Lymphotoxin β Receptor Triggering Induces Activation of the Nuclear Factor κB Transcription Factor in Some Cell Types

NFκB is a pleiotropic transcription factor capable of activating the expression of a great variety of genes critical for the immunoinflammatory response. Tumor necrosis factor α (TNFα) and lymphotoxin α (LTα, originally TNFβ) are potent nuclear factor κB (NFκB) activators in various cell types. The LTα molecule, in addition to being secreted as a soluble trimer, can also form membrane-anchored heterotrimers with the LTβ chain, another member of the TNF family. The LTα1β2 heterotrimer binds a specific receptor, called the LTβ receptor (LTβ-R), which is also a member of the TNF receptor family. Here, we show that engagement of LTβ-R with a soluble form of LTα1β2 or with a specific anti-LTβ-R agonistic monoclonal antibody CBE11 quickly induces activation of NFκB in HT-29 and WiDr human adenocarcinomas. LTβ-R triggering activates NFκB and induces proliferation in WI-38 human lung fibroblasts. No NFκB activation is observed in human umbilical vein endothelial cells, correlating with the inability of LTβ-R activation to induce expression of NFκB-dependent cell surface adhesion molecules. Thus, like several other members of the TNF receptor family, the LTβ-R can activate NFκB following receptor ligation in some but not all LTβ-R-positive cells.

For more than a decade, the TNF-LT 1 system has been known to be composed of two related inflammatory cytokines, TNF␣ (cachectin or tumor necrosis factor ␣) and LT or LT␣ (lymphotoxin ␣, also called TNF␤) which are active as soluble trimers (1)(2)(3)(4)(5). Both ligands engage two receptors, TNF-R55 (TNF-R 60 ) and TNF-R75 (TNF-R 80 ) (5), and appeared historically to have similar biological activities. Recently, however, the discovery of membrane-bound forms of these two ligands revealed interesting differences (6). TNF␣ can be retained on the cell membrane as an uncleaved type II membrane protein, whereas LT␣ was found to be processed yet anchored to the membrane by association with a 33-kDa protein named LT␤ (6). LT␤ is a member of the TNF-LT family (7), and its gene maps to the TNF-LT locus (8). Protein biochemistry studies showed that the LT␣ chain which composes the functional LT trimer (renamed LT␣3) can also form membrane-anchored heterotrimers with the LT␤ chain, in predominantly a LT␣1␤2 form with only minor amounts of the LT␣2␤1 form (9). More-over, a specific receptor (LT␤-R) for the LT␣ and LT␤ heterotrimers with homology to the TNF receptors has been identified and is a new member of the nerve growth factor and TNF receptor family (10). Interestingly, LT␤-R is expressed in a wide range of cell types, except lymphocytes, whereas, conversely, the expression of the ligand is restricted to activated lymphocytes (8 -10). Relatively little is known about the function of LT␤-R. The aberrant development of lymph nodes observed in the LT␣ knockout mice (11,12), which is not observed in the TNF-R55 or TNF-R75 knockout mice (13)(14)(15), and the ability of soluble decoy forms of the LT␤-R to block lymph node development 2 indicates a role for surface LT␣⅐LT␤ complexes in the development of the peripheral lymphoid system. Additionally, LT␣ and LT␤ signaling appears to be required for the expression of various cell surface adhesion proteins on cells in the marginal zones bordering the white pulp regions of the spleen (16). [2][3][4] It is a reasonable assumption that the LT␤-R mediates these activities. Intriguingly, these observations suggest possible parallels between TNF and LT actions in terms of their abilities to control expression of surface proteins involved in cell-cell adhesion on endothelial cells and marginal zone cells in the spleen, respectively.
To understand the biological consequences of LT␤-R signaling, we have analyzed in depth two different areas based on the known effects of TNF signaling. First, we have shown previously that receptor activation by either soluble ligands or agonistic antireceptor monoclonal antibodies induces the death of some adenocarcinoma cell lines (17). Second, in striking contrast to the TNF-R system, LT␤-R activation was shown to have little effect on the display of adhesion molecules on endothelial cells. 5 In light of these various observations, it would be useful to investigate which signal transduction pathways are affected by LT␤-R activation. Currently, little is known about LT␤-R activation other than the observation that the cytoplasmic domain of LT␤-R, like that of CD40, can bind to TNF receptor-associated factor 3, a member of a newly defined group of signal transduction molecules (18).
NFB is a pleiotropic transcription factor that is especially involved in the regulation of the expression of numerous genes critical for the immunoinflammatory response, e.g. interleukin 6 and 8, vascular cell adhesion molecule 1, and E-selectin (reviewed in Ref. 19). Many of these genes are expressed in fibroblasts and endothelial cells and play pivotal roles in host defense. The system is also exploited by some viruses, including human immunodeficiency virus, to drive their expression (19,20). Signaling through five members of the TNF receptor family has been shown to activate NFB, i.e. TNF-R55 (21-23), TNF-R75 (24,25), CD30 (26,27), Fas (28,29), and CD40 (25,30,31). Interestingly, signaling by two members of this family, CD30 and the TNF receptors, has also been shown to enhance human immunodeficiency virus expression (26,(32)(33)(34). Recent results suggest that LT␤-R, as well, triggers human immunodeficiency virus replication in infected monocytic cells. 6 In this report we show that LT␤-R signaling leads to NF-B activation in a cell type-restricted manner.

EXPERIMENTAL PROCEDURES
Cell Lines-The human colon adenocarcinomas WiDr and HT-29 and the human fibroblasts WI-38 were obtained from the American Type Culture Collection and grown in minimum Eagle's medium-Earles' salt medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (JRH Biosciences, Lenexa, KS), 2 mM glutamine, 100 unit/ml penicillin, 100 mg/ml streptomycin, 0.5 mM sodium pyruvate, 0.5 ϫ 100 ϫ nonessential amino acid solution (BioWhittaker, Walkersville, MD) at 37°C in 5% CO 2 . A consistently responsive subclone of the HT-29 line, called HT-29-14, was used in all of our experiments. WiDr and HT-29 cells are thought to be derived from the same patient (35). Human umbilical vein endothelial cells (HUVECs) were isolated using the method of Gimbrone (36) and serially passaged under the conditions modified and described by Thornton et al. (37). Normal human dermal fibroblasts and their optimized medium were purchased from Clonetics Corp. (San Diego, CA).
Reagents-Soluble recombinant LT␣3 (LT␣3 trimer) and LT␣1␤2 molecules were prepared as described previously (38). Briefly, Sf9 insect cells were infected with two recombinant baculoviruses that encode LT␣ and a soluble version of LT␤. Culture supernatants were passed over a series of TNF-R55 and LT␤-R affinity columns to purify LT␣1␤2 and LT␣2␤1. Human LT␣3 was prepared as previously detailed (39). The various LT trimers were also prepared with a LT␣D50N mutation that eliminates TNF-R binding (38). The LT␤-R-human IgG1 fusion protein has been described previously (9). The mouse anti-human LT␤-R antibodies CBE11 and BDA8 have also been described previously (17). A control mouse IgG1 monoclonal antibody MOPC 21 was obtained from Organon Pharmaceuticals. Antibodies against p50, p65, RelB, c-Rel, and p52 NFB subunits were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Nuclear Extracts and Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared as described by Digman et al. (40) with minor modifications. Briefly, cells were harvested and washed twice with phosphate-buffered saline. 1.5 to 2 ϫ 10 7 cells were resuspended in a hypotonic lysis buffer composed of 10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , and 10 mM KCl (buffer A) on ice. After a 20-min incubation, cells were homogenized by 20 strokes with a loose-fitting Dounce homogenizer. Nuclei were collected by centrifugation for 4 min at 6500 rpm at 4°C, and nuclear proteins were extracted with 3 pellet volume of high salt buffer composed of 20 mM Hepes, pH 7.9, 25% (v/v) glycerol, 0.42 M NaCl, 1.5 mM MgCl 2 , and 0.2 mM EDTA, pH 7.5 (buffer B). After a 1-h incubation on ice, samples were centrifuged (5 min, 6500 rpm, 4°C), and the supernatants were diluted in 3 volume of low salt buffer composed of 20 mM Hepes, pH 7.9, 20% (v/v) glycerol, 0.1 M KCl, 0.2 mM EDTA, pH 7.5, and 1% Nonidet P-40 (buffer C). The samples were used immediately for EMSA or stored at Ϫ80°C. Immediately before use, buffers A-C were supplemented with 1% of a fresh 0.1 M phenylmethylsulfonyl fluoride solution in ethanol and 1% of 1 M dithiothreitol (Fluka Chemie AG, Buchs, Switzerland). For EMSA, a synthetic double strand oligonucleotide containing two tandemly arranged NFB binding sites was used as a probe (5ЈATCAGGGACTTTCCGCTGGG-GACTTTCCG3Ј; oligonucleotides used in Figs. 1B and 2A). A control oligonucleotide containing a single mutated B site was included in the assay (5ЈAGGATGGGAGTGTGATATATCCTTGAT3Ј). These oligonucleotides were used in supershift experiments and in the experiment with HUVECs. In experiments with WiDr and WI-38 cells another set of oligonucleotides was used, showing less background with these cell types: wild type B oligonucleotide (5Ј-GATCCGAGGGGACTTTC-CGCTGGGGACTTTCCAGG-3Ј) and the corresponding mutated oligonucleotide (5Ј-GATCCGAGCTCACTTTCCGCTGCTCACTTTCCAGG-3Ј; oligonucleotides used in Figs. 1A and 2C). 60,000 cpm of the 32 Plabeled synthetic sequence were incubated with 5 l of each nuclear extract (calibrated to contain 10 g of protein as determined using the Bio-Rad protein assay for 45 min at ambient temperature in the presence of 2 g of poly(dI-dC) (Pharmacia Biotech Inc.) and 30 g of bovine serum albumin (41,42).
In the case of supershift experiments, nuclear extracts were incubated 30 min with the labeled oligonuclotide at room temperature and then incubated an additional 30 min with an antibody specific for one member of the rel family (2 g/reaction mix) at 4°C before loading the samples on the gel. Samples were electrophoresed on native 4% nongradient or gradient polyacrylamide gels from Integrated Separation Systems (Natick, MA) as indicated in the figure legends.
Proliferation Assays-WI-38 cells were cultured in a 96-well plate in 100 l of medium (1 ϫ 10 4 cells/well) in the presence of varying amounts of human TNF␣, LT␣3, LT␣1␤2, a mouse anti-human LT␤-R antibody CBE11, or a control mouse IgG. After a 72-h incubation at 37°C, cells were pulsed 6 h with 1 Ci/well (in 50 l) [methyl-3 H]thymidine (Du-Pont NEN) and harvested using a Tomtec (Orange, CT) cell harvester. The radioactivity was measured in a Betaplate liquid scintillation counter (Pharmacia Biotech).
Flow Cytometry-HUVEC and WI-38 cells were detached with 5 mM EDTA in phosphate-buffered saline, and cells were washed twice with phosphate-buffered saline, 0.05% NaN 3 , and 1% bovine serum albumin (FACS buffer). Cells were incubated 30 min on ice with 10 g/ml of mouse anti-human LT␤-R (BDA8) or a control mouse IgG, washed once in FACS buffer, and incubated 30 min on ice with phosphatidylethanolamine-conjugated donkey anti-mouse Ig (Jackson Immuno Research, West Grove, PA) at 5 g/ml in FACS buffer. Cells were washed twice in FACS buffer and analyzed on a FACScan cytofluorometer (Becton Dickinson). Cells were gated using forward versus side scatter to exclude dead cells and debris.

RESULTS AND DISCUSSION
NFB activation was analyzed by EMSA using two related human adenocarcinoma cell lines, HT-29 (clone 14) and WiDr, which have been shown to be sensitive to LT␣1␤2 in a cytotoxicity assay in the presence of interferon ␥ (17). In a first set of experiments, WiDr cells in culture were stimulated for various periods of time (from 5 min to 12 h) with LT␣1␤2, CBE11 (a mouse anti-human LT␤-R, agonist antibody), or a mouse IgG as a negative control. LT␤-R signaling in WiDr or HT-29 cells resulted in a fast activation of NFB, with a maximal signal occurring after 15 min of stimulation (Fig. 1, A and B, respectively). LT␤-R triggering is specifically mediating NFB activation because, NFB was activated with the anti-human LT␤-R agonistic monoclonal antibody CBE11 and not an isotype-matched control monoclonal antibody (Fig. 1A). The kinetics of activation of NFB through LT␤-R were very similar to the kinetics observed for TNF␣, LT␣3, CD40, and Fas in other cell types (28,43,44), suggesting similar mechanisms of activation. NFB activation was transient and disappeared after 12 h of stimulation (Fig. 1A). NFB is a heterodimer composed of p50 and p65 subunits, which are proteins of the Rel-NFB family (19). Homodimeric or heterodimeric combinations with other members of this family such as RelB, c-Rel, and p52 are referred to as NFB like (19,45). The question of which NFB dimer types were activated was addressed by supershift analysis using specific rabbit antisera to the p50, p65 (RelA), RelB, c-Rel, and p52 NFB subunits or a control antibody. The NFB bands resulting from TNF␣, LT␣3 or CBE11 stimulation, were supershifted by anti-p50 and anti-p65 antibodies (Fig. 1B). The supershift for p50 appeared weaker than for p65. LT␣1␤2 also activated a p50-p65 dimer in these cells (data not shown). Similar results were seen using the same WiDr extracts, except that only the anti-p65 antibody supershifted the NFB band obtained by the same three different stimuli (data not shown). These data suggest that TNF␣, LT␣3, LT␣1␤2, and CBE11 activated predominantly the same type of NFB dimer in a given cell type, i.e. p50-p65 NFB dimer in HT-29 cells, and probably a p65-p65 dimer in WiDr, presumably using similar pathways. As was noted for Fas-induced NFB activation (29), minor amounts of some other NFB forms appear in the WiDr cells (Fig. 1A). The p65-p50 form of NFB appears to be the common dimer activated by CD40 (25,31), CD30 (26,27), Fas (28), TNF-R55 (23), TNF-R75 (25), and LT␤-R, although CD40 (31) and CD30 (26) can activate additional NFB isoforms in lymphoid cells.
Because TNF␣ and NFB play fundamental roles in regulating the endothelium and fibroblasts, we examined the ability of the various LT forms to activate NFB in HUVECs and the nontransformed human diploid fibroblast cell line WI-38. Both HUVECs and WI-38 cells express LT␤-R on their cell surface (Fig. 2B). As expected, LT␣3 activated NFB in HUVECs and WI-38 cells (Fig. 2, A and C). LT␣1␤2 and CBE11 also activated NFB in WI-38 cells (Fig. 2C) but not in the HUVECs (Fig. 2A). Activation of NFB by LT␣1␤2 in WI-38 cells was inhibited by addition of soluble human LT␤-R and not by addition of soluble human TNF-R55 (Fig. 2C), indicating a LT␤-R specific event.
Moreover, LT␣1␤2 with a LT␣ D50N mutation (LT␣1D50N␤2), which would prevent the binding of a possible LT␣3 contaminant to the TNF receptor p55, also activated NFB with an intensity similar to that of LT␣1␤2 (Fig. 2C). Anti-human LT␤-R CBE11 antibody also specifically activated NFB in WI-38 cells (Fig. 2C). Therefore, like the HT29-WiDr system, NFB can also be activated in WI-38 cells specifically by LT␤-R cross-linking.
In addition to its well defined ability to induce cell death, TNF␣ has been shown to induce the proliferation of fibroblasts (46). In light of these dual effects on cell growth and death depending on cell type, we wished to determine whether LT␤-R signaling could also affect proliferation in a fibroblastoid cell line. Proliferation assays were performed using WI-38 fibroblasts stimulated for 72 h with TNF␣, LT␣3, LT␣1␤2, CBE11, and a mouse IgG as control. As shown in Fig. 3A, TNF␣, LT␣3, LT␣1␤2, and CBE11, but not the control antibody, stimulated WI-38 cell proliferation, demonstrating that the stimulation of LT␤-R induces proliferative signals in these cells. Both wild type LT␣1␤2 and LT␣1␤2 with a LT␣ D50N mutation equally stimulated the proliferation of WI-38 cells (Fig. 3B), again indicating specific LT␤-R signaling. The proliferative effects of LT␤-R signaling on nontransformed cells appear to be cell type specific, since LT␤-R triggering has no effect on HUVEC proliferation, 5 whereas TNF␣ inhibits their proliferation (47,48).
Specific stimulation of human LT␤-R triggers NFB activation in WiDr and HT-29 cells. LT␣1␤2 and CBE11 failed to activate NFB in HUVECs, which express reasonable levels of LT␤-R, whereas TNF␣ and LT␣3 were fully potent. These observations are consistent with the inefficiency of LT␣1␤2 or CBE11 to induce the NFB-dependent expression of E-selectin or vascular cell adhesion molecule 1 in HUVECs 5 and tend to exclude the LT␤-R system as a regulatory element in the endothelium-mediated inflammatory response. These observations are also supported by the results of systemic injection of LT␣1␤2 in mice, which showed no lethality in the dose range at which TNF␣ or LT␣3 are lethal. 5 The ability of LT␤-R to activate NFB could depend, in part, on LT␤-R expression levels in the cell lines used in our experiments, which are higher in WiDr, HT-29, and WI-38 cells compared with HU-VECs (Fig. 2B). However, previous work in the TNF system had shown that variations in receptor density alone cannot explain the responsiveness of various cells to TNF␣ (1), and we suspect that also in the LT system, a lower receptor density does not account for the nonresponsive behavior of HUVECs. These NFB results and the proliferation studies indicate that both LT␤-R signaling and function are cell type specific.
These studies also reveal an interesting phenomenon. It appears that varying levels of receptor cross-linking are required for the activation of various signal transduction pathways. For example, soluble CBE11 was not very potent in killing WiDr or HT-29 cells, whereas immobilized CBE11 was fully active in the same cytotoxicity assays (1), presumably by creating a superior level of receptor cross-linking in the latter case. However, in the case of NFB activation in these cells, soluble CBE11 was active. Likewise, soluble CBE11 was significantly less active than LT␣1␤2 in the activation of NFB in WI-38 cells (Fig. 3A). These signaling differences between soluble and cell surface ligands and soluble or immobilized agonist antibodies are examples of the complexity of receptor triggering, which has been previously observed for the TNF receptor system (3,49,50). Receptor cross-linking has been shown to be critical for signaling in the TNF and Fas system (3,51,52), and a similar observation was made for LT␤-R, using Fab fragments of agonistic antibodies (17). It is possible that the signal transduction elements mediating, for example,

FIG. 1. NFB activation in WiDR and HT-29 cells analyzed by electrophoretic mobility shift assay.
A, WiDr cells in culture were stimulated for the indicated time with human LT␣1␤2, 500 ng/ml, or an agonist antibody against human LT␤ receptor CBE11, 1 g/ml. A mouse IgG1 (MOPC 21), 1 g/ml, was used as negative control. Some samples were also incubated with a labeled oligo containing a mutated NFB binding site, as indicated. Samples were run on a 4% acrylamide gel. B, HT-29 cells in culture were incubated 15 min with human TNF␣, 20 ng/ml, human LT␣3, 100 ng/ml, or CBE11, 1 g/ml. Under each group, samples indicated as unstimulated refer to cells without any treatment. Then, the samples were incubated with 1-2 g/reaction mixture of a rabbit anti-RelA, anti-RelB, anti-c-Rel, anti-p50, or anti-p52 antibody for 30 min at 4°C. Rabbit IgG was used as control. Arrows, NFB bands and supershifted bands. Samples were run on a 4 -15% gradient gel. All the data were obtained from the same gel, but the lanes were rearranged for simplicity. NFB activation versus cell death require varying levels of oligomerization to trigger. Presumably immobilized antireceptor antibody and ligand would oligomerize the receptor to a greater state than the theoretical dimers resulting from soluble monoclonal antibodies. Although this model is plausible, a completely conflicting pattern was observed with HepG2 cells, in which a monoclonal anti-TNF-R55 antibody was active, whereas TNF␣ was inactive (50). Clearly, further analysis of the events stemming from differing levels of receptor oligomerization and the varying lifetimes of the oligomerized states is required to understand the receptor activation.
No correlation appears to exist between NFB activation and cell proliferation or death (28,29,53,54). Our data are also consistent with this hypothesis, since in vitro LT␤-R-mediated death in WiDr and HT-29 cells is interferon-␥-dependent, whereas NFB activation and the ability to induce WI-38 proliferation is interferon-␥-independent. These results suggest that different intracellular mechanisms are responsible for these phenomena. On the other hand, some common signaling pathway must be used by these receptors, since stimulation with TNF␣, LT␣3, LT␣1␤2, or CBE11 leads to the activation, with similar kinetics, of the same type of NFB dimer in HT-29 or WiDr cells. Presumably, multiple signal transduction path- rithmic scale (horizontal) are presented. C, WI-38 cells in culture were stimulated for 15 min with 100 ng/ml LT␣3 or 500 ng/ml LT␣1␤2 in the presence or absence of 10 g/ml human LT␤-R-human Fc or human TNF-R55-human Fc, 500 ng/ml LT␣1D50N␤2 (LT␣1␤2 containing a LT␣ D50N mutation), and 1 g/ml CBE11 anti-LT␤-R. The analysis was performed as indicated in Fig. 1A. Samples were run on a 4% acrylamide gel. ways are initiated following receptor cross-linking. The ability of the CBE11 monoclonal antibody to inhibit WiDr tumor growth in vivo is interferon-␥-independent, as is NFB induction by this monoclonal antibody (17). This disparity between in vitro and in vivo antibody effects on tumor growth underscores the importance of understanding the various signal transduction pathways activated in these systems.
The natural LT␣1␤2 ligand is only expressed on the surface of activated lymphocytes, whereas the receptor is expressed on nonlymphoid cells (55), therefore, in vivo, activation of NFB through LT␤-R probably occurs exclusively following cell-cell contact between lymphoid and nonlymphoid cells. NFB is a ubiquitous transcription factor, and its properties are extensively exploited by the immune system (19). Cell type-specific activation of NFB might therefore represent a unique and restricted way of local communication between activated lymphocytes transiting into the lymphoid organs and inflamed tissues and their stromal environment.