Distinct roles of the two tumor necrosis factor (TNF) receptors in modulating TNF and lymphotoxin alpha effects.

The role for the two tumor necrosis factor (TNF) receptors in discriminating TNF and lymphotoxin alpha (LTalpha) effects has been studied. TNF and LTalpha were equally mitogenic in Fs4 fibroblasts, which express a high amount of the p55 compared to the p75 TNF receptors (TNFRs). In contrast, TNF was more potent than LTalpha in mediating gene regulation and cytotoxicity in SW480-betaGal cells and KYM-1 cells, which have a high p75/p55 TNFR ratio. Both TNF and LTalpha showed comparable affinities for the two TNFRs. However, in contrast to LTalpha, TNF dissociated rapidly from the p75 TNFR, whereas both cytokines dissociated slowly from the p55 TNFR. Soluble p55 TNFR was much more potent than soluble p75 TNFR in inhibiting TNF cytotoxicity, whereas both soluble receptors moderately decreased LTalpha-mediated cytotoxicity with comparable efficacy. Antagonistic monoclonal antibodies against either TNFR types markedly inhibited TNF effects. However, only the p55 TNFR antagonistic antibody significantly decreased LTalpha-mediated cytotoxicity and cytomegalovirus promoter activation, whereas blocking of the p75 TNFR enhanced the LTalpha effects. These data suggest that whereas the p75 TNFR can both directly propagate TNF signals and "pass" TNF to the p55 TNFR, it attenuates LTalpha and may serve as a decoy receptor for this cytokine.

The role for the two tumor necrosis factor (TNF) receptors in discriminating TNF and lymphotoxin ␣ (LT␣) effects has been studied. TNF and LT␣ were equally mitogenic in Fs4 fibroblasts, which express a high amount of the p55 compared to the p75 TNF receptors (TNFRs). In contrast, TNF was more potent than LT␣ in mediating gene regulation and cytotoxicity in SW480-␤Gal cells and KYM-1 cells, which have a high p75/p55 TNFR ratio. Both TNF and LT␣ showed comparable affinities for the two TNFRs. However, in contrast to LT␣, TNF dissociated rapidly from the p75 TNFR, whereas both cytokines dissociated slowly from the p55 TNFR. Soluble p55 TNFR was much more potent than soluble p75 TNFR in inhibiting TNF cytotoxicity, whereas both soluble receptors moderately decreased LT␣-mediated cytotoxicity with comparable efficacy. Antagonistic monoclonal antibodies against either TNFR types markedly inhibited TNF effects. However, only the p55 TNFR antagonistic antibody significantly decreased LT␣-mediated cytotoxicity and cytomegalovirus promoter activation, whereas blocking of the p75 TNFR enhanced the LT␣ effects. These data suggest that whereas the p75 TNFR can both directly propagate TNF signals and "pass" TNF to the p55 TNFR, it attenuates LT␣ and may serve as a decoy receptor for this cytokine.
Tumor necrosis factor ␣ (TNF) 1 and lymphotoxin ␣ (LT␣, TNF␤) are pleiotropic cytokines which mediate a large variety of inflammatory, immunostimulatory, and antiviral responses (1). They are both members of the TNF ligand and receptor family, which now contains at least 12 ligand-receptor pairs (2). TNF and LT␣ are unique among this family in sharing the two TNF receptors, the p55 TNFR (CD120a) and the p75 TNFR (CD120b), both of which are expressed on most mammalian cells (3). Both TNF and LT␣ exist as homotrimeric molecules, each with the capability of complexing with three receptors (4,5), and both cytokines are believed to elicit their effects by the cross-linking of cell surface receptors (6 -8). LT␣ has also been shown to form a cell surface heterocomplex with LT␤ (9). While the LT␣ homotrimer binds to the p55 and p75 TNFRs, the heteromeric LT␣⅐LT␤ complex interacts with a TNFR-related protein (LT␤-R) (10). Gene knock-out experiments in mice suggest that the LT␣-LT␤ heteromer is involved in the development of the peripheral lymphoid tissues (11), while the two TNFRs are responsible for the resistance for certain types of bacterial pathogens (12,13).
The ratio of the two receptors on the cell surface varies among cells, i.e. some cell types have a high proportion of the p75 TNFR versus the p55 TNFR, and vice versa. Furthermore, the cell surface expression of the two TNFRs is independently regulated in many cell types. It has been shown, for instance, that activation of B cells results in a marked up-regulation of the p75 but not the p55 TNFR (14). Also, the generation of soluble receptors differs for the two TNFRs, because in vitro stimulation of monocytes with lipopolysacharide results in the specific secretion of the p75 TNFR, whereas administration of endotoxin to humans leads to increased serum levels of both types of soluble receptors (15,16). Both receptors can also be found in soluble forms in body fluids during various pathological and physiological conditions (17). However, the functional role of soluble TNFRs in vivo is yet to be elucidated since they have been reported to either attenuate or enhance TNF activity in vitro (18,19). Recombinant soluble receptors are currently under evaluation in the clinic in attempts to attenuate the harmful effects of TNF.
Despite strong redundancy, TNF and LT␣ have been reported to differ in their potency to exert biological effects in many cell types. Thus, TNF was more potent than LT␣ in inducing cytotoxicity toward MCF7 carcinoma cells and cytokine secretion by human fibroblasts, monocytes, and endothelial cells (20 -23). However, TNF and LT␣ were equally cytotoxic toward mouse WEHI clone 13 and L929 fibroblastic cell lines (24,25). LT␣, on the other hand, was more potent than TNF when assayed for the ability to induce cytotoxicity and apoptosis in oligodendrocyte cultures (25). Furthemore, LT␣, but not TNF, has been reported to act as a growth factor for Epstein-Barr virus-infected B cell lines (26). It is, however, not entirely clear to what extent these cytokines have differential physiological effects (27).
The specific roles of the two TNFRs in mediating the TNF and LT␣ signals are currently vividly debated (28,29). In many cell types, studies with antagonistic antibodies have indicated that both receptors are important in mediating TNF effects (30). Furthermore, a unique "passing model" has been suggested to explain the role of the p75 TNFR in mediating TNF responses (31). The basis for this model is that although TNF binds to the p75 TNFR with somewhat higher affinity than to the p55 TNFR, the rate of dissociation of TNF from the p75 TNFR is higher than from the p55 TNFR. This rapid rate of dissociation of the TNF⅐p75 TNFR complex may facilitate interaction of TNF with the p55 TNFR, suggesting a role of the p75 TNFR in passing TNF to the p55 TNFR, which has been postulated to be the main TNF signal transducer. On the other hand, also the p75 TNFR mediates several TNF effects such as proliferation of T-and B-cells (14,32), activation of the human CMV promoter and induction of NF-B (33), and cytotoxicity (8,34,35). Studies with agonistic antibodies specific for either of the receptors have indicated that at least in some cell types both receptors may induce similar effects, although possibly by using different intracellular signaling pathways (33)(34)(35). At present, the distinct functional roles of the two TNFRs, as well as their possible role in defining disparate TNF and LT␣ effects, are not clear.
Here we show that cells expressing a high proportion of the p75 TNFR are relatively resistant to the action of LT␣ and that LT␣, in contrast to TNF, has a slow rate of dissociation from the p75 TNFR. Blocking of the p75 TNFR with antagonistic antibodies inhibited TNF effects but enhanced LT␣ activities in cells expressing a high p75/p55 TNFR ratio. It suggests a mechanism by which the high p75 TNFR expression may inhibit LT␣ effects and indicates distinct specific roles for the two receptor types in discriminating between TNF and LT␣ effects on cells.

EXPERIMENTAL PROCEDURES
Materials-Human recombinant TNF-␣ and TNF-␤ (LT␣) were generously supplied by Genentech Inc. (South San Fransisco, CA) and had a specific activity of 7.6 ϫ 10 7 units/mg and 10 ϫ 10 7 units/mg, respectively, in the LM bioassay. Recombinant soluble TNF receptors were generously provided by Dr. Refaat Shalaby (Genentech Inc.) and were produced by transfecting insect cells with baculoviruses which contained genes encoding extracellular regions of the p55 and p75 TNFRs, respectively. The monoclonal antibodies utr-1, which inhibits binding of TNF to the p75 TNFR, as well as htr-5 and htr-9, which inhibit binding of TNF to the p55 TNFR, have been described before (36 -38). TNF and LT␣ were iodinated by the IODOGEN method (39) by incubating 10 g of cytokine with 2 g of solid 1,3,4,6-tetrachloro-3␣,6␣-diphenylglycoluril (Pierce) and 1 mCi of carrier-free Na 125 I (DuPont NEN, Bad Homburg, Germany; specific activity ϳ17 Ci/mg) for 15 min on ice. Labeled cytokine was separated from unincorporated iodine by chromatography over Sephadex G-25 (Pharmacia, Uppsala, Sweden) in PBS containing 0.1 mg/ml BSA. The specific activity of both cytokines was 60 -80 Ci/mmol in various preparations. Labeled cytokines were stored at Ϫ80°C in the presence of carrier protein (BSA) until use. Both TNF and LT␣ retained full biological activity after iodination when assayed in the WEHI clone 13 biological assay (24). It should be noted, however, that iodinated LT␣ appeared to lose its bioactivity more rapidly than iodinated TNF. Thus, the bioactivity of labeled cytokines were assayed in the WEHI clone 13 bioassay both before and after the experiments depicted here.
Cell Lines-The cell lines applied were the Fs4 human fibroblastic cell line (obtained from Dr. J Vilcek, The NYU Medical Center, New York), the human KYM-1 rhabdomyosarcoma cell line (kindly provided by Dr. A. Meager, The National Institute for Biological Standards and Control, South Mimms, UK), and the human adenocarcinoma cell line SW480-␤Gal (33). This cell line was prepared by transfecting the SW480 cells with pcDNAIneo (Invitrogen Corp., San Diego, CA) containing a promoterless ␤-galactosidase gene under the control of the CMV promoter. The cell lines were grown in RPMI 1640 (Life Technologies, Inc. Laboratories, Paisley, Scotland), supplemented with 2 mM L-glutamine, 10% heat-inactivated fetal calf serum (HyClone, Logan, UT), and 40 g/ml garamycin (culture medium, CM).
Binding of Iodinated TNF and LT␣ to Cells-For measurements of specific binding of cytokines to cells, confluent cells were incubated in 24-well tissue culture plates with the indicated amounts of iodinated cytokine in the absence or presence of 0.5 nM unlabeled cytokine or antibodies as indicated in PBS, 1 mg/ml BSA, 0.02% NaN 3 for 3 h at room temperature. The cells were washed three times in ice-cold PBS/ BSA/NaN 3 , solubilized in 0.5 ml/well 0.2 N KOH and counted in a Gamma spectrometer (Packard, Meriden, IL). Cell-associated radioactivity in the presence of excess unlabeled cytokine was defined as nonspecific binding and was subtracted from the values obtained without cold cytokine present to obtain the specific binding. The nonspecific binding was generally less than 10% of the specific binding. In experiments to measure the rate of dissociation, cells as above were incubated with saturating concentrations of iodinated cytokines, both in the presence and absence of utr-1 and htr-9, as above. The medium was then removed and PBS/BSA/NaN 3 containing 0.5 nM unlabeled cytokine was added. The cells were then incubated at room temperature for the indicated time periods, washed three times, and cell-associated radioactivity was assayed as described above. Shown are typical experiments out of at least three in each case.
Thymidine Incorporation Assay-Thymidine incorporation in Fs4 cells was measured as described (30)  Cytotoxic Assay-The cytotoxic response of WEHI clone 13 and KYM-1 cells was estimated as described earlier (24,35) with small modifications. Briefly, KYM-1 cells (3 ϫ 10 4 cells/well) were seeded into 96-well microtiter plates in 100 l of CM and incubated overnight. The medium was replenished, and the cells were incubated with cytokines and antibodies as indicated in the presence of 0.5 g/ml actinomycin D (Serva, Heidelberg, Germany) for 24 h at 37°C. Cell viability was then assayed by staining the cells with 0.5 mg/ml 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) for 4 h followed by dye extraction with isopropyl alcohol/HCl and measuring the absorbance of soluble dye at 570 nm. Cell viability was calculated according to the formula: cell viability (%) ϭ (B/A) ϫ 100, where A and B represent the A 570 in control and cytokine-treated cells, respectively.
␤-Galactosidase Assay-The activation of CMV promoter in the SW480-␤Gal cells was measured as described earlier (33). Briefly, cells were grown for 3 days at 37°C in microtiter wells (2 ϫ 10 4 cells per well) of 96-well flat-bottomed plates. Then, cytokines and antibodies were added as indicated, and incubation was continued for 5 h at 37°C. Cells were washed in PBS, and 100 l of 1.5 M clorophenol red-␤-D-galactopyranoside (Boehringer Mannheim, Germany) in Hanks' balanced salt solution containing 0.5% Nonidet P-40 (Sigma) were added. Following incubation for 20 min at 37°C, the reaction was stopped by addition of 100 l of 50 mM sodium carbonate (Sigma), and ␤-galactosidase activity was determined by measuring absorbance at 570 nM in a microplate reader (Bio-Rad Laboratories).

The Relative Potency of TNF and LT␣ in Mediating
Biological Responses-It is well known that in some cell lines TNF and LT␣ are equally potent in mediating a biological effect. Thus, TNF and LT␣ exhibited equal cytotoxic effect when added to WEHI clone 13 cells (Ref. 24 and data not shown). Furthermore, as shown in Fig. 1A, TNF and LT␣ were equally potent in inducing proliferation of the Fs4 human fibroblastic cell line. In contrast to this, LT␣ was less potent than TNF in other cell lines. Thus, as shown in Fig. 1B, about 10 -50-fold higher concentrations of LT␣ compared to TNF were needed to obtain a similar cytotoxicity toward the human rhabdomyosarcoma cell line KYM-1, when applying the same preparations of the cytokines as used in Fig. 1A. Furthermore, when assaying for activation of a CMV reporter construct in the human adenocarcinoma cell line SW480-␤Gal, 500-1000-fold higher concentrations of LT␣ were required to obtain an effect similar to that induced by TNF. Moreover, in SW480-␤Gal cells, the maximal response obtained with LT␣ was lower compared to TNF, and the dose of LT␣, inducing the half-maximal response, was about 1000-fold higher than that for TNF (Fig. 1C). The same difference between TNF and LT␣ was seen when directly assaying activation of the transcription factor NF-B in SW480-␤Gal cells, whereas the time required for the half-maximal response was indistinguishable between TNF and LT␣ (Fig. 2).
Taken together, these results indicate that although TNF and LT␣ are about equally potent in inducing responses in some cell lines, other cells are more or less resistant to the action of LT␣ while still being susceptible to the action of TNF.
The Expression of TNF Receptors in Fs4, KYM-1, and SW480-␤Gal Cells-In order to understand the large relative differences in the sensitivity of the cell lines to TNF and LT␣ described above, we speculated that this could correlate with possible differences in binding of the two cytokines to these cell lines. The total number of receptors as well as the binding to each receptor were estimated by performing binding experiments in the presence and in the absence of antibodies which specifically block the binding of TNF and LT␣ to one but not the other receptor. As shown in Fig. 3, both TNF and LT␣ bound specificially and in a saturable manner to these cell lines. When the binding experiments were performed in the presence of the p75 TNFR blocking antibody, utr-1, the binding of both TNF and LT␣ was reduced by 10 -20% in Fs4 cells, by 80 -90% in KYM-1 cells, and by more than 95% in SW480-␤Gal cells. The p55 TNFR antibody htr-9 inhibited the binding of both TNF and LT␣ to Fs4, KYM-1, and SW480-␤Gal cells by 90%, 10%, and 5%, respectively. These data demonstrate that the binding sites recognized by both TNF and LT␣ are mainly of the p55 type in Fs4 cells, whereas in KYM-1 and SW480-␤Gal cells, 90% and 95%, respectively, of the binding sites are of the p75 type.
Scatchard analysis of the data indicated that all three cell lines used here express approximately similar numbers of the p55 TNFR per cell (1000 -3000 receptors/cell). On the other hand, these cell lines express in the range from a few hundreds (Fs4) to more than one hundred thousands (SW480-␤Gal) of the p75 TNFR per cell (data not shown). Taken together, these results suggest that a high specific activity of LT␣ correlates with a high p55/p75 TNFR ratio, whereas the specific activity of TNF is not particularly affected by the p55/p75 ratio. It should also be noted that in all three cell lines examined, the sum of the binding of both TNF and LT␣ in the presence of utr-1 and htr-9 equals the specific binding in the absence of antibodies. Thus, there is no indication of the existence of a third type of binding sites for either TNF, or LT␣, which could explain the differences in their biological activity.
Affinity of TNF and LT␣ Binding to the p55 and p75 TNFR Expressed on Various Cell Types-It has been shown previously that TNF binds to the p75 TNFR with a dissociation constant of 0.3-0.5 nM, whereas the dissociation constant for TNF binding to the p55 TNFR has been reported to be 3-5 times higher (0.7-1.5 nM) (31, 34). Here we show Scatchard analysis of the

FIG. 1. Effects of TNF and LT␣ in Fs4 (A), KYM-1 (B), and SW480-␤Gal (C) cells. Cells were incubated as indicated with TNF (E)
and LT␣ (q). Responses were assayed as described under "Experimental Procedures" and were: incorporation of [ 3 H]thymidine in Fs4 cells, cytotoxicity measured as MTT release in KYM-1 cells, and activation of a ␤-galactosidase gene under the control of the CMV promoter in SW480-␤Gal cells.

FIG. 2. Mobilization of NF-B in SW480-␤Gal cells by TNF and LT␣.
Gel-shift assays were performed on nuclear extracts from SW480-␤Gal cells treated with the indicated concentrations of TNF or LT␣ for 1 h at 37°C (left-hand side) or with 10 ng/ml TNF and 100 ng/ml LT␣ for the indicated time periods in minutes (right-hand side). wt-B and m-B represent lanes where the nuclear extracts and the specific probe were incubated with excess amounts of unlabeled wild type NF-B and mutated NF-B probes, respectively, as described earlier (20).
binding of LT␣ toward the Fs4 and KYM-1 cells, both in the absence and in the presence of the antibodies utr-1 and htr-9 (Fig. 4, A and B). These data demonstrate that LT␣ binds also to the two TNFRs in a way comparable with TNF. Thus, for LT␣, we estimated K d values of 0.3 nM and 0.6 nM for the p75 and the p55 TNFRs, respectively. Furthermore, as in the case of TNF, LT␣ appeared to bind to both the p55 and the p75 TNFRs with comparable affinity constants in Fs4 and KYM-1 cells (data not shown).
Taken together, these data indicate that the difference in specific activity between TNF and LT␣ in Fs4 and KYM-1 cells cannot be explained by the differences in affinities of LT␣ and TNF to the p55 and p75 TNFRs.
Differences in the Rate of Dissociation of TNF and LT␣ from the p75 TNFR-The equilibrium binding experiments described above revealed no major differences between TNF and LT␣ which could explain the large differences in biological activity between the cell lines. To further examine the binding of these cytokines to their receptors, we therefore studied the dynamic interactions between receptors and ligands by measurements of the rates of dissociation.
As shown in Fig. 5A, TNF rapidly dissociated from the p75 TNFR cells. In striking contrast to this, LT␣ appeared to have a relatively slow rate of dissociation from this receptor, indicating that the on-off rate of the LT␣⅐p75 complex is much slower than the on-off rate of the TNF⅐p75 complex. The experiments shown were performed at room temperature and indicate at least a 6-fold difference between TNF and LT␣ in the rate of dissociation from the p75 TNFR at this temperature (t1 ⁄2 ϳ5 min for TNF versus more than 30 min for LT␣). The rate of dissociation for LT␣ and TNF from the p55 TNFR appeared to be indistinguishable, and somewhat lower (t1 ⁄2 Ͼ 60 min) than the rate of dissociation for LT␣ from the p75 TNFR (Fig. 5B). These data demonstrate that there are major differences between TNF and LT␣ in their interaction with the p75 TNFR, whereas the experiments reveal no differences in the interaction with the p55 TNFR.
Effect of Soluble TNFRs on TNF and LT␣ Activity-At least two predictions can be made on the basis of the results described above. First, soluble p55 TNFR may be more potent than soluble p75 TNFR in inhibiting TNF effects, and, second, both receptors could be equally potent in decreasing LT␣ activity. To test this, murine WEHI clone 13 fibroblasts were incubated with TNF or LT␣ alone, in the presence or in the absence of soluble recombinant TNFRs. As shown in Fig. 6, soluble p55 TNFR was more potent than soluble p75 TNFR in inhibiting TNF cytotoxic effect, whereas both receptors similarly decreased LT␣-mediated cytotoxicity. Interestingly, the inhibition of the LT␣ effect by both receptors was less pronounced than the inhibition of TNF activity by the p75 TNFR. Taken together, the results demonstrate that soluble p55 TNFR is more potent than soluble p75 TNFR in inhibiting TNF activity, whereas the LT␣-induced cytotoxicities were less potently, but equally, decreased by both soluble receptors.
Effect of Antagonistic TNFR Antibodies on TNF and LT␣ Activity-To test functional consequences of the difference in the rate of dissociation for TNF and LT␣ from the p75 TNFR, we applied antagonistic p55 and p75 TNFR monoclonal antibodies (htr-5 and utr-1, respectively). As shown in Fig. 7, A and C, both htr-5 and utr-1 markedly inhibited the TNF-induced CMV promoter activation in SW480-␤Gal cells and KYM-1 cytolysis. In contrast, only treatment of cells with htr-5 de-  Fig. 3 were plotted according to the method of Scatchard.

FIG. 5. Dissociation of TNF and LT␣ from the p75 (A) and the p55 TNFR (B).
In A were SW480-␤Gal cells incubated with saturating amounts of iodinated TNF (E) and LT␣ (F) in the presence of 10 g/ml htr-9 for 3 h. In B were Fs4 cells incubated with saturating concentrations of iodinated TNF (E) and LT␣ (F) in the presence of 10 g/ml utr-1 as above. Cells were washed, and unlabeled TNF and LT␣ were added to wells containing bound labeled TNF and LT␣, respectively. Cellassociated radioactivity was measured at the indicated time points thereafter as described under "Experimental Procedures."

FIG. 3. Specific binding of 125 I-TNF (A, C, and E) and 125 I-LT␣ (B, D, and F) to Fs4 (A and B), KYM-1 (C and D), and SW480-␤Gal (E and F) cells. Cells as described under "Experimental Procedures"
were incubated with the indicated concentrations of iodinated TNF or LT␣, in the absence or presence of either 0.5 nM unlabeled TNF or LT␣, 10 g/ml p75 TNFR specific utr-1 antibody, or 10 g/ml p55 TNFR specific htr-9 antibody. creased the ability of LT␣ to mediate biological responses. Interestingly, blocking of the p75 TNFR with utr-1 enhanced the LT␣-mediated activation of the reporter construct and its cytotoxic effect (Fig. 7, B and D). These data show that TNF uses both TNFR types for mediating signal transduction, whereas LT␣ involves only the p55 TNFR. In addition, these results indicate that blocking of the p75 TNFR inhibits the TNF-mediated responses but increases the LT␣ activities in cells expressing a high p75/p55 TNFR ratio. DISCUSSION Why TNF and LT␣ show, depending on the cell type, either similar or different biological effects remains one of the unresolved questions in the cytokine field. The present paper demonstrates a correlation of cell responsiveness to TNF and LT␣ with the expression of the p75 TNFR. Thus, TNF was 100-1000-fold more potent than LT␣ in cells expressing a high p75/p55 TNFR ratio, whereas both cytokines showed equal efficacy in evoking biological responses of cells expressing largely the p55 TNFR. The correlation of expression of a high proportion of the p75 TNFR with the relative resistance of cells to LT␣ is not restricted to the three cell lines described here. Also the human myeloma cell line OH-2, which predominantly expresses the p75 TNFR, is 100 -100-fold more resistant to the growth stimulatory action of LT␣ compared with TNF. 2 Furthermore, in U937 cells, which express about 80% p75 and 20% p55 TNFRs (41,42), LT␣ is required in at least 10-fold higher concentrations than TNF to obtain a similar cytotoxic effect (41). 3 These data, together with the results on the lack of homology in the receptor binding regions of TNF and LT␣ (4), their differences in trimer formation (43), and stability of trimers (44), suggest different modes of receptor interactions and/or receptor triggering caused by TNF and LT␣.
To verify this hypothesis, we first studied the interaction of TNF and LT␣ with the two TNFR by measuring their affinities and on-off rates of dissociation. Comparable K d values have been detected for both cytokines upon their binding to the two TNFRs. Together with previously published data (41,45), it suggests that different efficacy of TNF and LT␣ cannot be explained by their different affinities for the TNFRs. Of impor-tance, differences between TNF and LT␣ in the dynamic interaction with the receptors have been observed, which are not reflected in the measurements of affinity constants. Thus, TNF had a much more rapid on-off rate in the interaction with the p75 TNFR than with the p55 TNFR. In contrast, LT␣ showed only minor differences in its rate of dissociation from the p75 TNFR compared with that from the p55 TNFR and exhibited a significantly slower rate of dissociation from the p75 TNFR than TNF.
Secondly, the aforementioned difference between TNF and LT␣ in the dynamic interaction with the p75 TNFR has been confirmed by using soluble TNFRs. In accordance with the data reported by other groups (17,18), the soluble p55 TNFR was considerably more potent than the soluble p75 TNFR in inhibiting a TNF effect. Assuming that a rapid on-off rate of TNF in the interaction with soluble receptors increases the probability of interaction of TNF with a signal transducing cell surface receptor, the difference reported here can be explained by the differences in the rates of dissociation for TNF between the p55 and p75 TNFR. Also, in the case of LT␣, similar rates of dissociation in the interaction with the two TNFRs are reflected in similar abilities of the two soluble receptors in inhibiting the LT␣ cytotoxic effect. Interestingly, however, the inhibition of the LT␣ effect by either TNFR type was much less pronounced than the inhibition of the TNF activity by the p55 TNFR. These results confirm the finding that LT␣ has a lower potency to bind to the soluble form of the p55 TNFR when compared to TNF (46). Furthermore, in attempts to attenuate cytokine action in vivo, the implication of these results could be that in the case of TNF soluble p55 TNFR should be a more potent antagonist than the p75 TNFR, whereas perhaps none of the soluble receptors may be strong antagonists of LT␣ action in vivo.
Thirdly, the application of antagonistic TNFR antibodies in this study has revealed the different use of the p75 TNFR by TNF and LT␣. Blocking of both the p55 and p75 TNFRs led to a marked inhibition of the TNF effects in SW480-␤Gal and KYM-1 cells. In accordance with previously published observations (6,8,36,41), it indicates that TNF uses both TNFR types for signaling. In contrast, the antagonistic p75 TNFR antibody utr-1 potentiated the ability of LT␣ to mediate CMV promoter activation in SW480-␤Gal cells and to cause cytotoxic effect in KYM-1 cells, whereas blocking of the p55 TNFR inhibited the LT␣ effects. In line with this, only the p55 TNFR has been reported to mediate the LT␣ cytotoxic effect in U937 cells (41) and its ability to up-regulate the expression of adhesion molecules in human vascular endothelial cells (47) as well as the proliferative response of human primary fibroblasts (44), whereas both receptors were important for the TNF effects (36,41). Thus, despite the ability of LT␣ to bind with high affinity to the p75 TNFR (28), at least in some cell lines, this cytokine does not seem to mediate signal transduction through this receptor type.
The p75 TNFR has also been proposed to play an accessory function by passing TNF to the p55 TNFR due to its higher affinity, thereby providing more efficient signal transduction by the p55 TNFR (28). However, it is unlikely that the p75 TNFR mediates interaction of LT␣ with the p55 TNFR in SW480-␤Gal and KYM-1 cells, because in this case the on-off rate apparently is much slower. On the opposite, it could attenuate LT␣ and prevent its binding to the p55 TNFR. Indeed, antibodies which block the binding of LT␣ to the p75 TNFR without affecting the binding to the p55 TNFR potentiated the LT␣ effects but inhibited the TNF activities in SW480-␤Gal and KYM-1 cells. Thus, whereas the p75 TNFR is capable of both directly propagating TNF signals and "passing" TNF to the p55 TNFR, it appears to attenuate LT␣ and to serve as a "decoy" receptor for this cytokine in SW480-␤Gal and KYM-1 cells.
Under normal physiological conditions, LT␣ forms a heteromeric complex with LT␤ on the cell surface (9). It has been published that LT␤ can be found only in a membrane-associated form (40). A novel LT␤ specific receptor was identified, which binds to heterotrimers with the stoichiometry of LT␤2/ LT␣1 but not to those with the stoichiometry LT␤1/LT␣2 (10). Moreover, LT␣⅐LT␤ complexes with different stoichiometries can be distinguished by the LT␤R and the p55 TNFR (40). To the best of our knowledge, no binding between the LT␤R and LT␣ homotrimers has been reported. In addition, radiolabeled TNF can be displaced with excess cold LT␣ and vice versa, and the combination of the p55 and p75 TNFR antagonistic antibodies completely prevented the binding of both cytokines to the cell lines. 4 Thus, the binding data obtained in this study with soluble LT␣ give no indication of the involvement of LT␤ or the LT␤ specific receptor in the binding of homotrimeric LT␣ to the cell lines examined.
Interestingly, TNF and LT␣ exhibited comparable abilities to inhibit proliferation of the human breast carcinoma cell line BT-20, that had only low affinity receptors (suggesting the p55 TNFR), whereas in cell lines that had high affinity receptors (suggesting the p75 TNFR), TNF was 20 to 320 time more potent than LT␣ (45). It is plausible that the lack of LT␣ signaling through the p75 TNFR and the ability of this receptor to attenuate LT␣, found in the present study, could be the reasons explaining differential biological effects of these cytokines. In contrast, LT␣, but not TNF, acts as a growth factor for Epstein-Barr virus-infected B cell lines which express mainly the p75 TNFR (26). The reason for this discrepancy is not clear but could be related to the importance of the two TNFRs in mediating specific biological responses in various cell types. Further studies of the role of the two TNFRs in TNF and LT␣ signaling, especially in the cell types where the p75 TNFR is the main signal transducing molecule, are needed to bring a better understanding of the mechanisms behind differential or similar effects of these two cytokines.