Proc. Nati. Acad. Sci. USA Vol. 80, pp. 2031-2035, April 1983 Immunology

Macrophage activation: Dissociation of cytotoxic activity from Ia-A antigen expression ( /immune interferon/) ELLIOTT J. BLUMENTHAL, WALDEN K. ROBERTS, ADRIANA VASIL, AND DAVID W. TALMAGE Department of Microbiology and Immunology, University of Colorado Health Sciences Center, Denver, Colorado 80262 Contributed by David W. Talmage, December 20, 1982

ABSTRACT Peritoneal were obtained from MATERIALS AND METHODS DBA/2 mice that were untreated or after the injection of bacillus Cainette-Guerin (BCG), thioglycollate broth, proteose-peptone Mice. Our source of peritoneal macrophages was DBA/2 broth, or gamma-irradiated P-815 tumor cells. These macro- mice that were less than 3 months old (The Jackson Labo- phages were "activated" to become cytotoxic for a fibroblast cell ratory). line (L 929) by the addition of (LKs), lipopolysac- Elicitation of Peritoneal Macrophages. Peritoneal exudate charide (LPS), or fibroblast interferon (IFN-.3), and the expres- cells were collected by lavage with 5 ml of sterile Eagle min- sion of I region-associated antigens (Ia-Ad) on the macrophages imal essential medium. Resident cells were removed from un- was examined both before and after activation. Thioglycollate- treated mice. Cells were also obtained (a) 3-4 days after the elicited macrophages became Ia-A' when activated by LKs, but injection of 1 ml of 3% thioglycollate in broth that had been they remained la-A- when activated by LPS or IFN-f3. Resident "aged" for at least 6 months at 40C or (b) 3 days after intra- macrophages and proteose-peptone-elicited macrophages re- peritoneal injection of 1 ml of 1% proteose-peptone broth. mained Ia-A- when activated with LKs. Macrophages from BCG- The bacillus Calmette-Guerin (BCG)-elicited macrophages were infected mice were both Ia-A' and cytotoxic for tumor cells with- induced by injection of 0.1 ml of medium containing 2-3 X out further treatment. In contrast, macrophages from mice in- 107 BCG organisms into the tail vein. Four weeks later, 0.1 jected with gamma-irradiated P-815 mastocytoma cells were Ia- ml of purified protein derivative (2 mg/ml) was injected in- A+ but not cytotoxic, and these macrophages could not be made traperitoneally; 5 days later the exudate cells were removed cytotoxic by incubation with LKs. The cultured -like by lavage. The P-815-elicited macrophages were induced by cell lines P388D1 and WEHI-3 became Ia-A+ after incubation with intraperitoneal injection of 3 X 106 irradiated (5,000 R; 1 R LKs, and this treatment amplified the cytotoxicity of both cell lines. = 2.58 x 10-4 C/kg); P-815 tumor cells 5-6 days later the We conclude that a number of factors are important in deter- exudate cells were removed by lavage. All of the cell prep- mining whether Ia-A expression accompanies macrophage acti- arations were then washed three times with minimal essential vation and that Ia-A is irrelevant as a surface marker for mac- medium before counting and use. rophage activation. Macrophage Cell Lines. The two macrophage-like cell lines WEHI-3 and P388D1 were generously supplied by John Kap- I region-associated antigens (la) have been detected on distinct pler and Philippa Marrack (National Jewish Hospital and Re- subpopulations of macrophages. Most macrophages from the search Center, Denver, CO). These cells were grown in RPMI spleen, thymus, and liver are la', whereas la- macrophages medium supplemented with 0.5 mM 2-mercaptoethanol, pen- predominate in the peritoneal cavity (1-3). It has been shown icillin, streptomycin, and 10% fetal calf serum. that only the Ia+ subset of macrophages is effective in pre- Production of LKs. The LKs were prepared as described senting antigen to T cells (4, 5). In addition, Ia expression ap- (12). Briefly, spleens were removed from CF1 mice that were pears to be a prerequisite for macrophages to act as accessory infected or had been injected intravenously with 2 X 107 BCG cells for lymphokine (LK) production and prolif- organisms 18 days earlier. To the spleen cell suspension (usu- eration (6, 7). ally 5 x 10' cells per ml) was added concanavalin A (Con A) Recently, it has been demonstrated that LK preparations can (Pharmacia, Uppsala, Sweden) to 5 ,Ag/ml, and the cells were induce la- macrophages to become Ia+ in vitro (8-10). Similar incubated for 2 hr at 370C in a chamber containing 95% air/ of LKs are known to contain 5% CO2. The cells were then collected by centrifugation and preparations macrophage-acti- resuspended in minimal essential medium supplemented as vating factor (MAF) which stimulates macrophages to become above but without Con A, and the incubation was continued cytotoxic for tumor cells (11). This raises the questions of whether for 24 hr. The cells were then removed by centrifugation and the Ia' macrophage is the cytotoxic macrophage and whether the culture supernatants were combined and used as a crude La antigens can be used as a surface marker for macrophage ac- source of MAF and immune interferon (IFN-y). tivation. Purification of IFN-y. The spleen cell culture supernatant In this report, we present evidence indicating that the cy- from the LK preparation was partially purified by ammonium totoxic macrophage can be either Ia-A+ or Ia-A-, depending sulfate precipitation and passed over a Con A-Sepharose col- upon the procedures used for elicitation and activation of it. umn as described (12). Also, Ia-A was found not to be a useful marker for macrophage Assay for Macrophage Activation. The cytotoxicity assay activation because Ia-A expression was neither necessary nor we used for LK activation of macrophages has been described sufficient for macrophage cytotoxicity. Abbreviations: IFN-,B, fibroblast interferon (type I); IFN-y, immune The publication costs of this article were defrayed in part by page charge interferon (type II); LPS, lipopolysaccharide; BCG, bacillus Calmette- payment. This article must therefore be hereby marked "advertise- Guerin; MAF, macrophage-activating factor; LK, lymphokine; Con A, ment" in accordance with 18 U. S. C. §1734 solely to indicate this fact. concanavalin A. 2031 Downloaded by guest on September 24, 2021 2032 Immunology: Blumenthal et aL Proc. Natl. Acad. Sci. USA 80 (1983) in detail elsewhere (12, 13). Briefly, mouse L 929 cells were A Zeiss microscope was used to observe the fluorescence labeled with [3H]thymidine, trypsinized, and added to 16-mm of the cells, and photomicrographs were taken on Kodak culture wells of a cluster plate (Costar, Cambridge, MA) at a Ektachrome film (160 ASA tungsten). The exposure time for concentration of 1 x 105 cells per well. After incubation for the fluorescence microscopy under the blue was 5 min. another 24 hr, a suspension containing 4-10 X 105 peritoneal At least 200 cells were counted for each quantitation, and the macrophages per ml in minimal essential medium containing positive cells were characterized by having good peripheral 10% fetal calf serum was added. Following a 2-hr incubation, staining; the controls did not show any staining. the nonadherent cells were removed and the medium was re- placed with 0.5 ml of LKs diluted in minimal essential me- dium/10% fetal calf serum. These cultures were then incu- RESULTS bated for 48 hr at 370C in 95% air/5% C02, the plates were Effect of LKs on la-Ad Expression by Macrophages. Fig. removed, 5 ul of a 1 mg/ml solution of pancreatic DNase 1 demonstrates how different populations of macrophages stain (Worthington) was added to each well, and the plates were with the anti-Ia-Ad antiserum followed by the fluorescein-la- incubated an additional 20 min at 37TC. This terminal incu- beled anti-mouse globulin antiserum. Cells that did not re- bation with DNase facilitated the release of [3H]thymidine from ceive LKs during the incubation period were not stained (Fig. dead cells. The culture supernatants from each well were then 1A); and LK-induced cells stained brightly (Fig. 1B). We chose added to vials containing 5 ml of Biofluor (New England Nu- the 3-day period for staining the cells because we found that clear) and the radioactivity was determined by liquid scintil- la expression reaches its peak on this day and this correlates lation counting. Normally, <5% of the total radioactivity was with the findings of others (8). When normal mouse serum released in the absence of LKs whereas >50% was released was substituted for the anti-Ia-Ad antiserum in the procedure, when LKs were present during the assay. LK-treated cells appeared similar to those in Fig. IA. In all Antibodies. The hybridoma MK-D6 (anti-Ia-Ad) cell line was cases there were a few stained B cells that reacted directly generously provided by John Kappler and Philippa Marrack. with the goat anti-mouse IgG-fluorescein and thus were pos- The antibody produced by this cell line has been classified as itive in the normal serum controls. They were particularly belonging to the IgG2a class and has been mapped to Ia-Ad plentiful in the resident cell populations (Fig. 1C), but they by its reactivity with BLO.D2 or D2.GD but not BLO.LG. A were easily distinguished from macrophages by their small size. goat anti-mouse fluorescein-conjugated IgG (Meloy, Spring- There was no staining of either resident (Fig. 1C) or proteose- field, VA) was used as an indirect label to quantitate the Ia- peptone-elicited macrophages (Fig. 1D) when cultured for 3 Ad antigens on the macrophages. days in medium containing LKs. All of these macrophage Quantitation of Ia-Ad Antigens. Peritoneal macrophages populations had the same percentage of Ia-Ad-positive cells elicited by various methods were plated onto Lab-Tek tissue before the addition of LKs to the culture medium, the re- culture chamber slides (Lab-Tek, Naperville, IL) at a con- sponse after 3 days in culture was quite striking. In addition, centration of 2 x 105 cells per chamber. After a 3-hr period peritoneal macrophages elicited with gamma-irradiated P-815 in minimal essential medium/10% fetal calf serum for the cells tumor cells had a high percentage of Ia-Adpositive cells when to become adherent, the nonadherent. cells were removed by stained immediately after removal from the mice (Fig. 1E). washing twice with the same medium and the adherent cells These cells were very adherent, stained positively with a non- were either stained (time 0) or were incubated 1-3 days with specific esterase stain, and ingested latex beads, suggesting various "activating" agents and subsequently examined for Ia- that they were macrophages (data not shown). Ad antigen expression. Effect of LK Concentration on Macrophage Cytotoxicity Prior to the staining of the cells, the medium was aspirated and la-Ad Expression. We next looked at the correlation be- and replaced with 1% paraformaldehyde solution for 15 min cytotoxicity and the gen- at room temperature. The cells were then washed twice with tween LK activation of macrophage minimal essential medium/5% rabbit serum (Colorado Serum, eration of Ia-Ad antigens on the macrophage cell surface. As the Denver, CO) and allowed to equilibrate with minimal essen- concentration of LK was decreased, both the cytotoxic poten- tial medium/5% rabbit serum for 20 min to allow the rabbit tial and the Ia-Ad-inducing capability of the preparation were serum to bind to any available Fc receptors. After the 20-min lost (Fig. 2). We have assayed more than 10 different LK prep- incubation, the minimal essential medium/5% rabbit serum arations in this manner and, although both activities declined, was removed and fresh minimal essential medium/5% rabbit Ia-Ad expression was always induced at LK dilutions that elic- serum containing a 1:80 dilution of the MK-D6 ascites fluid ited no macrophage activation. In addition, 90% of both activ- was added to each well. The slide was kept at 4°C for 40 min. ities was lost after heating at 56°C for 1 hr or after treatment After this period the medium was removed from each well, of the LK preparation with pH 2 buffer at 4°C for 12 hr (results the cells were washed twice with cold minimal essential me- not shown). dium/5% rabbit serum, and then minimal essential medium/ Effects of LK on la-Ad Expression and Cytotoxicity of Var- 5% rabbit serum supplemented with a 1:30 dilution of the ious Macrophage Preparations. The previous sets of data sug- goat anti-mouse fluorescein-conjugated antibody was added gested that our LK preparation contained both Ia-Ad-inducing and the cells were incubated for another 40 min at 4°C. The capability and macrophage activating factor(s) (MAF) and that cells were then washed twice with minimal essential medium/ these effects occurred together. We next wanted to examine 5% rabbit serum to remove any unbound antibodies, the different macrophage populations, or macrophages that had been chamber was removed.from the slide, and a coverslip was placed elicited by different mechanisms, to determine whether they over the cells and sealed. As controls, either normal mouse would respond in a similar manner. The potential of the mac- serum was used in the place of the MK-D6 antibody or the rophages to express Ia-Ad and to become cytotoxic was depen- cells were stained only with the goat anti-mouse fluorescein- dent upon how the macrophages were elicited (Table I). Res- labeled antibody. It was found that B cells stained positively ident, proteose-peptone-elicited, and thioglycollage-elicited in these controls, but these were readily distinguishable from macrophages all were capable of becoming cytotoxic for tumor macrophages by their morphology (small, rounded cells) and cells, but only the thioglycollate-elicited macrophages ex- were not included with the macrophages that were counted. pressed Ia-Ad in response to LKs. In addition, WEHI-3 and Downloaded by guest on September 24, 2021 Immunology: Blumenthal et aL Proc. Nat Acad. Sci. USA 80 (1983) 2033

FIG. 1. Effect of LKs on Ia-Ad expression by macrophages: thio- glycollate-elicited macrophages cultured for 3 days in the absence (a, A) or presence (b, B) of LKs; resident macrophages cultured 3 days with LKs (c, C); proteose-peptone-elicited macrophages cultured 3 days with LKs (d, D); and gamma-irradiated P-815-elicited macrophages not cul- tured (e, E). Macrophages were photographed in white light (e-e) or in ultraviolet (blue) light (A-E) after staining with anti-Ia-Ad anti- serum followed by fluorescein-labeled goat anti-mouse IgG.

P388D1, the two macrophage-like cell lines were inducible for Ad antigen but were totally resistant to LK activation for tu- Ia-Ad expression as well as cytotoxicity. However, at the higher mor cell cytotoxicity. Finally, as reported by others (1416) effector macrophage concentrations, both cell lines were cy- we observed that BCG-elicited macrophages were both Ia-A totoxic for tumor cells without LK activation, although they positive and cytotoxic when assayed immediately after re- remained negative for expression of Ia-Ad. Furthermore, mac- moval from the mouse, even in the absence. of LKs. When rophages elicited by intraperitoneal injection of gamma-irra- these macrophages were incubated in the absence of LKs they diated P-815 mastocytoma cells were -positive initially for Ia- lost their Ia positivity. Downloaded by guest on September 24, 2021 2034 Immunology: Blumenthal et aL Proc. Natl. Acad. Sci. USA 80 (1983)

100 100 Table 2. Effects of IFN-y, IFN-13, and LPS on macrophage Ia - expression and cytotoxicity CZ 80 80 0~~~~~~~~~~~~~ 0 % released .4 ~~~~~~// [3Hlthymidine

0 Activating No macro- With macro- 60 0 factor added % Ia+ phages phages* a- a 40 OsCd None 7 2 5 ", 40 d) IFN-'y (10 units/ml) 100 3 74 10 IFN-MP (1,000 units/ml) 2 3 60 *ax LPS (20 3 4 32 20 ,g/ml) Pe *Tioglycollate-elicited peritoneal macrophages at a concentration of 2 x 105/cm2 were used for all assays. IFN-13 was obtained from Lee Biomolecular Research Labs (San Diego, CA) and was induced from 0 0.03 0.1 0.3 1 3 virally infected (Newcastle disease virus) fibroblast cells. % LK in culture medium wanted to determine whether different macrophage-activating FIG. 2. Effect of LK concentration on peritoneal macrophage cy- agents would show the same variability. To this end, we added totoxicity and Ia-Ad expression. Increasing amounts of a LK prepa- three different macrophage-activating agents-IFN-y, IFN-, ration were added to macrophage cultures; cytotoxicity was measured and LPS-to thioglycollate-elicited macrophages. All three ac- by the percentage of total [3H]thjmidine released and expression was tivating agents were able to generate a cytotoxic response to measured by percentage of Ia-A -positive cells. Each point represents tumor cells, but only IFN-y was able to induce the macro- the mean (SD, 20%) of triplicate determinations. phages to express Ia-Ad (Table 2). The IFN-y used in this ex- periment had been purified by us to a specific activity of 1 Effects of IFN-y, Fibroblast Interferon (IFN-fi), and Li- X 106 interferon units/mg of protein by fractionation of a LK popolysaccharide,(LPS) on Macrophage l-AAd Expression and preparation through selective ammonium sulfate precipitation Cytotoxicity. Because we observed that the type of elicitation and Con A-Sepharose column chromatography (12). Our re- used to induce macrophage populations was important in de- sults correspond with those of Steeg and Oppenheim (17) who termining whether Ia-Ad antigens were produced, we next also observed that purified IFN-y induced the expression of la antigens on macrophages. The concentrations shown in Ta- Table 1. Effects of LKs on Ia-Ad expression and cytotoxicity of ble 2 were found to be optimal for cytotoxicity, but even at various macrophage preparations concentrations up to 6,000 units of IFN-,B per ml there was % [3H]thymidine no evidence of la expression. % Ia-Ad positivet released* Macrophage On With With With With DISCUSSION preparation* day 0 no LK 2% LK§ no LK 2% LK In accord with results reported by others (18), we found that Resident 8 4 2 2 53 resident peritoneal macrophages or macrophages elicited with Proteose- thioglycollate or proteose-peptone did not express Ia, whereas peptone 5 5 8 4 67 macrophages that appeared in response to the injection of BCG Thioglycollate- or P-815 tumor cells were Ia'. Treatment of the Ia- mac- elicited 6 5 98 6 84 rophages with LKs activated all of these populations to be- WEHI-3 come for tumor but the concomitant appear- (1 x 105) 1 1 100 7 48 cytotoxic cells, WEHI-3 ance of surface Ta depended upon the method used for the (5 x 105) 1 1 100 19 46 original elicitation (Table 1). Ia expression was invariably seen P388D1 after- LK activation of thioglycollate-elicited macrophages, (1 X 104) 1 1 100 2 42 whereas no Ia could be detected on the surface of activated P388D1 resident or proteose-peptone-elicited macrophage popula- (1 X 105) 1 1 100 24 55 tions. Significant differences between these two populations P-815-elicited 58 8 15 5 4 have been reported with respect to 02 production (19), mem- BCG-elicitedl 66 18 100 87 82 brane-associated enzyme activity (20), and bactericidal capac- ity (21). We can speculate that only the recently recruited * All macrophage preparations except WEHI and P388D1 were assayed are less forIa expression and [3H]thymidine release at a macrophage density macrophages become Ia' (10), possibly because they of 2 x 105/cm2. For the exceptions, cell density (no./cm2) is shown in mature as, suggested by Hogg and Parish (22), and that we are parentheses. able to obtain these macrophages only with thioglycollate broth. tPercentages are based on counts of at least 200 cells in at least two Other investigators have used LKs to induce Ia expression different experiments; these values did not vary more than 15%. on peritoneal macrophages and have found that resident mac- t Thymidine release was measured 2 days after the addition of mac- rophages respond only weakly (10), whereas both thioglycol-. rophages to the labeled tumor cells. The mixtures were incubated with became or without 2% LKs. Numbers are the mean values of triplicate de- late- (9) and proteose-peptone-elicited macrophages terminations that did not vary more than 20% from this value. Ia+ after in vitro incubation with LKs. It is unclear whether 11a expression was assayed 3 days after incubation of macrophages the difference in Ia induction between our proteose-peptone- with or without 2% LKs. elicited macrophages and those of Steinman et aL(8) was due BCG-elicited macrophages were prepared, by the injection of 2-3 x to a difference in the health of the mice used, in the proteose- 107 organisms in 0.1 ml into the tail vein of mice. Four weeks later, peptone broth used, in the length of culture with LK (23), or 0.1 ml of purified protein derivative (2 mg/ml) was injected intra- we think that the im- peritoneally and 5 days later the cells were removed from the peri- in the Ia detection assay. However, toneal cavity by lavage, washed, counted, and placed onto the cham- portant points to be made are that the method of elicitation ber slides. can affect Ia expression, possibly due to differences in pros- Downloaded by guest on September 24, 2021 Immunology: Blumenthal et at Proc. Natl. Acad. Sci. USA 80 (1983) 2035 taglandin synthesis by the macrophage populations (24), and This work was supported by National Institute of Allergy and Infec- that macrophage cytotoxicity and Ia expression can be dis- tious Diseases Research Grant AI-03047 and American Cancer Society sociated. Grant IM-204. Thioglycollate-elicited macrophages that are activated by using IFN-p or LPS instead of LK remain la- (Table 2). This 1. Cowing, C., Schwartz, B. D. & Dickler, H. B. (1978)J. Immunol is additional evidence that the activated macrophage need not 120, 378-384. be Ia+ to be cytotoxic. 2. Beller, D. I. & Unanue, E. R. (1980)J. Immunol 124, 1433-1440. LK treatment induced Ia expression in both the WEHI-3 3. Richman, L. K., Klingenstein, R. J., Richman, J. A., Strober, W. and P388D1 cell lines and promoted cytotoxicity, although at & Berzofsky, J. A. (1979) J. Immunol 123, 2602-2609. 4. Yamashita, U. & Shevach, E. M. (1977) J. Immunol 119, 1584- relatively high cell concentrations LK was not necessary for 1588. target cell killing (Table 1). However, at lower cell concen- 5. Cowing, C., Pincus, S. H., Sachs, D. H. & Dickler, H. B. (1978) trations, LK activation was required for cytotoxicity, suggest- J. Immunol 121, 1680-1686. ing that with these cell lines LK treatment amplifies an in- 6. Farr, A. G., Wechter, W. J., Keily, J. M. & Unanue, E. R. (1979) herent low level of cytotoxicity. As expected from earlier work J. Immunol 122, 2405-2412. 7. Habu, S., Hayakawa, K. & Okumura, K. (1979) Cell Immunol 47, (14, 15), BCG infection induced macrophages that were both 416-423. Ia+ and cytotoxic without further treatment, probably as a 8. Steinman, R. M., Nogueira, N., Witmer, M. D., Tydings, J. D. consequence of the infection stimulating LK production in vivo. & Mellman, I. S. (1980)J. Exp. Med. 152, 1248-1261. In contrast, macrophages that were elicited by using irradi- 9. Steeg, P. S., Moore, R. N. & Oppenheim, J. J. (1980)J. Exp. Med. ated P-815 cells were Ia' upon removal from the mice but 152, 1734-1744. were not cytotoxic, and these macrophages could not be made 10. Scher, M. G., Unanue, E. R. & Beller, D. I. (1982) J. Immunol. 128, 447-450. cytotoxic by subsequent LK treatment. This presents another 11. Fidler, I. J. & Roz, A. (1981) in Lymphokines, ed. Pick, E. (Ac- interesting example of how tumor cells may escape immune ademic, New York), Vol. 3, pp. 345-363. surveillance-i.e., by paralyzing elicited macrophages so that 12. Roberts, W. K. & Vasil, A. (1982) J. Interferon Res. 2, 519-532. they are unable to give a cytotoxic response to LK signals. 13. Roberts, W. K. & Vasil, A. (1982) J. Immunol Methods 54, 371- This also shows that Ia expression is not sufficient for mac- 377. rophage cytotoxicity. 14. Marino, P. A., Whisnant, C. C. & Adams, D. 0. (1981)J. Exp. Med. 154, 77-87. In previous work, Roberts and Vasil (12) were unable to 15. McCarthy, M. E. & Zwilling, B. S. (1981) Cell. Immunol 60, 91- separate MAF and IFN-y activities by using techniques of 99. protein purification, protein inactivation, and differential LK 16. Nathan, C. F., Silverstein, S. C., Brukner, L. H. & Cohn, Z. A. induction. Using many of these same techniques, we also have (1979)J. Exp. Med. 149, 100-113. been unable to resolve Ia-inducing factor(s) from MAF/IFN- 17. Steeg, P. S. & Oppenheim, J. J. (1982) Fed. Proc. Fed. Am. Soc. y activities. In addition, we have investigated the LKs pro- Exp. Biol 41, 840 (abstr.). duced by 64 hybridoma clones and 29 subclones after 18. Scher, M. G., Beller, D. I. & Unanue, E. R. (1980)J. Exp. Med. stimulation with Con A (*) and found high qualitative and 152, 1684-1698. quantitative correlations among the production of MAF, IFN- 19. Soberman, R. J. & Karnovsky, M. L. (1981) in Lymphokines, ed. Pick, E. (Academic, New York), Vol. 3, pp. 11-31. y, and Ia-inducing factor. Thus, we think that the possibility 20. Edelson, P. (1981) in Lymphokines, ed. Pick, E. (Academic, New should be considered that the induction of la antigens, cy- York), Vol. 3, pp. 57-83. totoxicity, and the antiviral state represent different activities 21. Miake, S., Takeya, K., Matsumoto, T., Yoshikai, Y. & Nomoto, of a common factor. K. (1980)J. Reticuloend. Soc. 27, 421-427. 22. Hogg, N. & Parish, C. R. (1980) Immunology 41, 187-193. * Zlotnik, A., Roberts, W., Blumenthal, E., Marrack, P. & Kappler, J., 23. Beller, D. I. & Ho, K. (1982)J. Immunol 129, 971-976. T Cell Hybridomas as Sources of Lymphokines, Third International 24. Snyder, D. S. (1982) Fed. Proc. Fed. Am. Soc. Exp. Biol. 41, 815, Lymphokine Workshop, August 1982, Philadelphia. (abstr.). Downloaded by guest on September 24, 2021