Proc. Natl. Acad. Sci. USA Vol. 76, No. 12, pp. 6552-6556, December 1979

A human thymus-leukemia antigen defined by hybridoma monoclonal antibodies (acute lymphocytic leukemia/cell surface/tumor antigens) RONALD LEVY, JEANETTE DILLEY, ROBERT I. Fox, AND ROGER WARNKE The Howard Hughes Medical Institute Laboratories, and the Departments of Medicine and Pathology, Stanford University Medical Center, Stanford, California 94305 Communicated by Henry Kaplan, July 9, 1979

ABSTRACT A series of mouse hybridomas producing mo- MATERIALS AND METHODS noclonal antibodies against human acute lymphocytic leukemia (ALL) cells was generated and screened for tumor specificity. Human Cells. The leukemia cells used for immunization and Among 1200 primary cultures, 60 produced an antibody that for screening of antibodies were derived from the peripheral could distinguish between the immunizing leukemia cells and of a child (Dom) with T-cell ALL. The patient had a an isologous B lymphoblastoid cell line. Of these, two produced mediastinal mass and a blood cell count of 460,000 an antibody that detects an antigen expressed preferentially on ALL cells and on a subpopulation of normal cells found in the per mm3. Greater than 95% of these blast cells formed heat- cortex of the thymus. Other normal human lymphoid cells from stable sheep erythrocyte rosettes (7). Leukemia cells from patient lymph nodes, spleen, bone marrow, and peripheral blood express Dom and a series of other patients were purified from peri- only low levels of this antigen. High levels of this "thymus- pheral blood or bone marrow or Ficoll-Hypaque sedimenta- leukemia" antigen were found on T-ALL cells, T-ALL-derived tion (8) and stored in 10% dimethyl sulfoxide under liquid N2. cell lines, and some "null" ALL cells. By contrast, B-cell leuke- A B-cell LCL from patient Dom (LCL-Dom) was established mias, B lymphoblastoid cell lines, and normal and malignant myeloid cells contain either low or undetectable amounts of this by Henry Kaplan. This cell line is composed of a polyclonal antigen. The thymus-leukemia antigen has been isolated from mixture of K and X immunoglobulin-bearing cells (9) and it the membranes of leukemia cells by detergent solubilization contains the Epstein-Barr nuclear antigen (10) (Henry Kaplan, and subsequent immunoprecipitation with the monoclonal personal communication). The LCL-Dom cells were grown and antibody. Preliminary biochemical characterization shows the maintained in Dulbecco's modified Eagle's medium (high antigen to be associated with a polypeptide of Mr -28,000. glucose) (GIBCO), containing penicillin (100 units/ml), gluta- A major advance in cell surface and tumor immunology has mine (2 mM), and 15% fetal calf serum. Other cell lines used been the introduction of techniques for the production of ho- in these studies include the lines MOLT-4, 8402, and mogeneous monoclonal antibodies. It is now possible to sort HSB2 and the lines 8866, 8392, and SB. Their origins and through a complex mixture of antigens, such as cell surface characteristics have been described (11-13). Normal human molecules, with probes that see one antigen at a time. These thymus glands, lymph nodes, and spleens were obtained from techniques are especially applicable to the study of human cell fresh surgical specimens. surface antigens, particularly in the search for human tumor- Immunization Schemes. BALB/c mice were immunized specific antigens. Monoclonal antibodies can be obtained by and 1 week later were given an intraperitoneal booster with 107 the Klinman technique, which involves in vivo cloning of B cells glutaraldehyde-fixed (4) ALL-Dom cells. Three months later, by adoptive cell transfer and subsequent in vitro spleen frag- they were given an intravenous booster with another 107 ment culture (1). By using this method, we demonstrated that ALL-Dom cells. Three days later, 2 X 107 spleen cells from the mouse has within its antibody repertoire the capacity to these animals were transferred intravenously to syngeneic ir- distinguish among closely related human cell surface antigens, radiated (600 rads) mice along with another 107 ALL-Dom including HLA polymorphisms (2), B- and T-cell determinants cells. The recipient mice had been previously hyperimmunized (3), and tumor-associated antigens (4, 5). However, this tech- either with normal peripheral blood (PBL) (group nique was ultimately limited by the amount of antibody ob- I), LCL-Dom (group II), or nothing (group III, control). Hy- tainable from each spleen fragment culture. With the intro- bridomas were prepared from the spleen cells of each recipient duction of the hybridoma technique of Kohler and Milstein (6), group 1 week after transfer. it became possible to obtain monoclonal antibodies in large Cell Fusion and Culture. Spleen cells were fused with amounts. NSI/1-AG 4 cells (14) at a cell ratio of 2:1, respectively, by using In the present study, we generated a panel of individual 38% polyethylene glycol (Baker 1540) (5). After fusion, the cells hybridomas from mice immunized to human acute lympho- were resuspended in Kennet's modification of Dulbecco's cytic leukemia (ALL) cells. These antibodies were screened for medium containing hypoxanthine, aminopterin, and thymidine their ability to discriminate between the leukemia cells and an (15) and dispensed into 96-well tissue culture trays (Linbro) at isologous B lymphobastoid cell line (LCL). Among the dis- 1-2 X 105 cells per well along with 5 X 105 normal spleen cells. criminator antibodies, two define an antigen that is preferen- The medium was changed twice during the first 2 weeks to tially expressed on ALL cells and on a subpopulation of normal remove antibody released by unfused spleen cells and then was . collected for testing during the third week. Culture fluids were sampled, with attention being paid to avoid contamination of The publication costs of this article were defrayed in part by page each fluid by its neighbors-i.e., by washing or changing of charge payment. This article must therefore be hereby marked "ad- vertisement" in accordance with 18 U. S. C. §1734 solely to indicate Abbreviations: ALL, acute lymphocytic leukemia; LCL, lympho- this fact. blastoid cell line; PBL, peripheral blood lymphocytes. 6552 Downloaded by guest on September 25, 2021 Immunology: Levy et al. Proc. Natl. Acad. Sci. USA 76 (1979) 6553

pipettes between collections. These fluids were examined for the presence of antibody against the immunizing ALL-Doxn cells as well as the LCL-Dom cells by a cell-binding radcoim- munoassay (5), using purified 125I-labeled goat anti-mouse Fab fragment as a detecting reagent. Selected cultures were sub- It cloned by'plating limiting dilutions of the hybrid cells into the wells of 96-well plates along with normal mouse spleen cells. Immunofluorescence. Hybridoma antibodies were exam- ined by indirect immunofluorescence on viable cells in sus- pension, fixed cell smears, or frozen tissue sections prepared by described methods (9). A rhodamine conjugate of the same purified goat anti-mouse Fab used in the radioimmunoassay was employed. In some cases, an F(ab')2 fragment of this pu- -11 11 rified goat antibody was employed. E S S"- WNgS1E1 11S Labeling of Cell Membranes and Immunoprecipitation. Cell membranes were radio-iodinated by using lactoperoxidase and solubilized with Nonidet P-40 (Shell Chemical) as described (16). The solubilized 125I-labeled membrane preparation, 107 cell equivalents, was first subjected to a nonspecific immu- 2000 7 noprecipitation by the addition of human IgG (25 Mg) and goat

anti-human Fab at equivalence. To the cleared supernatant, 4000 - 100-200 ,l of culture fluid containing either a specific mono- clonal antibody or a myeloma protein of the same class was added. This was followed by the addition of normal mouse serum (2.5 pi) and goat anti-mouse Fab (preabsorbed on human 6000 IgG-Sepharose) at equivalence. The final precipitates were solubilized in 3% sodium dodecyl sulfate/2% mercaptoetha- _ LCL nol/0.05 M Tris-HCI, pH 6.8/5% glycerol and boiled. FIG. 1. Radioimmunoassay of hybridoma culture fluids. Fluids Gel Electrophoresis. Sodium dodecyl sulfate/polyacryl- from individual culture wells were assayed for binding activity against amide gel electrophoresis (17) was performed in slab gels by two different target cells from patient Dom: ALL (Upper) and LCL using either a 5-20% gradient of acrylamide or a uniform 10% (Lower). The data from one representative 96-well plate are dis- acrylamide concentration. Gels wer'e dried and autoradio- played. graphed by using Kodak XR2 film with a Cronex Lightning Plus intensifying screen (Dupont). groups the recipients had been hyperimmunized against normal Fluorescence-Activated Cell Sorter Analysis. Lymphocytes human lymphoid antigens-either PBL (group I) or LCL-Dom were incubated with monoclonal antibody (50 Ml per 5 X 106 (group II). This maneuver was an attempt to suppress the ex- cells) at 40C in the presence of 1 mM azide. The cells were pansion of B cell clones reactive to normal human antigens washed in fetal calf serum and gently resuspended in 50 Ml of while stimulating the expansion of clones reactive to antigens fluorescent goat anti-mouse Ig antibody. The stained cells were limited to the leukemia cell. A third group of recipients was not washed again in fetal calf serum, resuspended in Dulbecco's preimmunized. Similar numbers of total antibody-producing phosphate-buffered saline, and fixed in 1% formaldehyde. hybridomas were obtained in each group. However, a moderate Analysis of staining intensity was performed by using the flu- advantage was noted for groups I and II in terms of the per- orescence-activated cell sorter as described by Loken and centage of hybrids producing antibody that discriminated be- Herzenberg (18). tween the leukemia cell and the isologous B cell line-13.8 and 9.7%, respectively, vs. 5.2% in the control group. RESULTS Definition of a Thymus-Leukemia Antigen. On further Initial Screening. A total of 1877 culture wells, of which testing of the 60 discriminators, it became apparent that, al- approximately 60% contained growing hybrids at the time of though they failed to react with LCL-Dom, most of these an- fluid collection, was screened for antibody activity by radio- tibodies were strongly reactive against human normal PBL and immunoassay. Two different target cells were used for this presumably recognize normal T cell antigens. However, two initial screen ALL-Dom, the immunizing leukemia cells, and hybridomas designated 12E7 and 21D2 gave a reactivity pat- LCL-Dom, an isologous B-cell LCL. A representative portion tern that was distinctly different, distinguishing between the of the data from this screen is shown in Fig. 1. Antibody reactive leukemia cells on the one hand and the LCL-Dom and PBL on with the immunizing ALL cells was found in 619 (33%) of the the other (Fig. 2). These binding data, and absorption analyses culture wells. Of these, 148 appeared to contain an antibody (not shown), indicate that the antigen detected by antibodies that distinguished between the immunizing leukemia cell and 12E7 and 21D2 is present on PBL, but at a level less than 5% its isologous B cell line (shown as arrows in Fig. 1). With re- of that on ALL cells. Of note is that hybridomas 12E7 and 21D2 peated testing on these and subsequently collected fluids, 60 were both derived from group II, the animals that were cultures were confirmed to be continuously producing antibody hyperimmunized to LCL-Dom prior to receipt of ALL-Dom with preferential reactivity for the leukemia cell. This fre- immune cells. Hybridoma 12E7 that produces an IgG-1 anti- quency of 60/619 may be an underestimate of the true fre- body was selected for further study. quency of discriminators, because many of the cultures con- Indirect immunofluorescence was performed by using tained more than one clone. antibody 12E7 on viable cell suspensions and fixed cell smears. Three different spleen cell donors were used for cell fusion All of the ALL-Dom cells were brightly stained on the mem- in these experiments. All were radiated recipients of the same brane and in the cytoplasm, whereas PBL showed only faint adoptively transferred immune spleen cells. However, in two membrane staining. Similar stained preparations were analyzed Downloaded by guest on September 25, 2021 .6554 Immunology: Levy et al. Proc. Natl. Acad. Sci. USA 76 (1979)

E

1000 FIG. 4. Indirect immunofluorescence ofantibody 12E7 on a bone marrow smear from a patient with T-ALL. An F(ab')2 fragment of a purified goat anti-mouse Fab antibody was used as a rhodamine conjugate for the second-stage reagent. brightly stained and were easily detectable within a background 1:4 1:8 1:16 1:32 1:64 1:128 1:256 of unstained normal or weakly stained normal bone marrow Fluid dilution cells. Various normal lymphoid tissues were examined by indirect FIG. 2. Binding of antibody 12E7 to T-ALL cells (@), LCL cells (0), and PBL (A). Background of 200 cpm was subtracted from the immunofluorescence on frozen sections with antibodies 12E7 data. and 21D2. These antibodies gave weak staining of cells in spleen and lymph node. However, both these antibodies gave intense by using the fluorescence-activated cell sorter. Again, 100% of staining of cells in the cortex of the thymus gland (Fig. 5). By the ALL cells were very brightly stained and were easily dis- contrast, lymphoid cells in the medulla of the thymus failed to tinguishable from the PBL of normal donors (Fig. 3 Upper). stain with antibody 21D2 or 12E7. The binding of antibody The myeloma protein produced by P3/X63 cells was used as 12E7 to thymocytes in suspension is shown in Fig. 6 and com- a negative control (Fig. 3 Lower). Fluorescent staining by pared with its binding to several types of human leukemia cells. antibody 12E7 of a fixed bone marrow smear from a patient These data show that the average density of the antigen on with T-cell leukemia is shown in Fig. 4. The leukemia cells were thymocytes is comparable to that on T-ALL cells and greater than that on the T-ALL-derived cell line, 8402. It also shows 100 that the cells from one case of chronic lymphocytic leukemia ALL and those from one case of "null cell" ALL contain insignificant amounts of the antigen. 50 / PBL The reactivity of antibody 12E7, either by radioimmunoassay or by immunofluorescence, on a series of human normal and

n leukemia cells is summarized in Table 1. Strong reactivity of ...... CD IC*0lOt

St-L PBL ALL

L I I -J 0 10 20 30 Fluorescence intensity FIG. 3. Fluorescence intensity profiles of ALL cells or normal PBL with monoclonal antibodies. Cells were obtained from the blood by Ficoll-Hypaque sedimentation, mixed with monoclonal antibodies, FIG. 5. Indirect immunofluorescence on frozen section of normal counter-stained with fluorescent rabbit anti-mouse Ig antibody, and human thymus. An F(ab')2 fluorochrome was used as in Fig. 4. Two analyzed on the fluorescence-activated cell sorter. (Upper) 12E7 views are shown. Staining is limited to cells in the cortex (upper area antibody. (Lower) Control antibody. of pictures), sparing cells in the medulla (lower area of pictures). Downloaded by guest on September 25, 2021 Immunology: Levy et al. Proc. Natl. Acad. Sci. USA 76 (1979) 6555

E a

H -- s.

28,000 Mr o

L

1:16 1:32 1:64 Fluid dilution FIG. 6. Binding of antibody 12E7 to thymocytes (0), T-ALL cells (0), a T-ALL-derived cell line, 8402 (o), a "null cell" ALL (A), and a chronic lymphocytic leukemia (v). IgG 12E7 P3 12E7 P3 12E7 P3

this antibody appears to be limited to cortical thymocytes, acute PBL 8402 Dom lymphoid leukemias of the T-cell type, and most cases of acute lymphoid leukemias of "null cell type." FIG. 7. Sodium dodecyl sulfate/polyacrylamide gel electropho- Preliminary Characterization of the Thymus-Leukemia resis of membrane molecules immunoprecipitated with antibody Antigen. Cell membranes were labeled with 12'I by using lac- 12E7. 125I-Labeled cell membranes, 5 X 106 trichloroacetic acid- precipitable cpm, were immunoprecipitated either with the mono- toperoxidase, were solubilized with detergent, and were im- clonal antibody 12E7 or with the myeloma protein secreted by P3/X63 munoprecipitated with antibodies 21D2 and 12E7. Fig. 7 shows (P3), were solubilized in sodium dodecyl sulfate/2-mercaptoethanol, the results of gel electrophoresis of material immunoprecipi- and were electrophoresed on a 10% slab gel. L, light chain of Ig; H, tated by 12E7 from PBL, 8402 (a T-ALL derived cell line), and heavy chain of IgG. Dom (the original T-ALL cells used for immunization). In each case, a comparison is made with a control immunoprecipitate A specific band of -Mr 28,000 is apparent in the material in which P3/X63 fluid was used instead of 12E7. Ig is included immunoprecipitated from the leukemia cells. Identical results on the gel to provide molecular weight markers (L and H). were obtained when antibody 21D2 was used in place of 12E7 (not shown). No specific bands can be seen in the material de- Table 1. Reactivity of antibody 12E7 on normal and rived from PBL, nor could any be seen in material derived from leukemia cells chronic lymphocytic leukemia cells, LCL-Dom, or other B Levels of lymphoblastoid cell lines (not shown on this gel). Lymphoid cells TL antigen DISCUSSION ALL-Dom (the immunizing cell) High LCL-Dom (isologous B lymphoblastoid cell) Low In the present study, we sought to analyze possible antigenic Other B cell lines (SB, 8392, 8866) Low differences between human acute leukemia cells and normal PBL Low human cells. We adopted a screening strategy in which a Lymph node cells Low matched pair of genetically identical cells-one leukemic and Spleen cells Low one lymphoblastoid-was used in a radioimmunoassay to search Bone marrow cells Low for discriminator antibodies. This approach is similar to the Thymocytes strategy we previously employed to screen monoclonal anti- Cortex High bodies for recognition of human HLA polymorphic antigens Medulla Low (2). It requires that the antibody-producing cells be cloned prior T-All cell lines (MOLT-4, 8402, HSB2) High to testing, because the presence of nondiscriminating antibodies T cell acute leukemias and lymphoblastic lymphomas High obscure the discriminators. Furthermore, to be most effective, B cell leukemias (chronic lymphocytic leukemias) Low the pair of target cells should be as closely related as possible Myeloid leukemias (acute and chronic) None but different in the desired characteristic. Our initial screen "Null cell" leukemias (three out of four cases) High eliminated most of the antibodies from consideration and al- TL, thymus-leukemia. lowed us to focus on a manageable number of potentially in- Downloaded by guest on September 25, 2021 6556 Immunology: Levy et al. Proc. Natl. Acad. Sci. USA 76 (1979) teresting candidates. After a second screen with normal pe- Marilyn Pederson for tissue section immunofluorescence and pho- ripheral blood lymphocytes, two antibodies, 12E7 and 21D2, tography and Dr. Irving Weissman for helpful discussions. This work emerged as especially interesting. was supported by grants from the National Institutes of Health (CA The immunization method used in the present study was 21223-02 and AI-09072). R.L. is an investigator of the Howard Hughes based on the work of Moller, who showed that adoptive transfer Medical Institute. R.I.F. is a Fellow of the Arthritis Foundation. of specific B cells with antigen into irradiated recipients could greatly enhance B-cell expansion, but that the expansion could 1. Klinman, N. R. (1969) Immunochemistry 6,757-759. be specifically suppressed by prior immunization of the re- 2. Lampson, L. A., Levy, R., Grumet, F. C., Ness, D. & Pious, D. cipient (19). We attempted to stimulate the ALL-reactive B cells (1978) Nature (London) 271, 461-462. 3. Lampson, L. A., Royston, I. & Levy, R. (1977) J. Supramol. while suppressing the B cells reactive to normal human anti- Struct. 6, 441-448. gens. The maneuver resulted in a slight increase in the fre- 4. Levy, R. & Dilley, J. (1977) J. Immunol. 119,394-400. quency of discriminator hybridomas. 5. Levy, R., Dilley, J. & Lampson, L. A. (1978) in Current Topics Antibodies 12E7 and 21D2 detect an antigen present on ALL in Microbiology and Immunology, eds. Melchers, F., Potter, M. cells and on a population of normal cells found in the cortex of & Warner, N. L. (Springer, Berlin), Vol. 81, pp. 164-169. the thymus. This tissue distribution is reminiscent of the thy- 6. Kohler, G. & Milstein, C. (1975) Nature (London) 256, 495- mus-leukemia antigen of the mouse (20). The mouse thy- 497. mus-leukemia antigen is a 45,000 Mr protein that is associated 7. Sen, L. & Borella, L. (1975) N. Engl. J. Med. 292,828-832. on the cell surface with 32 microglobulin (21-23). By contrast, 8. Boyum, A. (1978) Scand. J. Clin. Lab. Invest. Suppl. 21 97, antigen described here appears 91-106. the human thymus-leukemia 9. Levy, R., Warnke, R., Dorfman, R. F. & Haimovich, J. (1977) to have a Mr of nt28,000 and it is not associated with f32 mi- J. Exp. Med. 145, 1014-1028. croglobulin. 10. Reedman, B. & Klein, G. (1973) Int. J. Cancer 11, 499-520. McMichael et al. (24) have recently described a monoclonal 11. Royston, I., Smith, R. W., Buell, D. N., Huang, E. & Pagano, J. antibody raised against human thymocytes that reacts with the S. (1974) Nature -(London) 251, 745-746. T-ALL-derived cell line MOLT-4. Their antibody immu- 12. Huang, C. C., Hou, Y., Woods, L. K., Moore, G. E. & Minowanda, noprecipitated a two-chain structure of Mr 45,000 and 11,000 J. (1974) J. Natl. Cancer Inst. 53, 655-658. from cell surfaces but the 11,000 Mr polypeptide 13. Pious, D., Hawley, P. & Forrest, G. (1973) Proc. Natl. Acad. Sci. was different from 32 microglobulin; no 28,000 Mr molecule USA 70, 1397-1400. was seen (24, 25). Chechik et al. (26) have described a 43,000 14. Kohler, G., Howe, S. C. & Milstein, C. (1977) Eur. J. Immunol. 6,231-236. Mr protein extractable from human thymocytes, T-cell lines, 15. Kennett, R. H., Denis, K. A., Tung, A. S. & Klinman, N. R. (1978) and T-ALL cells that copurified with a second chain of Mr in Current Topics in Microbiology and Immunology, eds. 23,000. However, they found this molecule to be only intra- Melchers, F., Potter, M. & Warner, N. L. (Springer, Berlin), Vol. cellular. The relationship between these different human 81, pp. 77-91. thymocyte antigens can be clarified only by a side-by-side 16. Fox, R. I. & Weissman, I. L. (1979) J. Immunol. 122, 1697- comparison. 1704. Monoclonal antibodies, such as 12E7 and 21D2, will be useful 17. Laemmli, U. K. (1970) Nature (London) 227,680. in a number of different areas of further investigation. Because 18. Loken, M. R. & Herzenberg, L. A. (1975) Ann. N.Y. Aca. Sci. 254, they react with cells from many, but not all, cases of ALL, they 163-171. will help to categorize different subtypes of leukemia and may 19. Moller, G. (1968) J. Exp. Med. 127,291-306. 20. Leckband, E. & Boyse, E. A. (1971) Science 172, 1258-1260. shed light on the tissue origin of these malignant cells. Cur- 21. Vitetta, E. S., Uhr, J. W. & Boyse, E. A. (1975) J. Immunol. 114, rently, the human ALLs are subdivided into T cell, pre-B cell, 252-254. and null cell types according to their expression of certain 22. Ostberg, L., Rask, L., Wigzell, H. & Peterson, S. A. (1975) Nature markers of normal differentiation (27-30). It is (London) 253, 735-736. already clear that the antigen defined here is not limited to just 23. Stanton, T. H., Bennett, J. C. & Wolcott, M. J. (1975) J. Immunol. the leukemias that carry other T-cell differentiation markers, 115, 1013-1017. because it is found on some "null" cell leukemias as well (Table 24. McMichael, A. J., Pilch, J. R., Galfre, G., Mason, D. Y., Fabre, 1). J. W. & Milstein, C. (1979) Eur. J. Immunol. 9,205-210. By using antibodies 12E7 and 21D2, leukemia cells can be 25. Ziegler, A. & Milstein, C. (1979) Nature (London) 279, 243- tissues. 244. detected in peripheral lymphoid These monoclonal 26. Chechik, B. E., Percy, M. E. & Gelfand, E. W. (1978) J. Nati. antibodies may, therefore, provide a sensitive means of moni- Cancer Inst. 60, 69-75. toring the patients with leukemia during therapy and for de- 27. Kersey, J. H., Sabad, A., Gajl-Peczalska, K., Hallgren, H. M., tecting early disease relapse by surveillance of bone marrow Yunis, E. J. & Nesbit, M. (1973) Science 182, 1355-1356. and blood for antigen-positive cells (31). Moreover, it should 28. Brown, G., Greaves, M. F., Lister, T. A., Rapson, N. & Papmi- be possible to develop a radioimmunoassay for the detection chael, M. (1974) Lancet i, 753-755. of free antigen, which may be even more sensitive than the 29. Brouet, J. C. & Seligmann, M. (1978) Cancer 42,817-827. detection of the leukemia cells per se. 30. Vogler, L. B., Crist, W. M., Bockman, D. E., Pearl, E. R., Lawton, A. R. & Copper, M. D. (1978) N. Engl. J. Med. 298,872-878. We are grateful to Dr. Henry Kaplan for the LCL-Dom cell line, 31. Brown, G., Capellaro, D. & Greaves, M. (1975) J. Natl. Cancer to Dr. Charles Bieber for normal human thymus tissue, and to Ms. Inst. 55, 1281-1289. Downloaded by guest on September 25, 2021