Discrete signaling regions in the lymphotoxin-beta receptor for tumor necrosis factor receptor-associated factor binding, subcellular localization, and activation of cell death and NF-kappaB pathways.

Lymphotoxin-beta receptor (LTbetaR), a member of the tumor necrosis factor receptor superfamily, is essential for the development and organization of secondary lymphoid tissue. Wild type and mutant LTbetaR containing successive truncations of the cytoplasmic domain were investigated by retrovirus-mediated gene transfer into HT29.14s and in 293T cells by transfection. Wild type receptors accumulated in perinuclear compartments and enhanced responsiveness to ligand-induced cell death and ligand-independent activation of NFkappaB p50 dimers. Coimmunoprecipitation and confocal microscopy mapped the TRAF3 binding site to amino acids PEEGDPG at position 389. However, LTbetaR truncated at position Pro(379) acted as a dominant positive mutant that down-modulated surface expression and recruited TRAF3 to endogenous LTbetaR. This mutant exhibited ligand-independent cell death and activated NF-kappaB p50 dimers. By contrast, truncation at Gly(359) created a dominant-negative mutant that inhibited ligand-induced cell death and activation of NF-kappaB p50/p65 heterodimers. This mutant also blocked accumulation of wild type receptor into perinuclear compartments, suggesting subcellular localization may be crucial for signal transduction. A cryptic TRAF-independent NF-kappaB activating region was identified. These mutants define discrete subregions of a novel proline-rich domain that is required for subcellular localization and signal transduction by the LTbetaR.

The lymphotoxin ␤ (LT␤) 1 receptor (LT␤R), a member of the tumor necrosis factor receptor (TNFR) superfamily, has emerged as a signaling system required for organization of lymphoid tissue (for reviews, see Refs. 1 and 2). The LT␤R binds two distinct but related ligands, the cell surface form of LT (3) and LIGHT (4). Surface LT is composed of two subunits, LT␣ and LT␤, arranged as a heterotrimers of either LT␣1␤2 (major form) or LT␣2␤1 stochiometry (5). The LT␤ subunit, a type II transmembrane protein, provides the membrane anchor for the ligand and the specificity for binding the LT␤R. The LT␣ subunit contributes primarily to the conformation of the heterotrimer (6) but can also form homotrimers that bind the two TNF receptors, TNFR1 (55-60 kDa; CD120a) and TNFR2 (80 kDa; CD120b). The second LT␤R ligand, LIGHT, a recently identified member of the TNF superfamily (4), forms homotrimers and interacts with another TNFR family member, the herpesvirus entry mediator (HVEM or HveA) (7), which also binds LT␣.
Although these ligands show significant cross-receptor specificity, each cytokine-receptor system plays distinct physiologic roles. Based on gene deletion studies, LT␤R, but not TNFR, is required for the differentiation of secondary lymph organs, Peyer's patches, and lymph nodes (8). More recent evidence indicates that progenitor cells crucial for the development of natural killer cells and dendritic cell compartmentalization require the LT␣␤-LT␤R system (9 -11). LT␤R signaling also acts in concert with the TNF/TNFR1 system for the organization of peripheral lymphoid tissue during immune responses (12)(13)(14)(15).
Signal transduction by the TNF receptors is initiated by the binding of specific trivalent ligands that induce aggregation of the receptors, which in turn recruit cytosolic proteins involved in the propagation of signals (16). LT␤R interacts with TNF receptor-associated factors (TRAFs), a family of zinc RING finger proteins with a C-terminal region that binds directly to the cytoplasmic tail of LT␤R and related receptors, such as CD40, CD30, and TNFR80 (17,18). LT␤R binds TRAF2, -3, -4, and -5, but not TRAF6 (19 -22). Binding of soluble recombinant LT␣1␤2 or anti-LT␤R antibodies rapidly induces the formation of a stable complex between TRAF3 and LT␤R (19). Forced overexpression of these receptors leads to aggregation and activation of signaling pathways independent of ligand, indicating the presence of regulatory mechanisms that normally limit receptor expression or spontaneous aggregation. In cell culture models, signaling through LT␤R induces cell death of certain adenocarcinoma tumor cells (23) and gene expression by activation of the p50/p65 form of nuclear factor B (NF-B) (19,24), a transcription factor involved in controlling expression of proinflammatory molecules, including chemokines (25) and integrins (26,27), and protection of cells from apoptotic death (28). Cell death and NF-B pathways bifurcate at the level of LT␤R-TRAF binding, since TRAF3 mutants block cell death signaling but not NF-B activation (19,29). TRAF2, -5, and -6 activate the NF-B pathway by members of the TNFR superfamily, and a common binding site for TRAF2, -3, and -5, the PVQET sequence, has been identified in CD40 (30,31), but this site is not readily apparent in the LT␤R. Identification of the regions in LT␤R involved in binding TRAF proteins will aid in understanding the mechanisms of signal propagation. Furthermore, it is unclear whether TRAFs mediate all of the signaling activities of this receptor, because recent findings show that mice with gene deletions in TRAF2, -3, or -5 contain a normal complement of lymph nodes (32)(33)(34).
We have identified the structural regions of the LT␤R required for TRAF binding and initiation of signaling that activates cell death and gene transcription pathways. A panel of LT␤R mutants was characterized that reveal several novel features of LT␤R signaling including three discrete regions within a short proline-rich sequence that control TRAF binding and receptor compartmentalization and regulate cell death and NF-B activation. These mutants should provide useful tools for dissecting the molecular components and pathways involved in physiologic roles dependent on the LT␤R.

MATERIALS AND METHODS
Cells and Reagents-Recombinant human TNF (35) and soluble LT␣1␤2 (36) produced with a truncated version of LT␤ lacking the cytosolic and transmembrane domains were provided by Jeffrey Browning (Biogen, Inc.). Mouse anti-Fas monoclonal antibody (mAb) CH11 (IgM) was obtained from MBL (Nagoya, Japan). M2 and M5 anti-Flag (IgG 1 ) mAb were obtained from Sigma. The anti-c-Myc monoclonal antibody 9E10 (IgG 1 ) was obtained from BABCO. Mouse anti-LT␤R antibodies, BDA8 (IgG 1 ) and BK11 (IgG 1 ) were provided by Biogen, Inc. HT29.14s is a clone of the HT29 adenocarcinoma cell line sensitive to death-inducing effects of LT and related cytokines (23). Human embryonic kidney-293 cell line and its derivative 293T cells, which express the SV40 large T antigen, were obtained from the American Type Culture Collection. All cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum. All lines routinely tested negative for mycoplasm using a PCRbased assay, the Mycoplasma Primer Set (Stratagene, La Jolla, CA).
Plasmid Construction-Retroviral plasmids directing the expression of the C-terminal truncated mutants of LT␤R were constructed by PCR using c-Myc-LT␤R-pBABE as a template (29). All deletion mutants were constructed using the common 5Ј-primer containing a BglII site, 5Ј-GACGAGAGATCTCTGGCTTCAGGAGCTGAATA-3Ј. A different 3Јprimer was used for each deletion mutant. Each 3Ј-primer includes a stop codon and a SalI site. The 3Ј-primers used were as follows: ⌬418, PCR was performed under the following conditions (Perkin-Elmer 9600 thermocycler): an initial 4 min at 94°C and then 30 cycles of 30 s at 94°C, 30 s at 60°C, and 80 s at 72°C followed by a final cycle for 10 min at 72°C. PCR products were generated using 30 ng of template plasmid and 5 units of Pfu DNA polymerase (Stratagene, La Jolla, CA). PCR products were purified using the Wizard PCR DNA Purification System (Promega, Madison, WI), digested with BglII and SalI, gelpurified and ligated into the BamHI and SalI sites of pBABEpuro. The integrity of the c-Myc-LT␤R coding region in each of the pBABEpuro deletion constructs was confirmed by sequencing both strands (ABI PRISM Dye Termination Cycle Sequencing kit) with an automated sequencer (ABI PRISM 310 genetic analyzer).
Moloney retroviral vectors were produced by transfection of the NX amphotrophic packaging cell line with the desired pBABE-derived construct as described (29). For production of control virus, NX cells were transfected with the empty pBABEpuro vector. Virus-containing supernatants were harvested after 48 h of transfection. Infection of the HT29.14s and 293 cell lines was performed as described previously (29).
Infected cells were selected in puromycin, and all assays described were performed within 1-2 weeks following selection.
Flow Cytometric Analysis-HT29.14s cells were detached from the culture vessel using 20 mM EDTA in phosphate-buffered saline (PBS), washed once in PBS, and then incubated with mouse isotype IgG control antibody, mouse anti-c-Myc epitope (9E10, IgG 1 ), or anti-human LT␤R (BKA11, IgG 1 ) at 10 g/ml in 50 l/well in cold binding buffer (PBS with 2% fetal bovine serum, 0.05% sodium azide) for 30 min on ice. After washing three times with cold binding buffer, 50 l of goat F(abЈ) 2 anti-mouse IgG conjugated to R-phycoerythrin (Southern Biotechnology Associates, Inc.) was added to wells and incubated for 30 min on ice. Cells were then washed as above and analyzed by flow cytometry (FACScan, Becton Dickenson).
Cell Viability Assay-Cell death in response to cytokines or antibodies was determined by a mitochondria dye reduction assay using 3-(4, 5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, which detects viable cells as described (37). After 3 days, the percentage of cell viability was calculated as a ratio of the dye absorbance (570 nm) by cells cultured with cytokines or antibodies to medium alone. After 3 days in culture, the A 570 for individual lines in medium alone ranged from 1.1 to 1.6. Statistical analysis for calculations of IC 50 and significance was calculated with GraphPad Prism and Instat software (San Diego, CA). Data are presented as the mean Ϯ S.D. of quadruplicate wells.
NF-B Activation-NF-B DNA binding interactions were performed by a electrophoretic gel shift assay as described (38) with modification (20) using the B site in the human immunodeficiency virus-1 enhancer. The composition of the activated NF-B complex was examined by supershift analysis with antiserum to Rel family members (Santa Cruz Biotechnology).
NF-B-dependent transcription was measured using a luciferase reporter construct (39). 293T cells seeded at 5 ϫ 10 5 cells/35-mm well were transfected by the CaPO 4 method. Briefly, 250 l of precipitate containing 4 g of LT␤R expression plasmid and 0.25 g of a NF-Bluciferase reporter plasmid containing two copies of the consensus B binding motif was added to the cells for an overnight incubation. Cells were washed once, and fresh growth medium was added for a 24-h incubation. Cells were extracted in 250 l of lysis buffer (25 mM Trisphosphate, pH 7.8, 2 mM dithiothreitol, 2 mM EDTA, 10% glycerol, 1% Triton X-100), and nuclei were removed by centrifugation at 7000 ϫ g. Immunofluoresence Confocal Microscopy-Twenty-four hours posttransfection, 293T cells were seeded in eight-well chamber slides (Lab-Tek) at 3 ϫ 10 4 cells/well and cultured for 18 -36 h at 37°C. For staining, wells were washed twice with PBS, fixed for 10 min at room temperature in freshly prepared 2% paraformaldhyde in PBS, pH 7.0, washed twice with PBS, and then permeabilized in methanol for 2 min at room temperature. Cells were washed in PBS and then blocked for 10 min at room temperature in PBS containing 3% BSA. Polyclonal goat anti-LT␤R TgG (19), diluted to a final concentration of 20 g/ml, and mouse anti-FLAG M2 to detect TRAF3 were diluted in PBS containing 3% BSA and 0.2% Triton X-100 (PBS/BSA/Triton). Primary antibodies were added to the wells to a final volume of 120 l/well and incubated in a humidified chamber at room temperature for 1 h. Wells were then washed three times in PBS/BSA/Triton buffer. Fluorescein isothiocyanate-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories) in combination with Texas Red-conjugated donkey antigoat IgG (Jackson ImmunoResearch Laboratories), were diluted to a final concentration of 1:200 in PBS/BSA/Triton in a final volume of 120 l/well. Slides were incubated in a humidified chamber at room temperature in the dark for 1 h and then washed three times in PBS/BSA/ Triton. The slides were mounted in 80% glycerol in PBS, sealed, and kept at 4°C in the dark for 1-7 days before visualization. Cells were observed with a Bio-Rad MRC-1024 confocal microscope with a krypton/ argon ion laser and a 60ϫ Nikon objective. Images were acquired using the LaserSharp operation system and were analyzed and manipulated in Adobe PhotoShop. Empty vector-transfected cells or cells stained with normal goat serum or mouse IgG isotype control were used for negative controls. Neither control exhibited background staining. Representative staining patterns were based on counting 200 cells.
LT␤R Structural Prediction-Secondary structural prediction for the LT␤R cytoplasmic domain was compiled by the Protein Sequence Analysis System (available on the World Wide Web) (40). The analysis was conducted on a subsequence of the LT␤R, residues Thr 242 -Asp 435 Cterminal of the transmembrane domain, using a type-1 discrete statespace model. The model assumes that the cytoplasmic domain is monomeric, single-domain, globular, and water-soluble and folds into well defined structural domains that do not pack against a hydrophobic surface.

Structural Requirements for LT␤R Expression and
Recruitment of TRAF3-The cytoplasmic domain of LT␤R is 194 amino acids in length and is predicted to belong to the ␣-␤ structural superclass (probability ϭ 0.9797) (Fig. 1, A and B). The polypeptide is predicted to emerge from the membrane as a stretch of ϳ120 residues that fold into three discrete helices, interspersed by ␤ strands, which precedes a proline-rich stretch (38%) of 36 residues (Pro 367 -Pro 403 ), likely to assume an elongated or kinked pedicle. The pedicle is followed by another ␤-strand conformation and a fourth helix at the C terminus. Glutathione S-transferase fusion proteins of the LT␤R cytoplasmic domain previously indicated that the binding site(s) for TRAF2, -3, and -5 and Hepatitis C virus core protein are all located within residues 338 -395, which spans the ␤-strand and pedicle regions (Ref. 41 and data not shown).
A series of mutants were constructed to determine the subregions of the LT␤R involved in TRAF binding and cellular responses. LT␤R mutants were made that successively truncate the C terminus through the pedicle and into the ␤-strand region and incorporate an N-terminal c-Myc epitope tag to distinguish mutant from endogenous receptor (Fig. 1C). HT29.14s cells trans-infected with retrovirus vectors expressing the c-Myc-LT␤R deletion mutants revealed striking differences in cell surface expression (Fig. 2). Cells expressing wild type c-Myc-LT␤R showed an ϳ2-3-fold increase over endogenous levels of surface LT␤R as estimated from the difference in specific fluorescence staining between anti-c-Myc and anti-LT␤R mAb used at saturating levels. Deletion through the C-terminal 36 residues (⌬389) showed no significant change in surface expression; however, the ⌬379 mutant expressed little or no staining by anti-c-Myc. Furthermore, ϳ90% of the endogenous LT␤R was also lost from the surface of ⌬379 mutant-FIG. 1. Structural features of the LT␤R and mutants. A, sequence of the human LT␤R cytoplasmic tail. The arrow indicates the region responsible for TRAF and hepatitis C virus core protein interactions. Putative protein kinase C (*) and protein kinase A (**) phosphorylation sites are shown. B, predicted secondary structure. Contour map of the predicted folds of the LT␤R cytoplasmic domain as calculated by the Protein Sequence Analysis System. The contour lines show probability increments of 0.1. C, LT␤R mutants. Mutants were constructed in a retrovirus expression vector by systematically deleting amino acid residues from the C-terminal cytoplasmic tail of the receptor. Each of these mutants contains an intact extracellular (Ecto) and transmembrane (TM) domain with an N-terminal c-Myc epitope tag. The LT␤R deletions are indicated by ⌬ followed by the initial deleted amino acid. expressing cells. No evidence was obtained that shedding accounted for the decrease in LT␤R expression on the cell surface, indicating that the protein is probably retained intracellular. By contrast, deletion of a further 10 residues (⌬369) led to increased expression on the cell surface of mutant and endogenous LT␤R. The ⌬359 mutant exhibited an ϳ5-fold increase in cell surface expression relative to the ⌬369 mutant and ϳ3-fold above wild type receptor.
In unmodified HT29.14s cells, TRAF3 co-immunoprecipitates with the LT␤R after brief treatment of cells with ligand (19). In contrast, precipitation with anti-c-Myc revealed that TRAF3 was specifically associated with wild type c-Myc-LT␤R independently of LT␣1␤2 (Fig. 3A). This result demonstrates that the modest overexpression of c-Myc-LT␤R in this cell type is sufficient for ligand-independent TRAF3 recruitment. The ⌬415 and ⌬403 mutants also associated with TRAF3; however, TRAF3 binding was absent in the ⌬345, ⌬359, and ⌬369 mutants. The ⌬415 and ⌬403 mutants also associated with TRAF3. However, the ⌬379 mutant co-immunoprecipitated with a species of TRAF3 that migrated as a doublet of a lower molecular mass (ϳ48 -55 kDa), indicating that TRAF3 is specifically affected by the ⌬379 mutation. The nature of this form of TRAF3 is unknown, but it could represent a proteolytic fragment among other possibilities. The analysis of TRAF3 binding was extended to HEK293 cells, including two additional mutants, ⌬396 and ⌬389, which showed that TRAF3 binding was specifically lost in ⌬389 mutant (Fig. 3B). This result locates the TRAF3 binding site to residues 389 PEEGDP.
Expression of LT␤R in HEK293 cells monitored by Western blot (Fig. 3C) revealed a single 69-kDa band for wild type receptor and proportionally smaller forms through mutant ⌬403. Mutants ⌬396 and ⌬389 were resolved as a tight doublet in which the smaller form became predominant in mutants ⌬379 and ⌬369. This suggests that post-translation modification (e.g., glycosylation) of the receptor was affected by the ⌬379 mutation. The ⌬359 and ⌬345 mutants were not detected by blotting, although expression was readily detected on the cell surface by fluorescence staining, indicating that the overall abundance of these two mutants was decreased.
The loss of endogenous LT␤R from the surface of HT29 cells and the coimmunoprecipitation of LT␤R and TRAF3 in the absence of ligand prompted us to investigate the subcellular location of these mutants. LT␤R and TRAF3 were investigated in the 293T cell line, which does not express detectable cell surface LT␤R. Following transfection of the LT␤R mutants into 293T cells, fluorescence staining analysis showed a pattern of expression similar to transinfected HT29.14s or HEK293 cells (data not shown). Confocal imaging revealed that the majority of the wild type LT␤R and mutants through ⌬369 accumulated as large clusters in perinuclear compartments (Fig. 4, a-f), and the diminished staining of cells that were fixed but not permeabilized indicated an intracellular location (data not shown). By contrast, ⌬359 exhibited a diffuse, primarily surface-staining pattern, indicating that 360 NIYIYNGPVL 369 is crucial for localization to these vesicles (Fig. 4g). TRAF3 expressed by itself exhibited a diffuse cytoplasmic staining in 293T cells, but when coexpressed with wild type LT␤R, it localized to the same perinuclear compartments, a pattern that was not altered by deletion of the LT␤R through ⌬396 (Fig. 4, h-j). As expected, TRAF3 failed to co-localize with the ⌬389 mutant and was dispersed throughout the cytosol; however, LT␤R remained in the perinuclear compartment (Fig. 4, k-n). Together, these results indicate that the TRAF3 binding site is defined by the ⌬389 mutant, and this region is distinct from the region (⌬359) controlling subcellular compartmentalization of the LT␤R. Interestingly, lysosome markers, cathepsin D or LAMP-1, and the endoplasmic reticulum marker AP-1 did not co-localize with the LT␤R (data not shown). This result indicates that these LT␤R/TRAF3-associated vesicles are unlikely to represent obstructed ER due to overexpression, nor do they appear to be a degradative end point. Further characterization of the subcellular compartments containing LT␤R is in progress.
That the ⌬379 mutant failed to co-localize or bind directly to TRAF3 in 293T cells but co-immunoprecipitated with TRAF3 in HT29.14s cells indicates that an indirect mechanism allows TRAF3 to associate with the ⌬379 mutant. A likely possibility is that ⌬379 mutant associates with endogenous LT␤R, which is complexed with TRAF3. This predicts that the ⌬379 mutant should coimmunoprecipitate in a complex with wild type LT␤R and TRAF3. To test if this association occurs, 293T cells were cotransfected with c-Myc-LT␤R⌬379, TRAF3-FLAG, and wild type LT␤R lacking an epitope tag. As predicted, the lysates subjected to immunoprecipitation with anti-c-Myc showed specific co-immunoprecipitation of TRAF3 by its association with c-Myc-LT␤R⌬379 only in the presence of wild type LT␤R (Fig.  5A). This association was also visualized by confocal microscopy (Fig. 5B), where TRAF3-FLAG colocalized with ⌬379 in Fig. 4. Co-localization of LT␤R and TRAF3. 293T cells were transfected with a total of 2 g of DNA/35-mm well as follows. a and h, empty vector; b and i, wild type LT␤R; c and j, ⌬396; d and k, ⌬389; e and l, ⌬379; f and m, ⌬369; g and n, ⌬359. Cells were cotransfected with empty vector (panels a-g) or with TRAF3-FLAG (panels h-n). LT␤R was stained with goat anti-LT␤R and TRAF3-FLAG with mouse anti-FLAG M2. Texas Red-conjugated donkey anti-goat IgG (red) and fluorescein isothiocyanate-conjugated donkey anti-mouse IgG (green) were used to visualize their respective antigens. Yellow indicates colocalization.
the perinuclear compartment only in the presence of wild type LT␤R. We further utilized this transfection system to analyze the effect of the ⌬359 dominant negative mutant. Confocal microscopy revealed that not only does the presence of ⌬359 inhibit TRAF3 from colocalizing with wild type LT␤R, but ⌬359 also inhibited the accumulation of wild type receptor in the perinuclear compartments.

The Effect of Dominant Positive and Negative Mutants of LT␤R on Cell Death
Signaling-HT29 cells expressing the c-Myc-LT␤R deletion mutants were treated with LT␣1␤2 or anti-LT␤R mAb, (CBE11), and cell viability was determined after 3 days. In this model, treatment of the cells with IFN-␥ is essential for apoptotic cell death induced by LT␣1␤2, TNF, or Fas (23,46). Additionally, the anti-LT␤R mAb CBE11 added in the soluble phase is normally not directly cytotoxic to HT29.14s cells unless combined with an additional anti-LT␤R antibody (5). Treatment with IFN-␥ of HT29.14s cells transduced with empty vector resulted in slight growth enhancement when compared with cells in medium, although together with LT␣1␤2 (1 nM) it induced a 50% decrease in cell viability, whereas LT␤R antibody was not cytotoxic (Fig. 6). By contrast, HT29.14s cells expressing wild type c-Myc-LT␤R responded to IFN-␥ treatment in the absence of LT␣1␤2 with ϳ30% loss of cell viability when compared with medium-treated controls. However, a substantial decrease in cell viability occurred following treatment with LT␣1␤2 or anti-LT␤R mAb. The ⌬410 and ⌬403 mutants responded to LT␣1␤2 treatment similar to wild type LT␤R. In contrast, cells expressing ⌬379 were exquisitely sensitive to treatment with IFN-␥ alone, displaying a Ͼ85% loss in cell viability. The effect of IFN-␥ and anti-LT␤R mAb was lost with a further 10-residue deletion (⌬369), indicating that the sequence 369 LGGPPGPGDL 378 is required for ligand-independent cell death response to IFN-␥. Indeed, the ⌬369 mutant displayed a phenotype similar to the cells transduced with empty vector control, suggesting that this mutant is inactive and that the cell death response due to ligand alone occurs via endogenous LT␤R. This is supported by the previous finding that ⌬369 itself does not bind TRAF3. In contrast to the ⌬369 mutant, ⌬359 and ⌬345 mutants were nonresponsive to LT␣1␤2 (Fig. 6).
The ligand-independent cell death induced by ⌬379, together with the finding that this mutant down-regulates endogenous LT␤R and indirectly co-immunoprecipitated with TRAF3, defines the behavior predicted for a dominant positive mutant. On the other side, the ⌬359 mutant acts as a dominant negative, perhaps due to increased cell surface expression relative to wild type receptor or by blockade of wild type receptor trafficking into perinuclear compartments or association of wild type LT␤R with TRAF3.
Ligand-dependent and -independent Activation of p50 and p50/p65 NF-B Complexes by the LT␤R-That NF-B activation by the LT␤R is not inhibited by dominant negative TRAF3 mutants indicates that signaling bifurcates at the level of the receptor (19). The c-Myc-LT␤R deletion mutants were examined for their ability to activate NF-B by electrophoretic mobility shift or NF-B-dependent reporter assays. HT29.14s cells expressing wild type c-Myc-LT␤R displayed constitutive NF-B binding to the human immunodeficiency virus long terminal repeat B site, in contrast to empty vector-infected cells, which required treatment with anti-LT␤R antibody (Fig. 7A, upper  panel). The constitutive B binding complex completely shifted with antibodies specific for the 50-kDa subunit (p50, NF-B) but not the 65-kDa subunit of NF-B (p65, RelA) (Fig. 7B,  upper panel). The constitutive p50 complex probably represents dimers of p50 (herein referred to as p50 dimers) but may also form a complex with members of the Rel family other than p65. Constitutive activation of p50 by c-Myc-LT␤R was lost in the ⌬396, indicating that the region controlling activation of p50 dimers is adjacent to the TRAF binding region. As expected, the ⌬379 mutant was active, which further supports the idea that this mutant acts as a dominant positive by activating endogenous receptors. The deletion mutant ⌬359 did not stimulate formation of constitutive p50 dimers.
Treatment of empty vector-infected HT29.14s with anti-LT␤R mAb resulted in the formation of NF-B binding complexes (resolved as a tight doublet) (Fig. 7A, middle panel), previously shown to supershift with antibodies to p65 and p50 (19). By contrast, anti-LT␤R treatment of cells expressing wild type c-Myc-LT␤R revealed an additional B band (compare lanes with vector and wild type in Fig. 7A, middle panel). Anti-p65 shifted the migration of the upper two complexes, demonstrating the presence of p65 subunit, but had no affect on the fast migrating band (Fig. 7B, lower panel). A nearly complete mobility shift of these bands occurred with anti-p50 antibody (the small fraction of residual band may represent p65 homodimers). This result indicates that both p50/p65 heterodimers and p50 dimers can be activated in HT29.14s cells by LT␤R signaling. Note that treatment of control HT29.14s cells (empty vector alone) with anti-LT␤R did not activate the p50 within this short time frame. Furthermore, the presence of p50 in anti-LT␤R-activated cells was dependent on the same region (⌬396) of the receptor as unstimulated cells (Fig. 7A, upper panel). Together, these results indicate that the pattern of NF-B bands in HT29.14s cells stimulated with anti-LT␤R is a composite of the rapidly induced p65/p50 dimers and constitutively formed p50 dimers. Analysis of the B bands in cells after treatment for 15 min with anti-LT␤R showed that formation of p50/p65 heterodimer was inhibited only by the ⌬359, further establishing this mutant as a dominant negative that inactivates the function of the endoge-nous LT␤R (Fig. 7B, middle panel).
As expected, none of these LT␤R mutations ablated the formation of p50/p65 complexes induced by TNF (Fig. 7A, lower  panel), and TNF had no detectable effect on the presence of the p50 dimers. However, TNF treatment of HT29.14s cells expressing the wild type c-Myc-LT␤R resulted in an enhanced activation of the p50/p65 NF-B complex (ϳ4-fold). TNF treatment of the remaining c-Myc-LT␤R deletion mutants, including the dominant negative mutant c-Myc-LT␤R⌬359, also resulted in enhanced activation of the p50/p65 NF-B heterodimer. These data suggest that TNF-induced activation of NF-B may cooperate with the LT␤R. Even the ⌬359 mutant, which is incapable of activating NF-B in response to LT␤R ligation, can contribute to the enhancement of NF-B activation by TNF, suggesting that additional sequences in the N-terminal proximal region are responsible for this NF-B enhancing activity.
The relatively complex pattern of NF-B activation in HT29.14s cells prompted us to examine the effect of these mutants in 293T cells. Expression of wild type receptor and mutants ⌬418 through ⌬396 conferred NF-B activation as measured by NF-B-dependent luciferase reporter (Fig. 7C). That ⌬396 was functional in this assay but not in HT29.14s is surprising. It is possible that overexpression in 293T can compensate for this mutation. The ⌬389 mutant, which deletes the TRAF3 recruitment domain, was inactive. However, significant activation of NF-B occurred with further truncation of the LT␤R including ⌬379, ⌬369, and ⌬359, although with a less robust signal, but was lost with the ⌬345 mutant. The lack of a dominant negative effect of ⌬359 in this system, as well as the absence of endogenous LT␤R for the ⌬379 to act through, implicates a TRAF-independent mechanism of NF-B activation by the LT␤R. This result indicates the presence of a cryptic NF-B activation site that is normally inhibited by the sequence defined by the ⌬389 mutant and confirms our suspicion of an additional region that can activate NF-B. The components involved in this NF-B pathway are currently being explored.

DISCUSSION
The LT␤R deletion mutants analyzed here define a subregion of the LT␤R cytoplasmic domain between Leu 359 and His 403 that functions as a key structural element for receptor compartmentalization and signal transduction (Fig. 8). This sequence encompasses a proline-rich region that is predicted to have an elongated or kinked conformation. Several of the deletion mutants were informative in that they defined discrete sequences with distinct functions including subcellular compartmentalization, TRAF binding, and regulation of cell death and activation of NF-B. The use of distinct cellular models was invaluable in realizing the effects of these mutations. One significant difference is that HT29.14s cells express endogenous LT␤R, whereas 293T cells do not, and this difference clearly affected the behavior of the ⌬379 and ⌬359 mutants.
Two distinct binding motifs for TRAF2, -3, and -5 are found in TNFR family members: PXQET in CD40 (PXQX(T/S) consensus motif in CD27, CD30, 4-1BB, OX40, and Epstein-Barr virus oncoprotein LMP-1) and SKEEC in TNFR2 (a similar motif is also in herpesvirus entry mediator and CD30) (18). Recent crystallographic studies of TRAF2 in complex with TRAF binding peptide from either CD40 (42) or TNFR2 (43) reveal the residues in these peptides that contact TRAF2 are distinct, although the binding affinity and conformation of the peptides are quite similar. The TRAF3 binding region in the LT␤R was localized to the sequence 389 PEEGDP, which is distinct from both CD40 and TNFR2 motifs. In preliminary results, mutation of both glutamate residues (Glu 391 -Glu 392 ) in the LT␤R was necessary to reduce TRAF3 by 80%, as well as TRAF2 and TRAF5 binding, suggesting that additional residues contribute to the LT␤R interaction with TRAF3. Additionally, the activation of NF-B p50 dimers, presumably a TRAF2or TRAF5-dependent process, required the sequence 396 PPGL-STH adjacent to the TRAF3 binding site, which suggests that the TRAF binding motif in the LT␤R may be more complex than the motifs in CD40 or TNFR2.
The ⌬379 mutant expressed in HT29.14s cells behaved as a dominant positive mutant that induced ligand-independent cell death and activation of NF-B in the absence of a functional TRAF binding site. As shown by cotransfection with wild type receptor, ⌬379 formed complexes with wild type LT␤R, which recruited TRAF3, and caused colocalization to the same subcellular compartment. Thus, the ⌬379-HT29.14s cells required only the requisite signal from IFN-␥ to induce cell death, indicating that the ⌬379-wild type complex is actively signaling. Here, we would envision that the ⌬379 sequence may provide a binding site for a regulatory protein that normally blocks spontaneous aggregation, perhaps analogous to SODD for TNFR1 (44,45). In this regard, expression of wild type LT␤R in HT29.14s cells, although only modestly increasing surface expression, enhanced the responsiveness to cell death, recruitment of TRAF3, and NF-B activation. These findings are consistent with the idea that exceeding a certain threshold of receptor density increases the probability that receptors will spontaneously aggregate and initiate limited signal transduction in the absence of ligand. Indeed, modest overexpression was sufficient for activation of p50 complexes of NF-B but not the p50/p65 complex. Furthermore, HT29.14s cells still required ligand to induce cell death, indicating that additional mechanism(s) prevented full receptor activation, which implicates the region spanning 379 PATPEPPYPI as a critical regulatory sequence.
The enhanced responsiveness of HT29.14s cells expressing wild type LT␤R or the other mutants (except for ⌬359) and the insensitivity of 293T cells to death induced by LT␤R precluded an unambiguous test to define the role of the TRAF binding site in the cell death pathway. However, previous results with dominant negative forms of TRAF3 indicate that recruitment and oligomerization of TRAF3 is important to specifically activate the LT␤R death pathway (29,46). How TRAF3 propagates the signal to the death pathway remains to be elucidated.
The ⌬359 mutant functioned as a dominant negative mutant in HT29.14s cells that suppressed the signaling action of endogenous LT␤R for cell death and NF-B activation. This mutant was expressed at increased levels on the cell surface, potentially mediating its dominant negative effect as a decoy receptor (i.e. retaining ligand binding but lacking signaling capacity). An alternative or contributing mechanism is suggested from confocal microscopy, which showed that the ⌬359 mutant effectively blocked accumulation of wild type LT␤R into the perinuclear compartments. This result implies that entry into this compartment may be necessary for signal transduction. That both dominant positive and negative mutants can interact with wild type receptors indicates that LT␤R contains a self-association domain that is membrane-proximal to ⌬359. Recent studies by Wu et al. (46)   Competition gel shift assays and supershift assays were performed with nuclear extracts (on samples used in A) isolated from HT29.14s cells expressing wild type c-Myc-LT␤R. As controls, NF-B-shifted complexes were incubated with a 10-fold molar excess of unlabeled NF-B (NFB) or mutated NF-B oligonucleotide (Mut) as indicated at the top. NF-B complexes were preincubated with antibody specific for either p65 or p50 subunits of NF-B prior to electrophoresis. Bands representing the p50 dimers and p50/p65 heterodimers of NF-B are indicated by arrows to the right. C, LT␤R mutants activate NF-B-dependent luciferase reporter. 293T cells were transfected with 4 g of the indicated LT␤R expression plasmid and 0.25 g of the NF-B-luciferase plasmid. Cells were lysed ϳ36 h after transfection, and the -fold activation in luciferase activity was calculated relative to transfection of 0.25 g of NF-Bluciferase plasmid plus 4 g of empty expression vector. from gene knockouts challenges this idea. TRAF2-, TRAF3-, or TRAF5-deficient mice exhibit secondary lymphoid tissue development, a phenotype that is observed by deletion of LT␤R (8). Although TRAF6 does not appear to bind LT␤R, mice deficient in TRAF6 fail to develop lymph nodes, a phenotype that is thought to be linked to the osteoclast differentiation factor (OPGL/RANK) pathway (47). The alymphoplasia (aly) mouse, which lacks lymph nodes, results from mutant NF-B-inducing kinase (48). TRAF2 and -5 can act as adapters coupling receptors to NF-B-inducing kinase, which then activates IB kinases, leading to activation of NF-B. TRAF2 and TRAF5 could be functionally redundant as adapters involved in signaling lymph node development, and here, a double knockout should reveal the expected phenotype. The finding that ⌬379, ⌬369, and ⌬359 initiated NF-B-dependent transcription in 293T cells suggests that another TRAF-independent NF-B-activating pathway is operative in the LT␤R. This effect was revealed only after truncation of the TRAF binding site (defined by ⌬389), implicating TRAF3 as a possible inhibitor of this cryptic NF-B-activating site. The possibility exists that this cryptic TRAF-independent, NF-B activation mechanism is involved in activating NF-B-inducing kinase and subsequent signaling for lymphoid organogenesis.
Different subregions of the LT␤R appear to be responsible for activating distinct forms of NF-B. The 396 PPGLSTPH sequence is crucial for the activation of constitutive p50 complexes as defined by the ⌬396 mutant. By contrast, the NF-B luciferase assay indicated that ⌬389, which lacks the TRAF3 binding sequence, was essential for activating NF-B. Because these sites are adjacent, the mutations may have disrupted the full site, creating high and low affinity sites not revealed by current TRAF binding assays, and thus the loss of constitutive p50 activation may reflect the need for a higher affinity interaction. The significance of LT␤R activation of p50 dimers is unclear, because p50 partners with several other Rel family members that can act as either suppressors or activators of transcription in tissue-specific contexts (49). However, distinct p50 complexes formed have different transactivating potentials that may in turn contribute to activation of specific subsets of genes.
The investigation of these mutants has revealed several new features of the LT␤R signal transduction pathways that illuminate the molecular steps involved in physiologic roles of this receptor.