Characterization of the Intracellular Domain of Receptor Activator of NF-kB (RANK) INTERACTION WITH TUMOR NECROSIS FACTOR RECEPTOR-ASSOCIATED FACTORS AND ACTIVATION OF NF-kB AND c-JUN N-TERMINAL KINASE*

Various members of the tumor necrosis factor (TNF) receptor superfamily interact directly with signaling molecules of the TNF receptor-associated factor (TRAF) family to activate nuclear factor kB (NF-kB) and the c-Jun N-terminal kinase (JNK) pathway. The receptor activator of NF-kB (RANK), a recently described TNF receptor family member, and its ligand, RANKL, promote survival of dendritic cells and differentiation of osteoclasts. RANK contains 383 amino acids in its intracellular domain (residues 234–616), which contain three putative TRAF-binding domains (termed I, II, and III). In this study, we set out to identify the region of RANK needed for interaction with TRAF molecules and for stimulation of NF-kB and JNK activity. We constructed epitope-tagged RANK (F-RANK616) and three C-terminal truncations, F-RANK330, F-RANK427, and FRANK530, lacking 85, 188, and 285 amino acids, respectively. From this deletion analysis, we determined that TRAF2, TRAF5, and TRAF6 interact with RANK at its C-terminal 85-amino acid tail; the binding affinity appeared to be in the order of TRAF2 > TRAF5 > TRAF6. Furthermore, overexpression of RANK stimulated JNK and NF-kB activation. When the C-terminal tail, which is necessary for TRAF binding, was deleted, the truncated RANK receptor was still capable of stimulating JNK activity but not NF-kB, suggesting that interaction with TRAFs is necessary for NF-kB activation but not necessary for activation of the JNK pathway.

RANK (for receptor activator of NF-B), a recently described novel TNFR family member, bears high similarity in its extracellular domain to CD40 (15). It consists of a 616-amino acid transmembrane receptor, of which 383 amino acids reside in the intracellular domain. The intracellular domain does not show any homology to any of the known TNFR family members. RANK mRNA is ubiquitously expressed in human tissues, but cell surface RANK is expressed only on dendritic cells, the CD4ϩ T cell line MP-1, and foreskin fibroblasts (15). CD40L greatly enhances expression of RANK on mature dendritic cells (15), suggesting a potential role for RANK in dendritic cell function.
The human and mouse ligands for RANK (RANKL) share 85% identity (15). This ligand consists of 317 residues and is a type II transmembrane protein, whose expression is restricted to primary T cells, T cell lines, and lymphoid tissue (15). Furthermore, RANKL was cloned independently by three groups as an osteoclast differentiation factor (16), as an apoptosisregulatory gene (TRANCE, for TNF-related activation-induced cytokine) (17), and as a ligand for the soluble TNFR family member osteoprotegerin (18). Overexpression of RANK and RANKL has been demonstrated to activate NF-B (15). RANKL was also shown to stimulate JNK activity in mouse thymocytes and T cell hybridomas, but not B cells (17), and was partially inhibited in thymocytes from dominant negative TRAF2 transgenic mice (19). Additionally, RANKL appears to enhance T cell growth and dendritic cell survival by up-regulation of Bcl-X L (15,17).
To date, there is no report to indicate the region of the RANK receptor necessary for activation of JNK and NF-B. Thus, we constructed various C-terminal truncations of RANK and transiently expressed them in human cultured cell lines to characterize their ability to interact with various TRAF family members and to activate JNK and NF-B. From this deletion analysis, we have identified specific regions of RANK that * This research was supported by the Clayton Foundation for Research. 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.

EXPERIMENTAL PROCEDURES
Reagents, Cell Lines, and Antibodies-HeLa, an epithelial carcinoma cell line, and 293, a human embryonic kidney cell line, were obtained from the American Type Culture Collection (Rockville, MD) and cultured in minimal essential medium supplemented with 10% fetal bovine serum and antibiotics. Affinity-purified rabbit anti-TRAF2 (SC-876, C-20) and anti-JNK1 (SC-474, C-17) antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Goat anti-rabbit IgGconjugated horseradish peroxidase was obtained from Bio-Rad. Anti-FLAG (monoclonal antibody M2) and anti-FLAG (M2)-conjugated agarose were obtained from Eastman Kodak Co. (New Haven, CT). Goat anti-mouse IgG-conjugated horseradish peroxidase was obtained from Transduction Laboratories (Lexington, KY). Protein A/G-Sepharose was obtained from Pierce.
Expression Plasmids-The complete cDNA for RANK (pSPORT3.0-TR8) was identified through a homology search of an expressed sequence tag cDNA data base (Human Genome Sciences, Inc., Rockville, MD) obtained from a primary dendritic cell cDNA library for proteins containing the cysteine-rich repeat characteristic of TNFR family members. This cDNA is identical to RANK (15). To generate FLAG-tagged RANK616, the 5Ј-primer CTAAGAAAGCTTTGTACCAGTGAGAAG-CAT and the 3Ј-primer GACGTAGTCGACTCAAGCCTTGGCCCCGCC were used in a PCR reaction with pSPORT3.0-TR8 to generate a PCR product that would encode residues 33-616 (lacking the signal sequence), which was cloned into the HindIII/SalI site of the expression vector pCMVFLAG1 (Eastman Kodak Co.). RANK deletion mutants were generated by PCR using the above 5Ј primer and the 3Ј primers (for RANK330: TCCTACGTCGACTCAGCTGACCAATGAGAGAGCAT-CCT; RANK427: AACGGCGTCGACTCAACTGTCCACCTCTTTTTGC-AA; and RANK530: CGCTGAGTCGACTCAGGAGTTACTTGTTTCCA-GTCAC) and cloned into the HindIII/SalI site of pCMVFLAG1. All plasmids were verified by automated DNA sequencing. Human TRAF2 cDNA (pcDNA3HisTRAF2) was a generous gift from Dr. T. Kamitani (University of Texas Health Science Center, Houston, TX). The complete cDNA for TRAF2 was cloned by PCR using primers containing BamHI (5Ј) and SalI (3Ј) sites and pcDNA3HisTRAF2 as a template. The TRAF2 PCR product was digested with BamHI/SalI and cloned into pRKmyc, resulting in pRKmycTRAF2. The plasmid encoding cDNA for TRAF6 (pSR␣-TRAF6) was a generous gift from Dr. S. Reddy (M. D. Anderson Cancer Center, Houston, TX). The cDNA for TRAF6 was digested from pSR␣-TRAF6 with KpnI/EcoRI and cloned into pB-S(KSϪ) to give rise to pBS-TRAF6.
Transient Transfections-HeLa (1.5 ϫ 10 6 cells/100-mm dish) and 293 (2 ϫ 10 6 cells/100-mm dish) cells were plated and transfected the next day with 7.5-10 g of expression vectors by using LipofectAMINE (Life Technologies, Inc.) as described by the manufacturer; transfection was allowed to proceed for an additional 24 h. Alternatively, 293 cells (0.6 ϫ 10 6 cells/well, 6-well plate) were plated and transfected the next day by the calcium phosphate method as described by the manufacturer (Life Technologies, Inc.). Cells were harvested 36 -40 h after transfection; half were analyzed for expression of epitope-tagged receptors and JNK activity, and the other half for NF-B. Lysates were prepared in lysis buffer (20 mM Tris, pH 8, 250 mM NaCl, 1 mM dithiothreitol, 2 mM EDTA, 1% Triton X-100, 10 g/ml leupeptin, 10 g/ml aprotinin, 0.5 mg/ml benzamidine, and 2 mM sodium vanadate). After a 30-min incubation on ice, the samples were cleared by centrifugation for 10 min. Protein was estimated using a Bio-Rad protein determination kit.
Western Blotting-Whole cell lysates (15 g) or proteins from immunoprecipitations were separated by 8.5% SDS-PAGE and electroblotted onto nitrocellulose membranes (Bio-Rad). Western blot analysis was performed using the indicated antibodies, and membranes were developed using the enhanced chemiluminescence (ECL) system (Amersham Pharmacia Biotech).
Immunoprecipitations and JNK Kinase Assays-From transiently transfected cells, lysates were prepared and immunoprecipitated using anti-FLAG-conjugated agarose or anti-JNK1 and protein A/G-Sepharose for 1 h. Where indicated, 35 S-labeled proteins were added to the lysate prior to immunoprecipitation. Beads were collected by centrifugation and washed four times in lysis buffer and then two times in kinase buffer (20 mM Tris, pH 8, 50 mM NaCl, and 1 mM dithiothreitol). For coimmunoprecipitation, proteins were eluted in SDS-sample buffer, boiled, and subjected to SDS-PAGE. JNK activity was analyzed using exogenous GST-Jun-(1-79) as a substrate as described previously (20). JNK activity and 35 S-labeled TRAF binding were quantitated using a PhosphorImager and Imagequant software (Molecular Dynamics, Sunnyvale, CA).
Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extracts were prepared from transfected cells essentially as described (20). Equivalent amounts of nuclear protein were used in an EMSA reaction with 32 P-labeled NF-B oligonucleotide from the human immunodeficiency virus-long terminal repeat as described (20). NF-B activation was quantitated using a PhosphorImager and Imagequant software.

RESULTS AND DISCUSSION
A human cDNA (TR8, for TNF receptor-like 8) encoding a TNFR-related protein was identified through a homology search of an expressed sequence tag cDNA library. The fulllength cDNA encodes a protein of 616 amino acid residues. The extracellular domain (residues 1-208) contains a signal sequence and the conserved cysteine-rich repeats characteristic of the TNFR family (21). The intracellular domain (residues 234 -616) is the largest of all the TNFR family members to date and contains no homology to other members of this family. This cDNA was found to be identical to a previously reported TNFR family member known as RANK (15).
Construction and Expression of Epitope-tagged RANK-To facilitate detection and immunoprecipitation of RANK in cultured cells, we constructed a FLAG epitope-tagged version of RANK in the plasmid pCMVFLAG1. The mature polypeptide encodes residues 33-616 (F-RANK616) with a FLAG epitope tag at its N terminus (Fig. 1A). To identify which region of the cytoplasmic domain is needed for signaling, we constructed three C-terminal deletions, designated F-RANK530, -427, and -330 (Fig. 1A) and lacking 85, 188, and 285 amino acids, respectively.
Most of the TNFR family members interact directly with various members of the TRAF family of signaling proteins. In some of these receptors, a consensus TRAF-binding motif (PX-QX(T/S)) is required to bind to TRAF2, TRAF3, and TRAF5 (7,13,14,22,23). Inspection of the intracellular domain of RANK suggests three potential TRAF-binding domains, two at the C terminus (TRAFII and III) and one in the middle of the intra- cellular domain (TRAFIII) (Fig. 1B). Furthermore, these putative TRAF binding domains in RANK are conserved between the mouse and human receptors. Transient expression of F-RANK and its deletion mutants was demonstrated in both HeLa and 293 cell lines (Fig. 2). As expected, the deletion mutants were expressed similarly in both cell lines; however, expression levels of the deletion mutants were typically less than those of the full-length receptors using similar amounts of expression vectors. 2 TRAF2, TRAF5, and TRAF6 Interact with the C Terminus of RANK-Because most TNFR family members utilize TRAFs as signaling components and RANK contains putative TRAFbinding domains, we examined the ability of RANK to interact with various TRAFs. We transiently transfected HeLa and 293 cells with vectors directing expression of F-RANK616 and F-RANK deletion mutants. After 24 -36 h, cell lysates were prepared, and epitope-tagged receptors were immunoprecipitated with anti-FLAG-conjugated agarose. Coprecipitation of endogenous TRAF2 was detected by Western blotting with anti-TRAF2 polyclonal antibodies. When expressed in HeLa (Fig.  3A, top) and 293 cells (Fig. 3A, bottom), only F-RANK616 and none of the F-RANK deletion mutants precipitated endogenous TRAF2. Membranes were also probed with anti-FLAG to ensure the precipitation of epitope-tagged receptors (data not shown).
To examine whether other TRAFs could interact with RANK, we transiently transfected 293 cells with F-RANK expression vectors. After 36 h, cell lysates were prepared and in vitro translated 35 S-labeled TRAF2, TRAF5, and TRAF6 were added to each of the lysates. The epitope-tagged receptors were immunoprecipitated with anti-FLAG-conjugated agarose, and bound proteins were eluted in SDS-sample buffer and subjected to SDS-PAGE. The bound 35 S-labeled TRAFs were detected by exposure of the dried SDS-PAGE gel to x-ray film. Like endogenous TRAF2, 35 S-labeled TRAF2 coprecipitated only with F-RANK616 and not with the deletion mutants (Fig.  3B, top). Similarly, 35 S-labeled TRAF5 (Fig. 3B, middle) and TRAF6 (Fig. 3B, bottom) coprecipitated with F-RANK616 and not with the deletion mutants. Quantitation of 35 S-labeled TRAF2, TRAF5, and TRAF6 bound to F-RANK616 showed 145-, 11-, and 5-fold increases, respectively, in binding relative to vector-transfected cells. Thus, we have shown that TRAF2, TRAF5, and TRAF6 interacted with RANK at its C-terminal 85 residues.
RANK Deletion Mutants Lacking TRAF Binding Domains (II and III) Activate JNK-TRAF2, TRAF5, and TRAF6 are involved in JNK activation (3) by various members of the TNFR family and the interleukin-1 receptor (5) (i.e. TRAF6). We tested whether RANK and the various C-terminal deletion mutants were capable of activating JNK. When overexpressed in cultured cell lines, most TNFR family members activate signal transduction pathways in the absence of ligand (2). Thus, we transiently transfected 293 cells with increasing amounts of F-RANK expression vectors. Cell lysates were prepared 36 h after transfection and analyzed for receptor expression by Western blotting with anti-FLAG antibodies (Fig. 4A). Furthermore, the cell lysates were assayed for JNK activation by immune complex kinase assays using GST-Jun-(1-79) as a substrate. Transient overexpression of F-RANK616 in 293 cells activated JNK (Fig. 4B). Furthermore, F-RANK530 and -427 deletion mutants, which lack 85 and 188 residues from the C terminus, respectively, could still activate JNK (Fig. 4B). However, C-terminal truncation of 285 residues (which leaves approximately 98 amino acids intact) could not activate JNK (Fig.  4B). From at least three independent transfection experiments, we found that F-RANK616, -530, and -427 could increase JNK activity between 4-and 10-fold, whereas F-RANK330 increased activity by no more than 1.5-fold relative to vector-transfected cells. These data suggest that F-RANK530 and F-RANK427 may stimulate JNK activation without binding directly to TRAFs. Because F-RANK330 had no significant effect on JNK activation, we tentatively localized a JNK activation domain between residues 330 and 427 within the cytoplasmic domain of RANK.
The C Terminus of RANK Is Necessary for NF-B Activation-According to gel mobility shift assays, overexpression of RANK in 293 cells activates NF-B (15). To explore whether RANK deletion mutants activate NF-B, we transiently transfected 293 cells with F-RANK616 and the F-RANK deletion mutants. Western blotting with anti-FLAG antibodies indicated expression of the epitope-tagged receptors (Fig. 5A). Analysis of NF-B by a gel mobility shift assay indicated that only F-RANK616 activated NF-B (Fig. 5B). None of the F-RANK deletions were capable of activating NF-B in three independent transient transfection experiments, even though from the same transfections F-RANK530 and -427 could activate JNK.
Our data are consistent with a previous report (15) indicating that transient overexpression of RANK in 293 cells induces NF-B. We further demonstrated by deletion of the C-terminal 85 residues that this domain is necessary for TRAF interaction and most likely NF-B activation as well. Whether the interaction between RANK and TRAFs is responsible for NF-B activation remains to be determined. Our data are in agreement with reports that show that TRAF2, TRAF5, and TRAF6 participate in NF-B activation by other TNFR family members (3).
Stimulation of mouse thymocytes or T-cells, but not B-cells, by RANKL/TRANCE induces JNK activation (17), which could be inhibited in thymocytes from transgenic mice expressing a dominant negative form of TRAF2 (19). Our deletion analysis of RANK provided evidence that RANK lacking the TRAF binding domain could still stimulate JNK activity. Furthermore, our deletion analysis implies that RANK residues between 330 and 427 are necessary for JNK activation. Thus, it appears that RANK can activate JNK in a TRAF-independent manner. This conclusion may seem contradictory to that reached following ligand stimulation of thymocytes from dominant negative TRAF2 transgenic mice (19), sible that RANK can activate the JNK pathway in both a TRAF-dependent and -independent fashion. Moreover, it is possible that other unidentified adaptor proteins and TRAFlike molecules are responsible for signaling by RANK.
In summary, RANK encodes the largest cytoplasmic domain (383 amino acids) of any TNFR family member identified thus far. For the first time, we provide evidence that TRAF2, TRAF5, and TRAF6 bind to the C-terminal 85 amino acids; however, TRAF2 appeared to bind better than TRAF5 and TRAF6. Furthermore, we demonstrated that deletion of the FIG. 3. Coprecipitation of TRAF2, TRAF5, and TRAF6 with F-RANK616. A, human HeLa (top) or embryonic kidney 293 (bottom) cells were transiently transfected with the indicated expression vectors as described under "Experimental Procedures." After 24 h, whole cell lysates were prepared, and epitope-tagged receptors were immunoprecipitated with anti-FLAG-conjugated agarose and washed, and bound proteins were eluted with SDS-sample buffer. Samples were subjected to SDS-PAGE, and coprecipitating TRAF2 was detected by Western blotting with anti-TRAF2 polyclonal antibodies. B, human 293 cells were transiently transfected with the indicated expression vectors as described under "Experimental Procedures," and after 36 h cell lysates were prepared. In vitro translated 35 S-TRAF2 (top), 35 S-TRAF5 (middle), or 35 S-TRAF6 (bottom) were added to the lysates, and the epitope-tagged receptors were immunoprecipitated as described in A. Samples were subjected to SDS-PAGE, and the dried gel was exposed to x-ray film for 24 h to detect bound 35 S-TRAF2, 35 S-TRAF5, and 35 S-TRAF6. IP, immunoprecipitate. 4. Overexpression of F-RANK616, -530, and -427 stimulate JNK activity. Human 293 cells were transiently transfected with 0.5, 1.5, and 3.0 g of the indicated expression vectors as described under "Experimental Procedures." After 36 h, cell lysates were prepared and subjected to Western blotting with anti-FLAG (A) and immunoprecipitated with anti-JNK1 antibodies. The activity of coprecipitating JNK1 was measured by an immune complex kinase assay (B) using exogenous GST-Jun-(1-79) as described under "Experimental Procedures." Phosphorylation of GST-Jun-(1-79) was quantitated on a PhosphorImager, and -fold activation was measured relative to the vector-transfected cells. Data are representative of three independent transfection experiments.

FIG.
FIG. 5. The C terminus of RANK is necessary for NF-B activation. Human 293 cells were transiently transfected with 0.5 and 2 g of the indicated expression vectors as described under "Experimental Procedures." After 36 h, cell lysates were prepared and subjected to Western blotting with anti-FLAG (A), and nuclear extracts were subjected to an EMSA (B) as described under "Experimental Procedures." NF-B binding was quantitated on a PhosphorImager, and -fold activation was measured relative to the vector-transfected cells. Data are representative of three independent transfection experiments. TRAF interaction motif at the C terminus did not diminish RANK's stimulation of JNK activity, suggesting that RANK could activate JNK in a TRAF-independent manner. However, deletion of the C-terminal 85 residues results in loss of NF-B activation. Thus, we have demonstrated that TRAF family members interact with the novel TNFR family member RANK and may participate in RANK signal transduction.