Herpesvirus Entry Mediator Ligand (HVEM-L), a Novel Ligand for HVEM/TR2, Stimulates Proliferation of T Cells and Inhibits HT29 Cell Growth*

Herpesvirus entry mediator (HVEM), a member of the tumor necrosis factor (TNF) receptor family, mediates herpesvirus entry into cells during infection. Upon overexpression, HVEM activates NF-κB and AP-1 through a TNF receptor-associated factor (TRAF)-mediated mechanism. Using an HVEM-Fc fusion protein, we screened soluble forms of novel TNF-related proteins derived from an expressed sequence tag data base. One of these, which we designated HVEM-L, specifically bound to HVEM-Fc with an affinity of 44 nm. This association was confirmed with soluble and membrane forms of both receptor and ligand. HVEM-L mRNA is expressed in spleen, lymph nodes, macrophages, and T cells and encodes a 240-amino acid protein. A soluble, secreted form of the protein stimulates proliferation of T lymphocytes during allogeneic responses, inhibits HT-29 cell growth, and weakly stimulates NF-κB-dependent transcription.

One member of this family, HVEM, also known as TR2 (23) and ATAR (24), was shown to be used as a co-receptor for herpesvirus infection (22). Despite a very short cytoplasmic domain, overexpression studies suggested that it may activate NF-B and AP-1 signaling pathways through engagement of TRAFs, in particular TRAF2 and TRAF5 (24,31). Furthermore, a soluble form of the receptor, HVEM-Fc, was shown to inhibit a mixed lymphocyte reaction, suggesting a role for this receptor or its ligand in T lymphocyte proliferation (23).
With the exception of the nerve growth factor receptor, all of the known ligands for these receptors share significant sequence identity to TNF and are therefore believed to share a similar structure. TNF and its closest homologues form trimers and are believed to signal by bringing together two or more receptors as suggested by the co-crystal structure of the ligand/ extracellular domain (32) and by the discovery that receptor monoclonal antibodies are often agonists. In order to further study the biological role of HVEM, we looked for potential ligands among novel TNF-like molecules discovered in an EST data base. A soluble form of one of these novel ligands, HVEM-L (for HVEM ligand), was shown to bind to a fusion protein of the extracellular domain of HVEM, HVEM-Fc, and subsequently this interaction was shown for membrane forms of the receptor as well. HVEM-L was further shown to activate NF-B, stimulate the proliferation of T lymphocytes, and inhibit growth of the adenocarcinoma, HT-29.

EXPERIMENTAL PROCEDURES
Cloning of HVEM-L-A partial cDNA sequence encoding the extracellular domain of HVEM-L was identified through a BLAST search of an assembled EST data base for homologues of TNF, and this was confirmed by sequencing of a cDNA from an oxidized low density lipoprotein-treated macrophage cDNA library containing the most 5Ј EST. The Gene Trapper Positive Selection (Life Technologies, Inc.) was used to obtain two full-length cDNAs from a human liver cDNA library in pCMVSport, which differ in their noncoding regions.
mRNA Analysis-Multiple Tissue Northern blots were purchased from CLONTECH (Palo Alto, CA). Total RNA was extracted from cell lines (in exponential growth phase) and primary cells with TriReagent (Molecular Research Center, Inc., Cincinnati, OH). Northern blotting, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  labeled probe preparation, and hybridization have been previously described (27).
Expression of Soluble HVEM-L (sHVEM-L)-To express sHVEM-L in CHO cells, a mammalian expression vector was constructed that contained the cytomegalovirus promoter, tPA signal sequence, 11 residues of human immunodeficiency virus gp120 followed by 6 His residues, BamHI/EcoRI sites, and a bovine growth hormone poly(A) site (pCTDND1D2). The extracellular domain of HVEM-L was engineered for expression through polymerase chain reaction with the primers 5Ј-CACGGATCCGACGACGACGACAAACA AGAGCGAAGGTCTCAC-GAGGTC-3Ј and 5Ј -GAATGAAGCCCCGAAAGTACCACACTA-GATCTATAGTCC-3Ј, digestion with BamHI/EcoRI, and insertion into the above vector. One primer includes an enterokinase cleavage site after the histidine repeats for tag removal. The vector was transfected into CHO cells, and single colonies were selected in nucleotide-and nucleoside-free medium by limiting dilution. The highest expressing cell line was selected through analysis of supernatants using an anti-gp120 peptide monoclonal antibody.
Purification of sHVEM-L and sHVEM-sHVEM-L was captured from 28 liters of CHO conditioned media by S Sepharose chromatography and further purified by nickel-nitrilotriacetic acid chromatography in 20 mM sodium phosphate, 150 mM sodium chloride, pH 7 (PBS, pH 7), followed by elution with 300 mM imidazole in PBS. The final product was purified by size exclusion chromatography (Superdex 200 column; Amersham Pharmacia Biotech) in 20 mM sodium phosphate, pH 7.
The purification of HVEM-Fc was described previously (23). When engineered with a factor Xa cleavage site between the HVEM extracellular domain and Fc region, sHVEM could be obtained from HVEM-Fc by cleavage with factor Xa and further protein A chromatography (33). The expression and purification of other Fc fusions was similar to HVEM-Fc (34).
Binding Studies by Immunoprecipitation-2 g of HVEM-Fc receptor was incubated with 250 ng of various purified soluble ligands in 250 l of 25 mM HEPES, pH 7.2, 0.25% bovine serum albumin, 0.01% Tween in RPMI 1640 (binding buffer) at 4°C. Protein A capture and detection of bound ligand by Western blot with anti-gp120 peptide or anti-His repeat (CLONTECH) monoclonal antibodies has been previously described (34).
To examine binding of HVEM-Fc to full-length HVEM-L, HEK 293 cells were transfected with HVEM-L DNA or vector control using lipofectamine (Life Technologies, Inc.). 48 h after transfection, regular growth medium was replaced with methionine, cysteine-free Dulbecco's modified Eagle's medium (Life Technologies, Inc.) for 1 h followed by the addition of 1.5 mCi of translabel [ 35 S]Met/Cys (ICN Biomedicals, Inc.) and 2% dialyzed fetal bovine serum. Methods for preparation of cell lysates, precipitation with HVEM-Fc, and analysis by SDS-polyacrylamide gel electrophoresis have been previously described (34).
Binding Studies by Surface Plasmon Resonance-The association and dissociation rates of the interaction of HVEM-L with captured HVEM-Fc was determined by surface plasmon resonance using a BIAcore 1000 (BIAcore Inc., Piscataway, NJ). The capture surface was a protein A (Pierce)-modified CM5 sensor chip (35). Sensorgrams were collected at 25°C and a flow rate of 30 l/min. The sensor surface was equilibrated with a buffer of 20 mM sodium phosphate, 150 mM NaCl, and 0.005% Tween 20, pH 7.4, and analyses were performed at 60 l/min and 25°C. HVEM-Fc was diluted into the above buffer to 1 g/ml, and a 60-l injection was passed over the capture surface, followed by a 240-l injection of HVEM-L. After the association phase, 600 s of dissociation data were collected. The surface was regenerated after each cycle. Sets of six analyte concentrations, 5.5-700 nM, and a buffer blank were collected and analyzed by nonlinear regression (36) using the BIAevaluation software, version 2.1. The first 10% of the dissociation data were fitted on the basis of both the simple AB 7A ϩ B and the biphasic A i B j 7 A i ϩ B j models. Fits with 2 values of Ͻ0.1 were obtained. The association data not limited by mass transport were fitted to A ϩ B 7 AB using the type 1 model.
Ultracentrifugation-Analytical ultracentrifuge data was collected in a Beckman XL-I analytical ultracentrifuge using double sector cells with sapphire windows (37). For analysis of data corresponding to a single macromolecular species, the primary data, absorbance at 230 nm versus radius, were fit to Equation 1.
Here, A tot is the total absorbance, and A m is the absorbance of the protein at the meniscus; M and v are the molecular mass and partial specific volume of the protein; is the solvent density; is the angular velocity; r m and r, respectively, are the measured and reference radial positions; and R and T, respectively, are the gas constant and the absolute temperature. "Base" is a term that signifies absorbing material that is nonsedimenting. The partial specific volume was estimated by the method of Cohn and Edsall for sHVEM-L to be 0.720 ml/g, and the solvent density (150 mM NaCl, 20 mM sodium phosphate, pH 7.4) was estimated to be to be 1.005 ml/g (38). The molecular mass for the sHVEM-L monomer determined by matrix-assisted laser desorption/ ionization mass spectrometry was 21,687 Da. The predicted mass from the sequence was 19,791 Da. The difference corresponded to ϳ11 mol of hexose/monomer. Carbohydrate was incorporated into the determination of v as suggested by Laue (38). The test for equilibrium was the superposition of data sets taken 4 h apart. NF-B-driven Luciferase Reporter Assay-U937 cells transfected with a plasmid encoding a luciferase reporter gene under the control of an IL-8 promoter (39) were seeded into 96-well plates at a density of 2 ϫ 10 5 cells/well. HVEM-L, HVEM-L preincubated for 15 min with a 10fold excess of HVEM-Fc or DR3-Fc, or PBS alone was added to different wells and incubated at 37°C for 5 h. After the removal of supernatant, cells were lysed in 20 l of 1ϫ lysis buffer (Promega, Madison, WI), gently shaken, incubated for 15 min at room temperature, and assayed for luciferase production in a MicroLumat LB 96 P luminometer (EG & G Berthold, Bad Wilbad, Germany).
Flow Cytometry-All surface staining was carried out using staining buffer consisting of PBS supplemented with 0.2% bovine serum albumin and 0.1% sodium azide. Cells were preincubated with 10 g of human IgG (Organon Teknika Corp., Durham, NC) for 10 min on ice to block nonspecific Fc binding. Biotinylated soluble receptor-Ig or unconjugated mouse mAb was then incubated with cells for 30 min at 4°C. Cells were washed and binding was detected using streptavidin/phycoerythrin for biotinylated receptor-Fc or goat anti-mouse antibody conjugated with fluorescein isothiocyanate (Sigma). After a further 30-min incubation at 4°C, cells were washed, fixed in 2% formaldehyde for 20 min at 4°C, and analyzed on a FACsort (Becton Dickinson Corp., San Jose, CA).
ELISAs-Immulon II ELISA plates (Dynatech Laboratories, Inc., Chantilly, VA) were coated with PBS containing receptor-Fc fusion protein at 10 g/ml overnight at 4°C. Plates were washed in buffer A (0.05% Tween 20, 0.02% sodium azide in PBS), blocked with buffer B (0.1% gelatin, 0.02% sodium azide in PBS) for 1 h, and washed with buffer A. sHVEM-L was then added to wells in buffer C (buffer A containing 0.1% gelatin) and incubated for 60 min at 37°C. HVEM-L binding was detected by incubating wells with 10 g/ml mouse anti-gp120 mAb to the epitope tag expressed on sHVEM-L in buffer C for 1 h at 37°C, washing, incubating with goat anti-mouse antibody conjugated with alkaline phosphatase (1:1000) in buffer C, and detecting with 1 mg/ml p-nitrophenyl phosphate. Absorbance was read at 405 nm using a Dynatech microplate reader (Dynatech Laboratories, Chantilly, VA).
Three-way Mixed Lymphocyte Reaction (MLR)-MLR proliferation assays were carried out as described previously (23). Dilutions of HVEM-L were added in quadruplicate to microtiter wells, and cells were cultured for 6 days at 37°C in 5% CO 2 . 1 Ci of [ 3 H]thymidine was added to wells for the last 6 h of culture, plates were harvested (Skatron, Sterling, VA), and thymidine incorporation was determined using a Wallac ␤-plate scintillaton counter (Wallac Inc., Gaithersburg, MA).
HT29 Assay-Proliferation assays using HT29 cells obtained from ATCC were carried out as described previously (40). Briefly, 5000 cells/well were seeded into 96-well flat bottomed plates (Falcon Labware, Franklin Lakes, NJ) together with 50 units/ml human interferon-␥ (R & D Systems, Minneapolis, MN). Cells were incubated with sHVEM-L for 90 h before the addition of [ 3 H]thymidine for the last 6 h of culture. Cells were harvested (Skatron Instruments, Chantilly, VA), and ␤-scintillation counting was carried out as described previously (23).

RESULTS
Identification of a Novel TNF Homologue-To identify potential ligands of HVEM and other novel TNF receptor homologues, we searched an assembled EST data base for homologues of TNF. One novel homologue was identified from several overlapping ESTs and was confirmed from sequencing of the cDNA containing the most 5Ј EST. Screening of a human liver cDNA library subsequently yielded a 1.49-kilobase pair cDNA, which encoded a complete open reading frame of 240 amino acids. This protein was designated HVEM-L based on subsequent studies (see below). Like other TNF homologues, the predicted protein is a type II membrane protein with a 37-amino acid intracellular domain, a single transmembranespanning region, and a 181-amino acid extracellular domain (Fig. 1A). There is one predicted N-linked glycosylation site at residue 102. Alignment of the extracellular domain of HVEM-L with members of the TNF family shows that it is closest to LT␤ and the closely related TNF␣ and LT␣ (Fig. 1, B and C).
Expression of Soluble TNF Homologue-We tested the receptor binding properties of this novel TNF homologue by express-  A-H). C, phylogenetic relationship of the extracellular domain of HVEM-L with closest TNF homologues. The alignment generated as above was further analyzed by DISTANCES and GROWTREE (Genetics Computer Group, Madison, WI). TL1 has been described previously (27).
ing the extracellular domain as a soluble, secreted protein. This was achieved by fusing the carboxyl-terminal 156 amino acids (amino acids 85-240) to a tissue plasminogen activator signal sequence followed by a peptide epitope from human immunodeficiency virus gp120, a hexahistidine purification tag, and an enterokinase cleavage site. The purified protein expressed in CHO cells and analyzed by reducing SDS-polyacrylamide gel electrophoresis ran as a single Coomassie-stained band with an apparent molecular mass of 24 kDa (Fig. 2a), but the monomer molecular mass found by matrix-assisted laser desorption/ionization mass spectrometry was 21,687 Da (data not shown). The difference between this value and that predicted from sequence (19,791 Da) suggests that the protein is glycosylated.
Analysis of the solution mass of sHVEM-L by analytical ultracentrifugation is shown in Fig. 2b. Equilibrium ultracentrifugation at 17,000 rpm (Fig. 2b, part B) gave a mass of 65,500 Ϯ 400 Da. Furthermore, time-derivative velocity sedimentation (41) (part C) yielded values for s 20, w and D 20, w of 4.17 Ϯ 0.04 S and 5.58 Ϯ 0.06 F, respectively. As such, a mass of 64,700 Ϯ 700 Da can be calculated from the Svedberg equation, s/D ϭ M(1 Ϫ v )/RT. Since the predicted mass for a trimer is 65,058 Da, these data are consistent with sHVEM-L being a stable trimer. There was no evidence for dissociation of the trimer to monomer or aggregation of the trimer to higher assembly states under the conditions used here.
Identification of the Ligand for HVEM-We examined the binding of the soluble TNF homologue described above to several TNF receptor family members by surface plasmon resonance. The protein bound to HVEM-Fc in a specific manner and so was named HVEM-L. When incubated with a panel of TNFR homologue fusions such as OPG-Fc, HVEM-Fc, DR3-Fc, DR4-Fc, and TRID-Fc and an unrelated Fc fusion, IL-5R-Fc, only HVEM-Fc was able to precipitate sHVEM-L as detected by Western blot with anti-gp120 mAb (Fig. 3A). Furthermore, the binding was competed with sHVEM (Fig. 3B). Similar specific binding could be demonstrated by an ELISA in which HVEM-Fc but not OPG-Fc was able to capture sHVEM-L as detected by anti-gp120 mAb (Fig. 4A).
The kinetic rate constants and the intrinsic dissociation constant were determined by surface plasmon resonance. HVEM-Fc was captured onto the sensor surface using protein A, and HVEM-L was passed over this surface. Analysis of the sensorgrams (Fig. 4B) shows that the association phase is monophasic with k a ϭ 5.4 ϫ 10 5 Ϯ 0.7 (M Ϫ1 s Ϫ1 ), whereas the dissociation phase is biphasic with two dissociation rates: k 1d ϭ 0.024 Ϯ 0.003 s Ϫ1 and k 2d ϭ 2.42 ϫ 10 Ϫ4 Ϯ 0.88 s Ϫ1 . The latter is due to saturation of HVEM-Fc on the surface by high concentrations of HVEM-L, which drives all of the binding toward a monovalent interaction. As HVEM-L dissociates from the surface, additional receptors are available to interact with a single HVEM-L trimer, and this decreases the dissociation rate. The intrinsic dissociation constant of HVEM-L trimer binding to HVEM-Fc calculated from the ratio k 1d /k a is 44 nM, and the avidity is approximately 100-fold tighter.
Binding of HVEM-Fc to HVEM-L-transfected COS Cells-To see if HVEM bound to the membrane form of HVEM-L, a vector containing the entire open reading frame of HVEM-L (Fig. 1A) was transfected into COS cells, and binding of HVEM to HVEM-L was examined by flow cytometry. Using a monoclonal antibody raised to sHVEM-L, high levels of HVEM-L surface expression were detected on 26% of transfected cells (Fig. 5E) compared with an isotype control mAb (Fig. 5F). No binding was detected on mock-transfected cells (data not shown). Similarly, HVEM-L-transfected COS cells bound significant levels of biotinylated HVEM-Fc (Fig. 5A), which was completely blocked by preincubation of HVEM-L-transfected cells with a 10-fold molar excess of unlabeled HVEM-Fc or sHVEM-L (Fig.  5, C and D, respectively). Furthermore, control Fc-biotin did not bind to HVEM-L-transfected cells (Fig. 5B), and biotinylated HVEM-Fc did not bind to mock-transfected cells (data not shown), confirming the specificity of this binding.
When the transfected COS cells were metabolically labeled with [ 35 S]methionine/cysteine, HVEM-Fc was able to precipitate a 30-kDa protein from HVEM-L-transfected but not mocktransfected cells (Fig. 3C). Binding was competed with unlabeled sHVEM. The molecular mass is higher than the predicted 26.4 kDa, again suggestive of glycosylation. These data indicate that HVEM can bind to both membrane and secreted forms of HVEM-L.
HVEM-L Induces an NF-B Promoter-driven Reporter Gene-It has been reported that overexpression of HVEM leads to activation of NF-B (24, 31). We therefore examined the FIG. 1-continued ability of HVEM-L to activate NF-B using an IL-8 promoterdriven luciferase gene transfected into U937 cells. U937 cells have been previously shown to express HVEM on their surface by flow cytometry (33). In response to TNF (5 ng/ml), this U937 cell line gives a Ͼ20-fold increase in luciferase gene expression (data not shown). The addition of increasing amounts of HVEM-L to these U937 cells led to a concentration-dependent increase in luciferase expression with an EC 50 of 130 nM, which saturated at 2-fold above the unstimulated level (Fig. 6). While small, this increase was specific, since it was completely blocked by the addition of excess HVEM-Fc but not DR3-Fc. Similar results were obtained with reporters containing either the IL-8 or the human immunodeficiency virus NF-B promoter element (data not shown). Thus, HVEM-L can weakly stimulate NF-B-driven transcription.
Tissue-specific Expression of HVEM-L-The expression of HVEM-L mRNA was examined in several tissues, cells, and cell lines by Northern blot. The 2.7-kilobase pair HVEM-L mRNA was abundantly expressed in spleen and lymph nodes, and lower expression was detected in peripheral blood lymphocytes, colon, small intestine, bone marrow, thymus, and lung (Fig. 7A). Low levels of expression were also detected in total RNA extracted from freshly isolated macrophages and myeloid cell lines such as KG1a and THP-1, but not PLB-985, HL60, U937, Jurkat, HUT78, Molt 3, K562, HEL, TF274, MG63, MCF7, HT29, monocytes, and coronary arterial endothelial cells (data not shown).
Consistent with the high spleen and lymph node expression, HVEM-L mRNA expression was readily detected in PMA-and phytohemagglutinin-treated primary CD4 ϩ and CD8 ϩ T cells but not in the respective resting T cells (Fig. 7B). The expression of HVEM-L was also induced by PMA in the myeloid cell lines PLB985, THP-1 (Fig. 7B), and HL60 (data not shown). Furthermore, HVEM-L ESTs were found predominantly from activated T cells and oxidized low density lipoprotein-induced macrophage cDNA libraries. These data suggest that HVEM-L expression is highly regulated in immune cells and might play a role in immune cell regulation.
Effect of sHVEM-L on an MLR-Both HVEM-L mRNA and HVEM mRNA and protein are expressed in T cells. Since HVEM-Fc (also known as TR2-Fc (23)) and HVEM monoclonal antibodies partially inhibit the proliferation of stimulated T cells and an MLR (23,33), it was possible that they did so by blocking endogenous HVEM-L activity. Hence, we examined the effect of sHVEM-L in a three-way mixed lymphocyte reaction, in which three donor cells are mixed together, and the resulting allogeneic proliferation was measured. sHVEM-L alone caused an enhancement of proliferation with an EC 50 of ϳ0.1 ng/ml, but at high doses this was reversed, suggesting down-regulation at high concentrations (Fig. 8). The enhanced proliferation was inhibited by excess HVEM-Fc (data not shown) as seen in the absence of sHVEM-L. The stimulatory effect of HVEM-L is consistent with the previously observed inhibitory effects of HVEM-Fc and HVEM monoclonal antibod-ies in this assay.
sHVEM-L Inhibits Proliferation of HT29 Adenocarcinoma Cells-The relatively close homology between HVEM-L and the lymphotoxins (Fig. 1C) suggested that these cytokines may have overlapping activities. To address this, the capacity of sHVEM-L to reduce proliferation of HT29 adenocarcinoma cells was examined. sHVEM-L inhibited HT29 cell proliferation in a dose-dependent manner with an IC 50 of 5 pM (Fig. 9A). In contrast, IL-4 had no activity. In a second experiment, HT29 cells were incubated with 500 pg/ml of sHVEM-L in the presence of varying concentrations of HVEM-Fc or control DR5-Fc (Fig. 9B). HVEM-Fc dose-dependently blocked sHVEM-L inhibition of proliferation, whereas DR5-Fc had no effect, indicating a specific receptor ligand interaction. The activity of HVEM-L was comparable in activity and magnitude with TNF␣ (IC 50 2 pM) and stronger than LT␣ (IC 50 100 pM) (Fig. 9C). DISCUSSION HVEM was originally identified as a co-receptor for herpesvirus infection (22) and subsequently shown to activate NF-B and AP-1 upon transfection in HEK293 cells (24,31). Furthermore, an HVEM-Fc fusion protein was shown to inhibit a mixed FIG. 4. Binding of HVEM-L to HVEM. A, ELISA. ELISA plates were coated with either HVEM-Fc or OPG-Fc at 10 g/ml. Increasing concentrations of sHVEM-L tagged with a gp120 epitope were incubated in the plate. sHVEM-L binding was detected using mouse anti-gp120 mAb followed by goat anti-mouse antibody-conjugated alkaline phosphatase. Absorbance was read at 405 nm. This experiment is a representative of three similar assays. B, surface plasmon resonance. Overlay of sensorgrams for the binding of HVEM-L by HVEM-Fc after subtraction of HVEM-Fc capture and the buffer blank is shown. The data for the dissociation rate constant and the association rate constant fitted well with biphasic and A ϩ B 7 AB models, respectively. lymphocyte proliferation assay (23), and more recent experiments have shown that both HVEM-Fc and monoclonal antibodies against HVEM are able to inhibit the proliferation, activation, and cytokine production of T cells (33). Consequently, it has been of great interest to determine how the regulation of T cell activity by this receptor compares and contrasts with that of other TNF receptor-related proteins expressed on these cells, such as TNFR1, TNFR2, Fas, CD27, CD30, OX40, and 4 -1BB (1). A key part of this understanding is the identification of the cognate ligand and where and how its expression is regulated.
In this paper, we have identified a ligand for HVEM, HVEM-L, which is a novel member of the TNF family. Like other TNF family members, HVEM-L is expressed as a type II membrane protein and, based on the characteristics of a secreted form of the protein, forms a trimer. As expected from previous studies of HVEM, HVEM-L stimulates NF-B dependent transcription and proliferation in an MLR. Several other TNF-like ligands can costimulate T cells, including TNF␣, LT␣, CD27L, CD30L, 4 -1BBL, and OX40L (42)(43)(44)(45). Like several of these, especially TNF␣ and LT␣, we found that the expression of HVEM-L is highly induced after activation of myeloid or T cells, but HVEM-L is not present in nonactivated cells. In contrast, HVEM is expressed constitutively in a number of cell types, and its expression is only weakly up-regulated during T cell activation (23), whereas several other TNF receptor homologues are more substantially regulated by activation (46 -49). There may also be differences in the kinetics of expression, subcellular localization, and response to specific stimuli that will need to be the subject of further investigation.
The weak activation of NF-B-dependent transcription by HVEM-L was surprising given the significant activation of NF-B obtained by overexpression after transfection into HEK293 cells (24,31). 2 The transcriptional activation did not depend on the particular NF-B promoter used, which suggests that the low level of activation is not due to differential stimulation of NF-B subunits. There is more than one possible explanation for these differences. The level of NF-B activation may depend upon the level of surface HVEM expression, which in the U937 cells used was low but detectable, 3 or it may may reflect a weak coupling of HVEM to NF-B in U937 cells versus HEK293 cells. Whether either reflects the physiology of natural cells will need to be explored further. The weak signaling might also reflect stimulation of additional HVEM-L receptor subunits. Indeed, while this paper was under review, another group published the identification of HVEM-L, which they called LIGHT (50) and showed that it bound to both HVEM and the LT␤ receptor.
We showed that HVEM-L inhibits proliferation of HT29 cells much like its closest homologues, LT␣, TNF␣, and LT␣␤ 2 . Since HT29 cells express LT␤R and HVEM, 3  death domain that is typically associated with apoptosis. Inhibition of proliferation by LT␣␤ 2 , which is mediated by the LT␤R, occurs progressively over 3 days, which is distinct from the more immediate induction of apoptosis commonly associ-ated with death domain-containing receptors. This distinction is also reflected in the different intracellular adapters utilized by these two receptors; LT␤R signals though TRAF3, whereas TNFR1 is known to signal through TRAF2 (51,52). These data suggest that members of the TNF/LT family may have overlapping properties but do not resolve the relative role of the different ligands and receptors.
One of the clearest ways to distinguish between the various TNF/LT members and their receptors has been through transgenic gene deletion experiments. TNF␣ and LT␣ share two receptors, TNFR1 and TNFR2, but LT␣ also binds to the LT␤R when it heterotrimerizes with LT␤ to form LT␣␤ 2 (53,54). Transgenic knockouts of either gene reveal that both are involved in germinal vesicle and lymph node development, although with important differences. TNF␣ Ϫ/Ϫ mice appear to have lost organized B cell follicles (55), whereas LT␣ Ϫ/Ϫ mice have lost both T and B cell follicles (56,57). It has been suggested that the additional LT␣ activity may be mediated by the LT␤ receptor (55), but this may be only the part of the story, since Mauri et al. (50) have shown that LT␣ also binds to HVEM in addition to HVEM-L binding to the LT␤R. Consequently, we are currently attempting to distinguish the role of HVEM-L through the generation of transgenic mice lacking HVEM-L.