NF-κB Activation in CD27 Signaling: Involvement of TNF Receptor-Associated Factors in Its Signaling and Identification of Functional Region of CD27¹

The CD27 Ag is a 120-kDa type I transmembrane phosphoglycoprotein composed of disulfide-linked 55-kDa monomers (1, 2, 3, 4) and belongs to the TNF receptor family, including TNF receptor (TNFR I and TNFR II), low affinity nerve growth factor receptor (p75^LNGFR), CD40, CD30, CD95 (Fas/Apo I), 4–1BB, and OX40 (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19). CD27 is expressed on medullary thymocytes (1, 20), on the surface of 75% of human peripheral blood T cells (equally distributed between CD4⁺ and CD8⁺ subsets) (1, 2), on a subpopulation of B cells (21, 22), and on NK cells (23). The activation of human T cells via the TCR-CD3 complex leads to expression of CD27, which peaks after 3 to 4 days and then gradually decreases after prolonged activation (24). CD70, a ligand for CD27, is a type II transmembrane glycoprotein belonging to the TNF family and is expressed by activated T and B cells (25, 26, 27). CD27/CD70 interaction augments the induction of T cell activation (28), B cell activation (IgG and IgM production) (29, 30), and NK cell killing activity (31).

The TNF receptor-associated factors (TRAFs)³ are composed of at least six members (TRAF1–6) (32, 33, 34, 35, 36, 37, 38, 39). TRAF1 and TRAF2 were found to associate with the cytoplasmic domain of TNFR II as a heterodimer (32). TRAF3 was identified as a molecule binding to cytoplasmic domain of CD40 (33, 34, 35), but was subsequently known to associate with the lymphotoxin-β receptor, TNFR II, and CD30. TRAF4 (CART1) was cloned from a breast carcinoma cell line, and this is the only TRAF that has been localized to the nucleus and may not be associated with a surface receptor (36). TRAF5 was shown to bind to the lymphotoxin-β receptor (37, 38). TRAF6 was identified as a mediator of the IL-1 signal (39). TRAF proteins share a conserved approximately 230-amino acid C-terminal TRAF domain (TRAF-N and TRAF-C) and also N-terminal ring finger and zinc fingers (except TRAF1). The TRAF-C domain is responsible for oligomerization and association to each molecule of the TNF receptor family. The TRAF-N domains of TRAF1 and TRAF2 have been shown to mediate the binding to members of the family of inhibitors of apoptosis proteins (40). Functionally, TRAF2 is a common signal mediator for TNFR II, CD40, and CD30, which lead to activation of the transcription factor, NF-κB (41, 42), while TRAF3 is a common signal inhibitor for them. It was recently reported that TRAF2 binds TRADD (the TNFR I-associated death domain-containing protein) (43) and mediates the activation of NF-κB by TNFR I (44). Malinin et al. have cloned NIK (NF-κB-inducing kinase) (45, 46), which is similar to mitogen-activated protein kinase kinase kinase and binds to TRAF2 and I-κB kinase-1 and -2 (IKK-1 and -2). These three kinases form a complex, and this complex phosphorylates I-κBα and I-κBβ, followed by their degradation (47, 48, 49).

In the current study, we demonstrated that CD27 signal activates NF-κB activity and suggested that this signal may be mediated by TRAF2, as the other member of TNFR family did, although CD27 is a unique member in that it forms disulfide-linked homodimer. In addition, TRAF3 displayed an inhibitory effect on this signal. This activated NF-κB complex contained at least two components, p50 and p65. Finally, we determined the essential cytoplasmic 13-amino acid alignment (residues 238–250 amino acids) of the CD27 molecule that is needed not only for the NF-κB-activating signal but also for its interaction with TRAFs. This 13-amino acid alignment contained the EEEG sequence, which is critical to CD30 or CD40 for their association with TRAF1 or TRAF2 (50, 51), and the PIQED sequence, which is similar to PXQXT and is critical to TNFR II or CD30 for their interaction with TRAF2 and TRAF3.

293 cells were maintained in DMEM with 10% (v/v) FCS.

Rabbit polyclonal anti-NF-κB p65 Ab (0.1 mg/ml; sc-109), rabbit polyclonal anti-NF-κB p50 Ab (0.1 mg/ml; sc-114), rabbit polyclonal anti-NF-κB p52 Ab (0.1 mg/ml; sc-297; Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti-human CD27 Ab (M-T271; Ancell Technology Research Products, Bayport, MN), mouse monoclonal anti-human CD70 Ab (HNE51; Immunotech, Luminy, France), mouse IgG1 (BioPur, Bubendorf, Switzerland), mouse anti-c-Myc mAb (OP-10; Oncogene, Cambridge MA), and horseradish peroxidase-coupled goat anti-rabbit IgG(H+L) (Zymed, South San Francisco, CA) were purchased.

DNA fragments encoding human CD27 (hCD27), human CD70 (hCD70), human TRAF1 (hTRAF1), and human TRAF3 (hTRAF3) were amplified by PCR from the human cDNA library prepared from activated mono-nuclear cell mRNA with the following specific primers: 5′hCD27 (5′-CAGCAAGCTTAACTGGGCACAGAAAGGAGCCG-3′), 3′hCD27 (5′-AGGTTCTAGATCAGGGGGAGCAGGCAGGC-3′), 5′hCD70 (5′-GCTCTCTAGACCCATCGCCCCTCCTGCGCTA-3′), 3′hCD70 (5′-TCAGTCTAGATGGTCAGGGGCGCACCCACTG-3′), 5′hTRAF1 (5′CCCTGAATTCGCCTCCAGCTCAGGCAGCAGTC3′), 3′hTRAF1 (5′-CCCGGGATCCCTAAGTGCTGGTCTCCACAATG-3′), 5′hTRAF3 (5′-TCCTGAATTCGAGTCGAGTAAAAAGATGGACT-3′), and 3′hTRAF3 (5′-CCTCGGATCCTACTTATCAGGGATCGGGCAGA-3′). DNA fragment encoding murine TRAF2 (mTRAF2) was amplified by PCR from a murine splenic cDNA library using the following primers: 5′mTRAF2 (5′-GTAAGGATCCTGGCTGGCTGCAGCCAGTGTGACTTC-3′) and 3′mTRAF2 (5′-GCAGGGATCCCTAGAGTCCTGTTAGGTCCACA-3′). The DNA fragment encoding dominant negative TRAF2 (dnTRAF2) (42), which lacks the N-terminal 86 amino acids of TRAF2 was also amplified by PCR from TRAF2 cDNA using primers 5′dnTRAF2 (5′-GGAAGAATTCCAGTCTATAAGCCCAGGAAGCC-3′) and 3′dnTRAF2 (5′-CCCGGGATCCCTAAGTGCTGGTCTCCACAATG-3′). For expression in mammalian cells, TRAFs were inserted into pCDM8 (52) with epitope tag (Myc) constructs at N-terminal. The epitope tag sequence for Myc was MEQKLISEEDLN. We designated the expression vector for the Myc-TRAF fusion sequence as pCDM-Myc-TRAFs. The cDNA for hCD27 was cloned into the pCDM8 vector. Subsequently, three C-terminus-truncated hCD27 mutants lacking 10, 23, or 41 amino acids (see Fig. 6 A) were generated by PCR. The cDNA for hCD70 was also cloned into the pEFBOS vector. All constructs were confirmed by partial sequencing.

p-55IgκLuc (53) containing three tandemly repeated κB motifs upstream of luciferase structural gene was a gift from Dr. T. Fujita (Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan). pCH110 (Pharmacia Biotech, Uppsala, Sweden) expressing the β-galactosidase gene was used to normalize the luciferase activity.

Basically, 293 cells (1 × 10⁶) were seeded on a 60-mm collagen-coated dish and transfected the next day with 6.5 μg of plasmid DNA/dish by the calcium phosphate method. After 4-h incubation with the DNA-calcium phosphate precipitates, the culture medium was changed to fresh medium. Electrophoretic mobility shift assays (EMSAs) and luciferase assays were performed 48 to 72 h after initiation of the transfection.

Nuclear extract preparation was performed with the modified Schreiber’s method (54). Cells were washed twice with PBS and resuspended in 400 μl of ice-cold lysis buffer containing 10 mM HEPES-KOH (pH 7.8), 10 mM KCl, 0.1 mM EDTA (pH 8.0), 1 mM DTT, and antiprotease mixture (0.5 mM PMSF and 2 μg/ml each of aprotinin, leupeptin, and pepstatin A; final concentrations). Following a 15-min incubation on ice, 25 μl of 10% Nonidet P-40 was added, and then cells were briefly vortexed and centrifuged at 5000 × g for 1 min at 4°C. The supernatants (nonnuclear fractions) were discarded, and the nuclear pellets were resuspended in 100 μl of ice-cold nuclear extraction buffer containing 50 mM HEPES-KOH (pH 7.8), 420 mM KCl, 0.1 mM EDTA (pH 8.0), 5 mM MgCl₂, 20% (v/v) glycerol, 1 mM DTT, and the antiprotease mixture; vortex briefly; and vigorously shaken on a shaking platform at 4°C for 15 min. After that, the extracted nuclear fractions were centrifuged for 15 min (15,000 × g, 4°C), and the resulting supernatants (the nuclear extracts) were aliquoted and stored at −70°C before use. Aliquots of the nuclear extracts were processed for protein content determination using Bradford reagent (Bio-Rad, Herculus, CA). Double-stranded oligonucleotides (5′-AGTTGAGGGGACTTTCCCAGGC-3′) containing NF-κB binding sites were ³²P labeled using polynucleotide kinase (Toyobo, Tokyo, Japan). Binding reactions were performed in 16 μl of binding buffer containing 25 mM HEPES-KOH (pH 7.8), 0.5 mM EDTA-NaOH (pH 8.0), 50 mM KCl, 10% (v/v) glycerol, 0.5 mM PMSF, 0.5 mM DTT, and 2 μg of double-stranded poly(dI-dC). Two micrograms of nuclear extract was then added to the binding mixture and allowed to equilibrate for 15 min at room temperature. Alternatively, for competition or supershift experiments, the above binding reactions were conducted in the presence of a 100-fold excess of cold competitors or 0.3 μg of anti-polyclonal Abs raised against members of the NF-κB/Rel family. Finally, >10⁵ cpm of ³²P-labeled double-stranded oligonucleotides containing the consensus NF-κB sequence were added to the binding mixtures and incubated for 30 min at room temperature. The resulting samples were separated on 4% polyacrylamide gels. After electrophoresis, gels were subsequently dried and exposed to Fuji medical x-ray film (Tokyo, Japan) at −80°C with intensifying screens.

After 2-day incubation of the transfection of 293 cells, the cells were immunostained with anti-human CD27 Ab (M-T271) or anti-human CD70 Ab (HNE51) and a FITC-conjugated goat anti-mouse IgG and analyzed by FACScan (Becton Dickinson, San Jose, CA) for CD27 or CD70 expression, respectively.

Luciferase activity assays were performed using the luciferase assay system (Promega, Madison, WI). Luminescence was quantitated by a chemiluminescence counter (Lumat LB9501, Berthold, Nashua, NH). β-Galactosidase activity was also measured using the β-galactosidase enzyme assay system (Promega).

293 cells (5 × 10⁶ cells) were transfected with each pCDM8-Myc-TRAF together with pCDM8-hCD27 (wild type (wt)) or hCD27 deletion mutant by the calcium phosphate method. Transfected cells were harvested 48 h after transfection. Cells were then washed with PBS and lysed in 400 μl of ice-cold 1% Nonidet P-40 lysis buffer containing 50 mM HEPES (pH 7.55), 150 mM NaCl, 0.4 mM EDTA, 10 μg/ml leupeptin, 1 μg/ml pepstatin A, 10 μg/ml aprotinin, and 100 μg/ml PMSF. Sixty microliters of each cell lysate (cell membrane and cytoplasm fractions) was mixed with 20 μl of 4× sample buffer (containing 0.0625 M Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, and 0.001% bromphenol blue), and 10 μl of each sample was used in SDS-PAGE analysis. One hundred and fifty microliters of each cell lysate was incubated with 1 μg of murine anti-human CD27 Ab (M-T271) or the same amount of isotype-matched murine Ig (control IgG) and 40 μl of protein G. The beads were washed four times with lysis buffer, and bound proteins were resuspended in 40 μl of 2× sample buffer. Ten microliters of each sample was fractioned by 4 to 20% gradient SDS-PAGE and transferred to an Immobilon membrane (Millipore, Bedford, MA). Western blot analysis to detect the Myc epitope was performed with anti-Myc Ab (from rabbit) and visualized with horseradish peroxidase-coupled goat anti-rabbit IgG(H+L) using the enhanced chemiluminescence (ECL) system according to the protocol of the manufacturer (Amersham, Aylesbury, U.K.).

Some members of the TNFR family have been reported to activate the nuclear transcriptional factor, NF-κB. Therefore, we examined whether CD27 could transduce the signal for NF-κB activation by monitoring the luciferase activities in 293 cells. 293 cells were transiently transfected with the CD27 expression vector (pCDM8-hCD27) or pCDM8 with no insert as a control vector. In pCDM8-hCD27-transfected 293 cells, NF-κB activation was only 2 to 5 times higher than that in control cells. However, when we cotransfected pCDM8-hCD27 together with pEFBOS-hCD70 in 293 cells, the activity was up-regulated to a level about 20 times higher than the control value (Fig. 1). In pEFBOS-hCD70-transfected 293 cells, no NF-κB activation was found. These results indicate that expression of CD27 alone is not sufficient for the full level of NF-κB activation in 293 cells, and additional CD70 stimulation augmented this signal, although the activation produced by the CD27/CD70 system is still lower than that produced by the CD30 signaling system. Expression profiles of hCD27 and hCD70 molecules in transient transformants were demonstrated in Figure 2.

To determine the involving NF-κB components, we performed EMSAs with the ³²P-labeled double-stranded NF-κB oligonucleotides (Fig. 3,A). EMSAs showed that CD70 did not activate NF-κB, CD27 alone did, and CD27 together with CD70 induced the most activity. These results are identical with that obtained by the luciferase activity assay system. In addition, each Ab to NF-κB subunit p65 or p50 caused induction of a supershift of the protein-NF-κB probe complex, but anti-p52 Ab had no effect (Fig. 3 B). These results indicate that the signal from CD27 can activate the NF-κB complex, which contains at least two components, p65 and p50, in 293 cells.

In the NF-κB activation signal of the TNFR II, CD40, and CD30 systems, either TRAF2 or TRAF3 works as an attaching molecule to the cytoplasmic region of the receptors and plays an important role in their signals (32, 33, 34, 35, 41, 42). We tried to determine whether the cytoplasmic domain of CD27 would bind to TRAFs by a yeast two-hybrid system. However, any obvious interactions between the cytoplasmic domain of CD27 and each TRAF (TRAF1–3) were not observed (data not shown). We speculated that the dimerization of CD27 may be necessary for the interaction of CD27 with TRAFs, because CD27 is a unique molecule among the TNF receptor family in that it is expressed in a homodimerized form. Therefore, we investigated the interactions between whole human CD27 and TRAFs by coimmunoprecipitation experiments. pCDM8-hCD27 and each pCDM8-Myc-TRAF were cotransfected in 293 cells, and coimmunoprecipitation of TRAFs with the CD27 molecule was demonstrated by anti-Myc immunoblot. The interaction between either TRAF2 or TRAF3 and the CD27 molecule was observed (Fig. 4). As for TRAF1, the possibility of association with the CD27 molecule was suspected, but the clear interaction was not detected because of its nonspecific background (Figs. 4 and 8). These results indicate that at least TRAF2 and TRAF3 can interact with the cytoplasmic region of the CD27 molecule and raise the possibility that CD27 might use TRAFs as effectors for NF-κB activation.

To elucidate the involvement of TRAFs in NF-κB activation in CD27 signaling, we cotransfected each pCDM8-TRAF with both pCDM8-hCD27 (wt) and pEFBOS-hCD70 and performed reporter assays, because the interaction between TRAFs and the CD27 molecule was shown in a coimmunoprecipitation experiment. From the data of the luciferase activity assays, intact TRAF2 augmented the NF-κB activation (Fig. 5). This result suggests that TRAF2 is a mediator of the signal induced by CD27/CD70 interaction. However, this result must be interpreted with caution, because overexpression of TRAF2 alone also induces NF-κB activation (Ref. 42, our data, and data not shown). Therefore, we constructed pCDM8-dnTRAF2. As we expected, dnTRAF2 clearly blocked the NF-κB activation triggered by CD27/CD70 interaction (Fig. 5). These two experiments indicate that the CD27 signal for NF-κB activation should be transmitted through at least TRAF2. TRAF1 also reduced this NF-κB activation, although the reduction level was partial. In contrast to TRAF2, TRAF3 blocked NF-κB activation completely, suggesting that TRAF3 works as a negative regulator in CD27 signaling, as demonstrated previously in TNFR II, CD40, and CD30 signaling.

To identify the cytoplasmic region of CD27 necessary for NF-κB activation, we constructed the expression vectors for the three kinds of CD27 mutants, lacking 10, 23, or 41 amino acids at the carboxyl-terminus (Fig. 6,A). All the mutations were confirmed by partial DNA sequencing, and the expression profiles of each mutant CD27 molecule and CD70 molecule in transiently cotransfected 293 cells were demonstrated in Figure 6,C. On each transformant, wild-type or mutant CD27 molecules and CD70 molecules were expressed. Then in cotransfected 293 cells with either pCDM8-hCD27 (wt) or one of the hCD27 deletion mutants and pEFBOS-hCD70, we measured NF-κB activities by monitoring luciferase activities. Deletion of 10 carboxyl-terminal amino acids did not affect NF-κB activation, but the mutant lacking 23 or 41 amino acids failed to activate NF-κB completely (Fig. 7). These results show that a 13-amino acid alignment (residues 238–250 amino acids) is critical for NF-κB activation by CD27/CD70 interaction. Subsequently, using CD27 (wt) or these CD27 deletion mutants, their interaction with each TRAF was examined by coimmunoprecipitation and Western blot analyses. In accordance with the results of luciferase assays, the interaction between TRAFs (TRAF2 and 3) and the CD27 (wt) molecule or the CD27 mutant lacking 10 carboxyl-terminal amino acids was detected; however, the obvious association of TRAFs (TRAF2 and 3) with the CD27 mutant lacking 23 amino acids was not observed (Fig. 8). Again, the high background of TRAF1 was observed (Fig. 8), and we failed to make it clear whether TRAF1 could bind to the CD27 molecule. From these results, the 13-amino acid alignment (residues 238–250 amino acids) is also critical for interaction with TRAFs as well as activation of NF-κB signaling.

CD27 is a member of TNFR family that includes two groups. One group has a death domain motif in its own receptor, and its activation induces apoptosis, and the other group induces NF-κB activation as an intracellular signal. Several reports suggested that CD27 belongs to the latter group by showing that CD27/CD70 interaction activates T, B, and NK cells, and CD27 does not have a death domain (28, 29, 30, 31). In this study we directly demonstrate that CD27/CD70 interaction induces NF-κB activation. Overexpression of CD27 molecules alone activates NF-κB signal in 293 cells, but the induction level is weaker than that of any other of this group’s molecules such as CD30 or CD40. Coexpression of CD70 together with CD27 remarkably augments this signal; therefore, CD70 stimulation is needed for sufficient NF-κB activation signal in 293 cells.

In the cascade of NF-κB activation by CD27, we suggest that TRAF2 may mediate this signal, and TRAF3 has an inhibitory effect on this signal, because either dominant negative TRAF2 or intact TRAF3 inhibits the CD27-mediated NF-κB action. This phenomenon was also seen in TNFR type II-, CD30-, and CD40-mediated signals. Interaction between TRAFs and these TNFR family members was demonstrated in several reports, by the yeast two-hybrid system, and by coimmunoprecipitation and Western blot analyses using mammalian cells. First, we performed the yeast two-hybrid assays, but we could not observe any interaction between the cytoplasmic region of CD27 and TRAFs. Because CD27 exists as a disulfide-bonded homodimer, we supposed that the yeast two-hybrid system could not work well. For this reason, we performed a coimmunoprecipitation experiment using mouse anti-human CD27 Ab (M-T271) and 293 cells transiently expressing whole CD27 molecule and either of the TRAFs. As a result, we observed the association between CD27 and TRAF2 or TRAF3. It is not clear why we could not detect the interaction between CD27 and TRAFs in the yeast two-hybrid system, but we think it is plausible that the dimerized form of the CD27 molecule might be necessary for the interaction between CD27 and each TRAF molecule. Additional experiments might be needed to prove our suggestion.

With the truncated forms of CD27, we determined the critical region involved in NF-κB activation. From the data of single transformants, CD27 (−23 amino acids) and CD27 (−41 amino acids) could not activate NF-κB completely, but CD27 (−10 amino acids) could activate it, although the activation level is lower than that of intact CD27. This might be caused by the slight decrease in CD27 expression (as shown in Fig. 6 B) and/or by the reduced affinity for TRAFs due to subtle conformational change by deletion of carboxyl-terminal 10 amino acids. From the reporter assays using the double transformant of CD27 mutant and CD70, the deletion of 10 carboxyl-terminal amino acids did not affect the CD27 signal, but the deletion of 23 amino acids almost completely abrogated the signal, indicating that this 13-amino acid alignment (residues 238–250 amino acids) is essential for the signal. From coimmunoprecipitation and Western blot analyses, this alignment is also necessary for interaction with TRAFs as well as NF-κB activation. There have been several reports demonstrating the essential domains of TNFR family in its signaling cascades (34, 50, 51, 55, 56). As for CD30, two functional domains exist at the C-terminus (50, 51): element 2A encompasses the sequence PEQET and binds to TRAF1, TRAF2, and TRAF3; element 2B, which is located distal of element 2A, encompasses the sequence EEEGKE and binds to TRAF1 and TRAF2, but not to TRAF3. CD40 contains the EEEGKE motif, and TNFR II contains the PVQET motif. The 13 amino acids critical for CD27 signaling include the EEEG motif, similar to EEEGKE, and there is also a PIQED sequence, similar to PEQET or PVQET at the distal region of EEEG motif. We show at least TRAF2 and TRAF3 can bind to this 13-amino acid sequence of CD27; therefore, it is supposed that TRAF2 may bind to CD27 at both motifs, and TRAF3 may bind at the PIQED motif, although the PIQED sequence is slightly different from the PEQET motif (also recognized as PXQXT). Single amino acid point mutation experiments have been reported as for the interaction between TRAFs and the CD30 molecule (57). From their reports, each single amino acid substitution of this PXQXT motif blocks their interaction, and other amino acid mutants that contain mutations around the intact PXQXT motif also partially or completely inhibit the interaction. Therefore, PXQXT is the core sequence, but the overall conformation including this core sequence is thought to be important for the association.

However, in the case of CD40, Thr²³⁴ (in the PVQET motif) has been shown to be important in the growth arrest signal in the murine thymoma cell line EL4 and the murine B lymphoma cell line M12 (55), suggesting that the phosphorylation by the TNFR family or some protein kinases is necessary for the interaction of TRAFs through the PXQXT motif, although there is no other suggestive evidence for their possibility. In CD27, the PIQED motif, similar to PXQXT, does not contain any amino acid residue that can be targeted by protein kinase. Therefore, we generated some mutants of CD27 molecules with point mutation at serine, threonine, or tyrosine residue around the critical region and examined the effect on NF-κB activity. We performed the reporter assays using cotransformants with mutant CD27 molecule and CD70 (wt), and we observed that in each transformant (Ser²⁴³→Ala, Thr²⁴⁴→Ala, or Tyr²⁵¹→Phe; Fig. 6 A), NF-κB activity slightly decreased compared with CD27 (wt), but not significantly (data not shown). This result indicates that at least these three mutated amino acids may not be essential for NF-κB activation.

In the CD27 signaling cascade, not as much information has been reported to date. Kobata et al. (58) demonstrated that CD27 ligation induces calcium mobilization, protein kinase C activation, and protein tyrosine kinase activation that causes tyrosine phosphorylation of cellular substrates, including ZAP-70. Here, we demonstrated that CD27 signaling cascade contains at least the TRAF2-NF-κB activation pathway. Recent reports indicated that TRAF2 activates NIK, and NIK, in turn, activates IKK-1 and IKK-2; these three kinases form a complex, resulting in the phosphorylation of I-κB and its degradation, which leads to NF-κB activation. At present, it is unlikely that the activation of TRAF-NF-κB activation cascade induces calcium mobilization, protein kinase C activation, or protein tyrosine kinase activation. Therefore, CD27 may also activate multiple signaling pathways as observed in tyrosine kinase receptors and cytokine receptors.

We thank Dr. T. Fujita (Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan) for generously providing us the vector p-55IgκLuc, and Drs. Y. Katada and T. Tatekawa (Department of Medicine III, Osaka University Medical School, Osaka, Japan) for technical support of FACS analysis.