Proc. Natl. Acad. Sci. USA Vol. 93, pp. 9437-9442, September 1996 Biochemistry

TRAF5, a novel receptor-associated factor family , mediates CD40 signaling (signal transduction/protein-protein interaction/yeast two-hybrid system) TAKAoMI ISHIDA*, TADASHI ToJo*, TSUTOMU AOKI*, NORIHIKO KOBAYASHI*, TSUKASA OHISHI*, TOSHIKI WATANABEt, TADASHI YAMAMOTO*, AND JUN-ICHIRO INOUE*t Departments of *Oncology and tPathology, The Institute of Medical Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato-ku, Tokyo 108, Japan Communicated by David Baltimore, Massachusetts Institute of Technology, Cambridge, MA, May 22, 1996 (received for review March 8, 1996)

ABSTRACT Signals emanating from CD40 play crucial called a death domain, suggesting that these receptors could roles in B-cell function. To identify molecules that transduce have either common or similar signaling mechanisms (13). CD40 signalings, we have used the yeast two-hybrid system to Biochemical purification of receptor-associated or the clone cDNAs encoding proteins that bind the cytoplasmic tail recently developed cDNA cloning system that uses yeast of CD40. A cDNA encoding a putative signal transducer genetic selection led to the discovery of two groups of signal protein, designated TRAF5, has been molecularly cloned. transducer molecules. Members of the first group are proteins TRAF5 has a tumor necrosis factor receptor-associated factor with a TRAF domain for TNFR2 and CD40 such as TRAF1, (TRAF) domain in its carboxyl terminus and is most homol- TRAF2 (17), and TRAF3, also known as CD40bp, LAP-1, or ogous to TRAF3, also known as CRAF1, CD40bp, or LAP-1, CRAF1 or CD40 receptor-associated factor (18-20). Those of a previously identified CD40-associated factor. The amino second group are proteins with a death domain for FAS and terminus has a RING finger domain, a cluster of fingers TNFR1 such as FADD (21), also known as MORT1 (22), RIP and a coiled-coil domain, which are also present in other (23), and TRADD (24). members of the TRAF family protein except for TRAF1. In CD40 signalings were reported to include modulation of the vitro binding assays revealed that TRAF5 associates with the activity of nonreceptor-type tyrosine kinases such as Lyn, Fyn, cytoplasmic tail of CD40, but not with the cytoplasmic tail of and Syk, activation of phosphatidylinositol-3-kinase, phos- tumor receptor factor receptor type 2, which associates with phorylation of phospholipase C-y2 (25-27), activation of the TRAF2. Based on analysis of the association between TRAF5 Rel/nuclear factorKB (NFKB) transcription factors (28), and and various CD40 mutants, residues 230-269 of CD40 are induction of the Bcl-XL, Cdk4, and Cdk6 proteins (29). It has required for the association with TRAF5. In contrast to also been shown that tyrosine kinase activity is essential for TRAF3, overexpression of TRAF5 activates transcription factor CD40-mediated rescue of the germinal center B cells from nuclear factor #cB. Furthermore, amino-terminally truncated (30). Among these signals, the Rel/NFKB activation forms ofTRAF5 suppress the CD40-mediated induction ofCD23 was shown to be mediated by TRAF2 (31). Although it was expression, as is the case with TRAF. These results suggest that demonstrated that all three known TRAF proteins can be TRAF5 and TRAF3 could be involved in both common and recruited to the cytoplasmic domain of CD40, the precise distinct signaling pathways emanating from CD40. mechanism by which CD40 mediates diverse signals remains to be elucidated. To characterize the initial stage of signaling by CD40, we have used the yeast two-hybrid system to identify CD40 is expressed in late B cells in bone marrow, mature B cDNAs encoding a novel protein containing a TRAF domain, cells, and certain accessory cells, including bone-marrow de- TRAF5, which is closely related to TRAF3 (18-20). We also rived dendritic cells and follicular dendritic cells (1-3), and is show that TRAF5 is involved in CD40 signaling linked to a receptor for CD40 ligand (CD40L) present on activated NFKB activation and CD23 expression. CD4+ T cells (4). Signaling through CD40 rescues B cells from apoptosis induced by crosslinking of the surface immunoglob- MATERIALS AND METHODS ulin M (IgM) complex (5) and also induces B cells to differ- Yeast Two-Hybrid System. DNA encoding the intracellular entiate and to undergo Ig isotype switching (6, 7). CD40L has domain of mouse CD40 (amino acids 216-306) was cloned into been shown to be defective in patients with X-linked hyper IgM the yeast LexA DNA-binding domain vector pBTM116. The syndrome, whose B cells do not form germinal centers and resulting plasmid, pBTM40cyt, was used as bait in two-hybrid produce only IgM and/or IgD (8). In addition, their B cells screens of a murine C57 Black Kaplan cDNA library fused to have the ability to switch from IgM to IgG, IgE, or IgA the activation domain of Gal4 in the pACT plasmid (CLON- production in vitro by the stimulation of CD40 (9). This genetic TECH). Seventy-two of the 2 x 106 transformants screened evidence strongly supports the idea that the intercellular grew in the absence of histidine and had detectable f3-galac- communication through CD40-CD40L is essential for germi- tosidase staining within 20 min of incubation. To remove nal center formation and Ig class switching. clones containing either TRAF3 or TRAF2, plasmids ex- CD40 is a member of the tumor necrosis factor receptor tracted from each yeast colony were subjected to Southern (TNFR) superfamily, which includes TNFR1 and TNFR2 (10, blotting probed with TRAF3 and TRAF2 cDNA. Ten clones, 11), lymphotoxin 13 receptor (12), Fas antigen (13), OX40 (14), which were not hybridized with either of two probes, were used CD30 (15), and the low-affinity nerve receptor (16), all of which share a ligand-binding domain composed of Abbreviations: TRAF, tumor necrosis factor receptor-associated fac- tandemly repeated cysteine-rich modules. Among these, Fas tor; NFKB, nuclear factor KB; CD40L, CD40 ligand; IgM, immuno- antigen and TNFR1 have significant similarity in their cyto- globulin M; TNFR, tumor necrosis factor receptor; GST, glutathione plasmic domain over 46 amino acids, a part of the domain S-transferase; CAT, chloramphenicol acetyltransferase. Data deposition: The sequence reported in this paper has been deposited in the GenBank database (accession no. D83528). The publication costs of this article were defrayed in part by page charge ITo whom reprint requests should be addressed at: Department of payment. This article must therefore be hereby marked "advertisement" in Oncology, The Institute of Medical Science, The University ofTokyo, accordance with 18 U.S.C. §1734 solely to indicate this fact. 4-6-1 Shirokanedai, Minato-ku, Tokyo 108, Japan. 9437 9438 Biochemistry: Ishida et al. Proc. Natl. Acad. Sci. USA 93 (1996) in a cotransformation assay with pBTM40cyt or control pBT- CD23 Expression Assay. Mouse WEHI-231 B-cell clones MLamin bait to reconfirm the specificity. Then these clones expressing the FLAG-C40-3 protein were generated by intro- were subjected to nucleotide sequencing. ducing pME-FLAG-C40-3 together with an expression vector cDNA Cloning and Northern Blotting. The cDNA insert of for the puromycin resistant (pApuro) followed by selec- approximately 1 kbp from two-hybrid clone C40-3 was used as tion in the presence of 0.5 ,ug/ml of puromycin (38). For a probe to screen mouse testis cDNA library in AZAPII controls, cells were transfected with pApuro alone and sub- (Stratagene, provided by M. Ohsugi) by standard methods. jected to selection. More than 10 clones were isolated and Two independent positive clones were obtained and subjected expression of FLAG-C40-3 was checked by Western blotting. to nucleotide sequencing by the BcaBest sequence system Every clone was confirmed to express normal levels of mouse (Takara Shuzo, Shiga). The total RNA from various tissue was CD40. WEHI-231 cell clones with or without expression of prepared as described (32). The purification of poly(A)+ RNA FLAG-C40-3 were stimulated with mouse CD40L-CD8 chi- was performed using oligo(dT) latex (Takara Shuzo). Seven meric protein (39) for 48 hr. For controls, medium was added micrograms of poly(A)+ RNA was separated by 1% formal- instead of stimulator. After stimulation, cells were stained with dehyde denaturing agarose gel and transferred to nylon mem- fluorescein isothiocyanate-conjugated anti-CD23 antibody brane (Hybond N, Amersham). The filter was incubated with followed by FACScan analysis using the Lysis II program. Side 32P-labeled C40-3 cDNA probe in hybridization buffer [0.2 M scatter values were also measured to confirm that the charac- NaHPO4, pH 7.2/1 mM EDTA/1% (wt/vol) BSA/7% (wt/vol) terization of cells was not dramatically changed during the SDS] at 65°C. The filter was finally washed with 0.5 x drug selection process. SSC/0.2% (wt/vol) SDS at 65°C for 30 min. Glutathione S-Transferase (GST) Fusion Protein and the RESULTS Binding Assay. Plasmids encoding CD40 mutants with Thr-254 Structure and Expression of TRAF5. From 2 x 106 clones to Ala-254 substitutions, carboxyl-terminal truncation, or in- of a murine C57 Black Kaplan T-cell lymphoma cDNA library, ternal deletion were generated by the method of Kunkel (33). 10 independent clones were isolated that met all specific DNA fragments encoding cytoplasmic tail of CD40 and its criteria for binding to the cytoplasmic tail of CD40 in yeast. mutants were subcloned into the GST fusion protein vector Nucleotide sequencing of all cDNA fragments revealed that pGEX2T (Pharmacia LKB) and transformed into the Esche- three (clones C40-3, C40-6, and C40-72) encoded portions of richia coli strain BL21. GST and GST fusion proteins were the same protein, which has a TRAF domain in its carboxyl prepared by standard methods (34), and the recombinant terminal region (Fig. 1A). Using the C40-3 cDNA as a probe, proteins were immobilized onto glutathione-agarose beads at cDNA encompassing an entire coding region was obtained a concentration of 0.2 mg/ml. To prepare cell extract, Cos-7 from a murine testis cDNA library. The longest cDNA clone cells (106) were transfected with SRa promoter-driven expres- contained 2105 bp, a size consistent with the 2.2 kb transcript sion vector pME18S (K. Maruyama, unpublished results) of as shown in Fig. 2. The complete nucleotide sequence has been the protein encoded by the C40-3 cDNA and tagged with deposited in the GenBank database. The open reading frame FLAG epitope (Eastman Kodak) at its amino terminus (pME- starts at nucleotide 104 and ends at nucleotide 1864. Because FLAG-C40-3). Cells were 36 hr after transfection, harvested, the first initiation codon, ATG, is located at nucleotides and lysed with TNE buffer [10 mM Tris-HCl, pH 7.8/1% 188-190, this cDNA can encode a protein of 558 amino acids (wt/vol) Nonidet P-40/0.15 M NaCi/10 mM iodoacetoamide/1 with a calculated molecular weight of 64,145. Homology mM EDTA/10 ,ug/ml searching against the Protein Identification Resource database aprotinin] followed by centrifugation. revealed that the encoded protein was most homologous to One-half of the lysate was incubated with 1 ,ug of GST fusion TRAF3 (Fig. 1B) (18-20). Since it is a fifth member of the protein bound to glutathione-agarose beads for 1 hr at 4°C. After TRAF family proteins (see below), we termed this protein washing the beads, samples were boiled in the presence of 0.1% "TRAF5." alignment of TRAF5 and TRAF3 SDS followed by immunoprecipitation using the anti-FLAG revealed that the two proteins share a common overall struc- antibody M2. The complexes were separated on polyacrylamide/ ture (Fig. 1 A and B). The TRAF5 sequence is also similar to 12.5% SDS gels and the FLAG-C40-3 protein was detected by that of TRAF1 and TRAF2 (17), which can associate with the Western blotting using the anti-FLAG antibody M2 and alkaline cytoplasmic tail of the TNF-a receptor type 2 (TNFR2). The phosphatase-coupled anti-mouse IgG. carboxyl terminus of TRAF5 is homologous to that of other Cell Culture, Plasmid, and Transient Transfection Assays. TRAF proteins for approximately 150 amino acids, wherein Human Jurkat T cells were cultured in RPMI 1640 medium TRAF5 is 66.2%, 60.8%, 60.0%, and 41.7% identical to supplemented with 10% fetal bovine serum. Human 293T TRAF3, TRAF1, TRAF2, and CART1(TRAF4) (41), respec- kidney cells were cultured in DMEM supplemented with 10% tively (Table 1). This domain, termed the TRAF-C domain, fetal bovine serum. cDNA for TRAF5 and TRAF3 cloned by functions as a protein-protein interaction domain. In addition PCR were inserted into the SRa-driven expression vector, to the TRAF-C domain, TRAF5 has three potential domains pME18S, to make pME-TRAF5 and pME-TRAF3, respec- including a coiled-coil domain (42), a cluster of five zinc tively. One microgram of the reporter plasmid, [KB]6TK-CAT fingers, and a RING finger domain (43) (Fig. 1A). The or [KBM]6TK-CAT (35), 1 gg of ,3-galactosidase expression putative RING finger domain contained a CX2CX11- vector driven by ,B-actin promoter (,B-actin/,B-gal) and the CX1HX2CX2CX12CX2D (C3HC3D) motif reminiscent of the indicated amounts of pME-TRAF5 or pME-TRAF3 were C3HC4 consensus sequence of the RING finger, which is cotransfected by the DEAE-dextran method into Jurkat cells. found in other known RING finger proteins, including TRAF3 The amount of DNA transfected was always adjusted to 5 ,ug and TRAF2. TRAF5 contains an aspartic acid residue instead with a control expression vector, pME18S. For 293T cells, 1 ,ig of the last cysteine residue of the C3HC4 motif. However, a of the reporter plasmid, 1 ,ug of 13-actin/p-gal, and the indi- new member of the TRAF family of proteins, cated amounts of pME-TRAF5 or pME-TRAF3 were cotrans- CART1(TRAF4), which is highly expressed in breast carci- fected by the calcium phosphate method. The amount of DNA nomas, contains a C3HC3D RING finger in its amino terminus transfected was always adjusted to 22 ,ug with a control like TRAF5 (41), suggesting that C3HC3D could also function expression vector, pME18S. Forty hours after transfection, cell as a RING finger. Five zinc fingers are present carboxyl extracts were prepared by freeze-thawing followed by centrifu- terminal to the RING finger. Four amino acids that would gation. ,B-galactosidase activity (36) was used to standardize contact metal are arranged in the pattern CX2-6CX1IHX3-7- transfection efficiency. Chloramphenicol acetyltransferase C(H), which is different from the classic containing (CAT) assays were performed for 1 hr at 37°C as described (37). CX24CX11,12HX24 (44). Almost identical patterns of zinc Biochemistry: Ishida et al. Proc. Natl. Acad. Sci. USA 93 (1996) 9439 A CO. RING finger o - c) = 3 a ~ > . ua NH - nlCzfln.gerfledo,..ITRAF-Cl -COOH

...... -^^~ ^ C40-3 . **E -."*:-'- C40-6 28S - C40-72 B 18S- a.w3.. .e:..,Sr.- TRAF5 TRAF5 1 ...... MAHSEEQAAVPCAFIRQNSGNSISLDFEPDTEYQFVEQLEERY 43 :'1- ...... :.:..: ..1*[1 :1::1 TRAF3 1 MESSKKMDAAGTLQPNPPLKLQPDRGAGSVLVPEQGGYKEKFVKTVEDKY 50 . RING.finger : 44 KCAFCHSVLHNPHQTGCGHRFCQQCIRSLRELNSVPICPVDKEVIKPQEV 93 f, * *s!: - II. (. 111-11I11111II I: *1 .1...... 11 ' p-actin 51 KCEKCRLVLCNPKQTECGHRFCESCMAALLSSSSPKCTACQESIIK.DKV 99 - - -- ,:

HI--~. - 94 FKDNCCKREVLNLHVYCKN.APGCNARIILGRFQDHLQH.CSFQAVPCPN 141 GAPDH

100 FKDNCCKREILALQVYCRNEGRGCAEQLTLGHLLVHLKNECQFEELPCLR 149 Zn finger .- 142 ESCREAMLRKDVKEHLSAYCRFREEKCLYCKRDIVVTNLQDHEENSCPAY 191 FIG. 2. Northern blot analysis of TRAF5 mRNA in mouse tissues. .l' :llll":t'. 11::l1..11 .( 1. ...1. 1..--1 1I- RNA extraction and blotting were performed as described. The 150 ADCKEKVLRKDLRDHVEKACKYREATCSHCKSQVPMIKLQKHEDTDCPCV 199 membrane filter was hybridized with murine TRAF5 probe (Top). An arrow indicates the of TRAF5 mRNA Positions of 192 PVSCPNRC.VQTIPRARVNEHLTVCPEAEQDCPFKHYGCTVKGKRGNLLE 240 position (2.2 kb). JIll"l 111' -..:.. .lll...*t..1 11 ..... 28S and 18S ribosomal RNA are indicated. To confirm the intactness 200 WSCPHKCSVQTLLRSELSAHLSECVNAPSTCSFKRYGCVFQGTNQQIKA 249 and the equal loading of RNA, the same filter was rehybridized with both /-actin (Middle) and glyceraldehyde 3-phosphate dehydrogenase 241 HERAALQDHMLLVLEKNYQLEQRISDLYQSLEQKESKIQQLAETVKKFEK 290 1I :1 1:.-. .1: . :, . (GAPDH, Bottom) probes, because these two are expressed in 250 HEASSAVQHVNLLKEWSNSLEKKVSLLQNESVEKNKSIQSLHNQICSFEI 299 somewhat different amounts in various tissues. coiled-coil 291 ELKQFTQMFGRNGTFLSNVQ. ALTSHTDKSAWLEAQVRQLLQIVNQQPSR 339 terize the CD40-TRAF5 the 1:··-: 1l:':':1. 1'* 1: interaction, cytoplasmic region 300 EIERQKEMLRNNESKILHLQRVIDSQAEKLKELDKEIRPFRQ...... NW 343 of wild-type and mutant CD40 were expressed as GST fusion proteins and immobilized to glutathione-agarose beads (Fig. 340 LDLRSLVDAVDSVKQRITQLEASD...... QRLVLLEGETSKHDAHINI 382 : ....Im .. Il:'II. III.:l .: 3A). Cell extracts prepared from Cos-7 cells transfected with 344 EEADSMKSSVESLQNRVTELESVDKSAGQAARNTGLLESQLSRHDQTLSV 393 an expression vector for the portion of TRAF5 encoded by the C40-3 cDNA and with FLAG 383 HKAQLNKNEERFKQLEGACYSGKLIWKVTDYRVKKREAVEGHTVSVFSQP 432 tagged epitope (FLAG-C40-3). 1. --- : 11. 11-1.11 1111 II: .1-111 1.-1:1::I111l The C40-3 protein contains the TRAF domain and part of the 394 HDIRLADMDLRFQVLETASYNGVLIWKIRDYKRRKQEAVMGKTLSLYSQP 443 zinc fingers, but does not have the RING finger domain (Fig. - TRAF-C Cell extracts were then mixed with GST fusion 433 FYTSRCGYRLCARAYLNGDGSGKGTHLSLYFVVMRGEFDSLLQWPFRQRV 482 1A). protein 111: II:1 . )l-l t11 lllJ11 111 11 1-111[: 1: attached to agarose beads and the FLAG-C40-3 protein 444 FYTGYFGYKMCARVYLNGDGMGKGTHLSLFFVIMRGEYDALLPWPFKQKV 493 precipitated with GST fusion protein beads was detected by 483 TLMLLDQSGKKNHIVETFKADPNSSSFKRPDGEMNIASGCPRFVSHSTLE 532 immunoprecipitation followed by Western blotting using the 111l:ll.....I . ..'1)1[1]:.11111111 .1... anti-FLAG M2 494 543 antibody (see Materials and Methods). Fusion TLMLMDQGSSRRHLGDAFKPDPNSSSFKKPTGEMNIASGCPVFVAQTVLE of the cytoplasmic tail of wild-type CD40 with GST efficiently 533 NSKNTYIKDDTLFLKVAVDLTDLEDL 558 interacted with TRAF5, whereas GST protein could not, 1: 1111111 1:11 11 .11[1 544 NG..TYIKDDTIFIKVIVDTSDLPD.. 566 indicating that the binding shown in this assay was specific (Fig. 3B). Their interaction was significantly reduced when Thr-254 FIG. 1. Structure of the TRAF5 protein. (A) Diagram of the of CD40 was converted to Ala (TA mutant; our amino acid domain structure of the TRAF5 protein. Three thick lines indicate numbering starts from the first methionine), an alteration regions encoded by the three independent cDNA fragments cloned known to disable CD40 signaling linked to growth inhibition using the yeast two-hybrid system. (B) Homology between TRAF5 and (45) (Fig. 3B). TRAF3, which mediates CD40 signals linked to TRAF3. The amino acid sequences were compared using 'Gap' as in CD23 expression, failed to associate with the TA mutant as implemented the University of Wisconsin GCG program (40). well This result that TRAF5 could also Identity (43.8%) and similarity (64.4%) of amino acids are indicated (18). strongly suggests by vertical lines and dots, respectively. The horizontal dotted lines mediate CD40 signaling. We next analyzed various deletion indicate gaps between the sequences of the two proteins. The RING mutants of CD40 (Fig. 3A). Carboxyl-terminally truncated finger, zinc finger, coiled-coil, and TRAF-C domains are indicated by mutants A270 could associate with TRAF5 as efficiently as arrows. Table 1. Comparison of the TRAF family proteins in the fingers are present in TRAF3, TRAF2, and CART1(TRAF4). TRAF-C domain of homology Use of the COILS algorithm (42) revealed a discrete coiled-coil domain spanning residues 251-403 and flanked by regions TRAFS TRAF3 TRAF1 TRAF2 TRAF4 without the potential to form coiled-coils. TRAF5 66.2 (83.4) 55.0 (75.8) 57.7 (77.9) 46.7 (67.1) The TRAF5 mRNA is highly expressed in lung and mod- TRAF3 60.0 (78.2) 60.8 (78.4) 41.7 (60.9) erately expressed in thymus, spleen, and kidney (Fig. 2). Low TRAF1 64.2 (81.8) 42.3 (60.4) levels of the transcripts were detected in brain and liver. TRAF2 46.6 (68.2) the were not detected in other However, transcripts tissues Data are presented as percentage amino acid sequence identity such as skeletal muscle, heart, small intestine, and testis by (conservatively substituted amino acid residues in parentheses) in the Northern blot analysis. The 2.2-kb size of the TRAF5 tran- TRAF-C domain. The following regions of each TRAF protein were script confirms that the cDNA clone represents a full-length used for the calculation of identity: amino acids 404-558 of mouse copy of TRAF5 mRNA. TRAF5, 415-566 of mouse TRAF3 (19), 260-409 of mouse TRAF1, Delineating the Domains in the Cytoplasmic Domain of 352-501 of mouse TRAF2 (17), and 308-470 of human CD40 Required for Its Interaction with TRAF5. To charac- CART1(TRAF4) (41). 9440 Biochemistry: Ishida et al. Proc. Natl. Acad. Sci. USA 93 (1996) A rying mutant KB sites ([KBM]6TK-CAT) was transfected under the same conditions as the wild-type reporter construct. In V, * ... KKVAK XKP -~~~ I~~~~~~~IQR Jurkat cells, TRAF5, but not TRAF3, activates KB-site- TA I :_ dependent transcription in a dose-dependent manner (Fig. A270 . 4A). TRAF5 also activated NFKB in 293T cells (Fig. 4B). A246 - However, the magnitude of NFKB activation was not large in A230 M 293T cells as in Jurkat cells, because NFKB is already activated A230-246 _ O A239-246 , _ to some extent without stimulation in 293T cells (Fig. 4B, see A220-239 pME18S columns). Importantly, this pre-activated endoge- nous NFKB activity was suppressed by the expression of TRAF3 (Fig. 4B). Thus, TRAF5 and TRAF3 expression have .& e , B z l b -A opposite effects on NFKB activation. -,, le lfll'. 44 V!. $' 191 EJ IAV AV, fsl AV, TRAF5 Lacking the RING Finger Suppressed the CD40- &I C.j I&I (.] 1 Mediated Induction of CD23 Expression. It has been previ- -46 ously demonstrated that the TRAF domain of TRAF3 inhibits FLAG-C40-3-0- , - - -30 the induction of CD23 expression caused by CD40 signaling (19). To determine whether TRAF5 is also involved in CD40- - . . . . -66 mediated induction of CD23 expression, WEHI-231 cell GST Fusion -46 clones, which stably expressed amino-terminally truncated Protein go-" __ does not -w_ -30 TRAF5 encoded by the C40-3 cDNA fragment, which have a RING finger domain and part of the zinc finger domain but has the TRAF domain, were generated (Fig. SA). Parental FIG. 3. Mapping of the TRAF5 binding domain of the cytoplasmic cells or cell clones expressing no FLAG-C40-3 expressed CD23 tail of CD40. (A) Structure of CD40 mutants. The amino acid after stimulation by the CD40L-CD8 chimeric stimulator for sequence of the cytoplasmic region of CD40 is shown on the top as 2 days (Fig. 5B, left). However, induction of CD23 expression WT. A boxed sequence indicates a death domain-like sequence that was scarcely observed with cell clones, which expressed de- consists of 46 amino acids conserved among TNFR1, Fas, and CD40. tectable levels of FLAG-C40-3 (Fig. 5B, right), indicating that A point mutation and deletions are indicated under the amino acid the amino-terminally truncated form of TRAF5 acts as a sequence. (B) Association of TRAF5 with various CD40 mutants. (Upper) Interaction of FLAG epitope-tagged TRAF5 (amino acids 185-558) with the intracellular regions of CD40 mutants and TNFR2 A 40 . T expressed as GST fusion proteins. A portion ofTRAF5 encoded by the I-- T . Jurkat C40-3 cDNA (see Fig. 1A) was tagged with FLAG-epitope and - 30 transiently expressed in Cos-7 cells. Extracts were incubated with . ---: was . :0 various GST fusion proteins and associated TRAF5 then analyzed E 20 -. as described. An arrow indicates the FLAG-epitope-tagged C40-3 i-.:,':- protein. (Lower) Analysis of GST and various GST fusion proteins. u 10 .-40 Two microliters of GST and various GST fusion proteins attached to f:f::: : .... .: agarose beads were separated on polyacrylamide/12.5% SDS gel and __ ...... _ visualized by Coomassie brilliant blue R-250 staining. ,:: 1.5 3 1.5 3 (qg) wild-type CD40, whereas A246 could not (Fig. 3B), leading to pME-TRAF5 pME-TRAF3 the conclusion that the region between 246 and 269 is required for the association with TRAF5. A small amount of TRAF5 was associated with A230-246, compared with A230 and A246, B lon which scarcely associate with TRAF5. A239-246 and A222-239 as 81o0 293T can associate with TRAF5 almost as efficiently as wild-type CD40. However, we could not rule out the possibility that some iE,:i. degradation products shown in the preparation of GSTA230- 9 246 could inhibit the binding of GSTA&230-246 to the C40-3 21 protein. Thus, we tentatively conclude that either 230-239 or 239-246 is additionally required for efficient association. The 0 ternary structure of the cytoplasmic domain of CD40 and T 10 20 10 20 (..) TRAF-C domain of TRAF5 should be extensively analyzed to pME-TRAF5 pME-TRAF3 understand their precise interaction. TRAF5 does not associ- ate with TNFR2 (Fig. 3B), whereas TRAF3 can weakly FIG. 4. Transcription factor NFKB is activated by TRAF5 overex- associate (20). pression. (A) Activation of NFKB in human Jurkat T-cell line. Jurkat TRAF5 Overexpression Activates NFKcB. One of the impor- cells (2 x 106) were transfected with 1 ,ug of [KB]6TK-CAT (shaded tant signals emanating from CD40 is activation of the tran- columns) or [KBM16TK-CAT (solid columns), 1 jig of pl3actin-f3gal, scription factor NFKB. It has been shown that TRAF2, but not the indicated amounts of pME-TRAF5 or pME-TRAF3, and enough pME18S control plasmid to give 5 jig of total DNA by the DEAE- TRAF3, mediates signals linked to NFKB activation. To dextran method. CAT assays were performed as described. Values examine the possible role of TRAF5 in CD40-mediated NFKB correspond to means SEM of at least three independent experi- activation, transient transfection experiments were performed ments. (B) Activation of NFKB in human 293T kidney cell line. 293T to determine whether TRAF5 expression might lead to acti- cells (106) were transfected with 1 ,ug of [KB]6TK-CAT (shaded vation of transcription from a KB-site-dependent reporter columns) or [KBM]6TK-CAT (solid columns), 1 ,g of pf3actin-f3gal, gene. The KB-site-dependent reporter construct ([KB]6TK- the indicated amounts of pME-TRAF5 or pME-TRAF3, and enough pME18S control plasmid to give 22 ,ug of total DNA by the calcium was cotransfected with a TRAF5 or TRAF3 CAT) (35) phosphate method. CAT assays were performed as described. The expression vector (pME-TRAF5 or pME-TRAF3) into human percentage values of conversion were calculated from the results that Jurkat T cells or human 293T kidney cell lines. To confirm the were in the linear range of CAT activity (<60%). Values correspond specificity of transcription, the same reporter construct car- to means + SEMs of at least three independent experiments. Biochemistry: Ishida et al. Proc. Natl. Acad. Sci. USA 93 (1996) 9441 A interact with the death domain containing molecule TRADD cF - c4 ON r- (47), indicating that crosstalk between TRAF proteins and C1 en 1 cn cn mn death domain proteins could mediate diverse signalings ema- nating from the TNFR family. Because TRAF1 associates with 46 -L --:1:~11-I... ..I...... FLAG TRAF2, CD40 could potentially recruit three members of the __ _ C40-3 TRAF domain family of proteins to transduce signals. Ectopic expression of amino-terminally truncated forms of TRAF2 or (kDa) TRAF3, lacking the RING finger, suppresses CD40-induced NFKB activation and CD23 expression, respectively (19, 31), suggesting that the RING finger could play a key role in signaling. However, the precise mechanisms by which CD40 0 transduces diverse signals remains to be elucidated. co In this study, we have molecularly cloned a cDNA encoding E~ a novel TRAF family protein, designated TRAF5, which binds -0 the cytoplasmic tail of CD40 using the yeast two-hybrid system. 0U) TRAF5 is most homologous to TRAF3; the two proteins z contain similar domain structures including RING finger, zinc finger, coiled-coil, and TRAF-C domains. Thus, it was antic- #27 #33 ipated that TRAF5 might act as a signal transducer, similar to TRAF3. However, TRAF5 and TRAF3 may be involved in 0 I distinct CD40 signaling pathways based on the following

1 results: (i) overexpression of TRAF5 activates NFKB, whereas w S35.8 a 4.6 overexpression of TRAF3 suppresses NFKB activity (Fig. 4); v 1 0 (ii) TRAF5 does not associate with TNFR2 (Fig. 3B), whereas #30 #39 cX TRAF3 can weakly associate (20) indicating that TRAF5 may be more specific for CD40 signaling; (iii) TRAF5 is highly 1101¢4 E expressed in lung, moderately in spleen, thymus, and kidney, IzI= I 1 U) but is expressed in very low levels in other tissue (Fig. 2), 5.4 whereas TRAF3 is expressed in every tissue examined (20). II* 17.4,4 _j However, expression of amino-terminally truncated forms of 'm both TRAF5 and TRAF3 suppressed the induction of CD23 #41 #57 expression triggered by CD40 signaling (Fig. 5 and ref. 19). u Thus, these two transducers are also involved in common CO) -I 1 signal transduction pathways, possibly forming a heterodimer I to cooperate. Although TRAF2, as well as TRAF5, associates I' 'JI.3 49 with CD40 and is involved in NFKB activation, these two I p 16, 1677F-l'k --RI..., -.0. ,--' LJ4 TRAF proteins could mediate signals linked to distinct bio- logical phenomena by interacting with both common signal CD23 Fluorescence transducers involved in NFKB activation and different signal transducers. Thus, effects of CD40 signaling on cell functions FIG. 5. TRAF5 lacking the RING finger suppresses CD40- could in part depend on the expression pattern of TRAF2 and mediated induction of CD23 expression. (A) Expression of FLAG- TRAF5. Because CD40 was shown to be expressed in many cell C40-3 in WEHI-231 cell clones. One hundred micrograms of extracts types, differences in the expression patterns of TRAF2 and were prepared from each cell clone indicated and expression of TRAF5 in various tissues could reflect the diverse effects of FLAG-C40-3 was checked as described. An arrow indicates the CD40 signaling on cell functions in different tissue. FLAG-C40-3 protein. (B) Suppression of CD40-mediated induction of Our previous work revealed that amino acids 246-269 of CD23 expression by ectopic expression of FLAG-C40-3. WEHI-231 CD40 are required for CD40-mediated induction of Bcl-XL and cell clones with or without FLAG-C40-3 expression were stimulated inhibition of apoptosis triggered by surface IgM signaling, and with a CD40-CD8 chimeric stimulator for 48 hr and then analyzed for amino 230-245 are for NFKB activation CD23 expression as described. The x and y axes represent CD23 that acids required fluorescence and side scatter, respectively. The percentages of cells (29). In vitro binding assays show that residues 230-269 of that express CD23 are indicated in the lower right-hand corner of each CD40 are required for efficient association with TRAF5. counter map. With regard to the CD23 expression of unstimulated These results lead to the hypothesis that TRAF5 mediates cells, results from the two representative clones are shown. Essentially signals linked to NFKB activation and also signals inhibiting the same results were obtained with other clones and parental apoptosis. In fact, NFKB is activated by the overexpression of WEHI-231 cells. TRAF5 (Fig. 4). Overexpression ofTRAF5 may induce dimer- ization ofTRAF5, which might mimic dimerization induced by dominant negative regulator of the CD40-mediated induction oligomerization of CD40 following CD40L stimulation. We of CD23 expression. have previously shown that the A246 mutant can activate NFKB as efficiently as wild-type CD40. However, here we show DISCUSSION that A246 associates with TRAF5 very weakly, inconsistent It has been shown that membrane receptors with no intrinsic with the result that TRAF5 activates NFKB. One possibility is enzymatic activity often use associated proteins to transduce that TRAF5 could indirectly associate with A246 via another biological signals. For example, the B-cell receptor, T-cell putative CD40-associated protein. Alternatively, oligomeriza- receptor, and IL-2 receptor 13-chain activate nonreceptor type tion of A246 induced upon CD40L stimulation could allow tyrosine kinases for signal transduction (46). The TNFR TRAF5 to be associated with A246. The role of TRAF5 and family, including Fas and CD40, has been shown to associate TRAF5 mutant proteins in inhibition of apoptosis induced by with downstream signal transducer molecules carrying either a surface IgM signaling is currently under investigation. TRAF or death domain (17-24). TRAF3 and TRAF2 have It has been shown that TRAF2 lacking the RING finger acts been shown to be directly associated with the cytoplasmic tail as a dominant-negative mutant with regard to NFKB activa- of CD40 through their TRAF domain (19, 31). TRAF2 can tion, indicating that the RING finger could be a critical motif 9442 Biochemistry: Ishida et al. Proc. Natl. Acad. Sci. USA 93 (1996) for interaction with downstream molecules. Since TRAF2, 13. Itoh, N., Yonehara, S., Ishii, A., Yonehara, M., Mizushima, S., TRAF3, and TRAF5 have a RING finger domain, it is possible Sameshima, M., Hase, A., Seto, Y. & Nagata, S. (1991) Cell 66, 233-243. that each RING finger interacts with distinct signaling mole- 14. Mallet, S., Fossum, S. & Barclay, A. N. (1990) EMBO J. 9, 1063-1068. cules. In this case, a dominant-negative type experiment may 15. Camerini, D., Walz, G., Loenen, W. A. M., Borst, J. & Seed, B. (1991) not allow us to fully dissect the signaling pathways. This model J. Immunol. 147, 3165-3169. predicts the existence of the molecules interacting with each 16. Johnson, D., Lanahan, A., Buck, C. R., Sehgal, A., Morgan, C., type of RING finger. Some proteins with a RING finger motif Mercer, E., Bothwell, M. & Chao, M. (1986) Cell 47, 545-554. 17. Rothe, M., Wong, S. C., Henzel, W. J. & Goeddel, D. V. (1994) Cell have been implicated in transcription or DNA recombination, 78, 681-692. including RPT-1 (48), SS-A/Ro (49), XNF7 (50), RING1 (51) 18. Hu, H. M., O'Rourke, K., Boguski, M. S. & Dixit, V. M. (1994)J. Biol. and RAD-18 (52). RING finger domain was shown to act as Chem. 269, 30069-30072. DNA binding domain (50, 52). The function of the zinc fingers 19. Cheng, G., Cleary, A. M., Ye, Z-e, Hong, D. I., Lederman, S. & could Baltimore, D. (1995) Science 267, 1494-1497. also be interesting, because it is known that zinc finger 20. Mosialos, G., Birkenbach, M., Yalamanchili, R., VanArsdale, T., domains can serve as DNA-binding or protein-protein inter- Ware, C. & Kieff, E. (1995) Cell 80, 389-399. action domains. Interestingly, a recently identified TRAF 21. Chinnaiyan, A. M., O'Rourke, K., Tewari, M. & Dixit, V. M. (1995) protein, CART1(TRAF4), was shown to be localized in nu- Cell 81, 505-512. cleus (41). These results suggest that the TRAF family pro- 22. Boldin, M. P., Varfolomeev, E, E., Pancer, Z., Mett, I. L., Camonis, J. H. & Wallach, D. (1995) J. Biol. Chem. 270, 7795-7798. teins, including TRAF5, could also translocate to the nucleus 23. Stanger, B. Z., Leder, P., Lee, T.-H., Kim, E. & Seed, B. (1995) Cell where they might directly interact with DNA. The coiled-coil 81, 513-523. domain could also function as a protein-protein interaction 24. Hsu, H., Xiong, J. & Goeddel, D. V. (1995) Cell 81, 495-504. domain, such as . Mutational analysis can be 25. Uckun, F. M., Schieven, G. L., Dibirdik, I., Chandan-Langlie, M., used to dissect the role of the RING finger, zinc finger, and Tuel-Ahlgren, L. & Ledbetter, J. A. (1991) J. Biol. Chem. 266, 17478-17485. coiled-coil domains. 26. Faris, M., Gaskin, F., Parsons, J. T. & Fu, S. M. (1994) J. Exp. Med. At least four members of the TRAF family of proteins could 179, 1923-1931. be recruited to the cytoplasmic region of CD40. Extensive 27. Ren, C. L., Morio, T., Fu, S. M. & Geha, R. S. (1994)J. Exp. Med. 179, studies to define complex formation of CD40 with four distinct 673-680. members of the TRAF 28. Berberich, I., Shu, G. L. & Clark, E. A. (1994) J. Immunol. 153, family of proteins and the discovery of 4357-4366. the downstream signal transducers will clarify the basis of 29. Ishida, T., Kobayashi, T., Tojo, T., Ishida, S., Yamamoto, T. & Inoue, CD40 signaling. J. (1995) J. Immunol. 155, 5527-5535. 30. Knox, K. A. & Gordon, J. (1993) Eur. J. Immunol. 23, 2578-2584. We thank Drs. P. Bartel and S. Fields for pBTM vector, Dr. T. 31. Rothe, M., Sarma, V., Dixit, V. M. & Goeddel, D. V. (1995) Science Watanabe for the expression plasmid for CD40L-CD8 chimeric 269, 1424-1427. protein, M. Ohsugi for mouse testis cDNA library, T. Tezuka and H. 32. Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162, 156-159. Nishizumi for valuable discussions, and Dr. Ronald M. Wisdom for 33. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA 82, 488-492. critical reading of the manuscript. This work was supported by a 34. Smith, D. B. & Johnson, K. S. (1988) Gene 67, 31-40. Grant-in-Aid for Scientific Research on Priority Areas and for Sci- 35. Inoue, J., Kerr, L. D., Ransone, L. J., Bengal, E., Hunter, T. & Verma, entific Research (B) from the Ministry of Education, Science, Sports I. M. (1991) Proc. Natl. Acad. Sci. USA 88, 3715-3719. and Culture of Japan, and a Grant-in-Aid from the Research Fellow- 36. Herbomel, P., Bourachot, B. & Yaniv, M. (1984) Cell 39, 653-662. ships of the Japan Society for the Promotion of Science for Young 37. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol. Cell. Biol. 2, 1044-1051. Scientists. 38. Takata, M., Sabe, H., Hata, A., Inazu, T., Hommma, Y., Nukada, T., 1. Stamenkovic, I., Clark, E. A. & Seed, B. (1989) EMBO J. 8, 1403- Yamamura, H. & Kurosaki, T. (1994) EMBO J. 13, 1341-1349. 1410. 39. Lane, P., Bocker, T., Hubele, S., Padovan, E., Lanzavecchia, A. & 2. Ling, N. R., MacLennan, I. C. M. & Mason, D. (1987) in Leukocyte McConnell, F. (1993) J. Exp. Med. 177, 1209-1213. Typing III, ed. McMichael, A. J. (Oxford Univ. Press, Oxford, U.K.), 40. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic Acids Res. 12, pp. 302-335. 387-395 3. Schriever, F., Freedman, A. S., Freeman, G., Messner, E., Lee, G., 41. Regnier, C. H., Tomasetto, C., Moog-Lutz, C., Chenard, M.-P., Delay, J. & Nadler, L. M. (1989) J. Exp. Med. 169, 2043-2048. Wendling, C., Basset, P. & Rio, M.-C. (1995) J. Biol. Chem. 270, 4. Armitage, R. J., Fanslow, W. C., Strockbine, L., Sato, T. A., Clifford, 25715-25721. K. N., Macduff, B. M., Anderson, D. M., Gimpel, S. D., Davis-Smith, 42. Lupas, A., van Dyke, M. & Stock, J. (1991) Science 252, 1162-1164. T., Maliszewski, C. R., Clark, E. A., Smith, C. A., Grabstein, K. H., 43. Barlow, P. N., Luisi, B., Milner, A., Elliott, M. & Everett, R. (1994) J. Mol. Biol. 237, 201-211. Cosman, D. & Spriggs, M. K. (1992) Nature (London) 357, 80-82. 44. Evans, R. M. & Hollenberg, S. M. (1988) Cell 52, 1-3. 5. Tsubata, T., Wu, J. & Honjo, T. (1993) Nature (London) 364,645-648. 45. Inui, S., Kaisho, T., Kikutani, H., Stamenkovic, I., Seed, B., Clark, 6. Jabara, H. H., Fu, S. M., Geha, R. S. & Vercelli, D. (1990) J. Exp. Med. E. A. & Kishimoto, T. (1990) Eur. J. Immunol. 20, 1747-1753. 172, 1861-1864. 46. Perlmutter, R. M., Levin, S. D., Appleby, M. W., Anderson, S. J. & 7. Defrance, T., Vanbervliet, B., Durand, I., Briolay, J. & Banchereau, Alberola-Ila, J. (1993) Annu. Rev. Immunol. 11, 451-499. J. (1992) Eur. J. Immunol. 22, 2831-2839. 47. Hsu, H., Shu, H.-B., Pan, M.-G. & Goeddel, D. V. (1996) Cell 84, 8. Kroczek, R. A., Graf, D., Brugnoni, D., Giliani, S., Korthauer, U., 299-308. Ugazio, A., Senger, G., Mages, H. W., Villa, A. & Notarangelo, L. D. 48. Patarca, R., Freeman, G. J., Schwartz, J., Singh, R. P., Kong, Q. T., (1994) Immunol. Rev. 138, 39-59. Murphy, E., Anderson, Y., Sheng, F. Y., Singh, P., Johnson, K. A., 9. Nonoyama, S., Hollenbaugh, D., Aruffo, A., Ledbetter, J. A. & Ochs, Guarnagia, S. M., Dufee, T., Blattner, F. & Cantor, H. (1988) Proc. H. D. (1993) J. Exp. Med. 178, 1097-1102. Natl. Acad. Sci. USA 85, 2733-2737. 10. Schall, T. J., Lewis, M., Koller, K. J., Lee, A., Rice, G. C., Wong, 49. Chan, E. K. L., Hamel, J. C., Buyon, J. P. & Tan, E. M. (1991)J. Clin. G. H. W., Gatanaga, T., Granger, G. A., Lentz, R., Raab, H., Kohr, Invest. 87, 68-76. W. J. & Goeddel, D. V. (1990) Cell 61, 361-370. 50. Reddy, B., Kloc, M. & Etkin, L. D. (1991) Dev. Biol. 148, 107-116. 11. Smith, K. A., Davis, T., Anderson, D., Solam, L., Beckmann, M. P., 51. Lovering, R., Hanson, I. M., Borden, K. L. B., Martin, S., O'Reilly, Jerzy, R., Dower, S. K., Cosman, D. & Goodwin, R. G. (1990) Science N. J., Evans, G. I., Rahman, D., Pappin, D. J. C., Trowsdale, J. & 248, 1019-1022. Freemont, P. S. (1993) Proc. Natl. Acad. Sci. USA 90, 2112-2116. 12. Ware, C. F., VanArsdale, T. L., Crowe, P. D. & Browning, J. L. (1995) 52. Jones, J. S., Weber, S. & Prakash, L. (1988) Nucleic Acids Res. 16, Cuirr. Top. Microbiol. Immunol. 198, 175-218. 7119-7131.