Membrane-associated Tumor Necrosis Factor Research Articles

Human B Cells Mediate Innate Anti-Cancer Cytotoxicity Through Concurrent Engagement of Multiple TNF Superfamily Ligands.

The essential innate immunity effector cells, natural killer and dendritic cells, express multiple plasma membrane-associated tumor necrosis factor (TNF) superfamily (TNFSF) ligands that, through simultaneous and synergistic engagement, mediate anti-cancer cytotoxicity. Here, we report that circulating B cells, mediators of adaptive humoral immunity, also mediate this innate anti-cancer immune mechanism. We show that resting human B cells isolated from peripheral blood induce apoptosis of, and efficiently kill a large variety of leukemia and solid tumor cell types. Single-cell RNA sequencing analyses indicate, and flow cytometry data confirm that B cells from circulation express transmembrane TNF, Fas ligand (FasL), lymphotoxin (LT) α1β2 and TNF-related apoptosis-inducing ligand (TRAIL). The cytotoxic activity can be inhibited by individual and, especially, combined blockade of the four transmembrane TNFSF ligands. B cells from tumor-bearing head and neck squamous cell carcinoma patients express lower levels of TNFSF ligands and are less cytotoxic than those isolated from healthy individuals. In conclusion, we demonstrate that B cells have the innate capacity to mediate anti-cancer cytotoxicity through concurrent activity of multiple plasma membrane-associated TNFSF ligands, that this mechanism is deficient in cancer patients and that it may be part of a general cancer immunosurveillance mechanism.

Crucial cytokine interactions in nitric oxide production induced by Mycoplasma arthritidis superantigen

Mycoplasma arthritidis causes autoimmune arthritis in rodents. It produces a superantigen (MAM) that simultaneously activates antigen presenting cells and T cells inducing nitric oxide and cytokine release. Nitric oxide is a key inducer and regulator of the immune system activation. Here, we investigated nitric oxide and cytokine production and interactions of these molecules in MAM-stimulated co-cultures of macrophages (J774A.1 cell line) with spleen lymphocytes. We found that: a) MAM-induced nitric oxide, interferon-gamma, membrane-associated tumor necrosis factor and interleukin-2 production in co-cultures of macrophages with lymphocytes from BALB/c and C3H/HePas but not from C57Bl/6 mice; b) production of nitric oxide was dependent on interferon-gamma whereas that of interferon-gamma was dependent on interleukin-2 and membrane-associated tumor necrosis factor; c) these cytokines up regulated MAM-induced nitric oxide production. Unraveling the mechanisms of cell activation induced by MAM might be helpful to design strategies to prevent immune system activation by superantigens and therefore in seeking amelioration of associated immunopathologies.

Simultaneous Induction of Apoptotic and Survival Signaling Pathways in Macrophage-Like THP-1 Cells by Shiga Toxin 1

Shiga toxins have been shown to induce apoptosis in many cell types. However, Shiga toxin 1 (Stx1) induced only limited apoptosis of macrophage-like THP-1 cells in vitro. The mechanisms regulating macrophage death or survival following toxin challenge are unknown. Differentiated THP-1 cells expressed tumor necrosis factor receptors and membrane-associated tumor necrosis factor alpha (TNF-alpha) and produced soluble TNF-alpha after exposure to Stx1. However, the cells were refractory to apoptosis induced by TNF-alpha, although the cytokine modestly increased apoptosis in the presence of Stx1. Despite the partial resistance of macrophage-like THP-1 cells to Stx1-mediated killing, treatment of these cells with Stx1 activated a broad array of caspases, disrupted the mitochondrial membrane potential (DeltaPsi(m)), and released cytochrome c into the cytoplasm. The DeltaPsi(m) values were greatest in cells that had detached from plastic surfaces. Specific caspase inhibitors revealed that caspase-3, caspase-6, caspase-8, and caspase-9 were primarily involved in apoptosis induction. The antiapoptotic factors involved in macrophage survival following toxin challenge include inhibitors of apoptosis proteins and X-linked inhibitor of apoptosis protein. NF-kappaB and JNK mitogen-activated protein kinases (MAPKs) appeared to activate survival pathways, while p38 MAPK was involved in proapoptotic signaling. The JNK and p38 MAPKs were shown to be upstream signaling pathways which may regulate caspase activation. Finally, the protein synthesis inhibitors Stx1 and anisomycin triggered limited apoptosis and prolonged JNK and p38 MAPK activation, while macrophage-like cells treated with cycloheximide remained viable and showed transient activation of MAPKs. Collectively, these data suggest that Stx1 activates both apoptotic and cell survival signaling pathways in macrophage-like THP-1 cells.

Systemic sclerosis Th2 cells inhibit collagen production by dermal fibroblasts via membrane-associated tumor necrosis factor alpha.

In systemic sclerosis (SSc; scleroderma), T cells infiltrate organs undergoing fibrotic changes and may participate in dysregulated production of collagen by fibroblasts. The objective of this study was to functionally characterize T cells infiltrating skin lesions in early SSc and investigate their capacity to affect production of type I collagen and interstitial collagenase (matrix metalloproteinase 1 [MMP-1]) by dermal fibroblasts. Four-color cytometric analysis was used to characterize subset distribution and production of interferon-gamma (IFN gamma) and interleukin-4 (IL-4) in T cell lines generated from the skin of patients with SSc. T cell clones were generated, and their capacity to modulate collagen and MMP-1 production by fibroblasts derived from patients with SSc and from normal individuals was assessed. Neutralizing reagents were used to identify T cell mediators involved in fibroblast modulation. The skin of individuals with early-stage SSc contained T cells preferentially producing high levels of IL-4. Cloned CD4+ Th2-like cells inhibited collagen production by normal fibroblasts. Th2 cell-dependent inhibition was, at least in part, contact-dependent, was essentially mediated by tumor necrosis factor alpha (TNF alpha), and was dominant over the enhancement induced by profibrotic IL-4 and transforming growth factor beta cytokines. The simultaneous induction of MMP-1 production confirmed the specificity of these observations. To be inhibitory, Th2 cells required activation by CD3 ligation. Th2 cells were less potent than were Th1 cells in inhibiting collagen production by normal fibroblasts via cell-to-cell interaction, and SSc fibroblasts were resistant to inhibition. These findings indicate that, despite their production of IL-4, Th2 cells reduce type I collagen synthesis by dermal fibroblasts because of the dominant effect of TNF alpha, and suggest that strategies based on TNF alpha blockade aimed at controlling fibrosis in SSc may be unwise.

Alterations in immunocyte tumor necrosis factor receptor and apoptosis in patients with congestive heart failure.

To examine systemic immune cell proinflammatory receptor expression and apoptosis in patients with congestive heart failure (CHF). Prior studies have demonstrated that CHF is associated with a chronic myocardial inflammatory state, including increased plasma proinflammatory cytokine and soluble receptor expression. By contrast, it has also been shown that tumor necrosis factor (TNF) receptor protein expression is decreased in the failing myocardium. However, no studies to date have examined systemic immune cell proinflammatory receptor expression or function as disease markers in patients with heart failure. Twenty-nine patients were studied prospectively over an 8-month period at a single institution. One group (n = 16) had a history of clinical symptoms of CHF and moderate to severe left ventricular dysfunction. The second group (n = 13) consisted of patients who had coronary artery disease without symptoms of CHF and documented preservation of left ventricular function. Blood samples were analyzed for polymorphonuclear cell (PMN) and monocyte TNF and CD95 membrane-associated receptor expression, spontaneous and CD95 (Fas)-mediated PMN apoptosis, and plasma cytokine and soluble TNF receptor levels. Isolated PMNs were incubated for 6 hours with or without CH 11, a CD95 agonist. Propidium iodide/RNAase staining and flow cytometry was used to assess apoptosis, defined as PMNs expressing hypodiploid DNA (<2 n DNA). Membrane-associated TNF receptor and CD95 were also measured by flow cytometry. Plasma levels of TNF, interleukin (IL)-6, IL-10, and soluble TNF receptors 1 and 2 were quantified using enzyme-linked immunosorbent assay. Compared to patients without CHF, circulating PMN and monocyte TNF receptor levels were significantly decreased in patients with CHF. By contrast, PMN and monocyte CD95 expression was not significantly changed in patients with CHF versus those without CHF. Patients with CHF had a 60% decrease in spontaneous PMN apoptosis compared to patients without CHF, whereas no significant difference in CD95-mediated apoptosis was observed between the two groups. Pearson-product movement correlation of monocyte TNF receptor expression and spontaneous PMN apoptosis rates versus patients' ejection fraction was performed and was statistically significant. Plasma levels of soluble TNF receptor 2 (p75) were elevated in CHF patients versus patients without CHF, while there was no significant difference in soluble TNF receptor 1 (p55), TNF, IL-6, and IL-10 between the two groups. These data demonstrate a systemic alteration in immune cell phenotype and apoptosis in patients with CHF. These findings provide support for the concept that inflammatory mediators either contribute to myocardial dysfunction or are elaborated systemically by left ventricular compromise. This present study suggests that immune cell TNF receptor expression and diminished PMN apoptosis may serve as biologic markers of myocardial failure.

A novel molecular mechanism modulating osteoclast differentiation and function

Osteoclasts, the multinucleated giant cells that resorb bone, develop from hematopoietic cells of the monocyte/macrophage lineage. Osteoblasts, as well as bone marrow stromal cells, support osteoclast development through a mechanism of cell-to-cell interaction with osteoclast progenitors. We recently purified and molecularly cloned osteoclastogenesis inhibitory factor (OCIF), which was identical to osteoprotegerin (OPG). OPG/OCIF, a secreted member of the tumor necrosis factor (TNF) receptor family, inhibited differentiation and activation of osteoclasts. A single class of high-affinity binding sites for OPG/OCIF appeared on a mouse bone marrow stromal cell line, ST2, in response to 1α,25-dihydroxyvitamin D3 [1,25(OH) 2D 3] and dexamethasone (Dex). When the binding sites were occupied by OPG/OCIF, ST2 cells failed to support the osteoclast formation from spleen cells. To identify an OPG/OCIF ligand, we screened a cDNA expression library of ST2 cells treated with 1,25(OH) 2D 3 and Dex using OPG/OCIF as a probe. The cloned molecule was found to be a member of the membrane-associated TNF ligand family, and it induced osteoclast formation from mouse and human osteoclast progenitors in the presence of macrophage colony-stimulating factor (M-CSF) in vitro. Expression of its gene in osteoblasts/stromal cells was upregulated by osteotropic factors, such as 1,25(OH) 2D 3, prostaglandin E 2 (PGE 2), parathyroid hormone (PTH), and interleukin (IL)-11. A polyclonal antibody against this protein, as well as OPG/OCIF, negated not only the osteoclastogenesis induced by the protein, but also bone resorption elicited by various osteotropic factors in a fetal mouse long bone culture system. These findings led us to conclude that the protein is osteoclast differentiation factor (ODF), a long sought-after ligand that mediates an essential signal to osteoclast progenitors for their differentiation into active osteoclasts. Recent analyses of ODF receptor demonstrated that RANK, a member of the TNF receptor family, is the signaling receptor for ODF in osteoclastogenesis, and that OPG/OCIF acts as a decoy receptor for ODF to compete against RANK. The discovery of ODF, OPG/OCIF, and RANK opens a new era in the investigation of the regulation of osteoclast differentiation and function.

Cloning and characterization of the gene encoding mouse osteoclast differentiation factor.

Osteoclast differentiation factor (ODF), a ligand for osteoclastogenesis inhibitory factor (OCIF)/ osteoprotegerin (OPG), is a member of the membrane-associated tumor necrosis factor (TNF) family and induces osteoclast-like cell formation in vitro. In the present study, mouse ODF genomic clones were isolated and sequenced to determine their gene structure. The mouse ODF gene is a single copy gene consisting of five exons and spans approximately 40kb of the mouse genome. The first exon encodes the intracellular and transmembrane domains. The extracellular region of ODF containing the TNF homologous domain is encoded by exons 1 through 5. The translation-termination codon and six polyadenylation signal residues are present in exon 5. A major transcription-initiation site is present 143 nucleotides upstream of the initiation-ATG codon. This genomic organization is similar to that of other members of the TNF family, especially the CD40 ligand.

Expression of the 75 kDA TNF receptor and its role in contact-mediated neuronal cell death

We previously demonstrated TNF toxicity, at high TNF doses or in the presence of actinomycin D, in the N1E-115 neuronal cell line (N1Es), which expresses only the 55 kDa TNF receptor (TNFR). To determine whether presence of the 75 kDa TNFR increases N1E sensitivity to TNF toxicity, cells were transfected with a 75 kDa TNFR expression construct. However, 75 kDa TNFR protein expression was undetectable in stably transfected N1Es. Further investigation revealed endogenous membrane-associated TNF in this neuronal line. Co-transfection with β-galactosidase and the 75 kDa TNFR or empty vector (pcDNA3) indicated cell loss in the 75 kDa TNFR-transfected population relative to vector-transfected populations, while inhibition of membrane-associated TNF with a neutralizing antibody led to increased 75 kDa TNFR expression in transiently transfected N1Es. We conclude that neutralization of membrane-associated TNF inhibits its interaction with the introduced 75 kDa TNFR, increasing neuronal survival and promoting 75 kDa TNFR expression. Induced 75 kDa TNFR expression in the presence of membrane-associated TNF and the 55 kDa TNFR results in lymphocyte cell death [J.K. Lazdins, M. Grell, M.R. Walker, K. Woods-Cook, P. Scheurich, K. Pfizenmaier, Membrane tumor necrosis factor (TNF) induced cooperative signaling of the TNFR60 and TNFR80 favors induction of cell death rather than virus production in HIV-infected T cells, J. Exp. Med. 185 (1997) 81–90]. This report demonstrates that membrane-associated TNF and the 75 kDa TNFR similarly contribute to neuronal cell death.

Osteoclast differentiation factor is a ligand for osteoprotegerin/osteoclastogenesis-inhibitory factor and is identical to TRANCE/RANKL.

Osteoclasts, the multinucleated cells that resorb bone, develop from hematopoietic cells of monocyte/macrophage lineage. Osteoclast-like cells (OCLs) are formed by coculturing spleen cells with osteoblasts or bone marrow stromal cells in the presence of bone-resorbing factors. The cell-to-cell interaction between osteoblasts/stromal cells and osteoclast progenitors is essential for OCL formation. Recently, we purified and molecularly cloned osteoclastogenesis-inhibitory factor (OCIF), which was identical to osteoprotegerin (OPG). OPG/OCIF is a secreted member of the tumor necrosis factor receptor family and inhibits osteoclastogenesis by interrupting the cell-to-cell interaction. Here we report the expression cloning of a ligand for OPG/OCIF from a complementary DNA library of mouse stromal cells. The protein was found to be a member of the membrane-associated tumor necrosis factor ligand family and induced OCL formation from osteoclast progenitors. A genetically engineered soluble form containing the extracellular domain of the protein induced OCL formation from spleen cells in the absence of osteoblasts/stromal cells. OPG/OCIF abolished the OCL formation induced by the protein. Expression of its gene in osteoblasts/stromal cells was up-regulated by bone-resorbing factors. We conclude that the membrane-bound protein is osteoclast differentiation factor (ODF), a long-sought ligand mediating an essential signal to osteoclast progenitors for their differentiation into osteoclasts. ODF was found to be identical to TRANCE/RANKL, which enhances T-cell growth and dendritic-cell function. ODF seems to be an important regulator in not only osteoclastogenesis but also immune system.

Direct cell/cell contact with stimulated T lymphocytes induces the expression of cell adhesion molecules and cytokines by human brain microvascular endothelial cells.

Upon inflammation, stimulated, but not resting T lymphocytes cross the blood-brain barrier and migrate into the central nervous system. This study shows that direct contact between stimulated T lymphocytes and human brain microvascular endothelial cells (HB-MVEC) induces phenotypic and functional changes on the latter cells. Plasma membranes isolated from stimulated T lymphocytes (S-PM) up-regulated the expression of intercellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), and E-selectin on isolated HB-MVEC. In addition, HB-MVEC activated by S-PM secreted interleukin (IL)-6 and IL-8. The levels of ICAM-1, E-selectin, IL-6, and IL-8 expressed in S-PM-activated HB-MVEC were similar to those observed with 1000 U/ml tumor necrosis factor (TNF). In contrast, VCAM-1 expression was 15% of that induced by TNF. Inhibitors of TNF diminished (< or = 45%), but did not abolish the expression of cell adhesion molecules and IL-6 induced by S-PM, IL-8 production being insignificantly affected (< or = 10%). This suggests that membrane-associated TNF was partially involved in HB-MVEC activation. The present study demonstrates that stimulated T lymphocytes are able to activate HB-MVEC upon direct cell contact. This novel mechanism of inducing the expression of cell adhesion molecules may prompt the initial adhesion of stimulated T lymphocytes to brain endothelium.

Parenteral Nutrition Alters Monocyte TNF Receptor Activity

The route of nutrient provision has been reported to influence some aspects of the host inflammatory response in both patient populations and normal subjects. The tumor necrosis factor receptor system is a complex regulatory mechanism that modulates the bioavailability of tumor necrosis factor (TNF). We sought to determine whether maintenance on total parenteral nutrition (TFN) can alter host response to endotoxin challenge, specifically as it relates to the TNF receptor system. Seventeen healthy men were randomized to receive either TPN ( n = 8) or a defined formula enteral diet (ENT, n = 9) prior to intravenous infusion of endotoxin (Lot EC-5, 20 U/kg). The subjects that received 1 week of antecedent TPN exhibited an increased heart rate and temperature and decreased mean arterial pressure post-LPS compared to those subjects maintained on enteral nutritional support. The TPN subjects also exhibited comparatively higher TNF and interleukin-6 levels in response to endotoxin. Monocyte TNF receptor levels decreased in both groups post-LPS, but TPN subjects exhibited consistently greater expression of this functional membrane-associated TNF receptor. After LPS, soluble tumor necrosis factor receptor II (sTNFr II, p75) peaked three times higher in TPN subjects than in ENT subjects. Conversely, sTNFr I (p55) was higher in the enterally fed group. From these studies it appears that antecedent TPN not only influences clinical manifestations of endotoxin but also modulates the regulation of all associated TNF receptors and shedding of soluble receptors.

Leukemic cell lysis by activated human macrophages: significance of membrane-associated tumor necrosis factor.

In this study, we analyzed the mechanism(s) of leukemic cell lysis by human macrophages. Peripheral blood monocyte‐derived macrophages were activated with recombinant interferon‐γ and lipopoly‐saccharide and their lytic activity against two leukemic cell lines (K562 and HL‐60 cells) was assessed by an 111In releasing assay. Activated macrophages lysed these leukemic cells, and the lytic activity against leukemic cells was almost completely inhibited by anti‐tumor necrosis factor (TNF) antibody. The macrophage‐lysate prepared from activated macrophages also exhibited significant lytic activity against leukemic cells; this lytic activity was inhibited by anti‐TNF antibody. The leukemic cells that we used for the cytotoxicity assays were resistant to recombinant TNF. The culture supernatant of activated macrophages did not show any lytic activity. These findings suggest that cell‐associated TNF plays a role in macrophage‐mediated cytotoxicity against leukemic cells.

Cytolytic activities of activated macrophages versus paraformaldehyde-fixed macrophages; soluble versus membrane-associated TNF

Membrane-associated tumor necrosis factor (TNF) and soluble TNF were compared as to their lytic activities, and as to the kinetics of their expression by macrophages activated with LPS and/or IFN-γ in the presence or absence of cycloheximide. EL 4 tumor cells, resistant and sensitive to lysis by recombinant TNF or membrane-associated TNF (paraformaldehyde (PF) -fixed activated macrophages) were used as targets. In the presence of cycloheximide the TNF-resistant S-EL4 cells were lysed by both TNFs. PF-fixed macrophages was cytolytic after 1 hr activation but not after 3 or more hours of activation. Their activity was totally inhibited by anti-TNF antibodies and was a composite of transmembrane (integral) TNF and soluble TNF conjugated to macrophage membrane TNF receptors. Treatment of the macrophages with glycine pH 3.0 buffer dissociated the conjugated TNF without affecting the integral membrane TNF. When macrophages were activated with LPS ± IFN-γ in the presence of cycloheximide or activated just with IFN-γ their activity after fixation with paraformaldehyde was no longer detected. Nonfixed macrophages under these conditions still remained cytotoxic. Tumor cell susceptibility to membrane-associated TNF activity, in contrast to recombinant (soluble) TNF, was greatly reduced in the presence of nicotinamide, an inhibitor of ADP-ribosyltransferase, suggesting that the mechanisms of lysis by these TNFs may be different. The lytic activity of both TNFs was found to be receptor-dependent in that tumor cells, whose TNF binding sites were “down-regulated” by TPA, were rendered resistant to lysis by both membrane-associated and soluble TNFs.

Antileishmanial defense in macrophages triggered by tumor necrosis factor expressed on CD4+ T lymphocyte plasma membrane.

In our studies of host defense to the intracellular protozoan Leishmania major, we uncovered a novel mechanism of antileishmanial defense that involves direct cell contact between effector CD4+ lymphocytes and Leishmania-infected macrophages. The mechanism is distinctive because it does not involve lymphokine secretion and induces no cytotoxic effects in the host cells; its expression is antigen-specific and genetically restricted. We now demonstrate that these effector CD4+ cells display tumor necrosis factor (TNF) on their surface and provide evidence that the membrane-associated TNF is involved in the activation of the antileishmanial defense. Using a Leishmania-specific cloned T-T cell hybridoma line (1B6; CD4+, T helper type 1) that activates antileishmanial defense in macrophages through cell contact and does not secrete TNF, we noted that only cells bearing surface TNF (TNF+), but not ones lacking surface TNF (TNF-), exerted these effects. Moreover, the antileishmanial effects excreted by TNF+ 1B6 cells as well as by lymph node CD4+ TNF+ lymphocytes could be blocked with anti-TNF antibody. We propose that membrane-associated TNF on CD4+ T cells may provide a mechanism of targeting activation signals to macrophages in an antigen-specific and genetically restricted manner.

A flow cytometric and immunofluorescence microscopic study of tumor necrosis factor production and localization in human monocytes

The production and localization of tumor necrosis factor (TNF) in human monocytes were investigated by using monoclonal and polyclonal antibodies against recombinant human TNF together with flow cytometry and immunofluorescence microscopy. Lipopolysaccharide (LPS) induced a rapid and transient accumulation of TNF in perinuclear vesicles which was detected 20 min after the addition of LPS. The fluorescence intensity of the vesicles peaked at 40 min of LPS exposure, concomitantly with the release of TNF into the medium. Thus, our results indicate that the secretion of TNF is typical for secretory proteins as it involves passage through the secretory apparatus. Additional studies demonstrated that plasma membrane-associated TNF could not be detected in live monocytes not exposed to LPS. However, after 90 min with LPS, a small population of monocytes expressed membrane-associated TNF, and by 24 hr approximately 50% of the monocytes displayed TNF on the plasma membrane. Furthermore, our results indicate that plasma membrane-associated TNF does not represent released TNF bound back to its own receptor. Thus, our findings support the view that TNF exists as a surface transmembrane protein in LPS-stimulated monocytes.

Plasma membrane-associated tumor necrosis factor. A non-integral membrane protein possibly bound to its own receptor.

Purified plasma membranes from LPS-activated human blood monocytes produced significant lysis and growth inhibition of the TNF-sensitive L929 murine fibroblast cell line. Unactivated human monocyte plasma membranes did not display either activity. Anti-TNF serum specifically inhibited the anti-tumor activity of activated monocyte membranes whereas anti-IL-1 serum or non-specific rabbit serum decreased neither the lysis nor growth inhibition of L929 cells. Membrane-associated TNF did not behave as an integral protein as it could be eluted from the plasma membranes by either high salt or low pH treatment. Plasma membranes cleared of membrane-associated TNF by high salt treatment were able to bind TNF, and this binding was specifically inhibited by preincubation of rTNF with specific anti-TNF serum. Western blot analysis of plasma membranes showed a membrane-associated TNF with a m. w. of approximately 17 kDa present only in the activated monocytes. When the plasma membranes were preincubated with the cross-linker agent dissuccinimidyl suberate, Western blot analysis revealed the presence of a TNF-binding protein with a Mr of approximately 102 kDa. These studies indicate that unlike IL-1, membrane-associated TNF is not an integral membrane protein and that TNF may be associated with the monocyte membrane by occupying the TNF R.

Cell-associated tumor necrosis factor (TNF) as a killing mechanism of activated cytotoxic macrophages.

Different macrophage populations were investigated for their abilities to secrete tumor necrosis factor (TNF) and to lyse TNF-susceptible tumor cells. In this way we could demonstrate that TNF-secretion, although a feature of all activated macrophage populations, is no absolute requirement for the killing of the TNF-sensitive Wehi 164 target. Macrophage cytotoxicity against this cell but not against the TNF-resistant P815 mastocytoma, was completely inhibitable by a specific anti-TNF serum also in the absence of measurable secreted TNF. Moreover the TNF-dependent lysis of tumor cells could also be performed by activated macrophages that had been fixed with paraformaldehyde before the addition of the target cells. In the indirect radioimmunoassay, TNF could be demonstrated on the surface of fixed effector cells. Our results must be interpreted in terms of membrane-associated TNF as the lytic principle for TNF-susceptible tumor cells.