(Received for publication, July 25, 1996, and in revised form, January 28, 1997)
From the Department of Pathology, Center for Immunology, Washington University School of Medicine, St. Louis, Missouri 63110
The p55 and p75 tumor necrosis factor receptors are known to mediate their effects on cells through distinct signaling pathways. Under certain circumstances, the two classes of TNF receptors cooperate with each another to produce enhanced cellular responses. The only molecular mechanism proposed thus far to explain this effect is the process of "ligand passing," whereby TNF is concentrated at cell surfaces by binding to p75 and then following dissociation from this receptor class binds with high efficiency to p55. Using the in vivo model of TNF-induced TNF receptor shedding we have uncovered a novel ligand-dependent interaction of the two TNF receptors that occurs upon exposure of cells to TNF. Using TNF receptor-specific monoclonal antibodies that bind TNF receptors in the presence or absence of ligand, we report that TNF induces the formation of heterocomplexes consisting of both p55 and p75 TNF receptors. Whereas immunoprecipitates from untreated or human TNF-treated cells formed with either p55 or p75 TNF receptor-specific monoclonal antibodies contained only the relevant TNF receptor class, anti-p55 or anti-p75 precipitated both receptor types from murine TNF-treated cells. Ligand-induced complex formation was transient, occurred at physiologically relevant concentrations of TNF, and occurred with receptors lacking intracellular domains or that contained irrelevant transmembrane domains. Formation of TNF receptor heterocomplexes may therefore 1) define a novel molecular mechanism of ligand passing and/or 2) contribute to cooperative TNF receptor signaling via the juxtaposition of the intracellular domains of the two receptor classes and the signaling proteins that they recruit.
TNF1 interacts with two distinct receptors of Mr 55,000 and 75,000, which are independently expressed on cell surfaces (1, 2). The p55 and p75 TNF receptors share 28% homology in their extracellular domains but show no homology in their intracellular domains (3). The TNF receptors belong to a receptor family that displays extracellular domain homology largely through conservation of cysteine-rich repeating sequences and includes the low affinity nerve growth factor receptor, Fas, OX-40, CD30, CD40, and 4BB1 (4).
Recent work (5-7) has revealed that the two receptor classes interact with distinct families of cellular proteins through their intracellular domains. These observations have led to the suggestion that the two TNF receptors utilize different signaling mechanisms. The ability of each TNF receptor to signal biologic responses in cells has been extensively studied during the past few years. Engagement of p55 is now known to be both necessary and sufficient to induce a variety of proinflammatory TNF-mediated cellular responses including cytotoxicity (8), induction of inducible nitric-oxide synthase (9) and manganous superoxide dismutase (10), expression of intercellular adhesion molecule 1 (ICAM-1) (11, 12), and anti-viral activity (13). In contrast, p75 appears to effect only a limited number of cellular responses, many of which are restricted to T cell populations and include effects on proliferation/cell viability (14, 15) and cytokine production (16).
In many biologic systems, p75 plays an accessory role in p55-mediated responses. In murine models, the engagement of both p55 and p75 by MuTNF leads to enhanced cellular responses compared with the selective engagement of p55 by HuTNF or p55 agonistic antibody. In addition, overexpression of p75 can result in the enhancement of p55-mediated signaling whereas p75-specific antagonist mAbs partially inhibit p55-induced responses (8, 16-18). To some extent, this cooperativity has been explained by "ligand passing," a process by which TNF is concentrated near the cell surface by selectively binding to p75 (because p75 binds ligand with a 10-20-fold higher affinity than p55 binds ligand), dissociates, and subsequently binds with increased efficiency to p55 (17).
Recent studies have shown that TNF can induce shedding of both classes of TNF receptors in mice (19). The biologic relevance of this response has been ascribed to a protective effect of soluble TNF receptors which can block TNF binding to other cell surface receptors and to the loss of cell surface receptors rendering the cell TNF-insensitive. The molecular basis of the shedding response has not yet been determined. In the current study, we analyzed the receptor requirements for TNF-induced p75 shedding in vivo and observed that p75 shedding could be mediated by engagement of p55 alone but not by p75 alone. However, simultaneous engagement of both p55 and p75 led to an enhanced shedding response. No level of p55 ligation could induce the amount of p75 shedding effected when both receptors were engaged. These results suggest that the cooperativity between p55 and p75 in this experimental model may extend beyond ligand passing.
One possible mechanism underlying this observation is that TNF may induce formation of a receptor heterocomplex containing both p55 and p75. However, previous studies by others have failed to demonstrate such a heterocomplex (17, 20, 21). In the current report we have used a unique set of monoclonal antibodies specific for murine p55 and p75 whose binding properties are not affected by the presence of ligand to demonstrate that murine TNF can indeed form heterocomplexes between p55 and p75 on the surface of intact cultured and primary cells. This result thereby suggests that 1) one mechanism of "ligand passing" may be a direct, contact-mediated hand-off of TNF from p75 to p55 in which a heterocomplex of p55 and p75 is an obligate intermediate and/or 2) formation of p55-p75 heterocomplexes and the subsequent juxtaposition of their intracellular domains could potentially lead to functional cooperativity between the signaling components that associate with the two types of TNF receptors.
Monoclonal antibodies specific for
murine TNF (22) and murine p55 and p75 (18) were produced as
described and biotinylated using the Enzotin reagent (ENZO Biochem,
Inc., New York, NY) according to the manufacturer's protocol. Purified
recombinant MuTNF
(1.2 × 107 units/mg) and
rHuTNF
(5.6 × 107 units/mg) were generously
supplied by Genentech, Inc., South San Francisco, CA. All reagents used
in this study contained less than 0.5 pg/ml endotoxin as determined by
the Limulus amebocyte lysate assay (BioWhittaker Inc.,
Walkerville, MD). Lipopolysaccharide (Escherichia coli
strain O127:B8) was purchased from Difco Laboratories (Detroit,
MI).
Balb/c ByJ female mice (6-7 weeks of age) were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice with genetic deficiency of the p55 TNF receptor were provided by Dr. Werner Lesslauer (Hoffman-La Roche, Basel, Switzerland). Mice deficient in p75 were provided by Dr. Mark Moore (Genentech, Inc.).
In Vivo TNF Receptor Shedding ExperimentsMice were injected intraperitoneally with 0.5-ml samples of monoclonal antibodies or pyrogen-free physiologic saline. At various time points thereafter lipopolysaccharide or purified recombinant murine or human TNF was diluted in 0.5 ml of pyrogen-free saline and was injected intraperitoneally into the mice. At specified time points mice were bled and the serum was immediately assayed for the presence of shed TNF receptor proteins by enzyme-linked immunosorbent assay. For detection of shed p75, the enzyme-linked immunosorbent assay consisted of plate-bound TR75-54 and biotinylated TR75-32 as the detection antibody (18, 28).
Scatchard Analysis of TNF Receptor ExpressionMuTNF was radiolabeled with Na125I (ICN, Irvine, CA) and IODO-BEADS (Pierce) according to the manufacturer's directions. The total number of TNF binding sites was assessed by offering 0.5, 1, 2, 4, 6, 8, 10, 15, 25, 50, 75, 100, or 125 ng of radiolabeled MuTNF to 4 million Meth A cells for 1 h at 4 °C. The numbers of p55- or p75-specific binding sites were quantitated by preincubating the cells for 1 h at 4 °C with a 200-fold molar excess of TR75-54 or 55R-593, which are blocking monoclonal antibodies to p75 or p55, respectively. Saturation of binding was achieved in all experiments, and regression of Scatchard analyses yielded correlation coefficients (r2) of 0.86 or greater. Meth A cells were calculated to express 21,700 total TNF binding sites, with 12,100 p55-specific sites and 7,800 p75-specific sites. The affinities (Kd) of p55 and p75 binding were 2.7 and 0.13 nM, respectively.
TNF Receptor CoprecipitationMeth A fibrosarcoma cells were grown in RPMI 1640 medium supplemented with 10% fetal calf serum, 1% L-glutamine, 1% penicillin/streptomycin, 1% sodium pyruvate, 1% non-essential amino acids, and washed twice in PBS. Thymocytes were harvested by gentle teasing with forceps and washed twice with PBS. Fifty million cells of either type were incubated with TNF for 5 min at 37 °C and immediately placed on ice. The cells were washed twice with cold PBS and then lysed in buffer containing 150 mM NaCl, 50 mM Tris (pH 7.5), 1% Brij 96, 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 5 mM iodoacetamide, 2.3 trypsin inhibitory units/ml aprotinin, 10 µg/ml leupeptin, 1 mM sodium orthovanadate, and 10 mM sodium fluoride for 20 min at 4 °C. Following lysis, the debris was removed by centrifugation at 11,000 × g for 10 min, and the samples were precleared with 20 µl of a 1:1 slurry of protein A-Sepharose 4 Fast Flow (Pharmacia Biotech Inc., Uppsala, Sweden). Five µg of nonblocking mAbs (55R-286 or TR75-89) were added, and the samples were incubated for 1 h at 4 °C. Sixty µl of protein A-Sepharose slurry was added, and incubation continued for 1 h at 4 °C. The Sepharose was washed four times in lysis buffer, suspended in 30 µl of Laemmli buffer, and heated at 70 °C for 5 min. Samples were then subjected to electrophoresis on nonreducing SDS-polyacrylamide gels and Western blotting for p55 or p75 conducted as described (18) using biotinylated forms of 55R-286 or TR75-54, respectively. Blots were developed using the ECL reagent (Amersham Life Science, Inc., Buckinghamshire, United Kingdom).
Generation of Truncated TNF Receptor ConstructsThe
cDNAs encoding the wild-type murine p55 and p75 TNF receptors in
the expression vector pRK5 were obtained from Dr. D. Goeddel (Tularik,
Inc., South San Francisco, CA). Polymerase chain reaction-directed
mutagenesis was utilized to generate cDNAs encoding cytoplasmically
truncated forms of p55 and p75. The oligonucleotides used for the
truncated p55 construct were 5
(5
-GCCTACTCGAGACCTGGTCCGATCATCTTAC-3
) and 3
(5
-CGAGGTCTAGACCTTACCTCCACCGGGGATATCG-3
). Oligonucleotides used in
the generation of the truncated p75 construct were 5
(5
-GCCTACTCGAGTCTAGCTCCAGGCACAAGGGC-3
) and 3
(5
-CGAGGTCTAGAGGTCACTTCTTTTTCCTCTGCACCAG-3
).
These oligonucleotides generate a XhoI site to the 5 end
and an XbaI site to the 3
end that were subsequently used
to ligate into the pSR
expression vector. The resulting cDNAs
allow for translation of mutant p55 and p75 receptor proteins that
contain only 6 or 4 residues of their respective intracellular domains. The sequences of all constructs were verified on both strands by ABI
Prism Dye Terminator Sequencing (Perkin-Elmer). Meth A cells were
electroporated-selected with 1 mg/ml G418, and the 5% highest
expressing cells were isolated by flow cytometry.
One million Meth A cells
were suspended in 1.5 of ml tissue culture medium and stimulated with
different doses of recombinant MuTNF or medium for 2 h at
37 °C. Nuclear extracts were prepared as described (23) and NF-
B
activation quantitated using an electrophoretic mobility shift assay
that employed a 32P-labeled 27-base pair oligonucleotide
probe derived from the promoter region of the Ig
gene (24).
It is now well established that TNF can effect shedding
of the p75 TNF receptor in vivo
(19).2 We therefore wanted to define the
role of each TNF receptor class in mediating the response. Serum from
untreated mice contained approximately 1 ng/ml of soluble p75 (Fig.
1A). Intraperitoneal injection of 7.5 µg MuTNF (the maximum tolerated dose) effected additional shedding
of soluble p75 which first became detectable within 30 min, reached
peak levels of 15 ng/ml by 3 h, and returned to constitutive
levels of 1 ng/ml by 24 h. When mice were treated with a lower
dose of MuTNF
(1 µg), the magnitude of p75 shedding was reduced by
75%, but the kinetics of the response remained unchanged (Fig.
1A).
To investigate whether p55 ligation was necessary for TNF- mediated p75
shedding, we examined whether murine TNF effected the shedding response
under circumstances where p55 ligation could not occur. MuTNF
induced an 11-fold increase in the level of circulating soluble p75 in
normal mice (17 ± 0.6 versus 1.5 ± 0.4 ng/ml for
mice treated with 7.5 µg of MuTNF or saline, respectively, Fig.
1B). In contrast, no significant increase was observed in TNF-treated p55-deficient mice (1.7 ± 0.4 versus
1.3 ± 0.01 ng/ml for MuTNF- or saline-treated mice,
respectively). In agreement with previous reports,
lipopolysaccharide-induced p75 shedding was not dependent on TNF (19)
and thus occurred to a comparable extent in p55-sufficient and
deficient mice. Similar results were obtained using normal Balb/cByJ
mice treated with blocking p55-specific mAb (55R-170). In this case
55R-170 effected a dose-dependent inhibition of TNF-induced
p75 shedding such that 85-90% of shedding was inhibited at the
highest dose of antibody used (450 µg/mouse, data not shown). These
results thereby demonstrate that p55 ligation is required for
TNF-induced p75 shedding.
To determine whether p55 engagement was sufficient to induce maximal p75 shedding in vivo, the experiments were repeated using two agonists which interact only with p55 and not p75: human TNF (25) and the agonistic p55-specific mAb, 55R-593 (18). In both cases, the agonists induced p75 shedding with kinetics similar to those effected by murine TNF (Fig. 1A). However, the maximal levels of p75 shedding were only 25% of those induced by the murine protein when used at the 7.5-µg dose. To explore whether the reduced p75 shedding induced by the p55 selective ligands was due either to an inherent reduction in affinity of TNF for its appropriate target cell (i.e. ligand passing) or a more subtle cooperativity between the two classes of TNF receptors, the in vivo shedding experiments were repeated using different agonist doses. Murine TNF effected a dose-dependent increase in shed p75 reaching levels of 14 ng/ml 3 h after challenge with a maximum tolerated dose of 7.5 µg/mouse (Fig. 1C). In contrast, both human TNF and the 55R-593 mAb induced lower levels of p75 shedding which reached plateau values of only 4 ng/ml even when the agonists were added at the extremely high doses of 300 and 450 µg/mouse, respectively. Moreover, pretreatment of mice with blocking mAbs to p75 attenuated the magnitude of MuTNF-induced shedding of p75 to that observed when the p55-specific ligands were used (HuTNF or 55R-593, data not shown). These observations indicated that p75 ligation, although insufficient alone, is important in effecting p75 receptor shedding. Since no amount of selective p55 ligation effected the level of p75 shedding induced upon engagement of both TNF receptors, the response could not be explained strictly by the model of ligand passing. We therefore considered the possibility that TNF receptor cooperativity could occur via formation of a ligand-induced p55-p75 receptor heterocomplex.
Demonstration That Murine TNF Induces Formation of Heterocomplexes of p55 and p75 at the Cell SurfaceWe previously described unique hamster mAbs that bound murine p55 or p75 TNF receptors in a manner that was independent of receptor occupancy by ligand. We therefore used these mAbs to investigate whether one TNF molecule can simultaneously interact with both p55 and p75. Immunoprecipitation/Western blot experiments were performed on the murine Meth A tumor cell line which expresses 12,100 p55 and 7,800 p75 receptor binding sites as detected by Scatchard analyses (data not shown). Cells were treated with PBS or TNF and then lysed, and immunoprecipitation was carried out with nonblocking TNF receptor-specific mAbs. Precipitates were subsequently analyzed for the presence of each TNF receptor by Western blotting.
TNF receptor antibody-generated immunoprecipitates of Meth A cells
treated with either PBS or HuTNF (which binds only to murine p55)
contained only the relevant TNF receptor class (Fig.
2, A and B, lanes 4,
5, 7, and 8). In contrast, p55 or p75
antibody-generated immunoprecipitates of Meth A cells exposed to MuTNF
(which binds both murine p55 and p75) contained both TNF receptor
classes. Specifically, anti-p75 immunoprecipitates of MuTNF-treated
cells contained p55 (Fig. 2A, lane 9), and
anti-p55 precipitates of MuTNF-treated cells contained p75 (Fig.
2B, lane 6). The identities of the coprecipitated
55- and 75-kDa components were confirmed by showing that the band
detected by Western blotting comigrated with the appropriate receptor
that was directly immunoprecipitated from cell lysates regardless of
cytokine treatment (p55: Fig. 2A, lanes 4-6;
p75: Fig. 2B, lanes 7-9). The specificity of the coprecipitation was further confirmed by two experiments. First, neither p55 nor p75 precipitates contained the murine
interferon- receptor
chain (Fig. 2C, lanes
4-9), thereby ruling out the possibility that the TNF receptor
precipitates contained irrelevant cell surface proteins. Second,
precipitates of the interferon-
receptor did not contain either p55
or p75 (Fig. 2, A and B, lanes 1-3).
Thus, murine TNF is capable of forming mixed TNF receptor complexes on
the surface of this murine tumor cell line.
Heterocomplex Formation Also Occurs in Primary Murine Cells
To investigate whether ligand-induced heterocomplex
formation occurs on normal cells as well as tumor cells, we repeated
the coprecipitation experiments using two different sources of primary murine cells. The results were identical to those obtained with Meth A
cells. Precipitates of p75 contained p55 only when thymocytes were
treated with MuTNF (Fig. 3A, lanes
3 and 4). Conversely, only p55 precipitates from
MuTNF-treated thymocytes contained p75 (Fig. 3B, lanes
1 and 2). Heterocomplex formation was also observed in
MuTNF-treated splenocytes derived from wild-type mice (data not shown).
As expected, heterocomplex formation was not observed in splenocytes
derived from mice deficient for p55 or p75 (data not shown), further
documenting the specificity of the immunoprecipitation/Western blot
analyses. Thus, ligand-dependent TNF receptor heterocomplex
formation occurs in primary as well as cultured cells.
TNF-dependent TNF Receptor Heterocomplexes Are Short-lived at the Cell Surface
To monitor the stability of ligand-induced TNF
receptor heterocomplexes, Meth A cells were treated with MuTNF, washed,
and then incubated in medium for various periods of time before being processed for p55 immunoprecipitation and p75 Western blotting. As seen
previously, no heterocomplexes were observed in the absence of MuTNF
but were obvious immediately following cellular exposure to MuTNF (Fig.
4A, lanes 1 and 2,
respectively). The amount of heterocomplex observed after 1 min of
incubation in TNF-free medium was comparable to that seen in
unincubated cells (Fig. 4A, lane 3) but decreased
rapidly thereafter (Fig. 4A, lanes 4-7).
Densitometric analysis revealed that the half-life of the complex was
approximately 3 min at 37 °C (Fig. 4B). Thus
ligand-induced heterocomplexes display a transient life span at the
cell surface.
The Intracellular and Transmembrane Domains of p55 and p75 Are Not Required for Heterocomplex Formation
To determine whether the intracellular domains of the two TNF receptors contributed to formation of TNF-induced heterocomplexes, we examined whether complex formation could proceed with cytoplasmically truncated receptor mutants. For this purpose we stably expressed cytoplasmically truncated forms of either p55 or p75 in Meth A cells, thereby generating cell lines that contained both wild-type full-length and truncated receptors.
Monoclonal antibodies specific for the extracellular domain of murine
p55 immunoprecipitated both full-length and truncated forms of this
receptor from transfected cells (Meth A-p55, Fig. 5A, lanes 4-6). When these cells
were treated with either PBS or HuTNF, p75 immunoprecipitates did not
contain either form of p55 (Fig. 5A, lanes 7-8).
However, following exposure of the cells to MuTNF, p75
immunoprecipitates contained both full-length and truncated forms of
p55 (Fig. 5A, lane 9). Conversely,
immunoprecipitation for p75 from Meth A cells expressing truncated p75
(Meth A-
p75) revealed the presence of both wild-type full-length and
truncated forms of p75 in these cells (Fig. 5B, lanes
7-9). Precipitation of p55 from MuTNF-treated cells led to the
coprecipitation of both full-length and truncated forms of p75 (Fig.
5B, lane 6). Importantly, the ratio of
full-length to truncated p75 receptor forms in the coprecipitate was
virtually identical to the ratio of full-length to truncated p75
receptor expressed on the cell surface, demonstrating that the
full-length receptor is not preferentially recruited into the complex.
This result thus rules out the possibility that heterocomplexes
contained an equal proportion of the wild-type and truncated receptors.
Thus the wild-type receptor present in the cell was not acting
catalytically to promote heterocomplex formation.
To monitor whether the transmembrane domains of p55 or p75 contributed
to heterocomplex formation, the coprecipitation experiments were
repeated using Meth A cells that overexpressed a protein that consisted
of the murine p75 extracellular domain, the murine interferon-
receptor
chain transmembrane domain and the first three
intracellular domain amino acids. Here again, precipitation of p55 from
MuTNF-treated cells resulted in the coprecipitation of both wild-type
and chimeric forms of p75 (data not shown).
These results demonstrate that the formation of p55-p75 heterocomplexes on cell surfaces does not require interactions between the intracellular or transmembrane domains of the receptors. Instead, it appears that the bridging of the extracellular domains by MuTNF is sufficient for heterocomplex formation. Nevertheless, attempts to generate a TNF-dependent heterocomplex with soluble forms of p55 and p75 have not succeeded. This result indicates that ligand-induced heterocomplex formation requires a specific topographical distribution or structural conformation of the receptor extracellular domains.
Formation of p55-p75 Heterocomplexes Occurs at Physiologically Relevant Concentrations of TNFThe previous experiments were
conducted on cells exposed to high concentrations of MuTNF (10 µg/ml). To determine whether ligand-dependent receptor
heterocomplex formation occurs at physiologic levels of MuTNF, the
coprecipitation experiments were repeated on cells exposed to different
concentrations of the murine ligand. Heterocomplex formation was first
detected by p55 immunoprecipitation/p75 Western blotting at a MuTNF
concentration of 1 ng/ml and reached maximal levels at ligand
concentrations of 10 ng/ml (Fig. 6A). These
results were confirmed by densitometry of the Western blots (Fig.
6B). We also examined NF-B activation in Meth A cells
exposed to the same range of concentrations of MuTNF used in the
immunoprecipitation/Western blot analyses. NF-
B activation was
detected at MuTNF concentrations of 0.01 ng/ml and reached maximum
levels at a TNF concentration of 10 ng/ml (Fig. 6C). Thus
heterocomplex formation can occur at a similar range of TNF
concentrations that triggers a typical TNF-induced response,
demonstrating that heterocomplexes are not due to an artifact of
artificially high ligand concentrations.
While previous in vivo studies using HuTNF suggested that engagement of p55 was sufficient for the TNF-induced shedding of p75 (19), we have shown herein that engagement of both p55 and p75 is required for maximal p75 shedding. In addition, we demonstrate that p75 engagement alone is not sufficient to induce its own shedding but rather acts in concert with p55 in a cooperative manner. We do not attribute this effect to a classical ligand passing mechanism, since the magnitude of p75 shedding induced by p55-selective ligands (HuTNF and p55 agonist mAbs) reaches a plateau that is only 30% of that induced by dual ligation of both receptors by MuTNF. If ligand passing were the mechanism underlying p75 involvement in the shedding process, comparable responses to HuTNF versus MuTNF would be expected by increasing the concentration of the p55-selective human TNF ligand. Such a response was not observed. We therefore conclude that some other mechanism of receptor cooperation is responsible for these observations.
Using a unique set of nonblocking mAbs to the two murine TNF receptors, we have shown in cultured and primary murine cells that MuTNF is capable of concomitantly interacting with p55 and p75, and thereby can form a hetero-receptor complex. Formation of p55-p75 heterocomplexes occurs at physiologically relevant concentrations of TNF and is transient in nature. Moreover, ligand-dependent bridging of p55 and p75 does not require the participation of the intracellular or transmembrane domains of the receptors.
The lack of a role for the intracellular and transmembrane domains of the TNF receptors in heterocomplex formation appears to conflict with the inability to demonstrate heterocomplexes with soluble receptors in solution. However, several factors could account for this apparent discrepancy. First, the interaction may require that the receptors be anchored in the membrane. Interactions between molecules imbedded in the membrane are likely to be more thermodynamically stable than interactions potentially forming between soluble molecules. Second, the membrane-anchored receptors may be held in close proximity to one another, thereby favoring heterocomplex formation. Third, the receptors may adapt a different conformation when they are not in a membrane which could render them unable to form heterocomplexes. So while the specific p55 and p75 transmembrane domains are not required per se, anchoring in the cell membrane domain appears to be necessary to facilitate the formation of heterocomplexes between the two receptor classes.
Importantly, whereas immunoprecipitation with anti-p55 co-precipitated almost all of the p75 protein, immunoprecipitation with anti-p75 co-precipitated only a fraction (5-10%) of the total p55 present. This can partially be attributed to the fact that the cells used in these experiments express almost twice as many p55 than p75 receptors. In addition, it is possible that the binding of the p75-specific antibody to the TNF-induced p55-p75 receptor heterocomplex results in a partial dissociation of p55.
Functional cooperativity between p55 and p75 TNF receptors is well documented. Until now, the only molecular explanation for this has been derived from the model of ligand passing proposed by Tartaglia et al. (17). In this model p75, which has a high affinity but fast off-rate for TNF, serves to concentrate low levels of TNF near the cell surface. Following TNF release from p75, ligand then binds to p55, which displays a lower ligand binding affinity (17). This model was proposed because a heterocomplex could not be demonstrated with the antibody reagents available at that time. Our study clearly demonstrates the ligand-induced formation of a receptor heterocomplex and thereby identifies a potential transient intermediate in the ligand passing process. Thus ligand passing may be the result of a direct, contact-mediated transfer of ligand from p75 to p55.
In addition, the ligand-induced interaction between p55 and p75 may have important implications for TNF signaling. Our p75 receptor shedding results indicate that p55 and p75 may be signaling in a cooperative manner, which could be accounted for by the formation of a hetero-receptor complex. In favor of this hypothesis is the recent observation that TRADD and TRAF proteins, which bind to p55 and p75 intracellular domains, respectively, are capable of interacting with one another (26). A TRADD-TRAF2 interaction has been observed at the p55 intracellular domain without the participation of p75 (27). However, the recruitment of TRAF2 to the p55-TRADD complex may be facilitated when TRAF2 can be actively recruited by p75 during heterocomplex formation with p55. The formation of p55-p75 heterocomplexes and the resulting juxtaposition of their respective intracellular domains may thus be a mechanism by which interaction between their respective signaling proteins is facilitated. It will be important to derive in vitro functional assays to examine this possibility. Unfortunately there exists at this time no in vitro model of TNF-induced receptor shedding in a murine cell line. Current efforts are underway to establish these models.