Recognition specificities, development, and possible biological function of natural killer cells in the mouse. I. Spleen focus forming assay for natural killer activity and analysis of lectin-like recognition structures on the surface of murine natural killer cells

A number of simple sugars have been tested and found to be effective in blocking lysis of YAC-1 tumor target cells by nonimmune murine natural killer (NK) effector cells. Using a spleen fragment culture system an assay has been developed which allows us to compare the inhibition of lysis observed in replicate culture wells prepared from cells contained in one spleen fragment (less than or equal to 1 X 10(6) cells). The inhibition pattern of any well was found to fall naturally into 1 of 25 (of the total 128 possible tested) patterns. Using this panel analysis of NK activity in individual mice of the same or different strain has been compared. Our data suggest that within any given strain the inhibition pattern of NK effector cells is quite uniform. Consistent differences are seen between strains which are interpreted in terms of a genetic control of the final expression of the NK recognition repertoire. In adult F1 hybrid individuals the pattern of recognition by NK cells is best considered a result of the codominant expression of genes contributed by each parent.     

Expression of surface lymphotoxin and tumor necrosis factor on activated T, B, and natural killer cells.

The expression of membrane-associated forms of lymphotoxin (LT) and TNF were examined on cell lines of T, B, and myeloid origin, IL-2 dependent T cell clones, and peripheral blood lymphocytes. Inducible and constitutive patterns of surface LT expression were found on T cells as exemplified by the II-23.D7, a CD4+T cell hybridoma, and HUT-78, a T cell lymphoma. Phorbol ester induced surface LT expression on Ramos, an EBV transformed B cell line, but at a slower rate of appearance when compared to the II-23.D7. Secretion of LT was rapidly inducible by phorbol ester in II-23.D7 and also in HUT-78 but with slower kinetics; surface LT expression continued in both lines after secretion had ceased. Low levels of membrane TNF were transiently induced on II-23.D7 and HUT-78, but none was observed on Ramos. Peripheral blood monocytes and some myeloid tumor lines did not express surface LT. Several T cell clones expressed surface LT after Ag-specific stimulation, and expression persisted several days. Stimulation through the TCR or by IL-2 rapidly induced surface LT on resting peripheral T cells and CD56+ NK cells; pokeweed mitogen activation induced expression on CD20+ B cells. Consistent with previous results, immunoprecipitation with anti-LT mAb showed that LT was complexed with a distinct 33 kDa glycoprotein (p33) on cells that expressed surface LT, whereas secreted LT was not associated with p33. Surface and secreted modes of LT expression by activated T, B, and NK cells suggests that LT can be utilized as either a localized or diffusible mediator in immune responses.     

The over-expression of P-glycoprotein in K-562 and DAUDI cells, is associated with a high susceptibility to NK and LAK cells.

We evaluated the susceptibility to natural killer (NK) or lymphokine activated killer (LAK) cell-mediated cytolysis of two pairs of drug sensitive/resistant tumor cell lines which were extensively characterized at phenotypic and genotypic level. In the DAUDI cell system, the acquired capability of tumor cell variants to grow in the presence of a relatively high concentration of vinblastine (VBL) is associated with a marked increase to NK and LAK susceptibility. In contrast in the K-562 cell system, no correlation between drug-resistance, P-glycoprotein expression and susceptibility to NK or LAK activity seems to occur.     

The effect of the Cmv-1 resistance gene, which is linked to the natural killer cell gene complex, is mediated by natural killer cells.

The resistance of mice to lethal infection by murine CMV (MCMV) is under complex host genetic control with contributions from both H-2 and non-H-2 genes. We have previously shown that an autosomal, non-MHC encoded gene, Cmv-1, controls MCMV replication in the spleen. We have investigated the mechanism by which the Cmv-1 resistance gene confers protection against MCMV infection. Using H-2 compatible irradiation bone marrow chimeras, the enhanced resistance to MCMV infection that is associated with the Cmv-1l allele in the C57BL background was shown to be mediated by an irradiation-sensitive bone marrow-derived cell population, or a factor produced by these cells. The lack of correlation between serum IFN titers and the strain distribution pattern of Cmv-1 in CXB recombinant inbred mouse strains suggests that IFN does not mediate resistance conferred by this gene. Similarly, the lack of effect of in vivo depletion of mature CD4+ and CD8+ T cells on virus replication in C57BL/6J mice indicates that T cells are unlikely to be involved. In contrast, in vivo depletion of NK cells by injection of the anti-NK1.1 mAb PK136 abrogated restricted splenic virus replication in C57BL/6J----BALB.B chimeric mice and in the Cmv-1l CXB strains. These data indicate that the effect of the Cmv-1 gene is mediated by NK cells. The significant augmentation in NK cell activity after MCMV infection of the susceptible Cmv-1h strains (BALB/cBy), CXBG/By, CXBH/By, CXBI/By, and CXBK/By) indicates the existence in these mice of NK cells that are functionally and phenotypically distinct from those in Cmv-1l strains. NK cells present in the Cmv-1h strains are unable to restrict efficiently splenic MCMV replication in vivo, possibly due to a lack of specificity for virus-infected target cells. Finally, flow cytometric analysis of NK1-1 expression in CXB and BXD RI mice together with MCMV replication studies in the BXD RI strains indicate that Cmv-1 is closely linked to NK1.1 and other loci that reside on a distal segment of murine chromosome 6 in a region that has recently been defined as the natural killer complex.     

Co-migration and internalization of transferrin and its receptor on K562 cells

The incorporation of iron into human cells involves the binding of diferric transferrin to a specific cell surface receptor. We studied the process of endocytosis in K562, a human erythroid cell line, by using tetramethylrhodamine isothiocyanate-labeled transferrin (TRITC- transferrin) and fluorescein isothiocyanate-labeled Fab fragments of goat antireceptor IgG preparation (FITC-Fab-antitransferrin receptor antibody). Because the antireceptor antibody and transferrin bind to different sites on the transferrin receptor molecule it was possible to simultaneously and independently follow ligand and receptor. At 4 degrees C, the binding of TRITC-transferrin or FITC-Fab antitransferrin receptor antibody exhibited diffuse membrane fluorescence. At 20 degrees C, the binding of TRITC-transferrin was followed by the rapid formation of aggregates. However, the FITC-Fab antitransferrin receptor did not show similar aggregation at 20 degrees C unless transferrin was present. In the presence of transferrin, the FITC-Fab antitransferrin receptor antibody formed aggregates at the same sites and within the same time period as TRITC transferrin, indicating co-migration. Although the diffuse surface staining of either label was removed by proteolysis, the larger aggregates were not susceptible to enzyme degradation, indicating that they were intracellular. The internal location of the aggregates was also demonstrated using permeabilized cells that had been preincubated with transferrin and fixed with 4% paraformaldehyde. These cells showed aggregated receptor in the interior of the cell when reacted with fluorescein-labeled antibody to the receptor. This indicated that the transferrin and the transferrin receptor co-internalize and migrate to the same structures within the cell.     

Fractionation of K-562 cells on the basis of their surface properties by partitioning in two-polymer aqueous-phase systems

The K-562 cell line is a culture of human leukemia stem cells originally derived from a patient with chronic myelogenous leukemia in blast crisis. We have subjected such cells, in the log phase of growth, to countercurrent distribution in a charge-sensitive dextran-polyethylene glycol aqueous-phase system, a method that fractionates cells on the basis of subtle differences in their surface properties, and found that: (1) The cell population is heterogeneous since it is composed of cells with different partition ratios. (2) There is a correlation between increasing cell partition ratios and increasing cell electrophoretic mobilities. (3) Cells under different parts of the distribution curve have dissimilar ratios of cells in different parts of the cell cycle, a phenomenon that may, at least partially, be the basis for the subfractionation of these cells. There is a clear tendency for cells in G0 + G1 + early S to decrease and for those in late S + G2 + M to increase with increasing partition ratios. (4) Sialic acid is a major surface charge component of the cells as evidenced by a dramatic drop in their partition ratios after treatment with neuraminidase.     

Fractionation of K-562 cells on the basis of their surface properties by partitioning in two-polymer aqueous-phase systems

The K-562 cell line is a culture of human leukemia stem cells originally derived from a patient with chronic myelogenous leukemia in blast crisis. We have subjected such cells, in the log phase of growth, to countercurrent distribution in a charge-sensitive dextran-polyethylene glycol aqueous-phase system, a method that fractionates cells on the basis of subtle differences in their surface properties, and found that: (1) The cell population is heterogeneous since it is composed of cells with different partition ratios. (2) There is a correlation between increasing cell partition ratios and increasing cell electrophoretic mobilities. (3) Cells under different parts of the distribution curve have dissimilar ratios of cells in different parts of the cell cycle, a phenomenon that may, at least partially, be the basis for the subfractionation of these cells. There is a clear tendency for cells in G₀+G₁+early S to decrease and for those in late S+G₂+M to increase with increasing partition ratios. (4) Sialic acid is a major surface charge component of the cells as evidenced by a dramatic drop in their partition ratios after treatment with neuraminidase.     

IFN-gamma treatment of K562 cells inhibits natural killer cell triggering and decreases the susceptibility to lysis by cytoplasmic granules from large granular lymphocytes

Pretreatment of human K562 leukemia cells with rIFN-alpha and rIFN-gamma resulted in decreased susceptibility to lysis by human peripheral blood NK cells. The reduction of NK-susceptibility after IFN treatment was not due to a general effect of IFN on the stability of the cell membrane because the susceptibility of K562 cells to lysis by antibodies plus C, distilled water, or lysolecithin was unaffected. Binding studies with effector cell preparations enriched for NK cells with large granular lymphocyte morphology revealed no difference in binding to control and IFN-gamma-treated target cells. The sensitivity to soluble NK cytotoxic factors was not affected significantly by the IFN treatment. In contrast, the susceptibility of IFN-treated target cells to the cytotoxic activity of purified cytoplasmic granules from a rat large granular lymphocyte tumor was significantly reduced, indicating that the IFN-induced resistance acted at the level of susceptibility to the lytic mechanism of NK cells. However, IFN-alpha was more effective than IFN-gamma in inducing resistance to the cytoplasmic granules although resulting in only a weak resistance in the cell-mediated cytotoxic assay. IFN-gamma but not IFN-alpha caused a reduction in the frequency of effector cells that had reoriented their Golgi apparatus toward their bound target cell. In addition, IFN-gamma treated K562 cells failed to elicit an influx of Ca2+ into effector cells. Taken together, the results suggest that IFN-gamma in addition to an increased resistance to the lytic molecules released by NK cells can also induce changes in the target cells which prevent the triggering and activation of the effector cell.     

The effect of natural killer cells on the development of syngeneic hematopoietic progenitors

To determine whether natural killer (NK) cells are involved in the regulation of hematopoiesis, well-characterized, cell sorter-purified NK cells were incubated with syngeneic bone marrow, and the effect of this interaction on the development of various hematopoietic progenitors was assessed. NK cells were obtained from the peritoneal exudates of CBA/J mice after i.p. infection with live Listeria monocytogenes (LM). These NK cells were nylon wool-nonadherent and were purified by using M1/70, a rat anti-murine macrophage monoclonal antibody, and a fluorescence-activated cell sorter (FACS). Syngeneic bone marrow was incubated overnight with these M1/70-purified NK cells. The cells were then assayed in vitro to determine the effect on the colony formation of the following hematopoietic progenitor cells: the myeloid progenitor that produces mixed granulocyte/macrophage colonies (CFU-G/M), the myeloid progenitor that is committed to macrophage differentiation (CFU-M), and the early erythroid progenitor that is known as the burst-forming unit-erythroid (BFU-E). The marrow cells, after incubation with NK cells, were also injected into lethally irradiated syngeneic recipients to assay for the splenic colony formation capacity of the trilineage myeloid stem cell (CFU-S). Although the formation of BFU-E-, CFU-G/M-, and CFU-M-derived colonies was not adversely affected by the exposure of syngeneic bone marrow to purified NK cells, there was a dramatic decrease in the number of CFU-S-derived colonies. Incubation with NK-depleted cells did not result in an inhibition of colony formation by the CFU-S. Mixing experiments showed that the M1/70-labeled NK cells exerted their effect directly on the CFU-S and not on any accessory cells. The effect of the NK cells on colony formation by the CFU-S could be blocked competitively and selectively by the addition, before incubation, of a classic murine NK tumor target, Yac-1. Another tumor line (WTS) that is poorly recognized by NK cells was less effective in blocking the inhibitory effect of NK cells on CFU-S. The demonstration that purified NK cells can selectively inhibit the development of the tripotential CFU-S may point to the importance of NK cells in the regulation of hematopoiesis, in the development of some types of marrow dysfunction, and in the failure of engraftment of transplanted bone marrow.     

Induction of activation antigens on human natural killer cells mediated through the Fc-gamma receptor

NK cells, defined here as lymphocytes bearing the CD16 Ag found on the NK cell Fc-gamma receptor (FcR), are known to enter a proliferative and activated state in response to stimulation with IL-2 as assessed by clonal expansion, short-term DNA synthesis, and de novo expression of lymphocyte-associated activation Ag. We have found that the FcR of NK cells acts as a signaling pathway through which IL-2-dependent activation may be greatly enhanced, allowing for more rapid induction of activation Ag and recruitment of an increased percentage of cells expressing surface markers of cellular activation. FcR-interactive agents, such as solid phase immobilized immune complexes or cross-linked CD16-specific mAb, work synergistically with rIL-2 to elicit a rapid expression of IL2R and transferrin receptors on greater than 50% of purified NK cells as early as day 3 of culture. IL-2 or FcR-interactive stimuli alone were weak or ineffective stimulators by comparison. In contrast to the induction of de novo activation Ag, DNA synthesis was elicited by IL-2 alone, but was not substantially or consistently enhanced by the subsequent addition of FcR-interactive stimuli.     

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.     

The effect of topoisomerase inhibitors on the expression of differentiation markers and cell cycle progression in human K-562 leukemia cells.

Treatment of human K-562-J leukemia cells for 1 h with the topoisomerase II-reactive drugs VP-16, VM-26, or mAMSA resulted in a dose-dependent inhibition of proliferation and in an increase in the percentage of cells staining positive for hemoglobin, a marker of erythroid differentiation. Staining for hemoglobin of up to about 60% of the cells was observed at 20 microM VP-16, 1 microM VM-26, and 8 microM mAMSA. Such treatment also caused a G2/M arrest in the cell cycle. Incubation of the cells with radiolabeled VP-16 indicated that the induced erythroid differentiation was not due to continuous cell exposure to a residual amount of the drug. VP-16-induced erythroid differentiation was also not affected by DNA, RNA, or protein synthesis inhibitors. Differentiation induction and the G2/M arrest evoked by VP-16, VM-26, and mAMSA were, however, reduced in the presence of novobiocin. Our results indicate that topo-reactive drugs that cause G2/M arrest in the K-562-J cell cycle can induce in these cells erythroid differentiation after a short and irreversible interaction with their target molecule(s).     

Human natural killer cells analyzed by B73.1, a monoclonal antibody blocking Fc receptor functions. II. Studies of B73.1 antibody-antigen interaction on the lymphocyte membrane

In this paper, we characterize the antigen recognized by the monoclonal antibody B73.1 and the modification occurring at the membrane of the positive cells after interaction with the antibody. The B73.1-defined antigen is a protein of 50,000 to 72,000 daltons that is sensitive to pronase but not to trypsin treatment. B73.1 antibody, and its F(ab')2 fragment, directly block, at high concentrations, the binding of IgG antibody-sensitized erythrocytes to the Fc receptors (FcR) of a subpopulation of lymphocytes and neutrophils. B73.1 antibody dissociates rapidly from the positive cells, but concomitant modulation of both B73.1 antigen and FcR is induced when cells are incubated in the continuous presence of antibody or when B73.1 antibody is cross-linked at the cell membrane with an anti-mouse immunoglobulin antiserum. Reaction of lymphocytes with immune complexes also induces modulation of both FcR and B73.1 antigen, without affecting the expression of other antigens on the positive cells. The possibility that the antigen is internalized and digested by the cell after reaction with the antibody is discussed. B73.1 antibody inhibits antibody-dependent cytotoxicity mediated by lymphocytes (K cells) and neutrophils, whereas it does not affect spontaneous cytotoxicity of NK cells. These results suggest the B73.1-defined antigen might be the FcR or a structure closely related to it on K/NK cells.     

Studies on the relationship of human natural killer and lymphokine-activated killer cells with lysosomal staining and analysis of surface marker phenotypes

We have previously demonstrated that natural killer (NK) cells are lysosome-rich and stain more intensely with lysosomotropic agents such as neutral red and quinacrine (Qu) than do non-NK cells. In this study we combined the quantitation of Qu staining with surface marker staining to define subpopulations of NK cells. While all NK activity was contained within the Qu+ population, most but not all NK cells expressed the surface marker CD16. A subpopulation of NK cells was found to be Qu+CD16- composed of medium- to large-sized cells with a granular appearance on Giemsa staining. Culture with interleukin-2 (IL-2) induced enhanced cytotoxicity in peripheral blood lymphocytes (PBL) against NK-sensitive and NK-resistant tumor cells. Like NK cells, these lymphokine-activated killer (LAK) cells were predominantly Qu+CD16+. However, some LAK cells were Qu+CD16-. The Qu+CD16+ cells were typical large granular lymphocytes (LGL). The Qu+CD16- cells were also large lymphocytes, more than 50% of which were proliferating. However, the granulation in some Qu+CD16- cells, as detected by Giemsa staining, was more prominent and numerous than others in the same population. No LAK activity was ever detected in Qu- cells, which were uniformly small lymphocytes. Quantitation of Qu staining in effector cells was therefore demonstrated to have a good correlation with NK and LAK functions, and with surface markers can help to characterize both types of cells. Moreover, these results indicate that both NK and LAK populations include a small subset of CD16- cells in each.     

Modification of the target cell surface with lipophilic glycoconjugates and interaction of the modified cells with natural killer cells

An experimental model system involving the modification of carbohydrate composition of the target cell surface with neoglycolipids was developed for studying the role of surface carbohydrates of target cells in the NK-cell-mediated cytotoxicity. The polymeric glycoconjugates of the Glyc-PAA-PEA and Glyc-PAA(Flu)-PEA types (where Glyc was an oligosaccharide residue, PAA poly(acrylamide) polymer, and PEA the phosphatidylethanolamine residue, and Flu fluorescein residue) capable of incorporation into the cell membrane were synthesized. The optimum structures of neoglycoconjugates and conditions for their incorporation into K562 and Raji cell lines, which differ in their sensitivity to the NK-cell-mediated lysis were selected. The mechanism of association of glycoconjugates with the plasma cell membrane and the kinetics of their elimination from the cell surface were investigated using the fluorescent-labeled Glyc-PAA(Flu)-PEA derivatives. The spatial accessibility of the carbohydrate ligands for the interaction with human NK cells was demonstrated. The target cells modified with the Le(x) trisaccharide were shown to be more sensitive to the cytotoxic effect of human NK cells than the intact cells. The English version of the paper: Russian Journal of Bioorganic Chemistry, 2004, vol. 30, no. 3; see also http://www.maik.ru.     

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.     

Adenovirus E1A gene induction of susceptibility to lysis by natural killer cells and activated macrophages in infected rodent cells.

Rodent cells immortalized by the E1A gene of nononcogenic adenoviruses are susceptible to lysis by natural killer (NK) cells and activated macrophages. This cytolysis-susceptible phenotype may contribute to the rejection of adenovirus-transformed cells by immunocompetent animals. Such increased cytolytic susceptibility has also been observed with infected rodent cells. This infection model provided a means to study the role of E1A gene products in induction of cytolytic susceptibility without cell selection during transformation. Deletion mutations outside of the E1A gene had no effect on adenovirus type 2 (Ad2) or Ad5 induction of cytolytic susceptibility in infected hamster cells, while E1A-minus mutant viruses could not induce this phenotype. E1A mutant viruses that induced expression of either E1A 12S or 13S mRNA in infected cells were competent to induce cytolytic susceptibility. Furthermore, there was a correlation between the accumulation of E1A gene products in Ad5-infected cells and the level of susceptibility of such target cells to lysis by NK cells. The results of coinfection studies indicated that the E1A gene products of highly oncogenic Ad12 could not complement the lack of induction of cytolytic susceptibility by E1A-minus Ad5 virus in infected cells and also could not block induction of this infected-cell phenotype by Ad5. These data suggest that expression of the E1A gene of nononcogenic adenoviruses may cause the elimination of infected cells by the immunologically nonspecific host inflammatory cell response prior to cellular transformation. The lack of induction of this cytolysis-susceptible phenotype by Ad12 E1A may result in an increased persistence of Ad12-infected cells in vivo and may lead to an increased Ad12-transformed cell burden for the host.     

Studies on the mechanism of natural killer cell-mediated cytotoxicity. VI. Characterization of human, rat, and murine natural killer cytotoxic factors

Recent evidence has implicated natural killer cytotoxic factors (NKCF) as the lytic mediators of NK cell-mediated cytotoxicity reactions. The objective of this study was to examine and compare some of the biochemical and functional characteristics of human, rat, and murine NKCF. Supernatants containing NKCF were generated by stimulating effector cells with Con A or U937 (for human PBL) or YAC-1 (for rodent spleen cells) and tested for cytotoxic activity in a 20-hour (rodent) or 24-hour (human) 51Cr release assay. NKCF activity was inactivated by heating to 63 degrees C, 8 M urea, pH 2, and reduction and alkylation. These factors were highly sensitive to trypsin, moderately sensitive to papain and resistant to neuraminidase. Adsorption of human NKCF to U937 cells is inhibited by mannose-6-phosphate and adsorption of rodent NKCF to YAC-1 cells is inhibited by alpha-methyl-D-mannoside and fructose-6-phosphate. Oxidation of NKCF with sodium periodate abolished lytic activity. Pretreatment of NKCF with Con A but not pretreatment of target cells inhibited lytic activity. NKCF activity eluted in a single broad band of apparent MW of 15,000-40,000 after fractionation by HPLC gel permeating chromatography. Pooled fractions containing NKCF activity were subjected to some of the same tests performed on whole supernatants. Test result with semipurified NKCF confirmed that these factors are inactivated by trypsin or sodium periodate and that mannose-6-phosphate inhibits their binding to target cells. There were no major differences observed in NKCF produced by the three different species whether stimulated by Con A or NK-sensitive tumor cells. The evidence indicates that NKCF are glycoproteins in which disulfide bonding is essential for lytic activity. Furthermore, it appears that carbohydrate residues expressed on NKCF molecules are involved in the binding of these factors to the target cell membrane.