ATAR, a novel tumor necrosis factor receptor family member, signals through TRAF2 and TRAF5.

Members of tumor necrosis factor receptor (TNFR) family signal largely through interactions with death domain proteins and TRAF proteins. Here we report the identification of a novel TNFR family member ATAR. Human and mouse ATAR contain 283 and 276 amino acids, respectively, making them the shortest known members of the TNFR superfamily. The receptor is expressed mainly in spleen, thymus, bone marrow, lung, and small intestine. The intracellular domains of human and mouse ATAR share only 25% identity, yet both interact with TRAF5 and TRAF2. This TRAF interaction domain resides at the C-terminal 20 amino acids. Like most other TRAF-interacting receptors, overexpression of ATAR activates the transcription factor NF-kappaB. Co-expression of ATAR with TRAF5, but not TRAF2, results in synergistic activation of NF-kappaB, suggesting potentially different roles for TRAF2 and TRAF5 in post-receptor signaling.

The TNFR 1 superfamily includes, among others, TNFR1, TNFR2, Fas, NGFR, CD40, and CD30 (1). Receptors within this group share a similar extracellular domain architecture of multiple cysteine-rich repeats, each containing about 40 amino acids with six cysteines. However, no significant homology is found within the intracellular domains with the exception of the death domain shared by TNFR1, Fas, and Wsl-1 (DR3) (2)(3)(4)(5). Post-receptor signaling mechanisms utilized by the TNFR superfamily have recently begun to emerge with the discovery of two distinct classes of receptor-associated proteins.
The death domain proteins TRADD, RIP, and FADD interact with the death domain of TNFR1 or Fas following ligandmediated receptor aggregation (6 -9). The N-terminal death effector domain of FADD then recruits Caspase-8 (Flice/ MACH), an ICE-like cysteine protease, to the receptor complex (6 -8). This process activates the apoptotic protease cascade and results in rapid onset of apoptosis.
The second group of signaling molecules, the TRAF proteins, directly interact with TNFR2, CD40, CD30, and lymphotoxin-␤ receptor (9 -12). No obvious motifs are shared by the cytoplasmic regions of these TRAF-binding receptors except for a few conserved amino acids. Six TRAF proteins have been reported, each containing a highly conserved TRAF domain at the C terminus and, with the exception of TRAF1, a lesser conserved N-terminal RING finger and zinc finger motif (9 -11, 13-15). Specific receptor recognition is mediated through the TRAF domain. Biological activities and post-receptor signaling mechanisms of the TRAF proteins are still obscure. TRAF2, TRAF5, and TRAF6, when overexpressed, activate NF-B (11,13,16). TRAF2 is also required for TNF-induced JNK activation (17). The N-terminal RING finger motif is necessary for both of these activities.
In the course of Amgen EST project, we discovered a novel TNFR-related protein we have named ATAR (another TRAFassociated receptor). Expression of ATAR is highly tissue-specific. To explore its biological function, we applied the yeast two-hybrid system to screen for signaling proteins that associate with the intracellular domain of ATAR. We found that ATAR interacts with both TRAF2 and TRAF5. Like most other TRAF-binding TNFR family members, ATAR also mediates NF-B activation.
Identification, cDNA Cloning, and RNase Protection-Mouse ATAR cDNA was identified during the Amgen EST Project as described (19). Full-length mouse and human clones were isolated by screening mouse intestinal and human bone marrow cDNA libraries, respectively, using mouse ATAR cDNA as a probe as described (19). RNase protection assay was performed with a commercially available kit according to the manufacturer's recommendations (Ambion).
Yeast Two-hybrid Screening-DNA encoding the intracellular domain (amino acids 224 -283) of ATAR was cloned in-frame with the GAL4 DNA-binding domain in pGBT9 vector, and the resulting plasmid was used as bait in two-hybrid screening of both human B cell and mouse lymph node libraries following the manufacturer's protocol (CLONTECH). Yeast DNA was recovered using the Yeast DNA Isolation System (Stratagene) and transformed into Escherichia coli by electroporation. Plasmids containing cDNA inserts were identified by restriction mapping and sequencing. Subsequent two-hybrid interaction assays were performed by co-transformation of plasmids into yeast HF7C strain and selection for histidine prototrophy and expression of ␤-galactosidase.
Antibody Generation-A PCR fragment encoding the ATAR extracellular domain was fused in frame upstream of human IgG1 Fc in a modified pCEP4 vector (Invitrogen). The resultant ATARFc/pCEP4 plasmid was transfected into 293-EBNA-1 cells using lipofectamine (Life Technologies, Inc.), and transfected cells were selected with 100 g/ml hygromycin. Drug resistant clones were pooled and cultured in serum-free medium for 72 h. ATARFc fusion protein was purified from the conditioned medium by protein G affinity chromatography as de-* 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 U.S.C. Section 1734 solely to indicate this fact.
scribed (20). The fusion protein was then used as antigen to raise rabbit anti-ATAR antiserum (Babco).
Transfection and Reporter Assay-Approximately 3 ϫ 10 5 293 cells/ well were seeded on 6-well plates. The next day, cells were transfected by the calcium phosphate precipitation method. After incubation for another 24 h, cells lysates were prepared and tested for luciferase activity using the Enhanced Luciferase Assay kit (Analytical Luminescence Laboratory). ␤-Galactosidase activity was measured with Galacto-Light (Tropix).
Co-immunoprecipitation and Western Analysis-Transfections were performed in 6-well plates as described above. 24 h after transfection, cells were lysed with 500 l of E1A buffer (50 mM Hepes [pH 7.6], 250 mM NaCl, 0.1% Nonidet P-40, 5 mM EDTA). Lysates were then incubated with 5 l of preimmune serum or antiserum against ATAR extracellular domain at 4°C for 1 h and then mixed with 20 l of a 1:1 slurry of protein A-Sepharose (Pharmacia Biotech Inc.) and incubated for another hour. After six washes with 1 ml of E1A buffer, the precipitates were fractionated on 12.5% SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membrane (Schleicher & Schuell). Western blot analysis was performed with an anti-FLAG monoclonal antibody (Kodak) and horseradish peroxidase-coupled rabbit antimouse immunoglobulin using enhanced chemiluminescence according to the protocol of the manufacturer (Amersham Corp.).

RESULTS
A murine cDNA sequence that appeared to encode a portion of a novel TNFR family member was identified during the Amgen EST program. A full-length cDNA isolated from a mouse intestinal cDNA library encodes a 276-amino acid pro-

FIG. 2. RNase protection assay of murine ATAR.
A 300-base pair (bp) 32 P-labeled fragment was used as probe in the RNase protection analysis of mouse total RNA samples as indicated. A 250-base pair protected fragment was generated after RNase treatment.

FIG. 3. ATAR Interaction with TRAF proteins.
A, human and mouse ATAR interact with TRAF5 and TRAF2. Expression vectors encoding full-length or deletion mutants of human or mouse ATAR intracellular domain fused to the GAL4 DNA-binding domain were cotransformed into HF7C yeast strain with vectors expressing GAL4 activation domain fused with TRAF1, TRAF2, TRAF3, or TRAF5 proteins. Yeast transformation mixtures were plated on synthetic dextrose plates lacking tryptophan, leucine, and histidine and containing 20 mM 3-aminotriazole. ϩ indicates growth on the selection plates and also expression of ␤-galactosidase. Each transformant was also plated on plates lacking tryptophan and leucine as a transformation efficiency control. wt, wild type. B, mutagenesis of TRAF-binding site of human ATAR. Four conserved amino acids within the TRAF-binding sites of human CD40 and ATAR were aligned. Each of the four amino acids within ATAR was replaced with alanine individually. The intracellular domain containing each point mutation was fused to the GAL4 DNAbinding domain, and the fusion protein was tested for its interaction with TRAF2 and TRAF5 in the yeast two-hybrid interaction assay as described above. (ϩ) indicates growth on the selection plates and also expression of ␤-galactosidase.  (lanes 1 and 3) or antibody against ATAR extracellular domain (lanes 2 and 4). Co-precipitating TRAF proteins were detected by immunoblot analysis with monoclonal antibody against Flag epitope. ATAR, TRAF5, and TRAF2 13472 tein we called ATAR that has a signal peptide and a single transmembrane region (Fig. 1A). The extracellular domain contains three cysteine-rich repeats and is most homologous to TNFR2. The receptor has a short cytoplasmic tail of only 46 amino acids. During the screen, we also isolated different spliced products that result in deletions within the extracellular domain or generation of soluble receptor forms (Fig. 1B). However, none of the soluble forms comprise complete cysteine repeats, suggesting they might be defective in ligand recognition. The human counterpart of the receptor was subsequently isolated from a bone marrow cDNA library. Amino acid comparisons demonstrated that the mouse and human homologs share 51 and 25% identity within their extracellular and intracellular domains, respectively.
Tissue distribution of mouse ATAR was examined in an RNase protection assay (Fig. 2). Among the tissues examined, ATAR mRNA was highly expressed in spleen, thymus, bone marrow, lung, and small intestine. Very low levels of expression were detected in brain, skeletal muscle, and stomach. Similar tissue distribution patterns were observed for human ATAR mRNA as determined by Northern blot analysis (data not shown).
To explore the biological function of ATAR, we used yeast two-hybrid screening to search for signaling proteins that interact with the ATAR intracellular domain. Human B lymphocyte and mouse lymph node libraries were screened with bait containing the GAL4 DNA-binding domain fused to the human ATAR intracellular domain. A total of 40 million transformants were screened and positive clones obtained encoded TRAF5 or TRAF2. All of these TRAF5 and TRAF2 clones contained a complete C-terminal TRAF domain, consistent with previous reports that this domain mediates receptor interaction. We also examined the interaction of ATAR with TRAF1 and TRAF3 by yeast two-hybrid interaction assays. No interaction of ATAR with TRAF1 or TRAF3 was detected in this assay (Fig. 3A).
Because human and mouse ATAR are poorly conserved within their intracellular domains, we tested the ability of mouse ATAR to interact with TRAF proteins. Like its human homolog, mouse ATAR interacted with TRAF2 and TRAF5 but not TRAF1 or TRAF3 (Fig. 3A). Sequence alignment revealed a stretch of nine conserved amino acids (residues 261-269 in mouse and 264 -272 in human) among otherwise largely unrelated intracellular domains (Fig. 1A), suggesting the importance of this short conserved region in TRAF recognition. To further localize the TRAF-interacting region, we fused residues 224 -263 or residues 264 -283 of the human ATAR intracellular domain to the GAL4 DNA-binding domain and tested their interaction with TRAF proteins. The GAL4DB-ATAR(264 -283) fusion protein containing the conserved nine-amino acid stretch was able to bind both TRAF5 and TRAF2. Therefore, the C-terminal 20 amino acids of ATAR were sufficient for TRAF interaction. Interestingly, the last four of these nine amino acids are conserved in the TRAF-interacting region of CD40 (Fig. 3B). Double mutations of CD40 Glu 253 and Thr 254 to alanine abolished the TRAF interaction (21). Each of the four conserved amino acids of ATAR was substituted with alanine, and each point mutant was tested for its interaction with TRAF2 and TRAF5 (Fig. 3B). The glutamic acid to alanine substitution at position 271 abolished ATAR interaction with the TRAF proteins. However, the other point mutants interacted as well as the wild type receptor with TRAF2 and TRAF5.
We then examined the interaction of ATAR with TRAF proteins in mammalian cells. An expression vector that directs the synthesis of full-length ATAR was co-transfected with a FLAGtagged TRAF2 or TRAF5 expression vector into human embryonic kidney 293 cells. Cell lysates from each transfection were immunoprecipitated with either preimmune serum or a polyclonal antibody against the extracellular domain of ATAR. Western analysis of the immunoprecipitates were performed with a monoclonal antibody against the FLAG epitope. Both Flag-TRAF2 and Flag-TRAF5 specifically co-precipitated with were transfected with 1 g of pELAM-luc reporter plasmid, 0.5 g of pRSV-␤gal, 1 g of ATAR expression vector, and the indicated amounts of TRAF5(205-558) expression vector and supplemented with vector DNA to reach a total of 3.5 g of DNA. Luciferase activities were measured as described above. C, synergistic activation of NF-B by ATAR and TRAF5 293 cells (3 ϫ 10 5 ) were transfected with 1 g of pELAM-luc reporter plasmid, 0.5 g of pRSV-␤gal, 1 g of indicated combinations of expression vectors for ATAR, TRAF5, or TRAF2, and vector DNA to reach a total of 3.5 g of DNA. Luciferase activities were measured as described above. Western analysis indicated similar protein expression levels in all samples (data not shown).

ATAR, TRAF5, and TRAF2
13473 ATAR (Fig. 4). Therefore ATAR is another TRAF-interacting TNFR family member. Interaction of ATAR with TRAF2 and TRAF5 suggests that like TNFR2 and CD40, ATAR may also mediate NF-B activation. To test this possibility, a NF-B-dependent reporter construct was transfected into 293 cells. Co-transfection with a ATAR expression vector resulted in an approximately 10-fold induction of luciferase activity as compared with the vector transfection control (Fig. 5A). This NF-B activation requires ATAR interaction with TRAF proteins, because expression of ATAR(E271A), the point mutant that no longer interacts with TRAF2 or TRAF5, failed to activate NF-B (Fig. 5A).
Both TRAF2 and TRAF5, when overexpressed, activate NF-B. This activity depends on their N-terminal RING finger and zinc finger motifs (11,16). We generated a TRAF5(205-558) truncation mutant lacking the RING and zinc finger motifs. In a dose-dependent experiment, co-expression of TRAF5(205-558) deletion mutant with ATAR in 293 cells blocks NF-B activation induced by ATAR overexpression (Fig. 5B). Similar inhibitory effects were observed when ATAR was co-expressed with TRAF2 RING and zinc finger deletion mutant (data not shown). These dominant negative inhibitory effects suggest potential involvement of both TRAF5 and TRAF2 in ATARmediated NF-B activation pathway.
To further examine the effects of TRAF5 and TRAF2 expression on ATAR-mediated NF-B activation, we co-expressed ATAR with full-length TRAF5 or TRAF2 proteins in 293 cells. Interestingly, we found that co-expression of ATAR and TRAF5 resulted in a synergistic activation of NF-B (Fig.  5C). On the contrary, no synergy was detected when ATAR was co-expressed with TRAF2 (Fig. 5C). We also examined ATAR(E271A), a point mutant that no longer binds with TRAF proteins. When co-expressed with TRAF5 or TRAF2, ATAR(E271A) did not affect NF-B activation induced by TRAF5 or TRAF2 (data not shown). This specific synergistic activation by ATAR and TRAF5 suggests that TRAF5 plays a unique role in ATAR signaling. DISCUSSION We report here the identification of a novel TNFR family member. ATAR is the smallest reported member of this receptor superfamily so far. The human and mouse homologs are poorly conserved (45% identity) as compared with other members within the family. For example, TNFR1, TNFR2, Fas, and CD40 share 65, 64, 51, and 63% identity in the two species, respectively. This poor sequence conservation is especially evident in the region of the short cytoplasmic tail (25% identity). However, we argued against the possibility that they may represent distinct receptors based on two factors: 1) no other human counterparts were isolated during cDNA library screening with mouse ATAR, and 2) distinct receptors within this family share approximately 25% between each other.
Despite the limited homology, both human and mouse ATAR intracellular domains interact with the same TRAF proteins. This TRAF-binding activity resides in the C-terminal 20 amino acids of human ATAR, because GAL4-ATAR(264 -283) is still able to interact with TRAF5. Interestingly, within this region is a stretch of nine amino acids well aligned with the corresponding region in the mouse receptor. Similar amino acid residues have been shown to be crucial in TRAF-recognition by TNFR2, CD40, and CD30. This finding should help shed light on the structural requirements of TRAF-receptor interactions.
Overexpression of ATAR in mammalian cells results in NF-B activation. This activation depends on the interaction of ATAR with TRAF proteins, because a TRAF-binding deficient mutant (E271A) fails to activate NF-B. In addition, we found that co-expression of ATAR with TRAF5 but not TRAF2 resulted in synergistic activation of NF-B. These results suggest that TRAF5 contributes to ATAR-mediated NF-B activation. However, it does not rule out involvement of TRAF2 in this process, because TRAF2 is ubiquitously expressed and might not be a limiting factor in 293 cells.
ATAR is identical to a recently reported protein (HVEM) that was identified as a receptor for mediating viral entry of herpes simplex virus-1 and -2 (22). It will be interesting to determine if this viral entry process activates ATAR and how cellular responses mediated by the receptor (e.g. NF-B activation) affect the viral life cycle. It is noteworthy that latent infection membrane protein-1 (LMP1) of Epstein-Barr virus also interacts with TRAF proteins through its C-terminal cytoplasmic tail. Possible pathogenic roles of ATAR and TRAF proteins in herpes simplex virus infection await further study.
ATAR is mainly expressed in spleen, thymus, bone marrow, and small intestine, all tissues in which lymphoid cells are enriched. This expression pattern suggests that ATAR plays a role in development or regulation of the immune system. This notion is supported by the fact that ATAR mediates activation of NF-B, which in turn activates a battery of genes crucial to immune regulation (23). Among the known TNF-like ligands tested, we did not detect interaction of ATAR with TNF␣, lymphotoxin ␣ and ␤, or CD40 ligand (data not shown). Identification of its cognate ligand will greatly improve our understanding of the physiological role of ATAR.