Expression of asialo GM1 and other antigens and glycolipids on natural killer cells and spleen leukocytes in virus-infected mice

The sensitivities of mouse natural killer (NK) cells to various antisera and complement were analyzed at different time points after acute lymphocytic choriomeningitis virus infection. Under these conditions NK cell activity peaks 3 days and virus-specific cytotoxic T cell activity 7 days after infection. The sensitivity of the cytotoxic activities to antibodies to asialo GM1 (AGM1), NK 1.2 alloantigen, and Ly 5 was in the order endogenous NK greater than day 3 NK greater than day 7 NK. Day 7 cytotoxic T cells were more resistant than day 7 NK to anti-AGM1 and to anti-NK 1.2, but more sensitive to anti-Ly 5. This decreased sensitivity of activated NK cells to antibodies and C' was examined in more detail for the AGM1 antigen. Antibody to AGM1 completely depleted NK cell activity in control, but not in day 3 lymphocytic choriomeningitis virus infected mice. However, mice treated before infection with antibody did not generate NK cell activity 3 days after infection. The mechanisms of the decreased sensitivity of activated NK cells to antibody to AGM1 was examined. High levels of antibody depleted activity, indicating that the effectors were not devoid of AGM1. Biochemical analyses of spleen leukocytes revealed marked increases in sialic acid, gangliosides, and neutral glycosphingolipids, including AGM1 in the order day 7 greater than day 3 greater than endogenous. Antibody to AGM1 was absorbed out by leukocytes in the order day 7 greater than day 3 greater than endogenous. Flow cytometry (FACS) analyses revealed marked shifts in the frequency and intensity of staining of cells with antibody to AGM1 in the order day 7 greater than day 3 greater than endogenous. All endogenous NK cell activity and all the large granular lymphocytes were associated with the brightest 5% of the total spleen leukocyte population. Day 3 and day 7 NK cell activity was also located in cells sorted by using the gate settings for the top 5% endogenous cells. However, there were marked increases in the number of the very bright cells in the order day 3 greater than day 7 greater than endogenous. These cell numbers correlate with the level of NK cell activity in these fractions. Thus, the decreased sensitivity of activated NK cells to antibody to AGM1 is not due to decreased expression of AGM1 on NK cells, but to a competition for antibody by greatly increased levels of AGM1 in infected spleen leukocytes.(ABSTRACT TRUNCATED AT 400 WORDS)     

A subset of asialo GM1+ cells play a protective role in the occurrence of graft-versus-host disease in mice

In three different murine models of bone marrow (BM) transplantation the capacity of asialo GM1+ cells to suppress graft-vs-host disease (GVHD) was investigated. In a first model, total lymphoid irradiation (TLI)-treated BALB/C mice were given 1 mg of anti-asialo GM1 antibody. This led to the disappearance of functional suppressor cells after TLI. Injections of anti-asialo GM1 into TLI-treated BALB/C mice before infusion of 30 x 10(6) fully allogeneic (C3H) BM cells, led to a significantly decreased survival rate as compared to TLI-treated mice injected with control serum before BM transplantation (survival 29 and 83%, respectively, at 120 days after transplantation, p = 0.0032 log rank). The mortality of the former group was due to GVHD as 1 degree all dying animals showed clinical and histologic signs of GVHD, 2 degrees all animals were chimeric and 3 degrees mice receiving no or syngeneic BALB/C BM had excellent survival rates excluding BM aplasia or increased susceptibility for infections as reason for the mortality of the allogeneic BM recipients. In a second model, asialo GM1+ cells were removed in vitro from the C3H BM inoculum before injection into lethally irradiated (9 Gy) BALB/C recipients. In mice kept in specific pathogen-free conditions, this procedure resulted into a significant mortality (12/12) as compared to mice receiving BM pretreated with control serum (1/12, p = 0.0001 log rank). When kept in conventional housing, GVHD occurred in both groups but much earlier in the group receiving anti-asialo GM1-treated BM (median survival time 6 vs 46 days for the control mice, p = 0.001 log rank). No animal receiving anti-asialo GM1 and treated with syngeneic BM died, thus excluding toxicity, increased susceptibility to infections, or decreased graft take as a cause of mortality. In a last model, asialo GM1 cells were removed from syngeneic BM in a BM transplantation model in which T cell-depleted syngeneic (BALB/C) and non-T cell-depleted allogeneic (C3H) BM was administered to lethally irradiated (9 Gy) BALB/C mice. Also in this model GVHD-related mortality only occurred in the group of mice receiving syngeneic BM from which asialo GM+ cells were depleted before infusion (3/12). Our experiments thus clearly show that asialo GM1+ cells from both recipient (the TLI model) as well as donor origin (the TBI experiments) can suppress the occurrence of GVHD.     

Suppression of B-cell differentiation by natural killer (asialo GM1+) cells in mice

Mice were injected intravenously with rabbit antiserum to ganglio-n-tetraosylceramide (asialo GM1, ASGM1), a neutral glycosphingolipid present at high quantities on the surface of natural killer (NK) cells. Spleen cells prepared from the mice were then examined for NK activity against YAC-1 targets, for phagocytic cells and by flow cytometric analysis for Thy1, Lyt1, Lyt2, ASGM1 and surface Ig (SIg) phenotypes. Administration of anti-ASGM1 in mice resulted in a complete depletion of NK activity and ASGM+1 cells in the spleen, but no changes in the proportions of Thy1+ cells and their Lyt1+ and Lyt2+ subsets and phagocytic cells. Corresponding to this selective depletion of ASGM+1 cells and NK activity, the spleen cells showed an increased number of SIg+ B cells and augmented mitogenic responses to B-cell but not T-cell mitogens. These NK-depleted spleen cells also showed production of pokeweek mitogen (PWM)-driven plaque-forming cells (PFC) to much higher levels than those of control spleens. In the spleens of mice treated with varying concentrations of anti-ASGM+1, a good correlation was found between the decreased NK activity and the enhanced PFC response. To directly test the possible suppressor activity of NK cells on PWM-induced PFC response, NK (ASGM+1) cells were highly purified from the spleen by a combination of Percoll gradients and cytolysis of T cells by monoclonal antibodies followed by indirect panning. When added to NK-depleted spleen cells, they suppressed the augmented PFC response of NK-depleted spleen cells, depending on the number of cells added. These results suggest that NK (ASGM+1) cells in mice exhibit a suppressor property on B cells, which are undergoing spontaneous or mitogen-induced differentiation.     

IL-4 and T cells are required for the generation of IgG1 isotype antibodies against cardiolipin

Infection with Mycobacterium tuberculosis induces Abs against a vast array of mycobacterial lipids and glycolipids. One of the most prominent lipid Ags recognized is cardiolipin (CL). The kinetics of the generation of anti-CL Abs during infection reveals that IgM titers to CL increase over time. Interestingly, at day 30 postinfection CL-specific IgG1 appears, an isotype usually dependent on T cell help. Using an immunization schedule with CL/anti-CL Ab complexes, which induces antiphospholipid syndrome in mice, we show that the generation of IgG1 to CL requires IL-4 and that optimal production is T cell dependent. IgG1 production to CL was impaired in nude (nu/nu) mice devoid in conventional T cells, but was not affected in mice deficient for either alphabeta TCR(+), gammadelta TCR(+), CD4(+), CD8(+), or NK1.1(+) T cells. We conclude that IgG1 production to CL depends on T cell help and IL-4, which can be provided by different T cell populations. This is the first report that IL-4 is indispensable for the induction of IgG1 Abs to lipid Ags.     

The role of IL-2 and T4+ cells in the generation of human influenza virus-specific CTL activity

Stimulation of human peripheral blood lymphocytes (PBL) with influenza A virus leads to the generation of virus-specific cytotoxic T lymphocyte (CTL) activity as well as natural killer (NK)-like activity. In this study, we show that exogenous IL-2 augments the in vitro generation of virus-specific CTL activity, only when added some days after the initiation of the culture. Apparently, the endogenously produced IL-2 can be a limiting factor in the in vitro generation of CTL activity. The increase of influenza virus-specific CTL activity after addition of exogenous IL-2 does not affect the restriction pattern of the CTL response. So, the preferential use of certain HLA antigens as restriction elements is not due to a limiting amount of endogenously produced IL-2. Depletion of T4+ cells completely abrogates the generation of virus-specific CTL activity. Addition of exogenous IL-2 to T4+-cell-depleted cultures fully restores the generation of HLA-restricted virus-specific CTL activity. We conclude that in the in vitro generation of virus-specific CTL activity in bulk cultures of human PBL the sole function of T4+ cells in human virus-specific CTL generation is the production of IL-2, no cognitive cell interaction of T8+ CTL precursors with T4+ cells is required, and in bulk cultures T8+ cells themselves are not able to produce sufficient amounts of IL-2 to ascertain the maturation of virus-specific CTL precursors into cytolytic T cells. Finally, we show that exogenous IL-2 also has a stimulatory effect on the NK-like or lymphokine-activated killer activity, which is always concomitantly induced in virus-specific CTL generation cultures, but has no influence on the levels of IFN produced in such cultures.     

Characterization of responding cells in oxidative mitogen stimulation. I. Ia+ cells and Ly-1+2+ cells are required for the proliferative response

Addition of anti-Ia sera to cultures of mouse spleen cells stimulated with oxidative mitogens, neuraminidase/galactose oxidase (NaGO) or sodium periodate (NaIO4), inhibits the subsequent proliferative response 30 to 70%. Anti-H-2K or D sera were not specifically inhibitory. Similar inhibition was seen when cells were pretreated with anti-Ia sera and washed before exposure to the mitogenic enzymes. Treatment with anti-Ia serum and complement depletes greater than 89% of the NaGO and the NaIO4 responses but 2% or less of the phytohemagglutinin (PHA) response. The response to NaGO was sensitive to depletion with anti-Thy-1 serum, rabbit anti-mouse brain serum, anti-Ly-1, or anti-Ly-2 serum. Mixtures of Ly-1 and Ly-2-depleted populations did not restore responsiveness. Thus both an Ia+ cell and an Ly-1+2+ T cell are required for [3H]TdR incorporation in response to NaGO treatment.     

Anti-glycoprotein D monoclonal antibody protects against herpes simplex virus type 1-induced diseases in mice functionally depleted of selected T-cell subsets or asialo GM1+ cells.

Passive transfer of a monoclonal antibody (MAb) specific for glycoprotein D (gD) is highly effective in preventing the development of herpes simplex virus type 1-induced stromal keratitis. In the present study, we investigated whether animals which had been functionally depleted of T-cell subsets or asialo GM1+ cells would continue to be responsive to MAb therapy. BALB/c mice were depleted of CD4+, CD8+, or asialo GM1+ cells by treatment with anti-L3T4, anti-Lyt 2.2, or anti-asialo GM1 antibodies, respectively. Functional depletion of CD4+ cells was documented by the loss of delayed-type hypersensitivity responsiveness, while CD8+ cell depletion was accompanied by abrogation of cytotoxic lymphocyte activity. Anti-asialo GM1 treatment led to the loss of natural killer cell lytic activity. Mice depleted of the desired cell population and infected on the scarified cornea with herpes simplex virus type 1 uniformly developed necrotizing stromal keratitis by 3 weeks postinfection. A single inoculation of anti-gD MAb (55 micrograms) given intraperitoneally 24 h postinfection strongly protected hosts depleted of CD4+ cells against stromal keratitis. Likewise, antibody treatment in CD8+ or asialo GM1+ cell-depleted hosts was as therapeutically effective as that seen in non-cell-depleted mice. We also observed that in cell-depleted mice, the virus spread into the central nervous system and caused encephalitis. The CD4+ cell-depleted mice were the most severely affected, as 100% developed fatal disease. Anti-gD MAb treatment successfully protected all (32 of 32) CD4+-, CD8+-, or asialo GM1(+)-depleted hosts against encephalitis. We therefore conclude that antibody-mediated prevention of stromal keratitis and encephalitis does not require the obligatory participation of CD4+, CD8+, or asialo GM1+ cells. However, when mice were simultaneously depleted of both CD4+ and CD8+ T-cell subsets, antibody treatment could not prevent fatal encephalitis. Thus, antibody can compensate for the functional loss of one but not two T-lymphocyte subpopulations.     

Inducing P-selectin ligand formation in CD8 T cells: IL-2 and IL-12 are active in vitro but not required in vivo

In vitro studies have demonstrated that IL-2 and IL-12 can support formation of P-selectin ligands (P-SelL) in activated T cells, ligands that are variably required for efficient lymphocyte recruitment to sites of inflammation. To ascertain whether these cytokines were required for P-SelL formation in vivo, TCR transgenic CD8 T cells specific for male Ag (HY) were transferred into male mice under conditions in which either IL-2 and/or IL-15 or IL-12Rp40 were absent. P-SelL formation at day 2 was unperturbed in HY-TCR IL-2(null) CD8 T cells responding in doubly deficient IL-2(null)IL-12(null) or IL-2(null)IL-15(null) male recipients. HY-specific CD8 T cell proliferative responses detected in both spleen and peritoneum occurred vigorously, but only splenic CD8 T cells up-regulated P-SelL, demonstrating that in vivo induction of P-SelL is an active, nonprogrammed event following T cell activation and that despite the efficacy of IL-2 and IL-12 in supporting P-SelL formation in vitro, these cytokines appear to be dispensable for this purpose in vivo.     

Natural killer-type lysis of thymocytes from young mice by normal spleen cells in vitro

Normal spleen cells from 6- to 10-week-old mice, enriched for natural killer (NK) cells on a discontinuous polyvinylpyrrolidone-silica (Percoll) gradient, lyse thymocytes of young mice (less than or equal to 19 days old) in a short-term 51Cr release assay. The highest NK-type activity was found in band 3 (density less than or equal to 1.077 g/ml) of a four-step gradient. In some experiments band 2 (density less than or equal to 1.070 g/ml) also showed NK activity. Activity was not unequivocably detectable in cells before separation or in bands 1 and 4. These results also show that the thymocyte sensitivity is dependent on the age of the target cell. Sensitivity of very young thymocytes (less than or equal to 7 days old) was higher than that of thymus cells from 8- to 19-day-old donors. Moreover, it seemed that syngeneic target thymocytes were lysed more effectively than allogeneic. Thus, an NK-type cell population may have the ability to lyse immature thymic target cells at an early stage of their differentiation. This could be of importance as a physiological mechanism for controlling the T cell repertoire and its reactivity.     

Requirement for non T-cells in the generation of cytotoxic T lymphocytes (CTL) in vitro. II. Characterization of the active cells in the spleen of nude mice.

Small numbers of LN cells will produce many more cytotoxic lymphocytes on in vitro culture with allogeneic stimulator cells if spleen cells from nu/nu mice are also present throughout the culture period. All cytotoxic cells produced are T cells and arise from precursors in the LN component. The nude spleen component appears to be providing a required non-T cell which has been lost from the LN component through dilution. At least two active subpopulations of cells, differing in sedimentation velocity, adherence properties, radiation sensitivity, and antigen recognition properties can be identified in the nu/nu spleen. The first, the dominant activity in normal nu/nu spleen, is nonadherent, radiation sensitive, and can synergize with either syngeneic or allogeneic LN cells provided both are different from the same alloantigens in the stimulator population. The second, found in nu/nu spleen cells precultured with alloantigen, sediments more rapidly, is adherent, and radiation resistant, and need not be allogeneic to the stimulator cells to synergize with LN cells. The first subpopulation may give rise to the second.     

Slo1 tail domains,but not the Ca2+ bowl,are required for the beta 1 subunit to increase the apparent Ca2+ sensitivity of BK channels

Functional large-conductance Ca(2+)- and voltage-activated K(+) (BK) channels can be assembled from four alpha subunits (Slo1) alone, or together with four auxiliary beta1 subunits to greatly increase the apparent Ca(2+) sensitivity of the channel. We examined the structural features involved in this modulation with two types of experiments. In the first, the tail domain of the alpha subunit, which includes the RCK2 (regulator of K(+) conductance) domain and Ca(2+) bowl, was replaced with the tail domain of Slo3, a BK-related channel that lacks both a Ca(2+) bowl and high affinity Ca(2+) sensitivity. In the second, the Ca(2+) bowl was disrupted by mutations that greatly reduce the apparent Ca(2+) sensitivity. We found that the beta1 subunit increased the apparent Ca(2+) sensitivity of Slo1 channels, independently of whether the alpha subunits were expressed as separate cores (S0-S8) and tails (S9-S10) or full length, and this increase was still observed after the Ca(2+) bowl was mutated. In contrast, beta1 subunits no longer increased Ca(2+) sensitivity when Slo1 tails were replaced by Slo3 tails. The beta1 subunits were still functionally coupled to channels with Slo3 tails, as DHS-I and 17 beta-estradiol activated these channels in the presence of beta1 subunits, but not in their absence. These findings indicate that the increase in apparent Ca(2+) sensitivity induced by the beta1 subunit does not require either the Ca(2+) bowl or the linker between the RCK1 and RCK2 domains, and that Slo3 tails cannot substitute for Slo1 tails. The beta1 subunit also induced a decrease in voltage sensitivity that occurred with either Slo1 or Slo3 tails. In contrast, the beta1 subunit-induced increase in apparent Ca(2+) sensitivity required Slo1 tails. This suggests that the allosteric activation pathways for these two types of actions of the beta1 subunit may be different.     

Effect of long-term treatment with platelet-activating factor on IL-1 and IL-2 production by rat spleen cells

To study the effect of the in vivo administration of platelet-activating factor (PAF) on cytokine production, alzet minipumps loaded with the mediator or solvent alone were connected to the jugular vein and placed under the skin of Sprague-Dawley rats. Over 7 days the animals received total doses of 0.5, 1, 4.5, 9, or 28 micrograms PAF or the solvent alone. The spleen mononuclear cells isolated from Ficoll gradients and the adherent cell fraction were separated before determination of basal and mitogen-stimulated IL-1 and IL-2 production, respectively. Adherent splenocytes from rats having received 28 micrograms PAF exhibited a decreased capability to produce IL-1, as compared to those from vehicle-treated animals. In contrast, adherent splenocytes from rats having received 9 and 4.5 micrograms PAF yielded higher amounts of released and cell-associated IL-1 activity upon LPS stimulation, as compared to those from solvent-treated animals. The PAF antagonist, BN 52021, given orally (5 mg/kg, twice a day throughout the experiments) inhibited the in vivo effect of 28 micrograms PAF. Statistically significant 144 +/- 43% (p less than 0.001, n = 5) and 73 +/- 33%, (p less than 0.01, n = 3) increases in IL-2 production were observed when whole spleen mononuclear cells from rats administered with 1 and 4.5 micrograms PAF, respectively, were stimulated with Con A. BN 52021 markedly inhibited the in vivo effect of 1 microgram PAF on the IL-2 release. Our study demonstrates that PAF can modulate immune functions in vivo and suggests that the specific PAF antagonist, BN 52021, may be used as an immunomodulatory agent.     

Role of accessory cells in B cell activation. IV. Ia+ accessory cells are required for the in vitro generation of thymic independent type 2 antibody responses to polysaccharide antigens

It has been previously shown that the in vitro antibody response to TNP-Ficoll requires the presence of adherent accessory cells. In order to determine if this characteristic was unique to TNP-Ficoll or a general feature of the TI-2 antibody responses, responses to the polysaccharide antigens TNP-Levan and TNP-Dextran were studied. Also, it was determined if the functionally relevant accessory cell expresses Ia determinants. Passage of spleen cells over Sephadex G-10 abrogated the response to TNP-Levan and TNP-Dextran as well as to TNP-Ficoll. Addition of adherent accessory cells to the G-10 passed spleen cells reconstituted the response to all 3 antigens. Pretreatment of the adherent accessory cells with a specific anti-Ia serum plus complement abrogated the ability of these cells to provide accessory cell function in the responses to all 3 antigens. Thus, an Ia-positive adherent accessory cell is required for the generation of TI-2 antibody responses to these polysaccharide antigens. This raises the possibility that genetic restrictions may exist between the Ia-positive accessory cell and the lymphocytes involved in the responses to TNP-Ficoll, TNP-Dextran, and TNP-Levan.     

Th1 and Th2 cells are required for both eosinophil- and neutrophil-associated airway inflammatory responses in mice

Most current animal models focus on eosinophil-mediated asthma, despite compelling evidence that a neutrophil-mediated disease occurs in some asthma patients. Using intranasal challenge of mice sensitized either orally or nasally with whole peanut protein extract in the presence of cholera toxin, we developed mouse models of eosinophil- and neutrophil-mediated asthma, respectively. In this study, mice deficient in Th1 (IL-12 and IFN-gamma) or Th2 (IL-4 and IL-13) pathways were used to characterize the role played by Th1 and Th2 cytokines during the initial priming phase in the two models. Antigen-specific Ab responses were controlled primarily by Th2 cytokines in mice sensitized by the oral route, whereas Th1 cytokines appeared to play a predominant role in mice sensitized by the nasal route. Furthermore, the absence of key Th1 or Th2 cytokines during the initial phase of priming reduced lung reactivity in both mouse models of airway inflammation.     

PEA3 and AP-1 are required for constitutive IL-8 gene expression in hepatoma cells

Interleukin-8 (IL-8) mRNA was constitutively expressed in human hepatoma cell line, HepG2 and in human hepatocellular carcinoma (HCC), which often form hypervascular tumors. The sequence 5'-AGGAAG-3' at -137 to -132 bp of IL-8 promoter was shown to be polyomavirus enhancer A binding protein-3 (PEA3) binding site, which can cooperate with activator protein-1 (AP-1). Both PEA3 and AP-1 are essential for constitutive IL-8 expression in HepG2 cells, determined by promoter assays. Moreover, PEA3 and IL-8 proteins coexisted in HCC tissues, but not in uninvolved liver tissues. It is possible PEA3 may have important roles in tumor progression and in angiogenesis in HCC.     

B cells are required for generation of protective effector and memory CD4 cells in response to Pneumocystis lung infection

B cell-deficient mice are susceptible to infection by Pneumocystis carinii f. sp. muris (PC). To determine whether this susceptibility is due to a requirement for B cells to prime T cells, we compared CD4 T cell responses to PC in bone marrow chimeric mice that express MHC class II (MHCII) on all APCs (wild-type (WT) chimeras) and in bone marrow chimeric mice that express MHCII on all APCs except B cells (MHCII(-/-) chimeras). Although PC was rapidly cleared by WT chimeric mice, PC levels remained high in chimeric mice that lacked MHCII on B cells. In addition, although T cells were primed in the draining lymph nodes of MHCII(-/-) chimeric mice, the number of activated CD4 T cells infiltrating the lungs of these mice was reduced relative to the number in the lungs of WT chimeras. We also adoptively transferred purified CD4 T cells from the draining lymph nodes of PC-infected normal or B cell-deficient mice into SCID mice. Mice that received CD4 cells from normal mice were able to mount a response to infection in the lungs and clear PC. However, mice that received CD4 cells from B cell-deficient mice had a delayed T cell response in the lungs and failed to control the infection. These data indicate that B cells play a vital role in generation of CD4(+) memory T cells in response to PC infection in the lungs.     

Anti-CD3 + IL-2-stimulated murine killer cells. In vitro generation and in vivo antitumor activity

The growth, phenotype, in vitro cytolytic characteristics, and in vivo antitumor activity of murine splenocytes stimulated with anti-murine CD3 mAb in combination with IL-2 as compared with IL-2 alone was investigated. When cultured for 12 days with anti-CD3 mAb + IL-2, murine splenocytes increased 100- to 4000-fold in number compared with only 6- to 20-fold for cultures stimulated with IL-2 alone. Anti-CD3 mAb + IL-2 activated cultures developed high lymphokine-activated killer activity against NK-resistant targets including the P815 mastocytoma cell line and fresh MCA 106 sarcoma. Peak cytotoxicity on a per cell basis developed by day 8 after anti-CD3 mAb + IL-2 activation. A large proportion of the total cytolytic activity of long term anti-CD3 mAb + IL-2-stimulated cultures was related to the presence of anti-CD3 in the assay, indicating enhancement of cytotoxicity by activated CD3+ T cells. Phenotypic analysis indicated that anti-CD3 mAb + IL-2-stimulated cultures contained heterogeneous populations of T cells with increased percentages of both CD4+ and CD8+ phenotypes compared with cultures stimulated with IL-2 alone. Anti-CD3 mAb + IL-2-stimulated cells were tested for their in vivo antitumor activity by using C57BL/6 mice bearing MCA 106 sarcoma pulmonary metastases. IL-2-activated murine killer cells were given in combination with in vivo IL-2 and indomethacin, the latter of which was shown to potentiate the antitumor effect of IL-2. When given on day 5 after tumor inoculation, cell doses as low as 5 x 10(6) anti-CD3 mAb + IL-2-stimulated cells per mouse significantly reduced the number of pulmonary metastases (p less than 0.005). Thus, activation with the combination of anti-CD3 mAb + IL-2 produces rapidly expanding cultures of cytolytic cells with demonstrated in vivo antitumor efficacy.