CA2035088A1 - Human monoclonal antibodies for brain tumor diagnosis and therapy - Google Patents

Abstract

ABSTRACT
The present invention relates to monoclonal antibodies directed against neuroectodermal tumors produced by human hybridoma cell lines. These hybridomas are produced by the fusion of peripheral blood lymphocytes, obtained from patients with different neuroectodermal tumors, and a myeloma-like cell line, TM-H2-SP2. Five hybridomas, designated as BT27/lA2, BT27/2A3, BT32/A6, BT34/A5, and BT54/B8, all produced monoclonal IgM in a range of 2.4 to 44 µg/ml, had a similar pattern of reactivity against a panel of human tumor cell lines, and did not react with normal human astrocytes. All five human monoclonal antibodies (HmAbs) recognized a subpopulation of tumor cells and preliminary antigen characterization indicated that these HmAbs are directed to cell surface glycolipids.
Cell sorting experiments suggest that the identified subpopulation may share certain properties with tumor stem cell. These HmAbs possess certain properties of reactivity that suggest potential roles for them in the future diagnosis and clinical management of human malignant gliomas.

Description

2 ~j C~

FIELD OF THE INVENTION
The present invention relates to human hybridoma cell lines and to the monoclonal antibodies produced by these cell lines.

BACKGROUN~ OF THE INVENTION
Malignant gliomas are the most common of all primary brain tumors. They are also among the most difficult to manage clinically~ Depending on their intracranial location, they can grow to a substantial size before causing any symptoms. Their biological behavior is notoriously aygressive, although they rarely metastasize outside of the CNS.
Surgery and radiotherapy alone are seldom curative; with the best of care, median survival is under one year. Moreover, desplte major efforts to introduce new adjunctive therapies, the prognosis for patients with malignant gliomas has remained essentially unchanged for the past thirty years.
Many ad;unctlve therapies, including immunotherapy, are either presently ineffective, or carry an unacceptable risk of serious morbidity. With the development of murine hybridomas by Kohler and Milstein (1975); however, tumor immunodiagnosis and immunotherapy with monoclonal antibodies (mAb) has emerged as a promising area of investigation.
Studies with murina mAbs have already contributed substantially to the understanding of gl~oma biology and heterogeneity. However, these reagents may not be ideally suited for clinical application in humans for a variety of reasons. In contrast, human mAbs offer the poss1bilities of greater specificity and biological compatibility than their murine counterparts.
Malignant gliomas are neuroepithelial tumors which commonly arise in the cerebral hemispheres. As a group, they include anaplastic astrocytoma, anaplastic oligodendroglioma, anaplastic ependymoma, and glioblastoma multiforme. These tumors all possess certain microscopic features in common such as dense cellularity, increased numbars of mitotic figures, nuclear pleomorphism with hyperchromasia, and cellular pleomorphism. In glioblastoma multiforme, the above features are present, as well as necrosis with pseudopalisades, capillary endothelial proliferation with pseudorosette formation, and mesenchymal proliferation. These tumors are heterogeneous and demonstrate both inter- and intra-tumor variability, which may well be the major determinant of therapeutic resistance. Since few markers exist which are selective for poorly differentiated glioma cells, i.e. stem cells, it would be highly beneficial to selctively target these cells within an otherwise heterogeneous population since failure to deplete a tumor of its stem cell compartment inevitably results in tumor recurrence.
It is well established that many patients with malignant gliomas are significantly immunosuppressed, especially with regards to cell-mediated functions, although few patients are cachectic. The finding of immunosuppression in the absence of significant systemic wasting would suggest that the former is due to tumor-intrinsic mechanisms, rather than a reflection of excessive tumor burden, as is usually the case. Humoral immune mechanlsms in patients with malignant gliomas also appear to be depressed, albeit they remain intact.
Malignant gliomas possess many antigens in common with the normal cells from which they were derived. As a result, it is virtually impossible to prepare conventional heteroantisera to glioma tissue without encountering some degree of cross-reactivity with normal adult brain. The shared antlgens may be either intracellular, membrane associated, or extracellular in nature.

The role of glycolipids and carbohydrate structures in cellular interaction, differentiation, and oncogenesis has received much attention lately. A
subclass of glycolipids known as gangliosides may play an important role in the regulation of cellular adhesion and proliferation. When a cell undergoes malignant transformation, there may be subtle qualitative and quantitative changes in the chemistry of its surface gangliosides.
Malignant gliomas are well vascularized tumors. However, there is brèakdown or total absence of the blood-brain-barrier that exists in a normal, healthy brain, in which the brain is normally 'privileged', or isolated from the effects of the immune system or other blood-borne agents. Because of this breakdown of the blood-brain-barrier in patients with these tumors, the immune system has access to the tumor cells.

BRIEF REFE~ENCE TO THE PRIOR A~T
Studies with murine mAbs raised against glioma cell lines were reported by Schnegg, et al. (1981b) and Bourdon, et al. (1983). Both groups immunized mice with cultured human glioma cell lines, and fused spleen cells with the mouse myeloma lines P3X63-Ag8 or P3X63-Ag8.653, respectively. The mAbs BF7, GE2, described by Schnegg et al (1981b) appear to be relatively glioma-restricted, whereas antibody CG12 (de Tribolet, et al. 1984) reacts wlth a spectrum of gliomas, melanomas, and neuroblastomas. Antibody 81C6 was produced and characterized by Bourdon, et al. (1983) and found to react with a glioma-mesenchymal e~tracellular matrix (GMEM) antigen expressed on gliomas, neuroblastomas, melanomas, sarcomas, and cultured fibroblasts. In addition to the above, the GMEM antigen was also found in normal liver sinusoids, spleen red pulp sinusoids, kidney medullary tubule interstitium, and glomerular mesangium. Whereas absorption of CG12 with normal adult and fetal brain abolished ~inding activity, neither BF7, GE2, nor 81C6 were affected by the same treatment.
Reports of human monoclonal antibodies (HmAbs) from patients with malignant gliomas have appeared in the literature. However, none of these antibodies have been well characterized.
Sikora, et al. (1982) reported the production of HmAbs reactive to 0.25% glutaraldehyde-fixed glioma cells. These investigators obtained intratumoral lymphocytes from 12 patients undergoing craniotomy for malignant glioma and fused them to the E8NA~ 8-azaguanine-resistant human lymphoblastoid line LICR-LON-HMy2, which is known to secrete gamma and lambda chains.
A total of 71 hybridomas were obtained from 5 patients.
No subcloning experiments were reported and no demonstratlon or proof of the monoclonality of the hybridomas is given. ~his study also reported cross-reactivity of the HmAbs to other tumor cell-lines. The disadvantages of the requirement for intratumoral cells from the human brain as the starting material are clear.
In a subsequent study by Sikora et al. (19~3), the authors concluded that, for the HmAb singled out for study, this HmAb was of a low affinity because of the high concentrations necessary to demonstrate reactivity with targets.
Other HmAbs cross-reactive with glioma-associated antigens have been produced and characterized.
It is an object of the present invention to provide human monoclonal antibodies to neurological tumors, useful in immunodiagnosis and with a po~ential for immunotherapy. It is a further object of the invention to provide a process for producing such monoclonal antibodies.

SUMMARY OF THE INVENTION
The present invention is based on the principle that human patients with neurological tumors should possess circulating B-lymphocytes with anti-glioma specificities. It is also based on the principle that tumor-reactive antibodies derived from glioma patients should recognize common, non-allelic determinants in human gliomas which might differ in specificities and distribution from those recognized by xenogeneic systems.
In the present invention, human-human hybridomas are used to prepare human monoclonal antibodies (HmAbs) to gliomas, the hybridomas being derlved from fusion of a human myeloma-like line and lymphocytes obtained from human patients suffering from gliomas. By appropriate culturing of the resultant hybridomas, effective HmAbs to human gliomas are obtained in useful quantities.
The HmAbs and the processes for obtaining them differ from the work reported by Sikora et al and referenced earlier, in that, inter alia, Sikora et al used intratumoral lymphocytes as opposed to peripheral blood lymphocytes, of glioma-suffering human patients to form the hybridomas. Also, Sikora et al used human lymphoblastoid cell lines as the fusion partner as opposed to human myeloma-like cell line used in the present invention.
Monoclonal antibodies derived from human cell lines as in this invention have slgnificant advantage over those derived from the cells of other mammalian species, particularly when intended for use in human subjects. It is believed that human source-derived mAbs have increased compatibility with and increased sensitivity of recognition for human target cells.
Thus, the mAbs of the present invention have beneficial application to their proposed range of uses, including ~t, A'l ~ F' i~

their use in diagnostic imaging for the radiolocalization of neurological tumors, and in immunotherapy by either direct action or by lin~ing these mAbs to chemotherapeutics which can then be targeted directly to the tumor cells.

DESCRIPTION OF THE PREFER~ED EMBODIMENTS

The present invention utilizes hybridomas derived from the fusion of a hum~n myeloma-like cell line and a lymphocyte obtained from the circulating blood of a person having a neuroectodermal tumor. The preferred myeloma-like cell lines are characterized by being deficient in the enzyme hypoxanthine-guanosine phosphoribosyl transferase (HGPRT) and by being an immunoglobulin non-secretor. The specific, most preferred such myeloma-like cell line is that designated TM-H2-SPZ. This is a sub-line of the parent line TM-H2 described by Sullivan et al (1982).
From the myeloma-like cell line TM-H~-SP2 there have been prepared, by cell fusion techniques, five novel hybridomas, each of which derives from a different lymphocyte extracted from the blood of a human patient having a neuroectodermal tumor, fused to TM-H2-SP2. These novel hybridomas, designated BT27/lA2;
BT27/2A3; BT32/A6; BT34/A5; and BT54/B8, all produce mAbs to gliomas, in relatively good yield. Viable samples of these novel hybridomas are in process of being deposited with the American Type Culture Collectlon (ATCC), Bethesda, Maryland, in accordance with the provisions of the Budapest Treaty.
A remarkable characteristic of the mAbs of the present invention is the high degree of similarity amongst the five mAbs, derived from the five hybridomas, despite being derived from cells obtained from four ~`J ~) 2.; ._. ~J ~J i~

different tumor patients, each having a different tumor type, history, or stage of tumor development.
The human myeloma-like cell line TM-H2-SP2 used in the preferred embodiment of the present 5 invention is the immunoglobulin non-secreting subline of the parent line, TM-H2, described by Sullivan, et al.
(1982). TM-H2 is a hypoxanthlne guanine phosphoribosyl transferase (EC 2.4.2.8) deficient (HGPRT-) derivative of an unknown human myeloma-like line, which was selected in 0.8% (w/w) methylcellulose for res$stance to 6-thioguanine (6 ~g/ml) and failure to grow in hypoxanthine-aminopterin-thy~id$ne (HAT) medium (Littlefield, 1964). TM-H2 produces and secretes up to 3 ~g/ml of IgG(K) in supernatant culture fluid when grown to a concentration of 5 x 10$ cells/ml in alpha-MEM
containing 10% FCS. The subline TM-H2-SP2 produces but does not secrete immunoglobulin, as determined by quantitative ELISA. TM-H2-SP2 has a 46, XX karyotype, and was grown continually in the presence of 6-thioguanine (6 ,ug/ml) to prevent HGPRT revertants. The line is negative for Epstein-Barr nuclear antigen (EBNA-), and therefore does not appear to be an EBV-immortalized derivative of lymphoid cells.

The invention i5 further illustrated in the following specific examples and experimental protocols and procedures.

MATERIALS AND METHODS
. . _ Derivation of Human Anti-Glioma Monoclonal Antibodies from Patients with Neurological Tumors Clinical Material Human patients were selected who were suspected of having a brain tumor on the basis of history, physical examination, and CT scan. From each such patient, approximately 40 ml of peripheral venous blood was withdrawn into four heparinized glass tubes (Vacutainer), and stored at room temperature until further use.
Within 2 hrs. of sampling, the lymphocytes were separated on a Ficoll-Hypaque density gradient (specific gravity 1.077 g/ml) by carefully layering two 20 ml aliquots of whole blood onto 20 ml of 5% (w/v) Ficoll-Hypaque and centrifuging at room temperature for 25 minutes at 1000 G. The lymphocytes were harvested from the interface, washed once with sterile phosphate-buffered saline (PBS) (8 g NaC1, 1.144 g Na2 HP04 (anhydrous), 0.2 g KH2 P04, and 0.2 g KCl in 1 liter deionized water and adjusted to pH 7.2), and re-suspended in 10 ml alpha minimum essential medium (alpha-MEM) containing 10~ (v/v) fetal calf serum (FCS;
Flow Laboratories, Rockville, MD), 292 mg/L L-glutamine, 44 mg/L L-asparagine, 100 U/ml penicillin, and 100 ,ug/ml streptomycin.
The lymphocytes were then counted in a hemacytometer after diluting 0.1 ml of the mixture with 0.9 ml of 2% glacial acetic acid, in order to lyse any remaining red blood cells. If human myeloma-like ~ ~' C`J ~' ~ ''` ~3 TM-H2-SP2 cells in logarithmic growth phase were avallable, a fuslon procedure would be carried out immediately (see below); otherwise the lymphocyte mixture was stored at 4C if it was felt that the TM-H2-SP2 line would be ready for fusion within the next 24 hrs. If no fusion procedure was to be carried out, the lymphocyte mlxture was re-suspended in 1 ml alpha-MEM + 10% FCS, and stored at -70C in a cryotube ~Nunclon, Denmark) containing 0.1 ml sterile dimethyl 10 sulfoxide (DMS0) as a preservatlve.
At the time of surgery, a portion of autologous tumor (1-5 gm) was removed and placed in sterile Elliott's solutlon (Abbott Laboratories Inc.) for preparation of KC1 extracts. After mincing with fine 15 scissors, fresh tumor was placed in a conical test tube and agitated overnight at 4C with 3 M KCl (1 ml/gm wet tissue) The crude cell lysate w~s diluted 1:10 with ddH20, and clarified by high speed centrifugation at 12,000 G for 20 min . at 4C. The extract was then 20 dialyzed and equilibrated to PBS over the next 6 hrs, and then filter-sterilized with a 0.22 ~m Millipore filter. The protein concentration of the tumor extract was determined using the method of Lowry et al. (1951), and the necessary dilutions made with PBS to bring the 25 final protein concentration to 100 ~g/ml. Aliquots were stored at -20C until ready for use in the screening assay (see below).

Human Glioma Cell Lines The human glioma cell lines used in this study were obtained from other research institutions.
All cell lines were grown in (alpha-MEM
containing 10~ fetal calf serum (FCS; Flow 35 Laboratories), 292 mg/l L-glutamine, 44 mg/l L-asparagine (anhydrous), 100 U/ml penicillin, and _ g _ 100 ~g/ml streptomycin in the presence of 5% C02 atmosphere at 37C. The cells were grown to confluence in standard tissue culture flasks (Nunclon, Denmark), and passaged in a split ratio of 1:2 using citrated PBS
(Gibco, Grand Island, NY) containing 0. l~ trypsin (Difco Laboratories, Detroit, MI) to remove the cells, and FCS to stop the action of the enzyme on the cells.

Fusion Procedure to Prepare Hvbridomas The method used was a variation of the original method of ~ohler and Milstein (1975).
Twenty-four hours prior to fusion, the TM-H2-SP2 cell line culture in mid- to late-logarithmic phase was dlluted 1:1 with fresh alpha-MEM containing 10% FCS and 6-thioguanine (6 ~g/ml). This was done to ensure a ma~ority of cells were cycling in growth phase. At the time of fusion, an aliquot of TM-H2-SP2 cells was stained with an equal volume of 0.16~ trypan blue, and the viable cell count was determined in a hemacytometer by the dye exclusion method (Gorer and O'Gorman, 1956).
If the viable cell count was below B0~ the cells were not used on that day.
The TM-H2-SP2 cells were then combined directly with the peripheral lymphocytes in a lymphocyte to myeloma cell ratio of 4:1, and pelleted together by centrifugation at 500 G for 5 min. The supernatant was discarded, and the cells were fused under serum-free conditions by the gradual addition of 1 ml of 50 polyethylene glycol (m.w. 1450) (Sigma, St. Louis, M0) diluted 1:1 in serum-free alpha-MEM over 1 minute (Pontevorco, 1976). This was followed by dilution over the next 4-5 minutes with lO ml of serum-free alpha-MEM, of which 2 ml ~ere added with gentle stirring over the first 2 minutes.

~PJ~

The fusion mixture was then pelleted by centrifugation as before, and washed twice with 10 ml of hypoxanthine-aminopterin-thymidine (HAT) medium containing 10% FCS, lO- 4 M hypoxanthine, 4 x 10- 7 M
aminopterin, 1.6 x 10-5 M thymidine, 292 mg/l L-glutamine, 44 mg/l L-asparagine, 100 U/ml penicillin, and 100 ~g/ml streptomycin. A viable cell count on 0.1 ml of the fusion mixture was performed as before, and the cells diluted to a concentration of either 1 x lOs or 2.5 x 105 viable myeloma cells/ml by addition of fresh HAT medium. The cells were then dispensed in 200 ,ul aliquots into 96-microwell trays (Linbro Laboratories Inc., McLean, VA; 0.28 cm2/well), and incubated at 37C
in 5% CO2 atmosphere.
Culture Conditions The hybridomas were maintained in HAT medium for 7 days post-fusion, with exchange of half the culture media on day 5, followed by re-feeding with fresh HA~ medium. On day 7, 50~ of the culture media was again removed, and the hybridomas re-fed with hypoxanthlne-thymidine (HT) medium (HAT medium minus aminopterin). This process of exchange feeding with HT
medium was continued every 5 days untll the hybridomas became microscopically visible (approximately 46 weeks post-fusion, at which point they were ready for testing by ELISA. Selected hybridomas were continuously propagated up to transfer into 25 cmZ culture flasks (Nunclon, Denmark) in a volume of 2 ml via 24 well trays (Linbro Laboratories Inc., McLean, VA), then stored at -70C in 10% FCS and 10% DMSO and alpha-MEM. During culture, all hybridomas were maintained con~inually in HT medium.

~ J ~ .; .,/; J 3 Tumor Extract ELISA

The screening method used was a modification of Douillard and Hoffman (1983). Briefly, 96-well polyvinyl chloride Microliter ELISA plates (Dynatech, Alexandria, VA) were incubated for 12-18 hours at 4C
with 200 ~l/well of 3 M KCl autologous tumor extract as described under "Clinical Material" at a concentration of 5 ~g proteln/ml (1 ,ug/well) in carbonate-bicarbonate 10 buffer, pH 9.6 (0.8 gm Na2CO3, 1.47 gm NaHC03, 0.1 gm NaN3 made up to 500 ml with ddH20). The plates were washed 3 times with PBS containing 0.05% (v/v) Tween-20R
(PBS-Tween; Sigma, St. Louis, M0) Then 50 ~1 of the hybridoma supernatant to be tested was added to 50 ,ul 15 PBS per well.
Culture supernatant from the parental myeloma line, TM-H2, containing IgG(K), was used as a nonspecific immunoglobulin control for the initial 15 fusions, whereas culture supernatant from one of the 20 hybridomas (BT27/2D2) containing 1-2 ~g/ml IgM was used as a control in subsequent screening assays.
After 2 hours incubation at room temperature, th~ plates were washed 3 times with PBS-Tween, and 200 ~l/well alkaline phosphatase (ALP)-conjugated goat 25 anti-human ~- and gamma-chain plus goat anti human kappa and lambda immunoglobulin light chains (TAG0, Burlingame, CA) were added at 1:2000 dilution in 1%
bovine serum albumin (BSA; Fraction V, Sigma, St. Louis, M0) in PBS. After a further 2 hours incubation at room 30 temperature, the plates were again washed 3 times with PBS-Tween, and developed with freshly prepared Sigma 104 phosphatase substrate (p-nitrophenyl phosphate disodium;
Sigma, St. Louis, M0), 1 mg/ml in diethanolamine buffer, pH 9.8 (97 ml diethanolamine, lO0 mg MgCl2.6H20, 0.2 gm 35 NaN3, ddH20 added to bring volume to lL) . The plates were left at room temperature for 20-30 minutes, then phosphatase activity was determined as absorbance at 405 nm using a Dynatech Microelisa autoreader (Dynatech, Alexandria, VA). A hybridoma supernatant was considered to be 'positive' if phosphatase activity exceeded the mean background level of wells with control culture supernatants by greater than 2 standard deviations.

Flxed Cells ELISA

The method used was adapted ~rom Suter, et al.
(1980). Flexible polyvinyl chloride Microliter ELISA
plates were coated with 100 ~l/well freshly prepared poly-L-lysine (50 ~g/ml in PBS), and allowed to stand at room temperature for 40 min. The outer rows and columns were not used because they tended to give false positive absorbance values for this particular ELISA assay. The poly-L-lysine was then removed, and the plates washed once with PBS. Freshly harvested human glioma cells, which had grown to confluence in alpha-MEM in a 175 cm2 tlssue culture flask, were removed in PBS with a rubber policeman and seeded into microwells at 105 cells per well in 100 ~1 PBS. The plates were allowed to incubate for 45 minutes at room temperature, then gently rinsed 3 times with PBS.
The cells were fixed by the addition o~ 100 ~l/well of 0.1% glutaraldehyde in PBS for 3 minutes, then gently r~nsed 3 times with PBS. To block any residual glutaraldehyde activity, 100 ~l/well of PBS
containing 0.2% gelatin 275 Bloom (Fisher, Montreal, QC) with 0.2% NaN3 (PBS-gelatin) was added, and the plates allowed to incubate at 4C for at least 12 hours before proceeding further with the assay. In this condition, the plates could be stored at 4C for up to 6 months.
On the day of the assay, the PBS-gelatin was removed by inversion of the plate, and lQO ~1 of hybridoma supernatant was added directly to each ~ J 3 ~ ~ ~ ~

microwell without prior washing. Control supernatants for this assay were the same as for the autologous tumor extract assay described in the previous section. The plates were incubated for 30 minutes at 37C, then cooled for 15 minutes at room temperature. They were washed 5 times with freshly prepared PBS containing 0.05% (v/v) Tween-20R and 0.1% (v/v) gelatin (PBS-Tween-gelatin), and the microwells filled with 100 ,ul of ALP-conjugated goat anti-human kappa and goat anti-human lambda immunoglo~ulin light chalns diluted l:1,000 each in P~S-Tween-gelatin.
After 30 minutes incubation at 37C and 15 minutes cooling at room temperature, the plates were washed 3 times with PBS-Tween-gelatin. Phosphatase activity was developed with the addition of 200 ~l/well Sigma 104 phosphatase substrate solution (l mg/ml) freshly made in diethanolamine bu~fer (97 ml diethanolamine, 800 ml deionized water, 0.2 g NaN3, and lO0 mg MgCl2.6H20, made up to 1 liter with deionized water and ad~usted to pH 9.8 with HCl~, and assessed as absorbance at 405 nm using an automated Dynatek Microelisa reader. Hybridoma microwells with phosphatase activity which exceeded mean background levels of ~he control supernatant wells by greater than 3 standard deviations were considered 'positive' ir. this assay.

Flow Cvtometry Confluent phase cultures of the cell lines were gently detached from tissue culture flasks with a rubber policeman and collected in ice cold PBS
containing 1% FCS (In general, all cell preparations were scraped. In some experiments, however, the cells were recovered by trypsinization, which did not result in any differences in labelling). Approximately l x 106 cells per sample were aliquoted into 10 ml test tubes, - 14 ~

~ ~ 3~

and pelleted by centrifugation for 5 min at 500 G. The PBS-l~ FCS was poured off and 300 ~1 of spent hybrldoma supernatant was added to each sample. After resuspending the pellet, the cells were placed on ice for 60 min.
Following 3 washes with 5 ml cold PBS-1% FCS, lO0 ,ul of FITC-con;ugated goat anti-human IgM (IgG fraction, ~u chain specific; Cappel Laboratories, Cochranville, PA), diluted 1:10 in sterile P~S, was added to each sample.
The cells were then placed on ice for 60 min, and washed 3 times w~th PBS-1% F~S as before. After the final wash, the cell pellet was resuspended in 500 ,ul of PBS-l~ FC~, and transported to the flow cytometer on ice.
The apparatus used in these experiments was a FACS-III fluorescence-activated cell sorter (Beckton-Dickinson FACS Systems, Mountain View, CA) linked to a custom-built microprocessor-based system capable of performing realtime correlated acquisition, storage, and display of multiparameter (3-parameter) data (Stewart and Price, 1986~. Prior to each experiment, the FACS-III was calibrated and standardized for small angle light scatter and fluorescence intensity with polystyrene latex microspheres and "Fluoresbrite"-carboxylated microspheres (Polysciences IncA, Warrington PA). Fluorescence intensity between 530 and 560 nm was measured on all samples using an excitation wavelength of 488 nm. Logarithmic small angle scatter, fluorescence, and 90 scatter data were collected on a minimum of 2 x 10~ gated cells from each sample, and the information stored on flexible diskettes. Fluorescein-labelled preparations are excited at 488 nm, and fluorescence is measured between 530 nm and 560 nm using a suitable combination of optical filters. Rhodamine-labelled preparations were excited at 514 nm, and fluorescence was measured above 620 nm.

rd ~ . J ~

Freezing and Thawing Procedures All cells (gliomas, hybridomas, etc.) were frozen in essentially the same manner. Approximately 107-iO~ cells were pelleted by centrifugation at 500 G
for 5 minutes, and the supernatants were discarded. The cells were resuspended in 1 ml of the usual tissue culture media, and pipetted into a 43 x 12.5 mm cryotube (Nunclon, Denmark) . Sterile DMSO (0.1 ml~ was added to 10 the cryotube, and the cells placed directly into the -70~C freezer. After 24 hours, the cells could be removed to liquid nitrogen for long term storage.
Cells were thawed by removing a cryotube from the -70C freezer and placing it directly in a 50C
15 water bath for 70 seconds. Before the ice pellet was fully dissolved, the contents of the cryotube were emptied into a screw-cap conical tube (Sarstedt, W.
Germany) containing 50 ml of tissue culture medium, and pelleted by centrifugation at 500 G for 5 minutes. The 20 supernatant was then discarded, and the cells re-suspended in fresh media and plated or placed in suspension culture according to the cells' requirements.

KarYotyping of_HYbridomas Metaphase chromosome spreads were prepared by first culturing =1 x 106 hybridoma cells overnight in 3 ml of HT medium containing 10~ FCS and 0.06 ~g/ml colcemid (Fisher). The next day, the cells were washed 30 in PBS and pelleted before resuspension in 3 ml 0.07 M
KCl for 20 min at room temperature. Then lO ml of methanol:acetic acid (3:1) was added, and the cells were pelleted at 1700 rpm. The cells were washed twice again with methanol:acetic acid- before re-suspension in 1 ml 35 of methanol:acetic acid. The mixture was dropped from a height of approximately 2 meters onto pre-warmed glass slides in order to burst the cells and spread their chromosomes. The cells were stained for 3 min with Giemsa (Fisher, Montreal, QC~, followed by mounting with Permount (Fisher, Montreal, QC). Fifteen to 20 metaphase spreads were counted to establish the chromosome number distribution.

Southern Blotting Genomic DNA was isolated from TM-H2-SP2, BT27/lA2, BT27/2A3, BT32/A6.5 (a subclone of BT32/A6), and BT27/2D2 cells according to Maniatis, et al. (1982).
T~n ~g of DNA were digested overnight at 37C with Bam HI and Hind III restriction enzymes (23 units/~g DNA;
Boehinger-Mannheim, West Germany) according to the manufacturer's specified conditions. After digestion, the DNA was subJected to electrophoresis on a 0.8% (w/v) agarose gel (Maniatis, et al. 1982), and transferred to a Genescreen P1US membrane (NEN, Boston, MA) by the usual method. Following prehybridlzation with herring sperm DNA (80ehingerMannheim, West Germany), the Southern blot was hybridized (Kaufmann, et al. 1985) overnight at 65~C with a 3 2 P-labelled JH probe (Oncor Inc., Gaithersburg, MD), specific activity of 5-6 x 108/~g. The probe includes the entire human JH region, and totals 5.6 kbp (Revetch, et al. 1981). Following washing, the blot was exposed to Kodak X-ray film with intensifying screens at -70C.

Characterization of Human Monoclonal Antibodies to -Glioma Cell Surface Antigens Determination of Immunoqlobulin _Isotype and Concentration Immunoglobulin chain class was determined by coating four sets of 96-well microliter ELISA plates with goat anti-human Ig (IgA + IgG + IgM) (Cappel Laboratories, Cochranville, PA ) at 1:4500 dilùtion.
Four replicates of the supernatants to be tested were added to each of the plates, and the latter incubated at room temperature for the usual 2 hrs. Spent culture media from the parental line, TM-H2, served as a source for kappa and gamma chains, and supernatant from the human B-lymphocyte line, RPMI 1788 (Huang and Moore, 1969), as a source for lambda and ,u chains. One plate was then labelled with ALP-con~ugated goat anti-human kappa chains, another with ALP-conjugated anti-human lambda chains, the third with ALP-con;ugated anti-human gamma chains, and the last plate with ALP-conjugated anti-human ~- chains. All ALP-con~ugated antisera (Tago Inc., Burlingame, CA) were used at a dilution of 1-2000 in P8S and 1% BSA. The ELISA plates were developed with Sigma 104 phosphatase substrate in the usual manner, and O.D. measurements performed at 450 nm.
Measurement of human IgM concentrations in spent hybridoma culture media from the 3 hybridomas was performed by quantitative ELISA some 10 months post-fusion for BT27/lA2 and BT27/2A3, and 9 months after the fusion date for BT32/A6. From the time of fusion to the date the ELISA was performed, the hybridomas were grown continuously in vitro for a total of 4-6 months.
On the day the ELISA was performed, the hybridomas were growing in logarithmic phase at a ~sJc.~

concentration of 5 x 105 cells/ml, in culture media which had been completely replenished 5 days previously.
Concentrations of Ig were determined by quantitative ELISA.
For each hybridoma supernatant to be tested, one 96-well ELISA plate was coated wlth goat anti-human Ig (IgA + IgG + IgM) (Cappel Laboratories, Cochranville, PA) at 1:4500 dilutlon. Affinity purified human IgM
(Cappel Laboratories, Cochranville, PA) at a concentration of 5.5 mg/ml was diluted in PBS to a final concentration of 500 ~g/ml. From the latter solution, dllutions in PBS were prepared with concentrations of 50 ug/ml, 5 ,ug/ml, 0.5 ~g/ml, and 0.05 ~g/ml.
The IgM standards were then applied to the ELISA plates, along with the supernatants to be tested.
ALP-con~ugated go~t anti-human IgM (~ chain specific;
Tago, Inc., Burlingame, CA) was added as a second antibody, and the plates developed with Sigma 104 phosphatase substrate in the usual manner. A standard curve of IgM concentration versus O.D. was plotted using a 4-cycle, semi-logarithmic scale, and the mean O.D.
values for the hybridoma supernatants converted to IgM
concentration.

Antibodv Dilution Curves Spent culture supernatants from each of the hybridomas were diluted in non-specific control IgM
(BT27/2D2) and reacted in an ELISA assay with the 3 M
KC1 extract of BT-27 (high grade astrocytoma).
Undiluted control supernatant from BT27/2D2 was also included in each ELISA plate for purposes of compa~ison.
For each ELISA plate, the mean background (BT27/2D2) O.D. value was subtracted from the mean O.D.
value for each dilution, and the difference (delta O~Do ) plotted as a function of dilution. Mean O.D. values which were greater than background were compared with the latter using Student's t-test for unpaired data.

Antibody Affinity ELISA

The method used is a modification of De Bernardo and Davies (1987). Briefly, a 96-well ELISA
plate was coated with the 3 M KCl extract of brain tumor BT-27 (high grade astrocytoma)~ The plate was washed 3 times with PBS containing 0.05% Tween-20R (PBS-Tween), and the wells filled with 50 ,ul of PBS. Fifty ~l of spent hybridoma culture supernatant from each of the 5 HmAbs (BT27/lA2, BT27/2A3, BT32/A6, BT34/A5, BT54/B8), as well as control IgM (BT27/2D2) were added to half of the wells. Fifty ,ul of PBS was added to the remaining wells.
After 14 hrs incubation at 4C, 50 ~l of fluid was withdrawn from the PBS-containing wells, and hybridoma supernatants added back. The ELISA plate was then incubated for a further 4 hrs at 4C, after which it was emptied, and washed 3 times with PBS-Tween.
ALP-conjugated goat anti-human IgM (,u chain specific;
Tago, Inc., Burlingame, CA) diluted 1:2000 in 1~ BSA-PBS
was then added to the wells, and the incubation continued for a further 2 hrs at room temperature. The plate was developed with Sigma 104 phosphatase substrate in the usual manner, and mean O.D. values for each supernatant and incubation time were calculated.
Statistlcal comparisons between different supernatants incubated for the same length of time were carried out using Student's t-test, or the Mann-Whitney ~-test, where two variances differed significantly according to the F-test (p<0.05). Comparisons of the same supernatant incubated for different lengths of time were also made. The results were plotted on a bar graph as mean differences ~ standard errors.

~ 3,~

Antigen Expression and Glioma Culture Densitv The human glioma cell line, SK-MG 1, was grown to confluence in T25 flasks (Nunc, Denmark) under standard conditions, and removed with trypsin. The cells were diluted in (alpha-MEM contalning 10% FCS and re-seeded into T25 flasks at varlable split ratios ranging from 1:2 to 1:8. After 72 hrs in culture without any exchanges of media, the old culture media was poured off, and ice cold PBS containing 1% FCS was added to the flasks. The cells were then gently scraped off with the aid of a rubber policeman, and labelled with either BT27/2A3, or spent culture supernatant from the parental myeloma line, TM-H2.
Rhodamine isothiocyanate (RITC)-conjugated goat antihuman IgM (F(ab')2 fraction, ,u chain specific;
Cappel Laboratories, Cochranville, PA) was used as a second antibody for the BT27/2A3-labelled cells, and RITC-con~ugated goat anti-human IgG (F(ab')2 fraction, gamma chain speciflc; Cappel Laboratories, Cochranville, PA) was used for the TM-H2-labelled cells. Flow cytometric analysis was then performed as described above.

PrimarY Tumor Explant Cultures Sterile surgical biopsy specimens of primary neuroectodermal tumors were collected in ice cold MEM
containing 20~ FCS. Approximately 0.5-1 g of specimen was finely minced with sterile scissors and 5-10 pieces measuring 1 mm3 were placed in T25 flasks containing 10 ml culture media. After 7 days of incubation under standard conditlons, the media was replenished and the cultures returned to the incubator. In 2-3 weeks, the explants formed confluent monolayers which could be passaged and analyzed by FCM.

~ 3 t,, Four such explants were labelled with BT27/2A3 and RITC-conjugated goat anti-human IgM (~ chain specific; Cappel Laboratories, Cochranville, PA). The percentage of reactive cells was estimated by overlapping the BT27/2A3 and control (BT27/2D2) fluorescence distributions (plotted to peak height), and determining the channel number where the two distributions crossed. The percentage of cells lying to the right of this plot for the control distribution was then subtracted from the BT27/2A3 distribution, to give a minimal estimate of the number of cells detected as significantly labelled by the HmAb.

Lipid Dot Blots and Immunochromato~raphy Total lipid extracts of cell lines to be analyz d were prepared according to Svennerholm and Fredman (1980). Briefly, approximately 5 x 108 cells were grown to confluence in standard 850 cm2 tissue culture roller bottles, scraped off in PBS, and homogenized in 3 ml ddH20 at 4 C. The homogenate was then combined with 10.8 ml methanol, and 5.4 ml chloroform, which was added dropwise with constant stirring. After mixin~ for 30 minutes at room temperature, the remaining unlysed cells and debris were spun down at 10,000 rpm for 10 min. The remaining supernatant was placed in a 37C water bath and evaporated to dryness under a constant stream of N2.
The residue was dissolved in 200 ~1 of 2:1 chloroform :
methanol, clarified by centrifugation in an Eppendorf microcentrifuge for 10 min, and stored at -20C in a tightly capped freezer vial (Nunclon, Denmark).
Dot blots were prepared by spotting 2 ,ul of each cell line to be tested onto Merck HPTLC alumlnum backed silica gel 60 plates (Applied Analytical Industries, Wilmington, NC). The plates were then J ~ J ~

dipped in a solution of 0.05% polyisobutyl-methylmethacrylate (Polyscience, Inc., Warrrington~ PA) in hexane for 30-60 seconds, and air dried for 15 min.
After blocking with 1% BSA in Tris-buffered saline (20mM
Tris-HCl, 0.5 M NaCl, pH 7.5; TBS ) for 2 hrs. at 4C, the blots were exposed to hybridoma supernatants for an addltional 2 hrs. at 4C, then washed with TBS six times. Peroxidase conjugated goat anti-human IgM (u chain specific; Kirkegaard and Perry, Gaithersburg, MD) 10 diluted 1:400 in 1% BSA-TBS was then added for a further 2 hrs. at 4C, and the plates washed once again in TBS.
The peroxidase reaction was developed with 3,3'-diamino-benzidine tetrahydrochloride (DAB) dissolved in 0.05 M
Tris-HCl (pH 7.5) at a final concentration of 1 mg/ml, 15 in the presence of 0.05% H20z. After rinsing with ddH20, the plates were dried with a hair dryer and photographed promptly.
For thin layer lmmunochromatograms, 4 ,ul of lipid extract was spotted onto each lane of the HPTLC
20 plates and dried under vacuum for at least 4 hrs. The plates were then chromatographed in chloroform:methanol:
water (60:35:8) and dried under vacuum overnight, at least 16 hrs. The plates were then coated with polyisobutylmethylmethacrylate as described above. The 25 plates were blocked with 5% BSA in TBS at 4C for 2 hrs.
The chromatograms were then exposed to hybridoma supernatants for an additional 2 hrs at 4C and washed as described above. Peroxidase conJugated gozt anti-human IgG SH~L chain specific) (BIO/CAN, Toronto, ON) 30 diluted 1:2000 or peroxidase con~ugated goat anti-human IgM (Fc specific) (BIO/CAN, Toronto, ON) diluted 1:2000 in 1% BSA-TBS was added for a further 2 hrs at 4C.
After washing again, the plates were developed as d~scribed for the lipid dot blots.

2V3~vJ~

Labelling of Normal Human Astrocytes Cultured normal human astrocytes were obtained as a gift from the laboratory of Dr. Jack Antel, Montreal Neurological Institute, Montreal, PQ. The cells were derived from patients undergoing craniotomy for epilepsy and were grown on glass coverslips. They were labelled on the coverslips using the same general method (reagents and time) as outlined in the section on Flow Cytometry with the exception that all washes were carried out on coverslips using PBS containing 1% FCS
and 0.02% (w/v) NaN3. The coverslips were wet-mounted using PBS with FCS and azide and viewed with a Leitz fluorescence microscope.
Cell Sorting An established culture of SK-MG-l (98th passage) waQ scraped with a sterile rubber policeman, resuspended in 1 ml of alpha-MEM ~ 10% FCS, and transported to the FCM on ice. After the usual calibration of the FCM, simultaneous fluorescence and small angle light scatter data were collected on 5 x 104 cells in order to establish 2-dimenslonal gates for the purpose of sorting. Each area was chosen so as to correspond to either a S-(small or low small angle light scatter) cell subpopulation or to a L-(large or high small angle light scatter) cell population. A total of 2 x 104 cells from each area were sorted and collected into 1 ml of alpha-MEM + 10% FCS and placed on ice.
Each sample of sorted cells was then diluted wlth fresh alpha-MEM + 10% FCS to a volume of 40 ml and distrlbuted into two 96-microwell plate-~, approx. 200 ~1 per well. The cells were fed every 5 days by exchanging half the old supernatant for fresh media. On the 10th day following the sort, the total number of colonies /

~ 3 ~ ' 3 j 3 containing >25 cells were counted in each of the 4 plates. An individual microwell might contain anywhere from 1 - 6 individually distlnguishable colonies.
Ten S-cell-derived and 10 L-cell-derived clones from wells with a single colony were then selected and passaged further into 24-well plates (Linbro; 2.01 cm2/well) in a volume of 1 ml alpha-MEM +
10% FCS. The S-cell clones were assigned the numbers S101 - SllO and the L-cell clones were assigned L101-L110. The clones were later tested for reactivity by FCM as described above.

Cell Cycle AnalYsis Confluent cultures of the human glioma cell line, SK-MG-l, were detached from tissue culture flasks with a rubber policeman and fixed in ice cold 70~ (v/v) ethanol for 30-60 min. Approximately 1 x 106 cells per sample were aliquoted into 10 ml test tubes, pelleted by centrifugation for 5 min at 500 G, and washed in PBS.
Twenty ,ul of RNase (Boehringer Mannheim, West Germany;
1 mg/ml in 10 mM Tris-HCl, 5 mM NaCl, pH 7.5) was added to each sample to digest RNA, along with 1 ml of propidium iodide (50 ,ug/ml in 10 mM Tris-HCl, 5 mM
MgCl2, pH 7.4). After incubating for 60 min at 37~C, the cells were placed on ice for an additional 30 min, then transported over to the FACS-III apparatus on ice.
After calibrating for linear small angle light scatter, red fluorescence, and DNA content using the human pseudodiploid cell line, HL-60, data was collected on a minimum of 2 x 104 gated cells and stored on flexible disks for subsequent analysis.

J ~ J ~j 3 Cell Morphology After labelling SK-MG-1 for FCM with BT27/2A3 and BT27/2D2 as desribed above, small angle light scatter vs. fluorescence data were collected on 5 x 10~
cells in order to establish 2-dimensional windows for the purposes of sorting. The windows were chosen in order to define morphology of:
1) S-cells which increase in fluorescence after 10 labelling with ~T27/2A3;
2) S-cells which remained unchanged in terms of fluorescence after labelling with BT27/2A3; and 3) L-cells.
Approximately 1 x 104 cells were sorted from 15 each window and collected in ice cold alpha-MEM + 10~
FCS. Samples from each sorted area, as well as total unsorted cells, were then centrifuged onto glass slides, air dried, and stained with Wright's stain (Fisher, Montreal, PQ). The cytocentrifuge preparations were 20 then viewed through a Leitz microscope and photographed with Fu~icolor HR film (ASA 400).

Karyoty~g_of SK-MG-1 Sublines Metaphase chromosome spreads were prepared as out,ined above. For each cell to be studied, one confluent T25 flask (Nunc, Denmark) was incubated overnight in 3 ml alpha-MEM + 10~ FCS and 0.06 ~g/ml colcemid (Gibco, Grand Island, NY). The remainder of 30 the procedure was followed exactly as outlined above (see "Karyotyping of Hybridomas").

.'~ ,J ~.~ .,' J

Cell fusions were performed, to make hybridomas from human myeloma-like cell line TM-H2-SP2 and peripheral blood lymphocytes (P~L) obtained from eight different brain tumor patients, coded 8T-24, BT-27, BT-32, ~T-34, BT-38, BT-39, BT-54 and BT-55. These patients, both male and female, ranged from 5 to 55 years of age and had a variety of neuroectodermal tumors including fibrillary astrocytoma, oligoastrocytoma, primitive neuroectodermal tumor, and glioblastoma. The fusions were performed as previously described.
Incubation of the cells in the 96-microwell trays was conducted using fresh tissue culture media and fetal calf serum (FCS), supplemented with L-glutamine (292 mg/l) and L-asparagine (44 mg/l). The hybridoma outgrowth, defined as the number of microwells containing macroscopic colon$es of greater than 50-100 cells (visible to the naked eye) divided by the total number seeded, and expressed as a percentage, ranged from 0~19.5%.
The number of PBL in all samples available for fusion varied from 6.0 x 106 to 5.8 x 107. Although the ratio of myeloma cells to lymphocytes remained constant at 1:4 for each fusion, the plating density that was used ranged from l.0 x 105 to 2.5 x 105 myeloma cells per ml of fusion mixture. The lower plating density was chosen in some fusions in attempts to encourage better initial monoclonality, whereas the higher plating density seemed to favor improvad hybridoma outgrowth.
The concentration of PEG was 50~ (v/v) in all fusions.
Typically, discernable hybridoma growth appeared from 4 to 6 weeks after a fusion, and new growth in microwells often continued to be observed for several we~ks thereafter. In general, those colonies of /Js~ 5 hybridomas which appeared earliest tended to bè the most stable. Not all macroscopic colonies visible at the end of the 96 microwell (0.28 cm2/well) culture period could be successfully propagated to amounts suitable for yrowth in 24 well dishes (2.01 cm2/well). In fusion BT-54, for example, of the seven colonies which originally grew in 96 microwells, only one was capable of prolonged growth in vitro. Instances of hybridoma growth failure occurring later than three months post-fusion were not observed.

The hybridomas were screened for the presence of human immunoglobulin and reactivity with 3 M KCl extracts of autologous tumors, or glutaraldehyde-fixed glioma cell lines.
A total of 1,121 wells containing growth of putative hybridomas from the fusions described in Example 1 were screened, of which 162 (14.5%) were ~ound to react with tumor extracts, or glioma cell lines. Two of the fusions, namely RT-27 and BT-32, were screened for reactivity with several glioma cell lines, and instances were found of individual microwells which tested positive in multiple ELISA assays.
In the last two fusions, BT-54 and BT-55, the ELISA assay was modified to detect only those microwells which contained reactive IgM species. This was accomplished by substituting culture supernatant from hybridoma BT27/2D2 for TM-H2 as a control, and adding alkaline phosphatase conjugated goat anti-human IgM as a second antibody. The reason for this was that analysis of immunoglobulin chains present in 34 hybridomas from fusions BT-24, BT-27, BT-32 and BT-34, indicated that all 34 contained IgM. Careful study of a few of these hybridomas indicated that the IgM alone was responsible for anti-tumor activity. Hybridoma BT27/2D2, derived from cells of patient BT-27, a hybridoma which was consistently negative in various ELISA tests of immune reactivity, was chosen as a non-specific IgM control (1-4 ,ug/ml).
The fact that all five HmAbs in this study were of the IgM isotype may be significant. This may be a result of the immunosuppressive mechanisms, discussed earlier, present in glioma patients which might prevent the elaboration of a secondary (IgG) immune response in these patients. Alternatively, this may simply be a reflection of the type of antigen which the five HmAbs recognize, i.e. a carbohydrate.

All five hybridoma supernatants were found to contain ,u heavy chains. Only BT34/A5 contained lambda light chains, the other four HmAbs contained kappa light chains. Each HmAb was shown to label the cell surface of human glioma line SK-MG-l using FCM and FITC~
conjugated goat anti-human IgM (IgG fraction, ~ chain specific; Cappel Laboratories, Cochranville, PA), indicating that the IgM molecules were binding to the tumor cell surface membrane (data not shown) . No gamma chains were detected in any of the hybridoma supernatants.
After approximately six months of maintenance in culture, the estimated IgM concentration for BT27/lA2 was 5.0 ,ug/ml; for BT27/2A3, it was 44 ,ug/ml, for BT32/A6 3.5 ,ug/ml, for BT34/A5 2.4 ~g/ml, and for BT54/B8 22.4 ~g/ml. After a further 6 months in continuous culture, IgM production levels were found to be comparable.

~ )J

A total of nine high immunoglobulin-producing hybridomas, from the above fusion series, as described in Example 1, were screened for reactivity with the human glioma line SK-MG-l, using the FACS-III Flow Cytometer. Five supernatants derived from the hybridomas, designated as BT27/lA2, BT27/2A3, BT32/A6, BT34/A5 and BT54/B8 were found to label this particular glioma cell line. All five contained tumor-reactive IgM
species. Thus, from eight fusions and a total of 59 hybridomas which were macroscopically visible at six weeks, only five (8.4~) reacted with the cell surface of a glioma cell line, and were capable of sustained growth in culture. It was noted that BT27/2A3 labelling of SK-MG-l is enhanced if the latter is maintained at a high cell density prior to testing.

Three of the five HmAbs were tested by ELISA
for their relative reactivities. At the initial concentration of supernatant fluid which was tested, the order of reactivity was BT27/2A3 > BT27/lA2 > BT32/A6, which is the same as the order of their respective IgM
concentrations (see Example 3). The final titre which gave O.D. readings significantly higher than control IgM
baseline was 1:16 for BT27/2A3, 1:4 for BT27/lA2, and 1:2 for BT32/A6. Since this assay involves the addition of 50 ,ul of hybridoma supernatant to 50 ,ul of PBS in the initial step, the initial dilution tested in the ELISA
was 1:2. There was no evidence of an initial plateau phase in any of the dilution curves, indicating that under the ELISA conditions which were used, the quantity of tumor extract was not a limiting factor.

~ J 3 r~ v O ( 1 . _ Flow cytometric analys:Ls of cultured human cell lines and strains was performed with the cells grown to confluence and maintained in that state with exchanges of fresh media at least 24 hrs prior to labelling. Suspension culture cell lines were maintained and used from high cell density cultures, i.e. near saturation. Screening was carried out according to the method outlined above in parallel on two or more separate occasions with established positive cell lines to verify maintenance of immune reactivity of the HmAbs.
Results of FCM screening with human cell lines are summarized in Tables 1 and 2. Several different classes of neuroectodermal, and non-neuroectodermal tumors and tissues were tested. All five HmAbs demonstrated a similar pattern of reactivity for the 30 cell lines which were studied, with few notable exceptions. For example, the human epithelial cerYical carcinoma cell line, ME180, reacted with BT27/lA2 and BT27/2A3, but not BT32/A6. Antibodies BT34/A5 and BT54/B8 both failed to react with the glioma cell line SKI-l, but labelled melanoma line M-4.
Only HmAb BT32/A6 had a singular pattern of reactivity. Antibodies BT27/lA2 and BT27/2A3 exhibited one pattern of reactivity, and antibodies BT34/A5 and BT54/B8 had another pattern of reactivity, which differed slightly from each other. None of the HmAbs reacted with any of the hematological cells lines which were tested.
Despite some minor differences in the pattern of reactivity against a panel of human tumor cell lines, all five HmAbs appear to be recognizing similar molecular substances, both in terms of biochemical composition and biological distribution even though ~ ~J

T~lo ~: R-act~ty o~ t~ao l~ab~

T~L~ ~Jg~e ~ ~ ~ C12~
Glioma SK-MG-l PoJltlv~b po~ltlvs po~itlve S~-MG-13 po~Cl~- po~ltl~- po iti~
SXI-l po~ltlv- po~ltlve positiv~
LN-21S n~g~tlv~ n-g~tlv~ negaCive LN-340 nay~lv- noqatl~- nagatlvo U-178 posltlvo positiv~ po~itiv~
U-373 po8itlv~ po~ltlvo po~itlve Melanoma M-4 ney~tlvd n~g-t$v~ neg~tiYe IGR 37 n~g-tlve n~gativa negative ~GR-39 neg~tivo neg~lv- negativ2 NeuroblaRtoma IMR-32 neg~tlv- n-ga~lv- n~gative S~-N-MC n-g~tlv~ nngatlva negative Retlnobla~toma Y-76 nog~tlv- n-gatlvo n~gative He~tologlcal c~ c n-g~tlv~ negat~v- negaelve K-562d n-g~tlvo nogatlvo negative HL-60- n-ga~lv- nng~tlv- negativ~
Other Tumor-~ H-L ~ pos~elv- po~ltlve posittvo M~l~O~ po~ltlve posi~iv~ negsCive c-33af n-g~tlvo n~gatlv~ n~gativ~
S~1166g n~g~tivn nagntiva n~g~tiv2 S~1~17g n-g~tlve n~gatlvoe n~gativ~

5~94Eg n g~tl~- nogatlvo negaeiv2 HS-29g ncga~lv- nagatlv- n~gativ~
J32h n-gatlv- nngatlvn negative Embryonlc HlF nogatlv n~gatlve n~gative Fibrobla-~t ~1-38 n~gativ8 n~g~tivo negative ~ootaot-~ nhybrldo~, bc~ VQr~ d-t-etod to hav~ bound HmAb abov- th~ b-e~ground l~vol o~ la~llln~ ~t~ eontrol ~nt$body, CT-e~ uk8~1a, dehronlc my-loeyt$c l~u~-mla, ~ACUt~ pro~yoloeytic l~uk-~la, ~opltholtal e-rvlcal e~re~no~ , geolonle ad~noe~reimoma, htranaltional bla~d-r e-ll carelne~

- 31a -Tablo 2: R-activity o~ H~Ab~ ~T34/AS and BT54/B8 T1~mor Type Cell Lin8 ;3T34~p5a ~ ~7/2A8 Glioma SK-MG-1 positiveb poaitive po~itive SKI-1 negative negative po~itive U-373 positive poYitive positive Melanoma M-4 positive po~itive negative Hematological CCRF-CEMC negative negative negative ~-562d negative negative negative HL-60e negative negative negative u-937f negative negative N.D.
Rajig negative negative N.D.
Other Tumors HeLah negative negative po~itive ME180h negative negative negative C-33Ah negative negative negative SW1417i negative negative negative sW1463i negative negative negative J82i negative negative negative Embryonic IMR-90 poqitive po9itive N.D.
Fibroblast WI-38 negative negative negative root~otoa: ahybridoma , bsee Table 4.1, c~ cell leukemia, dchronic myelocytlc leukemia, eacute promyelocytic leukemia, fhi3tiocytic lymphoma, monocyte-like, gB cell lymphoma, hepithelial cervical carcinoma, icolonic adenocarcimoma, itranqitional bladder cell carcinoma, N.D.: not done.

- 31b -~ i3 being derived from four different patients with different tumors.

Four tumor explants (glioblastoma multiforme, recurrent meningioma, medulloblastoma, and astrocytoma grade II) were adapted to tissue culture and studied by FCM after brief passages in vitro. The greatest apparent proportion of labelling occured with a recurrent meningioma (46%), whereas the lowest proportion of labelling (7%) was found in a low grade astrocytoma. These results indicate that BT27/2A3 is capable of reacting with primary glial and non-glial neurological tumors in the early stages of adaptation to tissue culture.

In order to get some indication of relative antibody affinity, a modification of the method of De Bernado and Davies (1987) was used. The results of ELISA
testing of the five HmAbs with tumor extract from BT-37 (glioblastoma multiforms) after either a long ~18 hr) or a short (4 hr) incubation at 4C, are expressed in terms of O.D. measurements. Comparison of the O.D. readings for each HmAb at the two timepoints indicates that there is a statistically significant (p<0.05) increase in O.D.
for BT32/A6 and BT54/B8 with the longer incubation. None of the other 3 HmAbs (BT27/lA2, BT27/2A3, and BT34/A5) showed a significant increase in reactivity after a longer incubation at 4C. These observations would suggest that BT32/A6 and BT54/B8 behave as a low affinity antibodies at low temperatures.
After the long incubation period (18 hrs at 4C), all five HmAbs confirmed positi~e (p<0.05) by F' ~

ELISA when compared to the control. After the short incubation time (4 hrs at 4C), however, only BT27/lA2, BT27/2A3, and BT34/A5 were significantly positive;
BT32/A6 and BT54/B8 did not react significantly, possibly because of their low affinity nature.

Dot blots of total lipid extracts from cultured neuroectodermal cell lines were examined.
LN-340 and M-4 are glioma and melanoma cell lines which have been previously shown to be unreactive with any of the five ~mAbs (see Example 6) . Compared to control HnAb BT27/2D2, there was some indication of potential reactivity for all five HmAbs when tested against glycolipids from the glioma cell line U~373. As for the glioma line SK-MG-1, all the HmAbs except for BT54/B8 seemed to exhibit some degree of reactivity to the lipid extract, although the quality of these dot blots was not entirely satisfactory.
Immunochromatography on total lipid extract of SK-MG-l was performed using BT34/A5 and the parental line, TM-H2, as control. A single specific band with Rf 0.60 was observed. As well, a non-specific band with Rf = 0.80 was evident in both the BT34/A5 and T~-H2 lanes. In separate experiments, similar results were obtained for BT27/lA2 and BT27/2A3 compared to BT27/2D2, i.e. a specific band at Rf = 0.60. Immunochromatography with HmAb BT54/B8 failed to reveal the presence of any specific banding pattern, but this result is consistent with the dot blot experiment.
These results suggest that the HmAbs recognize a determinant of a glycolipid, or ganglioside, which was not detected in association with glycoproteins.

'~'`i3 "'3~, Cultured normal human astrocytes were assessed for labelling with the HmAbs. All reagents were checked for immune reactivity against an established positive human glioma cell line (U-373). None of the HmAhs BT27/lA2, BT27/2A3, and BT32/A6 appeared to label cultured astrocytes from two different individuals.
Antibodies BT34/A5 and BT54/B8 also failed to label normal human astrocytes on at least one occasion.

The chromosomal content for each of the hybridomas was determined by karyotyping as previously described. The parental cell line, TM-H2-SP2, was also studied, and found to have a mean and modal number of 46 chromosomes, as previously reported (Sullivan, et al 1982). All three hybridomas tested were pseudodiploid, Z0 with BT27/lA2 and BT32/A6 having 2n to 3n number of chromosomes, and BT27/2A3 slightly less than 2n. The latter observation was confirmed by DNA flow cytometry.

Monoclonality was determined by Southern blot analysis, and revealed that hybridomas BT27/lA2, BT27/2A3, and BT32/A6.5, each possessed two rearranged bands bearing homology to the ~H gene region. Hybridoma BT27/2D2 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 for a TM-H2-SP2 type of rearrangement of the JH region in any of the hybridomas.
As a control, normal PBL DNA, which is composed of 60-70% T cell derived genetic material, was found to - 3g -yield a blot profile identical to placental (germline) DNA. There is also a common low molecular weight band containing an unrelated homologous sequence present in each of the lanes.
For conducting the Southern blot analysis, genomic DNA was isolated from hybridoma cells and digested with BAM Hl and HIND III restriction enzymes, then electrophoresed on a 0.8~ (w/v) agarose gel, hybridized and labelled with a JH region probe (ONCOR
Inc., Gaithersburg, Maryland), as described earlier.
The presence of two JH rearrangements in each of the three hybridomas is consistent with monoclonality.
The fusion partner chosen according to the preferred embodiment of this invention thus yields stable B cell hybridomas secreting IgM antibodies which react with autologous tumor extracts and glioma cell lines. Growth of the hybridomas, e.g. by supplementation with additional quantities of the amino acids L-glutamine and L-asparagine. This fusion partner TM-H2-SP2 appears to be superior to any previously reported fusion partner for normal PBL.

Biological S~nificance Of A Glioma Cell_Subpopulation Identified By Human Monoclonal Antibodies Reco~nition of a Subpopulation of Glioma Cells Initial studies of the human glioma line, SK-MG-l, using 2-dimensional FCM revealed two discrete subpopulations of cells. These consisted of a smaller subpopulation, both proportionally and also in terms of small angle light scatter, and a larger subpopulation, which contained the majority of the cells for this particular line. These will be referred to as S-cells and L-cells, respectively.

After reacting SK-MG-1 with either of the three HmAbs described in the previous chapters, there appeared to be selective labelling of only the S-cell subpopulation. This selective labelling was consistently observed over many experiments, and was also observed in many other tumor cell lines (e.g. HeLa) which also contained two discrete subpopulations.
Evidence suggests that the S-cell subpopulation consists of immature cells representative of the main stemline of SK-MG-l, and capable of stem cell-like behaviour. Cell size is often regarded as a characteristic feature of differentiation or anaplasia.
In many normal tissues, terminally differentiated 'end' cells tend to be larger than their smaller undifferentiated precursors.

EXAMPLE_14 Generation of Subpopulation-Derived Sublines from Earlier studies of monoclonally-derived sublines of SK-MG-1 involved the generation of randomly selected, unsorted clones. In these experiments, monoclonally derived sublines were established from sorted S- and L-cells of SK-MG-1. The flow cytcmetric properties of the resulting monoclonal S- and L-cell derived sublines (SlOl-10, and L101-10, respectively) were then compared to those of the parental line, SK-MG-1. Three important conclusions were drawn:
1) The colony forming efficiency in vitro for sor~ed S-and L-cells was equivalent (approximately 1%).
2) All S- and L-cell derived sublines were found to contain S- and L-cells.
3) The small angle light scatter (optical size) of the S-cells in the L-cell derived sublines was significantly lower (P<0.05) than the small angle light scatter of the 2 ~J i ~

S-cells in the parental line, SK-MG-l. In contrast, the FCM profile of the S-cell derived sublines appeared identical to that of SK-MG-1.
Table 3 compares the modal small angle light scatter channel number for both S- and L-cells in SK-MG-1 with 8 S-cell derived and 5 L-cell derived sublines. The L-cell peaks for SK-MG-1, S106, and L102 appear to map together, as do ths S-cell peaks for SK-MG-1 and S106. The S-cell peak for L102, however, is 10 seven channels lower (44.5~ lower after conversion to a linear scale) than the corresponding S-cell peak of SK-MG-1.
Typical FCM sortings of S- and L-cells were centrifuged onto glass slides and stained with Giemsa.
15 Qccasional mitotic figures were observed in FCM sortings of L-cells, but very rarely among the S-cells.
These results suggest that there exists a high degree of phenotypic similarity between the S-cell derived sublines and the parental line, SK-MG-1, 20 compared with the L-cell derived sublines. When total DNA content and cell cycle state of SK-MG-1 and S- and L-cell derived sublines was examined, S-cell derived sublines were more similar to SK-MG-1 than were L-cell derived sublines.

Cell Cycle Analysis of SK-MG-1 and Its Sublines Analysis of sorted S- and L-cells of SK-MG-1 30 demonstrated clearly that the S-cell subpopulation contained cells which were almost exclusively in the G1 or Go ('resting') portion of the cycle. The L-cell subpopulation, however, contained cells primarily in the G1 and S phases, suggesting that they were actively 35 dividing.

Table 3 Small angle light scatter dataa for SK-MG-1 and its small and large cell d~rived clones small cells larae cells SK-MG-1 (parental) 76.75 i .49 95.50 + 1.85 (n= 4) (n- 4) small cell clones 77.75 ~ .63 95.94 + 1.00 (n= 8) (n= 8) large cell clones 72.80 + 1.20b 98.00 + .95 (n= 5) (n= ~) aresults expressed as mean + standard error for modal channel number, 128-channel 1092 scale; n refers to the number of FACS distributions pooled to obtain these values bp<o.o5, compared with SK-MC,-1 using a two-tailed Student's t-test - 37a -No statistically significant differences were found between the DNA histograms of sorted S-cell derived clones and the parental line, SK-MG-1 (Table 4).
The S-cell derived clones therefore, again appear to resemble SK-MG-1 phenotypically. By comparison, the average DNA content (as estimated from the Go/G1 peak) of cells in the L-cell derived clones was statistically greater (p<0.05) than the parental line.
These results suggested that the L-cells may represent divergent hyperploid sublines of SK-MG-1. To verify this hypothesis, the karyotypes of SK-MG-l and its S- and L-cell derived clones were studied.

Karyotypic AnalYsis Of SK-MG-l and Its Sublines Metaphase chromosome spreads were prepared for SK-MG-l, and three different S- and L-cell derived sublines (Table 5). Statistical analysis revealed that:
1) There was no significant mean chromosomal difference between SX-MG-1 and its S-cell derived sublines;
however, 2) The variance among S-cell derived sublines was significantly lower than that of the parental line.
3) L-cell derived sublines had, on average, a higher ploidy number than SK-MG-l (compatible with the results of DNA labelling with propidium iodide).
4) One particular L-cell derived subline (Ll08), had a variance about the mean that was significantly greater (P<0.05) than for SK-MG-l.
These results suggest that the S-cell subpopulation of SK-MG-l possess properties resembling those of a hypothetical tumor stem cell. For example, the S-cells are phenotypically smaller and more anaplastic than the L-cells. When S-cells are individually sorted, they establish sublines which are Table g DNA contenta of SK MG-1 and its small and large cell derived clones smal! cell$ l~rge ce!/s SK-MG-1 (parental) 5Q.33 i .67 57.33 + .67 (n= 3) (n= 3) small cell clones 49.20 + .73 55.50 + .72 (n= 5) (n= 6) large cell clones 57.10 + .78b 63.33 + .g4b (n= 10) (n= 9) aethanol-fixed ce11s were stained with propidium iodide; results expressed as mean +
standard error for modal channel number, 128-channel linear scale; n refers to the number of FACS distributions pooled to obtain these values bp<o.oo5, compared with SK-MG-1 using a two-tailed Student's t-test - 38a -2 ., ~ v ~

Table 5 Metaphase chromosomes of SK MG-1 and its small and large cell derived clones mean+SD~ mode ranaQb ~c dfd p~

SK-MG-1 (parental) 57.4 + 4.2 56 46-68 S103 54.5 + .93 51 50-64 -0.66 19 21.009 S104 55.5 + 1.0 55 49-66 -0.44 19 18.109 S110 53.7 + 1.3 50.5 46-68 -0.82 l 9 10.959 L103 73.2 + 5.3 60 46-146 2.32f 38 1.58 L107 66.2 ~ 3.2 60.5 51-100 1.65 38 1.74 L108 96.2 + 10.7 74.5 46-~2D0 3.37~ 19 6.329 aresults expressed as mean + standard error for 20 metaphase spreads brange represents lhe lowest to highest chromosome number counted Ccalculated t-statistic ddegrees of freedom ecalculated F-statistic fp~o.oo5, compared with SK-MG-1, using a two-tailed S~udent's t-lest 9p~0.01, compared with SK-MG-1, using a two-tailed Ftest with df = (19,19) - 38b -similar in terms of cell size, DNA content, and chromosoma number to the parental line.
The L-cell subpopulation of SK-MG-1 is also capable of establishing colonies in vitro after sorting, a property which would not be expected i* L-cells were differentiated 'end cells'. The fact that these colonies also contain S- and L-cell subpopulations may be that the L-cell derived sublines appear to be from a different stemline than that which predominates in SK-MG-l. The L-cell subpopulation may, in fact, include a mixture of differentiated 'end' cells derived from the S-cell subpopulation, and divergent minority sublines of SK-MG-1 with stem cell-like properties. Evidence in support of this is given by the finding of a significantly lower chromosomal variance about the mean among the S-cell derived sublines than among the SK-MG-1 parental line.

Claims (12)

1. Human monoclonal antibodies recognizing glioma or astrocytoma tumor cells and produced by hybridoma cell lines resulting from the fusion of a human myeloma-like cell line which is HGRPT deficient and is an immunoglobulin non-secretor, with a human peripheral blood lymphocyte selected from the group of such lymphocytes consisting of BT27; BT32; BT34 and BT54.

2. Human monoclonal antibodies according to claim 1 of the IgM type.

3. Human monoclonal antibodies according to claim 2 in which the human myeloma-like cell line contributing to the hybridoma is TM-H2-SP2.

4. Human monoclonal antibodies according to claim 1 in which the hybridoma cell line is selected from the group consisting of BT27/lA2; BT27/2A3; BT32/A6;
BT34/A5; and BT54/B8.

5. Human monoclonal antibodies of the IgM type, recognizing glioma or astrocytoma tumor cells, and produced by the hybridoma cell line BT27/lA2.

6. Human monoclonal antibodies of the IgM type, recognizing glioma or astrocytoma tumor cells and produced by the hybridoma cell line BT27/2A3.

7. Human monoclonal antibodies of the IgM type, recognizing glioma or astrocytoma tumor cells, and produced by the hybridoma cell line BT32/A6.

8. A process of preparing human monoclonal antibodies recognizing glioma or astrocytoma tumor cells which comprises cultivating a hybridoma cell line selected from the group consisting of BT27/lA2;
BT27/2A3; BT32/A6; BT34/A5; and BT54/B8.

9. A method of differentiating between normal and malignant human brain cells which comprises contacting a human brain cell specimen with a human monoclonal antibody produced by a hybridoma cell line selected from the group consisting of BT27/lA2; BT27/2A3; BT32/A6;
BT34/A5; and BT54/B8.

10. A process of treating neuroectodermal tumors in a human patient which comprises administering to said patient at least one of the human monoclonal antibodies of claim 1 optionally linked to a therapeutic agent.

11. A method of identifying subpopulations of the human glioma cell line SK-MG-1 which comprises contacting said cell line with a human monoclonal antibody produced by a hybridoma cell line selected from the group consisting of BT27/lA2; BT27/2A3; BT32/A6;
BT34/A5; and BT54/B8.

12. A method as in claim 11, wherein the subpopulation consists of immature cells representative of the main stemline of SK-MG-1 and capable of stem cell-like behaviour.