Expression of the types A and B tumor necrosis factor (TNF) receptors is independently regulated, and both receptors mediate activation of the transcription factor NF-kappa B. TNF alpha is not needed for induction of a biological effect via TNF receptors.

The expression and biological function of the types A and B tumor necrosis factor (TNF) receptors were studied using three cell types. SW480T, HEp2, and HL60 cells had, respectively, mainly the type A, only the type B, and roughly similar amounts of both receptors. Dibutyric cAMP treatment induced a 3-6-fold increase in the amount of the type A receptor in HL60 cells without affecting the amount of the type B receptor. Expression of both receptors can thus be regulated independently. HEp2 and human umbilical vein endothelial cells only showed the type B receptor, and expression of the type A receptor could not be induced in these cells. HL60 cells showed, upon Scatchard analysis, a single binding site for TNF alpha, and its Kd may correspond to that of the type A receptor. The approximately 7-fold lower affinity of TNF alpha binding to the type B receptor of HL60 cells was only detected after blocking all TNF alpha binding to the type A receptor. Both the types A and B receptors mediated TNF alpha-induced activation of the transcription factor NF-kappa B. The agonistic antibody htr9 to the type B receptor also activated NF-kappa B. Thus, signal transduction via the type B receptor may only require interaction with the receptor's extracellular domain.

Tumor necrosis factor 01 (TNFa)' is a protein released by activated macrophages in response to external stimuli. TNFol was originally characterized as a protein that induces necrosis in certain tumors (Carswell et al., 1975). Later, it was recognized to be a cytokine, with diverse effects on different tissues and cells in uiuo as well as in vitro (for review, see Beutler and Cerami, 1986). Among these biological effects are (a) the activation of the endothelium, which leads to release of platelet-derived growth factor (Hajjar et al., 1987), to changes in the endothelial morphology and cellular organization (Stolpen et al., 1986), and to adhesion of, e.g. neutrophils, monocytes, and lymphocytes (Bevilacqua et al., 1985;Gamble et al., 1985); (b) the stimulation of human B-and T-cells, if TNFa is combined with other mitogens (Kehrl et al., 1987;Kuhweide et al., 1990); and (c) other immunomodulatory properties (Talmadge et al., 1988). The pleiotropic biological effects of TNFL~ are frequently enhanced synergistically by IFNr (Lewis et al., 1987) or, in the case of hemorrhagic necrosis and lethal shock, by bacterial lipopolysaccharides (Rothstein and Schreiber, 1988). TNFL~ is implicated in a wide variety of diseases (for review, see Klausner, 1987). It may be effective against certain cancers, plays a role in septic shock and cachexia (for reviews, see Beutler and Cerami, 1987;Oliff, 1988), is a mediator of inflammation and of various immunological reactions (for reviews, see Beutler and Cerami, 1986;Matthews and Neale, 1987;Old, 1988), and also seems to be important in some of the pathological effects seen in malaria (Clark, 1987;Grau et al., 1987;Bate et al., 1988).
A growing number of effects of TNFa are being characterized on the molecular level. TNFcv inhibits lipoprotein lipase in adipocytes (Torti et al., 1985), induces expression of class I major histocompatibility antigens (Collins et al., 1986), increases expression of epidermal growth factor receptors (Palombella et al., 1987), and activation of the c-myc oncogene (Kirstein and Baglioni, 1988). It also induces synthesis of inhibitors of plasminogen activator (Medcalf et al., 1988), superoxide dismutase (Wong and Goeddel, 1988), and the production of other cytokines such as interleukin-1, macrophage colony-stimulating factor, and granulocyte-macrophage colony-stimulating factor (Kaushansky et al., 1988). TNFcv (and interleukin-1) also strongly induce the expression of the vascular cell adhesion molecules, intercellular adhesion molecule-l Dustin and Springer, 1988), endothelial leukocyte adhesion molecule 1 (Bevilacqua et al., 1989), and vascular cell adhesion molecule-l (Osborn et al., 1989a), and also of the secreted platelet-activating factor (Bussolino et al., 1988), of a neutrophil chemotactic factor (Strieter et al., 1989), and of a macrophage-specific chemotaxin (Dixit et al., 1990). The last molecules may promote the local recruitment of cells such as neutrophils, leukocytes, monocytes, and macrophages, whereas the vascular cell adhesion molecules may initiate the tissue infiltration of these cells from the blood vessels. Recently, it was shown that incubation of cells with TNFol or interleukin-1 activated the transcription factor NF-KB (Osborn et al., 1989b;Lowenthal et al., 1989). The expression of at least some of the genes mentioned above may be mediated directly by NF-KB (for review, see Lenardo and Baltimore, 1989), but the importance of NF-KB in these processes is as yet unclear.
The majority of human cells have specific high affinity binding sites for TNFcv (Ku11 et al., 1985;Scheurich et al., 1986). We showed previously that at least two different TNFo( receptor proteins exist. The type A TNF receptor was characterized by a lOO-kDa cross-linked product between TNFa and the receptor protein. The type B receptor showed a major cross-linked product of 75 kDa and a minor product of 95 kDa (Hohmann et al., 1989). The 75-kDa product may contain one molecule of TNFa whereas the 95-kDa band contains two TNFa molecules (Smith and Baglioni, 1989). The type A receptor was the major TNF receptor on myeloid cells, of which at least HL60 cells contained some type B receptor in addition to the type A receptor. The type B receptor was found on several epithelial cell lines such as HEp2, MCF7, and HeLa cells. These cells did not contain any detectable type A receptor (Hohmann et al., 1989). The characterization of two receptor types was later confirmed and extended by Brockhaus et al. (1990) using receptor type-specific monoclonal antibodies.
The types A and B TNF receptors were shown to be proteins with apparent molecular masses of 75 and 55 kDa, respectively. Recently, two groups independently reported the cloning of the same TNF receptor gene corresponding to the type B receptor (Lotscher et al., 1990;Schall et al., 1990).
TNF receptors mediate the cellular effects of TNFa. A decrease in the number of cell surface receptors correlated with a decreased sensitivity to the cytolytic effect of TNFa in certain cells (Lehmann and Droge, 1986;Holtmann and Wallach, 1987), and half-maximal release of interleukin-1 by endothelial cells was seen with the same concentration of TNFn which showed half-maximal binding to the cells (Locksley et al., 1987). Recently, Espevik et al. (1990) reported the induction of several biological effects normally induced by TNFn by incubation of cells with an agonistic antibody to the type B TNF receptor.
In this paper, we determined the ratios of the types A and B TNF receptors on different cells and show that the expression of both receptor types can be regulated independently and that both receptors are biologically active.  (Jaffe et al., 1973), and cells derived from a single donor were analyzed in each experiment. The adherent HEp2 and SW480T cells were grown to near confluence before stimulation. The nonadherent HL60 cells were stimulated after reaching a density of l-1.5 x lo6 cells/ml. Cells were incubated with either recombinant human IFNr (Fountoulakis et al.. 1989) or TNFa (Hohmann et al., 1989). Both.cytokines were purified from Es&e&Z& coli cells as described.

MATERIALS
The concentrations used here refer to monomeric 1FN-y and TNFa.
The Co.) as described previously (Hohmann et al., 1989). After cross-linking, adherent cells were released by treatment with SDS-containing gel sample buffer. For Scatchard analysis, HL60 cells and HEp2 cells that were detached from the culture flask by treatment with 1 mM EDTA were incubated with the antibodies as described above. Instead of htr9 we used monoclonal antibody htr5, which is also specific for type B TNF receptor (Brockhaus et ai, 1990). Binding of I*?-TNFa and Scatchard analvsis were nerformed as described (Hohmann et al., 1989). To test for effects of dibutyric CAMP on the number of TNFa binding sites, HL60, HEpP, and HUVEC cells were incubated for 16 h with 1 mM dibutyric CAMP as described (Scheurich et aZ., 1989 with TNFcv or antibodies as indicated in the figure legends. HEpP or SW480T cells were stimulated after reaching near confluence in 75-cm* culture flasks. Nuclear extracts were prepared essentially as described (Dignam et al., 1983). HL60 cells were harvested by centrifugation of 5 min at 1,500 x g. Adherent HEp2 cells were released from the tissue culture flask by incubation with 1 mM EDTA, cells were then harvested by centrifugation, and washed once with phosphate-buffered saline. About 2 x lo7 cells were resuspended with 500 ~1 of a hypotonic lysis buffer (buffer A, Dignam et al., 1983). After 20 min, the cells were homogenized by 20 strokes with a loose fitting Dounce homogenizer. Nuclei were collected by centrifugation for 4 min at 6,500 rpm in a microcentrifuge (approximately 4,000 X g), and the proteins were extracted with 4 packed pellet volumes of high salt buffer (buffer C, Dignam et al., 1983). After 60 min, the samples were centrifuged as described above. The high salt extracts were diluted with 3 volumes of low salt buffer D (Dignam et al., 1983) containing 1% Nonidet P-40 and were used immediately for electrophoretic mobility shift assays using 4% polyacrylamide gels or kept frozen at -20 "C. Electrophoretic-mobility shift assays were performed as described (Sen and Baltimore, 1986). About 5,000 cpm of 32P end-labeled DdeI/ZZaeIII fragment of the K-light chain enhancer containing the NF-KB binding site were used per assay. A restriction fragment mutated in the NF-KB binding site, but otherwise identical to the wild-type fragment (Lenardo et al., 1987), was used to identify other possibly nonspecifically binding proteins. linking to the type A receptor of untreated HL60 cells was lanes 6 and 11) as reported earlier (Hohmann et al., 1989) and consistently found after htr9 treatment, but the explanation were also seen as minor products with SW480T cells (Fig. for this phenomenon is at present unclear. Pretreatment of MI). Binding and cross-linking of radioiodinated TNFa to all SW480T and HEp2 cells with utrl or htr9, respectively, cells were inhibited by more than 98% if cells were preincu-almost quantitatively inhibited the binding and cross-linking bated with a 300-fold molar excess of unlabeled TNFa (Fig. of TNFa to these cells (Fig. lA, lanes 3 and 9), indicating lA, lanes 2, 7, and 12). Preincubation of cells with the mono-that these cells contain mainly the type A (SW480T) or the clonal antibody utrl, directed against the type A TNF receptor type B (HEp2) receptor. As shown above, SW480T cells , inhibited the formation of the lOO-contain a small amount of the type B receptor. HL60 cells kDa cross-linked products on SW480T and HL60 cells (Fig. pretreated with utrl or htr9, respectively, showed in each case LA, lanes 3 and 13), but the 95 and 75kDa products on HEpP about 50% of the amount of specifically bound radiolabeled and HL60 (Fig. lA, lanes 8 and 13) and SW480T cells (Fig. TNFa when compared with untreated cells (Fig. lA, lanes 2 1, lB, lane 3) were still detected. Preincubation of cells with the 13, and 14; Fig. 2). Only incubation with a mixture of the utrl monoclonal antibody htr9, which is specific for the type B and htr9 antibodies completely abolished the binding of TNFn receptor (Hohmann et al., 1989;Brockhaus et al., 1990), to HL60 cells. HL60 cells may thus contain similar amounts completely inhibited the formation of the 95-and 75-kDa of the types A and B receptors, and the amount of the type A products on HEp2 (Fig. lA, lane 9), HL60 (Fig. lA, lane 14), receptor is therefore overestimated in cross-linking experiand SW480T cells (Fig. lB) Panel A, SW480T cells (lanes I-5), HEp2 cells (klnes 6-IO), and HL60 cells (lanes 11-15) were grown, and aliquots of each cell type were preincubated for 1 h at 0 "C either without any addition (lanes I, 6, and II), with TNFol (lanes 2, 7, and 12), or with the anti-TNF receptor antibodies (Brockhaus et al., 1990) utrl (lanes 3, 8, and 13), htr9 (lanes 4, 9, and 14), or with a mixture of both antibodies (lanes 5, 10, and 15). Then saturating amounts of '*"I-TNFa were added, and incubation was continued for 2 h at 0 "C, followed by cross-linking receptor-ligand complexes.

Three
SDS extracts of cells were analyzed by SDS-PAGE and autoradi-  Scheurich et al. (1989) showed that dibutyric CAMP reversibly enhanced the number of TNF receptors in HL60 cells. Treatment of HL60 cells with dibutyric CAMP resulted in a 3-fold increase in the total number of TNF receptors (Fig. 2, lanes 1 and 6) and in a drastic increase in the amount of the lOO-kDa cross-linked product. This band represented the type A receptor since its formation was blocked by pretreatment with utrl (Fig. 2, compare lanes 6 and 8) but not by htr9 incubation (Fig. 2,  compare lanes 6 and 9). The amount of the type B receptor was not significantly altered by the dibutyric CAMP treatment, as visualized after competition with utrl (Fig. 2, compare lanes 3 and 8). Again the formation of all cross-linked products was quantitatively inhibited by treatment with utrl plus htr9 (Fig. 2, lane 10). The increase in the total number of TNFcv binding sites after treatment with dibutyric CAMP is thus the result of an increased amount of the type A receptor. Binding of TNFol to the type A receptor in dibutyric CAMP-treated cells was also not detectably decreased by incubation with htr9, in contrast to in the untreated HL60 cells (Fig. 2, lanes 4 and 9, short exposure).
To verify the result described above, TNF receptors were immunoprecipitated by binding monoclonal antireceptor antibodies to intact cells cultivated in the absence or presence of dibutyric CAMP, washing the unbound antibody away and collecting the immunocomplexes after cell lysis. This procedure allowed immunoprecipitation of only those receptors that are exposed at the cell surface. TNF receptors were then subsequently visualized in ligand blots using radiolabeled TNFa. A drastically increased amount of the type A receptor was seen in the utrl immunoprecipitates of dibutyric CAMPtreated cells (Fig. 3A, compare lanes 2 and 5, short exposure).
The amount of the type B receptor was unaffected by the dibutyric CAMP treatment (Fig. 3A, compare lanes 3 and 6, long exposure). Multiple bands were seen in the utrl immunoprecipitates.
These bands are all derived from the type A receptor, as confirmed in Fig. 3B. Utrl immunoprecipitates were used for ligand blotting with radiolabeled TNFa, and incubation with the radiolabeled TNFa was done in the absence of any competitor (Fig. 3B, lane I), in the presence of excess htr9 (Fig. 3B, lane 2), or utrl antibody (Fig. 3B, lane  3), or unlabeled TNFa (Fig. 3B, lane 4). The fastest migrating band of about 50 kDa is likely to be a proteolytic degradation product (Hohmann et al., 1989;Brockhaus et al., 1990), the band of about 75 kDa represents the intact type A receptor, and the bands with higher molecular masses represent aggregates since they disappear upon extensive reduction (not shown). Thus, the expression of both receptor types can be regulated independently.
Dibutyric The amount of active NF-KB was measured in nuclear extracts using electrophoretic mobility shift assays and oligonucleotides that contain either an active binding site for NF-KB or a mutated, inactive NF-KB site (data for the mutant binding site are only shown in Fig. 7 Cells were cultivated in the absence (none) or presence of dibutyric CAMP as described (Scheurich et al., 1989) and were then processed exactly as described in the legend to Fig. 1 at 37" C drastically increased the amount of active NF-KB (Fig. 5A). Also, incubation of cells with the monoclonal antibody htr9, which is specific for the type B receptor (see above), increased the amount of active NF-KB in these cells (Fig. 5A,   lanes 3, 8, and 13 (lanes 1, 6, and II) or were incubated at 37 "C with 10 nM TNFa (lanes 2, 7, and 12) or 10 mg/ml anti-TNF receptor antibody htr9 (lanes 3,8, and 13). As controls, HL60 and HEp2 cells were also incubated with 1 rg/ml IFNy (lanes 4 and 9) or 10 pg/ml monoclonal antibody 45 (Garotta et al., 1990)   , did not amounts on HL60 cells (see above). Untreated HL60 cells, activate NF-KB in HL60 cells (Fig. 5B, lanes 1 and 4) and however, showed only a single binding site for TNFa with a completely blocked NF-KB activation by htr9 (Fig. 5B, lanes Kd of 5.1 x lo-" (Fig. 6A). Preincubation of HL60 cells with 2 and 5) but not its activation by TNFa (Fig. 5B, lanes 3 and htr5, which blocks all detectable binding of TNFa to the type 6). The agonistic effects of htr9 are therefore the result of B receptor (not shown), did not affect the affinity for TNFa binding of the antibody to the type B receptor and are not binding (Kd 5.2 x 10-l' for htr5-treated cells) but reduced the caused by a putative contaminant in the htr9 preparation, e.g. detected number of binding sites from 2,300 to 1,050 sites such as lipopolysaccharides.
This result also suggested that (compare Fig. 6, panels A and B). The higher affinity may either the activation of TNFa in htr5-treated HL60 cells is thus represent binding of TNF to the type A receptor. Preobtained via the type A receptor or that htr5 is not capable treatment of HL60 cells with utrl, which blocked all binding of blocking fully the biological effect of TNFa mediated by to the type A receptor (see Fig. l), reduced the number of the type B receptor even though all binding of TNFa to the binding sites for TNFa from 2,300 to 850 sites and decreased type B receptor was apparently blocked (Fig. 1). For HL60 the binding affinity for TNFa about 7-fold (& of 3.5 X 10-l') cells, the first possibility is most likely (see also "Discussion"). compared with untreated cells (Fig. 6,  Panels A-D, TNFol binding (left panels) and Scatchard analysis (right panels). Triplicate sample8 of untreated and antibody-pretreated HL60 cells (panels A-C) and of untreated HEp2 cells (panel D) were incubated with different concentrations of lz51-TNFa. Incubation was performed in the absence or presence of at Ieast a 300-fold excess of unlabeled TNFa (300 nM) to determine nonspecific binding.
The amount of lz61-TNFa bound to the cells was determined using a y-counter. The specific binding (difference between I?-TNFol binding in absence and presence of unlabeled TNFol) is shown in the left pan&s. Values represent the mean of triplicates f S.E. of total binding. The nonspecific binding was less than a few percent of the total binding. For incubations with monoclonal antireceptor antibodies, HL60 cells were incubated under the same conditions as used for Figs. 1-3. htr5 and htr9 behaved identically (not shown). Scatchard analysis (right panels) was performed using the computer program LIGAND (Munson, 1983 for the type B TNF receptor on HEp2 cells ( Kd of 3.4 x lo-"') ( Fig. 6D). Cross-linking of TNFa to utrl-pretreated HL60 cells confirmed that TNFa was only bound to the type B TNF receptor at all TNFa concentrations used (not shown). At present, however, it is unclear why only one binding site is detected on untreated HL60 cells. The Type A Receptor Mediates Activation of NF-KB by TNFa-All cell types studied by us thus far (Hohmann et al., 1989 and this paper) which express the type A receptor also express the type B receptor. Activity of the type A receptor was thus determined by specifically blocking binding of TNFa to the type A receptor using the anti-type A receptor antibody utrl. If the type A receptors were biologically active, we would expect that higher TNFcv concentrations would be necessary for activation of NF-KB after utrl pretreatment whereas the concentration dependence of NF-KB activation by htr9 should be unaffected. HL60 cells were either mock incubated or were incubated with the monoclonal antibody utrl under conditions that block TNFa binding to all type A receptors and change the affinity of TNFcv binding to that characteristic of the type B receptor (see above). In mock-incubated (untreated) HL60 cells, activation of NF-KB was found with TNFcv concentrations below 0.5 pM and was maximal with 4-8 pM TNFo (Fig. 7, panel A). In cells incubated with utrl, higher concentrations of TNFa were needed for activation of NF-KB, at least 2 PM, whereas maximal activation was seen with 16-32 pM TNFcv (Fig. 7, panel C). Quantitation (Fig. 7, panel E) of the results described above clearly shows that NF-KB activation in utrl-pretreated cells occurs via a site with lower affinity for TNFa; the angles of both curves are different. In addition, NF-KB activation by htr9 via the type B receptor was not affected by pretreatment with utrl (Fig. 7, panels B, D, and F). Thus, the type A TNF receptor mediates NF-KB activation by TNFa.

DISCUSSION
In this paper, we studied the expression and biological activity of the types A and B TNF receptors and determined whether TNF~Y itself is needed for receptor-mediated activation of the transcription factor NF-KB. Three cell lines used here have very different ratios of the types A and B receptors. SW480T cells, derived from the colon carcinoma cell line SW480 by selection for presence of a high number of TNF receptors,' had about 10 times more binding sites for TNFa compared with HL60 cells and showed mainly the lOO-kDa cross-linked product characteristic for the type A receptor. TNFcz binding to SW480T cells was almost quantitatively blocked by the monoclonal antibody utrl. SW480T cells thus have mainly the type A TNF receptor. SW480T cells also contain a small amount of the type B TNF receptor since prolonged exposure of the autoradiograms also showed the 95-and 75-kDa cross-linked products, the formation of which was specifically inhibited by antibodies to the type B receptor. In addition, the agonistic antibody htr9 to type B receptor also activated NF-KB (see below). HEp2 showed only the 95-and 75-kDa products characteristic for the type B receptor as shown previously (Hohmann et al., 1989) cells were incubated for 1 h at 37 "C without antibody (untreated, panels A and B) or with 10 gg of the antibody utrl against the type A TNF receptor per ml (utrl-treated, panels C and D). Under these conditions, all binding to the type A receptor was blocked (not shown and see Fig. 1). The indicated concentrations of TNFol or of the agonistic antibody htr9 against the type B receptor were then added, and incubation was continued for an additional 30 min. Nuclear extracts of identical cell aliquots were prepared and analyzed for NF-KB activity as described in the legend to Fig. 5. Autoradiograms of the gels are shown in panels A-D. Panels E and F, the results were quantified by excising the regions of the gels containing NF+B.DNA complexes and determining the radioactivity using Cerenkov counting. Circles and squares indicate NF-KB induction in untreated and utrl pretreated HL60 cells, respectively.
However, a small amount of the type A receptor was also detected in one experiment. At present it is unclear whether this represents contamination with other cell types or reflects individual differences caused by, e.g. stimulation of the immune system.
HL60 and HEp2 cells only showed a single binding site for TNFa and TNFP in Scatchard analysis (Hohmann et al., 1989(Hohmann et al., , 1990. The affinities of TNFol binding were about 7fold higher for HL60 than for HEp2 cells. The higher affinity of TNFcv binding seems to represent binding to the type A receptor. Preincubation of HL60 cells with the antibody htr5 blocked all detectable binding to the type B receptor and did not affect the affinity of TNFa binding. Preincubation of HL60 cells with the antibody utrl blocked all TNFa binding to the type A receptor and reduced the binding affinity for TNFo( to the affinity found with HEp2 cells. Thus, the lower affinity of TNF binding to the type B receptor can be detected in HL60 cells but only if binding to the type A receptor is blocked. In cross-linking experiments, no difference was observed between the amount of cross-linked products representing the type B receptor in untreated and utrl-treated cells. Thus, TNFa can be bound to the type B receptor in both conditions.
The reason for the apparent absence of the lower affinity binding site in untreated HL60 cells is unclear. It could suggest cooperativity in TNFcv binding and thus interaction between receptor types. However we did not obtain direct and conclusive evidence for receptor interaction. No co-immunoprecipitation of both receptor types was found using antibodies against each receptor type and a variety of different detergent conditions for solubilization of HL60 cell membranes.
Both the types A and B TNF receptors are functional in signal transduction.
The transcription factor NF-KB was activated by TNFa in HEp2 cells, in which only the type B receptor was detected. In addition, the antibody htr9 against the type B receptor also activated the transcription factor NF-KB in SW480T, HL60, and HEp2 cells, all of which have the type B receptor. Thus, the type B TNF receptor mediates NF-KB activation. NF-KB activation by htr9 was blocked by pretreatment with the antagonistic antibody htr5, confirming that the NF-KB activation is indeed mediated by the type B receptor and is not caused by an unknown contaminant in the htr9 preparation.
Pretreatment by htr5 of HL60 cells, which contain both receptor types, did not block the NF-KB activation by TNFa. The most logical explanation would be that NF-KB activation now occurs via the type A receptor. However, this result does not show directly that the type A receptor is also biologically active. NF-KB activation by TNFa (not shown), but not by htr9, was still found in htr5-treated HEp2 cells, although htr5 pretreatment blocked all detectable binding to the type B receptor in HL60 and HEp2 cells. Inhibition of TNFa-mediated NF-KB activation is apparently more difficult than blocking (most of the) receptor binding. We showed previously that occupation of only a minor fraction of all TNF receptors of HL60 and HEpX cells leads to maximal activation of NF-KB (Hohmann et al., 1990). The existence of other (minor) receptor types not detected by our experiments could also explain this observation and can, of course, not be excluded. The type A TNF receptor also mediates activation of NF-KB by TNFa. Pretreatment of HL60 cells with utrl blocked formation of the lOO-kDa product and reduced the Kd of binding of TNFa to the receptor to that specific for the type B receptor (see above). In addition, utrl pretreatment changed the concentration dependence of NF-KB activation by TNFa to that found in HEp2 cells (see also Hohmann et al., 1990) and was then characteristic for NF-KB activation by TNFa binding to a lower affinity binding site. In addition, NF-KB activation via the type B receptor by the antibody htr9 was not affected by utrl pretreatment. Of course, it remains to be established whether each of these receptors directly mediates the biological effects of TNFa or whether interaction with an as yet unknown receptor chain is necessary. At present, we also cannot exclude that signal transduction via the type A receptor requires interaction with the type B receptor; but as mentioned above, no conclusive evidence for interaction between the types A and B receptors is available.
The existence of drastically different ratios of the types A and B receptors on different cells suggested that expression of both receptor types is regulated independently and in a cell-specific manner. To test this hypothesis, we used the observation of Scheurich et al. (1989), who showed that TNF receptors on certain cells could be reversibly up-regulated by dibutyric CAMP. Five of the six cell lines for which they showed up-regulation of TNF receptors were shown previously by us to contain the type A receptor (HL60, U937, K562, SW480, and normal blood monocytes; EL4 cells were not tested by us) whereas at least one of the cells that did not show receptor up-regulation (HeLa) contained only the type B receptor (Hohmann et al., 1989;Brockhaus et al., 1990).
This suggested that dibutyric CAMP may specifically up-regulate the type A receptor. Here we showed that this is indeed the case and thus that the expression of the types A and B TNF receptors can be regulated independently. In addition, we show that dibutyric CAMP does not induce expression of the type A receptor in HEp2 cells and HUVEC, both of which only showed the type B receptor in our crosslinking experiments.
It seems most likely that the up-regulation of the type A receptor in HL60 cells is coupled to cell differentiation induced by overnight incubation with dibutyric CAMP (Chaplinski and Niedel, 1982). Overnight incubation with forskolin did not show a similar effect (not shown). Incubation with forskolin leads to a rapid and drastic increase in the levels of intracellular CAMP (not shown), but this effect might not last long enough to induce cell differentiation.
TNFa is internalized by cells and then degraded (Tsujimoto et al., 1985). The function of this internalization process was unknown. Here we showed that the agonistic antibody htr9 activated NF-KB via the type B TNF receptor (see above). Thus, triggering of the type B TNF receptor is necessary and sufficient for activation of NF-KB. In addition, htr9 also induced cytotoxicity in U937 cells, increased the proliferation of FS4 fibroblasts, and activated human endothelial cells (Espevik et al., 1990). These data show that neither TNFa nor internalization of TNFo( is required for the development of these biological effects. Internalization and degradation of TNFa by cells might only be necessary to remove receptorligand complexes although a possible function of TNFa internalization in other biological effects cannot be excluded. Recently, Smith et al. (1990) presented direct evidence for an intracellular role of TNFa in killing several target cells. Further work will be needed to determine the exact role of TNFL~ in various biological effects.