K, el al: Aberrant rearrangements contribute significantly to the allelic exclusion of immunoglobulin gene expression. Nature 290: 372-378, 198i 6. Cote ILl, Morrissey DM, Houghton AN, el al: Specificity analysis of human monoclonal antibodies reactive with cell surface and intracellular antigens. Proe Natl Acad Sci USA 83:2959-2963, 1986 7. Croce CM, Linnenbach A, Hall W, et al: Production of human hybridomas secreting antibodies to measles virus. Nature 288:488-489, 1980 8. DouiUard JY, Hoffman T: Enzyme-linked immunosorbent assay for screening monoclonal antibody production using enzyme-labeled second antibody. Methods Enzymol 92:168-174, 1983 9. Ford AM, Molgaard HV, Greaves MF, et al: lmmunoglobulin gene organisation and expression in Memopoietic stem cell leukaemia. EMBO J 2:997-1001, 1983 10. Gorer PA, O'Gorman P: The cytotoxic activity of isoantibodies in mice. Transplant Bull 3:142-143, 1956 11. Houghton AN, Brooks H, Cote RJ, et al: Detection of cell surface and intracellular antigens by human monoclonal antibodies. Hybrid cell lines derived from lymphocytes of patients with malignant melanoma. J Exp Med 158:53-65, 1983 12. Huang CC, Moore GE: Chromosomes of 14 hemapoietic cell lines derived from peripheral blood of persons with and without chromosome anomalies. JNCI 43:

1119-1128, 1969 13. Irie RF, Sze LL, Saxton RE: Human antibody to OFA-I, a tumor antigen, produced in vitro by Epstein-Burr virustransformed human B-lymphoid cell lines. Proe Natl Acad Sci USA 79:5666-5670, 1982 14. Katano M, Sidell N, Irie RF: Human monoelonal antibody to a neuroectodermal tumor antigen (OFA-I-2). Ann NY Acad Sd 417:427-434, 1983 15. Kaufmann G, Zannis-Hadjopoulos M, Martin RG: Cloning of nascent monkey DNA synthesized early in the cell cycle. Mol Cell Biol 5:721-727, 1985 J. Neurosurg. / Volume 76/April 1992

Human antiglioma monoclonal antibodies from astrocytic tumors 16. Korsmeyer SJ, Arnold A. Bakhshi A, el al: Immunoglobulin gene rearrangement and cell surface antigen expression in acute lymphocytic leukemias of T cell and 13cell precursor origins. J Clin Invest 71:301-313, 1983 17. Littlefield JW: Selection of hybrids from matings of fibroblasts in vitro and their presumed recombinants. Science 145:709-710, 1964 18. Livingston PO, Natoli EJ, Calves MJ, et al: Vaccines containing purified GM2 ganglioside elicit GM2 antibodies in melanoma patients. Proc Natl Aead Sei USA 84: 2911-2915, 1987 19. Lowry OH, Rosebrough N J, Fair AL, et al: Protein measurement with the Folin phenol reagent, d Biol Chem 193: 265-275, 1951 20. Mahaley MS Jr, Brooks WH, Roszman TL, et al: Immunobiology of primary intracranial tumors. Part 1. Studies of the cellular and humora] general immune competence of brain-tumor patients. J Neurosurg 46: 467-476, 1977 21. Mahaley MS Jr, Steinbok P, Aronin P, el al: Immunobiology of primary intracranial tumors. Part 4. Levamisole as an immune stimulant in patients and in the ASV glioma model. J Neurosurg 54:220-227, 198 l 22. Maniatis T, Fritsch EF, Sambrook J: Molecular Cloning: A Laboratory Manual. New York: Cold Spring Harbor Laboratories, 1982 23. Max EE: Immunogiobulins: molecular genetics, in Paul WE (ed): Fundamental Immunology. New York: Raven Press, 1984, pp 167-204 24. Nottenburg C, Weissman IL: C# gene rearrangement of mouse immunoglobulin genes in normal B cells occurs on both the expressed and nonexpressed chromosomes. Proc Natl Acad Sei USA 78:484-488, 1981 25. Olsson L, Kaplan HS: Human-human hybridomas producing monoclonal antibodies of predefined antigenic specificity. Proc Natl Acad SCi USA 77:5429-5431, 1980 26. Phillips J, Alderson T, Sikora K, et at: Localisation of malignant glioma by a radiolabelled human monoclonal antibody. J Neural Neurosurg Psychiatry 46:388-392, 1983 27. Pontecorvo G: Production of mammalian somatic cell

J. Neurosurg. / Volume 76/April, 1992

28.

29.

30. 31. 32. 33.

34.

35. 36.

hybrids by means of polyethylene glycol treatment. Somat Cell Genet 1:397-400, 1975 Ravetch JV, Siebenlist U, Korsmeyer S, et at: Structure of the human immunogiobulin ~ locus: characlerization of embryonic and rearranged J and D genes. Cell 27: 583-591, 1981 Rudders RA, Dhillon S. Krontiris TG: Clonal analysis of untreated non-Hodgkin's lymphoma utilizing immunoglobulin gene rearrangement and immunophenotype. Cancer Res 48:6272-6277, 1988 Salcman M: Survival in glioblastoma: historical perspective. Neurosurgery 7:435-439, 1980 Sikora K, Alderson T, Phillips J, et al: Human hybridomas from malignant gliomas. Lancet l:l 1-14, 1982 Stewart SS, Price GB: Realtime acquisition, storage, and display of correlated three-parameter flow cytometdc data. Cytometry 7:82-88, 1986 Sullivan AK, Teskey LM, Campling B, el al: The potential of hybridomas to study autoantibodies in man, in Russ L, Carlton D (eds): Hybddoma Technology. Toronto: Ortho Pharmaceuticals Canada, 1982, pp 63-68 Suter L, B~ggen J, Sorg C: Use of an enzyme-linked immunosorbent assay (ELISA) for screening of hybridoma antibodies against cell surface antigens. J lmmunol Methods 39:407-411, 1980 Svennerholm L, Fredman P: A procedure for the quantitative isolation of brain gangliosides. Biochim Biophys Acta 617:97-109, 1980 Tokumaru T, Catalano LW Jr: Elevation of serum immunoglobulin M (IgM) level in patients with brain tumors. Surg Neurol 4:17-21, 1975

Manuscript received February 8, 1991. Accepted in final form July 29, 1991. This work was presented in part at the Annual Meeting of the Canadian Congress of Neurological Sciences, June 27-30, 1990, Banff, Alberta, Canada. Address reprint requests to."Gerald B. Price, Ph.D., McGill Cancer Centre, 3655 Drummond Street, Montreal, Quebec H3G IY6, Canada.

669