Decreased natural killer activity in thalassemia major: a possible consequence of iron overload

We have investigated the natural killer (NK) activity of both fractionated (Percoll density gradient) and unfractionated mononuclear cells from patients with beta-thalassemia major who are iron overloaded as a consequence of chronic transfusion therapy. These patients were found to have significantly decreased NK activity against K562 targets at all effector:target ratios tested (p less than 0.001). Both splenectomized and nonsplenectomized patients had normal proportions of Leu-11b-staining (NK) cells. Since they had normal to elevated absolute white cell and lymphocyte counts, a change in the absolute number of NK cells could not account for the decreased killing. We also found that the decrease in NK activity was transfusion related (r = -0.603, p less than 0.005). To determine whether or not this decrease in NK activity could be related to iron overload, we preincubated patient effector cells with desferrioxamine (DFO) or 2,3-dihydroxybenzoic acid (DHB) for 6 hr before addition of K562 targets. Both of these iron-chelating agents consistently increased the NK activity of cells from thalassemia patients. DHB had the greater effect, being able to increase patient NK activity to virtually normal levels. On the other hand, preincubation of cells from normal controls with DHB caused only a slight increase in NK activity, while similar treatment with DFO had little or no effect. When target cells were preincubated with the chelating agents before addition of either normal or patient effector cells, no change in cytotoxicity was seen, demonstrating that the chelating agents act at the effector cell level. Furthermore, if the chelating agents were saturated with iron prior to preincubation with the effectors, no increase in the cytotoxicity of thalassemic NK cells was observed. These results indicate that thalassemia patients have a reversible, transfusion-related decrease in NK function which may arise as a consequence of iron overload.     

Natural killer (NK) cells are present in mice with severe combined immunodeficiency (scid)

Spleen cells from C.B- 17 scid mice with severe combined immunodeficiency disease exhibit natural killer cell (NK) activity against YAC lymphoma targets in a standard 4-hr 51Cr release assay. The cytolytic activity is demonstrable only at high effector to target ratios but can be augmented at least sevenfold by the interferon inducer poly I:C. The pattern of target lysis is specific, because splenocytes from poly I:C-primed C.B-17 scid mice lyse NK-sensitive YAC cells and not the insensitive P815 mastocytoma. The presence of several NK-associated antigens on C.B-17 scid splenocytes was tested by pretreating cells with the appropriate antiserum plus complement before testing for NK activity. The results indicate that a proportion of NK effectors in C.B-17 scid mice bear surface NK 2.1 and Asialo GM1 but are negative for Thy-1.     

Natural killer cell activity correlates with the amount of beta 2-microglobulin on human peripheral blood lymphocytes

Two cytotoxic assays, lectin-dependent cytotoxicity and natural killer (NK) cytotoxicity, were used to assess the competence of cord blood and neonatal peripheral blood mononuclear cell (PBMC) and T-cell cytotoxic reactions. The effect of exogenous interferon was also studied. Results were compared with cytotoxic capabilities of adult cells and cells from patients with primary immunodeficiency syndromes. Lectin-dependent cytotoxicity (LDCC), a property of both T and non-T cells, was assessed by lysis of chromium-labeled EL4 tumor target cells in the presence or absence of exogenous fibroblast interferon (IFN-β). Natural killer cytotoxicity was assessed by lysis of two different chromium-labeled tumor target cells, Molt 4f and K562 in the presence or absence of IFN-β. Lectin-dependent cytotoxicity (LDCC) of PBMC of cord blood (32 ± 4% SEM) and adult cells (36 ± 2% SEM) were equivalent but neonatal cells had slightly decreased LDCC (22 ± 3% lysis). T-depleted cells from cord or neonatal blood had increased LDCC but T-enriched (>95% sheep erythrocyte rosette-forming cells) from both cord (22 ± 3%) and neonatal blood (18 ± 5%) had significantly reduced LDCC compared to 55 ± 2% for adult T cells. This deficiency corrects with age and is near normal after age 2. Preincubation with IFN-β did not enhance LDCC of newborn or adult cells. The LDCC of some cord T cells was markedly reduced and was in the same low range as patients with severe combined immunodeficiency. Natural killer (NK) cytotoxicity of PBMC from cord and adult cells was equivalent at three effector:target ratios against the Molt 4f target but against the K562 target, cord PBMC had significantly less NK activity (22 ± 11 SD) compared to adult NK activity (50.5 ± 22.2 SD) at a 50:1 effector:target ratio. Similar differences were noted at 25:1 and 10:1 target:effector ratios. NK cytotoxicity against Molt 4f targets of adult cells was significantly enhanced by preincubation with IFN-β but NK of cord cells was only variably enhanced. By contrast, IFN-β enhanced NK against K562 targets of both adult and cord cells, adult greater (67.7 ± 20) than cord cells (37.8 ± 2.0). These T-cell effector deficiencies are in marked contrast to the vigorous proliferative responses of newborn T cells, and parallel deficiencies of certain neonatal lymphokines. These defects may explain the newborns' enhanced susceptibility to intracellular viruses and to congenital viral infections.     

Natural killer cells in patients with chronic alcoholism

The study of the activity of natural killer cells in patients with chronic alcoholism, stage II, has revealed its essential decrease in comparison with that in healthy subjects, which is particularly pronounced with effector/target cell ratio equal to 100:1 and 50:1 respectively. Besides, two subgroups can be differentiated among the total population of alcoholic patients: subgroup 1 with the activity of killer cells remaining practically intact and subgroup 2 with profound defects in the functioning of these cells. The presence of essential defects in natural killer cells of patients with the severe form of alcoholism is suggested.     

Characterization of the cellular basis for the inhibition of cytolytic effector cells by murine placenta

Direct suppression of cytolytic effector cell function by cells of the placenta may represent one mechanism that protects the "fetal allograft" from rejection by maternal transplantation immunity. Collagenase disaggregated murine placental cells block target cell lysis by natural killer, lymphokine-activated killer, and (CTL)-type killer cells. This inhibition is reversible and noncompetitive, similar to a previously described inhibitor of CTL found in spleens of mice undergoing an acute graft vs host (GVH) response. Velocity sedimentation separation of placental cells shows that the inhibitory activity is primarily associated with cells that cosediment with nucleated fetal erythrocytes. When these erythrocytes were lysed, an increased number of non-erythrocytic cells could be separated and under this circumstance, inhibitory activity was seen in association with either small white cells or fetal erythrocytes and with large white cells. There may be several cell populations in murine placenta that can inhibit cytolytic effector cells. The possible relevance of direct placental inhibition of cytolytic effectors to protection of the "fetal allograft" is discussed.     

The role of asparagine-linked carbohydrate in natural killer cell-mediated cytolysis

Chinese hamster ovary cell lines with specific lesions in the formation of glycoconjugates were tested for their sensitivity to lysis by interferon-boosted human natural killer cells. We report here that the type of asparagine-linked carbohydrate present on target cell glycoproteins determines their susceptibility to natural killer lysis. The targets tested were Chinese hamster ovary parent cells and Lec1, Lec2, and Lec8 mutants. Lec8 and Lec2 cells show an overall reduction of galactose and/or sialic acid in their glycoconjugates due to defects in the translocation of UDP-galactose and CMP-sialic acid, respectively. Due to a specific block in N-linked carbohydrate processing, Lec1 cells produce only high mannose-type oligosaccharides, but their glycolipids are identical to those of the parent. Both Lec2 and Lec8 mutants are more sensitive to natural killer lysis than the parent cells. This is consistent with their extensive reduction in cell surface sialic acid. Furthermore, Lec1 mutants are more susceptible to natural killer lysis than the parent cells. To confirm that the increased natural killer sensitivity of Lec1 cells was due to the modification of N-linked carbohydrate, parent cells were treated with swainsonine, a specific inhibitor of N-linked oligosaccharide processing. Swainsonine-treated parent cells are nearly as sensitive to natural killer lysis as the Lec1 mutants.     

Lysis of dengue virus-infected cells by natural cell-mediated cytotoxicity and antibody-dependent cell-mediated cytotoxicity

Peripheral blood mononuclear cells (PBMC) from humans without antibodies to dengue 2 virus lysed dengue 2 virus-infected Raji cells to a significantly greater degree than uninfected Raji cells. The addition of mouse anti-dengue antibody increased the lysis of dengue-infected Raji cells by PBMC. Dengue 2 immune human sera also increased lysis of dengue-infected Raji cells by PBMC. These results indicate that both PBMC-mediated cytotoxicity and antibody-dependent cell-mediated cytotoxicity (ADCC) can cause significant lysis of dengue-infected Raji cells. The lysis of infected Raji cells in the ADCC assay correlated with the dilution of dengue-specific antibody which was added, indicating the dengue virus specificity of the lysis of dengue virus-infected Raji cells. Alpha interferon (IFN alpha) was detected in the culture supernatant of PBMC and dengue-infected Raji cells. However, enhanced lysis of dengue-infected Raji cells by PBMC may not be due to the IFN produced, because neutralization of all IFN activity with anti-IFN alpha antibody did not decrease the lysis of dengue-infected cells, and effector cells pretreated with exogenous IFN alpha also lysed dengue-infected cells to a greater degree than uninfected cells. The effector cells responsible for lysis of dengue virus-infected Raji cells in the natural killer and ADCC assays were analyzed. Nonadherent PBMC caused more lysis than did adherent cells. Characterization of nonadherent cells with monoclonal antibodies showed that the predominant responsible effector cells were contained in OKM1+ and OKT3- fraction in the natural killer and ADCC assays.     

Effect of interferon γ on the sensitivity of bovine-papilloma-virus(BPV1)-transformed cell lines to cell-mediated cytotoxicity

Summary The effect of interferon (IFN) on the immunogenicity and immunosensitivity of mouse cell lines transformed by bovine papillomavirus type 1 (BPV1) DNA was examined in a syngeneic mouse model. The overnight incubation of BPV1-transformed cell lines with 100 IU/ml IFN did not affect their ability to induce the generation of cytotoxic effector cells but it clearly increased their sensitivity to lysis by interleukin-2-induced lymphokine-activated killer (LAK) cells and by nonspecific LAK-type effector cells induced by BPV-1-transformed cell lines. The treatment of two allogeneic lymphoid tumour cell lines, P815X2 and YAC-1, with IFN either decreased or had no effect on their sensitivity to LAK-cell-mediated lysis.     

Functional properties of lymphocyte subpopulations in hepatitis B virus infection. II. Cytotoxic effector cell killing of targets that naturally express hepatitis B surface antigen and liver-specific lipoprotein

Cytotoxic effector cell responsiveness to host and/or virus-determined hepatocyte surface membrane antigens has been postulated as an important pathogenetic determinant of hepatocellular injury in hepatitis B virus infection. Assuming that such effector cell populations would be detectable in peripheral blood, the present study was designed to examine 2 questions: first, whether target cells that normally express liver-specific protein (LSP) and hepatitis B surface antigen (HBsAg) are selectively destroyed by peripheral blood effector cells from patients with viral hepatitis; second, whether cytotoxic effector cell function emerges coincident with the development of defective suppressor cell activity in the same patients. No evidence of increased HBsAg or LSP specific cytotoxic effector cell activity was found in the peripheral blood natural killer (NK) or T killer cell populations of patients with acute or chronic viral hepatitis.     

Lysis of endothelial cells by autologous lymphokine-activated killer cells

The mechanisms of lysis of endothelial cells derived from human umbilical vein (HUVEC) by autologous lymphokine-activated killer (LAK) cells, generated from cord blood lymphocytes of the same donor, were investigated. Freshly isolated HUVEC as well as HUVEC cultured for several passages were efficiently lysed by autologous LAK cells, and their susceptibility to the LAK cells was almost the some as that of allogenic HUVEC. Complement-depletion experiments revealed that the lysis was mainly dependent on CD16-natural killer (NK) LAK cells. Pretreatment of HUVEC with recombinant interferon (rIFN) for 24 h made them resistant to lysis by autologous LAK cells, while pretreatment with either rIL-1. rTNF, or acidic or basic fibroblast growth factor did not alter the lytic sensitivity of HUVEC. The resistance of rIFN-treated HUVEC was specific to lysis by CD16⁺ NK LAK cells, and their lysis by CD3⁺ T-LAK cells was not significantly altered. Moreover, in comparison with control HUVEC or rIL-1-treated HUVEC, rIFN-treated HUVEC had a significantly less potent inhibitory effect on the lysis of untreated HUVEC, when used as an unlabeled target. This suggests that rIFN treatment may down-regulate the recognition of some molecules on HUVEC by rIL-2-activated NK cells. These data suggest that damage of the endothelium during LAK therapy is mainly dependent on LAK cells with a NK phenotype that can specifically recognize a certain molecule on autologous endothelial cells.     

Antitumor immunity and cellular cancer therapies

The identification of tumor specific antigens has provided important advance in tumor immunology. It is now established that specific cytotoxic T lymphocytes (CTL) and natural killer cells infiltrate tumor tissues and are effector cells able to control tumor growth. However, such a natural antitumor immunity has limited effects in cancer patients. Failure of host defenses against tumor is consecutive to several mechanisms which are becoming targets to design new immunotherapeutic approaches. CTL are critical components of the immune response to human tumors and induction of strong CTL responses is the goal of most current vaccine strategies. Effectiveness of cytokine therapy, cancer vaccines and injection of cells improving cellular immunity have been established in tumor grafted murine models. Clinical trials are underway. To day, interest is particularly focused on cell therapy: injected cells are either "ready to use" effector cells (lymphocytes) or antigen presenting cells able to induce a protective immune reaction in vivo (dendritic cells). The challenge ahead lie in the careful optimization of the most promising strategies in clinical situation.     

lack of suppressor cell activity for natural killer cells in infant, aged and a low responder strain of mice

We have examined the reported role of suppressor cells in the regulation of NK activity in mice with naturally low NK activity (infant and aged (C57 X A)F1 hybrids (CAF1) and low responder strain AKR mice). Possible suppressor activity was assayed by mixing, at a 1 : 1 ratio, spleen cells from low activity mice with spleen effector cells from normally active 8 to 10 wk old CAF1 mice. The lytic activity of the mixture was compared with the activity of effector cells diluted with medium alone or diluted 1 : 1 with "non-suppressor" population which served as a control for nonspecific decreases in lysis. The control or "filler" cells employed were suspensions of adult CAF1 thymus, thymus from adult mice exposed to 2,000 R, and adult CAF1 spleen cells cultured for 24 hours, a procedure that depleted NK activity. In no case was the activity observed in the presumed suppressor-effector mixture significantly lower than that observed in the filler-effector cell mixtures. Thus, in infant (1 to 2 wk) and aged (12 to 18 mo) CAF1 mice and in 8 to 10 wk old AKR mice, we found no evidence for specific cell-mediated suppression of natural cytotoxicity.     

The role of monocytes and natural killer cells in mediating antibody-dependent lysis of colorectal tumour cells

Monocytes and natural killer (NK) cells are known to be important effector cell populations in mediating antibody-dependent cell-mediated cytotoxicity (ADCC). Purified monocyte and NK effector cell populations, from normal and colorectal cancer (CRC) patients, together with a number of murine (17-1A and 323/A3) and their chimaeric (c17-1A) or humanised (3622W94) equivalents, and chimaeric (c) SF25 were compared for their ability to mediate ADCC of colorectal tumour cells. The chimaeric and humanised antibodies were significantly better at mediating tumour lysis than their murine equivalents with all-effector populations. When effector cells from CRC patients were used the cSF25 antibody was significantly better than 3622W94 (P < 0.02) which, in turn, was significantly better than c17-1A (P < 0.03). Depletion of NK cells produced a decrease in specific tumour lysis with all antibodies. In addition a higher rate of NK cell death was observed in CRC patients during the assay than in normal controls. The chimaeric and humanised antibodies stained a similar percentage of the HT-29 target cells (>80%), but 3622W94 bound to significantly more cells from primary tumour biopsies than cSF-25 (P = 0.001). Together, the results suggest that NK cells are the most important effector cell type mediating ADCC in vitro, that there is some impairment of NK function in CRC patients, and that cSF25 is the most potent antibody. For use in vivo the anti-Ep-CAM antibody 3622W94 would appear to be the most suitable reagent for further study. Received: 3 June 1999 / Accepted: 22 July 1999     

The anticancer drug,cisplatin, increases the naturally occurring cell-mediated lysis of tumor cells

Summary It has been proposed that a component of the antitumor potential of the chemotherapeutic agent, cisplatin, resides in the host's ability to respond to cisplatintreated tumor cells. Here we report that tumor cells that are normally resistant to lysis mediated by naturally occurring cytotoxic cells showed an increased sensitivity to lysis mediated by murine spleen cells or human peripheral blood monocytes and lymphocytes when cisplatin was added at the beginning of the lytic assay. This was shown for the lysis of both murine and human tumor cells. The pretreatment of tumor cells, but not effector cells with cisplatin caused an increase in lysis in the presence of murine spleen cells or human peripheral blood leukocytes, indicating that the effect of cisplatin is to reduce resistance to lysis by these effector cells. The lysis of tumor cells by naturally occurring cytotoxic cells was blocked by antibodies specific for tumor necrosis factor. In addition, the ability of cisplatin to increase lysis was seen with cells that are sensitive to natural cytotoxic cells, but not with cells that are sensitive to natural killer cells. These results suggest that the effector cells that mediate the lysis of these tumor cells in the presence of cisplatin are likely to be natural cytotoxic cells. The ability of cisplatin to increase the lysis of tumor cells by naturally occurring cytotoxic cells indicates that these cells may be a host defense mechanism that contributes to the anticancer potential of cisplatin.     

Natural killer cells in rhesus monkeys: properties of effector cells which lyse Raji targets

Spontaneously occurring natural killer cell activity of rhesus monkey peripheral blood mononuclear cells was assayed against five human cell lines, three of which were Epstein-Barr virus (EBV) positive, including the human B cell line Raji. The lytic activity to Raji cells was high, significantly higher than to any other cell line tested. Raji cells are normally insensitive to spontaneous lysis by human NK cells, and the contrasting vigor of the rhesus monkey cytolytic activity to Raji prompted us to investigate the properties of this effector cell. We found the effector cell-mediating lysis of Raji to be nonadherent and phagocytic with lytic activity slightly enhanced in the E-rosette-forming cell (ERFC+) fraction and decreased in the ERFC- fraction. Further isolation of FcIgG receptor-positive and FcIgG receptor-negative subsets by rosetting resulted in significant enrichment of NK activity to Raji in the positive fraction and a loss of activity in the negative fraction. Depletion studies with various monoclonal antibodies (mAb's) confirmed that nearly all lytic activity was contained in the CD16+ (Leu 11b+) population, while subsets of effector cells expressed CD2 (9.6) and CD8 (OKT8). Depletion of CD4 (OKT4)-, HLADR (OKIa)-, or LFA1 (MAC-1)-positive populations failed to reduce NK activity. We compared the phenotypic properties of alloimmune effector cells exhibiting specificity for allogeneic donor targets with those exhibiting lysis of Raji targets. Results indicated that allospecific cytotoxic T lymphocytes expressed a CD16-, CD2+ phenotype, a pattern distinct from that of the effector cell population recognizing Raji targets. The presence of CD2 mAb's in the culture had no effect on NK lytic activity. In contrast, mAbs CD8 and Leu 11b were inhibitory. This would suggest a functional role for CD8 and FcIgG molecules in the lysis of Raji cells by rhesus effectors. In summary, these studies describe a distinct population of effector cells in the blood of rhesus monkeys which exhibit spontaneous lytic activity to Raji cells and exhibit the properties of NK cells.     

In vitro modulation of natural killer cell activity in non-Hodgkin's lymphoma patients after therapy

Summary The depressed natural killer (NK) activity, antibody-dependent cellular cytotoxicity (ADCC) and NK cytotoxic factor cytotoxicity in untreated non-Hodgkin's lymphoma patients were found to be elevated after chemotherapy. In vitro treatment of the effector NK cells with interferon could augment the NK activity in normal subjects and treated patients to a comparable degree. Chemotherapy mainly affected the post-binding events in the NK cytotoxic process by causing an increase in the active killing potential of the NK cells. This study provides a better understanding of changes in the NK cytotoxic mechanism in non-Hodgkin's lymphoma patients and the role of interferon in this process.B. A. Mehta is a recipient of the Lady Tata Memorial Trust, India, Senior Scholarship     

Immunogenetic studies on the cell-mediated cytotoxicity in the clawed toadXenopus laevis

In order to test for further homologies between the MHC of mammals and amphibians, experiments were conducted to assess whether lower vertebrates such as anurans were able to generate killer cells after allogeneic stimulation. The generation of cytotoxic effector cells could be obtained in outbred families and clones of isogenic frogs after in vivo priming with either irradiated allogeneic lymphocytes or with an allogeneic skin graft, provided that the immune spleen cells were restimulated in vitro with the specific irradiated cells used for priming. Effector cells generated against a definedMHC haplotype could lyse targets having one of their two haplotypes in common with the stimulators. In contrast, no lysis was observed when the target cells differed from the specific stimulators by twoMHC haplotypes.The cytotoxic activity of the MLR-restimulated lymphocytes appeared to be mediated by T cells since passage of the effector spleen cells through a nylon wool column, under conditions which removedXenopus B lymphocytes, improved killing on a per cell basis. It therefore appears that the genes responsible for the highly specialized function of T killer cells have emerged early in evolution at least at the time of the emergence of the amphibians (300 million years ago) and that they were already linked to the MHC of this species. The MHC polymorphism inXenopus seems to be lower than in mammals as evidenced by the high frequency of cross-killing observations paralleled by the high frequency of MLR identical animals found in a large outbred population.     

Renal carcinoma cell lines inhibit natural killer activity via the CD94 receptor molecule

MHC class I molecules protect normal and transformed cells from lysis by natural killer (NK) cells through recognition of receptors expressed on leucocytes. Defects in NK cell activity and lymphokine activated killer (LAK) cell generation have been previously demonstrated in patients with renal cell carcinoma (RCC). However, to date, the importance of NK receptor/MHC class I interactions for immune evasion by RCC cells has not been described. In this study, human RCC cell lines (HTB46, HTB47, ACHN, CRL 1933 and HTB44) were found to be susceptible to lysis by both NK cells and interleukin-15 (IL-15)-derived LAK cells from normal donors in vitro. However, when NK cells were co-cultured with RCC cells their expression of the CD94 NK receptor molecule was significantly increased and their cytolytic activity against RCC targets was reduced. The cytolytic activity of NK cells was restored by the addition of IL-15, which further augmented the expression of CD94 on CD56+ NK cells. Disruption of NK receptor-MHC class I interactions by the addition of blocking antibodies to CD94 had no effect on the lysis of K562 or HTB47 targets by NK cells. However, the sensitivity of HTB46 cells to NK-mediated lysis was increased by blocking the CD94 receptor molecule, but only when the NK cells had not been previously co-cultured with RCC cells. This was independent of the presence of IL-15. These results show that RCC cells can inhibit NK activity via CD94 and suggest that disruption of interactions between receptor and ligand on RCC cells in vivo may augment the immune response against tumours by innate effector cells.     

Role of interferon in natural kill of HSV-1-infected fibroblasts

The production of interferon during natural killer (NK) assays against HSV-1-infected fibroblasts (NK(HSV-1)) was studied to determine whether this interferon was responsible for inducing the preferential lysis of herpes-virus-infected target cells over uninfected target cells. The interferon produced during NK(HSV-1) assays was analyzed and found to have the properties of HU-IFN-alpha. Little or no IFN was produced during NK assays against uninfected fibroblasts (NK(FS)) or K562 (NK(K562)) cells. Although the appearance of interferon in the culture supernatants seemed to parallel the development of cytotoxicity during NK(HSV-1) assays, the levels of cytotoxicity and IFN generated did not correlate, arguing against a strict quantitative dependence of cytotoxicity upon IFN production. NK(K562) and NK(FS) cytotoxicity developed with little or no production of IFN. When IFN-pretreated effector cells were used, there was still a preferential lysis of infected over uninfected target cells. This preferential lysis by IFN-treated effector cells of infected over uninfected targets was seen as early as 2 hr into the assay. Anti-IFN antibodies added to the NK assays, although neutralizing all the IFN produced during the assays, had no effect on NK(FS) or NK(K562) cytotoxic activity and caused a slightly reduction of NK(HSV-1) activity only in one of three experiments. We conclude that although IFN is generated during NK(HSV-1) assays, this IFN cannot solely account for the increased lysis of infected over uninfected cells and that NK(HSV-1) activity is in some other way dependent on the virus infection.     

Immunologic and clinical aspects of natural killer cells in human leukemia

We have studied peripheral-blood, splenic and bone marrow natural killer (NK) activity in patients with leukemia. These studies demonstrated that leukemic patients displayed defective NK activity in all of these tissues. However, NK defect could be corrected by culture of effector cells with interleukin-2 (IL-2). The phenotypic analysis of IL-2 cultures showed clearly the heterogeneity of lymphocyte subsets. The characterization studies demonstrated that CD56+, CD3- NK cells manifested most potent lysis of leukemia, CD56+, CD3+ T cells mediated some, but low, antileukemia activity and CD56-, CD3+ T lymphocytes were devoid of cytotoxicity.