Proc. Natl. Acad. Sci. USA Vol. 96, pp. 9803–9808, August 1999 Immunology

Targeted disruption of Traf5 causes defects in CD40- and CD27-mediated activation

HIROYASU NAKANO*†‡,SACHIKO SAKON*†,HARUHIKO KOSEKI†§,TOSHITADA TAKEMORI¶,KURISU TADA*, ࿣ MITSURU MATSUMOTO ,EIKO MUNECHIKA*, TSUYOSHI SAKAI**, TAKUJI SHIRASAWA**, HISAYA AKIBA*†, TETSUJI KOBATA*†,SYBIL M. SANTEE††,CARL F. WARE††,PAUL D. RENNERT‡‡,MASARU TANIGUCHI†§§, HIDEO YAGITA*†, AND KO OKUMURA*†

*Department of Immunology, Juntendo University, School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan; †Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Corporation, 2-3 Surugadai, Kanda, Chiyoda-ku, Tokyo 101-0062, Japan; Departments of §Molecular Embryology and §§Molecular Immunology, Graduate School of Chiba University, 1-8-1 Inohana, Chuo-ku, Chiba 260-0856, Japan; ¶Department of Immunology, ࿣ National Institute of Infectious Diseases, 1-23-1 Toyama, Shinjuku-ku, Tokyo 162-0052, Japan; Division of Informative Cytology, Institute for Enzyme Research, University of Tokushima, 3-18-15 Kuramoto, Tokushima 770-8503, Japan; **Department of Molecular Genetics, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakae-cho, Itabashi-ku, Tokyo 173-0015, Japan; ††Division of Molecular Immunology, La Jolla Institute for Allergy and Immunology, 10355 Science Center Drive, San Diego, CA 92121; and ‡‡Department of Immunology and Inflammation, Biogen, Inc., Cambridge, MA 02142

Edited by Elliott D. Kieff, Harvard University, Boston, MA, and approved June 23, 1999 (received for review February 4, 1999)

ABSTRACT TRAF5 [tumor necrosis factor (TNF) recep- family, MAP kinase͞ERK kinase kinase 1 (MEKK1) and tor-associated factor 5] is implicated in NF-␬B and c-Jun apoptosis signal-regulating kinase 1 (ASK1) (16, 18). These NH2-terminal kinase͞stress-activated kinase activa- results indicated that the two signaling pathways to activation tion by members of the TNF receptor superfamily, including of NF-␬B and JNK͞SAPK diverge downstream of TRAFs. CD27, CD30, CD40, and lymphotoxin-␤ receptor. To investi- Although these pathways have been investigated extensively, gate the functional role of TRAF5 in vivo, we generated other signaling cascades potentially mediated by TRAFs have TRAF5-deficient mice by gene targeting. Activation of either been largely unknown. Moreover, although both TRAF2 and NF-␬B or c-Jun NH2-terminal kinase͞stress-activated protein TRAF5 are recruited to CD27, CD30, CD40, OX40, lympho- kinase by tumor necrosis factor, CD27, and CD40 was not toxin-␤ receptor (LT-␤R), herpesvirus entry mediator, and ؊͞؊ ؊͞؊ abrogated in mice. However, traf5 B cells showed receptor activator of NF-␬B, it remains to be determined defects in proliferation and up-regulation of various surface whether TRAF2 and TRAF5 have individually specific func- molecules, including CD23, CD54, CD80, CD86, and Fas in tions or act redundantly in transmitting signals via these response to CD40 stimulation. Moreover, in vitro Ig produc- receptors (3). ؊͞؊ tion of traf5 B cells stimulated with anti-CD40 plus IL-4 Recent generation of TRAF2-deficient mice and transgenic was reduced substantially. CD27-mediated costimulatory sig- mice expressing a dominant negative form of TRAF2 ؊͞؊ nal also was impaired in traf5 T cells. Collectively, these (TRAF2-DN) revealed that TRAF2 is requisite for JNK͞ results demonstrate that TRAF5 is involved in CD40- and SAPK activation but not for NF-␬B activation by TNF (19, 20). CD27-mediated signaling. Whereas TRAF2- and TRAF3-deficient mice die earlier (19, 21), TRAF3-deficient mice showed impaired T-dependent Tumor necrosis factor receptor (TNFR) superfamily members immune response (21). Collectively, these results suggested transmit signals regulating proliferation, differentiation, and that each TRAF could act redundantly or specifically in apoptosis in various types of cells (1, 2). TNFR-associated particular signaling cascades. factors (TRAFs) emerged as a novel family of downstream To examine the functional role of TRAF5 in vivo,we mediators on the signal-transduction pathway of the TNFR generated TRAF5-deficient mice by gene targeting. Whereas superfamily (3). To date, six members of the TRAF family NF-␬BorJNK͞SAPK activation by CD27 or CD40 was not have been identified (4–12). All TRAFs, except for TRAF1, abrogated, we found that CD27- and CD40-mediated lym- Ϫ͞Ϫ are composed of N-terminal zinc RING finger and multiple phocyte activation was substantially impaired in traf5 zinc fingers, coiled-coil, and C-terminal receptor-binding . (TRAF) domains. With the exception of TRAF4, all the other TRAFs have been shown to interact directly with some MATERIALS AND METHODS members of the TNFR superfamily lacking death domains (3). TRAF2 also interacts indirectly with death domain receptors Cells. Murine mastcytoma P815, murine CD70-transfected via the adapter molecules, TRADD and RIP (13). Overex- P815 (CD70-P815), and murine CD80-transfected P815 pression of TRAF2, -5, and -6 activates NF-␬B, and truncated (CD80-P815) have been described (22) and were maintained TRAF2, -5, and –6, which lack Zn-binding domains, act as a in RPMI 1640 medium containing 10% FCS. dominant negative inhibitor in receptor-mediated NF-␬B ac- Generation of traf5؊͞؊ Mice by Gene Targeting. A genomic tivation (9, 10, 14), suggesting that these TRAFs are common DNA clone for traf5 was isolated by screening a 129͞Sv͞J mediators for NF-␬B activation by TNFR superfamily mem- mouse genomic DNA library with cDNA encoding a RING bers. It has been shown that TRAF2, -5, and -6 recruit finger domain of TRAF5 as a probe. The cDNA encoding the NF-␬B-inducing kinase (NIK), which, in turn, activates I␬B RING finger domain contained two separate exons, which kinases (15–17). TRAF2, -5, and -6 also participate in the were designated exon I and exon II. The targeting construct ͞ activation of c-Jun NH2-terminal kinase (JNK) stress- (T5-KO) was made by replacing the exon II encoding a activated protein kinase (SAPK), which is mediated by two members of the MAP kinase kinase kinase (MAPKKK) This paper was submitted directly (Track II) to the Proceedings office. Abbreviations: TNFR, tumor necrosis factor receptor; TRAF, TNFR- The publication costs of this article were defrayed in part by page charge associated factor; JNK, c-Jun NH2-terminal kinase; SAPK, stress- activated protein kinase; LT-␤R, lymphotoxin-␤ receptor; ES, embry- payment. This article must therefore be hereby marked ‘‘advertisement’’ in onic stem; LPS, lipopolysaccharide. accordance with 18 U.S.C. §1734 solely to indicate this fact. ‡To whom reprint requests should be addressed. E-mail: hnakano@ PNAS is available online at www.pnas.org. med.juntendo.ac.jp.

9803 Downloaded by guest on September 26, 2021 9804 Immunology: Nakano et al. Proc. Natl. Acad. Sci. USA 96 (1999)

C-terminal half of the RING finger domain and surrounding anti-CD23, anti-CD54, anti-CD80, anti-CD86, or anti-Fas introns with a pMC1neo gene cassette (Stratagene) in the mAb and analyzed by flow cytometry. reverse orientation to the endogenous traf5 gene (Fig. 1A). Electrophoretic Mobility-Shift Assay (EMSA). EMSA was Linealized T5-KO then was transfected into embryonic stem performed essentially as described (9). Briefly, 3 ϫ 107 thy- (ES) cells (R1 cells) by electroporation. Neomycin-resistant mocytes were incubated with agonistic anti-CD27 mAb (10 ES clones were selected by G418 and GANC as described (23). ␮g͞ml; PharMingen) on ice for 30 min. Then, the cells were Homologous recombinants were identified by Southern blot- washed with ice-cold PBS and crosslinked with prewarmed ting. Genomic DNA was isolated and digested with EcoRI or goat anti-hamster Igs (100 ␮g͞ml; Cappel) at 37°C for 15 min. BamHI, and Southern blots were hybridized with the 3Ј For CD40 or LPS stimulation, 2 ϫ 107 splenocytes were flanking probe (probe A) as shown in Fig. 1A. Germ-line incubated with anti-CD40 mAb (10 ␮g͞ml) at 37°C for 15 min chimeras were generated by the aggregation method as de- or LPS (5 ␮g͞ml) for 30 min. Then, the nuclear extracts were scribed (24). The resulting male chimeras were backcrossed prepared. The nuclear extracts were subjected to EMSA. with C57BL͞6J females, and germ-line transmission in F1 Reactions were subjected to 6% PAGE and analyzed on a Fuji traf5ϩ͞Ϫ mice was verified by Southern blot analysis. Geno- BAS2000 image analyzer. typing of the F2 mice was performed by PCR analysis of tail In Vitro Kinase Assay. In vitro kinase assay was performed ϫ 6 genomic DNA. The PCR primers (P1, 5Ј-GGG TCA TGC essentially as described (27). Briefly, 5 10 thymocytes were ␮ ͞ CAC TTG TTC GA-3Ј, and P2, 5Ј-ACC CAC ACG AGG incubated with agonistic anti-CD27 mAb (10 g ml) on ice for AAG GTC TGA-3Ј; see Fig. 1A) were designed to encompass 30 min. Then, the cells were washed with ice-cold PBS and ␮ ͞ the 0.7-kb fragment containing exon II and surrounding crosslinked with prewarmed goat anti-hamster Igs (100 g ml) ϫ introns, which was replaced by pMC1neo, resulting in a 1.2-kb at 37°C for 15 min. For CD40 stimulation, purified B cells (5 6 ␮ ͞ fragment in the targeted allele. 10 ) were incubated with anti-CD40 (10 g ml) at 37°C for 15 Western Blot Analysis. Total lung lysates were prepared min. The cells were lysed in the lysis buffer containing 1 mM from traf5ϩ͞ϩ, traf5ϩ͞Ϫ, and traf5Ϫ͞Ϫ littermates by homoge- NaF and 0.1 mM Na3VO4, incubated with rabbit polyclonal nization in a lysis buffer containing 20 mM Hepes (pH 7.2), 150 anti-JNK (Santa Cruz Biotechnology), and precipi- mM NaCl, 0.5% Triton-X100, 2 mM EDTA, 1 ␮g͞ml aproti- tated with protein G-Sepharose beads (Amersham Pharma- nin, 1 ␮g͞ml leupeptin, 1 mM PMSF, and 1 ␮g͞ml pepstatin. cia). The immunoprecipitates then were subjected to in vitro TRAF2 and TRAF5 were affinity-purified from the lung kinase assay by using GST-c-Jun (1–79) (a gift from E. Nishida, lysates by using glutathione-Sepharose (Amersham Pharma- Kyoto University) as a substrate. The reaction was stopped by addition of Laemmli’s sample buffer, and phosphorylated cia) preadsorbed with glutathione S-transferase (GST) or a ͞ fusion protein GST-LT-␤R containing the cytoplasmic region were subjected to 12% SDS PAGE and analyzed with of human LT-␤R (9). After three washes in the lysis buffer, the a Fuji BAS2000 image analyzer. In Vitro Ig Production. Small resting B cells (1 ϫ 105) were bound proteins were eluted and fractionated by 10% SDS͞ stimulated with anti-CD40 (10 ␮g͞ml) plus IL-4 (10 ng͞ml) for PAGE and blotted onto polyvinylidene fluoride membrane 6 days. Concentrations of IgM and IgG1 in the culture (Millipore). The Western blot was probed with polyclonal supernatants were measured by ELISA. In brief, plates were against N-terminal peptide (TRAF5-N) or C- coated with 5 ␮g͞ml monoclonal rat anti-mouse IgM or IgG1 terminal peptide of TRAF5 (TRAF5-C) (Santa Cruz Biotech- (PharMingen) overnight at 4°C. Plates were washed and nology). The blot then was developed by using an enhanced blocked with 1% BSA-PBS. Diluted culture supernatants were chemiluminescence system according to the manufacturer’s incubated for 1 hr. Then, plates were washed and bound IgM instructions (Amersham Pharmacia). As a control, the same and IgG1 were determined with biotin-conjugated rat anti- membrane was reprobed with a polyclonal antibody against mouse IgM or anti-IgG1 followed by avidin-biotinylated per- N-terminal peptide of TRAF2 (a gift from D. V. Goeddel, oxidase complex (Vector Laboratories). Tularik, South San Francisco) (25). Analysis of Humoral Immune Responses. Humoral response Flow Cytometric Analysis. Single-cell suspensions were to a T-dependent was analyzed essentially as described prepared from the thymus, spleen, lymph nodes, and bone (28). Briefly, traf5ϩ͞ϩ and traf5Ϫ͞Ϫ mice (8-12 weeks old) were marrow. Cells were incubated with FITC- or phycoerythrin immunized with (4-hydroxy-3-nitrophenyl-acetyl)-chicken (PE)-conjugated mAbs and analyzed by using a FACScan gamma globulin (molar ratio, 22:1; NP22-CG) adsorbed to (Becton Dickinson). The following mAbs from PharMingen alum. Immunization was performed by i.p. injection of 5 ␮gof were used: anti-B220, anti-CD4, anti-CD8, anti-CD3, anti- ͞ ␮ NP22-CG alum on day 0 and an i.p. boost with 5 gof CD23, anti-CD54, anti-Fas, anti-CD80, and anti-CD86. ͞ NP22-CG alumn on day 21. Sera were collected on days 7, 14, Proliferation of B and T Cells. Small resting B cells were 21, and 28. Total and high-affinity, NP-specific antibodies were prepared from splenocytes as described (22). Small resting B determined by an ELISA against NP22-BSA- or NP2-BSA- ϫ 5 ␮ ͞ cells (1 10 ) were stimulated with 10 g ml anti-CD40 mAb coated ELISA plates (Immulon), respectively. NP-specific (HM40–3; PharMingen), 10% volume of culture supernatant antibodies of IgM or IgG1 type were detected as described of cells producing murine CD40L-CD8 chimeric molecule above. Arbitrary units of NP-specific antibodies were calcu- ͞ ␮ ͞ (26), 10 ng ml IL-4 (R&D Systems), 10 g ml affinity-purified lated by using an NP-specific mAb C6 (for IgG1) and standard goat anti-mouse IgM antibody (Jackson ImmunoResearch immune sera (for IgM) as the standards. Statistical analysis was Laboratories), or 5 ␮g͞ml lipopolysaccharide (LPS; Sigma). performed by Student’s t test, Welch’s t test, or nonparametric Thymocytes (2 ϫ 105͞well) were cocultured with irradiated Mann–Whitney test. P815, CD70-P815, or CD80-P815 (2 ϫ 104͞well) in the pres- ence of anti-CD3 mAb (1 ␮g͞ml) as described (22). These cultures were pulsed for the last 8 hr of a 48-hr culture period RESULTS with 0.5 ␮Ci͞well of [3H]thymidine (Amersham Pharmacia). Normal Development of TRAF5-Deficient Mice. To eluci- Cells then were harvested onto glass-fiber filters, and the date the function of TRAF5 in vivo, we disrupted the murine radioactivity was measured in a micro-␤-counter (Micro Beta traf5 gene in ES cells. A targeting vector was designed to Plus; Wallac, Gaithersburg, MD). replace the exon II encoding the C-terminal portion of the Up-Regulation of Surface Molecules on B Cells. Small RING finger domain of TRAF5 with a neomycin-resistant resting B cells were cultured for 48 hr in the absence (un- gene (Fig. 1A). Heterozygous ES cell lines containing a stimulated) or presence of anti-CD40 mAb (10 ␮g͞ml) or LPS mutated traf5 allele were used to generate germ-line chimeras (5 ␮g͞ml). The cells then were stained with FITC-conjugated by the aggregation method (24), and traf5ϩ͞Ϫ animals were Downloaded by guest on September 26, 2021 Immunology: Nakano et al. Proc. Natl. Acad. Sci. USA 96 (1999) 9805

whether CD27-mediated costimulation was impaired in traf5Ϫ͞Ϫ thymocytes. Our previous study showed that murine CD70 or CD80 transfectants (CD70-P815 or CD80-P815) enhanced proliferation of anti-CD3-stimulated T cells (22). Traf5ϩ͞ϩ and traf5Ϫ͞Ϫ thymocytes then were cultured with P815, CD70-P815, or CD80-P815 in the presence of a subop- timal dose of anti-CD3 mAb as described (22). Although CD80-P815 enhanced proliferation of traf5ϩ͞ϩ and traf5Ϫ͞Ϫ thymocytes comparably, traf5Ϫ͞Ϫ thymocytes showed consis- tently lower responses than traf5ϩ͞ϩ thymocytes when costimu- lated by CD70-P815 (Fig. 2A). This indicated that CD27- mediated costimulation was impaired specifically in traf5Ϫ͞Ϫ thymocytes. We next examined whether CD27-mediated activation of NF-␬B and JNK͞SAPK was impaired in traf5Ϫ͞Ϫ thymocytes. Thymocytes from traf5ϩ͞ϩ and traf5Ϫ͞Ϫ mice were stimulated with agonistic anti-CD27 mAb followed by goat anti-hamster Igs to aggregate CD27. The cells then were assayed for activation of NF-␬B and JNK͞SAPK by EMSA and in vitro kinase reactions. Unexpectedly, the activation of either NF-␬B or JNK͞SAPK by CD27 was not altered significantly in traf5Ϫ͞Ϫ thymocytes (Fig. 2B). ؊͞؊ FIG. 1. Targeted disruption of the murine traf5 gene. (A) A portion CD40-Mediated Activation Is Impaired in traf5 B Cells. of the traf5 gene containing 5 exons (solid boxes) is indicated (Top). TRAF5 has been shown to interact with CD40 and truncated The targeting construct (Middle) was designed to replace exon II, TRAF5-inhibited surface up-regulation of CD23 by CD40 which encodes the C-terminus of the RING finger domain with a (12). Small resting B cells from traf5ϩ͞ϩ and traf5Ϫ͞Ϫ mice were pMC1neo cassette in a reverse orientation. (Bottom) The mutant locus resulting from homologous recombination. An extra EcoRI restriction stimulated with agonistic anti-CD40 mAb or LPS as a control. Both of these treatments enhanced the expression of CD23, site introduced by the pMC1neo cassette was used for the detection of ϩ͞ϩ homologous recombination. (B) Southern blot analysis of genomic CD54, CD80, CD86, and Fas on traf5 B cells. However, the DNA from wild-type and traf5ϩ͞Ϫ (clone 46, 48) ES cells. DNA was up-regulation of these molecules on anti-CD40-stimulated digested with EcoRI and hybridized with probe A as described in A. traf5Ϫ͞Ϫ B cells was reduced substantially when compared with (C) PCR analysis of genomic DNA from wild-type, traf5ϩ͞Ϫ, and ϩ͞ϩ Ϫ͞Ϫ traf5 B cells (Fig. 3A). On the other hand, the up-regulation traf5 animals. Tail DNA was amplified by PCR using P1 and P2 of these molecules on LPS-stimulated traf5Ϫ͞Ϫ B cells was primers as described in A.(D) Western blot analysis of TRAF5 protein ϩ͞ϩ ϩ͞Ϫ Ϫ͞Ϫ comparable to traf5 B cells. We next examined prolifera- expression. Total lung lysates from wild-type, traf5 , and traf5 tion of B cells in response to various stimuli. Proliferative mice were subjected to affinity purification by using glutathione beads Ϫ͞Ϫ preadsorbed with GST-LT-␤R fusion protein (L) or control beads responses of traf5 B cells to anti-CD40, CD40 ligand, and anti-CD40 plus anti-IgM were reduced substantially when preadsorbed with GST protein alone (G). The positive control lysate ϩ͞ϩ (P) was prepared from 293 cells transfected with TRAF5 expression compared with those of traf5 B cells (Fig. 3B). Moreover, vector. The bound materials were fractionated by SDS͞PAGE, and the addition of IL-4 partially restored the response. In contrast, Western blot was probed with polyclonal antibodies raised against a proliferative responses of traf5Ϫ͞Ϫ B cells to IL-4, anti-IgM, or peptide derived from TRAF5-N, TRAF5-C, or N-terminal LPS were comparable to those of traf5ϩ͞ϩ B cells. Collectively, TRAF2 (25). these results indicated a specific defect in CD40-mediated activation of traf5Ϫ͞Ϫ B cells. obtained. Heterozygous animals were interbred to obtain Upon stimulation with anti-CD40 plus IL-4 in vitro, B cells homozygotes. Homologous recombination and the genotypes differentiate and produce IgG1 as well as IgM. To examine were confirmed by Southern blot analysis (Fig. 1B)orPCR Ϫ͞Ϫ whether the defect in CD40-mediated signals of traf5 B (Fig. 1C). TRAF5 protein expression in tissue was determined by affinity purification by using GST-LT-␤R (9) and subse- quent Western blot analysis by using polyclonal antibodies against TRAF5-N, TRAF5-C, or N-terminal TRAF2 peptides. This analysis demonstrated the absence of intact or truncated TRAF5 protein in traf5Ϫ͞Ϫ animals and also showed that deletion of traf5 gene did not affect the expression of TRAF2 (Fig. 1D). TRAF5-deficient mice were born at the expected Mendelian ratios (ϩ͞ϩ:ϩ͞Ϫ:Ϫ͞Ϫϭ118:203:121). These mice were healthy and showed no obvious abnormalities through 24 weeks of age. There was no developmental defect of lymph nodes and Peyer’s patches, indicating that LT-␤R signaling during embryogenesis of secondary lymphoid organs FIG. 2. Impairment of CD27-mediated costimulation but not is competent (28). Flow cytometric analysis of the expression in NF-␬BorJNK͞SAPK activation. (A) Thymocytes isolated from of CD3, CD4, CD8, and B220 on thymocytes, splenocytes, or traf5ϩ͞ϩ (solid bars) and traf5Ϫ͞Ϫ(hatched bars) mice were stimulated lymph node cells showed that lymphocyte composition was not with anti-CD3 mAb in the presence of irradiated P815, P815-CD70, or altered in traf5Ϫ͞Ϫ mice when compared with traf5ϩ͞ϩ mice P815-CD80 for 48 hr. Incorporation of [3H]thymidine was measured Ϯ (data not shown). during the last 8 hr. Data are shown as mean cpm SEM from the CD27-Mediated Costimulation Is Impaired in traf5؊͞؊ T triplicate samples and represent one of five independent experiments with similar results. (B) Thymocytes isolated from traf5ϩ͞ϩ and Cells. Engagement of CD27 by its ligand CD70 or agonistic Ϫ͞Ϫ traf5 mice were stimulated with anti-CD27 followed by crosslinking anti-CD27 mAb provides a costimulatory signal for T cell with goat anti-hamster Igs for 15 min. EMSAs for NF-␬B activation proliferation, and both TRAF2 and TRAF5 are implicated in (Upper) and in vitro kinase assay for JNK͞SAPK activation (Lower) CD27-mediated activation of NF-␬B and JNK͞SAPK (27). were performed as described in Materials and Methods. B, oligonu- Because CD27 is expressed on most thymocytes, we asked cleotide probe bound to NF-␬B complexes. Downloaded by guest on September 26, 2021 9806 Immunology: Nakano et al. Proc. Natl. Acad. Sci. USA 96 (1999)

FIG. 3. CD40-mediated activation in vitro.(A) Up-regulation of CD23, CD54, CD80, CD86, and Fas by anti-CD40 or LPS stimulation. Purified B cells from traf5ϩ͞ϩ (thin line) and traf5Ϫ͞Ϫ (thick line) were stimulated with anti-CD40 mAb or LPS or were unstimulated (dotted line). After 48 hr, the cells were stained with FITC-labeled anti-CD23, anti-CD54, anti-CD80, anti-CD86, or anti-Fas mAb. Data represent one of three independent experiments with similar results. (B) Proliferative responses. Purified B cells isolated from traf5ϩ͞ϩ (solid bars) and traf5Ϫ͞Ϫ (hatched bars) mice were stimulated for 48 hr with anti-CD40 mAb, CD40L-CD8 chimeric protein (CD40L), IL-4, anti-IgM Ab, or LPS. Proliferation was assessed by [3H]thymidine incorporation during the last 8 hr. Data are shown as the mean cpm Ϯ SEM from triplicate samples and represent one of three independent experiments with similar results. (C) Splenocytes (for EMSA) or purified B cells (for in vitro kinase assay) isolated from traf5ϩ͞ϩ and traf5Ϫ͞Ϫ mice were stimulated with anti-CD40 mAb for 15 min or LPS for 30 min. EMSA for NF-␬B activation (Left) and in vitro kinase assay for JNK͞SAPK activation (Right) were performed as described in Materials and Methods. B, oligonucleotide probe bound to NF-␬B complexes.

cells could affect the differentiation, traf5ϩ͞ϩ or traf5Ϫ͞Ϫ B activation of NF-␬B and JNK͞SAPK by CD40 was impaired in cells were stimulated with anti-CD40 plus IL-4. Then, Ig traf5Ϫ͞Ϫ B cells. Splenocytes were stimulated with anti-CD40 production was assessed by isotype-specific ELISA. Produc- mAb or LPS for the indicated time for EMSA, and purified B Ϫ͞Ϫ tion of IgM as well as IgG1 by traf5 B cells was reduced cells were stimulated with anti-CD40 mAb for 15 min for in substantially when compared with traf5ϩ͞ϩ B cells (Table 1), vitro kinase assay. As shown in Fig. 3C, CD40-mediated suggesting that TRAF5 is involved in CD40-mediated Ig activation of NF-␬BorJNK͞SAPK apparently was not im- production and class switching. paired in traf5Ϫ͞Ϫ B cells. -It has been shown that CD40-mediated activation of JNK͞ Humoral Responses in traf5؊͞؊ Mice. CD40͞CD40L inter SAPK, but not NF-␬B, was impaired in B cells from action plays a crucial role in humoral responses to T- TRAF2-DN transgenic mice (20). We then examined whether dependent , especially for class switching and germinal

Table 1. In vitro Ig production of traf5ϩ͞ϩ and traf5Ϫ͞Ϫ B cells IgM, ng͞ml IgG1, ng͞ml ϩ͞ϩϪ͞Ϫϩ͞ϩϪ͞Ϫ Experiment 1 271.1 Ϯ 40.0 66.7 Ϯ 4.9 404.7 Ϯ 21.6 130.0 Ϯ 13.9 Experiment 2 327.0 Ϯ 34.1 137.3 Ϯ 7.2 319.8 Ϯ 19.3 222.5 Ϯ 2.9 Experiment 3 746.0 Ϯ 217.6 104.7 Ϯ 18.0 273.3 Ϯ 28.0 80.7 Ϯ 13.3 Purified B cells (1 ϫ 105) were stimulated with anti-CD40 (10 ␮g͞ml) plus IL-4 (10 ng͞ml). After 6 days, IgM and IgG1 concentrations in the culture supernatants were measured by ELISA. IgM and IgG1 from unstimulated cultures were undetectable. Three independent experiments are shown, and the differences between traf5ϩ͞ϩ and traf5Ϫ͞Ϫ are statistically significant (P Ͻ 0.001. Student’s t test). Downloaded by guest on September 26, 2021 Immunology: Nakano et al. Proc. Natl. Acad. Sci. USA 96 (1999) 9807

center formation (29, 30). The defect in CD40-mediated vealed redundant or unique functions of each molecule in Ϫ͞Ϫ activation in traf5 B cells in vitro prompted us to explore particular signaling cascades. TRAF5 and TRAF2 bind over- Ϫ͞Ϫ whether humoral responses were impaired in traf5 mice. lapping members of the TNFR superfamily (3). When over- We first examined the antibody production against a T- expressed, both activate NF-␬B- and JNK͞SAPK-signaling dependent antigen, NP22-CG (molar ratio, 22:1). Eight- to pathways (16, 27), yet TRAF5- and TRAF2-deficient mice 12-week-old traf5ϩ͞ϩ and traf5Ϫ͞Ϫ mice were injected i.p. with ␮ ␮ exhibit distinct phenotypes (19). Although CD40-mediated 50 g (high-dose protocol) or 5 g (low-dose protocol) of activation of NF-␬BorJNK͞SAPK was not affected in NP -CG adsorbed to alum. Each mouse was boosted with 5 ␮g Ϫ͞Ϫ 22 traf5 B cells, up-regulation of surface molecules and pro- of NP -CG 21 days later, and the sera were collected at the 22 liferation induced by CD40 were impaired. The defect of indicated time points in Fig. 4. Total IgM and IgG1 anti-NP- Ϫ͞Ϫ specific antibodies or high-affinity IgG1 anti-NP antibody was CD40-mediated signaling of traf5 B cells also was substan- determined by binding to densely (NP22-BSA) or sparsely tiated by the reduction of IgM and IgG1 production in vitro (NP2-BSA) NP-haptenated BSA, respectively. Ratios of high- upon stimulation with anti-CD40 plus IL-4. A previous study affinity IgG1 antibody to total IgG1 antibody also were showed that up-regulation of CD23, CD80, and Fas induced by calculated. Because we did not observe any significant differ- CD40 was not correlated with the ability of CD40 to activate ence in total IgM and IgG1 titers or high-affinity IgG1 titers NF-␬B (32). Although truncated TRAF3 inhibited up- ϩ͞ϩ Ϫ͞Ϫ between traf5 and traf5 mice (data not shown) in a regulation of CD23 by CD40 (5, 12), CD40-mediated signaling high-dose protocol, we next checked humoral response of including up-regulation of CD23 or proliferation was not Ϫ͞Ϫ Ϫ͞Ϫ Ϫ͞Ϫ traf5 mice to NP22-CG in a low-dose protocol. Traf5 ϩ͞ϩ impaired in B cells (21). Given that TRAF2, -3, and mice consistently produced more anti-NP IgM than traf5 -5 bind to the overlapping region of CD40 (12), truncated mice after primary immunization (Fig. 4A). Although there is TRAF3 could displace endogenous TRAF2 or TRAF5, re- a trend of lower antibody production of both total and Ϫ͞Ϫ sulting in impairment of CD23 up-regulation. Collectively, high-affinity IgG1 in traf5 mice compared with wild-type mice (Fig. 4B), the differences are not statistically significant these results indicated that TRAF5 is actually involved in these based on Student’s t test, Welch’s t test, or nonparametric signaling pathways. CD27-mediated costimulatory signal also was impaired in traf5Ϫ͞Ϫ T cells despite no significant impair- Mann–Whitney test (data not shown). However, there does ␬ ͞ appear to be a reduction in affinity maturation as evidenced by ment in CD27-mediated NF- B and JNK SAPK activation. ͞ ϭ high total ratio IgG1 at 4 weeks (P 0.0537 by Student’s t test Collectively, these results suggest that TRAF5 is involved in and P ϭ 0.0584 by Welch’s t test). These data are similar to CD40- and CD27-mediated signaling independently of NF-␬B those observed in LT-␤-deficient mice, in which there was no and JNK͞SAPK activation. difference in total anti-NP titer but a reduction in affinity CD40͞CD40L interaction is required for class switching and maturation (31). This decrease in affinity maturation also was formation (29, 30). The mild defect in affinity ␤ observed in LT- R-deficient animals at 21 days postimmuni- maturation of IgG1 antibodies to T-dependent antigen ob- zation (28). On the other hand, humoral response to a T- Ϫ͞Ϫ Ϫ͞Ϫ served in traf5 mice can directly reflect the impairment of independent antigen, NP-Ficoll, in traf5 mice was almost CD40-mediated B cell activation observed in vitro. It has been comparable to that in traf5ϩ͞ϩ mice (data not shown). Col- Ϫ͞Ϫ shown that the up-regulation of CD80 and CD86 by CD40- lectively, these results demonstrated that traf5 mice showed mediated stimulation increases the activity of B cells as a mild defect in humoral responses to a T-dependent antigen. antigen-presenting cells. Thus, the partial defect in up- regulation of CD80 and CD86 on traf5Ϫ͞Ϫ B cells also may be DISCUSSION relevant to the mild defect of humoral response to T- Recently, generated knock-out mice of various molecules that dependent antigen. Alternatively, because CD27 also has been mediate signaling via the TNFR superfamily members re- implicated in the regulation of humoral responses in vitro (33),

Ϫ͞Ϫ ϩ͞ϩ Ϫ͞Ϫ FIG. 4. Specific humoral responses and affinity maturation in traf5 mice. Traf5 (F) and traf5 (E) mice were immunized with 5 ␮g of NP22-CG͞alum on day 0 and boosted with the same dose on day 21. NP-specific IgM (A), total IgG1 (B), and high-affinity IgG1 (C) antibodies were measured at the indicated time points. IgM antibodies were determined by ELISA against densely haptenated BSA (NP22-BSA). Total and high-affinity IgG1 antibodies were determined by ELISA against NP22-BSA and NP2-BSA, respectively. The levels of anti-NP IgG1 are indicated Ϫ7 as arbitrary units by using an NP-specific IgG1 mAb of known affinity (C6; Kd ϭ 3.3 ϫ 10 M) as a standard. The levels of anti-NP IgM are indicated as arbitrary units determined by using an immune sera as a standard. In D, high-affinity͞total IgG1 ratios were calculated from the value in B and C for each individual. Bars indicate the means of four mice in each experimental group. Data represent one of two independent experiments with similar results. Downloaded by guest on September 26, 2021 9808 Immunology: Nakano et al. Proc. Natl. Acad. Sci. USA 96 (1999)

the impairment of CD27-mediated signaling also may be 7. Mosialos, G., Birkenbach, M., Yalamanchili, R., VanArsdale, T., involved. Impairment of T-dependent immune response also Ware, C. F. & Kieff, E. (1995) Cell 80, 389–399. was observed in traf3Ϫ͞Ϫ mice (30). Comparing the substantial 8. Regnier, C. H., Tomasetto, C., Moog-Lutz, C., Chenard, M.-P., reduction of in vitro Ig production of traf5Ϫ͞Ϫ B cells with the Wendling, C., Basset, P. & Rio, M.-C. (1995) J. Biol. Chem. 270, only mild defect of humoral response to T-dependent antigen, 25715–25721. other pathways could compensate the defect of CD40- 9. Nakano, H., Oshima, H., Chung, W., Williams-Abbot, L., Ware, Ϫ͞Ϫ C. F., Yagita, H. & Okumura, K. (1996) J. Biol. Chem. 271, mediated signals in traf5 mice in vivo. 14661–14664. Previous studies demonstrated that TRAF2 is not requisite 10. Cao, Z., Xiong, J., Takeuchi, M., Kurama, T. & Goeddel, D. V. for NF-␬B activation by TNF (19, 20). Our preliminary results ␬ (1996) Nature (London) 383, 443–446. based on NF- B-dependent reporter gene assay by using 293 11. Ishida, T., Mizushima, S., Azuma, S., Kobayashi, N., Tojo, T., cells showed that truncated TRAF5 inhibited NF-␬B activa- Suzuki, K., Aizawa, S., Watanabe, T., Mosialos, G., Kieff, E., et tion by TNF (data not shown), suggesting that TRAF5 could al. (1996) J. Biol. Chem. 271, 28745–28748. be involved in NF-␬B activation by TNF. We then asked 12. Ishida, T., Tojo, T., Aoki, T., Kobayashi, N., Ohishi, T., Wa- whether TNF-mediated NF-␬B activation was impaired in tanabe, T., Yamamoto, T. & Inoue, J.-I. (1996) Proc. Natl. Acad. traf5Ϫ͞Ϫ thymocytes or embryonic fibroblast (EF) cells. Un- Sci. USA 93, 9437–9442. expectedly, TNF-mediated NF-␬B activation was not impaired 13. Ashkenazi, A. & Dixit, V. M. (1998) Science 281, 1305–1308. in traf5Ϫ͞Ϫ thymocytes or EF cells (data not shown), suggesting 14. Rothe, M., Sarma, V., Dixit, V. M. & Goeddel, D. V. (1995) that TRAF5 is not requisite for TNF-mediated NF-␬B acti- Science 269, 1424–1427. vation. Alternatively, TRAF2 and TRAF5 might be function- 15. Malinin, N. L., Boldin, M. P., Kovalenko, A. V. & Wallach, D. ␬ (1997) Nature (London) 385, 540–544. ally redundant in TNF-mediated NF- B activation. Formally, 16. Song, H. Y., Regnier, C. H., Kirschning, C. J., Goeddel, D. V. & the possibility cannot be excluded that a molecule other than ␬ Rothe, M. (1997) Proc. Natl. Acad. Sci. USA 94, 9792–9796. TRAF2 or TRAF5 is involved in NF- B activation by the 17. Stancovski, I. & Baltimore, D. (1997) Cell 91, 299–302. TNFR superfamily to account for these results. Generation of 18. Nishitoh, H., Saitoh, M., Mochida, Y., Takeda, K., Nakano, H., TRAF2 and TRAF5 double-deficient mice will be required to Rothe, M., Miyazono, K. & Ichijo, H. (1998) Mol. Cell 2, 389–395. address this issue. 19. Yeh, W.-C., Shahinian, A., Speiser, D., Kraunus, J., Billia, F., As far as we examined, we could not detect any defect in Wakeham, A., de la Pompa, J. L., Ferrick, D., Hum, B., Iscove, LT-␤R-mediated signaling, such as vascular cell adhesion N., et al. (1997) Immunity 7, 715–725. molecule 1 induction and chemokine production, in traf5Ϫ͞Ϫ 20. Lee, S. Y., Reichlin, A., Santana, A., Sokol, K. A., Nussenzweig, EF cells stimulated by agonistic anti-LT-␤R mAb or recom- M. C. & Choi, Y. (1997) Immunity 7, 703–713. binant soluble LT␣1␤2 (data not shown). There was no defect 21. Xu, Y., Cheng, G. & Baltimore, D. (1996) Immunity 5, 407–415. of lymph node development and germinal center formation in 22. Oshima, H., Nakano, H., Nohara, C., Kobata, T., Nakajima, A., Jenkins, N. A., Gilbert, D. J., Copeland, N. G., Muto, T., Yagita, TRAF5-deficient mice (data not shown), suggesting that ␤ H., et al. (1998) Int. Immunol. 10, 517–526. TRAF5 might not be involved in LT- R-mediated signalings. 23. Wurst, W. & Joyner, A. L. (1993) in Gene Targeting: A Practical However, considering that TRAF2 and TRAF5 are implicated ␤ ␬ Approach (IRL, Oxford), pp. 33–61. in LT- R-mediated NF- B activation (9, 34) and lymph node 24. Nagy, A., Rossant, J., Nagy, R., Abramow-Newerly, W. & Roder, development was not impaired in TRAF2-deficient mice (19), J. C. (1993) Proc. Natl. Acad. Sci. USA 90, 8424–8428. TRAF2 and TRAF5 could be redundant in LT-␤R-mediated 25. Shu, H.-B., Takeuchi, M. & Goeddel, D. (1996) Proc. Natl. Acad. signaling. In addition, both TRAF2 and TRAF5 have been Sci. USA 93, 13973–13978. implicated in the signaling via CD30, HVEM, OX40, and 26. Lane, P., Brocker, T., Hubele, S., Padovan, E., Lanzavecchia, A. RANK (3). The traf5Ϫ͞Ϫ mice will be useful for investigating & McConnell, F. (1993) J. Exp. Med. 177, 1209–1213. further the signaling pathways via these receptors. 27. Akiba, H., Nakano, H., Nishinaka, S., Shindo, M., Kobata, T., Morimoto, C., Ware, C. F., Malinin, N. L., Wallach, D., Yagita, We thank Masahisa Shindo, Kazunori Hanaoka, Michiko Hanaoka, H. et al. (1998) J. Biol. Chem. 273, 13353–13358. Kasuyoshi Takeda, Machiko Atsuta, Atsushi Tajima, Masaaki Abe, 28. Futterer, A., Mink, K., Luz, A., Kosco-Vilbois, M. H. & Pfeffer, Yoshimasa Takahashi, Akio Nakane, and Hidehito Kuroyanagi for K. (1998) Immunity 9, 59–70. technical assistance and helpful discussions. We also thank Hidechika 29. Kawabe, T., Naka, T., Yoshida, K., Tanaka, T., Fujiwara, H., Azuma, Hideo Oshima, Noboru Motoyama, David V. Goeddel, Suematsu, S., Yoshida, N., Kishimoto, T. & Kikutani, H. (1994) Jeffrey L. Browning, Peter Lane, and Eisuke Nishida for reagents. This Immunity 1, 167–178. work was supported by grants from the Ministry of Education, Science, 30. Xu, J., Foy, T. M., Laman, J. D., Elliott, E. A., Dunn, J. J., and Culture, the Ministry of Health, Japan. Waldschmidt, T. J., Elsemore, J., Noelle, R. J. & Flavell, R. A. (1994) Immunity 1, 423–431. 1. Beutler, B. & van Huffel, C. (1994) Science 264, 667–668. 31. Koni, P. A., Sacca, R., Lawton, P., Browning, J. L., Ruddle, N. H. 2. Tracey, K. J. & Cerami, A. (1994) Annu. Rev. Med. 45, 491–503. & Flavell, R. A. (1997) Immunity 6, 491–500. 3. Arch, R. H., Gedrich, R. W. & Thompson, C. B. (1998) 32. Hostager, B. S., Hsing, Y., Harms, D. E. & Bishop, G. A. (1996) Dev. 12, 2821–2830. J. Immunol. 157, 1047–1053. 4. Rothe, M., Wong, S. C., Henzel, W. J. & Goeddel, D. V. (1994) 33. Kobata, T., Jacquot, S., Kozlowski, S., Agematsu, K., Schlossman, Cell 78, 681–692. S. F. & Morimoto, C. (1995) Proc. Natl. Acad. Sci. USA 92, 5. Cheng, G., Cleary, A. M., Ye, Z.-s., Hong, D. I., Lederman, S. & 11249–11253. Baltimore, D. (1995) Science 267, 1494–1498. 34. Nakano, H., Shindo, M., Yamada, K., Yoshida, M. C., Santee, 6. Hu, H. M., O’Rourke, K., Boguski, M. S. & Dixit, V. M. (1995) S. M., Ware, C. F., Jenkins, N. A., Gilbert, D. J., Yagita, H., J. Biol. Chem. 269, 30069–30072. Copeland, N. G., et al. (1997) Genomics 42, 26–32. Downloaded by guest on September 26, 2021