Role of TRAF3 and -6 in the activation of the NF-kappa B and JNK pathways by X-linked ectodermal dysplasia receptor.

X-linked ectodermal dysplasia receptor (XEDAR) is a recently isolated member of the tumor necrosis factor receptor family that has been shown to be highly expressed in ectodermal derivatives during embryonic development and binds to ectodysplasin-A2 (EDA-A2). By using a subclone of 293F cells with stable expression of XEDAR, we report that XEDAR activates the NF-kappaB and JNK pathways in an EDA-A2-dependent fashion. Treatment with EDA-A2 leads to the recruitment of TRAF3 and -6 to the aggregated XEDAR complex, suggesting a central role of these adaptors in the proximal aspect of XEDAR signaling. Whereas TRAF3 and -6, IKK1/IKKalpha, IKK2/IKKbeta, and NEMO/IKKgamma are involved in XEDAR-induced NF-kappaB activation, XEDAR-induced JNK activation seems to be mediated via a pathway dependent on TRAF3, TRAF6, and ASK1. Deletion and point mutagenesis studies delineate two distinct regions in the cytoplasmic domain of XEDAR, which are involved in binding to TRAF3 and -6, respectively, and play a major role in the activation of the NF-kappaB and JNK pathways. Taken together, our results establish a major role of TRAF3 and -6 in XEDAR signaling and in the process of ectodermal differentiation.

The ectodermal dysplasias are a heterogeneous group of genetic disorders that are identified by the absent or deficient function of at least two derivatives of ectoderm. Hypohidrotic ectodermal dysplasias (HED) 1 are a major subgroup of ectodermal dysplasias and are characterized by the triad of signs consisting of sparse hair, abnormal or missing teeth, and the inability to sweat (1). HED can be transmitted either as an X-linked disorder or morphologically indistinguishable autosomal dominant or recessive conditions in both humans and mouse (1). Mutations in ectodysplasin A, a novel ligand of the tumor necrosis factor family, were found to be responsible for the X-linked form of human anhidrotic ectodermal dysplasia (2,3) and Tabby (Ta) phenotype in mice (4). Subsequently, mutations in EDAR, a novel receptor of the tumor necrosis factor receptor family, were found in several families with autosomal dominant and recessive forms of anhidrotic ectodermal dysplasia and in downless (dl) mice (5,6). We (8) and others (7) have demonstrated that EDAR binds to a major isoform of ectodysplasin A, termed EDA-A1, and thus represented its physiological ligand. Recently, a homolog of EDAR, termed X-linked ectodermal dysplasia receptor (XEDAR), was discovered and was shown to bind to an alternatively spliced isoform of ectodysplasin, termed EDA-A2, which differs from EDA-A1 by two amino acids (7). XEDAR was also shown to activate the NF-B and ERK pathways upon transient transfection based overexpression. In order to further understand the signal transduction via XEADR, we have generated a subclone of 293F cells with stable expression of this receptor. By using this subclone, we have characterized the signal transduction via XEDAR upon treatment of cells with its physiological ligand, EDA-A2. We report that XEDAR activates NF-B and JNK pathways in a ligand-dependent fashion, and we establish the roles of TRAF3 and TRAF6 and the kinases of the IKK complex in the above processes. In addition, we have used deletion and point mutagenesis to delineate the regions and amino acid residues of the XEDAR cytoplasmic domain responsible for the above activities.

MATERIALS AND METHODS
Cell Lines and Reagents-293T cells were obtained from Dr. David Han (University of Washington, Seattle). 293F and 293 EBNA cells were obtained from Invitrogen. Rabbit polyclonal antibodies against IKK␣, IKK␤, NEMO/IKK␥, IB␣, TRAF2, TRAF3, TRAF6, FLAG, ␤-actin, JNK, and phospho-JNK were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against phosphorylated c-Jun, IKK␣/IKK␤, and IB␣ were obtained from Cell Signaling (Beverly, MA). FLAG and control mouse IgG beads were obtained from Sigma. The pull-down kinase assay kit for JNK activation was obtained from Cell Signaling, and the constructs for the Pathdetect luciferase reporter assay were purchased from Stratagene (La Jolla, CA).
Expression Constructs-XEDAR-L and XEDAR-s cDNA were amplified by reverse transcription-PCR using HaCat (human keratinocyte) cell line RNA as a template. The forward and reverse primers used for amplification also carried BamHI and SalI restriction enzyme sites at their 5Ј termini, respectively. The amplified product was subsequently cloned in a modified pSecTagA vector carrying a FLAG epitope tag downstream of the murine Ig signal peptide. The above FLAG-XE-DAR-L constructs were used to generate deletion mutants XEDAR⌬C37, XEDAR⌬C45, XEDAR⌬C53, XEDAR⌬C64, XEDAR⌬C69 * This work was supported by grants from the March of Dimes Foundation, National Institutes of Health Grant P30-AR41940-09, and the Department of Defense Breast Cancer Research Program Grant DAMD17-02-1-590, which is managed by the United States Army Medical Research and Materiel Command. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY152724.
A baculovirus construct encoding Myc-EDA-A1 has been described previously (8). Myc-EDA-A2 was constructed by cloning the nucleotide sequence corresponding to amino acids 134 -389 of the EDA-A2 isoform into a modified pFastBAC1 vector (Invitrogen), which contained a Myc epitope tag downstream of a baculovirus gp67 signal peptide (8). The sequences of all constructs were confirmed by automated fluorescent sequencing.
Myc-EDA-A1 and EDA-A2 proteins were produced by infection of Sf9 insect cells with the corresponding baculovirus constructs following the manufacturer's instructions (Invitrogen). Supernatants containing the secreted proteins were collected 60 h post-infection, filtered, virusremoved by ultracentrifugation, and stored at Ϫ70°C until used. Different batches were kept at same concentration by estimating secreted protein concentration by enzyme-linked immunosorbent assay using XEDAR-Fc as a probe and a purified preparation of EDA-A2 (R&D Systems) as a reference standard.
Generation of 293F-XEDAR-L Stable Cells-293F cells were infected with an empty retrovirus vector or one encoding FLAG-tagged XE-DAR-L. Twenty four hours after the infection, cells were selected with 500 g/ml G418. Individual clones were isolated by limiting dilution. Clones were screened by flow cytometry for the expression of FLAG-XEDAR-L using FLAG antibody.
Luciferase Reporter Assays-The NF-B reporter assay was performed essentially as described previously (8). Briefly, 293T cells were transfected in duplicate in a 24-well plate with the various test plasmids along with an NF-B/luciferase reporter construct (75 ng/well) and a Rous sarcoma virus promoter-driven ␤-galactosidase reporter construct (pRcRSV/LacZ; 75 ng). Cells were lysed 24 -30 h later, and extracts were used for the measurement of luciferase and ␤-galactosidase activities, respectively. In the case of experiments involving treatment of 293F-XEDAR cells with ligands, cells were transfected with the reporter plasmids as above and 12 h after transfection treated with control Sf9 supernatant (Control) or EDA-A2-containing supernatant for 9 h. In some experiments treatment with TNF-␣ (10 ng/ml) and IL-1␤ (10 ng/ml) was also used as a control. Cells were subsequently lysed and lysates used for reporter assays. Luciferase activity was normalized relative to ␤-galactosidase activity to control for the difference in the transfection efficiency.
For the c-Jun transcriptional activation assay, 293 EBNA cells or 293F-XEDAR cells were transfected in duplicate in a 24-well plate with various expression constructs (100 ng) along with a fusion transactivator plasmid containing the yeast Gal4 DNA-binding domain fused to transcription factor c-Jun (pFA-c-Jun) (50 ng), a reporter plasmid encoding the luciferase gene downstream of the Gal4 upstream activating sequence (pFR-luc) (500 ng), as well as ␤-galactosidase reporter construct (75 ng). Treatment, cell lysis, and luciferase assay were performed essentially as described above for NF-B reporter assays.
Co-immunoprecipitation Assays-For studying in vivo interaction, 5 ϫ 10 7 293F-XEDAR-L cells were treated with control supernatant or EDA-A2 for 10 min. Cells were subsequently lysed in 5 ml of buffer A (20 mM sodium phosphate (pH 7.4), 150 mM NaCl) containing 1% Triton X-100, and 1 EDTA-free mini-protease inhibitor tablet per 10 ml (Roche Molecular Biochemicals). Cell lysates were incubated for 1 h at 4°C with 50 l of FLAG or control mouse IgG beads precoated with a super-saturated casein solution. Beads were washed twice with buffer A, once with a high salt wash buffer (buffer A ϩ 500 mM NaCl), and again with buffer A. Bound proteins were eluted by boiling, separated by SDS-PAGE, transferred to a nitrocellulose membrane, and analyzed by Western blot. For studying the phosphorylation of different proteins in response to EDA-A2 treatment, 3 ϫ 10 5 XEDAR-L cells were plated in 6-well plates, Approximately 30 h later the cells were treated with EDA-A2 for different time intervals. Cells were subsequently lysed, and phosphorylation of proteins was detected by Western blot analysis using phospho-specific antibodies according to the manufacturer's instructions (Cell Signaling).

RESULTS
Isolation of XEDAR-L-We used reverse transcription-PCR to amplify XEDAR cDNA from RNA derived from HaCat (a human keratinocyte cell line). Sequencing of the cloned product revealed that it encoded an alternatively spliced isoform of XEDAR that differed from the published sequence (7) by the addition of 21 amino acids in the juxtamembrane region of the cytoplasmic domain (Fig. 1A). This isoform was designated XEDAR-L to distinguish it from the published sequence, which will be referred as XEDAR-s. Transient transfection of cDNAs encoding FLAG-tagged XEDAR-L or XEDAR-s isoforms in 293T cells led to equivalent activation of the NF-B pathway, as measured by a luciferase-based reporter assay (Fig. 1B).
EDA-A2 Induces IB␣ Phosphorylation and NF-B Activation in 293F-XEDAR Cells-In order to further characterize the various signaling activities of XEDAR, we used retrovirally mediated gene transfer to generate a subclone of 293F cells with stable expression of the FLAG-XEDAR-L isoform. After confirming the expression of FLAG-XEDAR-L using FACS, we tested the ability of this subclone to stimulate XEDAR signaling in an EDA-A2-dependent fashion. As shown in Fig. 2A, treatment of this subclone with EDA-A2 led to rapid phosphorylation and degradation of IB␣, which was evident within 5-10 min. A delayed phase of IB␣ phosphorylation was seen beginning ϳ1-h post-stimulation and probably represented phosphorylation of newly synthesized IB␣. The kinetics of IB␣ phosphorylation and degradation induced by EDA-A2 were similar to that induced by TNF-␣ ( Fig. 2A). Treatment of 293F-XEDAR cells with EDA-A2 was also accompanied by significant activation of the NF-B as measured by luciferasebased reporter assay (Fig. 2D) and electrophoretic mobility shift assay (Fig. 2B). No significant activation of the NF-B pathway was seen upon treatment of parental 293F cells with EDA-A2 (Fig. 2C) or the treatment of 293F-XEDAR cells with a control supernatant, suggesting that the above results are due to specific interaction of XEDAR with EDA-A2. 293T cells were transfected with the empty vector or indicated constructs (100 ng/well) along with an NF-B/luciferase reporter construct (75 ng/well) and an RSV/LacZ (␤-galactosidase) reporter construct (75 ng/well), and the experiment was performed as described under "Materials and Methods." The values shown are averages (mean Ϯ S.E.) of one representative experiment out of three in which each transfection was performed in duplicate.

Role of TRAFs in XEDAR-L-induced NF-B Activation-
After confirming the ability of 293F-XEDAR-L subclone to activate the NF-B pathway in an EDA-A2 dependent fashion, we used it to explore the downstream proteins involved in XEDAR signaling. TRAF family members have been shown to be involved in NF-B activation by different members of TNFR family (10,11). In order to characterize further the role of TRAFs in XEDAR-induced NF-B activation, we tested the ability of dominant-negative mutants of TRAF2, TRAF3, and TRAF6 to block EDA-A2-induced NF-B activation in 293F-XEDAR cells. As shown in Fig. 3, A and B, dominant-negative TRAF2, TRAF3, and TRAF6 were equally effective in blocking EDA-A2-induced NF-B activation. These results suggest the possibility that either TRAF2, TRAF3, or TRAF6 are involved in XEDAR-induced NF-B activation, or they play a mutually redundant role in this process.
TRAF3 and -6 Are Recruited to XEDAR in an EDA-A2dependent Fashion-To define further the role of TRAFs in XEDAR-induced signaling, we tested the interaction between XEDAR and various endogenously expressed TRAF family members upon treatment with EDA-A2. For this purpose, 293F-FLAG-XEDAR-L cells were treated with EDA-A2 or control supernatant, following which the cells were lysed and XEDAR immunoprecipitated with FLAG antibody beads, and the presence of TRAF family members in the immunoprecipitates was detected by Western blot analysis. As shown in Fig.   FIG. 2.

EDA-A2 induces IB␣ phosphorylation and NF-B activation in 293F-XEDAR-L cells.
A, 293F-XEDAR-L cells were treated with EDA-A2 (10 ng/ml) and TNF-␣ (10 ng/ml) for the indicated time intervals. Total cell lysates were subjected to Western blot analysis using phospho-specific and total IB␣ antibodies. B, electrophoretic mobility shift assay. 293F-XEDAR-L cells were treated with EDA-A2 for 1 h. Cells were subsequently lysed and nuclear extracts used for the gel shift assay as described previously (8). An arrow marks the position of the induced NF-B complex, and an asterisk marks the position of a constitutive complex. C, 293F cells were transfected with NF-B and ␤-galactosidase reporter plasmids as described in Fig. 1. Twelve hours after transfection cells were treated with control supernatant (Control), EDA-A2-containing supernatant, or TNF-␣ (10 ng/m) for 9 h and then lysed for the reporter assay as described under "Materials and Methods." The values shown are averages (mean Ϯ S.E.) of a representative of at least two independent experiments in which each transfection was performed in duplicate. D, 293F-XEDAR-L cells were transfected with NF-B and ␤-galactosidase reporter plasmids as described in Fig. 1. Twelve hours after transfection cells were treated with a different dose of EDA-A2 for 9 h and then lysed for the reporter assay. The values shown are averages (mean Ϯ S.E.) of a representative of at least two independent experiments in which each transfection was performed in duplicate.

FIG. 3. Mechanism of EDA-A2-induced NF-B.
A and B, dominant-negative mutants of TRAF2, TRAF3, and TRAF6 inhibit EDA-A2induced NF-B activity. 293F-XEDAR-L cells were transfected with the indicated plasmids together with the NF-B and ␤-galactosidase reporter plasmids. Twelve hours post-transfection, cells were treated with control Sf9 supernatant (Control), EDA-A2-containing supernatant, and TNF-␣ as indicated. Nine hours later cell lysates were prepared and used for reporter assays as described under "Materials and Methods." The amount of the inhibitor plasmids used was 500 ng/well, and the total amount of transfected DNA was kept constant by adding empty vector. The values shown are averages (mean Ϯ S.E.) of a representative of at least two independent experiments in which each transfection was performed in duplicate. C, endogenous TRAF3 and TRAF6 are recruited to XEDAR-L in an EDA-A2-dependent fashion, whereas TRAF2, IKK␣, and IKK␤ fail to do so. 293F-XEDAR-L cells were treated with control supernatant (Ϫ) and EDA-A2-containing supernatant (ϩ) for 10 min. Total cell lysates (L) were immunoprecipitated (IP) with FLAG beads (Sigma). Endogenously expressed co-immunoprecipitated proteins were detected by Western blot with the indicated antibodies. The TRAF3 antibody recognizes an epitope located at the C terminus of this molecule and is non-reactive with other TRAF family members.
3C, significant amounts of TRAF6 and TRAF3 were detected in the immunoprecipitate of EDA-A2-treated cells but were absent in any of the control treated samples. However, we have failed to detect an interaction between XEDAR and TRAF2 using the above assays (Fig. 3C). These results suggest that TRAF6 and TRAF3 are major adaptors involved in EDA-2induced XEDAR signaling, and the inhibitory effect of dominant-negative TRAF2 on EDA-A2-induced NF-B (Fig. 3A) may be related to the ability of the overexpressed protein to bind and sequester an essential component of NF-B pathway that is involved in TRAF2 as well as TRAF3/6 signaling (e.g. NIK).
Role of NIK in XEDAR-induced NF-B-NIK has been shown to be involved in the activation of the NF-B pathway by the members of TNFR and interleukin-1 receptor families (12).
To determine the role of NIK in XEDAR-induced NF-B, we used a dominant-negative inhibitor of this kinase. As shown in Fig. 4A, a C-terminal deletion mutant of NIK (NIK-2101) (8), was highly effective in blocking EDA-A2-and IL-1␤-induced NF-B activity while weakly blocking TNF-␣-induced activity.
Role of IKK Complex in XEDAR-induced NF-B-Mutations in NEMO/IKK␥ have been linked recently (13,14) to the pathogenesis of X-linked anhidrotic ectodermal dysplasia with immunodeficiency. Similarly, gene knockout of IKK␣/IKK1 leads to defects in ectodermal differentiation (15)(16)(17)(18)(19). We were interested in checking the involvement of these proteins of the IKK complex in XEDAR-induced NF-B activation. Western blot analysis of EDA-A2-treated 293F-XEDAR cells revealed the appearance of an IKK␣ band with reduced mobility, suggesting that XEDAR signaling leads to IKK␣ phosphorylation (Fig.  4E). This hypothesis was further supported by the examination of phosphorylation status of IKK␣ and IKK␤ using a phosphospecific antibody that recognizes phosphorylated IKK␣/IKK␤. As shown in Fig. 4F, this analysis revealed significant phosphorylation of IKK␣ beginning 5-10 min after treatment with EDA-A2, whereas weak phosphorylation of IKK␤ was detected after 40 min. The weak phosphorylation of IKK␤ could be due to poor reactivity of the phospho-IKK antibody toward this isoform. The involvement of the IKK complex proteins in XE-DAR-induced NF-B activation was further studied using dominant-negative mutants of the various component proteins. As shown in Fig. 4B, EDA-A2-induced NF-B activation in 293F-XEDAR cells was efficiently blocked by a dominant-negative mutant of NEMO/IKK␥, which suggests the possibility that a block in XEDAR signaling may contribute to the ectodermal manifestations of patients with X-linked anhidrotic ectodermal dysplasia and immunodeficiency. Consistent with a key role of the IKK complex in XEDAR-induced NF-B activation, kinasedeficient mutants of IKK1/IKK␣ and IKK2/IKK␤ significantly blocked EDA-A2-induced NF-B activation as well (Fig. 4B).
To confirm further the involvement of the IKK complex proteins in XEDAR-induced NF-B activation, we took advantage of the murine embryonic fibroblasts (MEF) derived from IKK␣-, IKK␤-, and NEMO/IKK␥-deficient animals. As shown in Fig.   FIG. 4. Mechanism of NF 4D, transient transfection of XEDAR led to significant NF-B activation in the wild-type MEFs as measured by a luciferasebased reporter assay, and this activity was completely blocked in IKK␤-and NEMO-deficient MEFs. In contrast, transfection of XEDAR in the IKK␣-null MEFs led to a weak NF-B activity as compared with vector-transfected cells. However, both the basal and XEDAR-induced level of NF-B activity in the IKK␣null cells was very low as compared with the wild-type cells. Finally, we studied the effect of a phosphorylation-resistant mutant of IB␣ (IB␣ S32A/S36A) on EDA-A2-induced NF-B. As shown in Fig. 4C, this mutant could almost completely block EDA-A2-mediated NF-B activation. Collectively, the above results indicate that XEDAR-induces NF-B activation via IKK complex-mediated phosphorylation and degradation of IB␣. A previous study (20) reported that the IKK complex is recruited to the TNFR1 in a TNF-␣-dependent fashion. However, we could not detect IKK1/IKK␣ or IKK2/IKK␤ in the receptor complex of EDA-A2-stimulated 293F-XEDAR-L cells (Fig. 3C).
Mutagenesis Analysis of XEDAR-L-induced NF-B Activa-tion-We used C-terminal deletion mutagenesis to map the domain of XEDAR-L responsible for NF-B activation (Fig. 5A). As shown in Fig. 5B, deletion mutant DC37, which is missing the C-terminal 37 amino acids, retains most of the NF-Binducing activity of the full-length protein. On the other hand, deletion mutants DC45, DC53, and DC64, which are missing the C-terminal 45, 53, and 64 amino acids, respectively, retained ϳ20% NF-B activity of the wild-type protein. Finally, almost no NF-B activity was detected in deletion mutants DC69 and DC80, which are missing the C-terminal 69 and 80 amino acid residues, respectively (Fig. 5B). These results suggest that the regions between amino acids 249 -254 and 273-281 are responsible for the NF-B activity of XEDAR, with the latter region accounting for most of this activity. Interestingly, the region between amino acid residues 249 and 254 contains the sequence PTQES that is homologous to the sequence PX-QE(T/S), which is the consensus binding motif for TRAF2, -3, and -5 (21,22). On the other hand the region between 273 and 281 contains the sequence PIECTE, which is homologous to the 293T cells were transfected with 10 ng of empty vector, XEDAR-L wild-type, or its mutants plasmids along with the reporter constructs, and the experiment was performed essentially as described for Fig. 1B. The values shown are averages (mean Ϯ S.E.) of a representative of two independent experiments in which each transfection was performed in duplicate. D, flow cytometric analysis demonstrating cell surface expression of the wild-type and mutant XEDAR proteins. 293T cells (5 ϫ 10 6 ) were transfected with 1 g of empty vector and different N-terminal FLAG-tagged XEDAR-L plasmids. Twenty four hours after transfection, cells were stained with a FLAG antibody (M2 FLAG) followed by (R)-phycoerythrin (PE)-labeled goat anti-mouse antibody and subsequently analyzed by flow cytometry. E, 293T cells (5 ϫ 10 6 ) were transfected with 50 ng of empty vector and different FLAG-XEDAR-L plasmids. After 24 h the cells were treated with 10 ng/ml EDA-A2 for 10 min, lysed, and immunoprecipitated (IP) with control or FLAG beads. Endogenous proteins that co-immunoprecipitated with the receptor were detected by Western blot analysis using the indicated antibodies. The blot was re-probed with an antibody against the FLAG epitope tag to demonstrate expression of various mutant constructs. EE/KK ϭ XEDAR-L E253K/E277K. consensus-binding motif PXEXX (aromatic/acidic) for TRAF6 (23). We generated point mutants in the above two regions to map more precisely the amino acid critical for this activity. Consistent with the results of deletion mutagenesis, a mutant containing a glutamine to lysine change at amino acid 253 (E253K) showed a marginal loss of NF-B activity, whereas a similar mutant at position 277 (E277K) showed a more significant loss. Finally, a double mutant, E253K/E277K (EE/KK) demonstrated almost a complete lack of the NF-B activity, confirming the importance of the two regions in NF-B activation (Fig. 5C).
It is conceivable that reduced NF-B activation by the various point mutants of XEDAR was due to improper folding and lack of cell surface expression of the mutant proteins. In order to rule out this possibility, we transiently transfected 293T cells with an empty vector or various FLAG-tagged XEDAR constructs, and we analyzed the surface expression of the receptors by immunofluorescence labeling of unfixed cells with an antibody against the FLAG tag. As shown in Fig. 5D, a flow cytometric analysis demonstrated that the wild-type and the various mutant XEDAR proteins could be readily detected on the surface of the transfected cells in nearly equivalent amounts, thereby ruling out the possibility of an artifact secondary to misfolding of the mutant proteins.
In order to explain the differential NF-B activity of the above point mutants, we studied their ability to interact with the TRAF molecules. For this purpose, we transfected very small amounts of the wild type or each of the mutant XEDAR plasmids into 293T cells and studied their ability to recruit the endogenous TRAF3 and -6 upon treatment with EDA-A2 (Fig.  5E). Consistent with the previous results, the wild-type receptor co-immunoprecipitated with TRAF6 and two major isoforms of TRAF3, which were ϳ62 and 53 kDa in mass. Interestingly, the point mutant E253K retained the ability to bind TRAF6 and the 62-kDa isoform of TRAF3 but failed to interact with the 53-kDa isoform of TRAF3. In contrast, the E277K mutant completely lost the ability to interact with TRAF6 and weakly interacted with the two TRAF3 isoforms. Finally, the double mutant E253K/E277K (EE/KK), which completely lacks the ability to activate NF-B, failed to interact with either TRAF6 or TRAF3 isoforms. Collectively, the above results support the involvement of TRAF6 and the 53-kDa isoform of TRAF3 in XEDAR-induced NF-B. Furthermore, these results indicate that the Glu-253 residue is critical for the recruitment of the 53-kDa isoform of TRAF3 and might contribute to the recruitment of the 62-kDa isoform of TRAF3 to XEDAR. On the other hand Glu-277 residue is critical for the recruitment of TRAF6 and contributes to the recruitment of the two TRAF3 isoforms.
XEDAR-L Activates the JNK Pathway-In addition to NF-B activation, different members of the TNFR family are also known to activate the JNK pathway. Therefore, we tested the ability of EDA-A2/XEDAR to activate this pathway by measuring the phosphorylation of JNKs, the terminal kinases of this pathway. As shown in Fig. 6A, treatment of 293F-XEDAR-L cells led to a rapid and significant increase in JNK1 and JNK2 phosphorylation as measured by Western blot analysis with a FIG. 6. EDA-A2 and XEDAR activate JNK. A, EDA-A2 induces phosphorylation of both JNK1 and JNK2. 293F-XEDAR-L cells were treated with EDA-A2 for the indicated times. Cell lysates were analyzed by Western blot using phospho-specific JNK and total JNK1 antibodies, respectively. B, EDA-A2 activates the JNK pathway. 293F-XEDAR-L cells were treated with EDA-A2 for the indicated time intervals, and JNK activation was measured by a pull-down JNK assay kit (Cell Signaling). GST-c-Jun coupled to agarose beads was used to both pull down the endogenously expressed JNK and as a substrate for activated JNK-induced phosphorylation. C, EDA-A2 activates c-Jun transcriptional activity. 293F-XEDAR-L cells were transfected with the pFA-c-Jun, pFR-luc, and pRcRSV/LacZ reporter plasmids. Twelve hours posttransfection, cells were treated with control Sf9 supernatant, EDA-A2containing supernatant, and TNF-␣ as indicated. Nine hours later, cell lysates were prepared and used for reporter assays as described under "Materials and Methods." EDA-A2 and TNF-␣ were used at 10 ng/ml. The values shown are averages (mean Ϯ S.E.) of a representative of two independent experiments in which each transfection was performed in duplicate. D, 293EBNA cells were transfected with the indicated plasmids, and the c-Jun transcriptional activity was measured as described under "Materials and Methods." The values shown are averages (mean Ϯ S.E.) of a representative of two independent experiments in which each transfection was performed in duplicate. phospho-JNK-specific antibody. Activation of the JNK pathway leads to the phosphorylation-induced activation of c-Jun transcription factor. As another measure of JNK activation, we tested the ability of EDA-A2 to induce phosphorylation of c-Jun in 293F-XEDAR-L cells using a "pull-down" kinase assay. Consistent with the above results, treatment with EDA-A2 led to rapid and strong phosphorylation of c-Jun in this assay (Fig.  6B). Activation of the JNK pathway by EDA-A2 was confirmed using a reporter assay in which luciferase expression was driven by JNK-mediated phosphorylation of the activation domain of transcription factor c-Jun fused to the GAL4 DNAbinding domain (Fig. 6C). Finally, transient transfection of XEDAR-L or XEDAR-s in 293EBNA cells led to a significant increase in c-Jun transcriptional activity (Fig. 6D), thus arguing against the possibility that EDA-A2 activates JNK via its interaction with some other TNFR family receptor. However, consistent with the previous results with NF-B activation, we did not observe any significant difference in the abilities of XEDAR-L and XEDA-s isoforms to activate the JNK pathway (Fig. 6D).
Mechanism of XEDAR-induced JNK Activation-We used the luciferase-based c-Jun transcriptional assay to understand the mechanism of JNK activation by XEDAR. As shown in Fig.  7A, EDA-A2-induced JNK activation in 293F-XEDAR cells could be effectively blocked by dominant-negative mutants of TRAF2 and TRAF6, which have been shown previously to block JNK activation via several members of the TNFR family (11,24). Similarly, a dominant-negative form of TRAF3 could block EDA-A2-induced JNK activation (Fig. 7B). As shown in Fig.  7C, EDA-A2-induced c-Jun transcriptional activation was also effectively blocked by dominant-negative mutants of ASK1 and JNK1, which are intermediate and terminal kinases of the JNK pathway, respectively (Fig. 7C). Finally, JBD-JIP1, a specific inhibitor of the JNK pathway (25), significantly blocked EDA-A2-induced JNK activity (Fig. 7C). Treatment with TNF␣ was used as a positive control for the above experiments. DN-ASK1 and JBD-JIP1 were also highly effective in blocking JNK activation induced by transient transfection of XEDAR-L in 293EBNA cells (Fig. 7D). However, consistent with our previously published results (8), DN-ASK1 failed to block TAJ-induced JNK activation, thereby suggesting that XEDAR and TAJ use distinct mechanisms for the activation of this pathway.
We next mapped the domain of XEDAR involved in JNK activation. As shown in Fig. 8A, XEDAR-L DC37 was as effective as the wild-type receptor in JNK activation, whereas the mutants DC45, DC53, and DC64 retained only about 20% JNK inducing activity of the wild-type receptor. In contrast, almost a complete lack of JNK activation was seen in the mutants XEDAR-L DC69 and DC80, respectively. Further analysis by point mutagenesis showed that E253K and E277K have reduced JNK activation, whereas the double mutant E253K/ E277K (EE/KK) has lost this activity completely. These results are very similar to the one shown above for the NF-B activa-

DISCUSSION
The members of the tumor necrosis factor family and their receptors have been known to play a central role in the regulation of cellular proliferation, activation, and programmed cell death (26). The recent discovery of mutations in EDA and EDAR in families with X-linked and autosomal forms of hypohidrotic ectodermal dysplasias has led to an increased appreciation of the role of this family in the regulation of embryonic development and epithelial morphogenesis (2,14,27,28). XE-DAR is a recently isolated homolog of EDAR and, like it, is highly expressed in the ectoderm during embryonic development (7). Therefore, it is conceivable that mutations in XEDAR and its downstream signaling components may be responsible for some forms of HED, which are not caused by mutations in EDAR or EDA.
The NF-B pathway has been shown to play an essential role in the process of ectodermal differentiation and hair follicle morphogenesis (6, 29 -31). We and others have demonstrated previously that EDAR activates the NF-B pathway (7,8), and this activity is defective in mutations associated with HED and downless phenotype, thereby suggesting that signaling via EDAR might be responsible for NF-B activation during ectodermal differentiation (6,14,31,32). In the present study, we demonstrate the ability of XEDAR to activate the NF-B pathway in a ligand-dependent fashion. Thus, XEDAR signaling might provide an alternative source of NF-B activation during ectodermal differentiation.
Although both EDAR and XEDAR activate the NF-B, they seem to utilize different proximal signaling intermediates. Recent studies (33,34) suggest the involvement of the death adaptor EDARADD/crinkled in EDAR-mediated NF-B activation and its lack of involvement in XEDAR-induced NF-B. In the present study we demonstrate the involvement of TRAF3 and -6 in the XEDAR-induced NF-B pathway. However, like other members of the TNFR family, both EDAR and XEDAR appear to depend on the IKK complex for activating the NF-B. Therefore, defects in ectodermal differentiation seen in patients with mutations in NEMO/IKK␥ might be due to inhibi-tion of signaling via both these receptors (14).
Our study suggests a major role of TRAF6 in XEDAR-induced NF-B activation. TRAF6 has been implicated previously in NF-B activation via the Toll receptors and IL-1 receptor pathway (35,36). A report (37) published while this manuscript was in preparation suggested the presence of hypohidrotic ectodermal dysplasia in TRAF6 Ϫ/Ϫ animals. These animals were found to have focal alopecia behind the ears, alopecia of the tail, a distinctive kink near the tip of their tail, and lack of sweat gland development, features also seen in Ta, dl, and crinkled (cr) mice. However, unlike Ta, dl, and cr mice TRAF6 Ϫ/Ϫ animals were found to have defect in sebaceous gland development as well. Taken together with our study, the above results support a role of XEDAR signaling in hair, sweat, and sebaceous gland development.
Unlike TRAF6, the role of TRAF3 in NF-B activation is controversial. Some of the earlier studies, based on transient transfection-based overexpression of full-length TRAF3, suggested lack of activation of the NF-B pathway by this adaptor protein. However, more recent studies (38,39) have documented the presence of multiple alternatively spliced isoforms of TRAF3, which, unlike the full-length isoform, are capable of NF-B activation upon transient transfection-based overexpression. In the present study, we have demonstrated EDA-A2dependent recruitment of 62-and 53-kDa isoforms of TRAF3 to XEDAR signaling complex. Interestingly, point mutagenesis studies suggest that the recruitment of the 53-kDa isoform to the XEDAR complex correlates with the NF-B-and JNKinducing ability.
In addition to NF-B, our results also demonstrate the ability of XEDAR to activate the JNK pathway. Although the role of JNK pathway in ectodermal differentiation is not well characterized, this pathway has been shown to be essential for lateral epithelial cell migration in Drosophila, a process essential for dorsal closure during embryogenesis (40 -42). It remains to be seen whether XEDAR-induced JNK activation plays a similar role during epithelial morphogenesis in mammals.
Whereas XEDAR resembles EDAR in the activation of the NF-B pathway, it resembles TAJ/TROY in its ability to activate the JNK pathway (7-9, 43). However, unlike TAJ/TROY, XEDAR-induced JNK activation is dependent on ASK1. Al- though we have previously reported that EDAR can activate the JNK pathway, this property is relatively weak as compared with XEDAR and TAJ/TROY (8,9). Similarly, although TAJ/ TROY has been reported to activate the NF-B pathway in one study (43), this activity was relatively weak, and we have failed to reproduce these results. Thus, EDAR, XEDAR, and TAJ/ TROY, three TNFR family members involved in ectodermal differentiation, differentially activate the NF-B and/or JNK pathway. The distinct signaling properties of the three receptors can be structurally explained by the lack of significant sequence homology in their cytoplasmic domains and their use of distinct proximal signaling intermediates. However, a biological explanation for the need of three receptors with distinctive, yet overlapping, signaling activities in the process of ectodermal differentiation will require further understanding of the downstream targets activated by the NF-B and JNK pathways and their spatial and temporal regulation. It is conceivable that these three receptors differentially control the morphogenesis of different types or stages of hair follicles. Alternatively, they may play distinct roles in the development of different ectodermal derivatives, as has been suggested by the defective development of sebaceous glands in with TRAF6 Ϫ/Ϫ animals, a feature not seen Ta, dl, or cr mice.