Human natural killer cytotoxic factor. Studies on its production, specificity, and mechanism of interaction with target cells

Human peripheral blood lymphocytes cultured in vitro for 2 days in serum-free conditions produced a natural killer (NK) cytotoxic factor (NKCF) which selectively killed NK-susceptible targets. Optimal release of NKCF was achieved under serum-free conditions, while the presence of fetal calf serum inhibited both the production and activity of the factor. Mechanistic studies with NKCF demonstrated that the factor could be adsorbed by the target cells within 6 h, with no further exposure to NKCF required for maximal levels of lysis of the treated targets after additional 30-48 h of incubation, as assessed by a 111I release microcytotoxicity assay. NKCF adsorption to target cells and its cytotoxic activity were inhibited by some phosphorylated sugars (mannose-6PO4 and glucose-6PO4), but not by fructose-6PO4 or nonphosphorylated sugars (mannose, glucose, galactose). These results suggest a role of sugar-6PO4 at the level of interaction of NKCF with NK target cells. This was further supported by the finding that inhibition of target cell glycosylation by tunicamycin also inhibited absorption of NKCF to the target cells and direct killing by NKCF. Therefore, it appears that NKCF is a large granular lymphocyte produced factor which produces lysis as a result of the interaction with glycosylated structures on target cell membranes. Purification studies were performed to begin biochemical characterization of human NKCF. The results indicated that NKCF has an apparent molecular weight between 20,000 and 40,000 dalton. Such approaches with radiolabeled NKCF should be useful for the further study of the biochemical characteristics of human NKCF and of its mechanism of action. The ability to elicit NKCF under serum-free conditions should facilitate its testing, purification, and biochemical characterization.     

Specific effect of anti-transferrin antibodies on natural killer cells directed against tumor cells. Evidence for the transferrin receptor being one of the target structures recognized by NK cells

Treatment of PBL or Percoll-isolated LGL with anti-transferrin antibodies plus complement reduced their natural killing activity against K-562 cells between 30 and 70%. The same antibodies inhibited natural cytotoxicity when added directly to the assay. Similar depletion or inhibition of NK cytotoxicity was observed when using HeLa cells as targets. The decrease or inhibition by transferrin antibodies was less marked when IFN-treated PBL or LGL as effector cells were used. The inhibition of anti-transferrin antibodies seems to be located at the level of the effector cell population. When PBL but not target K-562 cells were pretreated with anti-transferrin antibodies and were washed before use in the assay, cytotoxicity was decreased by 50%. In addition, about 80% of the LGL positively selected on anti-transferrin plates stained with Leu-11. Furthermore, no reduction by anti-transferrin antibodies plus complement treatment of PBL or LGL, or inhibition by antibodies alone, was observed when the cells were tested against HSV-1-infected cells. Membrane extracts from LGL inhibited NK cytotoxicity against K-562 or HeLa cells. Moreover, the inhibitory component of this extract was removed by anti-transferrin IgG but not by control IgG. These results are in agreement with the recent hypothesis that NK cells recognize the transferrin receptor in tumor target cells, because both the transferrin receptor and anti-transferrin antibodies may share a similar structure that interacts with the NK cells.     

Studies on the mechanism of low natural killer cell activity in infant and aged mice

NK activity in mice is high between about 6 and 10 weeks of age. In contrast, infant mice and mice older than 12-14 weeks of age usually have quite low or undetectable NK activity. Studies were performed to analyze the mechanisms underlying this characteristic age-related regulation of NK activity. Spleen cells from infant mice did not develop appreciable NK activity upon incubation for 12-18 h with either interferon (IFN) or interleukin-2 (IL-2). Analysis of the frequency of IL-2-dependent progenitors of NK cells, in a limiting dilution assay, also indicated that the spleens of infant mice are deficient in precursors of NK cells. In contrast, spleen cells from old mice (30 weeks old) developed substantial levels of NK activity upon incubation with either IFN or IL-2, and they showed a frequency of IL-2-dependent progenitors of effector cells that was similar to that of young mice. Both infant and old mice had plastic-adherent suppressor cells in their spleens, which could strongly inhibit NK activity. In addition, both infant and old mouse spleen cells contained nonadherent suppressor cells, which had a higher density on Percoll gradients than NK cells. Thus, several factors appear to contribute to the age-related regulation of NK activity in mice.     

Cytotoxic action of natural killer cells on human tumor cells

Natural killer cell cytotoxicity was studied in a 18-hour 51Cr-release assay in the cultures of human tumor target cells: K562 leukemia and lung adenocarcinoma (LAC) cells. The mean cytotoxic value was similar for K562 and LAC cells: 36.13 +/- 3.23% and 40.78 +/- 3.43%, respectively, although significant individual variability was recorded. The similar cytolytic action of blood mononuclear cells (MNC) on the two tumor lines was observed in 30% of normal donors. MNC from 30% donors produced more pronounced lytic action on K562 cells while MNC from other 30% donors lysed mainly LAC cells. In the competitive inhibition test cold K562 cells more effectively than cold LAC cells suppressed the MNC-induced lysis of both K562 and LAC cells.     

Studies on the mechanism of cortisol inhibition of human natural killer cell activity: effects of calcium entry blockers and calmodulin antagonists

The role of Ca2+ in mediating the inhibition by glucocorticoids of human natural killer (NK) activity was investigated using Ca2+ entry blockers (verapamil and its desmethoxy-derivatives LU46973 and LU47093) and calmodulin antagonists (pimozide and two naphthalenesulfopamide derivatives, W-7 and W-13). Peripheral blood mononuclear (PBM) cell preparations were incubated for 20 h with 1 x 10(-6) M cortisol and these agents in various combinations (concentration range: 1 x 10(-7) - 1 x 10(-5) M) and then assayed in a direct 4-h cytolytic assay using 51Cr-labeled K 562 target cells. Exposure to cortisol led to a significant reduction of NK cell activity (about 50% with respect to the spontaneous activity). Ca2+ entry blockers displayed per se a dose-dependent depressive effect on cytotoxicity and gave significant enhancement of cortisol-dependent inhibition. Calmodulin antagonists were per se minimally effective but clearly amplified the cortisol-mediated inhibition. Raising extracellular Ca2+ by CaCl2 or intracellular Ca2+ by the ionophore A23187 yelded an appreciable reduction of these effects. Our data are compatible with the view that extracellular and intracellular Ca2+ play a role in the control of human NK cell activity. Moreover, it is conceivable that the mechanisms involved in glucocorticoid inhibition of NK cell activity involve Ca2+-dependent pathways.     

Alteration of inhibitory and activating natural killer cell receptor expression on T cells in human immunodeficiency virus-infected Chinese

T cell expression of NKRs can trigger or inhibit cell‐mediated cytotoxicity. However, few studies on T lymphocyte NKR expression in HIV infection exist. Here, we examined the expression patterns of NKG2D, NKG2A, and KIR3DL1 on CD8⁺ and CD3⁺CD8^? cells by multicolor flow cytometry in groups of patients with HIV, AIDS or HAART‐treated AIDS, as well as HIV‐negative normal controls. Individual analysis of KIR3DL1 on CD3⁺CD8⁺ or CD3⁺CD8^? cells revealed no significant differences among any of the groups (P > 0.05). In contrast, the percentage of NKG2A⁺NKG2D^?CD8⁺ T cells was higher in the AIDS group than in the HIV‐negative normal control group (P < 0.01). Meanwhile, the prevalence of NKG2D⁺NKG2A^?CD8⁺T cells was lower in the AIDS group than in HIV‐negative normal controls (P < 0.001). Similar results were also observed for the percentage of NKG2A⁺NKG2D^? on CD3⁺CD8^?cells. However, in contrast to CD8⁺ T cells, the frequencies of NKG2D⁺NKG2A^? on CD3⁺CD8^? cells were higher in AIDS and HIV patients than in HIV‐negative normal controls (P < 0.01, P < 0.05, respectively). The percentage of NKG2A⁺NKG2D^?CD8⁺ T cells was negatively correlated with CD4⁺ T cell counts (r=?0.499, P < 0.01), while the percentage of NKG2D⁺NKG2A^?CD8⁺ T cells was positively correlated with CD4⁺ T cell counts (r= 0.494, P < 0.01). The percentage of NKG2D⁺NKG2A^?CD3⁺CD8^? T cells was also positively correlated with viral load (r= 0.527, P < 0.01) and negatively correlated with CD4⁺ T cell counts (r=?0.397, P < 0.05). Finally, HAART treatment reversed the changes in NKR expression caused by HIV infection. These results indicate that the expression of NKRs on T cells may be correlated with HIV disease progression.     

Phosphatidylinositol metabolism accompanies early activation events in tumor target cell-stimulated human natural killer cells

This study examined the role of phospholipid metabolism in human natural killer (NK) cells upon activation by tumor target cells(TC). The effector cell (EC) population consisted of peripheral blood lymphocytes enriched for NK cells. Upon a 5-min exposure of EC to the NK-sensitive tumor TC K562 and U937, nearly four- and threefold increases in the incorporation of 32P into phosphatidylinositol (PI) occurred, respectively. In contrast, no increase in 32P incorporation into PI was seen when two NK-resistant TC were used. In addition, little or no change in the incorporation of 32P into phosphatidylcholine, phosphatidylethanolamine, or phosphatidylserine took place with any of the above TC. Depletion of Leu 11b-positive cells abolished the increase in 32P incorporation into PI when K562 were used in the phospholipid assay. Furthermore, labeling kinetics of this phospholipid turnover showed that it occurred less than 5 min following exposure to NK-sensitive TC and that phosphatidic acid, a breakdown product of phosphoinositides, was produced during this 5-min period. These results indicated that metabolism of a phosphoinositide took place and that it occurred in association with early activation events in NK cells. Quercetin and dibutyryladenosine-cyclic monophosphate (dbcAMP) plus theophylline exerted profound inhibitory effects on both NK activity and PI metabolism, suggesting a linkage between the two events. The inhibitors had no effect on target cell-binding capacity, indicating that the inhibition occurred postbinding. PI metabolism took place in the absence of extracellular calcium even though NK activity was completely abolished under the same conditions. Thus, we have shown PI metabolism, but not other phospholipids, to occur in human NK cells upon exposure to NK-sensitive TC, in association with early activation events. This event was independent of extracellular calcium and could be inhibited by quercetin or dbcAMP plus theophylline.     

Radiosensitivity of human natural killer cells: binding and cytotoxic activities of natural killer cell subsets

The sensitivity of human natural killer (NK) cell activities (both binding and killing) after exposure of peripheral blood mononuclear cells to different doses of gamma radiation was studied. A panel of monoclonal antibodies was used to identify the NK and T-lymphocyte subsets and to evaluate their radiosensitivity. Peripheral blood mononuclear cells were irradiated with low (2-6 Gy) and high (10-30 Gy) doses and NK cell binding and cytotoxic activity against K562 target cells were studied after 3 h and 48 h in culture. The primary damage to NK cell activity was identified at the postbinding level and affected mainly the lytic machinery. After 48 h culture postirradiation, an overall depression of cytotoxic activity was observed, but ionizing radiation produced either a selection of the more cytotoxic NK cell subsets, which therefore might be considered more resistant to radiation damage than the less cytotoxic NK cells, or a long-term stimulation of cytotoxic activity in surviving cells.     

Decay-accelerating factor expression on either effector or target cells inhibits cytotoxicity by human natural killer cells.

Previous studies have shown that freshly isolated CD16+ NK cells are deficient in the expression of decay-accelerating factor (DAF), or CD55, a membrane regulator of C3 activation. In this study we investigated the significance, for NK cell-mediated lysis, of DAF expression on the target and effector cells. The effect of DAF expression on the susceptibility of NK cell targets was investigated by several means: first, DAF- cell lines were cloned from K562; second, the cloned DAF- cells were reconstituted with exogenous purified DAF; and third, anti-DAF F(ab')2 was used to block DAF function on the DAF+ K562 cells. Consistently, the presence of DAF in the target cell membrane, either naturally occurring or experimentally incorporated, afforded the target cell protection against lysis, and this protection could be blocked with anti-DAF. Similarly, targets for antibody-dependent cell-mediated cytotoxicity with exogenous DAF incorporated in their plasma membrane became less sensitive to antibody-dependent cell-mediated cytotoxicity by NK cells compared with the same target cells without incorporated DAF in their membranes. DAF incorporated in the plasma membranes of the effector NK cells made the NK cells less effective at killing K562 targets. The known function of DAF is to regulate C3 activation, and we were able to demonstrate that the isolated NK cell is capable of releasing C3. It is also possible that the participation of DAF in NK cell function represents a new, noncomplement-dependent function for DAF.     

Renewal of natural killer cells in mice having elevated natural killer cell activity

The present study was designed to examined the dynamics of splenic natural killer (NK) cells under two conditions of enhanced NK cell activity: (1) CBA/J mice given polyinosinic-polycytidylic acid (poly-I:C), an NK-cell-enhancing agent, and 62) untreated athymic nude (nu/nu) mice. The 'total NK cell activity' of the spleen (percentage specific lysis corrected for changes in organ cellularity) increased 5-fold and 2.7-fold after poly-I:C treatment for 1 day and 4 days, respectively. An injection of hydroxyurea (HU), a cell-cycle-toxic drug, given together with either poly-I:C or saline to CBA/J mice resulted in both cases in a 25% reduction in total NK cell activity 1 day later. This suggests that the renewal rate of nondividing NK cells is similar in poly-I:C-treated and saline-injected mice, and that the NK-enhancing effect of poly-I:C is not due to a stimulation of proliferation among NK cell precursors. HU administered simultaneously with poly-I:C or saline for 4 days eliminated NK cell activity in both cases, indicating that spleen NK cell activity is mediated almost entirely by newly formed (less than or equal to 4 days) cells. In nude mice, NK cell activity was assayed at various intervals after an HU depletion period of 2 days. NK depletion was initially more rapid in nu/nu mice than in control (nu/+) mice, although equally profound, and the subsequent recovery of NK cell activity after cessation of HU was also more rapid than in control (nu/+) mice.(ABSTRACT TRUNCATED AT 250 WORDS)     

Cobalt-mediated dimerization of the human natural killer cell inhibitory receptor

Upon engagement of specific class I major histocompatibility complex (MHC) molecules on target cells, inhibitory receptors on natural killer (NK) cells deliver a negative signal that prevents the target cell lysis by NK cells. In humans, killer cell immunoglobulin-related receptors (KIR) with two immunoglobulin-like domains (KIR2D) modulate the lysis of target cells bearing specific HLA-C alleles (Moretta, A., Vitale, M., Bottino, C., Orengo, A. M., Morelli, L., Augugliaro, R., Barbaresi, M., Ciccone, E., and Moretta, L. (1993) J. Exp. Med. 178, 597-604). The transduction of inhibitory signals by KIR2D molecules is impaired by the zinc chelator, 1,10-phenanthroline, and mutation of a putative zinc-binding site (Rajagopalan, S., and Long, E. O. (1998) J. Immunol. 161, 1299-1305), but the mechanism by which zinc may affect the function of KIR remains unknown. In this study, the inhibitory NK receptor KIR2DL1 was discovered to dimerize in the presence of Co(2+) as observed on native gel electrophoresis and by gel filtration column chromatography. Furthermore, Co(2+)-mediated KIR2DL1 dimer binds to HLA-Cw4 with higher affinity than the wild type KIR2DL1 monomer. Replacement of the amino-terminal His residue by Ala abolishes the ability of KIR2DL1 to bind Co(2+), indicating that Co(2+)-mediated KIR2DL1 dimerization involves pairing of the D1 domain. Although not observed on native gels, the inhibitory receptor KIR2DL1 can be chemically cross-linked into dimers in the presence of Zn(2+) and its related divalent metal ions, suggesting that Co(2+)-mediated dimerization of KIR2DL1 may mimic a weaker interaction between KIR2DL1 and zinc in vivo.     

Avian influenza virus directly infects human natural killer cells and inhibits cell activity

Natural killer (NK) cell is a key component of innate immunity and plays an important role in host defense against virus infection by directly destroying infected cells. Influenza is a respiratory disease transmitted in the early phase of virus infection. Evasion of host innate immunity including NK cells is critical for the virus to expand and establish a successful acute infection. Previously, we showed that human influenza H1N1 virus infects NK cells and induces cell apoptosis, as well as inhibits NK cell activity. In this study, we further demonstrated that avian influenza virus also directly targeted NK cells as an immunoevasion strategy. The avian virus infected human NK cells and induced cell apoptosis. In addition, avian influenza virion and HA protein inhibited NK cell cytotoxicity. This novel strategy has obvious advantages for avian influenza virus, allowing the virus sufficient time to expand and subsequent spread before the onset of the specific immune response. Our findings provide an important clue for the immunopathogenesis of avian influenza, and also suggest that direct targeting NK cells may be a common strategy used by both human and avian influenza viruses to evade NK cell immunity.

Open image in new window

     

Target cell specificity of human natural killer cells. II. Apparent change with activation

The activity of natural killer (NK) cells can be augmented by incubation with interferons, or with other compounds, such as staphylococcal protein A, which stimulate interferon production. In the experiments described here we compared the patterns of lytic activity of human lymphoid effector cells, before (NK-B) and after (NK-A) short-term activation. Target cells used were K562, Clone I (a partially NK-resistant K562 variant), and those that had been preincubated with neuraminidase or trypsin. The results obtained include the following: proteases, but not neuraminidase, decreased lysis by NK-B, but not by NK-A, of both K562 and Clone I in the standard Cr-release assay. In the single cell assay, trypsin minimally decreased conjugate formation and the fraction of bound cells that were killed by NK-B, but did not reverse the increased lytic efficiency of NK-A. In the cold-target competition assay, Clone I, which does not compete as well with K562 for NK-B, did so equally well for NK-A. Trypsinized targets, regardless of their equal sensitivity to lysis by NK-A, were not as active competitors for NK-A. We conclude that the most reasonable interpretation is that K562 cells bear surface structures which can induce release of lytic mediators from NK-A under conditions that are not sufficient to stimulate NK-B. Although it appears that NK-A may respond to a smaller number of the same target molecules recognized by NK-B, the process must be better defined at the molecular level to exclude the possibility that there are qualitative differences between the proposed recognition structures for these two states of NK activity.     

Studies on the mechanism of natural killer cell-mediated cytotoxicity. VII. functional comparison of human natural killer cytotoxic factors with recombinant lymphotoxin and tumor necrosis factor

The present study was undertaken to evaluate the possible contribution of other cytokines to the lytic activity of NKCF-containing supernatants. We compared some of the functional properties of human NKCF and purified recombinant human rLT and rTNF. It was found that the target cell specificity of rLT was quite different from NKCF in that rLT was neither species specific nor NK specific. Furthermore, antibodies against rLT did not affect the lytic activity of NKCF. These results demonstrate that LT does not significantly contribute to the lytic activity mediated by NKCF. The target specificity of rTNF was found to be related to that of NKCF with the exception of one NK-resistant cell line that was lysed by rTNF in a 20-hr 51Cr-release assay. However, rTNF was not toxic to any of the target cells tested as assessed by trypan blue exclusion in a 20-hr assay unless the targets were labeled with 51Cr. In contrast, NKCF did kill target cells as detected by trypan blue exclusion that were not labeled with 51Cr. Further analysis of this mechanistic difference in the lytic activity of rTNF and NKCF revealed that rTNF in combination with either cycloheximide or mitomycin C but not IFN-gamma could lyse unlabeled U937 target cells. In addition, pretreatment of U937 target cells with nonradioactive Na2CrO4 at concentrations equivalent to that used to 51Cr-labeled cells resulted in their susceptibility to lysis by rTNF as assessed by trypan blue exclusion. These findings suggest that lysis of several susceptible target cells in 20 hr by rTNF requires the presence of additional agents that may be sublethally toxic and/or inhibitory to macromolecular synthesis. Antibody inhibition studies revealed that anti-TNF mediated from partial to complete inhibition of lysis of U937 by unfractionated supernatants containing NKCF. However, fractionation of such supernatants on chromatofocusing columns yielded two distinct peaks of activity eluting in the pH range of 5 to 6 and 7 to 8. Anti-TNF could inhibit the acidic form of NKCF but not the neutral form. It is concluded that NKCF activity is mediated in part by TNF or an antigenically related molecule as well as some other distinct factor(s). The lack of consistent inhibition of NK CMC by anti-TNF suggests that TNF alone is not sufficient to mediate NK activity, or else it is inaccessible to the added antibody.     

Characterization and utilization of a monoclonal antibody inhibiting porcine natural killer cell activity for isolation of natural killer and killer cells

A mAb, porcine NK-inhibitory mAb (PNK-I) that inhibits porcine NK activity without affecting antibody-dependent cellular cytotoxicity (ADCC) has been developed. PNK-I acts at the level of the effector cell and inhibition of NK activity is independent of complement. Inhibitory effects are seen against various human and murine NK-susceptible targets. Addition of PNK-I antibody up to 60 min after assay initiation was effective at inhibiting NK activity. Furthermore PNK-I does not inhibit E:T conjugation and inhibits during the Ca2(+)-dependent phase of NK cytolysis. PNK-I Ag is present on virtually all PBL showing a bimodal distribution with 74% "dim" and 15% "bright" by flow cytometry. Monocytes and granulocytes stain with an intermediate intensity with greater than 90% and 95% staining positively, respectively. F(ab')2 fragments of PNK-I antibody show identical staining and functional activity as the whole molecule indicating that PNK-I acts independently of FcR. PNK-I immunoprecipitates molecules of molecular mass of 166, 155, 95 kDa under reducing and nonreducing conditions. PNK-I appears to be recognizing an epitope on a CD18 molecule. The CD18 molecule (beta-chain of CD11a,b,c) is ubiquitous on the surface of leukocytes and is implicated in a variety of cellular functions. Dim and bright populations were sorted and assessed functionally for NK and ADCC activity. It is demonstrated that PNK-I+ bright lymphocytes contain all detectable NK and ADCC activity in porcine PBL. Furthermore PNK-I+ bright lymphocytes contain the cytokine responsive NK cells capable of stimulation by IL-2, porcine NK-activating factor, and porcine natural killer-enhancing mAb. PNK-I+ dim cells were devoid of all baseline as well as inducible NK and ADCC activity. Giemsa stain of sorted populations show PNK-I+ bright cells containing the large granular lymphocytes whereas dim are devoid of these. Two color analysis show that PT4+ cells are PNK-I+ dim whereas PT8+ lymphocytes are divided between PNK-I+ bright and dim populations. Our results indicate that we are able to isolate all active as well as inducible NK and ADCC effector cells from porcine PBL based on relative Ag expression of CD18. Therefore quantitative as well as qualitative antigen expression is important in NK/ADCC-mediated cytotoxicity.     

Role of interferon in human natural killer activity against target cells infected with HSV-1

Human natural killer (NK) cells show high cytotoxic activity against target cells infected with herpes simplex virus type 1 (HSV-1). Substantial amounts of interferon (IFN) were generated in co-cultures of NK effector cells and infected target cells; however, the cytotoxic activity seen against a specific infected cell target did not correlate with the amount of IFN induced. The production of IFN increased steadily from 4 to 18 hr of co-culture, as did NK activity; however, IFN production peaked 4 hr later than NK activity. Pretreatment of NK effector cells with exogenous IFN increased cytotoxic activity against all targets tested, but the differential pattern of reactivity against cells infected with wild type and mutant viruses was unaltered. When effector cells were treated with the RNA synthesis inhibitor actinomycin D before co-culture with virus-infected targets, IFN production was markedly reduced, without a concomitant reduction in cytotoxicity. Similarly, the addition of anti-IFN antiserum to co-cultures greatly decreased the available IFN present, but had no effect on NK activity. We conclude that the induction of cytotoxic activity in co-cultures of NK effector cells and HSV-1-infected target cells is independent of the induction of IFN.     

Lysis of human solid tumor cells by lymphokine-activated natural killer cells

The ability of NK cells to lyse noncultured solid tumor cells was investigated, and the results were compared with lysis of K562. Purified NK cell fractions separated by either Percoll centrifugation or a cell sorter exhibited higher level of lysis against noncultured melanoma cells than did NK-depleted cell fractions. However, the level of lysis was low (less than 10% lysis). Adding recombinant interleukin 2 (rIL 2) to the 4-hr assay induced significant lysis (more than 10%) of noncultured melanoma cells in 18 of 23 (78%) Percoll-enriched NK cell fractions and seven of 11 (64%) sorted Leu-11a+ cells at an E:T ratio of 80 and 10, respectively. In contrast, only two of 13 (14%) PBMC, five of 17 (29%) Percoll-decreased NK cell fractions, and one of 12 (8%) sorted Leu-11a- cells lysed noncultured melanomas in the presence of rIL 2. rIL 2 induced NK cells to lyse noncultured lung and breast cancer cells, as well as melanoma tumors. Exposure of NK cells to 2000 rad radiation abrogated the rIL 2-induced cytotoxicity against noncultured melanomas. Preculture of PBMC for 18 hr with recombinant interferon-gamma (rIFN-gamma) resulted in a modest level of lysis of non-cultured melanomas by sorted Leu-11a+ cells. Adding rIL 2 to the assay increased the cytotoxic activity in both rIFN-gamma-activated Leu-11a+ and Leu-7+ NK subsets. The level of noncultured tumor lysis correlated well with that of K562 lysis in all of the experiments. Purified NK cell fractions in rIL 2 cultures increased cytotoxic activity against noncultured tumor cells with incubation time for up to 3 days, and the level of NK cell-mediated lysis was dependent on both doses of rIL 2 and length of incubation. In contrast, both NK-depleted and sorted Leu-11a- cells demonstrated very low levels of solid tumor lysis after 3-day cultures with a high dose of rIL 2. Killer cell precursors induced by 3-day cultures of sorted cell fractions with rIL 2 and rIFN-gamma were found in both Leu-11a+ and Leu-7+ NK subsets, but not Leu-4+ or Leu-3a+ T lymphocytes. These results indicate that NK cells become cytotoxic for noncultured solid tumor cells by a brief contact with rIL 2, and increase cytotoxic activity after culture with rIL 2.